NRAM_Controller_10_Final_Dataset5.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "NRAM_Controller_10-Final.ipynb",
      "version": "0.3.2",
      "provenance": [],
      "collapsed_sections": [
        "rGVz5HwlhDtX",
        "OjRq1i1ngydI",
        "T4ApKdVugtiT",
        "3fdYDwRbgclC",
        "3BXUJ9lmiEja",
        "u6c9aQ_vidJ2",
        "ufUl6gMukbfv",
        "0XIeyQKgA7rR",
        "eyRqTvVqBf34",
        "CqOyepkSHrSB",
        "bfhbghdOjijr",
        "zlBQ73u7R0eh",
        "Hnd_YcNF1HyU",
        "fEINnQwk_SbF",
        "Q5Yp8kFGREb4",
        "HFGqpuDkRIw9",
        "vhZ8G5qeRZSW",
        "hLhcTRDPsmzq",
        "nh3ousI_stj8",
        "8qRl86D6O1zu",
        "THPWe7-J1jNf",
        "ESmNV7HR1wj1",
        "k76s7cjDNoB4",
        "edz6c1hoZJ43"
      ]
    },
    "language_info": {
      "codemirror_mode": {
        "name": "ipython",
        "version": 3
      },
      "file_extension": ".py",
      "mimetype": "text/x-python",
      "name": "python",
      "nbconvert_exporter": "python",
      "pygments_lexer": "ipython3",
      "version": "3.7.0"
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    }
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "colab_type": "text",
        "id": "Jiig7MT1g6X7"
      },
      "source": [
        "# NRAM Implementation"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "mzV_aXjjAT9G",
        "colab_type": "text"
      },
      "source": [
        "### --------------------Dataset 5 Implemenation--------------------"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "colab_type": "text",
        "id": "my1537BG-tt8"
      },
      "source": [
        "### Status\n",
        "Completed\n",
        "*   Base Framework\n",
        "*   Code Cleanup\n",
        "*   Neural Network model - MLP, LSTM, LSTM + MLP\n",
        "*  Dataset 1\n",
        "*  Not Gate\n",
        "* Update model to run more gates\n",
        "* Add read/write module\n",
        "* Dataset 2 for read/write module \n",
        "* Dataset 3 for copy\n",
        "* Dataset 4 for basic bucket sort\n",
        "* Dataset 5 larger bucket sort list\n",
        "* Dataset 6 larger bucket sort list\n",
        "\n",
        "To-Do - Priority -2\n",
        "* LSTM - check hidden states\n",
        "* Check Final Error Summation\n",
        "\n",
        "\n",
        "To-Do - Priority -3\n",
        "* Gradient Clipping\n",
        "* Gradient Noise\n",
        "* Entropy\n",
        "* Limit the value of log\n",
        "* Enforcing Distribution Constraint\n",
        "\n",
        "\n",
        "#### Errors\n",
        "* Training\n",
        "* Testing - same size problem\n",
        "* Testing - different sized problem\n",
        "\n",
        "\n",
        "#### 5 Different parameters\n",
        "* Layer size\n",
        "* Learning rate\n",
        "\n",
        "\n",
        "#### Plot\n",
        "Y-axis error\n",
        "X-axis training iteration\n",
        "\n",
        "### Ablation Study\n",
        "After training, perturb the model (added noise, removed layers, etc.) or the data (added noise, shuffled input, etc). Show the impacts of the ablation on the three types of errors, for the 5 different hyper-parameter settings. Randomize the ablations and show aggregated error results (e.g., with histograms or boxplots)\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "colab_type": "text",
        "id": "rGVz5HwlhDtX"
      },
      "source": [
        "## Initial Setup"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "colab_type": "text",
        "id": "OjRq1i1ngydI"
      },
      "source": [
        "### Import libraries"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "j5SUYTG7mkYz",
        "colab": {}
      },
      "source": [
        "from __future__ import print_function\n",
        "\n",
        "import numpy as np\n",
        "import tensorflow as tf\n",
        "\n",
        "from tensorflow.keras import layers\n",
        "\n",
        "import matplotlib.pyplot as plt\n",
        "\n",
        "tf.enable_eager_execution()"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "colab_type": "text",
        "id": "T4ApKdVugtiT"
      },
      "source": [
        "### Create Modules / Gates"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "zGWfXinzlzuM",
        "colab": {}
      },
      "source": [
        "def zeroModule(a, M):\n",
        "    zero = np.asarray([0])\n",
        "    return tf.one_hot(zero, M)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "vMr3C4Bpm_F5",
        "colab": {}
      },
      "source": [
        "def oneModule(a, M):\n",
        "    one = np.asarray([1])\n",
        "    return tf.one_hot(one, M)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "nha3eY8X1IhT",
        "colab": {}
      },
      "source": [
        "def twoModule(a, M):\n",
        "    two = np.asarray([2])\n",
        "    return tf.one_hot(two, M)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "kMZFHeLn1NTQ",
        "colab": {}
      },
      "source": [
        "def incModule(a, M):\n",
        "  return tf.roll(a, shift=1, axis=1)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "5YjJ8iexM5bC",
        "colab": {}
      },
      "source": [
        "def decModule(a, M):\n",
        "  return tf.roll(a, shift=-1, axis=1)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "mVGHQM0eDGfu",
        "colab": {}
      },
      "source": [
        "# def incModule1(a, M):\n",
        "#   aVal = tf.argmax(a, axis=1)\n",
        "#   addVal = tf.mod(tf.add(aVal, 1), a.shape[1])\n",
        "#   return tf.one_hot(addVal, a.shape[1])"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "_p8cSsZSF5bq",
        "colab": {}
      },
      "source": [
        "def addModule(a, b, M):\n",
        "  aVal = tf.argmax(a, axis=1)\n",
        "  bVal = tf.argmax(b, axis=1)\n",
        "  addVal = tf.mod(tf.add(aVal, bVal), a.shape[1])\n",
        "  return tf.one_hot(addVal, a.shape[1])"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "NGTn_A4w9UCi",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# def addModule2(a, b, M):\n",
        "#     bVal = tf.roll(b, shift=-M, axis=1)\n",
        "    \n",
        "#     rows = [tf.roll(b, shift=shift+1, axis=1) for shift in range(M)]\n",
        "#     print(rows)\n",
        "#     bVal = tf.stack(rows)\n",
        "#     print(bVal)\n",
        "#     bVal = tf.reshape(bVal, shape=[4,4,1])\n",
        "#     print(bVal)\n",
        "# #     return bVal\n",
        "#     return tf.tensordot(a, bVal, axes=1)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "2KSYBdSx9UCm",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# addModule2(tf.one_hot([2], M), tf.one_hot([1], M), M)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "ntBMShoxLb_E",
        "colab": {}
      },
      "source": [
        "def subModule(a, b, M):\n",
        "  aVal = tf.argmax(a, axis=1)\n",
        "  bVal = tf.argmax(b, axis=1)\n",
        "  subtractVal = tf.mod(tf.subtract(aVal, bVal), M)\n",
        "  return tf.one_hot(subtractVal, M)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "vDTe2xahNgaz",
        "colab": {}
      },
      "source": [
        "def lessModule(a, b, M):\n",
        "  aVal = tf.argmax(a, axis=1)\n",
        "  bVal = tf.argmax(b, axis=1)\n",
        "  lessVal = tf.less(aVal, bVal)\n",
        "  return tf.one_hot(tf.cast(lessVal, tf.int64), M)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "2g3ZQKmUQpJM",
        "colab": {}
      },
      "source": [
        "def lessOrEqualThenModule(a, b, M):\n",
        "  aVal = tf.argmax(a, axis=1)\n",
        "  bVal = tf.argmax(b, axis=1)\n",
        "  lessVal = tf.less_equal(aVal, bVal)\n",
        "  return tf.one_hot(tf.cast(lessVal, tf.int64), M)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "vHjdqWraQ0iw",
        "colab": {}
      },
      "source": [
        "def equalModule(a, b, M):\n",
        "  aVal = tf.argmax(a, axis=1)\n",
        "  bVal = tf.argmax(b, axis=1)\n",
        "  equalVal = tf.equal(aVal, bVal)\n",
        "  return tf.one_hot(tf.cast(equalVal, tf.int64), M)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "33Gou21oRRYD",
        "colab": {}
      },
      "source": [
        "def minimumModule(a, b, M):\n",
        "  aVal = tf.argmax(a, axis=1)\n",
        "  bVal = tf.argmax(b, axis=1)\n",
        "  minimumVal = tf.minimum(aVal, bVal)\n",
        "  return tf.one_hot(tf.cast(minimumVal, tf.int64), M)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "sNFEl44sRrpi",
        "colab": {}
      },
      "source": [
        "def maximumModule(a, b, M):\n",
        "  aVal = tf.argmax(a, axis=1)\n",
        "  bVal = tf.argmax(b, axis=1)\n",
        "  maximumVal = tf.maximum(aVal, bVal)\n",
        "  return tf.one_hot(tf.cast(maximumVal, tf.int64), M)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "colab_type": "text",
        "id": "3fdYDwRbgclC"
      },
      "source": [
        "### Test Modules"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "609r8cUKwgF2",
        "colab": {}
      },
      "source": [
        "#Note : Temporary M Values\n",
        "M = 20"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "Qdf9Kx1aDtpN",
        "colab": {}
      },
      "source": [
        "# incModule1(tf.one_hot([1, 2], M), M)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "ArqxHhD0UcKJ",
        "outputId": "53ab329b-c07f-4b17-ce62-8a1d03c2c93a",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 122
        }
      },
      "source": [
        "incModule(tf.one_hot([1, 2], M), M)"
      ],
      "execution_count": 19,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<tf.Tensor: id=7, shape=(2, 20), dtype=float32, numpy=\n",
              "array([[0., 0., 1., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0.,\n",
              "        0., 0., 0., 0.],\n",
              "       [0., 0., 0., 1., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0.,\n",
              "        0., 0., 0., 0.]], dtype=float32)>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 19
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "fK0ApcZ6VJcN",
        "outputId": "275ca5ef-6a86-4030-f912-63464c4382cb",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 122
        }
      },
      "source": [
        "decModule(tf.one_hot([4, 9], M), M)"
      ],
      "execution_count": 20,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<tf.Tensor: id=16, shape=(2, 20), dtype=float32, numpy=\n",
              "array([[0., 0., 0., 1., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0.,\n",
              "        0., 0., 0., 0.],\n",
              "       [0., 0., 0., 0., 0., 0., 0., 0., 1., 0., 0., 0., 0., 0., 0., 0.,\n",
              "        0., 0., 0., 0.]], dtype=float32)>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 20
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "ycjL_7EpI-t-",
        "outputId": "1efc8713-bd93-4bfc-d810-c2ff95a57a82",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 122
        }
      },
      "source": [
        "addModule(tf.one_hot([1, 2], M), tf.one_hot([1, 2], M), M)"
      ],
      "execution_count": 21,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<tf.Tensor: id=38, shape=(2, 20), dtype=float32, numpy=\n",
              "array([[0., 0., 1., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0.,\n",
              "        0., 0., 0., 0.],\n",
              "       [0., 0., 0., 0., 1., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0.,\n",
              "        0., 0., 0., 0.]], dtype=float32)>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 21
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "DhQoWIsD9UDu",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# addModule2(tf.one_hot([2], M), tf.one_hot([1], M), M)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "dax-xXeMgufM",
        "outputId": "b415f0b0-d395-4312-cf3e-d8837be29cca",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 54
        }
      },
      "source": [
        "addModule(np.array([[0., 0., 0., 0.5, 0.5]]), np.array([[0., 0., 0., 1., 0.]]), M)"
      ],
      "execution_count": 23,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<tf.Tensor: id=52, shape=(1, 5), dtype=float32, numpy=array([[0., 1., 0., 0., 0.]], dtype=float32)>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 23
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "colab_type": "text",
        "id": "PvK-yReagQkw"
      },
      "source": [
        "**Note: One hot of values greater than M is 0**"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "M8TZ2g58gRgR",
        "outputId": "e442360a-ae63-4366-b326-8581290599b7",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 122
        }
      },
      "source": [
        "tf.one_hot([7, 8], M)"
      ],
      "execution_count": 24,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<tf.Tensor: id=58, shape=(2, 20), dtype=float32, numpy=\n",
              "array([[0., 0., 0., 0., 0., 0., 0., 1., 0., 0., 0., 0., 0., 0., 0., 0.,\n",
              "        0., 0., 0., 0.],\n",
              "       [0., 0., 0., 0., 0., 0., 0., 0., 1., 0., 0., 0., 0., 0., 0., 0.,\n",
              "        0., 0., 0., 0.]], dtype=float32)>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 24
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "IvnYQqPsLl20",
        "outputId": "78d15468-5452-410a-8887-d23e16484063",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 122
        }
      },
      "source": [
        "subModule(tf.one_hot([4, 8], M), tf.one_hot([6, 7], M), M)"
      ],
      "execution_count": 25,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<tf.Tensor: id=80, shape=(2, 20), dtype=float32, numpy=\n",
              "array([[0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0.,\n",
              "        0., 0., 1., 0.],\n",
              "       [0., 1., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0.,\n",
              "        0., 0., 0., 0.]], dtype=float32)>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 25
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "KRczko6xMStv",
        "outputId": "f2e3433c-f20f-475c-a33e-5882124c445d",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 122
        }
      },
      "source": [
        "incModule(tf.one_hot([4, 9], M), M)"
      ],
      "execution_count": 26,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<tf.Tensor: id=89, shape=(2, 20), dtype=float32, numpy=\n",
              "array([[0., 0., 0., 0., 0., 1., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0.,\n",
              "        0., 0., 0., 0.],\n",
              "       [0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 1., 0., 0., 0., 0., 0.,\n",
              "        0., 0., 0., 0.]], dtype=float32)>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 26
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "7FmUCjtIND3u",
        "outputId": "7e489cd8-9932-4018-d04e-7f0842878a3c",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 122
        }
      },
      "source": [
        "decModule(tf.one_hot([4, 9], M), M)"
      ],
      "execution_count": 27,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<tf.Tensor: id=98, shape=(2, 20), dtype=float32, numpy=\n",
              "array([[0., 0., 0., 1., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0.,\n",
              "        0., 0., 0., 0.],\n",
              "       [0., 0., 0., 0., 0., 0., 0., 0., 1., 0., 0., 0., 0., 0., 0., 0.,\n",
              "        0., 0., 0., 0.]], dtype=float32)>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 27
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "_SkTggdrOe64",
        "outputId": "ae499acb-ed57-4fe2-d4b6-624039a55cbb",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 122
        }
      },
      "source": [
        "lessModule(tf.one_hot([8, 5], M), tf.one_hot([6, 7], M), M)"
      ],
      "execution_count": 28,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<tf.Tensor: id=119, shape=(2, 20), dtype=float32, numpy=\n",
              "array([[1., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0.,\n",
              "        0., 0., 0., 0.],\n",
              "       [0., 1., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0.,\n",
              "        0., 0., 0., 0.]], dtype=float32)>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 28
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "k0d33tqhRDgU",
        "outputId": "d2a3de8a-06ff-4273-9416-51912e577020",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 122
        }
      },
      "source": [
        "lessOrEqualThenModule(tf.one_hot([8, 5], M), tf.one_hot([8, 7], M), M)"
      ],
      "execution_count": 29,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<tf.Tensor: id=140, shape=(2, 20), dtype=float32, numpy=\n",
              "array([[0., 1., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0.,\n",
              "        0., 0., 0., 0.],\n",
              "       [0., 1., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0.,\n",
              "        0., 0., 0., 0.]], dtype=float32)>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 29
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "7fK2kzaWRDsD",
        "outputId": "fabece90-779f-493f-863b-d31a14411507",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 122
        }
      },
      "source": [
        "equalModule(tf.one_hot([8, 5], M), tf.one_hot([8, 7], M), M)"
      ],
      "execution_count": 30,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<tf.Tensor: id=161, shape=(2, 20), dtype=float32, numpy=\n",
              "array([[0., 1., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0.,\n",
              "        0., 0., 0., 0.],\n",
              "       [1., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0.,\n",
              "        0., 0., 0., 0.]], dtype=float32)>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 30
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "pigT9MXeRnGU",
        "outputId": "018b8796-1e05-436a-fc73-b7f2481a4fcd",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 122
        }
      },
      "source": [
        "minimumModule(tf.one_hot([8, 5], M), tf.one_hot([8, 7], M), M)"
      ],
      "execution_count": 31,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<tf.Tensor: id=181, shape=(2, 20), dtype=float32, numpy=\n",
              "array([[0., 0., 0., 0., 0., 0., 0., 0., 1., 0., 0., 0., 0., 0., 0., 0.,\n",
              "        0., 0., 0., 0.],\n",
              "       [0., 0., 0., 0., 0., 1., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0.,\n",
              "        0., 0., 0., 0.]], dtype=float32)>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 31
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "colab_type": "text",
        "id": "3BXUJ9lmiEja"
      },
      "source": [
        "### Circuit Implementation"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "5M1uj2sjWZ5u",
        "colab": {}
      },
      "source": [
        "# def dotProduct(a, b):\n",
        "#   print(\"a \",a.shape)\n",
        "#   print(\"b \", b.shape)\n",
        "#   if len(a.shape) == 2:\n",
        "#     return tf.tensordot(a.T, b, axes=1)\n",
        "#   else:\n",
        "#     output = []\n",
        "#     for i in range(a.shape[0]):\n",
        "#       aT = tf.transpose(a[i])\n",
        "#       output.append(tf.tensordot(aT, b[i], axes=1))\n",
        "#     return tf.stack(output)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "geOH2iAy1BAj",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "def dotProduct(a, b):\n",
        "  if len(b.shape) > 1:\n",
        "    print(\"error dotProduct shape: \", b.shape)\n",
        "  aT = tf.transpose(a)\n",
        "  aT = tf.reshape(aT, [a.shape[0],a.shape[2],a.shape[1]])\n",
        "  return tf.tensordot(aT, b, axes=1)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "NwIjw_u3YfAm",
        "outputId": "df4bc02b-09cc-47a4-fd64-9eb5a3a3d435",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 54
        }
      },
      "source": [
        "a = np.array([[\n",
        "    [1., 2., 3.],\n",
        "    [4., 5., 6.],\n",
        "]])\n",
        "\n",
        "b = np.array([1., 0.])\n",
        "dotProduct(a, b)\n",
        "\n",
        "\n",
        "a = np.array([[\n",
        "    [1., 2., 3.],\n",
        "    [4., 5., 6.],\n",
        "]])\n",
        "\n",
        "b = np.array([0., 1.])\n",
        "dotProduct(a, b)"
      ],
      "execution_count": 34,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<tf.Tensor: id=216, shape=(1, 3), dtype=float64, numpy=array([[4., 5., 6.]])>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 34
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ZD4UW1C5nBIY",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# a = np.array([\n",
        "#     [\n",
        "#       [1., 2., 3.],\n",
        "#       [4., 5., 6.],\n",
        "#     ],\n",
        "#     [\n",
        "#       [1., 2., 3.],\n",
        "#       [4., 5., 6.],\n",
        "#     ],\n",
        "# ])\n",
        "\n",
        "# b = np.array([\n",
        "#       [1., 0.],\n",
        "#       [0., 1.]\n",
        "#     ])\n",
        "# dotProduct(a, b)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "colab_type": "text",
        "id": "u6c9aQ_vidJ2"
      },
      "source": [
        "#### Gates List"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "szypKQYPHhz_",
        "colab": {}
      },
      "source": [
        "gates = [incModule, decModule, addModule]"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "3t2_sRNQLxQA",
        "outputId": "b07d3efd-2ab7-4104-a730-85d6438d2953",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 122
        }
      },
      "source": [
        "gates[0](tf.one_hot([1, 9], M), M)"
      ],
      "execution_count": 37,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<tf.Tensor: id=225, shape=(2, 20), dtype=float32, numpy=\n",
              "array([[0., 0., 1., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0.,\n",
              "        0., 0., 0., 0.],\n",
              "       [0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 1., 0., 0., 0., 0., 0.,\n",
              "        0., 0., 0., 0.]], dtype=float32)>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 37
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "colab_type": "text",
        "id": "ufUl6gMukbfv"
      },
      "source": [
        "#### Gate Controller Output"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "colab_type": "text",
        "id": "nkJ0U4JQkx6Y"
      },
      "source": [
        "Runs the input which is selected by controller values into the gate passed and returns the output from the gate"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "zzzDSRfSMCG3",
        "colab": {}
      },
      "source": [
        "def gateOutput(gate, inputs, controllerValues, M):\n",
        "  if(controllerValues.shape[0] > 1):\n",
        "    gateInputA = dotProduct(inputs, controllerValues[0])\n",
        "    gateInputB = dotProduct(inputs, controllerValues[1])\n",
        "    output = gate(gateInputA, gateInputB, M)\n",
        "  else:\n",
        "    gateInputA = dotProduct(inputs, controllerValues[0])\n",
        "    output = gate(gateInputA, M)\n",
        "  \n",
        "  return output"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "WJtRToqMb-t6",
        "colab": {}
      },
      "source": [
        "ip = np.array([[\n",
        "  [1., 0., 0., 0., 0.],\n",
        "  [0., 1., 0., 0., 0.],\n",
        "  [0., 0., 0., 1., 0.],\n",
        "  [0., 1., 0., 0., 0.]\n",
        " ]])"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "tHk-AU-JcPG1",
        "outputId": "7195523d-d47f-487a-b72a-35aa193d8752",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 54
        }
      },
      "source": [
        "cp = np.array([[0., 0., 0., 1.],\n",
        "              [0., 0., 1., 0.]])\n",
        "cp.shape"
      ],
      "execution_count": 40,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "(2, 4)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 40
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "colab_type": "text",
        "id": "-v88H8xMIJNb"
      },
      "source": [
        "Run gate to check output - Adds ip 4th value with 3rd value"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "-QwYscutXBAM",
        "colab": {}
      },
      "source": [
        "outputVal = gateOutput(gates[2], ip, cp, M)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "2Yib6hWAprpV",
        "outputId": "57820c97-ffc2-4bfe-fdc8-1c803bca30bc",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 54
        }
      },
      "source": [
        "outputVal"
      ],
      "execution_count": 42,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<tf.Tensor: id=271, shape=(1, 5), dtype=float32, numpy=array([[0., 0., 0., 0., 1.]], dtype=float32)>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 42
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "v6GT03BdqqW3",
        "outputId": "5bda6505-cd20-440c-c922-67db0fbfcde2",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 122
        }
      },
      "source": [
        "tf.dtypes.cast(ip, tf.float32)"
      ],
      "execution_count": 43,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<tf.Tensor: id=274, shape=(1, 4, 5), dtype=float32, numpy=\n",
              "array([[[1., 0., 0., 0., 0.],\n",
              "        [0., 1., 0., 0., 0.],\n",
              "        [0., 0., 0., 1., 0.],\n",
              "        [0., 1., 0., 0., 0.]]], dtype=float32)>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 43
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "colab_type": "text",
        "id": "SFsUtDGcytne"
      },
      "source": [
        "For adding the output to the registers"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "mw9hvmJKp4Y0",
        "outputId": "0e866bad-816c-42e6-b214-9b1067355f67",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 139
        }
      },
      "source": [
        "tf.concat([ip, [outputVal]], axis=1)"
      ],
      "execution_count": 44,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<tf.Tensor: id=280, shape=(1, 5, 5), dtype=float64, numpy=\n",
              "array([[[1., 0., 0., 0., 0.],\n",
              "        [0., 1., 0., 0., 0.],\n",
              "        [0., 0., 0., 1., 0.],\n",
              "        [0., 1., 0., 0., 0.],\n",
              "        [0., 0., 0., 0., 1.]]])>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 44
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "wtlsZZwqzQDH",
        "outputId": "e11a86d4-e1d4-45b0-f376-89455200f141",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 88
        }
      },
      "source": [
        "zeroModule([], M)"
      ],
      "execution_count": 45,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<tf.Tensor: id=286, shape=(1, 20), dtype=float32, numpy=\n",
              "array([[1., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0.,\n",
              "        0., 0., 0., 0.]], dtype=float32)>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 45
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "colab_type": "text",
        "id": "0XIeyQKgA7rR"
      },
      "source": [
        "#### Input Register"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "g6LRxD2_0Dpf",
        "colab": {}
      },
      "source": [
        "def registerValueModule(a):\n",
        "    aVal = np.asarray([a])\n",
        "    return tf.one_hot(aVal, M)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "M2eRTYos0Lib",
        "outputId": "311370a4-fe20-4220-8741-6bf5a3ae8574",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 88
        }
      },
      "source": [
        "registerValueModule(2)"
      ],
      "execution_count": 47,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<tf.Tensor: id=292, shape=(1, 20), dtype=float32, numpy=\n",
              "array([[0., 0., 1., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0.,\n",
              "        0., 0., 0., 0.]], dtype=float32)>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 47
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "rIsnozvvyq1X",
        "colab": {}
      },
      "source": [
        "registers = tf.concat([registerValueModule(0), registerValueModule(1), registerValueModule(3)], axis=0)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "colab_type": "text",
        "id": "K_Y2_qNYBMOY"
      },
      "source": [
        "**Registers = [0,1,3]**"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "o0-ZLeDezsj9",
        "colab": {}
      },
      "source": [
        "registers = tf.reshape(registers, shape=[1, registers.shape[0], registers.shape[1]])"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "yFAr3khI06Cs",
        "outputId": "a86e53f0-d9f7-4116-cfde-84ac6544a508",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 156
        }
      },
      "source": [
        "registers"
      ],
      "execution_count": 50,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<tf.Tensor: id=312, shape=(1, 3, 20), dtype=float32, numpy=\n",
              "array([[[1., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0.,\n",
              "         0., 0., 0., 0.],\n",
              "        [0., 1., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0.,\n",
              "         0., 0., 0., 0.],\n",
              "        [0., 0., 0., 1., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0.,\n",
              "         0., 0., 0., 0.]]], dtype=float32)>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 50
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "colab_type": "text",
        "id": "eyRqTvVqBf34"
      },
      "source": [
        "#### Update data"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "-aMLqPU7h049",
        "colab": {}
      },
      "source": [
        "def runTimeStep(gates, registers, controllerValues_, registercontrollerValues_):\n",
        "  updatedRegisters = registers\n",
        "  for i in range(len(gates)):\n",
        "    output = gateOutput(gates[i], updatedRegisters, controllerValues_[i], M)\n",
        "    updatedRegisters = tf.concat([updatedRegisters, [output]], axis=1)\n",
        "\n",
        "  output_registers = []\n",
        "\n",
        "  for i in range(registers.shape[1]):\n",
        "    output = dotProduct(updatedRegisters, registercontrollerValues_[i])\n",
        "    output_registers.append(output)\n",
        "  \n",
        "  output_registers = tf.stack(output_registers, axis=1)\n",
        "  \n",
        "  return output_registers"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "colab_type": "text",
        "id": "6g-9SmRHGEbw"
      },
      "source": [
        "**Circuit Testing 1**"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "JAQOP0RvF7GV",
        "outputId": "611bc4df-8a42-4620-c4e4-36b4c619b75d",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 207
        }
      },
      "source": [
        "# registers = [2, 1, 3]\n",
        "registers = tf.concat([registerValueModule(2), registerValueModule(1), registerValueModule(3)], axis=0)\n",
        "registers = tf.reshape(registers, shape=[1, registers.shape[0], registers.shape[1]])\n",
        "\n",
        "print(\"Input registers: \", tf.argmax(registers, axis=2))\n",
        "\n",
        "# Gatest -> Increment, Decrement, Addition\n",
        "gates = [incModule, decModule, addModule]\n",
        "\n",
        "# Increment input r1 -> decrement output o1 -> add output o2 + input r1\n",
        "controllerValues_ = [\n",
        "    [[0, 1, 0]],\n",
        "    [[0, 0, 0, 1]],\n",
        "    [[0, 0, 0, 0, 1],\n",
        "     [0, 1, 0, 0, 0]]\n",
        "]\n",
        "\n",
        "# Save the input r2 -> output r0, input r0 -> output r1, last gate output -> output r2\n",
        "registerControllerValues_ = [\n",
        "    [0, 0, 1, 0, 0, 0],\n",
        "    [1, 0, 0, 0, 0, 0],\n",
        "    [0, 0, 0, 0, 0, 1]\n",
        "]\n",
        "\n",
        "controllerValues = []\n",
        "for value in controllerValues_:\n",
        "  controllerValues.append(tf.dtypes.cast(value, tf.float32))\n",
        "\n",
        "registerControllerValues = []\n",
        "for value in registerControllerValues_:\n",
        "  registerControllerValues.append(tf.dtypes.cast(value, tf.float32))\n",
        "\n",
        "# Calling the function to run the complete circuit\n",
        "output_registers = runTimeStep(gates, registers, controllerValues, registerControllerValues)\n",
        "\n",
        "print(\"\\noutput_registers:\\n\", output_registers)\n",
        "print(\"\\noutput_registers:\\n\", tf.argmax(output_registers, axis=2))"
      ],
      "execution_count": 52,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Input registers:  tf.Tensor([[2 1 3]], shape=(1, 3), dtype=int64)\n",
            "\n",
            "output_registers:\n",
            " tf.Tensor(\n",
            "[[[0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            "  [0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            "  [0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]]], shape=(1, 3, 20), dtype=float32)\n",
            "\n",
            "output_registers:\n",
            " tf.Tensor([[3 2 2]], shape=(1, 3), dtype=int64)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "colab_type": "text",
        "id": "7niScH_Df2Oo"
      },
      "source": [
        "**Circuit Testing 2**"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "Joyn2Q0kuAKE",
        "outputId": "9bd77876-cf9c-47e8-a35b-b76289ac41f8",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 479
        }
      },
      "source": [
        "gates = [addModule]\n",
        "\n",
        "controllerValues_ = [\n",
        "    [\n",
        "        [1, 0, 0],\n",
        "        [0, 1, 0]\n",
        "    ]\n",
        "]\n",
        "\n",
        "registerControllerValues_ = [\n",
        "    [0, 0, 0, 1],\n",
        "    [0, 1, 0, 0],\n",
        "    [0, 0, 1, 0]\n",
        "]\n",
        "\n",
        "\n",
        "controllerValues = []\n",
        "for value in controllerValues_:\n",
        "  controllerValues.append(tf.dtypes.cast(value, tf.float32))\n",
        "\n",
        "registerControllerValues = []\n",
        "for value in registerControllerValues_:\n",
        "  registerControllerValues.append(tf.dtypes.cast(value, tf.float32))\n",
        "\n",
        "# First set of input values\n",
        "registers = tf.concat([registerValueModule(1), registerValueModule(1), registerValueModule(3)], axis=0)\n",
        "registers = tf.reshape(registers, shape=[1, registers.shape[0], registers.shape[1]])\n",
        "\n",
        "print(\"Input register set 1: \", tf.argmax(registers, axis=2))\n",
        "\n",
        "print(\"\\nRunning input r0 , r1 -> add -> output r0\")\n",
        "\n",
        "# Calling the function to run the complete circuit\n",
        "output_registers = runTimeStep(gates, registers, controllerValues, registerControllerValues)\n",
        "\n",
        "print(\"\\noutput_registers:\\n\", output_registers)\n",
        "print(\"\\noutput_registers:\\n\", tf.argmax(output_registers, axis=2))\n",
        "\n",
        "\n",
        "# Second set of input values\n",
        "registers = tf.concat([registerValueModule(2), registerValueModule(1), registerValueModule(3)], axis=0)\n",
        "registers = tf.reshape(registers, shape=[1, registers.shape[0], registers.shape[1]])\n",
        "\n",
        "print(\"\\n\\nInput register set 2: \", tf.argmax(registers, axis=2))\n",
        "\n",
        "print(\"\\nRunning input r0 , r1 -> add -> output r0\")\n",
        "# Calling the function to run the complete circuit\n",
        "output_registers = runTimeStep(gates, registers, controllerValues, registerControllerValues)\n",
        "\n",
        "print(\"\\noutput_registers:\\n\", output_registers)\n",
        "print(\"\\noutput_registers:\\n\", tf.argmax(output_registers, axis=2))"
      ],
      "execution_count": 53,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Input register set 1:  tf.Tensor([[1 1 3]], shape=(1, 3), dtype=int64)\n",
            "\n",
            "Running input r0 , r1 -> add -> output r0\n",
            "\n",
            "output_registers:\n",
            " tf.Tensor(\n",
            "[[[0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            "  [0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            "  [0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]]], shape=(1, 3, 20), dtype=float32)\n",
            "\n",
            "output_registers:\n",
            " tf.Tensor([[2 1 3]], shape=(1, 3), dtype=int64)\n",
            "\n",
            "\n",
            "Input register set 2:  tf.Tensor([[2 1 3]], shape=(1, 3), dtype=int64)\n",
            "\n",
            "Running input r0 , r1 -> add -> output r0\n",
            "\n",
            "output_registers:\n",
            " tf.Tensor(\n",
            "[[[0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            "  [0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            "  [0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]]], shape=(1, 3, 20), dtype=float32)\n",
            "\n",
            "output_registers:\n",
            " tf.Tensor([[3 1 3]], shape=(1, 3), dtype=int64)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "CqOyepkSHrSB",
        "colab_type": "text"
      },
      "source": [
        "### Memory"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "aUfwkdVhHtc2",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "def readMemory(pointer, memory, M):\n",
        "  # pointer -> 1,M\n",
        "  # memory -> M, M\n",
        "  # output -> 1, M\n",
        "  val = tf.tensordot(pointer, memory, axes=1)\n",
        "  return val\n",
        "\n",
        "def registerModule(a, M):\n",
        "    aVal = np.asarray([a])\n",
        "    return tf.one_hot(aVal, M)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "pLUCpxF9HuW2",
        "colab_type": "code",
        "outputId": "978663e9-3285-4a62-8b24-517ac734026b",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 258
        }
      },
      "source": [
        "registers = []\n",
        "for _ in range(10):\n",
        "  registers.append(registerModule(_+2, 10))\n",
        "registers = tf.concat(registers, axis=0)\n",
        "print(registers)\n",
        "\n",
        "pointer = np.array([[0., 0., 0., 0., 0., 0.5, 0.5, 0., 0., 0.]], dtype=\"float32\")\n",
        "\n",
        "print(pointer)\n",
        "\n",
        "val = readMemory(pointer, registers, 10)\n",
        "\n",
        "print(val)"
      ],
      "execution_count": 55,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "tf.Tensor(\n",
            "[[0. 0. 1. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 1. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 1. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 1. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 1. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 1. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 1.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]], shape=(10, 10), dtype=float32)\n",
            "[[0.  0.  0.  0.  0.  0.5 0.5 0.  0.  0. ]]\n",
            "tf.Tensor([[0.  0.  0.  0.  0.  0.  0.  0.5 0.5 0. ]], shape=(1, 10), dtype=float32)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Yprjj7u-HzAm",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "def writeMemory(pointer, value, memory, M):\n",
        "  # newValToWrite -> M, M\n",
        "  # mask -> M, M\n",
        "  # M:=(J−p)JT ·M+paT\n",
        "  \n",
        "  newValToWrite = tf.tensordot(tf.transpose(pointer), value, axes=1)\n",
        "\n",
        "  J = np.ones([1,M], dtype=\"float32\")\n",
        "  erase = tf.transpose(J-pointer)\n",
        "  \n",
        "  mask = tf.tensordot(erase, J, axes=1)\n",
        "  \n",
        "  oldMemory = mask * memory\n",
        "\n",
        "  return oldMemory + newValToWrite"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "kj3rl2guH195",
        "colab_type": "code",
        "outputId": "7bcbdcf7-e86c-42be-e482-cd0b48b84577",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 394
        }
      },
      "source": [
        "registers = []\n",
        "for _ in range(10):\n",
        "  registers.append(registerModule(_, 10))\n",
        "memory = tf.concat(registers, axis=0)\n",
        "memory = np.array(memory, dtype=\"float32\")\n",
        "print(memory)\n",
        "\n",
        "pointer = np.array([[0., 1., 0., 0., 0., 0., 0., 0., 0., 0.]], dtype=\"float32\")\n",
        "value    = np.array([[0., 0., 0., 0., 0., 0., 0., 1., 0., 0.]], dtype=\"float32\")\n",
        "\n",
        "outMemory = writeMemory(pointer, value, memory, 10)\n",
        "print(outMemory)"
      ],
      "execution_count": 57,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "[[1. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 1. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 1. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 1. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 1. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 1. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 1. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 1. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 1.]]\n",
            "tf.Tensor(\n",
            "[[1. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 1. 0. 0.]\n",
            " [0. 0. 1. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 1. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 1. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 1. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 1. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 1. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 1.]], shape=(10, 10), dtype=float32)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "colab_type": "text",
        "id": "bfhbghdOjijr"
      },
      "source": [
        "## NRAM Controller Model"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "colab_type": "text",
        "id": "zlBQ73u7R0eh"
      },
      "source": [
        "#### NRAM Controller MLP Model"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "A5bIwnH5zNie",
        "colab": {}
      },
      "source": [
        "def gateOutputNN(gate, registers, controllerValues, M):\n",
        "  if(len(controllerValues) > 1):\n",
        "    gateInputA = dotProduct(registers, controllerValues[0][0])\n",
        "    gateInputB = dotProduct(registers, controllerValues[1][0])\n",
        "    output = gate(gateInputA, gateInputB, M)\n",
        "  else:\n",
        "    gateInputA = dotProduct(registers, controllerValues[0][0])\n",
        "    output = gate(gateInputA, M)\n",
        "  return output"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "QFI7ujB2jPeZ",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# Handle extra b size\n",
        "def dotProductMemory(a, b):\n",
        "  aT = tf.transpose(a)\n",
        "  aT = tf.reshape(aT, [a.shape[0],a.shape[2],a.shape[1]])\n",
        "  return tf.tensordot(aT, b[0], axes=1)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "GjfXD0Jbt-da",
        "colab": {}
      },
      "source": [
        "def runTimeStepNN(gates, registers, memory, readModule, writeModule, controllerValues_, registercontrollerValues_):\n",
        "  updatedRegisters = registers\n",
        "\n",
        "  M = memory.shape[0]\n",
        "  pointer = dotProductMemory(updatedRegisters, controllerValues_[0][0])\n",
        "  value = readModule(pointer, memory, M)\n",
        "\n",
        "  value= tf.reshape(value, shape=[value.shape[0], 1, value.shape[1]])\n",
        "  updatedRegisters = tf.concat([updatedRegisters, value], axis=1)\n",
        "  \n",
        "  for i in range(len(gates)):\n",
        "    output = gateOutputNN(gates[i], updatedRegisters, controllerValues_[i+1], M)\n",
        "    output= tf.reshape(output, shape=[updatedRegisters.shape[0], 1, updatedRegisters.shape[2]])\n",
        "    updatedRegisters = tf.concat([updatedRegisters, output], axis=1)\n",
        "  \n",
        "  # controllerValues_[-1] => [Pointer, Value]\n",
        "  writePointer = dotProductMemory(updatedRegisters, controllerValues_[-1][0])\n",
        "  writeValue = dotProductMemory(updatedRegisters, controllerValues_[-1][1])\n",
        "  \n",
        "  outputMemory = writeMemory(writePointer, writeValue, memory, M)\n",
        "  writePointer = tf.reshape(writePointer, shape=[writePointer.shape[0], 1, writePointer.shape[1]])\n",
        "  updatedRegisters = tf.concat([updatedRegisters, writePointer], axis=1)\n",
        "\n",
        "  outputRegisters = []\n",
        "  for i in range(registers.shape[1]):\n",
        "    output = dotProduct(updatedRegisters, registercontrollerValues_[i][0])\n",
        "    outputRegisters.append(output)\n",
        "  \n",
        "  outputRegisters = tf.stack(outputRegisters, axis=1)\n",
        "  \n",
        "  return outputRegisters, outputMemory"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "bH6eayDVWVg3",
        "colab": {}
      },
      "source": [
        "class NRAMController(tf.keras.Model):\n",
        "   def __init__(self, hidden_dimensions=100, gateControllerDimensions=[[3], [4,4]], registerControllerDimensions=[4, 4, 4]):\n",
        "      super(NRAMController, self).__init__()\n",
        "      self.Hidden1 = tf.keras.layers.Dense(hidden_dimensions, input_shape=(3,), use_bias=True, activation=\"relu\")\n",
        "      self.Hidden2 = tf.keras.layers.Dense(hidden_dimensions, input_shape=(hidden_dimensions,), use_bias=True, activation=\"relu\")\n",
        "\n",
        "      self.gateControllerDimensions = gateControllerDimensions\n",
        "      self.registerControllerDimensions = registerControllerDimensions\n",
        "\n",
        "      # Gate Inputs Controller values \n",
        "      self.GateControllerOutput = []\n",
        "      for gateControllerValue in gateControllerDimensions:\n",
        "        GateController = []\n",
        "        for dim in gateControllerValue:\n",
        "          GateController.append(tf.keras.layers.Dense(dim, input_shape=(hidden_dimensions,), use_bias=True, activation=\"softmax\"))\n",
        "        self.GateControllerOutput.append(GateController)\n",
        "\n",
        "      # Register Output Controller Values\n",
        "      self.RegisterController = []\n",
        "      for i in registerControllerDimensions:\n",
        "        self.RegisterController.append(tf.keras.layers.Dense(i, input_shape=(hidden_dimensions,), use_bias=True, activation=\"softmax\"))\n",
        "\n",
        "      # Complete status\n",
        "      self.CompletedStatus = tf.keras.layers.Dense(1, input_shape=(hidden_dimensions,), use_bias=True, activation=\"sigmoid\")\n",
        "   \n",
        "   def call(self, input_registers):\n",
        "      # hidden layers\n",
        "      hidden1 = self.Hidden1(input_registers)\n",
        "      hidden2 = self.Hidden2(hidden1)\n",
        "      \n",
        "      # To get gate controller values\n",
        "      gateControllerOutput = []\n",
        "      for i in range(len(self.gateControllerDimensions)):\n",
        "        gateController = []\n",
        "        for j in range(len(self.gateControllerDimensions[i])):\n",
        "          gateController.append(self.GateControllerOutput[i][j](hidden2))\n",
        "        gateControllerOutput.append(gateController)\n",
        "         \n",
        "      registerControllerOutput = []\n",
        "      # To get register controller values\n",
        "      for i in range(len(self.registerControllerDimensions)):\n",
        "        registerControllerOutput.append(self.RegisterController[i](hidden2))\n",
        "\n",
        "      # final completed status\n",
        "      completed = self.CompletedStatus(hidden2)\n",
        "\n",
        "      return gateControllerOutput, registerControllerOutput, completed"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "colab_type": "text",
        "id": "Hnd_YcNF1HyU"
      },
      "source": [
        "#### Example Implementation"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "RNaT8HpdoCZ4",
        "colab": {}
      },
      "source": [
        "# Creating instance of the contoller network\n",
        "nramController = NRAMController(hidden_dimensions=100, gateControllerDimensions=[[3], [4], [5], [6,6], [7, 7]], registerControllerDimensions=[8, 8, 8]);"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "X4F4zm_G_hUB",
        "outputId": "4b44a307-3e38-4c2a-a5be-45419cfd4120",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 428
        }
      },
      "source": [
        "# Running the controller\n",
        "gateControllerOutput, registerControllerOutput, completedStatus = nramController(tf.Variable(tf.ones((1,3))))\n",
        "\n",
        "print(\"\\nThe output from the neural network controller without training \\n\")\n",
        "gateControllerOutput"
      ],
      "execution_count": 63,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/tensorflow/python/ops/resource_variable_ops.py:642: colocate_with (from tensorflow.python.framework.ops) is deprecated and will be removed in a future version.\n",
            "Instructions for updating:\n",
            "Colocations handled automatically by placer.\n",
            "\n",
            "The output from the neural network controller without training \n",
            "\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "[[<tf.Tensor: id=1000, shape=(1, 3), dtype=float32, numpy=array([[0.3166075, 0.4237229, 0.2596697]], dtype=float32)>],\n",
              " [<tf.Tensor: id=1027, shape=(1, 4), dtype=float32, numpy=array([[0.20408267, 0.33518848, 0.22614823, 0.23458059]], dtype=float32)>],\n",
              " [<tf.Tensor: id=1054, shape=(1, 5), dtype=float32, numpy=\n",
              "  array([[0.21607451, 0.21540017, 0.20172255, 0.17757514, 0.18922766]],\n",
              "        dtype=float32)>],\n",
              " [<tf.Tensor: id=1081, shape=(1, 6), dtype=float32, numpy=\n",
              "  array([[0.158515  , 0.1752745 , 0.15319951, 0.14739434, 0.17360762,\n",
              "          0.19200905]], dtype=float32)>,\n",
              "  <tf.Tensor: id=1108, shape=(1, 6), dtype=float32, numpy=\n",
              "  array([[0.16984208, 0.19801897, 0.13187848, 0.1358704 , 0.16106479,\n",
              "          0.20332521]], dtype=float32)>],\n",
              " [<tf.Tensor: id=1135, shape=(1, 7), dtype=float32, numpy=\n",
              "  array([[0.12622438, 0.1355994 , 0.1274946 , 0.1530594 , 0.15377821,\n",
              "          0.16954228, 0.13430169]], dtype=float32)>,\n",
              "  <tf.Tensor: id=1162, shape=(1, 7), dtype=float32, numpy=\n",
              "  array([[0.16870417, 0.15780962, 0.12734526, 0.13887663, 0.1189144 ,\n",
              "          0.15182911, 0.13652082]], dtype=float32)>]]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 63
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "MZfmwP8Jx4Oq",
        "outputId": "b23ac918-281c-4f64-f9c2-78674cf99899",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 190
        }
      },
      "source": [
        "registerControllerOutput"
      ],
      "execution_count": 64,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "[<tf.Tensor: id=1189, shape=(1, 8), dtype=float32, numpy=\n",
              " array([[0.135936  , 0.14026052, 0.11946642, 0.12741882, 0.10818893,\n",
              "         0.10859721, 0.13993882, 0.12019324]], dtype=float32)>,\n",
              " <tf.Tensor: id=1216, shape=(1, 8), dtype=float32, numpy=\n",
              " array([[0.12422334, 0.10248131, 0.12454563, 0.1479609 , 0.12437946,\n",
              "         0.1467319 , 0.12937258, 0.10030489]], dtype=float32)>,\n",
              " <tf.Tensor: id=1243, shape=(1, 8), dtype=float32, numpy=\n",
              " array([[0.14644666, 0.12546715, 0.1081463 , 0.11810425, 0.14152135,\n",
              "         0.1256183 , 0.1302898 , 0.10440611]], dtype=float32)>]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 64
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "Hnfjd145x7dC",
        "outputId": "d3a97579-1693-4f7e-a83b-f98e3167f030",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 54
        }
      },
      "source": [
        "completedStatus"
      ],
      "execution_count": 65,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<tf.Tensor: id=1270, shape=(1, 1), dtype=float32, numpy=array([[0.4600103]], dtype=float32)>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 65
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "colab_type": "text",
        "id": "fEINnQwk_SbF"
      },
      "source": [
        "##### Input Controller Values is the output of NN being applied to the circuit for one time step"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "fzX2cCz7RFlP",
        "colab_type": "code",
        "outputId": "821e30d3-4dbb-403e-cccc-ea207bb62de9",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 377
        }
      },
      "source": [
        "r_ = []\n",
        "for _ in range(M):\n",
        "  r_.append(registerModule(_, M))\n",
        "memory = tf.concat(r_, axis=0)\n",
        "memory = np.array(memory, dtype=\"float32\")\n",
        "print(memory)"
      ],
      "execution_count": 66,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "[[1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1.]]\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "pMgQHz42llTj",
        "outputId": "659c7d2d-2e11-47e6-c105-7690ea43c594",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1805
        }
      },
      "source": [
        "gates = [incModule, decModule, addModule]\n",
        "\n",
        "# First set of input values\n",
        "registers = tf.concat([registerValueModule(0), registerValueModule(0), registerValueModule(0)], axis=0)\n",
        "registers = tf.reshape(registers, shape=[1, registers.shape[0], registers.shape[1]])\n",
        "\n",
        "print(\"Input register set 1: \", tf.argmax(registers, axis=2))\n",
        "\n",
        "print(\"\\nRunning input r0 -> increment -> output r0\")\n",
        "\n",
        "# Calling the function to run the complete circuit\n",
        "output_registers, memory = runTimeStepNN(gates, registers, memory, readMemory, writeMemory, gateControllerOutput, registerControllerOutput)\n",
        "\n",
        "print(\"\\nOutput Memory\")\n",
        "print(memory)\n",
        "\n",
        "print(\"\\noutput_registers:\\n\", output_registers)\n",
        "print(\"\\noutput_registers:\\n\", tf.argmax(output_registers, axis=2))"
      ],
      "execution_count": 67,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Input register set 1:  tf.Tensor([[0 0 0]], shape=(1, 3), dtype=int64)\n",
            "\n",
            "Running input r0 -> increment -> output r0\n",
            "\n",
            "Output Memory\n",
            "tf.Tensor(\n",
            "[[0.82847035 0.08428195 0.         0.         0.         0.\n",
            "  0.         0.         0.         0.         0.         0.\n",
            "  0.         0.         0.         0.         0.         0.\n",
            "  0.         0.08724774]\n",
            " [0.11656185 0.8645083  0.         0.         0.         0.\n",
            "  0.         0.         0.         0.         0.         0.\n",
            "  0.         0.         0.         0.         0.         0.\n",
            "  0.         0.01892992]\n",
            " [0.         0.         1.         0.         0.         0.\n",
            "  0.         0.         0.         0.         0.         0.\n",
            "  0.         0.         0.         0.         0.         0.\n",
            "  0.         0.        ]\n",
            " [0.         0.         0.         1.         0.         0.\n",
            "  0.         0.         0.         0.         0.         0.\n",
            "  0.         0.         0.         0.         0.         0.\n",
            "  0.         0.        ]\n",
            " [0.         0.         0.         0.         1.         0.\n",
            "  0.         0.         0.         0.         0.         0.\n",
            "  0.         0.         0.         0.         0.         0.\n",
            "  0.         0.        ]\n",
            " [0.         0.         0.         0.         0.         1.\n",
            "  0.         0.         0.         0.         0.         0.\n",
            "  0.         0.         0.         0.         0.         0.\n",
            "  0.         0.        ]\n",
            " [0.         0.         0.         0.         0.         0.\n",
            "  1.         0.         0.         0.         0.         0.\n",
            "  0.         0.         0.         0.         0.         0.\n",
            "  0.         0.        ]\n",
            " [0.         0.         0.         0.         0.         0.\n",
            "  0.         1.         0.         0.         0.         0.\n",
            "  0.         0.         0.         0.         0.         0.\n",
            "  0.         0.        ]\n",
            " [0.         0.         0.         0.         0.         0.\n",
            "  0.         0.         1.         0.         0.         0.\n",
            "  0.         0.         0.         0.         0.         0.\n",
            "  0.         0.        ]\n",
            " [0.         0.         0.         0.         0.         0.\n",
            "  0.         0.         0.         1.         0.         0.\n",
            "  0.         0.         0.         0.         0.         0.\n",
            "  0.         0.        ]\n",
            " [0.         0.         0.         0.         0.         0.\n",
            "  0.         0.         0.         0.         1.         0.\n",
            "  0.         0.         0.         0.         0.         0.\n",
            "  0.         0.        ]\n",
            " [0.         0.         0.         0.         0.         0.\n",
            "  0.         0.         0.         0.         0.         1.\n",
            "  0.         0.         0.         0.         0.         0.\n",
            "  0.         0.        ]\n",
            " [0.         0.         0.         0.         0.         0.\n",
            "  0.         0.         0.         0.         0.         0.\n",
            "  1.         0.         0.         0.         0.         0.\n",
            "  0.         0.        ]\n",
            " [0.         0.         0.         0.         0.         0.\n",
            "  0.         0.         0.         0.         0.         0.\n",
            "  0.         1.         0.         0.         0.         0.\n",
            "  0.         0.        ]\n",
            " [0.         0.         0.         0.         0.         0.\n",
            "  0.         0.         0.         0.         0.         0.\n",
            "  0.         0.         1.         0.         0.         0.\n",
            "  0.         0.        ]\n",
            " [0.         0.         0.         0.         0.         0.\n",
            "  0.         0.         0.         0.         0.         0.\n",
            "  0.         0.         0.         1.         0.         0.\n",
            "  0.         0.        ]\n",
            " [0.         0.         0.         0.         0.         0.\n",
            "  0.         0.         0.         0.         0.         0.\n",
            "  0.         0.         0.         0.         1.         0.\n",
            "  0.         0.        ]\n",
            " [0.         0.         0.         0.         0.         0.\n",
            "  0.         0.         0.         0.         0.         0.\n",
            "  0.         0.         0.         0.         0.         1.\n",
            "  0.         0.        ]\n",
            " [0.         0.         0.         0.         0.         0.\n",
            "  0.         0.         0.         0.         0.         0.\n",
            "  0.         0.         0.         0.         0.         0.\n",
            "  1.         0.        ]\n",
            " [0.10419302 0.016346   0.         0.         0.         0.\n",
            "  0.         0.         0.         0.         0.         0.\n",
            "  0.         0.         0.         0.         0.         0.\n",
            "  0.         0.87946093]], shape=(20, 20), dtype=float32)\n",
            "\n",
            "output_registers:\n",
            " tf.Tensor(\n",
            "[[[0.76875854 0.12667203 0.         0.         0.         0.\n",
            "   0.         0.         0.         0.         0.         0.\n",
            "   0.         0.         0.         0.         0.         0.\n",
            "   0.         0.10456941]\n",
            "  [0.7274418  0.13980417 0.         0.         0.         0.\n",
            "   0.         0.         0.         0.         0.         0.\n",
            "   0.         0.         0.         0.         0.         0.\n",
            "   0.         0.1327541 ]\n",
            "  [0.72622365 0.15757674 0.         0.         0.         0.\n",
            "   0.         0.         0.         0.         0.         0.\n",
            "   0.         0.         0.         0.         0.         0.\n",
            "   0.         0.11619954]]], shape=(1, 3, 20), dtype=float32)\n",
            "\n",
            "output_registers:\n",
            " tf.Tensor([[0 0 0]], shape=(1, 3), dtype=int64)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "10bMXmezUw_f",
        "colab_type": "code",
        "outputId": "73617d01-5be7-4810-eebc-aa0af9229b88",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1431
        }
      },
      "source": [
        "memory"
      ],
      "execution_count": 68,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<tf.Tensor: id=1616, shape=(20, 20), dtype=float32, numpy=\n",
              "array([[0.82847035, 0.08428195, 0.        , 0.        , 0.        ,\n",
              "        0.        , 0.        , 0.        , 0.        , 0.        ,\n",
              "        0.        , 0.        , 0.        , 0.        , 0.        ,\n",
              "        0.        , 0.        , 0.        , 0.        , 0.08724774],\n",
              "       [0.11656185, 0.8645083 , 0.        , 0.        , 0.        ,\n",
              "        0.        , 0.        , 0.        , 0.        , 0.        ,\n",
              "        0.        , 0.        , 0.        , 0.        , 0.        ,\n",
              "        0.        , 0.        , 0.        , 0.        , 0.01892992],\n",
              "       [0.        , 0.        , 1.        , 0.        , 0.        ,\n",
              "        0.        , 0.        , 0.        , 0.        , 0.        ,\n",
              "        0.        , 0.        , 0.        , 0.        , 0.        ,\n",
              "        0.        , 0.        , 0.        , 0.        , 0.        ],\n",
              "       [0.        , 0.        , 0.        , 1.        , 0.        ,\n",
              "        0.        , 0.        , 0.        , 0.        , 0.        ,\n",
              "        0.        , 0.        , 0.        , 0.        , 0.        ,\n",
              "        0.        , 0.        , 0.        , 0.        , 0.        ],\n",
              "       [0.        , 0.        , 0.        , 0.        , 1.        ,\n",
              "        0.        , 0.        , 0.        , 0.        , 0.        ,\n",
              "        0.        , 0.        , 0.        , 0.        , 0.        ,\n",
              "        0.        , 0.        , 0.        , 0.        , 0.        ],\n",
              "       [0.        , 0.        , 0.        , 0.        , 0.        ,\n",
              "        1.        , 0.        , 0.        , 0.        , 0.        ,\n",
              "        0.        , 0.        , 0.        , 0.        , 0.        ,\n",
              "        0.        , 0.        , 0.        , 0.        , 0.        ],\n",
              "       [0.        , 0.        , 0.        , 0.        , 0.        ,\n",
              "        0.        , 1.        , 0.        , 0.        , 0.        ,\n",
              "        0.        , 0.        , 0.        , 0.        , 0.        ,\n",
              "        0.        , 0.        , 0.        , 0.        , 0.        ],\n",
              "       [0.        , 0.        , 0.        , 0.        , 0.        ,\n",
              "        0.        , 0.        , 1.        , 0.        , 0.        ,\n",
              "        0.        , 0.        , 0.        , 0.        , 0.        ,\n",
              "        0.        , 0.        , 0.        , 0.        , 0.        ],\n",
              "       [0.        , 0.        , 0.        , 0.        , 0.        ,\n",
              "        0.        , 0.        , 0.        , 1.        , 0.        ,\n",
              "        0.        , 0.        , 0.        , 0.        , 0.        ,\n",
              "        0.        , 0.        , 0.        , 0.        , 0.        ],\n",
              "       [0.        , 0.        , 0.        , 0.        , 0.        ,\n",
              "        0.        , 0.        , 0.        , 0.        , 1.        ,\n",
              "        0.        , 0.        , 0.        , 0.        , 0.        ,\n",
              "        0.        , 0.        , 0.        , 0.        , 0.        ],\n",
              "       [0.        , 0.        , 0.        , 0.        , 0.        ,\n",
              "        0.        , 0.        , 0.        , 0.        , 0.        ,\n",
              "        1.        , 0.        , 0.        , 0.        , 0.        ,\n",
              "        0.        , 0.        , 0.        , 0.        , 0.        ],\n",
              "       [0.        , 0.        , 0.        , 0.        , 0.        ,\n",
              "        0.        , 0.        , 0.        , 0.        , 0.        ,\n",
              "        0.        , 1.        , 0.        , 0.        , 0.        ,\n",
              "        0.        , 0.        , 0.        , 0.        , 0.        ],\n",
              "       [0.        , 0.        , 0.        , 0.        , 0.        ,\n",
              "        0.        , 0.        , 0.        , 0.        , 0.        ,\n",
              "        0.        , 0.        , 1.        , 0.        , 0.        ,\n",
              "        0.        , 0.        , 0.        , 0.        , 0.        ],\n",
              "       [0.        , 0.        , 0.        , 0.        , 0.        ,\n",
              "        0.        , 0.        , 0.        , 0.        , 0.        ,\n",
              "        0.        , 0.        , 0.        , 1.        , 0.        ,\n",
              "        0.        , 0.        , 0.        , 0.        , 0.        ],\n",
              "       [0.        , 0.        , 0.        , 0.        , 0.        ,\n",
              "        0.        , 0.        , 0.        , 0.        , 0.        ,\n",
              "        0.        , 0.        , 0.        , 0.        , 1.        ,\n",
              "        0.        , 0.        , 0.        , 0.        , 0.        ],\n",
              "       [0.        , 0.        , 0.        , 0.        , 0.        ,\n",
              "        0.        , 0.        , 0.        , 0.        , 0.        ,\n",
              "        0.        , 0.        , 0.        , 0.        , 0.        ,\n",
              "        1.        , 0.        , 0.        , 0.        , 0.        ],\n",
              "       [0.        , 0.        , 0.        , 0.        , 0.        ,\n",
              "        0.        , 0.        , 0.        , 0.        , 0.        ,\n",
              "        0.        , 0.        , 0.        , 0.        , 0.        ,\n",
              "        0.        , 1.        , 0.        , 0.        , 0.        ],\n",
              "       [0.        , 0.        , 0.        , 0.        , 0.        ,\n",
              "        0.        , 0.        , 0.        , 0.        , 0.        ,\n",
              "        0.        , 0.        , 0.        , 0.        , 0.        ,\n",
              "        0.        , 0.        , 1.        , 0.        , 0.        ],\n",
              "       [0.        , 0.        , 0.        , 0.        , 0.        ,\n",
              "        0.        , 0.        , 0.        , 0.        , 0.        ,\n",
              "        0.        , 0.        , 0.        , 0.        , 0.        ,\n",
              "        0.        , 0.        , 0.        , 1.        , 0.        ],\n",
              "       [0.10419302, 0.016346  , 0.        , 0.        , 0.        ,\n",
              "        0.        , 0.        , 0.        , 0.        , 0.        ,\n",
              "        0.        , 0.        , 0.        , 0.        , 0.        ,\n",
              "        0.        , 0.        , 0.        , 0.        , 0.87946093]],\n",
              "      dtype=float32)>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 68
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "colab_type": "text",
        "id": "Q5Yp8kFGREb4"
      },
      "source": [
        "#### NRAM Controller with LSTM V1"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "99U-TGkmDqCF",
        "colab": {}
      },
      "source": [
        "class NRAMControllerLSTM(tf.keras.Model):\n",
        "  def __init__(self, hidden_dimensions=256, gateControllerDimensions=[[3], [4,4]], registerControllerDimensions=[4, 4, 4]):\n",
        "    super(NRAMControllerLSTM, self).__init__()\n",
        "    self.lstm = layers.LSTM(hidden_dimensions, return_sequences=True, return_state=True)\n",
        "\n",
        "    self.gateControllerDimensions = gateControllerDimensions\n",
        "    self.registerControllerDimensions = registerControllerDimensions\n",
        "\n",
        "    # Gate Inputs Controller values \n",
        "    self.GateControllerOutput = []\n",
        "    for gateControllerValue in gateControllerDimensions:\n",
        "      GateController = []\n",
        "      for dim in gateControllerValue:\n",
        "        GateController.append(tf.keras.layers.Dense(dim, input_shape=(hidden_dimensions,), use_bias=True, activation=\"softmax\"))\n",
        "      self.GateControllerOutput.append(GateController)\n",
        "\n",
        "    # Register Output Controller Values\n",
        "    self.RegisterController = []\n",
        "    for i in registerControllerDimensions:\n",
        "      self.RegisterController.append(tf.keras.layers.Dense(i, input_shape=(hidden_dimensions,), use_bias=True, activation=\"softmax\"))\n",
        "\n",
        "    self.CompletedStatus = tf.keras.layers.Dense(1, input_shape=(hidden_dimensions,), use_bias=True, activation=\"sigmoid\")\n",
        "\n",
        "  def call(self, input_registers):\n",
        "    inputValue = tf.reshape(input_registers, shape=[input_registers.shape[0], 1, input_registers.shape[1]])\n",
        "    lstm_output, state_h, state_c  = self.lstm(inputValue)\n",
        "\n",
        "    # To get gate controller values\n",
        "    gateControllerOutput = []\n",
        "    for i in range(len(self.gateControllerDimensions)):\n",
        "      gateController = []\n",
        "      for j in range(len(self.gateControllerDimensions[i])):\n",
        "        o = self.GateControllerOutput[i][j](lstm_output)\n",
        "        gateController.append(tf.reshape(o, shape=[o.shape[0],o.shape[2]]))\n",
        "      gateControllerOutput.append(gateController)\n",
        "\n",
        "    registerControllerOutput = []\n",
        "    # To get register controller values\n",
        "    for i in range(len(self.registerControllerDimensions)):\n",
        "      o = self.RegisterController[i](lstm_output)\n",
        "      registerControllerOutput.append(tf.reshape(o, shape=[o.shape[0],o.shape[2]]))\n",
        "\n",
        "    # final completed status\n",
        "    completed = self.CompletedStatus(lstm_output)\n",
        "\n",
        "    return gateControllerOutput, registerControllerOutput, completed"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "lSJKOt7tQ8dj",
        "outputId": "be42ef23-301a-419d-852a-eca45dd7b3e3",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 139
        }
      },
      "source": [
        "# Creating instance of the contoller network\n",
        "nramControllerLSTM = NRAMControllerLSTM();\n",
        "\n",
        "# Running the controller\n",
        "gateControllerOutput, registerControllerOutput, completedStatus = nramControllerLSTM(tf.Variable(tf.ones((1,3))))\n",
        "\n",
        "print(\"\\nThe output from the neural network controller without training \\n\")\n",
        "gateControllerOutput"
      ],
      "execution_count": 70,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "\n",
            "The output from the neural network controller without training \n",
            "\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "[[<tf.Tensor: id=2069, shape=(1, 3), dtype=float32, numpy=array([[0.34617248, 0.33220032, 0.3216272 ]], dtype=float32)>],\n",
              " [<tf.Tensor: id=2114, shape=(1, 4), dtype=float32, numpy=array([[0.24884419, 0.24564016, 0.25285354, 0.25266215]], dtype=float32)>,\n",
              "  <tf.Tensor: id=2159, shape=(1, 4), dtype=float32, numpy=array([[0.25097144, 0.25398603, 0.25092143, 0.24412115]], dtype=float32)>]]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 70
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "VnHsOkcDAwJO",
        "colab_type": "code",
        "outputId": "c99ca6e6-ac17-49ca-a67f-83ac35e7ba97",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 88
        }
      },
      "source": [
        "registerControllerOutput"
      ],
      "execution_count": 71,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "[<tf.Tensor: id=2204, shape=(1, 4), dtype=float32, numpy=array([[0.24674706, 0.24614383, 0.25540006, 0.25170904]], dtype=float32)>,\n",
              " <tf.Tensor: id=2249, shape=(1, 4), dtype=float32, numpy=array([[0.24570797, 0.24756736, 0.26077256, 0.24595214]], dtype=float32)>,\n",
              " <tf.Tensor: id=2294, shape=(1, 4), dtype=float32, numpy=array([[0.2549216 , 0.24656099, 0.24827492, 0.25024247]], dtype=float32)>]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 71
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "colab_type": "text",
        "id": "HFGqpuDkRIw9"
      },
      "source": [
        "#### NRAM Controller with LSTM V2"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "F7uo7cYrSdSQ",
        "colab": {}
      },
      "source": [
        "class NRAMControllerLSTM2(tf.keras.Model):\n",
        "   def __init__(self, hidden_dimensions=256, gateControllerDimensions=[[3], [4,4]], registerControllerDimensions=[4, 4, 4]):\n",
        "      super(NRAMControllerLSTM2, self).__init__()\n",
        "#       self.lstmHidden = layers.CuDNNLSTM(hidden_dimensions, return_sequences=True, return_state=True)\n",
        "      self.lstmHidden = layers.LSTM(hidden_dimensions, return_sequences=True, return_state=True)\n",
        "\n",
        "#       self.output_dimensions = output_dimensions\n",
        "\n",
        "#       self.Output = []\n",
        "#       self.SoftmaxOutput = []\n",
        "#       # Gate Inputs Controller values and Register Output Controller Values\n",
        "#       for dimension in output_dimensions:\n",
        "#         self.Output.append(layers.LSTM(dimension, return_sequences=True, return_state=True))\n",
        "#         self.SoftmaxOutput.append(layers.Activation(\"softmax\"))\n",
        "\n",
        "      self.gateControllerDimensions = gateControllerDimensions\n",
        "      self.registerControllerDimensions = registerControllerDimensions\n",
        "\n",
        "      # Gate Inputs Controller values \n",
        "      self.GateControllerOutput = []\n",
        "      self.SoftmaxGateOutput= []\n",
        "      for gateControllerValue in gateControllerDimensions:\n",
        "        GateController = []\n",
        "        SoftmaxLayer = []\n",
        "        for dim in gateControllerValue:\n",
        "          GateController.append(layers.LSTM(dim, return_sequences=True, return_state=True))\n",
        "          SoftmaxLayer.append(layers.Activation(\"softmax\"))\n",
        "        self.GateControllerOutput.append(GateController)\n",
        "        self.SoftmaxGateOutput.append(SoftmaxLayer)\n",
        "\n",
        "      # Register Output Controller Values\n",
        "      self.RegisterController = []\n",
        "      self.SoftmaxRegisterOutput = []\n",
        "      for i in registerControllerDimensions:\n",
        "        self.RegisterController.append(layers.LSTM(i, return_sequences=True, return_state=True))\n",
        "        self.SoftmaxRegisterOutput.append(layers.Activation(\"softmax\"))\n",
        "\n",
        "      self.CompletedStatus = tf.keras.layers.Dense(1, input_shape=(hidden_dimensions,), use_bias=True, activation=\"sigmoid\")\n",
        "   \n",
        "   def call(self, input_registers):\n",
        "      inputValue = tf.reshape(input_registers, shape=[input_registers.shape[0], 1, input_registers.shape[1]])\n",
        "      lstm_output, state_h, state_c  = self.lstmHidden(inputValue)\n",
        "\n",
        "      # To get gate controller values\n",
        "      gateControllerOutput = []\n",
        "      for i in range(len(self.gateControllerDimensions)):\n",
        "        gateController = []\n",
        "        for j in range(len(self.gateControllerDimensions[i])):\n",
        "          o, h, c = self.GateControllerOutput[i][j](lstm_output)\n",
        "          gateController.append(self.SoftmaxGateOutput[i][j](tf.reshape(o, shape=[o.shape[0],o.shape[2]])))\n",
        "        gateControllerOutput.append(gateController)\n",
        "\n",
        "      registerControllerOutput = []\n",
        "      # To get register controller values\n",
        "      for i in range(len(self.registerControllerDimensions)):\n",
        "        o, h, c = self.RegisterController[i](lstm_output)\n",
        "        registerControllerOutput.append(self.SoftmaxRegisterOutput[i](tf.reshape(o, shape=[o.shape[0],o.shape[2]])))\n",
        "\n",
        "#       # To get controller values\n",
        "#       output = []\n",
        "#       for i in range(len(self.output_dimensions)):\n",
        "#         o, h, c = self.Output[i](lstm_output)\n",
        "#         lstmOp = tf.reshape(o, shape=[o.shape[0],o.shape[2]])\n",
        "#         output.append(self.SoftmaxOutput[i](lstmOp))\n",
        "\n",
        "      # final completed status\n",
        "      completed = self.CompletedStatus(lstm_output)\n",
        "\n",
        "      return gateControllerOutput, registerControllerOutput, completed"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "kMGbwAJxSgis",
        "outputId": "bc560f39-8a50-45ff-f838-e3fe66085b7f",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 139
        }
      },
      "source": [
        "# Example with one gate and 3 input/output registers\n",
        "outputVals = [3, 4, 4, 4]\n",
        "\n",
        "# Creating instance of the contoller network\n",
        "nramControllerLSTM = NRAMControllerLSTM2();\n",
        "\n",
        "# Running the controller\n",
        "gateControllerOutput, registerControllerOutput, completedStatus = nramControllerLSTM(tf.Variable(tf.ones((1,3))))\n",
        "\n",
        "print(\"\\nThe output from the neural network controller without training \\n\")\n",
        "gateControllerOutput"
      ],
      "execution_count": 73,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "\n",
            "The output from the neural network controller without training \n",
            "\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "[[<tf.Tensor: id=3008, shape=(1, 3), dtype=float32, numpy=array([[0.33342683, 0.33112642, 0.33544672]], dtype=float32)>],\n",
              " [<tf.Tensor: id=3339, shape=(1, 4), dtype=float32, numpy=array([[0.24929503, 0.2512825 , 0.24875504, 0.25066748]], dtype=float32)>,\n",
              "  <tf.Tensor: id=3670, shape=(1, 4), dtype=float32, numpy=array([[0.24815004, 0.2497499 , 0.25161406, 0.25048593]], dtype=float32)>]]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 73
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "pJT2qRHHHFlC",
        "colab_type": "code",
        "outputId": "e0ad4968-7335-45a7-a0b7-610cb2a3f418",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 88
        }
      },
      "source": [
        "registerControllerOutput"
      ],
      "execution_count": 74,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "[<tf.Tensor: id=4001, shape=(1, 4), dtype=float32, numpy=array([[0.2531984 , 0.24763887, 0.24969175, 0.24947096]], dtype=float32)>,\n",
              " <tf.Tensor: id=4332, shape=(1, 4), dtype=float32, numpy=array([[0.2495757 , 0.24955481, 0.25083593, 0.25003344]], dtype=float32)>,\n",
              " <tf.Tensor: id=4663, shape=(1, 4), dtype=float32, numpy=array([[0.2501385 , 0.24775039, 0.25151163, 0.25059947]], dtype=float32)>]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 74
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "colab_type": "text",
        "id": "vhZ8G5qeRZSW"
      },
      "source": [
        "## Dataset Implementation"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ORziqbqo7Mg_",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# To select dataset\n",
        "dataSelected = 5"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "hLhcTRDPsmzq",
        "colab_type": "text"
      },
      "source": [
        "##### Bucketsort"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "JBDJWrXBf-vg",
        "colab_type": "code",
        "outputId": "9ff67d4d-8c0c-433f-86dd-118f95a8dc4c",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 105
        }
      },
      "source": [
        "# Python3 program to sort an array \n",
        "# using bucket sort \n",
        "def insertionSort(b): \n",
        "\tfor i in range(1, len(b)): \n",
        "\t\tup = b[i] \n",
        "\t\tj = i - 1\n",
        "\t\twhile j >=0 and b[j] > up: \n",
        "\t\t\tb[j + 1] = b[j] \n",
        "\t\t\tj -= 1\n",
        "\t\tb[j + 1] = up\t \n",
        "\treturn b\t \n",
        "\t\t\t\n",
        "def bucketSort(x, M): \n",
        "\tarr = [] \n",
        "\tslot_num = 3 # 10 means 10 slots, each \n",
        "\t\t\t\t# slot's size is 0.1 \n",
        "\tfor i in range(slot_num): \n",
        "\t\tarr.append([]) \n",
        "\t\t\n",
        "\t# Put array elements in different buckets \n",
        "\tfor j in x: \n",
        "\t\tindex_b = int((slot_num * j)/M) \n",
        "\t\tarr[index_b].append(j) \n",
        "\t\n",
        "\t# Sort individual buckets \n",
        "\tfor i in range(slot_num): \n",
        "\t\tarr[i] = insertionSort(arr[i]) \n",
        "\t\n",
        "\t# concatenate the result \n",
        "\tk = 0\n",
        "\tfor i in range(slot_num): \n",
        "\t\tfor j in range(len(arr[i])): \n",
        "\t\t\tx[k] = arr[i][j] \n",
        "\t\t\tk += 1\n",
        "\treturn x, arr\n",
        "\n",
        "# Driver Code \n",
        "x = [1, 4, 3, 6, \n",
        "\t2, 5, 7, 9, 8] \n",
        "print(\"Sorted Array is\") \n",
        "print(bucketSort(x, 10))\n",
        "\n",
        "\n",
        "# Driver Code \n",
        "x = [1, 14, 13, 6, \n",
        "\t12, 15, 7, 19, 8] \n",
        "print(\"Sorted Array is\") \n",
        "print(bucketSort(x, 20))"
      ],
      "execution_count": 76,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Sorted Array is\n",
            "([1, 2, 3, 4, 5, 6, 7, 8, 9], [[1, 2, 3], [4, 5, 6], [7, 8, 9]])\n",
            "Sorted Array is\n",
            "([1, 6, 7, 8, 12, 13, 14, 15, 19], [[1, 6], [7, 8, 12, 13], [14, 15, 19]])\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "nh3ousI_stj8",
        "colab_type": "text"
      },
      "source": [
        "##### Test data Generator"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "OuamNiOVsr9g",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# # Initalize the memory\n",
        "# inputMemory = np.zeros((M,M), dtype=\"float32\")\n",
        "# targetMemory = np.zeros((M,M), dtype=\"float32\")\n",
        "\n",
        "# # Pointer to the buckets\n",
        "# p1 = 8\n",
        "# p2 = 10\n",
        "# p3 = 12\n",
        "\n",
        "# # Bucket 1\n",
        "# inputMemory[0] = registerModule(p1, M)\n",
        "# targetMemory[0] = registerModule(p1, M)\n",
        "\n",
        "# # Bucket 2\n",
        "# inputMemory[1] = registerModule(p2, M)\n",
        "# targetMemory[1] = registerModule(p2, M)\n",
        "\n",
        "# # Bucket 3\n",
        "# inputMemory[2] = registerModule(p3, M)\n",
        "# targetMemory[2] = registerModule(p3, M)\n",
        "\n",
        "# # We can give sorted values into the function bucketSort\n",
        "# _, bucketsArray = bucketSort(np.arange(1, 20), 20)\n",
        "# # print(bucketsArray)\n",
        "\n",
        "\n",
        "# print(\"\\nindex:\")\n",
        "# randIndex = np.arange(3)\n",
        "# np.random.shuffle(randIndex)\n",
        "# print(randIndex)\n",
        "\n",
        "# for i in range(3):\n",
        "#   # memory index starts from 4 to 6 -> 4,5,6\n",
        "#   bucketIndex = randIndex[i]\n",
        "#   val = np.random.choice(bucketsArray[bucketIndex], 1)[0]\n",
        "#   print(val)\n",
        "#   inputMemory[i+4] = registerModule(val, M)\n",
        "#   targetMemory[i+4] = registerModule(val, M)\n",
        "  \n",
        "#   if bucketIndex == 0:\n",
        "#     #bucket1\n",
        "#     targetMemory[p1] = registerModule(val, M)\n",
        "#   elif bucketIndex == 1:\n",
        "#     #bucket2\n",
        "#     targetMemory[p2] = registerModule(val, M)    \n",
        "#   elif bucketIndex == 2:\n",
        "#     #bucket3\n",
        "#     targetMemory[p3] = registerModule(val, M)\n",
        "\n",
        "# print(np.argmax(inputMemory, axis=1))\n",
        "# print(np.argmax(targetMemory, axis=1))"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "n8L5FeUwOh9N",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# ## [20x20] -> 3 Pointers -> 4 Input Elements -> 2 element in each bucket -> 3,  1,4,  1,2, 1,2, 1,2\n",
        "# # Initalize the memory\n",
        "# inputMemory = np.zeros((M,M), dtype=\"float32\")\n",
        "# targetMemory = np.zeros((M,M), dtype=\"float32\")\n",
        "\n",
        "# # Pointer to the buckets\n",
        "# p1 = 9\n",
        "# p2 = 12\n",
        "# p3 = 15\n",
        "\n",
        "# # Bucket 1\n",
        "# inputMemory[0] = registerModule(p1, M)\n",
        "# targetMemory[0] = registerModule(p1, M)\n",
        "\n",
        "# # Bucket 2\n",
        "# inputMemory[1] = registerModule(p2, M)\n",
        "# targetMemory[1] = registerModule(p2, M)\n",
        "\n",
        "# # Bucket 3\n",
        "# inputMemory[2] = registerModule(p3, M)\n",
        "# targetMemory[2] = registerModule(p3, M)\n",
        "\n",
        "# # We can give sorted values into the function bucketSort\n",
        "# _, bucketsArray = bucketSort(np.arange(1, 20), 20)\n",
        "# # print(bucketsArray)\n",
        "\n",
        "\n",
        "# print(\"\\nindex:\")\n",
        "# randIndex = np.arange(3)\n",
        "# np.random.shuffle(randIndex)\n",
        "# print(randIndex)\n",
        "\n",
        "# for i in range(3):\n",
        "#   # memory index starts from 4 to 6 -> 4,5,6,7\n",
        "#   bucketIndex = randIndex[i]\n",
        "#   val = np.random.choice(bucketsArray[bucketIndex], 1)[0]\n",
        "#   print(val)\n",
        "#   inputMemory[i+4] = registerModule(val, M)\n",
        "#   targetMemory[i+4] = registerModule(val, M)\n",
        "  \n",
        "#   if bucketIndex == 0:\n",
        "#     #bucket1\n",
        "#     targetMemory[p1] = registerModule(val, M)\n",
        "#   elif bucketIndex == 1:\n",
        "#     #bucket2\n",
        "#     targetMemory[p2] = registerModule(val, M)    \n",
        "#   elif bucketIndex == 2:\n",
        "#     #bucket3\n",
        "#     targetMemory[p3] = registerModule(val, M)\n",
        "\n",
        "# # 4th element\n",
        "# print(\"\\nindex:\")\n",
        "# randIndex = np.arange(3)\n",
        "# np.random.shuffle(randIndex)\n",
        "# print(randIndex)\n",
        "\n",
        "# for i in range(1):\n",
        "#   # memory index starts from 4 to 6 -> 4,5,6,7\n",
        "#   bucketIndex = randIndex[i]\n",
        "#   val = np.random.choice(bucketsArray[bucketIndex], 1)[0]\n",
        "#   print(val)\n",
        "#   inputMemory[i+7] = registerModule(val, M)\n",
        "#   targetMemory[i+7] = registerModule(val, M)\n",
        "  \n",
        "#   if bucketIndex == 0:\n",
        "#     #bucket1\n",
        "#     targetMemory[p1+1] = registerModule(val, M)\n",
        "#   elif bucketIndex == 1:\n",
        "#     #bucket2\n",
        "#     targetMemory[p2+1] = registerModule(val, M)    \n",
        "#   elif bucketIndex == 2:\n",
        "#     #bucket3\n",
        "#     targetMemory[p3+1] = registerModule(val, M)\n",
        "\n",
        "# print(np.argmax(inputMemory, axis=1))\n",
        "# print(np.argmax(targetMemory, axis=1))"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "wyzaNXZKwpqR",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# ## [20x20] -> 3 Pointers -> 6 Input Elements -> 2 element in each bucket -> 3,  1,6,  1,2, 1,2, 1,2\n",
        "# # Initalize the memory\n",
        "# inputMemory = np.zeros((M,M), dtype=\"float32\")\n",
        "# targetMemory = np.zeros((M,M), dtype=\"float32\")\n",
        "\n",
        "# # Pointer to the buckets\n",
        "# p1 = 11\n",
        "# p2 = 14\n",
        "# p3 = 17\n",
        "\n",
        "# # Bucket 1\n",
        "# inputMemory[0] = registerModule(p1, M)\n",
        "# targetMemory[0] = registerModule(p1, M)\n",
        "\n",
        "# # Bucket 2\n",
        "# inputMemory[1] = registerModule(p2, M)\n",
        "# targetMemory[1] = registerModule(p2, M)\n",
        "\n",
        "# # Bucket 3\n",
        "# inputMemory[2] = registerModule(p3, M)\n",
        "# targetMemory[2] = registerModule(p3, M)\n",
        "\n",
        "# # We can give sorted values into the function bucketSort\n",
        "# _, bucketsArray = bucketSort(np.arange(1, 20), 20)\n",
        "# # print(bucketsArray)\n",
        "\n",
        "\n",
        "# print(\"\\nindex:\")\n",
        "# randIndex = np.arange(3)\n",
        "# np.random.shuffle(randIndex)\n",
        "# print(randIndex)\n",
        "\n",
        "# for i in range(3):\n",
        "#   # memory index starts from 4 to 6 -> 4,5,6,7\n",
        "#   bucketIndex = randIndex[i]\n",
        "#   val = np.random.choice(bucketsArray[bucketIndex], 1)[0]\n",
        "#   print(val)\n",
        "#   inputMemory[i+4] = registerModule(val, M)\n",
        "#   targetMemory[i+4] = registerModule(val, M)\n",
        "  \n",
        "#   if bucketIndex == 0:\n",
        "#     #bucket1\n",
        "#     targetMemory[p1] = registerModule(val, M)\n",
        "#   elif bucketIndex == 1:\n",
        "#     #bucket2\n",
        "#     targetMemory[p2] = registerModule(val, M)    \n",
        "#   elif bucketIndex == 2:\n",
        "#     #bucket3\n",
        "#     targetMemory[p3] = registerModule(val, M)\n",
        "\n",
        "# # 4th element\n",
        "# print(\"\\nindex:\")\n",
        "# randIndex = np.arange(3)\n",
        "# np.random.shuffle(randIndex)\n",
        "# print(randIndex)\n",
        "\n",
        "# for i in range(3):\n",
        "#   # memory index starts from 4 to 6 -> 4,5,6,7\n",
        "#   bucketIndex = randIndex[i]\n",
        "#   val = np.random.choice(bucketsArray[bucketIndex], 1)[0]\n",
        "#   print(val)\n",
        "#   inputMemory[i+7] = registerModule(val, M)\n",
        "#   targetMemory[i+7] = registerModule(val, M)\n",
        "  \n",
        "#   if bucketIndex == 0:\n",
        "#     #bucket1\n",
        "#     targetMemory[p1+1] = registerModule(val, M)\n",
        "#   elif bucketIndex == 1:\n",
        "#     #bucket2\n",
        "#     targetMemory[p2+1] = registerModule(val, M)    \n",
        "#   elif bucketIndex == 2:\n",
        "#     #bucket3\n",
        "#     targetMemory[p3+1] = registerModule(val, M)\n",
        "\n",
        "# print(np.argmax(inputMemory, axis=1))\n",
        "# print(np.argmax(targetMemory, axis=1))"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "8qRl86D6O1zu",
        "colab_type": "text"
      },
      "source": [
        "### Dataset 4"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "HZQqDkzgs0R2",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "def repeat(size, inputSet, targetSet):\n",
        "  inputRepeatSet = []\n",
        "  targetRepeatSet = []\n",
        "  \n",
        "  for _ in range(size):\n",
        "    arrIndex = np.arange(len(inputSet))\n",
        "    np.random.shuffle(arrIndex)\n",
        "    \n",
        "    for j in arrIndex:\n",
        "      inputRepeatSet.append(inputSet[j])\n",
        "      targetRepeatSet.append(targetSet[j])\n",
        "\n",
        "  return inputRepeatSet, targetRepeatSet\n",
        "\n",
        "def dataset4(repeats, M):\n",
        "  inputSet = []\n",
        "  targetSet = []\n",
        "  \n",
        "  for _ in range(repeats):\n",
        "    # Initalize the memory\n",
        "    inputMemory = np.zeros((M,M), dtype=\"float32\")\n",
        "    targetMemory = np.zeros((M,M), dtype=\"float32\")\n",
        "\n",
        "    # Pointer to the buckets\n",
        "    p1 = 8\n",
        "    p2 = 10\n",
        "    p3 = 12\n",
        "\n",
        "    # Bucket 1\n",
        "    inputMemory[0] = registerModule(p1, M)\n",
        "    targetMemory[0] = registerModule(p1, M)\n",
        "\n",
        "    # Bucket 2\n",
        "    inputMemory[1] = registerModule(p2, M)\n",
        "    targetMemory[1] = registerModule(p2, M)\n",
        "\n",
        "    # Bucket 3\n",
        "    inputMemory[2] = registerModule(p3, M)\n",
        "    targetMemory[2] = registerModule(p3, M)\n",
        "\n",
        "    # We can give sorted values into the function bucketSort\n",
        "    _, bucketsArray = bucketSort(np.arange(1, 20), 20)\n",
        "    # print(bucketsArray)\n",
        "\n",
        "\n",
        "    print(\"\\nindex:\")\n",
        "    randIndex = np.arange(3)\n",
        "    np.random.shuffle(randIndex)\n",
        "    print(randIndex)\n",
        "\n",
        "    for i in range(3):\n",
        "      # memory index starts from 4 to 6 -> 4,5,6\n",
        "      bucketIndex = randIndex[i]\n",
        "      val = np.random.choice(bucketsArray[bucketIndex], 1)[0]\n",
        "      inputMemory[i+4] = registerModule(val, M)\n",
        "      targetMemory[i+4] = registerModule(val, M)\n",
        "\n",
        "      if bucketIndex == 0:\n",
        "        #bucket1\n",
        "        targetMemory[p1] = registerModule(val, M)\n",
        "      elif bucketIndex == 1:\n",
        "        #bucket2\n",
        "        targetMemory[p2] = registerModule(val, M)    \n",
        "      elif bucketIndex == 2:\n",
        "        #bucket3\n",
        "        targetMemory[p3] = registerModule(val, M)\n",
        "\n",
        "    print(np.argmax(inputMemory, axis=1))\n",
        "    print(np.argmax(targetMemory, axis=1))\n",
        "\n",
        "    inputSet.append(inputMemory)\n",
        "    targetSet.append(targetMemory)\n",
        "  return inputSet, targetSet"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "iRKxGfBvx-FP",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 4287
        },
        "outputId": "bad86f0e-d72e-418b-ec01-0deebe35b383"
      },
      "source": [
        "inputSet, targetSet = dataset4(50, M)"
      ],
      "execution_count": 81,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "[ 8 10 12  0 14  6 11  0  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 8 10 12  0 14  6 11  0  6  0 11  0 14  0  0  0  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "[ 8 10 12  0  3  7 15  0  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 8 10 12  0  3  7 15  0  3  0  7  0 15  0  0  0  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "[ 8 10 12  0 10  6 18  0  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 8 10 12  0 10  6 18  0  6  0 10  0 18  0  0  0  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "[ 8 10 12  0 14  8  6  0  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 8 10 12  0 14  8  6  0  6  0  8  0 14  0  0  0  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "[ 8 10 12  0 12 18  4  0  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 8 10 12  0 12 18  4  0  4  0 12  0 18  0  0  0  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "[ 8 10 12  0  3 16  9  0  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 8 10 12  0  3 16  9  0  3  0  9  0 16  0  0  0  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "[ 8 10 12  0 18  6 13  0  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 8 10 12  0 18  6 13  0  6  0 13  0 18  0  0  0  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "[ 8 10 12  0  6  9 17  0  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 8 10 12  0  6  9 17  0  6  0  9  0 17  0  0  0  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "[ 8 10 12  0 10 17  3  0  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 8 10 12  0 10 17  3  0  3  0 10  0 17  0  0  0  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "[ 8 10 12  0  2 19 10  0  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 8 10 12  0  2 19 10  0  2  0 10  0 19  0  0  0  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "[ 8 10 12  0  5 15  8  0  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 8 10 12  0  5 15  8  0  5  0  8  0 15  0  0  0  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "[ 8 10 12  0  2 18 13  0  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 8 10 12  0  2 18 13  0  2  0 13  0 18  0  0  0  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "[ 8 10 12  0 19  2 11  0  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 8 10 12  0 19  2 11  0  2  0 11  0 19  0  0  0  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "[ 8 10 12  0  7  6 15  0  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 8 10 12  0  7  6 15  0  6  0  7  0 15  0  0  0  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "[ 8 10 12  0  7 17  1  0  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 8 10 12  0  7 17  1  0  1  0  7  0 17  0  0  0  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "[ 8 10 12  0  2 19 10  0  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 8 10 12  0  2 19 10  0  2  0 10  0 19  0  0  0  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "[ 8 10 12  0  4 18 11  0  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 8 10 12  0  4 18 11  0  4  0 11  0 18  0  0  0  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "[ 8 10 12  0  9  4 16  0  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 8 10 12  0  9  4 16  0  4  0  9  0 16  0  0  0  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "[ 8 10 12  0  9 17  4  0  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 8 10 12  0  9 17  4  0  4  0  9  0 17  0  0  0  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "[ 8 10 12  0 17  9  2  0  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 8 10 12  0 17  9  2  0  2  0  9  0 17  0  0  0  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "[ 8 10 12  0  2 14  7  0  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 8 10 12  0  2 14  7  0  2  0  7  0 14  0  0  0  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "[ 8 10 12  0 18 10  1  0  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 8 10 12  0 18 10  1  0  1  0 10  0 18  0  0  0  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "[ 8 10 12  0 17  6 12  0  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 8 10 12  0 17  6 12  0  6  0 12  0 17  0  0  0  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "[ 8 10 12  0 16 12  6  0  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 8 10 12  0 16 12  6  0  6  0 12  0 16  0  0  0  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "[ 8 10 12  0  1 19  9  0  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 8 10 12  0  1 19  9  0  1  0  9  0 19  0  0  0  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "[ 8 10 12  0 19  2  7  0  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 8 10 12  0 19  2  7  0  2  0  7  0 19  0  0  0  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "[ 8 10 12  0  4 12 19  0  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 8 10 12  0  4 12 19  0  4  0 12  0 19  0  0  0  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "[ 8 10 12  0  8 16  3  0  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 8 10 12  0  8 16  3  0  3  0  8  0 16  0  0  0  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "[ 8 10 12  0  7  6 15  0  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 8 10 12  0  7  6 15  0  6  0  7  0 15  0  0  0  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "[ 8 10 12  0  8  4 15  0  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 8 10 12  0  8  4 15  0  4  0  8  0 15  0  0  0  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "[ 8 10 12  0 14  3 10  0  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 8 10 12  0 14  3 10  0  3  0 10  0 14  0  0  0  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "[ 8 10 12  0 12 14  6  0  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 8 10 12  0 12 14  6  0  6  0 12  0 14  0  0  0  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "[ 8 10 12  0  8 14  5  0  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 8 10 12  0  8 14  5  0  5  0  8  0 14  0  0  0  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "[ 8 10 12  0 14  5  9  0  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 8 10 12  0 14  5  9  0  5  0  9  0 14  0  0  0  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "[ 8 10 12  0  3 17  7  0  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 8 10 12  0  3 17  7  0  3  0  7  0 17  0  0  0  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "[ 8 10 12  0  3 17 11  0  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 8 10 12  0  3 17 11  0  3  0 11  0 17  0  0  0  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "[ 8 10 12  0 13  1 18  0  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 8 10 12  0 13  1 18  0  1  0 13  0 18  0  0  0  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "[ 8 10 12  0 16 10  3  0  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 8 10 12  0 16 10  3  0  3  0 10  0 16  0  0  0  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "[ 8 10 12  0 13  6 14  0  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 8 10 12  0 13  6 14  0  6  0 13  0 14  0  0  0  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "[ 8 10 12  0 16 10  5  0  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 8 10 12  0 16 10  5  0  5  0 10  0 16  0  0  0  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "[ 8 10 12  0 16  5  9  0  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 8 10 12  0 16  5  9  0  5  0  9  0 16  0  0  0  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "[ 8 10 12  0  8 14  3  0  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 8 10 12  0  8 14  3  0  3  0  8  0 14  0  0  0  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "[ 8 10 12  0 13  1 15  0  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 8 10 12  0 13  1 15  0  1  0 13  0 15  0  0  0  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "[ 8 10 12  0 11 17  1  0  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 8 10 12  0 11 17  1  0  1  0 11  0 17  0  0  0  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "[ 8 10 12  0 15  2  9  0  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 8 10 12  0 15  2  9  0  2  0  9  0 15  0  0  0  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "[ 8 10 12  0  3 17 13  0  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 8 10 12  0  3 17 13  0  3  0 13  0 17  0  0  0  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "[ 8 10 12  0  2 11 15  0  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 8 10 12  0  2 11 15  0  2  0 11  0 15  0  0  0  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "[ 8 10 12  0  4 19 10  0  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 8 10 12  0  4 19 10  0  4  0 10  0 19  0  0  0  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "[ 8 10 12  0  4 15 12  0  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 8 10 12  0  4 15 12  0  4  0 12  0 15  0  0  0  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "[ 8 10 12  0  9  3 19  0  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 8 10 12  0  9  3 19  0  3  0  9  0 19  0  0  0  0  0  0  0]\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ioGw30_ky17i",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# dataSelected = 4\n",
        "if dataSelected == 4:\n",
        "  ## Generate dataset\n",
        "  inputSet, targetSet = dataset4(50, M)\n",
        "\n",
        "  inputTrainSet, targetTrainSet = repeat(40, inputSet[0:40], targetSet[0:40])\n",
        "\n",
        "  print(len(inputTrainSet))\n",
        "  print(inputSet[0].shape)\n",
        "  print(targetSet[0].shape)\n",
        "\n",
        "  inputTestSet, targetTestSet = inputSet[40:50], targetSet[40:50]\n",
        "\n",
        "  print(len(inputTestSet))\n",
        "  # for (input_data, target_data) in zip(inputSet, targetSet):\n",
        "  #   print(input_data)\n",
        "  #   print(target_data)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "THPWe7-J1jNf",
        "colab_type": "text"
      },
      "source": [
        "### Dataset 5"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "HOMOe-Zx01UT",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "def repeat(size, inputSet, targetSet):\n",
        "  inputRepeatSet = []\n",
        "  targetRepeatSet = []\n",
        "  \n",
        "  for _ in range(size):\n",
        "    arrIndex = np.arange(len(inputSet))\n",
        "    np.random.shuffle(arrIndex)\n",
        "    \n",
        "    for j in arrIndex:\n",
        "      inputRepeatSet.append(inputSet[j])\n",
        "      targetRepeatSet.append(targetSet[j])\n",
        "\n",
        "  return inputRepeatSet, targetRepeatSet\n",
        "\n",
        "def dataset5(repeats, M):\n",
        "  inputSet = []\n",
        "  targetSet = []\n",
        "  \n",
        "  for _ in range(repeats):\n",
        "    ## [20x20] -> 3 Pointers -> 4 Input Elements -> 2 element in each bucket -> 3,  1,4,  1,2, 1,2, 1,2\n",
        "    # Initalize the memory\n",
        "    inputMemory = np.zeros((M,M), dtype=\"float32\")\n",
        "    targetMemory = np.zeros((M,M), dtype=\"float32\")\n",
        "\n",
        "    # Pointer to the buckets\n",
        "    p1 = 9\n",
        "    p2 = 12\n",
        "    p3 = 15\n",
        "\n",
        "    # Bucket 1\n",
        "    inputMemory[0] = registerModule(p1, M)\n",
        "    targetMemory[0] = registerModule(p1, M)\n",
        "\n",
        "    # Bucket 2\n",
        "    inputMemory[1] = registerModule(p2, M)\n",
        "    targetMemory[1] = registerModule(p2, M)\n",
        "\n",
        "    # Bucket 3\n",
        "    inputMemory[2] = registerModule(p3, M)\n",
        "    targetMemory[2] = registerModule(p3, M)\n",
        "\n",
        "    # We can give sorted values into the function bucketSort\n",
        "    _, bucketsArray = bucketSort(np.arange(1, 20), 20)\n",
        "    # print(bucketsArray)\n",
        "\n",
        "\n",
        "    print(\"\\nindex:\")\n",
        "    randIndex = np.arange(3)\n",
        "    np.random.shuffle(randIndex)\n",
        "    print(randIndex)\n",
        "\n",
        "    for i in range(3):\n",
        "      # memory index starts from 4 to 6 -> 4,5,6,7\n",
        "      bucketIndex = randIndex[i]\n",
        "      val = np.random.choice(bucketsArray[bucketIndex], 1)[0]\n",
        "      print(val)\n",
        "      inputMemory[i+4] = registerModule(val, M)\n",
        "      targetMemory[i+4] = registerModule(val, M)\n",
        "\n",
        "      if bucketIndex == 0:\n",
        "        #bucket1\n",
        "        targetMemory[p1] = registerModule(val, M)\n",
        "      elif bucketIndex == 1:\n",
        "        #bucket2\n",
        "        targetMemory[p2] = registerModule(val, M)    \n",
        "      elif bucketIndex == 2:\n",
        "        #bucket3\n",
        "        targetMemory[p3] = registerModule(val, M)\n",
        "\n",
        "    # 4th element\n",
        "    print(\"\\nindex:\")\n",
        "    randIndex = np.arange(3)\n",
        "    np.random.shuffle(randIndex)\n",
        "    print(randIndex)\n",
        "\n",
        "    for i in range(1):\n",
        "      # memory index starts from 4 to 6 -> 4,5,6,7\n",
        "      bucketIndex = randIndex[i]\n",
        "      val = np.random.choice(bucketsArray[bucketIndex], 1)[0]\n",
        "      print(val)\n",
        "      inputMemory[i+7] = registerModule(val, M)\n",
        "      targetMemory[i+7] = registerModule(val, M)\n",
        "\n",
        "      if bucketIndex == 0:\n",
        "        #bucket1\n",
        "        targetMemory[p1+1] = registerModule(val, M)\n",
        "      elif bucketIndex == 1:\n",
        "        #bucket2\n",
        "        targetMemory[p2+1] = registerModule(val, M)    \n",
        "      elif bucketIndex == 2:\n",
        "        #bucket3\n",
        "        targetMemory[p3+1] = registerModule(val, M)\n",
        "\n",
        "    print(np.argmax(inputMemory, axis=1))\n",
        "    print(np.argmax(targetMemory, axis=1))\n",
        "\n",
        "    inputSet.append(inputMemory)\n",
        "    targetSet.append(targetMemory)\n",
        "  return inputSet, targetSet"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "CO15-Dol6z4N",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 102088
        },
        "outputId": "bbfeaa04-523a-4bb1-89ae-b3b47547ff97"
      },
      "source": [
        "# dataSelected = 5\n",
        "if dataSelected == 5:\n",
        "  ## Generate dataset\n",
        "  inputSet, targetSet = dataset5(500, M)\n",
        "  inputTrainSet, targetTrainSet = repeat(20, inputSet[0:40], targetSet[0:40])\n",
        "\n",
        "  print(len(inputTrainSet))\n",
        "  print(inputSet[0].shape)\n",
        "  print(targetSet[0].shape)\n",
        "\n",
        "  inputTestSet, targetTestSet = inputSet[40:500], targetSet[40:500]\n",
        "\n",
        "  # for (input_data, target_data) in zip(inputSet, targetSet):\n",
        "  #   print(input_data)\n",
        "  #   print(target_data)"
      ],
      "execution_count": 84,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "4\n",
            "18\n",
            "10\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "16\n",
            "[ 9 12 15  0  4 18 10 16  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  4 18 10 16  0  4  0  0 10  0  0 18 16  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "14\n",
            "8\n",
            "6\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "15\n",
            "[ 9 12 15  0 14  8  6 15  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 14  8  6 15  0  6  0  0  8  0  0 14 15  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "6\n",
            "18\n",
            "12\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "16\n",
            "[ 9 12 15  0  6 18 12 16  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  6 18 12 16  0  6  0  0 12  0  0 18 16  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "7\n",
            "6\n",
            "17\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "14\n",
            "[ 9 12 15  0  7  6 17 14  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  7  6 17 14  0  6  0  0  7  0  0 17 14  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "9\n",
            "1\n",
            "14\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "15\n",
            "[ 9 12 15  0  9  1 14 15  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  9  1 14 15  0  1  0  0  9  0  0 14 15  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "14\n",
            "4\n",
            "11\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "16\n",
            "[ 9 12 15  0 14  4 11 16  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 14  4 11 16  0  4  0  0 11  0  0 14 16  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "19\n",
            "3\n",
            "13\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "12\n",
            "[ 9 12 15  0 19  3 13 12  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 19  3 13 12  0  3  0  0 13 12  0 19  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "8\n",
            "16\n",
            "3\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "7\n",
            "[ 9 12 15  0  8 16  3  7  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  8 16  3  7  0  3  0  0  8  7  0 16  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "15\n",
            "5\n",
            "8\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "16\n",
            "[ 9 12 15  0 15  5  8 16  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 15  5  8 16  0  5  0  0  8  0  0 15 16  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "13\n",
            "4\n",
            "17\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "18\n",
            "[ 9 12 15  0 13  4 17 18  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 13  4 17 18  0  4  0  0 13  0  0 17 18  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "17\n",
            "13\n",
            "2\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "18\n",
            "[ 9 12 15  0 17 13  2 18  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 17 13  2 18  0  2  0  0 13  0  0 17 18  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "17\n",
            "2\n",
            "13\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "6\n",
            "[ 9 12 15  0 17  2 13  6  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 17  2 13  6  0  2  6  0 13  0  0 17  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "17\n",
            "1\n",
            "9\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "18\n",
            "[ 9 12 15  0 17  1  9 18  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 17  1  9 18  0  1  0  0  9  0  0 17 18  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "7\n",
            "3\n",
            "15\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "12\n",
            "[ 9 12 15  0  7  3 15 12  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  7  3 15 12  0  3  0  0  7 12  0 15  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "19\n",
            "8\n",
            "5\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "4\n",
            "[ 9 12 15  0 19  8  5  4  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 19  8  5  4  0  5  4  0  8  0  0 19  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "15\n",
            "8\n",
            "4\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "13\n",
            "[ 9 12 15  0 15  8  4 13  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 15  8  4 13  0  4  0  0  8 13  0 15  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "1\n",
            "10\n",
            "15\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "9\n",
            "[ 9 12 15  0  1 10 15  9  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  1 10 15  9  0  1  0  0 10  9  0 15  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "1\n",
            "17\n",
            "11\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "15\n",
            "[ 9 12 15  0  1 17 11 15  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  1 17 11 15  0  1  0  0 11  0  0 17 15  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "6\n",
            "8\n",
            "19\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "18\n",
            "[ 9 12 15  0  6  8 19 18  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  6  8 19 18  0  6  0  0  8  0  0 19 18  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "12\n",
            "2\n",
            "18\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "19\n",
            "[ 9 12 15  0 12  2 18 19  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 12  2 18 19  0  2  0  0 12  0  0 18 19  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "11\n",
            "3\n",
            "15\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "17\n",
            "[ 9 12 15  0 11  3 15 17  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 11  3 15 17  0  3  0  0 11  0  0 15 17  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "4\n",
            "19\n",
            "12\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "2\n",
            "[ 9 12 15  0  4 19 12  2  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  4 19 12  2  0  4  2  0 12  0  0 19  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "16\n",
            "13\n",
            "5\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "5\n",
            "[ 9 12 15  0 16 13  5  5  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 16 13  5  5  0  5  5  0 13  0  0 16  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "4\n",
            "16\n",
            "11\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "15\n",
            "[ 9 12 15  0  4 16 11 15  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  4 16 11 15  0  4  0  0 11  0  0 16 15  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "19\n",
            "12\n",
            "3\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "11\n",
            "[ 9 12 15  0 19 12  3 11  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 19 12  3 11  0  3  0  0 12 11  0 19  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "13\n",
            "19\n",
            "5\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "2\n",
            "[ 9 12 15  0 13 19  5  2  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 13 19  5  2  0  5  2  0 13  0  0 19  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "7\n",
            "14\n",
            "6\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "14\n",
            "[ 9 12 15  0  7 14  6 14  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  7 14  6 14  0  6  0  0  7  0  0 14 14  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "1\n",
            "9\n",
            "14\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "19\n",
            "[ 9 12 15  0  1  9 14 19  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  1  9 14 19  0  1  0  0  9  0  0 14 19  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "14\n",
            "8\n",
            "2\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "16\n",
            "[ 9 12 15  0 14  8  2 16  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 14  8  2 16  0  2  0  0  8  0  0 14 16  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "13\n",
            "1\n",
            "17\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "10\n",
            "[ 9 12 15  0 13  1 17 10  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 13  1 17 10  0  1  0  0 13 10  0 17  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "9\n",
            "6\n",
            "14\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "6\n",
            "[ 9 12 15  0  9  6 14  6  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  9  6 14  6  0  6  6  0  9  0  0 14  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "1\n",
            "8\n",
            "18\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "11\n",
            "[ 9 12 15  0  1  8 18 11  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  1  8 18 11  0  1  0  0  8 11  0 18  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "17\n",
            "3\n",
            "10\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "12\n",
            "[ 9 12 15  0 17  3 10 12  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 17  3 10 12  0  3  0  0 10 12  0 17  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "7\n",
            "5\n",
            "14\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "5\n",
            "[ 9 12 15  0  7  5 14  5  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  7  5 14  5  0  5  5  0  7  0  0 14  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "3\n",
            "17\n",
            "8\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "4\n",
            "[ 9 12 15  0  3 17  8  4  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  3 17  8  4  0  3  4  0  8  0  0 17  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "9\n",
            "3\n",
            "18\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "16\n",
            "[ 9 12 15  0  9  3 18 16  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  9  3 18 16  0  3  0  0  9  0  0 18 16  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "11\n",
            "15\n",
            "4\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "10\n",
            "[ 9 12 15  0 11 15  4 10  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 11 15  4 10  0  4  0  0 11 10  0 15  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "4\n",
            "11\n",
            "15\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "15\n",
            "[ 9 12 15  0  4 11 15 15  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  4 11 15 15  0  4  0  0 11  0  0 15 15  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "12\n",
            "14\n",
            "1\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "1\n",
            "[ 9 12 15  0 12 14  1  1  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 12 14  1  1  0  1  1  0 12  0  0 14  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "7\n",
            "2\n",
            "17\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "2\n",
            "[ 9 12 15  0  7  2 17  2  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  7  2 17  2  0  2  2  0  7  0  0 17  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "2\n",
            "11\n",
            "16\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "13\n",
            "[ 9 12 15  0  2 11 16 13  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  2 11 16 13  0  2  0  0 11 13  0 16  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "16\n",
            "9\n",
            "6\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "13\n",
            "[ 9 12 15  0 16  9  6 13  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 16  9  6 13  0  6  0  0  9 13  0 16  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "17\n",
            "11\n",
            "5\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "11\n",
            "[ 9 12 15  0 17 11  5 11  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 17 11  5 11  0  5  0  0 11 11  0 17  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "14\n",
            "2\n",
            "7\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "18\n",
            "[ 9 12 15  0 14  2  7 18  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 14  2  7 18  0  2  0  0  7  0  0 14 18  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "1\n",
            "8\n",
            "15\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "10\n",
            "[ 9 12 15  0  1  8 15 10  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  1  8 15 10  0  1  0  0  8 10  0 15  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "6\n",
            "15\n",
            "10\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "19\n",
            "[ 9 12 15  0  6 15 10 19  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  6 15 10 19  0  6  0  0 10  0  0 15 19  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "11\n",
            "19\n",
            "2\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "5\n",
            "[ 9 12 15  0 11 19  2  5  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 11 19  2  5  0  2  5  0 11  0  0 19  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "15\n",
            "6\n",
            "13\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "3\n",
            "[ 9 12 15  0 15  6 13  3  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 15  6 13  3  0  6  3  0 13  0  0 15  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "3\n",
            "19\n",
            "8\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "18\n",
            "[ 9 12 15  0  3 19  8 18  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  3 19  8 18  0  3  0  0  8  0  0 19 18  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "2\n",
            "9\n",
            "15\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "6\n",
            "[ 9 12 15  0  2  9 15  6  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  2  9 15  6  0  2  6  0  9  0  0 15  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "1\n",
            "13\n",
            "18\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "13\n",
            "[ 9 12 15  0  1 13 18 13  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  1 13 18 13  0  1  0  0 13 13  0 18  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "12\n",
            "19\n",
            "6\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "14\n",
            "[ 9 12 15  0 12 19  6 14  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 12 19  6 14  0  6  0  0 12  0  0 19 14  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "17\n",
            "9\n",
            "4\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "8\n",
            "[ 9 12 15  0 17  9  4  8  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 17  9  4  8  0  4  0  0  9  8  0 17  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "6\n",
            "13\n",
            "17\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "5\n",
            "[ 9 12 15  0  6 13 17  5  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  6 13 17  5  0  6  5  0 13  0  0 17  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "18\n",
            "13\n",
            "5\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "15\n",
            "[ 9 12 15  0 18 13  5 15  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 18 13  5 15  0  5  0  0 13  0  0 18 15  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "18\n",
            "3\n",
            "11\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "18\n",
            "[ 9 12 15  0 18  3 11 18  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 18  3 11 18  0  3  0  0 11  0  0 18 18  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "17\n",
            "13\n",
            "5\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "15\n",
            "[ 9 12 15  0 17 13  5 15  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 17 13  5 15  0  5  0  0 13  0  0 17 15  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "12\n",
            "16\n",
            "6\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "19\n",
            "[ 9 12 15  0 12 16  6 19  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 12 16  6 19  0  6  0  0 12  0  0 16 19  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "1\n",
            "7\n",
            "19\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "16\n",
            "[ 9 12 15  0  1  7 19 16  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  1  7 19 16  0  1  0  0  7  0  0 19 16  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "9\n",
            "3\n",
            "16\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "9\n",
            "[ 9 12 15  0  9  3 16  9  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  9  3 16  9  0  3  0  0  9  9  0 16  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "9\n",
            "3\n",
            "17\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "6\n",
            "[ 9 12 15  0  9  3 17  6  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  9  3 17  6  0  3  6  0  9  0  0 17  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "3\n",
            "13\n",
            "16\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "2\n",
            "[ 9 12 15  0  3 13 16  2  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  3 13 16  2  0  3  2  0 13  0  0 16  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "18\n",
            "3\n",
            "13\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "10\n",
            "[ 9 12 15  0 18  3 13 10  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 18  3 13 10  0  3  0  0 13 10  0 18  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "19\n",
            "8\n",
            "3\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "19\n",
            "[ 9 12 15  0 19  8  3 19  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 19  8  3 19  0  3  0  0  8  0  0 19 19  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "2\n",
            "13\n",
            "16\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "3\n",
            "[ 9 12 15  0  2 13 16  3  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  2 13 16  3  0  2  3  0 13  0  0 16  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "18\n",
            "6\n",
            "9\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "14\n",
            "[ 9 12 15  0 18  6  9 14  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 18  6  9 14  0  6  0  0  9  0  0 18 14  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "9\n",
            "4\n",
            "18\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "4\n",
            "[ 9 12 15  0  9  4 18  4  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  9  4 18  4  0  4  4  0  9  0  0 18  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "7\n",
            "14\n",
            "6\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "10\n",
            "[ 9 12 15  0  7 14  6 10  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  7 14  6 10  0  6  0  0  7 10  0 14  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "16\n",
            "2\n",
            "12\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "7\n",
            "[ 9 12 15  0 16  2 12  7  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 16  2 12  7  0  2  0  0 12  7  0 16  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "2\n",
            "17\n",
            "12\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "15\n",
            "[ 9 12 15  0  2 17 12 15  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  2 17 12 15  0  2  0  0 12  0  0 17 15  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "14\n",
            "13\n",
            "3\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "3\n",
            "[ 9 12 15  0 14 13  3  3  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 14 13  3  3  0  3  3  0 13  0  0 14  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "16\n",
            "11\n",
            "1\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "2\n",
            "[ 9 12 15  0 16 11  1  2  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 16 11  1  2  0  1  2  0 11  0  0 16  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "2\n",
            "7\n",
            "16\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "6\n",
            "[ 9 12 15  0  2  7 16  6  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  2  7 16  6  0  2  6  0  7  0  0 16  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "14\n",
            "13\n",
            "3\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "16\n",
            "[ 9 12 15  0 14 13  3 16  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 14 13  3 16  0  3  0  0 13  0  0 14 16  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "11\n",
            "4\n",
            "18\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "11\n",
            "[ 9 12 15  0 11  4 18 11  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 11  4 18 11  0  4  0  0 11 11  0 18  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "12\n",
            "17\n",
            "2\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "2\n",
            "[ 9 12 15  0 12 17  2  2  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 12 17  2  2  0  2  2  0 12  0  0 17  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "2\n",
            "9\n",
            "19\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "10\n",
            "[ 9 12 15  0  2  9 19 10  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  2  9 19 10  0  2  0  0  9 10  0 19  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "6\n",
            "18\n",
            "9\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "1\n",
            "[ 9 12 15  0  6 18  9  1  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  6 18  9  1  0  6  1  0  9  0  0 18  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "13\n",
            "2\n",
            "17\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "8\n",
            "[ 9 12 15  0 13  2 17  8  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 13  2 17  8  0  2  0  0 13  8  0 17  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "9\n",
            "4\n",
            "16\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "12\n",
            "[ 9 12 15  0  9  4 16 12  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  9  4 16 12  0  4  0  0  9 12  0 16  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "16\n",
            "6\n",
            "12\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "5\n",
            "[ 9 12 15  0 16  6 12  5  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 16  6 12  5  0  6  5  0 12  0  0 16  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "15\n",
            "8\n",
            "1\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "1\n",
            "[ 9 12 15  0 15  8  1  1  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 15  8  1  1  0  1  1  0  8  0  0 15  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "2\n",
            "9\n",
            "19\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "3\n",
            "[ 9 12 15  0  2  9 19  3  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  2  9 19  3  0  2  3  0  9  0  0 19  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "8\n",
            "19\n",
            "6\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "16\n",
            "[ 9 12 15  0  8 19  6 16  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  8 19  6 16  0  6  0  0  8  0  0 19 16  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "7\n",
            "15\n",
            "1\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "12\n",
            "[ 9 12 15  0  7 15  1 12  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  7 15  1 12  0  1  0  0  7 12  0 15  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "11\n",
            "2\n",
            "16\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "6\n",
            "[ 9 12 15  0 11  2 16  6  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 11  2 16  6  0  2  6  0 11  0  0 16  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "10\n",
            "2\n",
            "17\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "18\n",
            "[ 9 12 15  0 10  2 17 18  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 10  2 17 18  0  2  0  0 10  0  0 17 18  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "9\n",
            "3\n",
            "15\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "7\n",
            "[ 9 12 15  0  9  3 15  7  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  9  3 15  7  0  3  0  0  9  7  0 15  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "8\n",
            "17\n",
            "6\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "8\n",
            "[ 9 12 15  0  8 17  6  8  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  8 17  6  8  0  6  0  0  8  8  0 17  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "1\n",
            "9\n",
            "14\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "16\n",
            "[ 9 12 15  0  1  9 14 16  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  1  9 14 16  0  1  0  0  9  0  0 14 16  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "13\n",
            "19\n",
            "5\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "9\n",
            "[ 9 12 15  0 13 19  5  9  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 13 19  5  9  0  5  0  0 13  9  0 19  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "10\n",
            "3\n",
            "17\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "15\n",
            "[ 9 12 15  0 10  3 17 15  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 10  3 17 15  0  3  0  0 10  0  0 17 15  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "13\n",
            "15\n",
            "1\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "16\n",
            "[ 9 12 15  0 13 15  1 16  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 13 15  1 16  0  1  0  0 13  0  0 15 16  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "5\n",
            "17\n",
            "9\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "19\n",
            "[ 9 12 15  0  5 17  9 19  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  5 17  9 19  0  5  0  0  9  0  0 17 19  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "4\n",
            "12\n",
            "14\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "16\n",
            "[ 9 12 15  0  4 12 14 16  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  4 12 14 16  0  4  0  0 12  0  0 14 16  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "4\n",
            "8\n",
            "15\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "13\n",
            "[ 9 12 15  0  4  8 15 13  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  4  8 15 13  0  4  0  0  8 13  0 15  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "18\n",
            "3\n",
            "7\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "16\n",
            "[ 9 12 15  0 18  3  7 16  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 18  3  7 16  0  3  0  0  7  0  0 18 16  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "15\n",
            "11\n",
            "3\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "15\n",
            "[ 9 12 15  0 15 11  3 15  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 15 11  3 15  0  3  0  0 11  0  0 15 15  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "1\n",
            "14\n",
            "13\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "3\n",
            "[ 9 12 15  0  1 14 13  3  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  1 14 13  3  0  1  3  0 13  0  0 14  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "6\n",
            "17\n",
            "11\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "18\n",
            "[ 9 12 15  0  6 17 11 18  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  6 17 11 18  0  6  0  0 11  0  0 17 18  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "13\n",
            "19\n",
            "6\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "17\n",
            "[ 9 12 15  0 13 19  6 17  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 13 19  6 17  0  6  0  0 13  0  0 19 17  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "16\n",
            "3\n",
            "11\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "2\n",
            "[ 9 12 15  0 16  3 11  2  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 16  3 11  2  0  3  2  0 11  0  0 16  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "4\n",
            "14\n",
            "8\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "2\n",
            "[ 9 12 15  0  4 14  8  2  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  4 14  8  2  0  4  2  0  8  0  0 14  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "5\n",
            "11\n",
            "19\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "18\n",
            "[ 9 12 15  0  5 11 19 18  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  5 11 19 18  0  5  0  0 11  0  0 19 18  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "7\n",
            "15\n",
            "5\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "6\n",
            "[ 9 12 15  0  7 15  5  6  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  7 15  5  6  0  5  6  0  7  0  0 15  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "17\n",
            "6\n",
            "13\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "5\n",
            "[ 9 12 15  0 17  6 13  5  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 17  6 13  5  0  6  5  0 13  0  0 17  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "7\n",
            "14\n",
            "5\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "6\n",
            "[ 9 12 15  0  7 14  5  6  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  7 14  5  6  0  5  6  0  7  0  0 14  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "8\n",
            "2\n",
            "19\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "2\n",
            "[ 9 12 15  0  8  2 19  2  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  8  2 19  2  0  2  2  0  8  0  0 19  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "9\n",
            "17\n",
            "5\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "11\n",
            "[ 9 12 15  0  9 17  5 11  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  9 17  5 11  0  5  0  0  9 11  0 17  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "4\n",
            "15\n",
            "7\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "9\n",
            "[ 9 12 15  0  4 15  7  9  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  4 15  7  9  0  4  0  0  7  9  0 15  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "9\n",
            "14\n",
            "3\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "8\n",
            "[ 9 12 15  0  9 14  3  8  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  9 14  3  8  0  3  0  0  9  8  0 14  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "12\n",
            "5\n",
            "14\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "1\n",
            "[ 9 12 15  0 12  5 14  1  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 12  5 14  1  0  5  1  0 12  0  0 14  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "8\n",
            "15\n",
            "6\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "10\n",
            "[ 9 12 15  0  8 15  6 10  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  8 15  6 10  0  6  0  0  8 10  0 15  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "11\n",
            "4\n",
            "14\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "3\n",
            "[ 9 12 15  0 11  4 14  3  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 11  4 14  3  0  4  3  0 11  0  0 14  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "6\n",
            "14\n",
            "10\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "17\n",
            "[ 9 12 15  0  6 14 10 17  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  6 14 10 17  0  6  0  0 10  0  0 14 17  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "3\n",
            "17\n",
            "10\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "16\n",
            "[ 9 12 15  0  3 17 10 16  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  3 17 10 16  0  3  0  0 10  0  0 17 16  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "10\n",
            "5\n",
            "19\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "2\n",
            "[ 9 12 15  0 10  5 19  2  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 10  5 19  2  0  5  2  0 10  0  0 19  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "19\n",
            "11\n",
            "6\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "3\n",
            "[ 9 12 15  0 19 11  6  3  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 19 11  6  3  0  6  3  0 11  0  0 19  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "11\n",
            "1\n",
            "17\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "18\n",
            "[ 9 12 15  0 11  1 17 18  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 11  1 17 18  0  1  0  0 11  0  0 17 18  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "3\n",
            "7\n",
            "18\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "6\n",
            "[ 9 12 15  0  3  7 18  6  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  3  7 18  6  0  3  6  0  7  0  0 18  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "13\n",
            "14\n",
            "5\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "15\n",
            "[ 9 12 15  0 13 14  5 15  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 13 14  5 15  0  5  0  0 13  0  0 14 15  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "5\n",
            "13\n",
            "19\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "18\n",
            "[ 9 12 15  0  5 13 19 18  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  5 13 19 18  0  5  0  0 13  0  0 19 18  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "1\n",
            "14\n",
            "9\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "1\n",
            "[ 9 12 15  0  1 14  9  1  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  1 14  9  1  0  1  1  0  9  0  0 14  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "11\n",
            "17\n",
            "6\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "9\n",
            "[ 9 12 15  0 11 17  6  9  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 11 17  6  9  0  6  0  0 11  9  0 17  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "2\n",
            "11\n",
            "19\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "4\n",
            "[ 9 12 15  0  2 11 19  4  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  2 11 19  4  0  2  4  0 11  0  0 19  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "3\n",
            "8\n",
            "15\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "8\n",
            "[ 9 12 15  0  3  8 15  8  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  3  8 15  8  0  3  0  0  8  8  0 15  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "17\n",
            "5\n",
            "7\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "1\n",
            "[ 9 12 15  0 17  5  7  1  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 17  5  7  1  0  5  1  0  7  0  0 17  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "10\n",
            "16\n",
            "6\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "9\n",
            "[ 9 12 15  0 10 16  6  9  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 10 16  6  9  0  6  0  0 10  9  0 16  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "16\n",
            "5\n",
            "8\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "4\n",
            "[ 9 12 15  0 16  5  8  4  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 16  5  8  4  0  5  4  0  8  0  0 16  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "18\n",
            "5\n",
            "10\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "12\n",
            "[ 9 12 15  0 18  5 10 12  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 18  5 10 12  0  5  0  0 10 12  0 18  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "11\n",
            "14\n",
            "3\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "14\n",
            "[ 9 12 15  0 11 14  3 14  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 11 14  3 14  0  3  0  0 11  0  0 14 14  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "11\n",
            "18\n",
            "5\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "6\n",
            "[ 9 12 15  0 11 18  5  6  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 11 18  5  6  0  5  6  0 11  0  0 18  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "9\n",
            "5\n",
            "14\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "15\n",
            "[ 9 12 15  0  9  5 14 15  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  9  5 14 15  0  5  0  0  9  0  0 14 15  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "1\n",
            "15\n",
            "12\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "4\n",
            "[ 9 12 15  0  1 15 12  4  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  1 15 12  4  0  1  4  0 12  0  0 15  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "3\n",
            "15\n",
            "12\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "15\n",
            "[ 9 12 15  0  3 15 12 15  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  3 15 12 15  0  3  0  0 12  0  0 15 15  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "5\n",
            "16\n",
            "12\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "19\n",
            "[ 9 12 15  0  5 16 12 19  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  5 16 12 19  0  5  0  0 12  0  0 16 19  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "19\n",
            "3\n",
            "11\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "1\n",
            "[ 9 12 15  0 19  3 11  1  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 19  3 11  1  0  3  1  0 11  0  0 19  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "6\n",
            "15\n",
            "8\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "4\n",
            "[ 9 12 15  0  6 15  8  4  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  6 15  8  4  0  6  4  0  8  0  0 15  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "8\n",
            "3\n",
            "19\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "19\n",
            "[ 9 12 15  0  8  3 19 19  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  8  3 19 19  0  3  0  0  8  0  0 19 19  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "12\n",
            "3\n",
            "19\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "7\n",
            "[ 9 12 15  0 12  3 19  7  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 12  3 19  7  0  3  0  0 12  7  0 19  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "11\n",
            "4\n",
            "19\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "13\n",
            "[ 9 12 15  0 11  4 19 13  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 11  4 19 13  0  4  0  0 11 13  0 19  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "4\n",
            "12\n",
            "19\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "15\n",
            "[ 9 12 15  0  4 12 19 15  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  4 12 19 15  0  4  0  0 12  0  0 19 15  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "8\n",
            "1\n",
            "18\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "16\n",
            "[ 9 12 15  0  8  1 18 16  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  8  1 18 16  0  1  0  0  8  0  0 18 16  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "13\n",
            "16\n",
            "6\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "4\n",
            "[ 9 12 15  0 13 16  6  4  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 13 16  6  4  0  6  4  0 13  0  0 16  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "3\n",
            "13\n",
            "18\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "16\n",
            "[ 9 12 15  0  3 13 18 16  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  3 13 18 16  0  3  0  0 13  0  0 18 16  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "5\n",
            "12\n",
            "18\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "8\n",
            "[ 9 12 15  0  5 12 18  8  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  5 12 18  8  0  5  0  0 12  8  0 18  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "1\n",
            "11\n",
            "18\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "6\n",
            "[ 9 12 15  0  1 11 18  6  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  1 11 18  6  0  1  6  0 11  0  0 18  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "14\n",
            "7\n",
            "3\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "17\n",
            "[ 9 12 15  0 14  7  3 17  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 14  7  3 17  0  3  0  0  7  0  0 14 17  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "5\n",
            "16\n",
            "9\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "18\n",
            "[ 9 12 15  0  5 16  9 18  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  5 16  9 18  0  5  0  0  9  0  0 16 18  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "10\n",
            "5\n",
            "14\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "16\n",
            "[ 9 12 15  0 10  5 14 16  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 10  5 14 16  0  5  0  0 10  0  0 14 16  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "10\n",
            "18\n",
            "4\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "5\n",
            "[ 9 12 15  0 10 18  4  5  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 10 18  4  5  0  4  5  0 10  0  0 18  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "13\n",
            "1\n",
            "18\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "12\n",
            "[ 9 12 15  0 13  1 18 12  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 13  1 18 12  0  1  0  0 13 12  0 18  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "19\n",
            "12\n",
            "2\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "8\n",
            "[ 9 12 15  0 19 12  2  8  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 19 12  2  8  0  2  0  0 12  8  0 19  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "12\n",
            "5\n",
            "18\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "10\n",
            "[ 9 12 15  0 12  5 18 10  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 12  5 18 10  0  5  0  0 12 10  0 18  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "19\n",
            "4\n",
            "8\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "11\n",
            "[ 9 12 15  0 19  4  8 11  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 19  4  8 11  0  4  0  0  8 11  0 19  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "2\n",
            "14\n",
            "12\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "17\n",
            "[ 9 12 15  0  2 14 12 17  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  2 14 12 17  0  2  0  0 12  0  0 14 17  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "2\n",
            "17\n",
            "11\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "14\n",
            "[ 9 12 15  0  2 17 11 14  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  2 17 11 14  0  2  0  0 11  0  0 17 14  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "17\n",
            "10\n",
            "3\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "7\n",
            "[ 9 12 15  0 17 10  3  7  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 17 10  3  7  0  3  0  0 10  7  0 17  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "3\n",
            "17\n",
            "11\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "8\n",
            "[ 9 12 15  0  3 17 11  8  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  3 17 11  8  0  3  0  0 11  8  0 17  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "8\n",
            "14\n",
            "6\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "8\n",
            "[ 9 12 15  0  8 14  6  8  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  8 14  6  8  0  6  0  0  8  8  0 14  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "13\n",
            "18\n",
            "6\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "6\n",
            "[ 9 12 15  0 13 18  6  6  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 13 18  6  6  0  6  6  0 13  0  0 18  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "12\n",
            "17\n",
            "3\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "11\n",
            "[ 9 12 15  0 12 17  3 11  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 12 17  3 11  0  3  0  0 12 11  0 17  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "9\n",
            "5\n",
            "18\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "13\n",
            "[ 9 12 15  0  9  5 18 13  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  9  5 18 13  0  5  0  0  9 13  0 18  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "9\n",
            "1\n",
            "15\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "8\n",
            "[ 9 12 15  0  9  1 15  8  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  9  1 15  8  0  1  0  0  9  8  0 15  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "15\n",
            "10\n",
            "3\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "7\n",
            "[ 9 12 15  0 15 10  3  7  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 15 10  3  7  0  3  0  0 10  7  0 15  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "5\n",
            "19\n",
            "12\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "2\n",
            "[ 9 12 15  0  5 19 12  2  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  5 19 12  2  0  5  2  0 12  0  0 19  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "16\n",
            "12\n",
            "4\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "17\n",
            "[ 9 12 15  0 16 12  4 17  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 16 12  4 17  0  4  0  0 12  0  0 16 17  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "3\n",
            "15\n",
            "10\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "19\n",
            "[ 9 12 15  0  3 15 10 19  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  3 15 10 19  0  3  0  0 10  0  0 15 19  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "5\n",
            "12\n",
            "19\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "18\n",
            "[ 9 12 15  0  5 12 19 18  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  5 12 19 18  0  5  0  0 12  0  0 19 18  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "10\n",
            "17\n",
            "5\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "18\n",
            "[ 9 12 15  0 10 17  5 18  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 10 17  5 18  0  5  0  0 10  0  0 17 18  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "17\n",
            "8\n",
            "2\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "2\n",
            "[ 9 12 15  0 17  8  2  2  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 17  8  2  2  0  2  2  0  8  0  0 17  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "18\n",
            "5\n",
            "9\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "3\n",
            "[ 9 12 15  0 18  5  9  3  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 18  5  9  3  0  5  3  0  9  0  0 18  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "3\n",
            "16\n",
            "10\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "18\n",
            "[ 9 12 15  0  3 16 10 18  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  3 16 10 18  0  3  0  0 10  0  0 16 18  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "13\n",
            "17\n",
            "6\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "15\n",
            "[ 9 12 15  0 13 17  6 15  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 13 17  6 15  0  6  0  0 13  0  0 17 15  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "17\n",
            "9\n",
            "1\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "1\n",
            "[ 9 12 15  0 17  9  1  1  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 17  9  1  1  0  1  1  0  9  0  0 17  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "13\n",
            "17\n",
            "2\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "2\n",
            "[ 9 12 15  0 13 17  2  2  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 13 17  2  2  0  2  2  0 13  0  0 17  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "9\n",
            "6\n",
            "19\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "18\n",
            "[ 9 12 15  0  9  6 19 18  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  9  6 19 18  0  6  0  0  9  0  0 19 18  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "5\n",
            "17\n",
            "12\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "17\n",
            "[ 9 12 15  0  5 17 12 17  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  5 17 12 17  0  5  0  0 12  0  0 17 17  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "1\n",
            "12\n",
            "14\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "19\n",
            "[ 9 12 15  0  1 12 14 19  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  1 12 14 19  0  1  0  0 12  0  0 14 19  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "2\n",
            "11\n",
            "16\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "6\n",
            "[ 9 12 15  0  2 11 16  6  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  2 11 16  6  0  2  6  0 11  0  0 16  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "17\n",
            "13\n",
            "1\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "7\n",
            "[ 9 12 15  0 17 13  1  7  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 17 13  1  7  0  1  0  0 13  7  0 17  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "17\n",
            "5\n",
            "8\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "7\n",
            "[ 9 12 15  0 17  5  8  7  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 17  5  8  7  0  5  0  0  8  7  0 17  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "15\n",
            "5\n",
            "10\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "17\n",
            "[ 9 12 15  0 15  5 10 17  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 15  5 10 17  0  5  0  0 10  0  0 15 17  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "19\n",
            "13\n",
            "1\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "18\n",
            "[ 9 12 15  0 19 13  1 18  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 19 13  1 18  0  1  0  0 13  0  0 19 18  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "6\n",
            "16\n",
            "8\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "9\n",
            "[ 9 12 15  0  6 16  8  9  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  6 16  8  9  0  6  0  0  8  9  0 16  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "16\n",
            "2\n",
            "12\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "2\n",
            "[ 9 12 15  0 16  2 12  2  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 16  2 12  2  0  2  2  0 12  0  0 16  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "18\n",
            "4\n",
            "13\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "19\n",
            "[ 9 12 15  0 18  4 13 19  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 18  4 13 19  0  4  0  0 13  0  0 18 19  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "9\n",
            "4\n",
            "18\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "18\n",
            "[ 9 12 15  0  9  4 18 18  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  9  4 18 18  0  4  0  0  9  0  0 18 18  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "10\n",
            "18\n",
            "1\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "15\n",
            "[ 9 12 15  0 10 18  1 15  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 10 18  1 15  0  1  0  0 10  0  0 18 15  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "14\n",
            "11\n",
            "2\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "11\n",
            "[ 9 12 15  0 14 11  2 11  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 14 11  2 11  0  2  0  0 11 11  0 14  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "19\n",
            "3\n",
            "12\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "12\n",
            "[ 9 12 15  0 19  3 12 12  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 19  3 12 12  0  3  0  0 12 12  0 19  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "4\n",
            "18\n",
            "13\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "9\n",
            "[ 9 12 15  0  4 18 13  9  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  4 18 13  9  0  4  0  0 13  9  0 18  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "10\n",
            "1\n",
            "15\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "11\n",
            "[ 9 12 15  0 10  1 15 11  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 10  1 15 11  0  1  0  0 10 11  0 15  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "17\n",
            "4\n",
            "7\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "3\n",
            "[ 9 12 15  0 17  4  7  3  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 17  4  7  3  0  4  3  0  7  0  0 17  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "14\n",
            "1\n",
            "10\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "10\n",
            "[ 9 12 15  0 14  1 10 10  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 14  1 10 10  0  1  0  0 10 10  0 14  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "12\n",
            "6\n",
            "16\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "10\n",
            "[ 9 12 15  0 12  6 16 10  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 12  6 16 10  0  6  0  0 12 10  0 16  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "3\n",
            "14\n",
            "13\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "12\n",
            "[ 9 12 15  0  3 14 13 12  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  3 14 13 12  0  3  0  0 13 12  0 14  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "7\n",
            "14\n",
            "1\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "1\n",
            "[ 9 12 15  0  7 14  1  1  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  7 14  1  1  0  1  1  0  7  0  0 14  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "11\n",
            "17\n",
            "4\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "4\n",
            "[ 9 12 15  0 11 17  4  4  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 11 17  4  4  0  4  4  0 11  0  0 17  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "2\n",
            "7\n",
            "18\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "15\n",
            "[ 9 12 15  0  2  7 18 15  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  2  7 18 15  0  2  0  0  7  0  0 18 15  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "17\n",
            "5\n",
            "7\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "14\n",
            "[ 9 12 15  0 17  5  7 14  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 17  5  7 14  0  5  0  0  7  0  0 17 14  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "5\n",
            "12\n",
            "15\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "5\n",
            "[ 9 12 15  0  5 12 15  5  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  5 12 15  5  0  5  5  0 12  0  0 15  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "8\n",
            "19\n",
            "4\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "18\n",
            "[ 9 12 15  0  8 19  4 18  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  8 19  4 18  0  4  0  0  8  0  0 19 18  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "19\n",
            "1\n",
            "8\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "3\n",
            "[ 9 12 15  0 19  1  8  3  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 19  1  8  3  0  1  3  0  8  0  0 19  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "3\n",
            "19\n",
            "7\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "10\n",
            "[ 9 12 15  0  3 19  7 10  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  3 19  7 10  0  3  0  0  7 10  0 19  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "5\n",
            "19\n",
            "7\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "10\n",
            "[ 9 12 15  0  5 19  7 10  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  5 19  7 10  0  5  0  0  7 10  0 19  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "5\n",
            "11\n",
            "15\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "15\n",
            "[ 9 12 15  0  5 11 15 15  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  5 11 15 15  0  5  0  0 11  0  0 15 15  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "19\n",
            "3\n",
            "8\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "11\n",
            "[ 9 12 15  0 19  3  8 11  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 19  3  8 11  0  3  0  0  8 11  0 19  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "7\n",
            "2\n",
            "18\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "8\n",
            "[ 9 12 15  0  7  2 18  8  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  7  2 18  8  0  2  0  0  7  8  0 18  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "12\n",
            "17\n",
            "1\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "14\n",
            "[ 9 12 15  0 12 17  1 14  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 12 17  1 14  0  1  0  0 12  0  0 17 14  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "10\n",
            "16\n",
            "2\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "1\n",
            "[ 9 12 15  0 10 16  2  1  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 10 16  2  1  0  2  1  0 10  0  0 16  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "14\n",
            "9\n",
            "6\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "14\n",
            "[ 9 12 15  0 14  9  6 14  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 14  9  6 14  0  6  0  0  9  0  0 14 14  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "16\n",
            "2\n",
            "13\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "16\n",
            "[ 9 12 15  0 16  2 13 16  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 16  2 13 16  0  2  0  0 13  0  0 16 16  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "14\n",
            "7\n",
            "5\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "4\n",
            "[ 9 12 15  0 14  7  5  4  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 14  7  5  4  0  5  4  0  7  0  0 14  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "5\n",
            "18\n",
            "10\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "8\n",
            "[ 9 12 15  0  5 18 10  8  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  5 18 10  8  0  5  0  0 10  8  0 18  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "11\n",
            "18\n",
            "6\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "13\n",
            "[ 9 12 15  0 11 18  6 13  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 11 18  6 13  0  6  0  0 11 13  0 18  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "6\n",
            "17\n",
            "9\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "13\n",
            "[ 9 12 15  0  6 17  9 13  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  6 17  9 13  0  6  0  0  9 13  0 17  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "5\n",
            "13\n",
            "15\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "2\n",
            "[ 9 12 15  0  5 13 15  2  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  5 13 15  2  0  5  2  0 13  0  0 15  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "5\n",
            "10\n",
            "15\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "1\n",
            "[ 9 12 15  0  5 10 15  1  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  5 10 15  1  0  5  1  0 10  0  0 15  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "7\n",
            "14\n",
            "4\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "9\n",
            "[ 9 12 15  0  7 14  4  9  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  7 14  4  9  0  4  0  0  7  9  0 14  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "14\n",
            "9\n",
            "4\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "19\n",
            "[ 9 12 15  0 14  9  4 19  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 14  9  4 19  0  4  0  0  9  0  0 14 19  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "2\n",
            "19\n",
            "9\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "19\n",
            "[ 9 12 15  0  2 19  9 19  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  2 19  9 19  0  2  0  0  9  0  0 19 19  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "10\n",
            "5\n",
            "19\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "12\n",
            "[ 9 12 15  0 10  5 19 12  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 10  5 19 12  0  5  0  0 10 12  0 19  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "5\n",
            "13\n",
            "15\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "3\n",
            "[ 9 12 15  0  5 13 15  3  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  5 13 15  3  0  5  3  0 13  0  0 15  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "13\n",
            "15\n",
            "2\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "1\n",
            "[ 9 12 15  0 13 15  2  1  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 13 15  2  1  0  2  1  0 13  0  0 15  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "7\n",
            "14\n",
            "5\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "18\n",
            "[ 9 12 15  0  7 14  5 18  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  7 14  5 18  0  5  0  0  7  0  0 14 18  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "6\n",
            "18\n",
            "13\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "2\n",
            "[ 9 12 15  0  6 18 13  2  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  6 18 13  2  0  6  2  0 13  0  0 18  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "4\n",
            "19\n",
            "8\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "11\n",
            "[ 9 12 15  0  4 19  8 11  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  4 19  8 11  0  4  0  0  8 11  0 19  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "16\n",
            "8\n",
            "5\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "9\n",
            "[ 9 12 15  0 16  8  5  9  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 16  8  5  9  0  5  0  0  8  9  0 16  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "1\n",
            "16\n",
            "7\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "6\n",
            "[ 9 12 15  0  1 16  7  6  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  1 16  7  6  0  1  6  0  7  0  0 16  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "12\n",
            "16\n",
            "3\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "15\n",
            "[ 9 12 15  0 12 16  3 15  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 12 16  3 15  0  3  0  0 12  0  0 16 15  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "9\n",
            "1\n",
            "19\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "18\n",
            "[ 9 12 15  0  9  1 19 18  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  9  1 19 18  0  1  0  0  9  0  0 19 18  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "8\n",
            "18\n",
            "6\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "3\n",
            "[ 9 12 15  0  8 18  6  3  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  8 18  6  3  0  6  3  0  8  0  0 18  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "12\n",
            "2\n",
            "17\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "11\n",
            "[ 9 12 15  0 12  2 17 11  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 12  2 17 11  0  2  0  0 12 11  0 17  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "4\n",
            "12\n",
            "15\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "4\n",
            "[ 9 12 15  0  4 12 15  4  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  4 12 15  4  0  4  4  0 12  0  0 15  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "12\n",
            "16\n",
            "2\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "14\n",
            "[ 9 12 15  0 12 16  2 14  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 12 16  2 14  0  2  0  0 12  0  0 16 14  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "12\n",
            "4\n",
            "14\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "2\n",
            "[ 9 12 15  0 12  4 14  2  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 12  4 14  2  0  4  2  0 12  0  0 14  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "12\n",
            "17\n",
            "1\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "12\n",
            "[ 9 12 15  0 12 17  1 12  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 12 17  1 12  0  1  0  0 12 12  0 17  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "19\n",
            "10\n",
            "6\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "2\n",
            "[ 9 12 15  0 19 10  6  2  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 19 10  6  2  0  6  2  0 10  0  0 19  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "6\n",
            "19\n",
            "7\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "11\n",
            "[ 9 12 15  0  6 19  7 11  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  6 19  7 11  0  6  0  0  7 11  0 19  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "11\n",
            "16\n",
            "2\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "2\n",
            "[ 9 12 15  0 11 16  2  2  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 11 16  2  2  0  2  2  0 11  0  0 16  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "17\n",
            "4\n",
            "11\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "15\n",
            "[ 9 12 15  0 17  4 11 15  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 17  4 11 15  0  4  0  0 11  0  0 17 15  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "9\n",
            "2\n",
            "14\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "10\n",
            "[ 9 12 15  0  9  2 14 10  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  9  2 14 10  0  2  0  0  9 10  0 14  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "3\n",
            "8\n",
            "16\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "12\n",
            "[ 9 12 15  0  3  8 16 12  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  3  8 16 12  0  3  0  0  8 12  0 16  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "16\n",
            "2\n",
            "12\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "18\n",
            "[ 9 12 15  0 16  2 12 18  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 16  2 12 18  0  2  0  0 12  0  0 16 18  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "14\n",
            "12\n",
            "2\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "8\n",
            "[ 9 12 15  0 14 12  2  8  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 14 12  2  8  0  2  0  0 12  8  0 14  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "16\n",
            "6\n",
            "11\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "18\n",
            "[ 9 12 15  0 16  6 11 18  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 16  6 11 18  0  6  0  0 11  0  0 16 18  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "17\n",
            "7\n",
            "4\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "17\n",
            "[ 9 12 15  0 17  7  4 17  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 17  7  4 17  0  4  0  0  7  0  0 17 17  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "10\n",
            "15\n",
            "3\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "15\n",
            "[ 9 12 15  0 10 15  3 15  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 10 15  3 15  0  3  0  0 10  0  0 15 15  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "5\n",
            "7\n",
            "17\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "1\n",
            "[ 9 12 15  0  5  7 17  1  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  5  7 17  1  0  5  1  0  7  0  0 17  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "11\n",
            "16\n",
            "3\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "12\n",
            "[ 9 12 15  0 11 16  3 12  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 11 16  3 12  0  3  0  0 11 12  0 16  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "1\n",
            "8\n",
            "15\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "4\n",
            "[ 9 12 15  0  1  8 15  4  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  1  8 15  4  0  1  4  0  8  0  0 15  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "5\n",
            "9\n",
            "18\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "12\n",
            "[ 9 12 15  0  5  9 18 12  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  5  9 18 12  0  5  0  0  9 12  0 18  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "15\n",
            "9\n",
            "6\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "17\n",
            "[ 9 12 15  0 15  9  6 17  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 15  9  6 17  0  6  0  0  9  0  0 15 17  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "19\n",
            "5\n",
            "8\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "18\n",
            "[ 9 12 15  0 19  5  8 18  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 19  5  8 18  0  5  0  0  8  0  0 19 18  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "11\n",
            "2\n",
            "16\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "19\n",
            "[ 9 12 15  0 11  2 16 19  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 11  2 16 19  0  2  0  0 11  0  0 16 19  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "7\n",
            "17\n",
            "2\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "3\n",
            "[ 9 12 15  0  7 17  2  3  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  7 17  2  3  0  2  3  0  7  0  0 17  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "6\n",
            "16\n",
            "10\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "3\n",
            "[ 9 12 15  0  6 16 10  3  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  6 16 10  3  0  6  3  0 10  0  0 16  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "10\n",
            "19\n",
            "4\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "8\n",
            "[ 9 12 15  0 10 19  4  8  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 10 19  4  8  0  4  0  0 10  8  0 19  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "13\n",
            "17\n",
            "6\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "15\n",
            "[ 9 12 15  0 13 17  6 15  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 13 17  6 15  0  6  0  0 13  0  0 17 15  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "1\n",
            "11\n",
            "15\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "4\n",
            "[ 9 12 15  0  1 11 15  4  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  1 11 15  4  0  1  4  0 11  0  0 15  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "3\n",
            "17\n",
            "8\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "19\n",
            "[ 9 12 15  0  3 17  8 19  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  3 17  8 19  0  3  0  0  8  0  0 17 19  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "19\n",
            "2\n",
            "10\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "17\n",
            "[ 9 12 15  0 19  2 10 17  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 19  2 10 17  0  2  0  0 10  0  0 19 17  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "18\n",
            "1\n",
            "13\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "18\n",
            "[ 9 12 15  0 18  1 13 18  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 18  1 13 18  0  1  0  0 13  0  0 18 18  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "2\n",
            "13\n",
            "18\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "15\n",
            "[ 9 12 15  0  2 13 18 15  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  2 13 18 15  0  2  0  0 13  0  0 18 15  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "3\n",
            "10\n",
            "19\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "5\n",
            "[ 9 12 15  0  3 10 19  5  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  3 10 19  5  0  3  5  0 10  0  0 19  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "9\n",
            "2\n",
            "14\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "15\n",
            "[ 9 12 15  0  9  2 14 15  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  9  2 14 15  0  2  0  0  9  0  0 14 15  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "5\n",
            "18\n",
            "13\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "8\n",
            "[ 9 12 15  0  5 18 13  8  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  5 18 13  8  0  5  0  0 13  8  0 18  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "17\n",
            "13\n",
            "3\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "19\n",
            "[ 9 12 15  0 17 13  3 19  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 17 13  3 19  0  3  0  0 13  0  0 17 19  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "17\n",
            "1\n",
            "7\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "2\n",
            "[ 9 12 15  0 17  1  7  2  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 17  1  7  2  0  1  2  0  7  0  0 17  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "3\n",
            "9\n",
            "15\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "2\n",
            "[ 9 12 15  0  3  9 15  2  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  3  9 15  2  0  3  2  0  9  0  0 15  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "11\n",
            "1\n",
            "18\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "5\n",
            "[ 9 12 15  0 11  1 18  5  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 11  1 18  5  0  1  5  0 11  0  0 18  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "12\n",
            "19\n",
            "2\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "19\n",
            "[ 9 12 15  0 12 19  2 19  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 12 19  2 19  0  2  0  0 12  0  0 19 19  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "11\n",
            "16\n",
            "5\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "17\n",
            "[ 9 12 15  0 11 16  5 17  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 11 16  5 17  0  5  0  0 11  0  0 16 17  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "4\n",
            "19\n",
            "7\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "11\n",
            "[ 9 12 15  0  4 19  7 11  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  4 19  7 11  0  4  0  0  7 11  0 19  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "13\n",
            "16\n",
            "5\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "1\n",
            "[ 9 12 15  0 13 16  5  1  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 13 16  5  1  0  5  1  0 13  0  0 16  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "12\n",
            "5\n",
            "16\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "14\n",
            "[ 9 12 15  0 12  5 16 14  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 12  5 16 14  0  5  0  0 12  0  0 16 14  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "15\n",
            "7\n",
            "2\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "15\n",
            "[ 9 12 15  0 15  7  2 15  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 15  7  2 15  0  2  0  0  7  0  0 15 15  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "15\n",
            "4\n",
            "8\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "7\n",
            "[ 9 12 15  0 15  4  8  7  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 15  4  8  7  0  4  0  0  8  7  0 15  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "14\n",
            "13\n",
            "2\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "1\n",
            "[ 9 12 15  0 14 13  2  1  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 14 13  2  1  0  2  1  0 13  0  0 14  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "9\n",
            "1\n",
            "16\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "17\n",
            "[ 9 12 15  0  9  1 16 17  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  9  1 16 17  0  1  0  0  9  0  0 16 17  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "16\n",
            "8\n",
            "5\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "5\n",
            "[ 9 12 15  0 16  8  5  5  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 16  8  5  5  0  5  5  0  8  0  0 16  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "7\n",
            "17\n",
            "6\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "16\n",
            "[ 9 12 15  0  7 17  6 16  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  7 17  6 16  0  6  0  0  7  0  0 17 16  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "9\n",
            "4\n",
            "16\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "8\n",
            "[ 9 12 15  0  9  4 16  8  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  9  4 16  8  0  4  0  0  9  8  0 16  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "6\n",
            "9\n",
            "18\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "15\n",
            "[ 9 12 15  0  6  9 18 15  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  6  9 18 15  0  6  0  0  9  0  0 18 15  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "9\n",
            "17\n",
            "5\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "16\n",
            "[ 9 12 15  0  9 17  5 16  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  9 17  5 16  0  5  0  0  9  0  0 17 16  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "13\n",
            "1\n",
            "14\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "8\n",
            "[ 9 12 15  0 13  1 14  8  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 13  1 14  8  0  1  0  0 13  8  0 14  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "15\n",
            "10\n",
            "4\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "15\n",
            "[ 9 12 15  0 15 10  4 15  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 15 10  4 15  0  4  0  0 10  0  0 15 15  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "12\n",
            "1\n",
            "18\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "4\n",
            "[ 9 12 15  0 12  1 18  4  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 12  1 18  4  0  1  4  0 12  0  0 18  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "11\n",
            "14\n",
            "4\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "3\n",
            "[ 9 12 15  0 11 14  4  3  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 11 14  4  3  0  4  3  0 11  0  0 14  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "17\n",
            "12\n",
            "6\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "3\n",
            "[ 9 12 15  0 17 12  6  3  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 17 12  6  3  0  6  3  0 12  0  0 17  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "4\n",
            "10\n",
            "15\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "11\n",
            "[ 9 12 15  0  4 10 15 11  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  4 10 15 11  0  4  0  0 10 11  0 15  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "17\n",
            "5\n",
            "10\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "18\n",
            "[ 9 12 15  0 17  5 10 18  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 17  5 10 18  0  5  0  0 10  0  0 17 18  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "11\n",
            "16\n",
            "1\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "8\n",
            "[ 9 12 15  0 11 16  1  8  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 11 16  1  8  0  1  0  0 11  8  0 16  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "14\n",
            "9\n",
            "4\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "14\n",
            "[ 9 12 15  0 14  9  4 14  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 14  9  4 14  0  4  0  0  9  0  0 14 14  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "11\n",
            "5\n",
            "19\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "14\n",
            "[ 9 12 15  0 11  5 19 14  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 11  5 19 14  0  5  0  0 11  0  0 19 14  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "4\n",
            "18\n",
            "13\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "6\n",
            "[ 9 12 15  0  4 18 13  6  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  4 18 13  6  0  4  6  0 13  0  0 18  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "18\n",
            "8\n",
            "3\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "13\n",
            "[ 9 12 15  0 18  8  3 13  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 18  8  3 13  0  3  0  0  8 13  0 18  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "13\n",
            "16\n",
            "2\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "12\n",
            "[ 9 12 15  0 13 16  2 12  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 13 16  2 12  0  2  0  0 13 12  0 16  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "13\n",
            "16\n",
            "4\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "4\n",
            "[ 9 12 15  0 13 16  4  4  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 13 16  4  4  0  4  4  0 13  0  0 16  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "2\n",
            "7\n",
            "14\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "19\n",
            "[ 9 12 15  0  2  7 14 19  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  2  7 14 19  0  2  0  0  7  0  0 14 19  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "3\n",
            "12\n",
            "18\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "5\n",
            "[ 9 12 15  0  3 12 18  5  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  3 12 18  5  0  3  5  0 12  0  0 18  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "16\n",
            "6\n",
            "10\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "14\n",
            "[ 9 12 15  0 16  6 10 14  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 16  6 10 14  0  6  0  0 10  0  0 16 14  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "1\n",
            "14\n",
            "9\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "16\n",
            "[ 9 12 15  0  1 14  9 16  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  1 14  9 16  0  1  0  0  9  0  0 14 16  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "12\n",
            "18\n",
            "3\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "17\n",
            "[ 9 12 15  0 12 18  3 17  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 12 18  3 17  0  3  0  0 12  0  0 18 17  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "12\n",
            "15\n",
            "6\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "7\n",
            "[ 9 12 15  0 12 15  6  7  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 12 15  6  7  0  6  0  0 12  7  0 15  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "4\n",
            "19\n",
            "8\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "16\n",
            "[ 9 12 15  0  4 19  8 16  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  4 19  8 16  0  4  0  0  8  0  0 19 16  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "2\n",
            "11\n",
            "16\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "7\n",
            "[ 9 12 15  0  2 11 16  7  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  2 11 16  7  0  2  0  0 11  7  0 16  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "4\n",
            "9\n",
            "19\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "10\n",
            "[ 9 12 15  0  4  9 19 10  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  4  9 19 10  0  4  0  0  9 10  0 19  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "3\n",
            "14\n",
            "12\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "9\n",
            "[ 9 12 15  0  3 14 12  9  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  3 14 12  9  0  3  0  0 12  9  0 14  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "11\n",
            "15\n",
            "2\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "17\n",
            "[ 9 12 15  0 11 15  2 17  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 11 15  2 17  0  2  0  0 11  0  0 15 17  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "14\n",
            "2\n",
            "8\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "16\n",
            "[ 9 12 15  0 14  2  8 16  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 14  2  8 16  0  2  0  0  8  0  0 14 16  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "1\n",
            "12\n",
            "14\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "15\n",
            "[ 9 12 15  0  1 12 14 15  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  1 12 14 15  0  1  0  0 12  0  0 14 15  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "12\n",
            "2\n",
            "16\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "8\n",
            "[ 9 12 15  0 12  2 16  8  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 12  2 16  8  0  2  0  0 12  8  0 16  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "18\n",
            "1\n",
            "10\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "18\n",
            "[ 9 12 15  0 18  1 10 18  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 18  1 10 18  0  1  0  0 10  0  0 18 18  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "17\n",
            "6\n",
            "12\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "18\n",
            "[ 9 12 15  0 17  6 12 18  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 17  6 12 18  0  6  0  0 12  0  0 17 18  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "13\n",
            "19\n",
            "1\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "14\n",
            "[ 9 12 15  0 13 19  1 14  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 13 19  1 14  0  1  0  0 13  0  0 19 14  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "13\n",
            "6\n",
            "14\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "14\n",
            "[ 9 12 15  0 13  6 14 14  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 13  6 14 14  0  6  0  0 13  0  0 14 14  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "5\n",
            "11\n",
            "17\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "19\n",
            "[ 9 12 15  0  5 11 17 19  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  5 11 17 19  0  5  0  0 11  0  0 17 19  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "8\n",
            "6\n",
            "17\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "6\n",
            "[ 9 12 15  0  8  6 17  6  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  8  6 17  6  0  6  6  0  8  0  0 17  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "18\n",
            "12\n",
            "6\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "18\n",
            "[ 9 12 15  0 18 12  6 18  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 18 12  6 18  0  6  0  0 12  0  0 18 18  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "16\n",
            "7\n",
            "3\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "1\n",
            "[ 9 12 15  0 16  7  3  1  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 16  7  3  1  0  3  1  0  7  0  0 16  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "13\n",
            "6\n",
            "16\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "16\n",
            "[ 9 12 15  0 13  6 16 16  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 13  6 16 16  0  6  0  0 13  0  0 16 16  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "8\n",
            "5\n",
            "16\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "3\n",
            "[ 9 12 15  0  8  5 16  3  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  8  5 16  3  0  5  3  0  8  0  0 16  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "2\n",
            "18\n",
            "8\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "10\n",
            "[ 9 12 15  0  2 18  8 10  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  2 18  8 10  0  2  0  0  8 10  0 18  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "13\n",
            "6\n",
            "16\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "3\n",
            "[ 9 12 15  0 13  6 16  3  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 13  6 16  3  0  6  3  0 13  0  0 16  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "9\n",
            "1\n",
            "14\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "6\n",
            "[ 9 12 15  0  9  1 14  6  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  9  1 14  6  0  1  6  0  9  0  0 14  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "6\n",
            "16\n",
            "8\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "12\n",
            "[ 9 12 15  0  6 16  8 12  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  6 16  8 12  0  6  0  0  8 12  0 16  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "7\n",
            "15\n",
            "4\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "17\n",
            "[ 9 12 15  0  7 15  4 17  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  7 15  4 17  0  4  0  0  7  0  0 15 17  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "14\n",
            "8\n",
            "3\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "14\n",
            "[ 9 12 15  0 14  8  3 14  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 14  8  3 14  0  3  0  0  8  0  0 14 14  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "2\n",
            "11\n",
            "18\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "17\n",
            "[ 9 12 15  0  2 11 18 17  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  2 11 18 17  0  2  0  0 11  0  0 18 17  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "12\n",
            "3\n",
            "15\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "12\n",
            "[ 9 12 15  0 12  3 15 12  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 12  3 15 12  0  3  0  0 12 12  0 15  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "11\n",
            "16\n",
            "5\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "14\n",
            "[ 9 12 15  0 11 16  5 14  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 11 16  5 14  0  5  0  0 11  0  0 16 14  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "10\n",
            "5\n",
            "15\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "13\n",
            "[ 9 12 15  0 10  5 15 13  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 10  5 15 13  0  5  0  0 10 13  0 15  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "12\n",
            "5\n",
            "16\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "5\n",
            "[ 9 12 15  0 12  5 16  5  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 12  5 16  5  0  5  5  0 12  0  0 16  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "9\n",
            "1\n",
            "18\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "18\n",
            "[ 9 12 15  0  9  1 18 18  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  9  1 18 18  0  1  0  0  9  0  0 18 18  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "6\n",
            "8\n",
            "15\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "2\n",
            "[ 9 12 15  0  6  8 15  2  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  6  8 15  2  0  6  2  0  8  0  0 15  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "11\n",
            "1\n",
            "15\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "17\n",
            "[ 9 12 15  0 11  1 15 17  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 11  1 15 17  0  1  0  0 11  0  0 15 17  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "11\n",
            "18\n",
            "5\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "8\n",
            "[ 9 12 15  0 11 18  5  8  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 11 18  5  8  0  5  0  0 11  8  0 18  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "11\n",
            "2\n",
            "16\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "19\n",
            "[ 9 12 15  0 11  2 16 19  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 11  2 16 19  0  2  0  0 11  0  0 16 19  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "4\n",
            "19\n",
            "11\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "10\n",
            "[ 9 12 15  0  4 19 11 10  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  4 19 11 10  0  4  0  0 11 10  0 19  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "8\n",
            "19\n",
            "6\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "7\n",
            "[ 9 12 15  0  8 19  6  7  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  8 19  6  7  0  6  0  0  8  7  0 19  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "11\n",
            "3\n",
            "16\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "15\n",
            "[ 9 12 15  0 11  3 16 15  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 11  3 16 15  0  3  0  0 11  0  0 16 15  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "16\n",
            "4\n",
            "9\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "8\n",
            "[ 9 12 15  0 16  4  9  8  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 16  4  9  8  0  4  0  0  9  8  0 16  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "16\n",
            "3\n",
            "11\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "9\n",
            "[ 9 12 15  0 16  3 11  9  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 16  3 11  9  0  3  0  0 11  9  0 16  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "17\n",
            "2\n",
            "8\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "2\n",
            "[ 9 12 15  0 17  2  8  2  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 17  2  8  2  0  2  2  0  8  0  0 17  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "15\n",
            "3\n",
            "8\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "18\n",
            "[ 9 12 15  0 15  3  8 18  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 15  3  8 18  0  3  0  0  8  0  0 15 18  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "1\n",
            "7\n",
            "17\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "11\n",
            "[ 9 12 15  0  1  7 17 11  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  1  7 17 11  0  1  0  0  7 11  0 17  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "16\n",
            "11\n",
            "5\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "13\n",
            "[ 9 12 15  0 16 11  5 13  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 16 11  5 13  0  5  0  0 11 13  0 16  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "2\n",
            "7\n",
            "16\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "5\n",
            "[ 9 12 15  0  2  7 16  5  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  2  7 16  5  0  2  5  0  7  0  0 16  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "12\n",
            "5\n",
            "17\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "1\n",
            "[ 9 12 15  0 12  5 17  1  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 12  5 17  1  0  5  1  0 12  0  0 17  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "18\n",
            "10\n",
            "6\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "3\n",
            "[ 9 12 15  0 18 10  6  3  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 18 10  6  3  0  6  3  0 10  0  0 18  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "17\n",
            "10\n",
            "4\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "6\n",
            "[ 9 12 15  0 17 10  4  6  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 17 10  4  6  0  4  6  0 10  0  0 17  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "11\n",
            "15\n",
            "4\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "19\n",
            "[ 9 12 15  0 11 15  4 19  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 11 15  4 19  0  4  0  0 11  0  0 15 19  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "9\n",
            "1\n",
            "16\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "1\n",
            "[ 9 12 15  0  9  1 16  1  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  9  1 16  1  0  1  1  0  9  0  0 16  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "3\n",
            "9\n",
            "14\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "4\n",
            "[ 9 12 15  0  3  9 14  4  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  3  9 14  4  0  3  4  0  9  0  0 14  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "16\n",
            "5\n",
            "13\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "17\n",
            "[ 9 12 15  0 16  5 13 17  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 16  5 13 17  0  5  0  0 13  0  0 16 17  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "19\n",
            "5\n",
            "11\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "19\n",
            "[ 9 12 15  0 19  5 11 19  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 19  5 11 19  0  5  0  0 11  0  0 19 19  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "14\n",
            "3\n",
            "10\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "5\n",
            "[ 9 12 15  0 14  3 10  5  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 14  3 10  5  0  3  5  0 10  0  0 14  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "6\n",
            "18\n",
            "13\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "15\n",
            "[ 9 12 15  0  6 18 13 15  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  6 18 13 15  0  6  0  0 13  0  0 18 15  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "1\n",
            "18\n",
            "8\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "16\n",
            "[ 9 12 15  0  1 18  8 16  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  1 18  8 16  0  1  0  0  8  0  0 18 16  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "2\n",
            "8\n",
            "19\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "14\n",
            "[ 9 12 15  0  2  8 19 14  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  2  8 19 14  0  2  0  0  8  0  0 19 14  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "11\n",
            "6\n",
            "14\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "11\n",
            "[ 9 12 15  0 11  6 14 11  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 11  6 14 11  0  6  0  0 11 11  0 14  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "4\n",
            "19\n",
            "10\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "5\n",
            "[ 9 12 15  0  4 19 10  5  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  4 19 10  5  0  4  5  0 10  0  0 19  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "1\n",
            "10\n",
            "17\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "19\n",
            "[ 9 12 15  0  1 10 17 19  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  1 10 17 19  0  1  0  0 10  0  0 17 19  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "1\n",
            "17\n",
            "12\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "2\n",
            "[ 9 12 15  0  1 17 12  2  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  1 17 12  2  0  1  2  0 12  0  0 17  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "3\n",
            "13\n",
            "18\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "8\n",
            "[ 9 12 15  0  3 13 18  8  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  3 13 18  8  0  3  0  0 13  8  0 18  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "3\n",
            "17\n",
            "12\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "12\n",
            "[ 9 12 15  0  3 17 12 12  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  3 17 12 12  0  3  0  0 12 12  0 17  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "4\n",
            "18\n",
            "8\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "6\n",
            "[ 9 12 15  0  4 18  8  6  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  4 18  8  6  0  4  6  0  8  0  0 18  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "16\n",
            "3\n",
            "8\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "14\n",
            "[ 9 12 15  0 16  3  8 14  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 16  3  8 14  0  3  0  0  8  0  0 16 14  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "16\n",
            "11\n",
            "3\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "13\n",
            "[ 9 12 15  0 16 11  3 13  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 16 11  3 13  0  3  0  0 11 13  0 16  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "3\n",
            "15\n",
            "8\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "18\n",
            "[ 9 12 15  0  3 15  8 18  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  3 15  8 18  0  3  0  0  8  0  0 15 18  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "10\n",
            "19\n",
            "1\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "13\n",
            "[ 9 12 15  0 10 19  1 13  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 10 19  1 13  0  1  0  0 10 13  0 19  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "17\n",
            "3\n",
            "12\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "14\n",
            "[ 9 12 15  0 17  3 12 14  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 17  3 12 14  0  3  0  0 12  0  0 17 14  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "18\n",
            "1\n",
            "10\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "15\n",
            "[ 9 12 15  0 18  1 10 15  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 18  1 10 15  0  1  0  0 10  0  0 18 15  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "6\n",
            "14\n",
            "7\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "9\n",
            "[ 9 12 15  0  6 14  7  9  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  6 14  7  9  0  6  0  0  7  9  0 14  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "2\n",
            "7\n",
            "14\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "19\n",
            "[ 9 12 15  0  2  7 14 19  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  2  7 14 19  0  2  0  0  7  0  0 14 19  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "15\n",
            "8\n",
            "2\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "10\n",
            "[ 9 12 15  0 15  8  2 10  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 15  8  2 10  0  2  0  0  8 10  0 15  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "7\n",
            "1\n",
            "19\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "10\n",
            "[ 9 12 15  0  7  1 19 10  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  7  1 19 10  0  1  0  0  7 10  0 19  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "8\n",
            "16\n",
            "6\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "6\n",
            "[ 9 12 15  0  8 16  6  6  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  8 16  6  6  0  6  6  0  8  0  0 16  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "1\n",
            "18\n",
            "13\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "1\n",
            "[ 9 12 15  0  1 18 13  1  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  1 18 13  1  0  1  1  0 13  0  0 18  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "13\n",
            "3\n",
            "17\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "11\n",
            "[ 9 12 15  0 13  3 17 11  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 13  3 17 11  0  3  0  0 13 11  0 17  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "14\n",
            "3\n",
            "12\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "19\n",
            "[ 9 12 15  0 14  3 12 19  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 14  3 12 19  0  3  0  0 12  0  0 14 19  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "9\n",
            "15\n",
            "1\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "19\n",
            "[ 9 12 15  0  9 15  1 19  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  9 15  1 19  0  1  0  0  9  0  0 15 19  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "15\n",
            "8\n",
            "5\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "11\n",
            "[ 9 12 15  0 15  8  5 11  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 15  8  5 11  0  5  0  0  8 11  0 15  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "4\n",
            "14\n",
            "7\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "16\n",
            "[ 9 12 15  0  4 14  7 16  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  4 14  7 16  0  4  0  0  7  0  0 14 16  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "4\n",
            "9\n",
            "19\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "11\n",
            "[ 9 12 15  0  4  9 19 11  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  4  9 19 11  0  4  0  0  9 11  0 19  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "3\n",
            "7\n",
            "19\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "5\n",
            "[ 9 12 15  0  3  7 19  5  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  3  7 19  5  0  3  5  0  7  0  0 19  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "1\n",
            "15\n",
            "8\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "13\n",
            "[ 9 12 15  0  1 15  8 13  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  1 15  8 13  0  1  0  0  8 13  0 15  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "3\n",
            "9\n",
            "19\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "5\n",
            "[ 9 12 15  0  3  9 19  5  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  3  9 19  5  0  3  5  0  9  0  0 19  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "6\n",
            "17\n",
            "12\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "3\n",
            "[ 9 12 15  0  6 17 12  3  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  6 17 12  3  0  6  3  0 12  0  0 17  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "19\n",
            "5\n",
            "7\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "3\n",
            "[ 9 12 15  0 19  5  7  3  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 19  5  7  3  0  5  3  0  7  0  0 19  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "18\n",
            "8\n",
            "5\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "11\n",
            "[ 9 12 15  0 18  8  5 11  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 18  8  5 11  0  5  0  0  8 11  0 18  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "9\n",
            "14\n",
            "4\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "4\n",
            "[ 9 12 15  0  9 14  4  4  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  9 14  4  4  0  4  4  0  9  0  0 14  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "18\n",
            "1\n",
            "7\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "9\n",
            "[ 9 12 15  0 18  1  7  9  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 18  1  7  9  0  1  0  0  7  9  0 18  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "10\n",
            "5\n",
            "16\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "15\n",
            "[ 9 12 15  0 10  5 16 15  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 10  5 16 15  0  5  0  0 10  0  0 16 15  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "11\n",
            "1\n",
            "19\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "16\n",
            "[ 9 12 15  0 11  1 19 16  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 11  1 19 16  0  1  0  0 11  0  0 19 16  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "16\n",
            "10\n",
            "6\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "2\n",
            "[ 9 12 15  0 16 10  6  2  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 16 10  6  2  0  6  2  0 10  0  0 16  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "4\n",
            "12\n",
            "17\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "10\n",
            "[ 9 12 15  0  4 12 17 10  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  4 12 17 10  0  4  0  0 12 10  0 17  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "9\n",
            "16\n",
            "4\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "12\n",
            "[ 9 12 15  0  9 16  4 12  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  9 16  4 12  0  4  0  0  9 12  0 16  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "15\n",
            "5\n",
            "10\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "5\n",
            "[ 9 12 15  0 15  5 10  5  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 15  5 10  5  0  5  5  0 10  0  0 15  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "6\n",
            "10\n",
            "15\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "7\n",
            "[ 9 12 15  0  6 10 15  7  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  6 10 15  7  0  6  0  0 10  7  0 15  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "6\n",
            "14\n",
            "10\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "11\n",
            "[ 9 12 15  0  6 14 10 11  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  6 14 10 11  0  6  0  0 10 11  0 14  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "10\n",
            "14\n",
            "3\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "15\n",
            "[ 9 12 15  0 10 14  3 15  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 10 14  3 15  0  3  0  0 10  0  0 14 15  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "8\n",
            "2\n",
            "15\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "14\n",
            "[ 9 12 15  0  8  2 15 14  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  8  2 15 14  0  2  0  0  8  0  0 15 14  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "7\n",
            "17\n",
            "4\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "11\n",
            "[ 9 12 15  0  7 17  4 11  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  7 17  4 11  0  4  0  0  7 11  0 17  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "13\n",
            "17\n",
            "3\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "9\n",
            "[ 9 12 15  0 13 17  3  9  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 13 17  3  9  0  3  0  0 13  9  0 17  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "13\n",
            "14\n",
            "2\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "4\n",
            "[ 9 12 15  0 13 14  2  4  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 13 14  2  4  0  2  4  0 13  0  0 14  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "15\n",
            "11\n",
            "4\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "9\n",
            "[ 9 12 15  0 15 11  4  9  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 15 11  4  9  0  4  0  0 11  9  0 15  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "2\n",
            "10\n",
            "19\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "13\n",
            "[ 9 12 15  0  2 10 19 13  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  2 10 19 13  0  2  0  0 10 13  0 19  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "9\n",
            "1\n",
            "18\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "5\n",
            "[ 9 12 15  0  9  1 18  5  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  9  1 18  5  0  1  5  0  9  0  0 18  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "12\n",
            "2\n",
            "16\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "18\n",
            "[ 9 12 15  0 12  2 16 18  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 12  2 16 18  0  2  0  0 12  0  0 16 18  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "12\n",
            "1\n",
            "16\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "1\n",
            "[ 9 12 15  0 12  1 16  1  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 12  1 16  1  0  1  1  0 12  0  0 16  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "19\n",
            "4\n",
            "11\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "3\n",
            "[ 9 12 15  0 19  4 11  3  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 19  4 11  3  0  4  3  0 11  0  0 19  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "12\n",
            "15\n",
            "5\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "3\n",
            "[ 9 12 15  0 12 15  5  3  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 12 15  5  3  0  5  3  0 12  0  0 15  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "11\n",
            "4\n",
            "16\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "7\n",
            "[ 9 12 15  0 11  4 16  7  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 11  4 16  7  0  4  0  0 11  7  0 16  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "15\n",
            "11\n",
            "6\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "6\n",
            "[ 9 12 15  0 15 11  6  6  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 15 11  6  6  0  6  6  0 11  0  0 15  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "8\n",
            "2\n",
            "15\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "16\n",
            "[ 9 12 15  0  8  2 15 16  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  8  2 15 16  0  2  0  0  8  0  0 15 16  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "11\n",
            "14\n",
            "4\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "11\n",
            "[ 9 12 15  0 11 14  4 11  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 11 14  4 11  0  4  0  0 11 11  0 14  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "8\n",
            "17\n",
            "3\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "3\n",
            "[ 9 12 15  0  8 17  3  3  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  8 17  3  3  0  3  3  0  8  0  0 17  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "14\n",
            "6\n",
            "10\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "19\n",
            "[ 9 12 15  0 14  6 10 19  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 14  6 10 19  0  6  0  0 10  0  0 14 19  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "16\n",
            "6\n",
            "12\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "14\n",
            "[ 9 12 15  0 16  6 12 14  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 16  6 12 14  0  6  0  0 12  0  0 16 14  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "11\n",
            "16\n",
            "2\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "9\n",
            "[ 9 12 15  0 11 16  2  9  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 11 16  2  9  0  2  0  0 11  9  0 16  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "1\n",
            "11\n",
            "16\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "15\n",
            "[ 9 12 15  0  1 11 16 15  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  1 11 16 15  0  1  0  0 11  0  0 16 15  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "6\n",
            "9\n",
            "17\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "3\n",
            "[ 9 12 15  0  6  9 17  3  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  6  9 17  3  0  6  3  0  9  0  0 17  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "1\n",
            "18\n",
            "12\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "7\n",
            "[ 9 12 15  0  1 18 12  7  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  1 18 12  7  0  1  0  0 12  7  0 18  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "10\n",
            "19\n",
            "6\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "18\n",
            "[ 9 12 15  0 10 19  6 18  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 10 19  6 18  0  6  0  0 10  0  0 19 18  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "3\n",
            "11\n",
            "14\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "7\n",
            "[ 9 12 15  0  3 11 14  7  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  3 11 14  7  0  3  0  0 11  7  0 14  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "18\n",
            "7\n",
            "4\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "8\n",
            "[ 9 12 15  0 18  7  4  8  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 18  7  4  8  0  4  0  0  7  8  0 18  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "5\n",
            "14\n",
            "8\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "13\n",
            "[ 9 12 15  0  5 14  8 13  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  5 14  8 13  0  5  0  0  8 13  0 14  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "4\n",
            "15\n",
            "9\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "3\n",
            "[ 9 12 15  0  4 15  9  3  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  4 15  9  3  0  4  3  0  9  0  0 15  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "10\n",
            "16\n",
            "3\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "15\n",
            "[ 9 12 15  0 10 16  3 15  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 10 16  3 15  0  3  0  0 10  0  0 16 15  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "9\n",
            "19\n",
            "4\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "15\n",
            "[ 9 12 15  0  9 19  4 15  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  9 19  4 15  0  4  0  0  9  0  0 19 15  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "5\n",
            "14\n",
            "10\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "18\n",
            "[ 9 12 15  0  5 14 10 18  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  5 14 10 18  0  5  0  0 10  0  0 14 18  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "6\n",
            "14\n",
            "12\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "4\n",
            "[ 9 12 15  0  6 14 12  4  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  6 14 12  4  0  6  4  0 12  0  0 14  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "14\n",
            "2\n",
            "10\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "14\n",
            "[ 9 12 15  0 14  2 10 14  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 14  2 10 14  0  2  0  0 10  0  0 14 14  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "3\n",
            "15\n",
            "10\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "2\n",
            "[ 9 12 15  0  3 15 10  2  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  3 15 10  2  0  3  2  0 10  0  0 15  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "11\n",
            "6\n",
            "17\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "17\n",
            "[ 9 12 15  0 11  6 17 17  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 11  6 17 17  0  6  0  0 11  0  0 17 17  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "14\n",
            "8\n",
            "4\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "16\n",
            "[ 9 12 15  0 14  8  4 16  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 14  8  4 16  0  4  0  0  8  0  0 14 16  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "13\n",
            "6\n",
            "14\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "12\n",
            "[ 9 12 15  0 13  6 14 12  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 13  6 14 12  0  6  0  0 13 12  0 14  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "13\n",
            "16\n",
            "5\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "5\n",
            "[ 9 12 15  0 13 16  5  5  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 13 16  5  5  0  5  5  0 13  0  0 16  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "1\n",
            "15\n",
            "11\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "7\n",
            "[ 9 12 15  0  1 15 11  7  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  1 15 11  7  0  1  0  0 11  7  0 15  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "4\n",
            "19\n",
            "11\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "1\n",
            "[ 9 12 15  0  4 19 11  1  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  4 19 11  1  0  4  1  0 11  0  0 19  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "17\n",
            "2\n",
            "9\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "9\n",
            "[ 9 12 15  0 17  2  9  9  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 17  2  9  9  0  2  0  0  9  9  0 17  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "1\n",
            "11\n",
            "14\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "4\n",
            "[ 9 12 15  0  1 11 14  4  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  1 11 14  4  0  1  4  0 11  0  0 14  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "17\n",
            "5\n",
            "12\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "12\n",
            "[ 9 12 15  0 17  5 12 12  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 17  5 12 12  0  5  0  0 12 12  0 17  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "10\n",
            "3\n",
            "15\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "2\n",
            "[ 9 12 15  0 10  3 15  2  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 10  3 15  2  0  3  2  0 10  0  0 15  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "1\n",
            "19\n",
            "13\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "3\n",
            "[ 9 12 15  0  1 19 13  3  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  1 19 13  3  0  1  3  0 13  0  0 19  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "1\n",
            "17\n",
            "7\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "6\n",
            "[ 9 12 15  0  1 17  7  6  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  1 17  7  6  0  1  6  0  7  0  0 17  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "1\n",
            "15\n",
            "10\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "6\n",
            "[ 9 12 15  0  1 15 10  6  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  1 15 10  6  0  1  6  0 10  0  0 15  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "7\n",
            "17\n",
            "3\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "16\n",
            "[ 9 12 15  0  7 17  3 16  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  7 17  3 16  0  3  0  0  7  0  0 17 16  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "19\n",
            "11\n",
            "1\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "6\n",
            "[ 9 12 15  0 19 11  1  6  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 19 11  1  6  0  1  6  0 11  0  0 19  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "5\n",
            "16\n",
            "11\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "8\n",
            "[ 9 12 15  0  5 16 11  8  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  5 16 11  8  0  5  0  0 11  8  0 16  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "11\n",
            "17\n",
            "4\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "4\n",
            "[ 9 12 15  0 11 17  4  4  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 11 17  4  4  0  4  4  0 11  0  0 17  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "14\n",
            "5\n",
            "10\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "1\n",
            "[ 9 12 15  0 14  5 10  1  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 14  5 10  1  0  5  1  0 10  0  0 14  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "3\n",
            "7\n",
            "18\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "14\n",
            "[ 9 12 15  0  3  7 18 14  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  3  7 18 14  0  3  0  0  7  0  0 18 14  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "18\n",
            "5\n",
            "8\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "8\n",
            "[ 9 12 15  0 18  5  8  8  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 18  5  8  8  0  5  0  0  8  8  0 18  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "11\n",
            "17\n",
            "6\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "5\n",
            "[ 9 12 15  0 11 17  6  5  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 11 17  6  5  0  6  5  0 11  0  0 17  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "8\n",
            "5\n",
            "18\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "4\n",
            "[ 9 12 15  0  8  5 18  4  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  8  5 18  4  0  5  4  0  8  0  0 18  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "7\n",
            "5\n",
            "18\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "6\n",
            "[ 9 12 15  0  7  5 18  6  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  7  5 18  6  0  5  6  0  7  0  0 18  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "18\n",
            "13\n",
            "5\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "16\n",
            "[ 9 12 15  0 18 13  5 16  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 18 13  5 16  0  5  0  0 13  0  0 18 16  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "6\n",
            "11\n",
            "16\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "1\n",
            "[ 9 12 15  0  6 11 16  1  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  6 11 16  1  0  6  1  0 11  0  0 16  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "15\n",
            "6\n",
            "9\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "12\n",
            "[ 9 12 15  0 15  6  9 12  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 15  6  9 12  0  6  0  0  9 12  0 15  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "8\n",
            "2\n",
            "19\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "6\n",
            "[ 9 12 15  0  8  2 19  6  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  8  2 19  6  0  2  6  0  8  0  0 19  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "14\n",
            "7\n",
            "3\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "18\n",
            "[ 9 12 15  0 14  7  3 18  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 14  7  3 18  0  3  0  0  7  0  0 14 18  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "8\n",
            "15\n",
            "4\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "2\n",
            "[ 9 12 15  0  8 15  4  2  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  8 15  4  2  0  4  2  0  8  0  0 15  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "11\n",
            "3\n",
            "14\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "2\n",
            "[ 9 12 15  0 11  3 14  2  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 11  3 14  2  0  3  2  0 11  0  0 14  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "5\n",
            "10\n",
            "15\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "6\n",
            "[ 9 12 15  0  5 10 15  6  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  5 10 15  6  0  5  6  0 10  0  0 15  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "8\n",
            "16\n",
            "4\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "8\n",
            "[ 9 12 15  0  8 16  4  8  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  8 16  4  8  0  4  0  0  8  8  0 16  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "13\n",
            "14\n",
            "1\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "19\n",
            "[ 9 12 15  0 13 14  1 19  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 13 14  1 19  0  1  0  0 13  0  0 14 19  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "10\n",
            "19\n",
            "3\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "2\n",
            "[ 9 12 15  0 10 19  3  2  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 10 19  3  2  0  3  2  0 10  0  0 19  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "18\n",
            "4\n",
            "10\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "19\n",
            "[ 9 12 15  0 18  4 10 19  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 18  4 10 19  0  4  0  0 10  0  0 18 19  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "17\n",
            "4\n",
            "12\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "4\n",
            "[ 9 12 15  0 17  4 12  4  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 17  4 12  4  0  4  4  0 12  0  0 17  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "17\n",
            "6\n",
            "11\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "3\n",
            "[ 9 12 15  0 17  6 11  3  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 17  6 11  3  0  6  3  0 11  0  0 17  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "18\n",
            "5\n",
            "8\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "12\n",
            "[ 9 12 15  0 18  5  8 12  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 18  5  8 12  0  5  0  0  8 12  0 18  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "11\n",
            "3\n",
            "15\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "18\n",
            "[ 9 12 15  0 11  3 15 18  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 11  3 15 18  0  3  0  0 11  0  0 15 18  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "5\n",
            "16\n",
            "10\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "3\n",
            "[ 9 12 15  0  5 16 10  3  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  5 16 10  3  0  5  3  0 10  0  0 16  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "7\n",
            "15\n",
            "5\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "11\n",
            "[ 9 12 15  0  7 15  5 11  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  7 15  5 11  0  5  0  0  7 11  0 15  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "12\n",
            "14\n",
            "2\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "15\n",
            "[ 9 12 15  0 12 14  2 15  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 12 14  2 15  0  2  0  0 12  0  0 14 15  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "6\n",
            "12\n",
            "18\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "11\n",
            "[ 9 12 15  0  6 12 18 11  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  6 12 18 11  0  6  0  0 12 11  0 18  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "6\n",
            "12\n",
            "15\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "18\n",
            "[ 9 12 15  0  6 12 15 18  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  6 12 15 18  0  6  0  0 12  0  0 15 18  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "17\n",
            "4\n",
            "12\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "16\n",
            "[ 9 12 15  0 17  4 12 16  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 17  4 12 16  0  4  0  0 12  0  0 17 16  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "6\n",
            "14\n",
            "8\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "10\n",
            "[ 9 12 15  0  6 14  8 10  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  6 14  8 10  0  6  0  0  8 10  0 14  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "8\n",
            "1\n",
            "16\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "7\n",
            "[ 9 12 15  0  8  1 16  7  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  8  1 16  7  0  1  0  0  8  7  0 16  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "7\n",
            "3\n",
            "16\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "16\n",
            "[ 9 12 15  0  7  3 16 16  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  7  3 16 16  0  3  0  0  7  0  0 16 16  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "1\n",
            "19\n",
            "13\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "9\n",
            "[ 9 12 15  0  1 19 13  9  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  1 19 13  9  0  1  0  0 13  9  0 19  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "3\n",
            "19\n",
            "7\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "5\n",
            "[ 9 12 15  0  3 19  7  5  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  3 19  7  5  0  3  5  0  7  0  0 19  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "2\n",
            "17\n",
            "12\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "17\n",
            "[ 9 12 15  0  2 17 12 17  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  2 17 12 17  0  2  0  0 12  0  0 17 17  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "12\n",
            "3\n",
            "14\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "7\n",
            "[ 9 12 15  0 12  3 14  7  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 12  3 14  7  0  3  0  0 12  7  0 14  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "8\n",
            "14\n",
            "4\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "3\n",
            "[ 9 12 15  0  8 14  4  3  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  8 14  4  3  0  4  3  0  8  0  0 14  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "13\n",
            "5\n",
            "19\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "5\n",
            "[ 9 12 15  0 13  5 19  5  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 13  5 19  5  0  5  5  0 13  0  0 19  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "19\n",
            "9\n",
            "6\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "8\n",
            "[ 9 12 15  0 19  9  6  8  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 19  9  6  8  0  6  0  0  9  8  0 19  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "14\n",
            "12\n",
            "5\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "1\n",
            "[ 9 12 15  0 14 12  5  1  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 14 12  5  1  0  5  1  0 12  0  0 14  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "2\n",
            "7\n",
            "16\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "13\n",
            "[ 9 12 15  0  2  7 16 13  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  2  7 16 13  0  2  0  0  7 13  0 16  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "15\n",
            "10\n",
            "2\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "4\n",
            "[ 9 12 15  0 15 10  2  4  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 15 10  2  4  0  2  4  0 10  0  0 15  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "7\n",
            "4\n",
            "16\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "3\n",
            "[ 9 12 15  0  7  4 16  3  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  7  4 16  3  0  4  3  0  7  0  0 16  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[2 0 1]\n",
            "15\n",
            "3\n",
            "13\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "13\n",
            "[ 9 12 15  0 15  3 13 13  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 15  3 13 13  0  3  0  0 13 13  0 15  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 2 1]\n",
            "2\n",
            "14\n",
            "9\n",
            "\n",
            "index:\n",
            "[2 1 0]\n",
            "17\n",
            "[ 9 12 15  0  2 14  9 17  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  2 14  9 17  0  2  0  0  9  0  0 14 17  0  0  0]\n",
            "\n",
            "index:\n",
            "[1 2 0]\n",
            "10\n",
            "15\n",
            "4\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "13\n",
            "[ 9 12 15  0 10 15  4 13  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0 10 15  4 13  0  4  0  0 10 13  0 15  0  0  0  0]\n",
            "\n",
            "index:\n",
            "[0 1 2]\n",
            "1\n",
            "10\n",
            "16\n",
            "\n",
            "index:\n",
            "[1 0 2]\n",
            "11\n",
            "[ 9 12 15  0  1 10 16 11  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            "[ 9 12 15  0  1 10 16 11  0  1  0  0 10 11  0 16  0  0  0  0]\n",
            "800\n",
            "(20, 20)\n",
            "(20, 20)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "ESmNV7HR1wj1",
        "colab_type": "text"
      },
      "source": [
        "### Dataset 6"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "cv9IDi-C1tf0",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "def repeat(size, inputSet, targetSet):\n",
        "  inputRepeatSet = []\n",
        "  targetRepeatSet = []\n",
        "  \n",
        "  for _ in range(size):\n",
        "    arrIndex = np.arange(len(inputSet))\n",
        "    np.random.shuffle(arrIndex)\n",
        "    \n",
        "    for j in arrIndex:\n",
        "      inputRepeatSet.append(inputSet[j])\n",
        "      targetRepeatSet.append(targetSet[j])\n",
        "\n",
        "  return inputRepeatSet, targetRepeatSet\n",
        "\n",
        "def dataset6(repeats, M):\n",
        "  inputSet = []\n",
        "  targetSet = []\n",
        "  \n",
        "  for _ in range(repeats):\n",
        "    ## [20x20] -> 3 Pointers -> 6 Input Elements -> 2 element in each bucket -> 3,  1,6,  1,2, 1,2, 1,2\n",
        "    # Initalize the memory\n",
        "    inputMemory = np.zeros((M,M), dtype=\"float32\")\n",
        "    targetMemory = np.zeros((M,M), dtype=\"float32\")\n",
        "\n",
        "    # Pointer to the buckets\n",
        "    p1 = 11\n",
        "    p2 = 14\n",
        "    p3 = 17\n",
        "\n",
        "    # Bucket 1\n",
        "    inputMemory[0] = registerModule(p1, M)\n",
        "    targetMemory[0] = registerModule(p1, M)\n",
        "\n",
        "    # Bucket 2\n",
        "    inputMemory[1] = registerModule(p2, M)\n",
        "    targetMemory[1] = registerModule(p2, M)\n",
        "\n",
        "    # Bucket 3\n",
        "    inputMemory[2] = registerModule(p3, M)\n",
        "    targetMemory[2] = registerModule(p3, M)\n",
        "\n",
        "    # We can give sorted values into the function bucketSort\n",
        "    _, bucketsArray = bucketSort(np.arange(1, 20), 20)\n",
        "    # print(bucketsArray)\n",
        "\n",
        "\n",
        "    print(\"\\nindex:\")\n",
        "    randIndex = np.arange(3)\n",
        "    np.random.shuffle(randIndex)\n",
        "    print(randIndex)\n",
        "\n",
        "    for i in range(3):\n",
        "      # memory index starts from 4 to 6 -> 4,5,6,7\n",
        "      bucketIndex = randIndex[i]\n",
        "      val = np.random.choice(bucketsArray[bucketIndex], 1)[0]\n",
        "      print(val)\n",
        "      inputMemory[i+4] = registerModule(val, M)\n",
        "      targetMemory[i+4] = registerModule(val, M)\n",
        "\n",
        "      if bucketIndex == 0:\n",
        "        #bucket1\n",
        "        targetMemory[p1] = registerModule(val, M)\n",
        "      elif bucketIndex == 1:\n",
        "        #bucket2\n",
        "        targetMemory[p2] = registerModule(val, M)    \n",
        "      elif bucketIndex == 2:\n",
        "        #bucket3\n",
        "        targetMemory[p3] = registerModule(val, M)\n",
        "\n",
        "    # 4th element\n",
        "    print(\"\\nindex:\")\n",
        "    randIndex = np.arange(3)\n",
        "    np.random.shuffle(randIndex)\n",
        "    print(randIndex)\n",
        "\n",
        "    for i in range(3):\n",
        "      # memory index starts from 4 to 6 -> 4,5,6,7\n",
        "      bucketIndex = randIndex[i]\n",
        "      val = np.random.choice(bucketsArray[bucketIndex], 1)[0]\n",
        "      print(val)\n",
        "      inputMemory[i+7] = registerModule(val, M)\n",
        "      targetMemory[i+7] = registerModule(val, M)\n",
        "\n",
        "      if bucketIndex == 0:\n",
        "        #bucket1\n",
        "        targetMemory[p1+1] = registerModule(val, M)\n",
        "      elif bucketIndex == 1:\n",
        "        #bucket2\n",
        "        targetMemory[p2+1] = registerModule(val, M)    \n",
        "      elif bucketIndex == 2:\n",
        "        #bucket3\n",
        "        targetMemory[p3+1] = registerModule(val, M)\n",
        "\n",
        "    print(np.argmax(inputMemory, axis=1))\n",
        "    print(np.argmax(targetMemory, axis=1))\n",
        "\n",
        "    inputSet.append(inputMemory)\n",
        "    targetSet.append(targetMemory)\n",
        "  return inputSet, targetSet"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "lschFQKT6kq2",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# dataSelected = 6"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "EAk1RaXf6gQN",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "if dataSelected == 6:\n",
        "  ## Generate dataset\n",
        "  inputSet, targetSet = dataset6(50, M)\n",
        "  inputTrainSet, targetTrainSet = repeat(40, inputSet[0:40], targetSet[0:40])\n",
        "\n",
        "  print(len(inputTrainSet))\n",
        "  print(inputSet[0].shape)\n",
        "  print(targetSet[0].shape)\n",
        "\n",
        "  inputTestSet, targetTestSet = inputSet[40:50], targetSet[40:50]\n",
        "\n",
        "  # for (input_data, target_data) in zip(inputSet, targetSet):\n",
        "  #   print(input_data)\n",
        "  #   print(target_data)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "colab_type": "text",
        "id": "D_9C5PA4RLA8"
      },
      "source": [
        "## Train Model"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "k76s7cjDNoB4",
        "colab_type": "text"
      },
      "source": [
        "### Training Function"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "lUGX_C7Juetl",
        "colab": {}
      },
      "source": [
        "optimizer = tf.train.AdamOptimizer(\n",
        "    learning_rate=0.001,\n",
        "    beta1=0.9,\n",
        "    beta2=0.999,\n",
        "    epsilon=1e-08\n",
        ")"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "q8eUsG9T3WBO",
        "colab": {}
      },
      "source": [
        "def trainModel(inputSet, targetSet, gates, nramController, readMemory, writeMemory, epochs=2, max_timesteps=5, min_error=1e-5):\n",
        "  loss_history = []\n",
        "\n",
        "  for epoch in range(epochs):\n",
        "      batch = 0\n",
        "      for (input_data, target_data) in zip(inputSet, targetSet):\n",
        "          cost = 0\n",
        "          total_prob_complete = 0\n",
        "          prob_incomplete = 1\n",
        "          \n",
        "          with tf.GradientTape() as tape:\n",
        "            registers = tf.concat([registerValueModule(0), registerValueModule(0), registerValueModule(0)], axis=0)\n",
        "            registers = tf.reshape(registers, shape=[1, registers.shape[0], registers.shape[1]])\n",
        "            memory = input_data\n",
        "           \n",
        "            for timestep in range(max_timesteps):\n",
        "              register_data = []\n",
        "              registersNumpy = registers.numpy()\n",
        "              for i in range(registersNumpy.shape[1]):\n",
        "                register_data.append(registersNumpy[0][i][0])\n",
        "              register_data = np.array([register_data], dtype=\"float32\")\n",
        "              \n",
        "              gateControllerOutput, registerControllerOutput, status = nramController(register_data)\n",
        "              output_registers, memory = runTimeStepNN(gates, registers, memory, readMemory, writeMemory, gateControllerOutput, registerControllerOutput)\n",
        "              \n",
        "              registers = output_registers\n",
        "              \n",
        "              # update the probability when its the last time step\n",
        "              if timestep == max_timesteps:\n",
        "                  prob_complete = 1 - total_prob_complete\n",
        "              else:\n",
        "                  prob_complete = status * prob_incomplete\n",
        "\n",
        "              # Accumulate the probability that a result has been produced.\n",
        "              total_prob_complete += prob_complete\n",
        "\n",
        "              # update the probability for incomplete status\n",
        "              prob_incomplete *= 1 - status\n",
        "\n",
        "              # Cost for this timestep.\n",
        "              cost_current_timestep_registers = 0\n",
        "\n",
        "              # Add epsilon to every log to avoid having inf in costs.\n",
        "              epsilon = tf.constant(1e-35, shape=[output_registers.shape[0], output_registers.shape[2]])\n",
        "\n",
        "              for r in range(memory.shape[0]):\n",
        "                  y = target_data[r]\n",
        "                  yPredicted = memory[r]\n",
        "                  loss = tf.math.reduce_sum((y * tf.math.log(tf.math.add(yPredicted, epsilon))), axis=1)\n",
        "                  cost_current_timestep_registers += tf.reshape(loss, [loss.shape[0], 1])\n",
        "\n",
        "              cost_now = prob_complete * cost_current_timestep_registers\n",
        "              cost -= cost_now\n",
        "            final_cost = tf.math.reduce_sum(cost)\n",
        "          \n",
        "          # store the loss history \n",
        "          loss_history.append(final_cost.numpy())\n",
        "\n",
        "          # fetch the trainable variables\n",
        "          variables = nramController.variables\n",
        "\n",
        "          # calculate the gradient\n",
        "          grads = tape.gradient(final_cost, variables)\n",
        "\n",
        "          # update the weights of the network\n",
        "          optimizer.apply_gradients(zip(grads, variables), global_step=tf.train.get_or_create_global_step())\n",
        "        \n",
        "          batch += 1\n",
        "          print(\"\\nEpoch:\", epoch, \" batch:\", batch, \" final cost:\", final_cost.numpy())\n",
        "          \n",
        "          if final_cost.numpy() < min_error:\n",
        "            print(\"\\nExit: Epoch:\", epoch, \" batch:\", batch, \" final cost:\", final_cost.numpy())\n",
        "            return loss_history, nramController\n",
        "          \n",
        "  return loss_history, nramController"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "colab_type": "text",
        "id": "1MZlsNaiRRiZ"
      },
      "source": [
        "### Run the model"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "colab_type": "text",
        "id": "pTOo1kur3AMO"
      },
      "source": [
        "##### LSTM Model"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "VF01xnaTWuVq",
        "outputId": "76eb9fa9-9257-41c0-8fd2-d587781155ac",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 54417
        }
      },
      "source": [
        "## Run me\n",
        "gateDimensions = [[3], [4], [5], [6], [7], [8,8], [9,9], [10], [11,11], [12,12], [13,13], [14,14], [15,15], [16,16]]\n",
        "registerDimensions = [17, 17, 17]\n",
        "\n",
        "nramController = NRAMControllerLSTM(hidden_dimensions=256, gateControllerDimensions=gateDimensions, registerControllerDimensions=registerDimensions)\n",
        "\n",
        "gates = [zeroModule, oneModule, twoModule, incModule, addModule, subModule, decModule, lessModule, lessOrEqualThenModule, equalModule, minimumModule, maximumModule]\n",
        "\n",
        "loss_history, model = trainModel(inputTrainSet, targetTrainSet, gates, nramController, readMemory, writeMemory, min_error=1e-5)\n"
      ],
      "execution_count": 124,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Epoch: 0  batch: 1  final cost: 184.66348\n",
            "\n",
            "Epoch: 0  batch: 2  final cost: 169.5034\n",
            "\n",
            "Epoch: 0  batch: 3  final cost: 168.98502\n",
            "\n",
            "Epoch: 0  batch: 4  final cost: 147.75818\n",
            "\n",
            "Epoch: 0  batch: 5  final cost: 73.7142\n",
            "\n",
            "Epoch: 0  batch: 6  final cost: 53.118397\n",
            "\n",
            "Epoch: 0  batch: 7  final cost: 36.51227\n",
            "\n",
            "Epoch: 0  batch: 8  final cost: 47.26234\n",
            "\n",
            "Epoch: 0  batch: 9  final cost: 32.38112\n",
            "\n",
            "Epoch: 0  batch: 10  final cost: 26.960358\n",
            "\n",
            "Epoch: 0  batch: 11  final cost: 9.111071\n",
            "\n",
            "Epoch: 0  batch: 12  final cost: 24.160513\n",
            "\n",
            "Epoch: 0  batch: 13  final cost: 4.0272775\n",
            "\n",
            "Epoch: 0  batch: 14  final cost: 7.2524247\n",
            "\n",
            "Epoch: 0  batch: 15  final cost: 1.9853479\n",
            "\n",
            "Epoch: 0  batch: 16  final cost: 1.4382572\n",
            "\n",
            "Epoch: 0  batch: 17  final cost: 2.29822\n",
            "\n",
            "Epoch: 0  batch: 18  final cost: 1.7802283\n",
            "\n",
            "Epoch: 0  batch: 19  final cost: 1.2534322\n",
            "\n",
            "Epoch: 0  batch: 20  final cost: 0.6874942\n",
            "\n",
            "Epoch: 0  batch: 21  final cost: 0.24124357\n",
            "\n",
            "Epoch: 0  batch: 22  final cost: 0.44582963\n",
            "\n",
            "Epoch: 0  batch: 23  final cost: 0.2529522\n",
            "\n",
            "Epoch: 0  batch: 24  final cost: 0.107571155\n",
            "\n",
            "Epoch: 0  batch: 25  final cost: 0.2280885\n",
            "\n",
            "Epoch: 0  batch: 26  final cost: 0.08320916\n",
            "\n",
            "Epoch: 0  batch: 27  final cost: 0.09313906\n",
            "\n",
            "Epoch: 0  batch: 28  final cost: 0.052341208\n",
            "\n",
            "Epoch: 0  batch: 29  final cost: 0.08874317\n",
            "\n",
            "Epoch: 0  batch: 30  final cost: 0.025849547\n",
            "\n",
            "Epoch: 0  batch: 31  final cost: 0.041931283\n",
            "\n",
            "Epoch: 0  batch: 32  final cost: 0.053040847\n",
            "\n",
            "Epoch: 0  batch: 33  final cost: 0.016140852\n",
            "\n",
            "Epoch: 0  batch: 34  final cost: 0.013776964\n",
            "\n",
            "Epoch: 0  batch: 35  final cost: 0.01341797\n",
            "\n",
            "Epoch: 0  batch: 36  final cost: 0.027626999\n",
            "\n",
            "Epoch: 0  batch: 37  final cost: 0.018668778\n",
            "\n",
            "Epoch: 0  batch: 38  final cost: 0.0092590805\n",
            "\n",
            "Epoch: 0  batch: 39  final cost: 0.013499773\n",
            "\n",
            "Epoch: 0  batch: 40  final cost: 0.010694552\n",
            "\n",
            "Epoch: 0  batch: 41  final cost: 0.011562994\n",
            "\n",
            "Epoch: 0  batch: 42  final cost: 0.0075392453\n",
            "\n",
            "Epoch: 0  batch: 43  final cost: 0.0067780227\n",
            "\n",
            "Epoch: 0  batch: 44  final cost: 0.0060486495\n",
            "\n",
            "Epoch: 0  batch: 45  final cost: 0.011805361\n",
            "\n",
            "Epoch: 0  batch: 46  final cost: 0.009101653\n",
            "\n",
            "Epoch: 0  batch: 47  final cost: 0.0052753305\n",
            "\n",
            "Epoch: 0  batch: 48  final cost: 0.014395567\n",
            "\n",
            "Epoch: 0  batch: 49  final cost: 0.010117217\n",
            "\n",
            "Epoch: 0  batch: 50  final cost: 0.009405483\n",
            "\n",
            "Epoch: 0  batch: 51  final cost: 0.0052727154\n",
            "\n",
            "Epoch: 0  batch: 52  final cost: 0.010202304\n",
            "\n",
            "Epoch: 0  batch: 53  final cost: 0.004992067\n",
            "\n",
            "Epoch: 0  batch: 54  final cost: 0.010812052\n",
            "\n",
            "Epoch: 0  batch: 55  final cost: 0.012094798\n",
            "\n",
            "Epoch: 0  batch: 56  final cost: 0.0074545867\n",
            "\n",
            "Epoch: 0  batch: 57  final cost: 0.010684966\n",
            "\n",
            "Epoch: 0  batch: 58  final cost: 0.004025596\n",
            "\n",
            "Epoch: 0  batch: 59  final cost: 0.003596856\n",
            "\n",
            "Epoch: 0  batch: 60  final cost: 0.004639619\n",
            "\n",
            "Epoch: 0  batch: 61  final cost: 0.005383305\n",
            "\n",
            "Epoch: 0  batch: 62  final cost: 0.0034026776\n",
            "\n",
            "Epoch: 0  batch: 63  final cost: 0.007343346\n",
            "\n",
            "Epoch: 0  batch: 64  final cost: 0.006535676\n",
            "\n",
            "Epoch: 0  batch: 65  final cost: 0.004345456\n",
            "\n",
            "Epoch: 0  batch: 66  final cost: 0.0034497166\n",
            "\n",
            "Epoch: 0  batch: 67  final cost: 0.003898018\n",
            "\n",
            "Epoch: 0  batch: 68  final cost: 0.0040943804\n",
            "\n",
            "Epoch: 0  batch: 69  final cost: 0.010259076\n",
            "\n",
            "Epoch: 0  batch: 70  final cost: 0.0070727793\n",
            "\n",
            "Epoch: 0  batch: 71  final cost: 0.003822871\n",
            "\n",
            "Epoch: 0  batch: 72  final cost: 0.0034652513\n",
            "\n",
            "Epoch: 0  batch: 73  final cost: 0.0096967695\n",
            "\n",
            "Epoch: 0  batch: 74  final cost: 0.0070395805\n",
            "\n",
            "Epoch: 0  batch: 75  final cost: 0.00438553\n",
            "\n",
            "Epoch: 0  batch: 76  final cost: 0.0074507627\n",
            "\n",
            "Epoch: 0  batch: 77  final cost: 0.0041924804\n",
            "\n",
            "Epoch: 0  batch: 78  final cost: 0.006196041\n",
            "\n",
            "Epoch: 0  batch: 79  final cost: 0.008630846\n",
            "\n",
            "Epoch: 0  batch: 80  final cost: 0.0067314794\n",
            "\n",
            "Epoch: 0  batch: 81  final cost: 0.004126616\n",
            "\n",
            "Epoch: 0  batch: 82  final cost: 0.0035305466\n",
            "\n",
            "Epoch: 0  batch: 83  final cost: 0.0036202888\n",
            "\n",
            "Epoch: 0  batch: 84  final cost: 0.006653345\n",
            "\n",
            "Epoch: 0  batch: 85  final cost: 0.0054305475\n",
            "\n",
            "Epoch: 0  batch: 86  final cost: 0.00338851\n",
            "\n",
            "Epoch: 0  batch: 87  final cost: 0.009755926\n",
            "\n",
            "Epoch: 0  batch: 88  final cost: 0.0032940556\n",
            "\n",
            "Epoch: 0  batch: 89  final cost: 0.0059995786\n",
            "\n",
            "Epoch: 0  batch: 90  final cost: 0.0065762736\n",
            "\n",
            "Epoch: 0  batch: 91  final cost: 0.003728841\n",
            "\n",
            "Epoch: 0  batch: 92  final cost: 0.0039244746\n",
            "\n",
            "Epoch: 0  batch: 93  final cost: 0.005535734\n",
            "\n",
            "Epoch: 0  batch: 94  final cost: 0.0065968963\n",
            "\n",
            "Epoch: 0  batch: 95  final cost: 0.008794863\n",
            "\n",
            "Epoch: 0  batch: 96  final cost: 0.003786154\n",
            "\n",
            "Epoch: 0  batch: 97  final cost: 0.0040876884\n",
            "\n",
            "Epoch: 0  batch: 98  final cost: 0.0037067425\n",
            "\n",
            "Epoch: 0  batch: 99  final cost: 0.0068517346\n",
            "\n",
            "Epoch: 0  batch: 100  final cost: 0.0028967604\n",
            "\n",
            "Epoch: 0  batch: 101  final cost: 0.0037702462\n",
            "\n",
            "Epoch: 0  batch: 102  final cost: 0.0058871224\n",
            "\n",
            "Epoch: 0  batch: 103  final cost: 0.0030027912\n",
            "\n",
            "Epoch: 0  batch: 104  final cost: 0.0032131874\n",
            "\n",
            "Epoch: 0  batch: 105  final cost: 0.0094291875\n",
            "\n",
            "Epoch: 0  batch: 106  final cost: 0.008906606\n",
            "\n",
            "Epoch: 0  batch: 107  final cost: 0.0034126774\n",
            "\n",
            "Epoch: 0  batch: 108  final cost: 0.0069076177\n",
            "\n",
            "Epoch: 0  batch: 109  final cost: 0.0064560976\n",
            "\n",
            "Epoch: 0  batch: 110  final cost: 0.0044861683\n",
            "\n",
            "Epoch: 0  batch: 111  final cost: 0.00920492\n",
            "\n",
            "Epoch: 0  batch: 112  final cost: 0.0067123817\n",
            "\n",
            "Epoch: 0  batch: 113  final cost: 0.0063185003\n",
            "\n",
            "Epoch: 0  batch: 114  final cost: 0.007990413\n",
            "\n",
            "Epoch: 0  batch: 115  final cost: 0.007868316\n",
            "\n",
            "Epoch: 0  batch: 116  final cost: 0.0028384894\n",
            "\n",
            "Epoch: 0  batch: 117  final cost: 0.005651757\n",
            "\n",
            "Epoch: 0  batch: 118  final cost: 0.0032020542\n",
            "\n",
            "Epoch: 0  batch: 119  final cost: 0.0036145754\n",
            "\n",
            "Epoch: 0  batch: 120  final cost: 0.0072766375\n",
            "\n",
            "Epoch: 0  batch: 121  final cost: 0.006235948\n",
            "\n",
            "Epoch: 0  batch: 122  final cost: 0.0062814835\n",
            "\n",
            "Epoch: 0  batch: 123  final cost: 0.0031489367\n",
            "\n",
            "Epoch: 0  batch: 124  final cost: 0.0061252452\n",
            "\n",
            "Epoch: 0  batch: 125  final cost: 0.005566746\n",
            "\n",
            "Epoch: 0  batch: 126  final cost: 0.0042948816\n",
            "\n",
            "Epoch: 0  batch: 127  final cost: 0.003505981\n",
            "\n",
            "Epoch: 0  batch: 128  final cost: 0.0027510037\n",
            "\n",
            "Epoch: 0  batch: 129  final cost: 0.0064408435\n",
            "\n",
            "Epoch: 0  batch: 130  final cost: 0.0030161124\n",
            "\n",
            "Epoch: 0  batch: 131  final cost: 0.007549383\n",
            "\n",
            "Epoch: 0  batch: 132  final cost: 0.0034880424\n",
            "\n",
            "Epoch: 0  batch: 133  final cost: 0.00551558\n",
            "\n",
            "Epoch: 0  batch: 134  final cost: 0.0033254612\n",
            "\n",
            "Epoch: 0  batch: 135  final cost: 0.003074223\n",
            "\n",
            "Epoch: 0  batch: 136  final cost: 0.0033204623\n",
            "\n",
            "Epoch: 0  batch: 137  final cost: 0.005994563\n",
            "\n",
            "Epoch: 0  batch: 138  final cost: 0.005976906\n",
            "\n",
            "Epoch: 0  batch: 139  final cost: 0.004918516\n",
            "\n",
            "Epoch: 0  batch: 140  final cost: 0.008315454\n",
            "\n",
            "Epoch: 0  batch: 141  final cost: 0.0027081235\n",
            "\n",
            "Epoch: 0  batch: 142  final cost: 0.0050196936\n",
            "\n",
            "Epoch: 0  batch: 143  final cost: 0.0029440033\n",
            "\n",
            "Epoch: 0  batch: 144  final cost: 0.0025858232\n",
            "\n",
            "Epoch: 0  batch: 145  final cost: 0.0030731075\n",
            "\n",
            "Epoch: 0  batch: 146  final cost: 0.0027999089\n",
            "\n",
            "Epoch: 0  batch: 147  final cost: 0.0033369693\n",
            "\n",
            "Epoch: 0  batch: 148  final cost: 0.0086737275\n",
            "\n",
            "Epoch: 0  batch: 149  final cost: 0.0052547236\n",
            "\n",
            "Epoch: 0  batch: 150  final cost: 0.0068458407\n",
            "\n",
            "Epoch: 0  batch: 151  final cost: 0.003522395\n",
            "\n",
            "Epoch: 0  batch: 152  final cost: 0.0029394196\n",
            "\n",
            "Epoch: 0  batch: 153  final cost: 0.0061237114\n",
            "\n",
            "Epoch: 0  batch: 154  final cost: 0.006245962\n",
            "\n",
            "Epoch: 0  batch: 155  final cost: 0.008578557\n",
            "\n",
            "Epoch: 0  batch: 156  final cost: 0.003231907\n",
            "\n",
            "Epoch: 0  batch: 157  final cost: 0.0033351004\n",
            "\n",
            "Epoch: 0  batch: 158  final cost: 0.007827728\n",
            "\n",
            "Epoch: 0  batch: 159  final cost: 0.0071572997\n",
            "\n",
            "Epoch: 0  batch: 160  final cost: 0.008390659\n",
            "\n",
            "Epoch: 0  batch: 161  final cost: 0.0027973484\n",
            "\n",
            "Epoch: 0  batch: 162  final cost: 0.004802311\n",
            "\n",
            "Epoch: 0  batch: 163  final cost: 0.0083837295\n",
            "\n",
            "Epoch: 0  batch: 164  final cost: 0.0024993413\n",
            "\n",
            "Epoch: 0  batch: 165  final cost: 0.0032489812\n",
            "\n",
            "Epoch: 0  batch: 166  final cost: 0.0070045516\n",
            "\n",
            "Epoch: 0  batch: 167  final cost: 0.0030985752\n",
            "\n",
            "Epoch: 0  batch: 168  final cost: 0.007773858\n",
            "\n",
            "Epoch: 0  batch: 169  final cost: 0.0050799525\n",
            "\n",
            "Epoch: 0  batch: 170  final cost: 0.0046246615\n",
            "\n",
            "Epoch: 0  batch: 171  final cost: 0.0037686992\n",
            "\n",
            "Epoch: 0  batch: 172  final cost: 0.0059277993\n",
            "\n",
            "Epoch: 0  batch: 173  final cost: 0.0023869155\n",
            "\n",
            "Epoch: 0  batch: 174  final cost: 0.0055911867\n",
            "\n",
            "Epoch: 0  batch: 175  final cost: 0.008155846\n",
            "\n",
            "Epoch: 0  batch: 176  final cost: 0.0066308198\n",
            "\n",
            "Epoch: 0  batch: 177  final cost: 0.0054158885\n",
            "\n",
            "Epoch: 0  batch: 178  final cost: 0.002847637\n",
            "\n",
            "Epoch: 0  batch: 179  final cost: 0.002399715\n",
            "\n",
            "Epoch: 0  batch: 180  final cost: 0.0053697345\n",
            "\n",
            "Epoch: 0  batch: 181  final cost: 0.005645747\n",
            "\n",
            "Epoch: 0  batch: 182  final cost: 0.0053562\n",
            "\n",
            "Epoch: 0  batch: 183  final cost: 0.0028606346\n",
            "\n",
            "Epoch: 0  batch: 184  final cost: 0.0029821238\n",
            "\n",
            "Epoch: 0  batch: 185  final cost: 0.0025376882\n",
            "\n",
            "Epoch: 0  batch: 186  final cost: 0.002682643\n",
            "\n",
            "Epoch: 0  batch: 187  final cost: 0.0023515336\n",
            "\n",
            "Epoch: 0  batch: 188  final cost: 0.0047649615\n",
            "\n",
            "Epoch: 0  batch: 189  final cost: 0.0030914138\n",
            "\n",
            "Epoch: 0  batch: 190  final cost: 0.0077222055\n",
            "\n",
            "Epoch: 0  batch: 191  final cost: 0.005223474\n",
            "\n",
            "Epoch: 0  batch: 192  final cost: 0.002872264\n",
            "\n",
            "Epoch: 0  batch: 193  final cost: 0.007053769\n",
            "\n",
            "Epoch: 0  batch: 194  final cost: 0.0045252596\n",
            "\n",
            "Epoch: 0  batch: 195  final cost: 0.002561254\n",
            "\n",
            "Epoch: 0  batch: 196  final cost: 0.0025556618\n",
            "\n",
            "Epoch: 0  batch: 197  final cost: 0.0027961673\n",
            "\n",
            "Epoch: 0  batch: 198  final cost: 0.0053054746\n",
            "\n",
            "Epoch: 0  batch: 199  final cost: 0.002451845\n",
            "\n",
            "Epoch: 0  batch: 200  final cost: 0.005902724\n",
            "\n",
            "Epoch: 0  batch: 201  final cost: 0.0049255244\n",
            "\n",
            "Epoch: 0  batch: 202  final cost: 0.0028377513\n",
            "\n",
            "Epoch: 0  batch: 203  final cost: 0.0021677364\n",
            "\n",
            "Epoch: 0  batch: 204  final cost: 0.0051599066\n",
            "\n",
            "Epoch: 0  batch: 205  final cost: 0.005883083\n",
            "\n",
            "Epoch: 0  batch: 206  final cost: 0.005076796\n",
            "\n",
            "Epoch: 0  batch: 207  final cost: 0.002700063\n",
            "\n",
            "Epoch: 0  batch: 208  final cost: 0.002432536\n",
            "\n",
            "Epoch: 0  batch: 209  final cost: 0.00709659\n",
            "\n",
            "Epoch: 0  batch: 210  final cost: 0.0025690151\n",
            "\n",
            "Epoch: 0  batch: 211  final cost: 0.002390857\n",
            "\n",
            "Epoch: 0  batch: 212  final cost: 0.0049681114\n",
            "\n",
            "Epoch: 0  batch: 213  final cost: 0.0020657864\n",
            "\n",
            "Epoch: 0  batch: 214  final cost: 0.0023664867\n",
            "\n",
            "Epoch: 0  batch: 215  final cost: 0.0062378887\n",
            "\n",
            "Epoch: 0  batch: 216  final cost: 0.0024993375\n",
            "\n",
            "Epoch: 0  batch: 217  final cost: 0.0025734704\n",
            "\n",
            "Epoch: 0  batch: 218  final cost: 0.002046286\n",
            "\n",
            "Epoch: 0  batch: 219  final cost: 0.0065135844\n",
            "\n",
            "Epoch: 0  batch: 220  final cost: 0.0052655307\n",
            "\n",
            "Epoch: 0  batch: 221  final cost: 0.004451664\n",
            "\n",
            "Epoch: 0  batch: 222  final cost: 0.003040486\n",
            "\n",
            "Epoch: 0  batch: 223  final cost: 0.002625727\n",
            "\n",
            "Epoch: 0  batch: 224  final cost: 0.0042320224\n",
            "\n",
            "Epoch: 0  batch: 225  final cost: 0.0019328261\n",
            "\n",
            "Epoch: 0  batch: 226  final cost: 0.0035446836\n",
            "\n",
            "Epoch: 0  batch: 227  final cost: 0.003733331\n",
            "\n",
            "Epoch: 0  batch: 228  final cost: 0.0035014306\n",
            "\n",
            "Epoch: 0  batch: 229  final cost: 0.0020254652\n",
            "\n",
            "Epoch: 0  batch: 230  final cost: 0.0036305096\n",
            "\n",
            "Epoch: 0  batch: 231  final cost: 0.0021858225\n",
            "\n",
            "Epoch: 0  batch: 232  final cost: 0.003872756\n",
            "\n",
            "Epoch: 0  batch: 233  final cost: 0.003987452\n",
            "\n",
            "Epoch: 0  batch: 234  final cost: 0.0047909883\n",
            "\n",
            "Epoch: 0  batch: 235  final cost: 0.0022905958\n",
            "\n",
            "Epoch: 0  batch: 236  final cost: 0.0055287937\n",
            "\n",
            "Epoch: 0  batch: 237  final cost: 0.0018413218\n",
            "\n",
            "Epoch: 0  batch: 238  final cost: 0.0019683854\n",
            "\n",
            "Epoch: 0  batch: 239  final cost: 0.00307934\n",
            "\n",
            "Epoch: 0  batch: 240  final cost: 0.004640471\n",
            "\n",
            "Epoch: 0  batch: 241  final cost: 0.0016759159\n",
            "\n",
            "Epoch: 0  batch: 242  final cost: 0.0030101931\n",
            "\n",
            "Epoch: 0  batch: 243  final cost: 0.003520315\n",
            "\n",
            "Epoch: 0  batch: 244  final cost: 0.0022457668\n",
            "\n",
            "Epoch: 0  batch: 245  final cost: 0.0018302873\n",
            "\n",
            "Epoch: 0  batch: 246  final cost: 0.003624693\n",
            "\n",
            "Epoch: 0  batch: 247  final cost: 0.0020169467\n",
            "\n",
            "Epoch: 0  batch: 248  final cost: 0.0020302827\n",
            "\n",
            "Epoch: 0  batch: 249  final cost: 0.0040573594\n",
            "\n",
            "Epoch: 0  batch: 250  final cost: 0.0033596947\n",
            "\n",
            "Epoch: 0  batch: 251  final cost: 0.0033430723\n",
            "\n",
            "Epoch: 0  batch: 252  final cost: 0.001753098\n",
            "\n",
            "Epoch: 0  batch: 253  final cost: 0.0024754924\n",
            "\n",
            "Epoch: 0  batch: 254  final cost: 0.0031705808\n",
            "\n",
            "Epoch: 0  batch: 255  final cost: 0.0017791608\n",
            "\n",
            "Epoch: 0  batch: 256  final cost: 0.004047986\n",
            "\n",
            "Epoch: 0  batch: 257  final cost: 0.0020297198\n",
            "\n",
            "Epoch: 0  batch: 258  final cost: 0.0014233731\n",
            "\n",
            "Epoch: 0  batch: 259  final cost: 0.0040597664\n",
            "\n",
            "Epoch: 0  batch: 260  final cost: 0.0018272576\n",
            "\n",
            "Epoch: 0  batch: 261  final cost: 0.0019701405\n",
            "\n",
            "Epoch: 0  batch: 262  final cost: 0.0015446906\n",
            "\n",
            "Epoch: 0  batch: 263  final cost: 0.002083367\n",
            "\n",
            "Epoch: 0  batch: 264  final cost: 0.003446351\n",
            "\n",
            "Epoch: 0  batch: 265  final cost: 0.0015739822\n",
            "\n",
            "Epoch: 0  batch: 266  final cost: 0.002190752\n",
            "\n",
            "Epoch: 0  batch: 267  final cost: 0.0036213207\n",
            "\n",
            "Epoch: 0  batch: 268  final cost: 0.001515811\n",
            "\n",
            "Epoch: 0  batch: 269  final cost: 0.0024593899\n",
            "\n",
            "Epoch: 0  batch: 270  final cost: 0.0018458113\n",
            "\n",
            "Epoch: 0  batch: 271  final cost: 0.0029516649\n",
            "\n",
            "Epoch: 0  batch: 272  final cost: 0.0019629844\n",
            "\n",
            "Epoch: 0  batch: 273  final cost: 0.0023500335\n",
            "\n",
            "Epoch: 0  batch: 274  final cost: 0.0028489316\n",
            "\n",
            "Epoch: 0  batch: 275  final cost: 0.0018228428\n",
            "\n",
            "Epoch: 0  batch: 276  final cost: 0.0021549147\n",
            "\n",
            "Epoch: 0  batch: 277  final cost: 0.002625706\n",
            "\n",
            "Epoch: 0  batch: 278  final cost: 0.0016201657\n",
            "\n",
            "Epoch: 0  batch: 279  final cost: 0.0024722447\n",
            "\n",
            "Epoch: 0  batch: 280  final cost: 0.0020305372\n",
            "\n",
            "Epoch: 0  batch: 281  final cost: 0.0027354944\n",
            "\n",
            "Epoch: 0  batch: 282  final cost: 0.0023998409\n",
            "\n",
            "Epoch: 0  batch: 283  final cost: 0.002988711\n",
            "\n",
            "Epoch: 0  batch: 284  final cost: 0.0021583012\n",
            "\n",
            "Epoch: 0  batch: 285  final cost: 0.002753862\n",
            "\n",
            "Epoch: 0  batch: 286  final cost: 0.0018770348\n",
            "\n",
            "Epoch: 0  batch: 287  final cost: 0.0020871535\n",
            "\n",
            "Epoch: 0  batch: 288  final cost: 0.0022536004\n",
            "\n",
            "Epoch: 0  batch: 289  final cost: 0.0018142923\n",
            "\n",
            "Epoch: 0  batch: 290  final cost: 0.0024260317\n",
            "\n",
            "Epoch: 0  batch: 291  final cost: 0.001381718\n",
            "\n",
            "Epoch: 0  batch: 292  final cost: 0.0017805601\n",
            "\n",
            "Epoch: 0  batch: 293  final cost: 0.0014056514\n",
            "\n",
            "Epoch: 0  batch: 294  final cost: 0.0023704076\n",
            "\n",
            "Epoch: 0  batch: 295  final cost: 0.0018312386\n",
            "\n",
            "Epoch: 0  batch: 296  final cost: 0.0017654548\n",
            "\n",
            "Epoch: 0  batch: 297  final cost: 0.0015790567\n",
            "\n",
            "Epoch: 0  batch: 298  final cost: 0.001952782\n",
            "\n",
            "Epoch: 0  batch: 299  final cost: 0.0016282309\n",
            "\n",
            "Epoch: 0  batch: 300  final cost: 0.00159145\n",
            "\n",
            "Epoch: 0  batch: 301  final cost: 0.0035681156\n",
            "\n",
            "Epoch: 0  batch: 302  final cost: 0.0020023296\n",
            "\n",
            "Epoch: 0  batch: 303  final cost: 0.001185857\n",
            "\n",
            "Epoch: 0  batch: 304  final cost: 0.0015863082\n",
            "\n",
            "Epoch: 0  batch: 305  final cost: 0.0016084355\n",
            "\n",
            "Epoch: 0  batch: 306  final cost: 0.0029580023\n",
            "\n",
            "Epoch: 0  batch: 307  final cost: 0.0016348461\n",
            "\n",
            "Epoch: 0  batch: 308  final cost: 0.0018356817\n",
            "\n",
            "Epoch: 0  batch: 309  final cost: 0.0012779507\n",
            "\n",
            "Epoch: 0  batch: 310  final cost: 0.0024017184\n",
            "\n",
            "Epoch: 0  batch: 311  final cost: 0.0018063516\n",
            "\n",
            "Epoch: 0  batch: 312  final cost: 0.0013565761\n",
            "\n",
            "Epoch: 0  batch: 313  final cost: 0.0017274738\n",
            "\n",
            "Epoch: 0  batch: 314  final cost: 0.0018502696\n",
            "\n",
            "Epoch: 0  batch: 315  final cost: 0.0023615453\n",
            "\n",
            "Epoch: 0  batch: 316  final cost: 0.0016228004\n",
            "\n",
            "Epoch: 0  batch: 317  final cost: 0.001726402\n",
            "\n",
            "Epoch: 0  batch: 318  final cost: 0.001154914\n",
            "\n",
            "Epoch: 0  batch: 319  final cost: 0.0015988175\n",
            "\n",
            "Epoch: 0  batch: 320  final cost: 0.0017019465\n",
            "\n",
            "Epoch: 0  batch: 321  final cost: 0.0018964407\n",
            "\n",
            "Epoch: 0  batch: 322  final cost: 0.0016981661\n",
            "\n",
            "Epoch: 0  batch: 323  final cost: 0.0015433631\n",
            "\n",
            "Epoch: 0  batch: 324  final cost: 0.0013297853\n",
            "\n",
            "Epoch: 0  batch: 325  final cost: 0.0014203459\n",
            "\n",
            "Epoch: 0  batch: 326  final cost: 0.0016864388\n",
            "\n",
            "Epoch: 0  batch: 327  final cost: 0.0015807825\n",
            "\n",
            "Epoch: 0  batch: 328  final cost: 0.0016841198\n",
            "\n",
            "Epoch: 0  batch: 329  final cost: 0.0017879073\n",
            "\n",
            "Epoch: 0  batch: 330  final cost: 0.0018042705\n",
            "\n",
            "Epoch: 0  batch: 331  final cost: 0.0028335822\n",
            "\n",
            "Epoch: 0  batch: 332  final cost: 0.0016885857\n",
            "\n",
            "Epoch: 0  batch: 333  final cost: 0.0016602714\n",
            "\n",
            "Epoch: 0  batch: 334  final cost: 0.0033621504\n",
            "\n",
            "Epoch: 0  batch: 335  final cost: 0.0012114726\n",
            "\n",
            "Epoch: 0  batch: 336  final cost: 0.0015366307\n",
            "\n",
            "Epoch: 0  batch: 337  final cost: 0.0015705352\n",
            "\n",
            "Epoch: 0  batch: 338  final cost: 0.0014389276\n",
            "\n",
            "Epoch: 0  batch: 339  final cost: 0.0014245443\n",
            "\n",
            "Epoch: 0  batch: 340  final cost: 0.0015522491\n",
            "\n",
            "Epoch: 0  batch: 341  final cost: 0.0014631946\n",
            "\n",
            "Epoch: 0  batch: 342  final cost: 0.0012754914\n",
            "\n",
            "Epoch: 0  batch: 343  final cost: 0.0017733233\n",
            "\n",
            "Epoch: 0  batch: 344  final cost: 0.0017054586\n",
            "\n",
            "Epoch: 0  batch: 345  final cost: 0.0021388624\n",
            "\n",
            "Epoch: 0  batch: 346  final cost: 0.0011034713\n",
            "\n",
            "Epoch: 0  batch: 347  final cost: 0.0015939123\n",
            "\n",
            "Epoch: 0  batch: 348  final cost: 0.0015475624\n",
            "\n",
            "Epoch: 0  batch: 349  final cost: 0.0019806763\n",
            "\n",
            "Epoch: 0  batch: 350  final cost: 0.0018510781\n",
            "\n",
            "Epoch: 0  batch: 351  final cost: 0.0022014766\n",
            "\n",
            "Epoch: 0  batch: 352  final cost: 0.001735666\n",
            "\n",
            "Epoch: 0  batch: 353  final cost: 0.0010848355\n",
            "\n",
            "Epoch: 0  batch: 354  final cost: 0.002145186\n",
            "\n",
            "Epoch: 0  batch: 355  final cost: 0.0016527686\n",
            "\n",
            "Epoch: 0  batch: 356  final cost: 0.0023766884\n",
            "\n",
            "Epoch: 0  batch: 357  final cost: 0.0017602614\n",
            "\n",
            "Epoch: 0  batch: 358  final cost: 0.0021214643\n",
            "\n",
            "Epoch: 0  batch: 359  final cost: 0.0012180218\n",
            "\n",
            "Epoch: 0  batch: 360  final cost: 0.0014450527\n",
            "\n",
            "Epoch: 0  batch: 361  final cost: 0.0015065256\n",
            "\n",
            "Epoch: 0  batch: 362  final cost: 0.0012110507\n",
            "\n",
            "Epoch: 0  batch: 363  final cost: 0.0013631708\n",
            "\n",
            "Epoch: 0  batch: 364  final cost: 0.0014321081\n",
            "\n",
            "Epoch: 0  batch: 365  final cost: 0.001512062\n",
            "\n",
            "Epoch: 0  batch: 366  final cost: 0.0020493374\n",
            "\n",
            "Epoch: 0  batch: 367  final cost: 0.0016127448\n",
            "\n",
            "Epoch: 0  batch: 368  final cost: 0.0015719908\n",
            "\n",
            "Epoch: 0  batch: 369  final cost: 0.0020324509\n",
            "\n",
            "Epoch: 0  batch: 370  final cost: 0.0016887258\n",
            "\n",
            "Epoch: 0  batch: 371  final cost: 0.0018550314\n",
            "\n",
            "Epoch: 0  batch: 372  final cost: 0.0010436748\n",
            "\n",
            "Epoch: 0  batch: 373  final cost: 0.0014357186\n",
            "\n",
            "Epoch: 0  batch: 374  final cost: 0.0014169834\n",
            "\n",
            "Epoch: 0  batch: 375  final cost: 0.0015674634\n",
            "\n",
            "Epoch: 0  batch: 376  final cost: 0.0012984404\n",
            "\n",
            "Epoch: 0  batch: 377  final cost: 0.0016489105\n",
            "\n",
            "Epoch: 0  batch: 378  final cost: 0.0016729652\n",
            "\n",
            "Epoch: 0  batch: 379  final cost: 0.0012836208\n",
            "\n",
            "Epoch: 0  batch: 380  final cost: 0.00106128\n",
            "\n",
            "Epoch: 0  batch: 381  final cost: 0.002025779\n",
            "\n",
            "Epoch: 0  batch: 382  final cost: 0.0014237239\n",
            "\n",
            "Epoch: 0  batch: 383  final cost: 0.001579436\n",
            "\n",
            "Epoch: 0  batch: 384  final cost: 0.0025672081\n",
            "\n",
            "Epoch: 0  batch: 385  final cost: 0.0015693379\n",
            "\n",
            "Epoch: 0  batch: 386  final cost: 0.0030671218\n",
            "\n",
            "Epoch: 0  batch: 387  final cost: 0.0012155632\n",
            "\n",
            "Epoch: 0  batch: 388  final cost: 0.0015172864\n",
            "\n",
            "Epoch: 0  batch: 389  final cost: 0.0015202736\n",
            "\n",
            "Epoch: 0  batch: 390  final cost: 0.001170292\n",
            "\n",
            "Epoch: 0  batch: 391  final cost: 0.0013941535\n",
            "\n",
            "Epoch: 0  batch: 392  final cost: 0.001633835\n",
            "\n",
            "Epoch: 0  batch: 393  final cost: 0.0016004036\n",
            "\n",
            "Epoch: 0  batch: 394  final cost: 0.0015075263\n",
            "\n",
            "Epoch: 0  batch: 395  final cost: 0.0014132627\n",
            "\n",
            "Epoch: 0  batch: 396  final cost: 0.0010068\n",
            "\n",
            "Epoch: 0  batch: 397  final cost: 0.0021601515\n",
            "\n",
            "Epoch: 0  batch: 398  final cost: 0.0013855409\n",
            "\n",
            "Epoch: 0  batch: 399  final cost: 0.0011628572\n",
            "\n",
            "Epoch: 0  batch: 400  final cost: 0.0019201328\n",
            "\n",
            "Epoch: 0  batch: 401  final cost: 0.0012332726\n",
            "\n",
            "Epoch: 0  batch: 402  final cost: 0.0014922626\n",
            "\n",
            "Epoch: 0  batch: 403  final cost: 0.0014991424\n",
            "\n",
            "Epoch: 0  batch: 404  final cost: 0.0015845216\n",
            "\n",
            "Epoch: 0  batch: 405  final cost: 0.0015613232\n",
            "\n",
            "Epoch: 0  batch: 406  final cost: 0.0013565791\n",
            "\n",
            "Epoch: 0  batch: 407  final cost: 0.0021132417\n",
            "\n",
            "Epoch: 0  batch: 408  final cost: 0.0013859586\n",
            "\n",
            "Epoch: 0  batch: 409  final cost: 0.0013753867\n",
            "\n",
            "Epoch: 0  batch: 410  final cost: 0.0013124873\n",
            "\n",
            "Epoch: 0  batch: 411  final cost: 0.001559958\n",
            "\n",
            "Epoch: 0  batch: 412  final cost: 0.0024330963\n",
            "\n",
            "Epoch: 0  batch: 413  final cost: 0.0013338114\n",
            "\n",
            "Epoch: 0  batch: 414  final cost: 0.0012102164\n",
            "\n",
            "Epoch: 0  batch: 415  final cost: 0.0018611423\n",
            "\n",
            "Epoch: 0  batch: 416  final cost: 0.0015268555\n",
            "\n",
            "Epoch: 0  batch: 417  final cost: 0.0011095103\n",
            "\n",
            "Epoch: 0  batch: 418  final cost: 0.0013364325\n",
            "\n",
            "Epoch: 0  batch: 419  final cost: 0.0014224941\n",
            "\n",
            "Epoch: 0  batch: 420  final cost: 0.00095826853\n",
            "\n",
            "Epoch: 0  batch: 421  final cost: 0.0018443676\n",
            "\n",
            "Epoch: 0  batch: 422  final cost: 0.0011142874\n",
            "\n",
            "Epoch: 0  batch: 423  final cost: 0.0010849482\n",
            "\n",
            "Epoch: 0  batch: 424  final cost: 0.0013106252\n",
            "\n",
            "Epoch: 0  batch: 425  final cost: 0.0009409527\n",
            "\n",
            "Epoch: 0  batch: 426  final cost: 0.0011690726\n",
            "\n",
            "Epoch: 0  batch: 427  final cost: 0.0009615432\n",
            "\n",
            "Epoch: 0  batch: 428  final cost: 0.001504936\n",
            "\n",
            "Epoch: 0  batch: 429  final cost: 0.001406471\n",
            "\n",
            "Epoch: 0  batch: 430  final cost: 0.0016510899\n",
            "\n",
            "Epoch: 0  batch: 431  final cost: 0.0013254285\n",
            "\n",
            "Epoch: 0  batch: 432  final cost: 0.0028087953\n",
            "\n",
            "Epoch: 0  batch: 433  final cost: 0.0014185056\n",
            "\n",
            "Epoch: 0  batch: 434  final cost: 0.0017689272\n",
            "\n",
            "Epoch: 0  batch: 435  final cost: 0.0017786452\n",
            "\n",
            "Epoch: 0  batch: 436  final cost: 0.0012425466\n",
            "\n",
            "Epoch: 0  batch: 437  final cost: 0.0014140034\n",
            "\n",
            "Epoch: 0  batch: 438  final cost: 0.0013688849\n",
            "\n",
            "Epoch: 0  batch: 439  final cost: 0.001378101\n",
            "\n",
            "Epoch: 0  batch: 440  final cost: 0.0010681847\n",
            "\n",
            "Epoch: 0  batch: 441  final cost: 0.0010574494\n",
            "\n",
            "Epoch: 0  batch: 442  final cost: 0.0012433997\n",
            "\n",
            "Epoch: 0  batch: 443  final cost: 0.0017405523\n",
            "\n",
            "Epoch: 0  batch: 444  final cost: 0.0014405604\n",
            "\n",
            "Epoch: 0  batch: 445  final cost: 0.0013473262\n",
            "\n",
            "Epoch: 0  batch: 446  final cost: 0.0009402585\n",
            "\n",
            "Epoch: 0  batch: 447  final cost: 0.0014596503\n",
            "\n",
            "Epoch: 0  batch: 448  final cost: 0.0011341909\n",
            "\n",
            "Epoch: 0  batch: 449  final cost: 0.0011244419\n",
            "\n",
            "Epoch: 0  batch: 450  final cost: 0.0012554197\n",
            "\n",
            "Epoch: 0  batch: 451  final cost: 0.0017379812\n",
            "\n",
            "Epoch: 0  batch: 452  final cost: 0.0012850787\n",
            "\n",
            "Epoch: 0  batch: 453  final cost: 0.0013336317\n",
            "\n",
            "Epoch: 0  batch: 454  final cost: 0.0012457592\n",
            "\n",
            "Epoch: 0  batch: 455  final cost: 0.0013889766\n",
            "\n",
            "Epoch: 0  batch: 456  final cost: 0.00087968184\n",
            "\n",
            "Epoch: 0  batch: 457  final cost: 0.0015914536\n",
            "\n",
            "Epoch: 0  batch: 458  final cost: 0.0013579067\n",
            "\n",
            "Epoch: 0  batch: 459  final cost: 0.0012088136\n",
            "\n",
            "Epoch: 0  batch: 460  final cost: 0.0010037881\n",
            "\n",
            "Epoch: 0  batch: 461  final cost: 0.0010560966\n",
            "\n",
            "Epoch: 0  batch: 462  final cost: 0.0013426833\n",
            "\n",
            "Epoch: 0  batch: 463  final cost: 0.0008721594\n",
            "\n",
            "Epoch: 0  batch: 464  final cost: 0.0012815481\n",
            "\n",
            "Epoch: 0  batch: 465  final cost: 0.0012335439\n",
            "\n",
            "Epoch: 0  batch: 466  final cost: 0.0011716514\n",
            "\n",
            "Epoch: 0  batch: 467  final cost: 0.0014593891\n",
            "\n",
            "Epoch: 0  batch: 468  final cost: 0.0016614613\n",
            "\n",
            "Epoch: 0  batch: 469  final cost: 0.0021790008\n",
            "\n",
            "Epoch: 0  batch: 470  final cost: 0.0011129291\n",
            "\n",
            "Epoch: 0  batch: 471  final cost: 0.0011950862\n",
            "\n",
            "Epoch: 0  batch: 472  final cost: 0.0009939754\n",
            "\n",
            "Epoch: 0  batch: 473  final cost: 0.0012748545\n",
            "\n",
            "Epoch: 0  batch: 474  final cost: 0.0013356856\n",
            "\n",
            "Epoch: 0  batch: 475  final cost: 0.0011769194\n",
            "\n",
            "Epoch: 0  batch: 476  final cost: 0.0013443686\n",
            "\n",
            "Epoch: 0  batch: 477  final cost: 0.0016707926\n",
            "\n",
            "Epoch: 0  batch: 478  final cost: 0.0013638396\n",
            "\n",
            "Epoch: 0  batch: 479  final cost: 0.0025420042\n",
            "\n",
            "Epoch: 0  batch: 480  final cost: 0.0018168219\n",
            "\n",
            "Epoch: 0  batch: 481  final cost: 0.0013251571\n",
            "\n",
            "Epoch: 0  batch: 482  final cost: 0.001125856\n",
            "\n",
            "Epoch: 0  batch: 483  final cost: 0.0011658831\n",
            "\n",
            "Epoch: 0  batch: 484  final cost: 0.0016299428\n",
            "\n",
            "Epoch: 0  batch: 485  final cost: 0.0014915955\n",
            "\n",
            "Epoch: 0  batch: 486  final cost: 0.0015980725\n",
            "\n",
            "Epoch: 0  batch: 487  final cost: 0.0011487945\n",
            "\n",
            "Epoch: 0  batch: 488  final cost: 0.0008761023\n",
            "\n",
            "Epoch: 0  batch: 489  final cost: 0.0011264384\n",
            "\n",
            "Epoch: 0  batch: 490  final cost: 0.0009455167\n",
            "\n",
            "Epoch: 0  batch: 491  final cost: 0.0011792703\n",
            "\n",
            "Epoch: 0  batch: 492  final cost: 0.001122138\n",
            "\n",
            "Epoch: 0  batch: 493  final cost: 0.0015525522\n",
            "\n",
            "Epoch: 0  batch: 494  final cost: 0.0011998842\n",
            "\n",
            "Epoch: 0  batch: 495  final cost: 0.0017398427\n",
            "\n",
            "Epoch: 0  batch: 496  final cost: 0.0008002828\n",
            "\n",
            "Epoch: 0  batch: 497  final cost: 0.0015723007\n",
            "\n",
            "Epoch: 0  batch: 498  final cost: 0.001142643\n",
            "\n",
            "Epoch: 0  batch: 499  final cost: 0.0024037596\n",
            "\n",
            "Epoch: 0  batch: 500  final cost: 0.0009177115\n",
            "\n",
            "Epoch: 0  batch: 501  final cost: 0.0010114555\n",
            "\n",
            "Epoch: 0  batch: 502  final cost: 0.0012559922\n",
            "\n",
            "Epoch: 0  batch: 503  final cost: 0.0009507758\n",
            "\n",
            "Epoch: 0  batch: 504  final cost: 0.0013210949\n",
            "\n",
            "Epoch: 0  batch: 505  final cost: 0.0009897286\n",
            "\n",
            "Epoch: 0  batch: 506  final cost: 0.0012109673\n",
            "\n",
            "Epoch: 0  batch: 507  final cost: 0.0012203509\n",
            "\n",
            "Epoch: 0  batch: 508  final cost: 0.0011052879\n",
            "\n",
            "Epoch: 0  batch: 509  final cost: 0.0011300615\n",
            "\n",
            "Epoch: 0  batch: 510  final cost: 0.0012667944\n",
            "\n",
            "Epoch: 0  batch: 511  final cost: 0.001177051\n",
            "\n",
            "Epoch: 0  batch: 512  final cost: 0.0008768422\n",
            "\n",
            "Epoch: 0  batch: 513  final cost: 0.0011413577\n",
            "\n",
            "Epoch: 0  batch: 514  final cost: 0.0009897852\n",
            "\n",
            "Epoch: 0  batch: 515  final cost: 0.0012059245\n",
            "\n",
            "Epoch: 0  batch: 516  final cost: 0.0010778037\n",
            "\n",
            "Epoch: 0  batch: 517  final cost: 0.0019144986\n",
            "\n",
            "Epoch: 0  batch: 518  final cost: 0.00074461906\n",
            "\n",
            "Epoch: 0  batch: 519  final cost: 0.0012076718\n",
            "\n",
            "Epoch: 0  batch: 520  final cost: 0.0011239718\n",
            "\n",
            "Epoch: 0  batch: 521  final cost: 0.0012013364\n",
            "\n",
            "Epoch: 0  batch: 522  final cost: 0.001061464\n",
            "\n",
            "Epoch: 0  batch: 523  final cost: 0.0012463739\n",
            "\n",
            "Epoch: 0  batch: 524  final cost: 0.0010855907\n",
            "\n",
            "Epoch: 0  batch: 525  final cost: 0.0014270846\n",
            "\n",
            "Epoch: 0  batch: 526  final cost: 0.0010418394\n",
            "\n",
            "Epoch: 0  batch: 527  final cost: 0.002223487\n",
            "\n",
            "Epoch: 0  batch: 528  final cost: 0.0011208677\n",
            "\n",
            "Epoch: 0  batch: 529  final cost: 0.0014176593\n",
            "\n",
            "Epoch: 0  batch: 530  final cost: 0.00092190824\n",
            "\n",
            "Epoch: 0  batch: 531  final cost: 0.0007163051\n",
            "\n",
            "Epoch: 0  batch: 532  final cost: 0.00106783\n",
            "\n",
            "Epoch: 0  batch: 533  final cost: 0.00082676794\n",
            "\n",
            "Epoch: 0  batch: 534  final cost: 0.0013704104\n",
            "\n",
            "Epoch: 0  batch: 535  final cost: 0.0009820524\n",
            "\n",
            "Epoch: 0  batch: 536  final cost: 0.0008250669\n",
            "\n",
            "Epoch: 0  batch: 537  final cost: 0.0011070154\n",
            "\n",
            "Epoch: 0  batch: 538  final cost: 0.0018001704\n",
            "\n",
            "Epoch: 0  batch: 539  final cost: 0.00081629434\n",
            "\n",
            "Epoch: 0  batch: 540  final cost: 0.0010614885\n",
            "\n",
            "Epoch: 0  batch: 541  final cost: 0.0010590337\n",
            "\n",
            "Epoch: 0  batch: 542  final cost: 0.0011499695\n",
            "\n",
            "Epoch: 0  batch: 543  final cost: 0.0010988471\n",
            "\n",
            "Epoch: 0  batch: 544  final cost: 0.0007571464\n",
            "\n",
            "Epoch: 0  batch: 545  final cost: 0.0013623147\n",
            "\n",
            "Epoch: 0  batch: 546  final cost: 0.00148905\n",
            "\n",
            "Epoch: 0  batch: 547  final cost: 0.000997458\n",
            "\n",
            "Epoch: 0  batch: 548  final cost: 0.0012546894\n",
            "\n",
            "Epoch: 0  batch: 549  final cost: 0.0006723623\n",
            "\n",
            "Epoch: 0  batch: 550  final cost: 0.0010979156\n",
            "\n",
            "Epoch: 0  batch: 551  final cost: 0.001074248\n",
            "\n",
            "Epoch: 0  batch: 552  final cost: 0.0009942446\n",
            "\n",
            "Epoch: 0  batch: 553  final cost: 0.00094523275\n",
            "\n",
            "Epoch: 0  batch: 554  final cost: 0.0008592288\n",
            "\n",
            "Epoch: 0  batch: 555  final cost: 0.0009393629\n",
            "\n",
            "Epoch: 0  batch: 556  final cost: 0.00088505086\n",
            "\n",
            "Epoch: 0  batch: 557  final cost: 0.00093559694\n",
            "\n",
            "Epoch: 0  batch: 558  final cost: 0.0010667387\n",
            "\n",
            "Epoch: 0  batch: 559  final cost: 0.00090994104\n",
            "\n",
            "Epoch: 0  batch: 560  final cost: 0.0008167357\n",
            "\n",
            "Epoch: 0  batch: 561  final cost: 0.0009634888\n",
            "\n",
            "Epoch: 0  batch: 562  final cost: 0.00076378346\n",
            "\n",
            "Epoch: 0  batch: 563  final cost: 0.001319106\n",
            "\n",
            "Epoch: 0  batch: 564  final cost: 0.00087767164\n",
            "\n",
            "Epoch: 0  batch: 565  final cost: 0.0010462111\n",
            "\n",
            "Epoch: 0  batch: 566  final cost: 0.0014268915\n",
            "\n",
            "Epoch: 0  batch: 567  final cost: 0.0009806821\n",
            "\n",
            "Epoch: 0  batch: 568  final cost: 0.0009783682\n",
            "\n",
            "Epoch: 0  batch: 569  final cost: 0.001039389\n",
            "\n",
            "Epoch: 0  batch: 570  final cost: 0.0008106056\n",
            "\n",
            "Epoch: 0  batch: 571  final cost: 0.0019558063\n",
            "\n",
            "Epoch: 0  batch: 572  final cost: 0.0013026247\n",
            "\n",
            "Epoch: 0  batch: 573  final cost: 0.0008392759\n",
            "\n",
            "Epoch: 0  batch: 574  final cost: 0.000953232\n",
            "\n",
            "Epoch: 0  batch: 575  final cost: 0.0010915251\n",
            "\n",
            "Epoch: 0  batch: 576  final cost: 0.0010492367\n",
            "\n",
            "Epoch: 0  batch: 577  final cost: 0.0008784641\n",
            "\n",
            "Epoch: 0  batch: 578  final cost: 0.0010849375\n",
            "\n",
            "Epoch: 0  batch: 579  final cost: 0.001204782\n",
            "\n",
            "Epoch: 0  batch: 580  final cost: 0.00088719046\n",
            "\n",
            "Epoch: 0  batch: 581  final cost: 0.0008430877\n",
            "\n",
            "Epoch: 0  batch: 582  final cost: 0.00071138\n",
            "\n",
            "Epoch: 0  batch: 583  final cost: 0.0011929206\n",
            "\n",
            "Epoch: 0  batch: 584  final cost: 0.0008395232\n",
            "\n",
            "Epoch: 0  batch: 585  final cost: 0.0006050175\n",
            "\n",
            "Epoch: 0  batch: 586  final cost: 0.0015841525\n",
            "\n",
            "Epoch: 0  batch: 587  final cost: 0.00095651724\n",
            "\n",
            "Epoch: 0  batch: 588  final cost: 0.0008501472\n",
            "\n",
            "Epoch: 0  batch: 589  final cost: 0.0007017266\n",
            "\n",
            "Epoch: 0  batch: 590  final cost: 0.00097262696\n",
            "\n",
            "Epoch: 0  batch: 591  final cost: 0.0011572563\n",
            "\n",
            "Epoch: 0  batch: 592  final cost: 0.0012700097\n",
            "\n",
            "Epoch: 0  batch: 593  final cost: 0.0009415898\n",
            "\n",
            "Epoch: 0  batch: 594  final cost: 0.0008295987\n",
            "\n",
            "Epoch: 0  batch: 595  final cost: 0.00092707435\n",
            "\n",
            "Epoch: 0  batch: 596  final cost: 0.00073036226\n",
            "\n",
            "Epoch: 0  batch: 597  final cost: 0.00057982723\n",
            "\n",
            "Epoch: 0  batch: 598  final cost: 0.0010006555\n",
            "\n",
            "Epoch: 0  batch: 599  final cost: 0.0008359484\n",
            "\n",
            "Epoch: 0  batch: 600  final cost: 0.0009947425\n",
            "\n",
            "Epoch: 0  batch: 601  final cost: 0.0009884135\n",
            "\n",
            "Epoch: 0  batch: 602  final cost: 0.0009055354\n",
            "\n",
            "Epoch: 0  batch: 603  final cost: 0.0007890031\n",
            "\n",
            "Epoch: 0  batch: 604  final cost: 0.001216054\n",
            "\n",
            "Epoch: 0  batch: 605  final cost: 0.00097262\n",
            "\n",
            "Epoch: 0  batch: 606  final cost: 0.0008221085\n",
            "\n",
            "Epoch: 0  batch: 607  final cost: 0.00056081027\n",
            "\n",
            "Epoch: 0  batch: 608  final cost: 0.0009587163\n",
            "\n",
            "Epoch: 0  batch: 609  final cost: 0.0012860167\n",
            "\n",
            "Epoch: 0  batch: 610  final cost: 0.0009078651\n",
            "\n",
            "Epoch: 0  batch: 611  final cost: 0.00078674254\n",
            "\n",
            "Epoch: 0  batch: 612  final cost: 0.00091485004\n",
            "\n",
            "Epoch: 0  batch: 613  final cost: 0.00087819103\n",
            "\n",
            "Epoch: 0  batch: 614  final cost: 0.0017075284\n",
            "\n",
            "Epoch: 0  batch: 615  final cost: 0.0011760992\n",
            "\n",
            "Epoch: 0  batch: 616  final cost: 0.00054478477\n",
            "\n",
            "Epoch: 0  batch: 617  final cost: 0.0007961062\n",
            "\n",
            "Epoch: 0  batch: 618  final cost: 0.00092598936\n",
            "\n",
            "Epoch: 0  batch: 619  final cost: 0.00078131957\n",
            "\n",
            "Epoch: 0  batch: 620  final cost: 0.0010753581\n",
            "\n",
            "Epoch: 0  batch: 621  final cost: 0.00075935345\n",
            "\n",
            "Epoch: 0  batch: 622  final cost: 0.0014054863\n",
            "\n",
            "Epoch: 0  batch: 623  final cost: 0.00062565476\n",
            "\n",
            "Epoch: 0  batch: 624  final cost: 0.0007406029\n",
            "\n",
            "Epoch: 0  batch: 625  final cost: 0.0010362663\n",
            "\n",
            "Epoch: 0  batch: 626  final cost: 0.0007688277\n",
            "\n",
            "Epoch: 0  batch: 627  final cost: 0.00096749136\n",
            "\n",
            "Epoch: 0  batch: 628  final cost: 0.0006137532\n",
            "\n",
            "Epoch: 0  batch: 629  final cost: 0.0007427349\n",
            "\n",
            "Epoch: 0  batch: 630  final cost: 0.00079687\n",
            "\n",
            "Epoch: 0  batch: 631  final cost: 0.0006464125\n",
            "\n",
            "Epoch: 0  batch: 632  final cost: 0.0007184446\n",
            "\n",
            "Epoch: 0  batch: 633  final cost: 0.0008209764\n",
            "\n",
            "Epoch: 0  batch: 634  final cost: 0.0007861267\n",
            "\n",
            "Epoch: 0  batch: 635  final cost: 0.0010041926\n",
            "\n",
            "Epoch: 0  batch: 636  final cost: 0.0009248057\n",
            "\n",
            "Epoch: 0  batch: 637  final cost: 0.0006903037\n",
            "\n",
            "Epoch: 0  batch: 638  final cost: 0.001125291\n",
            "\n",
            "Epoch: 0  batch: 639  final cost: 0.00058587914\n",
            "\n",
            "Epoch: 0  batch: 640  final cost: 0.00077281345\n",
            "\n",
            "Epoch: 0  batch: 641  final cost: 0.00072578277\n",
            "\n",
            "Epoch: 0  batch: 642  final cost: 0.00087657\n",
            "\n",
            "Epoch: 0  batch: 643  final cost: 0.00070992514\n",
            "\n",
            "Epoch: 0  batch: 644  final cost: 0.0015498254\n",
            "\n",
            "Epoch: 0  batch: 645  final cost: 0.0009744793\n",
            "\n",
            "Epoch: 0  batch: 646  final cost: 0.0007581616\n",
            "\n",
            "Epoch: 0  batch: 647  final cost: 0.0009042369\n",
            "\n",
            "Epoch: 0  batch: 648  final cost: 0.0004863973\n",
            "\n",
            "Epoch: 0  batch: 649  final cost: 0.00073308067\n",
            "\n",
            "Epoch: 0  batch: 650  final cost: 0.00057181827\n",
            "\n",
            "Epoch: 0  batch: 651  final cost: 0.00085335906\n",
            "\n",
            "Epoch: 0  batch: 652  final cost: 0.0006848921\n",
            "\n",
            "Epoch: 0  batch: 653  final cost: 0.0009962504\n",
            "\n",
            "Epoch: 0  batch: 654  final cost: 0.0007835105\n",
            "\n",
            "Epoch: 0  batch: 655  final cost: 0.0009194319\n",
            "\n",
            "Epoch: 0  batch: 656  final cost: 0.0006958456\n",
            "\n",
            "Epoch: 0  batch: 657  final cost: 0.00047363338\n",
            "\n",
            "Epoch: 0  batch: 658  final cost: 0.0010327225\n",
            "\n",
            "Epoch: 0  batch: 659  final cost: 0.00068166654\n",
            "\n",
            "Epoch: 0  batch: 660  final cost: 0.0008427445\n",
            "\n",
            "Epoch: 0  batch: 661  final cost: 0.0012434635\n",
            "\n",
            "Epoch: 0  batch: 662  final cost: 0.0010576762\n",
            "\n",
            "Epoch: 0  batch: 663  final cost: 0.00064565276\n",
            "\n",
            "Epoch: 0  batch: 664  final cost: 0.0007135635\n",
            "\n",
            "Epoch: 0  batch: 665  final cost: 0.0008094166\n",
            "\n",
            "Epoch: 0  batch: 666  final cost: 0.0007447189\n",
            "\n",
            "Epoch: 0  batch: 667  final cost: 0.00075661304\n",
            "\n",
            "Epoch: 0  batch: 668  final cost: 0.00067798933\n",
            "\n",
            "Epoch: 0  batch: 669  final cost: 0.00053055113\n",
            "\n",
            "Epoch: 0  batch: 670  final cost: 0.0005326875\n",
            "\n",
            "Epoch: 0  batch: 671  final cost: 0.00076320214\n",
            "\n",
            "Epoch: 0  batch: 672  final cost: 0.0010333892\n",
            "\n",
            "Epoch: 0  batch: 673  final cost: 0.00060080364\n",
            "\n",
            "Epoch: 0  batch: 674  final cost: 0.00069902686\n",
            "\n",
            "Epoch: 0  batch: 675  final cost: 0.0006176215\n",
            "\n",
            "Epoch: 0  batch: 676  final cost: 0.00074575737\n",
            "\n",
            "Epoch: 0  batch: 677  final cost: 0.0006295544\n",
            "\n",
            "Epoch: 0  batch: 678  final cost: 0.0010699823\n",
            "\n",
            "Epoch: 0  batch: 679  final cost: 0.0006539342\n",
            "\n",
            "Epoch: 0  batch: 680  final cost: 0.0009440701\n",
            "\n",
            "Epoch: 0  batch: 681  final cost: 0.0010417842\n",
            "\n",
            "Epoch: 0  batch: 682  final cost: 0.00043550864\n",
            "\n",
            "Epoch: 0  batch: 683  final cost: 0.001372824\n",
            "\n",
            "Epoch: 0  batch: 684  final cost: 0.00071091333\n",
            "\n",
            "Epoch: 0  batch: 685  final cost: 0.00094401994\n",
            "\n",
            "Epoch: 0  batch: 686  final cost: 0.0006486088\n",
            "\n",
            "Epoch: 0  batch: 687  final cost: 0.00073173456\n",
            "\n",
            "Epoch: 0  batch: 688  final cost: 0.0007157242\n",
            "\n",
            "Epoch: 0  batch: 689  final cost: 0.00062995794\n",
            "\n",
            "Epoch: 0  batch: 690  final cost: 0.0006296722\n",
            "\n",
            "Epoch: 0  batch: 691  final cost: 0.0010528374\n",
            "\n",
            "Epoch: 0  batch: 692  final cost: 0.0006020397\n",
            "\n",
            "Epoch: 0  batch: 693  final cost: 0.00066714984\n",
            "\n",
            "Epoch: 0  batch: 694  final cost: 0.0006478646\n",
            "\n",
            "Epoch: 0  batch: 695  final cost: 0.001133641\n",
            "\n",
            "Epoch: 0  batch: 696  final cost: 0.00064961036\n",
            "\n",
            "Epoch: 0  batch: 697  final cost: 0.00048757053\n",
            "\n",
            "Epoch: 0  batch: 698  final cost: 0.00059834437\n",
            "\n",
            "Epoch: 0  batch: 699  final cost: 0.0007358432\n",
            "\n",
            "Epoch: 0  batch: 700  final cost: 0.00041070895\n",
            "\n",
            "Epoch: 0  batch: 701  final cost: 0.0009104294\n",
            "\n",
            "Epoch: 0  batch: 702  final cost: 0.00073125324\n",
            "\n",
            "Epoch: 0  batch: 703  final cost: 0.0005611769\n",
            "\n",
            "Epoch: 0  batch: 704  final cost: 0.00068083534\n",
            "\n",
            "Epoch: 0  batch: 705  final cost: 0.0010228861\n",
            "\n",
            "Epoch: 0  batch: 706  final cost: 0.0005717538\n",
            "\n",
            "Epoch: 0  batch: 707  final cost: 0.00046909036\n",
            "\n",
            "Epoch: 0  batch: 708  final cost: 0.00061879796\n",
            "\n",
            "Epoch: 0  batch: 709  final cost: 0.000565587\n",
            "\n",
            "Epoch: 0  batch: 710  final cost: 0.00091977045\n",
            "\n",
            "Epoch: 0  batch: 711  final cost: 0.0008521702\n",
            "\n",
            "Epoch: 0  batch: 712  final cost: 0.0004625423\n",
            "\n",
            "Epoch: 0  batch: 713  final cost: 0.0008042451\n",
            "\n",
            "Epoch: 0  batch: 714  final cost: 0.0009551482\n",
            "\n",
            "Epoch: 0  batch: 715  final cost: 0.0007628273\n",
            "\n",
            "Epoch: 0  batch: 716  final cost: 0.00062419806\n",
            "\n",
            "Epoch: 0  batch: 717  final cost: 0.00076358474\n",
            "\n",
            "Epoch: 0  batch: 718  final cost: 0.0005630411\n",
            "\n",
            "Epoch: 0  batch: 719  final cost: 0.00073513965\n",
            "\n",
            "Epoch: 0  batch: 720  final cost: 0.00078055065\n",
            "\n",
            "Epoch: 0  batch: 721  final cost: 0.0007818229\n",
            "\n",
            "Epoch: 0  batch: 722  final cost: 0.0009825551\n",
            "\n",
            "Epoch: 0  batch: 723  final cost: 0.0005527299\n",
            "\n",
            "Epoch: 0  batch: 724  final cost: 0.0006597313\n",
            "\n",
            "Epoch: 0  batch: 725  final cost: 0.00060651486\n",
            "\n",
            "Epoch: 0  batch: 726  final cost: 0.00076185225\n",
            "\n",
            "Epoch: 0  batch: 727  final cost: 0.00086811214\n",
            "\n",
            "Epoch: 0  batch: 728  final cost: 0.0005962018\n",
            "\n",
            "Epoch: 0  batch: 729  final cost: 0.0005132102\n",
            "\n",
            "Epoch: 0  batch: 730  final cost: 0.0011896526\n",
            "\n",
            "Epoch: 0  batch: 731  final cost: 0.00067000143\n",
            "\n",
            "Epoch: 0  batch: 732  final cost: 0.0005916277\n",
            "\n",
            "Epoch: 0  batch: 733  final cost: 0.0006258418\n",
            "\n",
            "Epoch: 0  batch: 734  final cost: 0.00057353335\n",
            "\n",
            "Epoch: 0  batch: 735  final cost: 0.0006143577\n",
            "\n",
            "Epoch: 0  batch: 736  final cost: 0.00055258325\n",
            "\n",
            "Epoch: 0  batch: 737  final cost: 0.00066593447\n",
            "\n",
            "Epoch: 0  batch: 738  final cost: 0.00061926217\n",
            "\n",
            "Epoch: 0  batch: 739  final cost: 0.00036000554\n",
            "\n",
            "Epoch: 0  batch: 740  final cost: 0.0008333973\n",
            "\n",
            "Epoch: 0  batch: 741  final cost: 0.00041893902\n",
            "\n",
            "Epoch: 0  batch: 742  final cost: 0.0009840254\n",
            "\n",
            "Epoch: 0  batch: 743  final cost: 0.0008787961\n",
            "\n",
            "Epoch: 0  batch: 744  final cost: 0.0008695859\n",
            "\n",
            "Epoch: 0  batch: 745  final cost: 0.00068148767\n",
            "\n",
            "Epoch: 0  batch: 746  final cost: 0.00069549005\n",
            "\n",
            "Epoch: 0  batch: 747  final cost: 0.00054634246\n",
            "\n",
            "Epoch: 0  batch: 748  final cost: 0.0004956019\n",
            "\n",
            "Epoch: 0  batch: 749  final cost: 0.0007726179\n",
            "\n",
            "Epoch: 0  batch: 750  final cost: 0.0004074276\n",
            "\n",
            "Epoch: 0  batch: 751  final cost: 0.00048922724\n",
            "\n",
            "Epoch: 0  batch: 752  final cost: 0.0004965952\n",
            "\n",
            "Epoch: 0  batch: 753  final cost: 0.0003433486\n",
            "\n",
            "Epoch: 0  batch: 754  final cost: 0.00068359304\n",
            "\n",
            "Epoch: 0  batch: 755  final cost: 0.000397912\n",
            "\n",
            "Epoch: 0  batch: 756  final cost: 0.0005290192\n",
            "\n",
            "Epoch: 0  batch: 757  final cost: 0.00051704986\n",
            "\n",
            "Epoch: 0  batch: 758  final cost: 0.0005246269\n",
            "\n",
            "Epoch: 0  batch: 759  final cost: 0.00089094805\n",
            "\n",
            "Epoch: 0  batch: 760  final cost: 0.00087976886\n",
            "\n",
            "Epoch: 0  batch: 761  final cost: 0.0006722748\n",
            "\n",
            "Epoch: 0  batch: 762  final cost: 0.00079477194\n",
            "\n",
            "Epoch: 0  batch: 763  final cost: 0.0005461079\n",
            "\n",
            "Epoch: 0  batch: 764  final cost: 0.00047118898\n",
            "\n",
            "Epoch: 0  batch: 765  final cost: 0.0007800845\n",
            "\n",
            "Epoch: 0  batch: 766  final cost: 0.00038848355\n",
            "\n",
            "Epoch: 0  batch: 767  final cost: 0.0003842326\n",
            "\n",
            "Epoch: 0  batch: 768  final cost: 0.0008699781\n",
            "\n",
            "Epoch: 0  batch: 769  final cost: 0.00057439046\n",
            "\n",
            "Epoch: 0  batch: 770  final cost: 0.00047377677\n",
            "\n",
            "Epoch: 0  batch: 771  final cost: 0.0007392304\n",
            "\n",
            "Epoch: 0  batch: 772  final cost: 0.00087198743\n",
            "\n",
            "Epoch: 0  batch: 773  final cost: 0.00050816673\n",
            "\n",
            "Epoch: 0  batch: 774  final cost: 0.00044391502\n",
            "\n",
            "Epoch: 0  batch: 775  final cost: 0.00060124154\n",
            "\n",
            "Epoch: 0  batch: 776  final cost: 0.0006446023\n",
            "\n",
            "Epoch: 0  batch: 777  final cost: 0.0003187041\n",
            "\n",
            "Epoch: 0  batch: 778  final cost: 0.00049130013\n",
            "\n",
            "Epoch: 0  batch: 779  final cost: 0.00054278574\n",
            "\n",
            "Epoch: 0  batch: 780  final cost: 0.0005481881\n",
            "\n",
            "Epoch: 0  batch: 781  final cost: 0.000492628\n",
            "\n",
            "Epoch: 0  batch: 782  final cost: 0.0006703585\n",
            "\n",
            "Epoch: 0  batch: 783  final cost: 0.0005989148\n",
            "\n",
            "Epoch: 0  batch: 784  final cost: 0.0006178381\n",
            "\n",
            "Epoch: 0  batch: 785  final cost: 0.0007931594\n",
            "\n",
            "Epoch: 0  batch: 786  final cost: 0.00064017496\n",
            "\n",
            "Epoch: 0  batch: 787  final cost: 0.00044230671\n",
            "\n",
            "Epoch: 0  batch: 788  final cost: 0.0003105922\n",
            "\n",
            "Epoch: 0  batch: 789  final cost: 0.0008261284\n",
            "\n",
            "Epoch: 0  batch: 790  final cost: 0.0005030573\n",
            "\n",
            "Epoch: 0  batch: 791  final cost: 0.0008514707\n",
            "\n",
            "Epoch: 0  batch: 792  final cost: 0.0004774601\n",
            "\n",
            "Epoch: 0  batch: 793  final cost: 0.00054145494\n",
            "\n",
            "Epoch: 0  batch: 794  final cost: 0.00048729498\n",
            "\n",
            "Epoch: 0  batch: 795  final cost: 0.00043556574\n",
            "\n",
            "Epoch: 0  batch: 796  final cost: 0.0004975113\n",
            "\n",
            "Epoch: 0  batch: 797  final cost: 0.00035146903\n",
            "\n",
            "Epoch: 0  batch: 798  final cost: 0.0007750352\n",
            "\n",
            "Epoch: 0  batch: 799  final cost: 0.00097216637\n",
            "\n",
            "Epoch: 0  batch: 800  final cost: 0.0005064609\n",
            "\n",
            "Epoch: 1  batch: 1  final cost: 0.00053071394\n",
            "\n",
            "Epoch: 1  batch: 2  final cost: 0.00077088707\n",
            "\n",
            "Epoch: 1  batch: 3  final cost: 0.00082497636\n",
            "\n",
            "Epoch: 1  batch: 4  final cost: 0.000491192\n",
            "\n",
            "Epoch: 1  batch: 5  final cost: 0.00040425416\n",
            "\n",
            "Epoch: 1  batch: 6  final cost: 0.0004170121\n",
            "\n",
            "Epoch: 1  batch: 7  final cost: 0.0004149677\n",
            "\n",
            "Epoch: 1  batch: 8  final cost: 0.00050217507\n",
            "\n",
            "Epoch: 1  batch: 9  final cost: 0.000603287\n",
            "\n",
            "Epoch: 1  batch: 10  final cost: 0.0004796193\n",
            "\n",
            "Epoch: 1  batch: 11  final cost: 0.0002907591\n",
            "\n",
            "Epoch: 1  batch: 12  final cost: 0.0007031252\n",
            "\n",
            "Epoch: 1  batch: 13  final cost: 0.00045000928\n",
            "\n",
            "Epoch: 1  batch: 14  final cost: 0.0005606693\n",
            "\n",
            "Epoch: 1  batch: 15  final cost: 0.00033577488\n",
            "\n",
            "Epoch: 1  batch: 16  final cost: 0.00050146476\n",
            "\n",
            "Epoch: 1  batch: 17  final cost: 0.00057745987\n",
            "\n",
            "Epoch: 1  batch: 18  final cost: 0.0007827292\n",
            "\n",
            "Epoch: 1  batch: 19  final cost: 0.00074562617\n",
            "\n",
            "Epoch: 1  batch: 20  final cost: 0.00052582467\n",
            "\n",
            "Epoch: 1  batch: 21  final cost: 0.00040873516\n",
            "\n",
            "Epoch: 1  batch: 22  final cost: 0.00071163266\n",
            "\n",
            "Epoch: 1  batch: 23  final cost: 0.00048252038\n",
            "\n",
            "Epoch: 1  batch: 24  final cost: 0.0003989276\n",
            "\n",
            "Epoch: 1  batch: 25  final cost: 0.00070365553\n",
            "\n",
            "Epoch: 1  batch: 26  final cost: 0.00047943\n",
            "\n",
            "Epoch: 1  batch: 27  final cost: 0.0005454036\n",
            "\n",
            "Epoch: 1  batch: 28  final cost: 0.00043982646\n",
            "\n",
            "Epoch: 1  batch: 29  final cost: 0.00078105315\n",
            "\n",
            "Epoch: 1  batch: 30  final cost: 0.00042707718\n",
            "\n",
            "Epoch: 1  batch: 31  final cost: 0.00057991745\n",
            "\n",
            "Epoch: 1  batch: 32  final cost: 0.0007669039\n",
            "\n",
            "Epoch: 1  batch: 33  final cost: 0.00026997644\n",
            "\n",
            "Epoch: 1  batch: 34  final cost: 0.00031481546\n",
            "\n",
            "Epoch: 1  batch: 35  final cost: 0.00042329155\n",
            "\n",
            "Epoch: 1  batch: 36  final cost: 0.000881307\n",
            "\n",
            "Epoch: 1  batch: 37  final cost: 0.000595921\n",
            "\n",
            "Epoch: 1  batch: 38  final cost: 0.00030948504\n",
            "\n",
            "Epoch: 1  batch: 39  final cost: 0.00044416342\n",
            "\n",
            "Epoch: 1  batch: 40  final cost: 0.0005519184\n",
            "\n",
            "Epoch: 1  batch: 41  final cost: 0.00046782236\n",
            "\n",
            "Epoch: 1  batch: 42  final cost: 0.00041399425\n",
            "\n",
            "Epoch: 1  batch: 43  final cost: 0.00044008193\n",
            "\n",
            "Epoch: 1  batch: 44  final cost: 0.00026241143\n",
            "\n",
            "Epoch: 1  batch: 45  final cost: 0.000587517\n",
            "\n",
            "Epoch: 1  batch: 46  final cost: 0.0004511851\n",
            "\n",
            "Epoch: 1  batch: 47  final cost: 0.00037181296\n",
            "\n",
            "Epoch: 1  batch: 48  final cost: 0.0007198241\n",
            "\n",
            "Epoch: 1  batch: 49  final cost: 0.00067563576\n",
            "\n",
            "Epoch: 1  batch: 50  final cost: 0.00073488895\n",
            "\n",
            "Epoch: 1  batch: 51  final cost: 0.00030381884\n",
            "\n",
            "Epoch: 1  batch: 52  final cost: 0.00085255044\n",
            "\n",
            "Epoch: 1  batch: 53  final cost: 0.00040584544\n",
            "\n",
            "Epoch: 1  batch: 54  final cost: 0.0007229424\n",
            "\n",
            "Epoch: 1  batch: 55  final cost: 0.0007442583\n",
            "\n",
            "Epoch: 1  batch: 56  final cost: 0.0004538378\n",
            "\n",
            "Epoch: 1  batch: 57  final cost: 0.00065227435\n",
            "\n",
            "Epoch: 1  batch: 58  final cost: 0.00029384656\n",
            "\n",
            "Epoch: 1  batch: 59  final cost: 0.00041103855\n",
            "\n",
            "Epoch: 1  batch: 60  final cost: 0.00036604717\n",
            "\n",
            "Epoch: 1  batch: 61  final cost: 0.0005252053\n",
            "\n",
            "Epoch: 1  batch: 62  final cost: 0.00039873034\n",
            "\n",
            "Epoch: 1  batch: 63  final cost: 0.00050325895\n",
            "\n",
            "Epoch: 1  batch: 64  final cost: 0.0004798279\n",
            "\n",
            "Epoch: 1  batch: 65  final cost: 0.00043658575\n",
            "\n",
            "Epoch: 1  batch: 66  final cost: 0.00028948457\n",
            "\n",
            "Epoch: 1  batch: 67  final cost: 0.00035262562\n",
            "\n",
            "Epoch: 1  batch: 68  final cost: 0.00044978556\n",
            "\n",
            "Epoch: 1  batch: 69  final cost: 0.00073106796\n",
            "\n",
            "Epoch: 1  batch: 70  final cost: 0.0005039559\n",
            "\n",
            "Epoch: 1  batch: 71  final cost: 0.00024713666\n",
            "\n",
            "Epoch: 1  batch: 72  final cost: 0.00033756183\n",
            "\n",
            "Epoch: 1  batch: 73  final cost: 0.0006521961\n",
            "\n",
            "Epoch: 1  batch: 74  final cost: 0.0005083861\n",
            "\n",
            "Epoch: 1  batch: 75  final cost: 0.00043205707\n",
            "\n",
            "Epoch: 1  batch: 76  final cost: 0.00052620406\n",
            "\n",
            "Epoch: 1  batch: 77  final cost: 0.00034551704\n",
            "\n",
            "Epoch: 1  batch: 78  final cost: 0.00042219373\n",
            "\n",
            "Epoch: 1  batch: 79  final cost: 0.0006542961\n",
            "\n",
            "Epoch: 1  batch: 80  final cost: 0.0005273737\n",
            "\n",
            "Epoch: 1  batch: 81  final cost: 0.0003410808\n",
            "\n",
            "Epoch: 1  batch: 82  final cost: 0.00040063343\n",
            "\n",
            "Epoch: 1  batch: 83  final cost: 0.00023854242\n",
            "\n",
            "Epoch: 1  batch: 84  final cost: 0.0005204712\n",
            "\n",
            "Epoch: 1  batch: 85  final cost: 0.00042739924\n",
            "\n",
            "Epoch: 1  batch: 86  final cost: 0.0002720079\n",
            "\n",
            "Epoch: 1  batch: 87  final cost: 0.0006889822\n",
            "\n",
            "Epoch: 1  batch: 88  final cost: 0.00032238686\n",
            "\n",
            "Epoch: 1  batch: 89  final cost: 0.00040927186\n",
            "\n",
            "Epoch: 1  batch: 90  final cost: 0.00047198337\n",
            "\n",
            "Epoch: 1  batch: 91  final cost: 0.00042481872\n",
            "\n",
            "Epoch: 1  batch: 92  final cost: 0.00033662957\n",
            "\n",
            "Epoch: 1  batch: 93  final cost: 0.00040853338\n",
            "\n",
            "Epoch: 1  batch: 94  final cost: 0.0006573494\n",
            "\n",
            "Epoch: 1  batch: 95  final cost: 0.0006025349\n",
            "\n",
            "Epoch: 1  batch: 96  final cost: 0.0003619533\n",
            "\n",
            "Epoch: 1  batch: 97  final cost: 0.00040895032\n",
            "\n",
            "Epoch: 1  batch: 98  final cost: 0.00026878464\n",
            "\n",
            "Epoch: 1  batch: 99  final cost: 0.00060845696\n",
            "\n",
            "Epoch: 1  batch: 100  final cost: 0.00036626274\n",
            "\n",
            "Epoch: 1  batch: 101  final cost: 0.00036055566\n",
            "\n",
            "Epoch: 1  batch: 102  final cost: 0.00041207127\n",
            "\n",
            "Epoch: 1  batch: 103  final cost: 0.00026534873\n",
            "\n",
            "Epoch: 1  batch: 104  final cost: 0.00022682532\n",
            "\n",
            "Epoch: 1  batch: 105  final cost: 0.00065116497\n",
            "\n",
            "Epoch: 1  batch: 106  final cost: 0.0006143111\n",
            "\n",
            "Epoch: 1  batch: 107  final cost: 0.00032188254\n",
            "\n",
            "Epoch: 1  batch: 108  final cost: 0.0004943961\n",
            "\n",
            "Epoch: 1  batch: 109  final cost: 0.0004729576\n",
            "\n",
            "Epoch: 1  batch: 110  final cost: 0.00047063225\n",
            "\n",
            "Epoch: 1  batch: 111  final cost: 0.0006750669\n",
            "\n",
            "Epoch: 1  batch: 112  final cost: 0.0005137485\n",
            "\n",
            "Epoch: 1  batch: 113  final cost: 0.00046135142\n",
            "\n",
            "Epoch: 1  batch: 114  final cost: 0.00063283194\n",
            "\n",
            "Epoch: 1  batch: 115  final cost: 0.0006666983\n",
            "\n",
            "Epoch: 1  batch: 116  final cost: 0.0004674189\n",
            "\n",
            "Epoch: 1  batch: 117  final cost: 0.00043402688\n",
            "\n",
            "Epoch: 1  batch: 118  final cost: 0.00032967085\n",
            "\n",
            "Epoch: 1  batch: 119  final cost: 0.00042745032\n",
            "\n",
            "Epoch: 1  batch: 120  final cost: 0.0007186404\n",
            "\n",
            "Epoch: 1  batch: 121  final cost: 0.00061653217\n",
            "\n",
            "Epoch: 1  batch: 122  final cost: 0.00045322394\n",
            "\n",
            "Epoch: 1  batch: 123  final cost: 0.0004712166\n",
            "\n",
            "Epoch: 1  batch: 124  final cost: 0.00049783196\n",
            "\n",
            "Epoch: 1  batch: 125  final cost: 0.0004248701\n",
            "\n",
            "Epoch: 1  batch: 126  final cost: 0.00054668495\n",
            "\n",
            "Epoch: 1  batch: 127  final cost: 0.00037196474\n",
            "\n",
            "Epoch: 1  batch: 128  final cost: 0.00045146167\n",
            "\n",
            "Epoch: 1  batch: 129  final cost: 0.00063667004\n",
            "\n",
            "Epoch: 1  batch: 130  final cost: 0.00024110937\n",
            "\n",
            "Epoch: 1  batch: 131  final cost: 0.0006404014\n",
            "\n",
            "Epoch: 1  batch: 132  final cost: 0.00037528836\n",
            "\n",
            "Epoch: 1  batch: 133  final cost: 0.00038198815\n",
            "\n",
            "Epoch: 1  batch: 134  final cost: 0.0004465231\n",
            "\n",
            "Epoch: 1  batch: 135  final cost: 0.0002590087\n",
            "\n",
            "Epoch: 1  batch: 136  final cost: 0.00030079143\n",
            "\n",
            "Epoch: 1  batch: 137  final cost: 0.0004335129\n",
            "\n",
            "Epoch: 1  batch: 138  final cost: 0.00044137967\n",
            "\n",
            "Epoch: 1  batch: 139  final cost: 0.0003854423\n",
            "\n",
            "Epoch: 1  batch: 140  final cost: 0.00057160266\n",
            "\n",
            "Epoch: 1  batch: 141  final cost: 0.00023982652\n",
            "\n",
            "Epoch: 1  batch: 142  final cost: 0.00037309693\n",
            "\n",
            "Epoch: 1  batch: 143  final cost: 0.0002849418\n",
            "\n",
            "Epoch: 1  batch: 144  final cost: 0.00042998232\n",
            "\n",
            "Epoch: 1  batch: 145  final cost: 0.0003139578\n",
            "\n",
            "Epoch: 1  batch: 146  final cost: 0.0002565755\n",
            "\n",
            "Epoch: 1  batch: 147  final cost: 0.00040118516\n",
            "\n",
            "Epoch: 1  batch: 148  final cost: 0.0006222288\n",
            "\n",
            "Epoch: 1  batch: 149  final cost: 0.0004195749\n",
            "\n",
            "Epoch: 1  batch: 150  final cost: 0.0006744928\n",
            "\n",
            "Epoch: 1  batch: 151  final cost: 0.00039841371\n",
            "\n",
            "Epoch: 1  batch: 152  final cost: 0.00030376538\n",
            "\n",
            "Epoch: 1  batch: 153  final cost: 0.0004949205\n",
            "\n",
            "Epoch: 1  batch: 154  final cost: 0.00045295493\n",
            "\n",
            "Epoch: 1  batch: 155  final cost: 0.0005999841\n",
            "\n",
            "Epoch: 1  batch: 156  final cost: 0.00030430316\n",
            "\n",
            "Epoch: 1  batch: 157  final cost: 0.00028428057\n",
            "\n",
            "Epoch: 1  batch: 158  final cost: 0.0005521286\n",
            "\n",
            "Epoch: 1  batch: 159  final cost: 0.0006032995\n",
            "\n",
            "Epoch: 1  batch: 160  final cost: 0.0006154624\n",
            "\n",
            "Epoch: 1  batch: 161  final cost: 0.00024342985\n",
            "\n",
            "Epoch: 1  batch: 162  final cost: 0.0003616324\n",
            "\n",
            "Epoch: 1  batch: 163  final cost: 0.0006048007\n",
            "\n",
            "Epoch: 1  batch: 164  final cost: 0.00041678347\n",
            "\n",
            "Epoch: 1  batch: 165  final cost: 0.00027712336\n",
            "\n",
            "Epoch: 1  batch: 166  final cost: 0.0005904707\n",
            "\n",
            "Epoch: 1  batch: 167  final cost: 0.0002936835\n",
            "\n",
            "Epoch: 1  batch: 168  final cost: 0.0005454454\n",
            "\n",
            "Epoch: 1  batch: 169  final cost: 0.00039255156\n",
            "\n",
            "Epoch: 1  batch: 170  final cost: 0.00036294252\n",
            "\n",
            "Epoch: 1  batch: 171  final cost: 0.000489102\n",
            "\n",
            "Epoch: 1  batch: 172  final cost: 0.0004295393\n",
            "\n",
            "Epoch: 1  batch: 173  final cost: 0.0003990346\n",
            "\n",
            "Epoch: 1  batch: 174  final cost: 0.00040612198\n",
            "\n",
            "Epoch: 1  batch: 175  final cost: 0.0005659197\n",
            "\n",
            "Epoch: 1  batch: 176  final cost: 0.000579658\n",
            "\n",
            "Epoch: 1  batch: 177  final cost: 0.00040442694\n",
            "\n",
            "Epoch: 1  batch: 178  final cost: 0.00026685034\n",
            "\n",
            "Epoch: 1  batch: 179  final cost: 0.00021499846\n",
            "\n",
            "Epoch: 1  batch: 180  final cost: 0.00043971417\n",
            "\n",
            "Epoch: 1  batch: 181  final cost: 0.00045371667\n",
            "\n",
            "Epoch: 1  batch: 182  final cost: 0.00038862458\n",
            "\n",
            "Epoch: 1  batch: 183  final cost: 0.00039120438\n",
            "\n",
            "Epoch: 1  batch: 184  final cost: 0.00032442017\n",
            "\n",
            "Epoch: 1  batch: 185  final cost: 0.00020660923\n",
            "\n",
            "Epoch: 1  batch: 186  final cost: 0.00027827814\n",
            "\n",
            "Epoch: 1  batch: 187  final cost: 0.00022636505\n",
            "\n",
            "Epoch: 1  batch: 188  final cost: 0.00033616633\n",
            "\n",
            "Epoch: 1  batch: 189  final cost: 0.0003551891\n",
            "\n",
            "Epoch: 1  batch: 190  final cost: 0.0005621056\n",
            "\n",
            "Epoch: 1  batch: 191  final cost: 0.0005161366\n",
            "\n",
            "Epoch: 1  batch: 192  final cost: 0.0003119816\n",
            "\n",
            "Epoch: 1  batch: 193  final cost: 0.00050032773\n",
            "\n",
            "Epoch: 1  batch: 194  final cost: 0.00037452855\n",
            "\n",
            "Epoch: 1  batch: 195  final cost: 0.00024697272\n",
            "\n",
            "Epoch: 1  batch: 196  final cost: 0.0002677729\n",
            "\n",
            "Epoch: 1  batch: 197  final cost: 0.00035127503\n",
            "\n",
            "Epoch: 1  batch: 198  final cost: 0.00054860726\n",
            "\n",
            "Epoch: 1  batch: 199  final cost: 0.00039016447\n",
            "\n",
            "Epoch: 1  batch: 200  final cost: 0.0005924386\n",
            "\n",
            "Epoch: 1  batch: 201  final cost: 0.00038734823\n",
            "\n",
            "Epoch: 1  batch: 202  final cost: 0.00025162654\n",
            "\n",
            "Epoch: 1  batch: 203  final cost: 0.00022044801\n",
            "\n",
            "Epoch: 1  batch: 204  final cost: 0.00054771994\n",
            "\n",
            "Epoch: 1  batch: 205  final cost: 0.00055131654\n",
            "\n",
            "Epoch: 1  batch: 206  final cost: 0.0004395796\n",
            "\n",
            "Epoch: 1  batch: 207  final cost: 0.00030546676\n",
            "\n",
            "Epoch: 1  batch: 208  final cost: 0.0002425312\n",
            "\n",
            "Epoch: 1  batch: 209  final cost: 0.0005581743\n",
            "\n",
            "Epoch: 1  batch: 210  final cost: 0.00026435143\n",
            "\n",
            "Epoch: 1  batch: 211  final cost: 0.00026671815\n",
            "\n",
            "Epoch: 1  batch: 212  final cost: 0.00039982362\n",
            "\n",
            "Epoch: 1  batch: 213  final cost: 0.00020169688\n",
            "\n",
            "Epoch: 1  batch: 214  final cost: 0.00024828376\n",
            "\n",
            "Epoch: 1  batch: 215  final cost: 0.0004958265\n",
            "\n",
            "Epoch: 1  batch: 216  final cost: 0.0003704729\n",
            "\n",
            "Epoch: 1  batch: 217  final cost: 0.0003054494\n",
            "\n",
            "Epoch: 1  batch: 218  final cost: 0.0002112101\n",
            "\n",
            "Epoch: 1  batch: 219  final cost: 0.0005296111\n",
            "\n",
            "Epoch: 1  batch: 220  final cost: 0.000573244\n",
            "\n",
            "Epoch: 1  batch: 221  final cost: 0.00049449207\n",
            "\n",
            "Epoch: 1  batch: 222  final cost: 0.00043697451\n",
            "\n",
            "Epoch: 1  batch: 223  final cost: 0.0003357758\n",
            "\n",
            "Epoch: 1  batch: 224  final cost: 0.00041083444\n",
            "\n",
            "Epoch: 1  batch: 225  final cost: 0.00035776003\n",
            "\n",
            "Epoch: 1  batch: 226  final cost: 0.00031859017\n",
            "\n",
            "Epoch: 1  batch: 227  final cost: 0.00035207276\n",
            "\n",
            "Epoch: 1  batch: 228  final cost: 0.0003595474\n",
            "\n",
            "Epoch: 1  batch: 229  final cost: 0.00019073553\n",
            "\n",
            "Epoch: 1  batch: 230  final cost: 0.00032050465\n",
            "\n",
            "Epoch: 1  batch: 231  final cost: 0.00025175477\n",
            "\n",
            "Epoch: 1  batch: 232  final cost: 0.0003632387\n",
            "\n",
            "Epoch: 1  batch: 233  final cost: 0.00036717957\n",
            "\n",
            "Epoch: 1  batch: 234  final cost: 0.00051800854\n",
            "\n",
            "Epoch: 1  batch: 235  final cost: 0.00033062426\n",
            "\n",
            "Epoch: 1  batch: 236  final cost: 0.0005306127\n",
            "\n",
            "Epoch: 1  batch: 237  final cost: 0.00035679206\n",
            "\n",
            "Epoch: 1  batch: 238  final cost: 0.00036454343\n",
            "\n",
            "Epoch: 1  batch: 239  final cost: 0.0003227214\n",
            "\n",
            "Epoch: 1  batch: 240  final cost: 0.0004854343\n",
            "\n",
            "Epoch: 1  batch: 241  final cost: 0.00020249517\n",
            "\n",
            "Epoch: 1  batch: 242  final cost: 0.00031343254\n",
            "\n",
            "Epoch: 1  batch: 243  final cost: 0.00040174028\n",
            "\n",
            "Epoch: 1  batch: 244  final cost: 0.00023284713\n",
            "\n",
            "Epoch: 1  batch: 245  final cost: 0.00018579795\n",
            "\n",
            "Epoch: 1  batch: 246  final cost: 0.0005148022\n",
            "\n",
            "Epoch: 1  batch: 247  final cost: 0.00024516517\n",
            "\n",
            "Epoch: 1  batch: 248  final cost: 0.00024616407\n",
            "\n",
            "Epoch: 1  batch: 249  final cost: 0.0005484004\n",
            "\n",
            "Epoch: 1  batch: 250  final cost: 0.00047399595\n",
            "\n",
            "Epoch: 1  batch: 251  final cost: 0.00041196827\n",
            "\n",
            "Epoch: 1  batch: 252  final cost: 0.0002319596\n",
            "\n",
            "Epoch: 1  batch: 253  final cost: 0.0004160933\n",
            "\n",
            "Epoch: 1  batch: 254  final cost: 0.00035698022\n",
            "\n",
            "Epoch: 1  batch: 255  final cost: 0.00035477016\n",
            "\n",
            "Epoch: 1  batch: 256  final cost: 0.0005051817\n",
            "\n",
            "Epoch: 1  batch: 257  final cost: 0.00034519625\n",
            "\n",
            "Epoch: 1  batch: 258  final cost: 0.00019822882\n",
            "\n",
            "Epoch: 1  batch: 259  final cost: 0.00051443576\n",
            "\n",
            "Epoch: 1  batch: 260  final cost: 0.00021958633\n",
            "\n",
            "Epoch: 1  batch: 261  final cost: 0.0003157246\n",
            "\n",
            "Epoch: 1  batch: 262  final cost: 0.00018394085\n",
            "\n",
            "Epoch: 1  batch: 263  final cost: 0.0003141713\n",
            "\n",
            "Epoch: 1  batch: 264  final cost: 0.0004941388\n",
            "\n",
            "Epoch: 1  batch: 265  final cost: 0.0003397065\n",
            "\n",
            "Epoch: 1  batch: 266  final cost: 0.00031064503\n",
            "\n",
            "Epoch: 1  batch: 267  final cost: 0.00051375676\n",
            "\n",
            "Epoch: 1  batch: 268  final cost: 0.00033368706\n",
            "\n",
            "Epoch: 1  batch: 269  final cost: 0.00034504145\n",
            "\n",
            "Epoch: 1  batch: 270  final cost: 0.00027165355\n",
            "\n",
            "Epoch: 1  batch: 271  final cost: 0.0005006017\n",
            "\n",
            "Epoch: 1  batch: 272  final cost: 0.00034111785\n",
            "\n",
            "Epoch: 1  batch: 273  final cost: 0.0003518915\n",
            "\n",
            "Epoch: 1  batch: 274  final cost: 0.00046309896\n",
            "\n",
            "Epoch: 1  batch: 275  final cost: 0.00027585172\n",
            "\n",
            "Epoch: 1  batch: 276  final cost: 0.00030371142\n",
            "\n",
            "Epoch: 1  batch: 277  final cost: 0.00046786736\n",
            "\n",
            "Epoch: 1  batch: 278  final cost: 0.00023605298\n",
            "\n",
            "Epoch: 1  batch: 279  final cost: 0.00036371383\n",
            "\n",
            "Epoch: 1  batch: 280  final cost: 0.0003311678\n",
            "\n",
            "Epoch: 1  batch: 281  final cost: 0.00049446\n",
            "\n",
            "Epoch: 1  batch: 282  final cost: 0.00036076814\n",
            "\n",
            "Epoch: 1  batch: 283  final cost: 0.00048162357\n",
            "\n",
            "Epoch: 1  batch: 284  final cost: 0.0003364812\n",
            "\n",
            "Epoch: 1  batch: 285  final cost: 0.0004831414\n",
            "\n",
            "Epoch: 1  batch: 286  final cost: 0.00032274966\n",
            "\n",
            "Epoch: 1  batch: 287  final cost: 0.00038664014\n",
            "\n",
            "Epoch: 1  batch: 288  final cost: 0.0003327334\n",
            "\n",
            "Epoch: 1  batch: 289  final cost: 0.00026447076\n",
            "\n",
            "Epoch: 1  batch: 290  final cost: 0.00044352486\n",
            "\n",
            "Epoch: 1  batch: 291  final cost: 0.00017168382\n",
            "\n",
            "Epoch: 1  batch: 292  final cost: 0.00028577415\n",
            "\n",
            "Epoch: 1  batch: 293  final cost: 0.0001663977\n",
            "\n",
            "Epoch: 1  batch: 294  final cost: 0.00046301834\n",
            "\n",
            "Epoch: 1  batch: 295  final cost: 0.00032712118\n",
            "\n",
            "Epoch: 1  batch: 296  final cost: 0.0002065117\n",
            "\n",
            "Epoch: 1  batch: 297  final cost: 0.00027905477\n",
            "\n",
            "Epoch: 1  batch: 298  final cost: 0.00028743874\n",
            "\n",
            "Epoch: 1  batch: 299  final cost: 0.00022028829\n",
            "\n",
            "Epoch: 1  batch: 300  final cost: 0.0002849449\n",
            "\n",
            "Epoch: 1  batch: 301  final cost: 0.0004847267\n",
            "\n",
            "Epoch: 1  batch: 302  final cost: 0.00036625483\n",
            "\n",
            "Epoch: 1  batch: 303  final cost: 0.00017600489\n",
            "\n",
            "Epoch: 1  batch: 304  final cost: 0.00019769807\n",
            "\n",
            "Epoch: 1  batch: 305  final cost: 0.00030498166\n",
            "\n",
            "Epoch: 1  batch: 306  final cost: 0.00041673824\n",
            "\n",
            "Epoch: 1  batch: 307  final cost: 0.00021458989\n",
            "\n",
            "Epoch: 1  batch: 308  final cost: 0.00035288004\n",
            "\n",
            "Epoch: 1  batch: 309  final cost: 0.00031119023\n",
            "\n",
            "Epoch: 1  batch: 310  final cost: 0.00046794664\n",
            "\n",
            "Epoch: 1  batch: 311  final cost: 0.00028404366\n",
            "\n",
            "Epoch: 1  batch: 312  final cost: 0.00020372163\n",
            "\n",
            "Epoch: 1  batch: 313  final cost: 0.0004235427\n",
            "\n",
            "Epoch: 1  batch: 314  final cost: 0.00030834865\n",
            "\n",
            "Epoch: 1  batch: 315  final cost: 0.00046528006\n",
            "\n",
            "Epoch: 1  batch: 316  final cost: 0.00021438343\n",
            "\n",
            "Epoch: 1  batch: 317  final cost: 0.000314296\n",
            "\n",
            "Epoch: 1  batch: 318  final cost: 0.00017592433\n",
            "\n",
            "Epoch: 1  batch: 319  final cost: 0.0003078641\n",
            "\n",
            "Epoch: 1  batch: 320  final cost: 0.00024694361\n",
            "\n",
            "Epoch: 1  batch: 321  final cost: 0.00036573043\n",
            "\n",
            "Epoch: 1  batch: 322  final cost: 0.00025249718\n",
            "\n",
            "Epoch: 1  batch: 323  final cost: 0.00019608653\n",
            "\n",
            "Epoch: 1  batch: 324  final cost: 0.00020324162\n",
            "\n",
            "Epoch: 1  batch: 325  final cost: 0.00027846068\n",
            "\n",
            "Epoch: 1  batch: 326  final cost: 0.00024671596\n",
            "\n",
            "Epoch: 1  batch: 327  final cost: 0.00030817883\n",
            "\n",
            "Epoch: 1  batch: 328  final cost: 0.00043026748\n",
            "\n",
            "Epoch: 1  batch: 329  final cost: 0.00033757178\n",
            "\n",
            "Epoch: 1  batch: 330  final cost: 0.00030860893\n",
            "\n",
            "Epoch: 1  batch: 331  final cost: 0.00041895921\n",
            "\n",
            "Epoch: 1  batch: 332  final cost: 0.00031474826\n",
            "\n",
            "Epoch: 1  batch: 333  final cost: 0.00020185279\n",
            "\n",
            "Epoch: 1  batch: 334  final cost: 0.00048314413\n",
            "\n",
            "Epoch: 1  batch: 335  final cost: 0.00031650672\n",
            "\n",
            "Epoch: 1  batch: 336  final cost: 0.0002162553\n",
            "\n",
            "Epoch: 1  batch: 337  final cost: 0.0002135726\n",
            "\n",
            "Epoch: 1  batch: 338  final cost: 0.00028781936\n",
            "\n",
            "Epoch: 1  batch: 339  final cost: 0.00030979858\n",
            "\n",
            "Epoch: 1  batch: 340  final cost: 0.00021286141\n",
            "\n",
            "Epoch: 1  batch: 341  final cost: 0.00028341473\n",
            "\n",
            "Epoch: 1  batch: 342  final cost: 0.00016101282\n",
            "\n",
            "Epoch: 1  batch: 343  final cost: 0.00028398162\n",
            "\n",
            "Epoch: 1  batch: 344  final cost: 0.00028449326\n",
            "\n",
            "Epoch: 1  batch: 345  final cost: 0.00036621114\n",
            "\n",
            "Epoch: 1  batch: 346  final cost: 0.00017487611\n",
            "\n",
            "Epoch: 1  batch: 347  final cost: 0.00032002787\n",
            "\n",
            "Epoch: 1  batch: 348  final cost: 0.00032701535\n",
            "\n",
            "Epoch: 1  batch: 349  final cost: 0.0004589835\n",
            "\n",
            "Epoch: 1  batch: 350  final cost: 0.00046595736\n",
            "\n",
            "Epoch: 1  batch: 351  final cost: 0.00046248065\n",
            "\n",
            "Epoch: 1  batch: 352  final cost: 0.00043865747\n",
            "\n",
            "Epoch: 1  batch: 353  final cost: 0.0001720074\n",
            "\n",
            "Epoch: 1  batch: 354  final cost: 0.0004711436\n",
            "\n",
            "Epoch: 1  batch: 355  final cost: 0.00035394137\n",
            "\n",
            "Epoch: 1  batch: 356  final cost: 0.00044710137\n",
            "\n",
            "Epoch: 1  batch: 357  final cost: 0.00031605654\n",
            "\n",
            "Epoch: 1  batch: 358  final cost: 0.00046888954\n",
            "\n",
            "Epoch: 1  batch: 359  final cost: 0.0001601796\n",
            "\n",
            "Epoch: 1  batch: 360  final cost: 0.00029462797\n",
            "\n",
            "Epoch: 1  batch: 361  final cost: 0.0002058712\n",
            "\n",
            "Epoch: 1  batch: 362  final cost: 0.00015897848\n",
            "\n",
            "Epoch: 1  batch: 363  final cost: 0.00027334533\n",
            "\n",
            "Epoch: 1  batch: 364  final cost: 0.00029216276\n",
            "\n",
            "Epoch: 1  batch: 365  final cost: 0.00030712702\n",
            "\n",
            "Epoch: 1  batch: 366  final cost: 0.00035070712\n",
            "\n",
            "Epoch: 1  batch: 367  final cost: 0.0003248127\n",
            "\n",
            "Epoch: 1  batch: 368  final cost: 0.0002995214\n",
            "\n",
            "Epoch: 1  batch: 369  final cost: 0.00045253383\n",
            "\n",
            "Epoch: 1  batch: 370  final cost: 0.0003510109\n",
            "\n",
            "Epoch: 1  batch: 371  final cost: 0.00043533617\n",
            "\n",
            "Epoch: 1  batch: 372  final cost: 0.00016409415\n",
            "\n",
            "Epoch: 1  batch: 373  final cost: 0.00018412748\n",
            "\n",
            "Epoch: 1  batch: 374  final cost: 0.0004101215\n",
            "\n",
            "Epoch: 1  batch: 375  final cost: 0.00033735862\n",
            "\n",
            "Epoch: 1  batch: 376  final cost: 0.00027104543\n",
            "\n",
            "Epoch: 1  batch: 377  final cost: 0.00026608753\n",
            "\n",
            "Epoch: 1  batch: 378  final cost: 0.00030211572\n",
            "\n",
            "Epoch: 1  batch: 379  final cost: 0.0002911487\n",
            "\n",
            "Epoch: 1  batch: 380  final cost: 0.0002975805\n",
            "\n",
            "Epoch: 1  batch: 381  final cost: 0.00043347172\n",
            "\n",
            "Epoch: 1  batch: 382  final cost: 0.00030588868\n",
            "\n",
            "Epoch: 1  batch: 383  final cost: 0.0002651323\n",
            "\n",
            "Epoch: 1  batch: 384  final cost: 0.00039048996\n",
            "\n",
            "Epoch: 1  batch: 385  final cost: 0.00041029835\n",
            "\n",
            "Epoch: 1  batch: 386  final cost: 0.00044822288\n",
            "\n",
            "Epoch: 1  batch: 387  final cost: 0.00026205814\n",
            "\n",
            "Epoch: 1  batch: 388  final cost: 0.00018600652\n",
            "\n",
            "Epoch: 1  batch: 389  final cost: 0.00022698486\n",
            "\n",
            "Epoch: 1  batch: 390  final cost: 0.00014838255\n",
            "\n",
            "Epoch: 1  batch: 391  final cost: 0.00028412015\n",
            "\n",
            "Epoch: 1  batch: 392  final cost: 0.00043121015\n",
            "\n",
            "Epoch: 1  batch: 393  final cost: 0.00028387902\n",
            "\n",
            "Epoch: 1  batch: 394  final cost: 0.00023117324\n",
            "\n",
            "Epoch: 1  batch: 395  final cost: 0.00019569263\n",
            "\n",
            "Epoch: 1  batch: 396  final cost: 0.00016056043\n",
            "\n",
            "Epoch: 1  batch: 397  final cost: 0.00041995183\n",
            "\n",
            "Epoch: 1  batch: 398  final cost: 0.00019778761\n",
            "\n",
            "Epoch: 1  batch: 399  final cost: 0.00018561457\n",
            "\n",
            "Epoch: 1  batch: 400  final cost: 0.00044394657\n",
            "\n",
            "Epoch: 1  batch: 401  final cost: 0.000286535\n",
            "\n",
            "Epoch: 1  batch: 402  final cost: 0.00033045956\n",
            "\n",
            "Epoch: 1  batch: 403  final cost: 0.0004060417\n",
            "\n",
            "Epoch: 1  batch: 404  final cost: 0.00033951757\n",
            "\n",
            "Epoch: 1  batch: 405  final cost: 0.0002589728\n",
            "\n",
            "Epoch: 1  batch: 406  final cost: 0.0003000386\n",
            "\n",
            "Epoch: 1  batch: 407  final cost: 0.00041537362\n",
            "\n",
            "Epoch: 1  batch: 408  final cost: 0.00019274387\n",
            "\n",
            "Epoch: 1  batch: 409  final cost: 0.00019232032\n",
            "\n",
            "Epoch: 1  batch: 410  final cost: 0.00039798097\n",
            "\n",
            "Epoch: 1  batch: 411  final cost: 0.00029220118\n",
            "\n",
            "Epoch: 1  batch: 412  final cost: 0.00037794668\n",
            "\n",
            "Epoch: 1  batch: 413  final cost: 0.0002769212\n",
            "\n",
            "Epoch: 1  batch: 414  final cost: 0.00025655847\n",
            "\n",
            "Epoch: 1  batch: 415  final cost: 0.00032755948\n",
            "\n",
            "Epoch: 1  batch: 416  final cost: 0.00027590472\n",
            "\n",
            "Epoch: 1  batch: 417  final cost: 0.00014359463\n",
            "\n",
            "Epoch: 1  batch: 418  final cost: 0.00019191738\n",
            "\n",
            "Epoch: 1  batch: 419  final cost: 0.0002799164\n",
            "\n",
            "Epoch: 1  batch: 420  final cost: 0.00015516677\n",
            "\n",
            "Epoch: 1  batch: 421  final cost: 0.00041499035\n",
            "\n",
            "Epoch: 1  batch: 422  final cost: 0.00025293994\n",
            "\n",
            "Epoch: 1  batch: 423  final cost: 0.00014613243\n",
            "\n",
            "Epoch: 1  batch: 424  final cost: 0.00017262627\n",
            "\n",
            "Epoch: 1  batch: 425  final cost: 0.00015331304\n",
            "\n",
            "Epoch: 1  batch: 426  final cost: 0.0002579188\n",
            "\n",
            "Epoch: 1  batch: 427  final cost: 0.00028362725\n",
            "\n",
            "Epoch: 1  batch: 428  final cost: 0.000417349\n",
            "\n",
            "Epoch: 1  batch: 429  final cost: 0.00021649228\n",
            "\n",
            "Epoch: 1  batch: 430  final cost: 0.00040915952\n",
            "\n",
            "Epoch: 1  batch: 431  final cost: 0.000285253\n",
            "\n",
            "Epoch: 1  batch: 432  final cost: 0.00042677528\n",
            "\n",
            "Epoch: 1  batch: 433  final cost: 0.0003021527\n",
            "\n",
            "Epoch: 1  batch: 434  final cost: 0.00042422258\n",
            "\n",
            "Epoch: 1  batch: 435  final cost: 0.0004275431\n",
            "\n",
            "Epoch: 1  batch: 436  final cost: 0.00026694356\n",
            "\n",
            "Epoch: 1  batch: 437  final cost: 0.00024928036\n",
            "\n",
            "Epoch: 1  batch: 438  final cost: 0.00017548655\n",
            "\n",
            "Epoch: 1  batch: 439  final cost: 0.00021916971\n",
            "\n",
            "Epoch: 1  batch: 440  final cost: 0.00017627898\n",
            "\n",
            "Epoch: 1  batch: 441  final cost: 0.00014022735\n",
            "\n",
            "Epoch: 1  batch: 442  final cost: 0.000388218\n",
            "\n",
            "Epoch: 1  batch: 443  final cost: 0.0004217318\n",
            "\n",
            "Epoch: 1  batch: 444  final cost: 0.00026900475\n",
            "\n",
            "Epoch: 1  batch: 445  final cost: 0.00027348642\n",
            "\n",
            "Epoch: 1  batch: 446  final cost: 0.0002811905\n",
            "\n",
            "Epoch: 1  batch: 447  final cost: 0.00032632935\n",
            "\n",
            "Epoch: 1  batch: 448  final cost: 0.00025196475\n",
            "\n",
            "Epoch: 1  batch: 449  final cost: 0.0002564221\n",
            "\n",
            "Epoch: 1  batch: 450  final cost: 0.00029030052\n",
            "\n",
            "Epoch: 1  batch: 451  final cost: 0.00042893222\n",
            "\n",
            "Epoch: 1  batch: 452  final cost: 0.0002845381\n",
            "\n",
            "Epoch: 1  batch: 453  final cost: 0.00021878324\n",
            "\n",
            "Epoch: 1  batch: 454  final cost: 0.00018533165\n",
            "\n",
            "Epoch: 1  batch: 455  final cost: 0.00039630552\n",
            "\n",
            "Epoch: 1  batch: 456  final cost: 0.00015119379\n",
            "\n",
            "Epoch: 1  batch: 457  final cost: 0.000408805\n",
            "\n",
            "Epoch: 1  batch: 458  final cost: 0.00031986862\n",
            "\n",
            "Epoch: 1  batch: 459  final cost: 0.000169682\n",
            "\n",
            "Epoch: 1  batch: 460  final cost: 0.00014363744\n",
            "\n",
            "Epoch: 1  batch: 461  final cost: 0.00025316636\n",
            "\n",
            "Epoch: 1  batch: 462  final cost: 0.00024975662\n",
            "\n",
            "Epoch: 1  batch: 463  final cost: 0.00015179804\n",
            "\n",
            "Epoch: 1  batch: 464  final cost: 0.0001752046\n",
            "\n",
            "Epoch: 1  batch: 465  final cost: 0.00018536871\n",
            "\n",
            "Epoch: 1  batch: 466  final cost: 0.00026733192\n",
            "\n",
            "Epoch: 1  batch: 467  final cost: 0.00042446595\n",
            "\n",
            "Epoch: 1  batch: 468  final cost: 0.0003210096\n",
            "\n",
            "Epoch: 1  batch: 469  final cost: 0.00037183074\n",
            "\n",
            "Epoch: 1  batch: 470  final cost: 0.00028030603\n",
            "\n",
            "Epoch: 1  batch: 471  final cost: 0.0001872926\n",
            "\n",
            "Epoch: 1  batch: 472  final cost: 0.00017605451\n",
            "\n",
            "Epoch: 1  batch: 473  final cost: 0.00021357709\n",
            "\n",
            "Epoch: 1  batch: 474  final cost: 0.000304702\n",
            "\n",
            "Epoch: 1  batch: 475  final cost: 0.00026968314\n",
            "\n",
            "Epoch: 1  batch: 476  final cost: 0.00024874794\n",
            "\n",
            "Epoch: 1  batch: 477  final cost: 0.00041426686\n",
            "\n",
            "Epoch: 1  batch: 478  final cost: 0.00028781942\n",
            "\n",
            "Epoch: 1  batch: 479  final cost: 0.0004224803\n",
            "\n",
            "Epoch: 1  batch: 480  final cost: 0.00040232006\n",
            "\n",
            "Epoch: 1  batch: 481  final cost: 0.00024570536\n",
            "\n",
            "Epoch: 1  batch: 482  final cost: 0.0002620347\n",
            "\n",
            "Epoch: 1  batch: 483  final cost: 0.00018125224\n",
            "\n",
            "Epoch: 1  batch: 484  final cost: 0.00040530448\n",
            "\n",
            "Epoch: 1  batch: 485  final cost: 0.0004002317\n",
            "\n",
            "Epoch: 1  batch: 486  final cost: 0.00041800257\n",
            "\n",
            "Epoch: 1  batch: 487  final cost: 0.00018142635\n",
            "\n",
            "Epoch: 1  batch: 488  final cost: 0.00027620097\n",
            "\n",
            "Epoch: 1  batch: 489  final cost: 0.00016378198\n",
            "\n",
            "Epoch: 1  batch: 490  final cost: 0.00016957062\n",
            "\n",
            "Epoch: 1  batch: 491  final cost: 0.00027588592\n",
            "\n",
            "Epoch: 1  batch: 492  final cost: 0.00025949645\n",
            "\n",
            "Epoch: 1  batch: 493  final cost: 0.00030675376\n",
            "\n",
            "Epoch: 1  batch: 494  final cost: 0.0002628952\n",
            "\n",
            "Epoch: 1  batch: 495  final cost: 0.0003874253\n",
            "\n",
            "Epoch: 1  batch: 496  final cost: 0.00014444438\n",
            "\n",
            "Epoch: 1  batch: 497  final cost: 0.00041240334\n",
            "\n",
            "Epoch: 1  batch: 498  final cost: 0.00037595764\n",
            "\n",
            "Epoch: 1  batch: 499  final cost: 0.0004024066\n",
            "\n",
            "Epoch: 1  batch: 500  final cost: 0.00013250651\n",
            "\n",
            "Epoch: 1  batch: 501  final cost: 0.0002407944\n",
            "\n",
            "Epoch: 1  batch: 502  final cost: 0.0003747662\n",
            "\n",
            "Epoch: 1  batch: 503  final cost: 0.00023802313\n",
            "\n",
            "Epoch: 1  batch: 504  final cost: 0.00039489949\n",
            "\n",
            "Epoch: 1  batch: 505  final cost: 0.00024236937\n",
            "\n",
            "Epoch: 1  batch: 506  final cost: 0.00029951538\n",
            "\n",
            "Epoch: 1  batch: 507  final cost: 0.00028282637\n",
            "\n",
            "Epoch: 1  batch: 508  final cost: 0.00027259107\n",
            "\n",
            "Epoch: 1  batch: 509  final cost: 0.00016256221\n",
            "\n",
            "Epoch: 1  batch: 510  final cost: 0.00030602963\n",
            "\n",
            "Epoch: 1  batch: 511  final cost: 0.00023124184\n",
            "\n",
            "Epoch: 1  batch: 512  final cost: 0.00013302616\n",
            "\n",
            "Epoch: 1  batch: 513  final cost: 0.00019774033\n",
            "\n",
            "Epoch: 1  batch: 514  final cost: 0.000257865\n",
            "\n",
            "Epoch: 1  batch: 515  final cost: 0.00024987783\n",
            "\n",
            "Epoch: 1  batch: 516  final cost: 0.00017091008\n",
            "\n",
            "Epoch: 1  batch: 517  final cost: 0.00034246105\n",
            "\n",
            "Epoch: 1  batch: 518  final cost: 0.0001395653\n",
            "\n",
            "Epoch: 1  batch: 519  final cost: 0.00026660226\n",
            "\n",
            "Epoch: 1  batch: 520  final cost: 0.00020196459\n",
            "\n",
            "Epoch: 1  batch: 521  final cost: 0.00026630153\n",
            "\n",
            "Epoch: 1  batch: 522  final cost: 0.0001704152\n",
            "\n",
            "Epoch: 1  batch: 523  final cost: 0.00038972188\n",
            "\n",
            "Epoch: 1  batch: 524  final cost: 0.00016079363\n",
            "\n",
            "Epoch: 1  batch: 525  final cost: 0.00039459125\n",
            "\n",
            "Epoch: 1  batch: 526  final cost: 0.00026895368\n",
            "\n",
            "Epoch: 1  batch: 527  final cost: 0.00038871236\n",
            "\n",
            "Epoch: 1  batch: 528  final cost: 0.00022795695\n",
            "\n",
            "Epoch: 1  batch: 529  final cost: 0.00037716058\n",
            "\n",
            "Epoch: 1  batch: 530  final cost: 0.00023241733\n",
            "\n",
            "Epoch: 1  batch: 531  final cost: 0.00013725276\n",
            "\n",
            "Epoch: 1  batch: 532  final cost: 0.00024848027\n",
            "\n",
            "Epoch: 1  batch: 533  final cost: 0.00015876914\n",
            "\n",
            "Epoch: 1  batch: 534  final cost: 0.00028808726\n",
            "\n",
            "Epoch: 1  batch: 535  final cost: 0.00015257223\n",
            "\n",
            "Epoch: 1  batch: 536  final cost: 0.00012581641\n",
            "\n",
            "Epoch: 1  batch: 537  final cost: 0.00028737314\n",
            "\n",
            "Epoch: 1  batch: 538  final cost: 0.00033047234\n",
            "\n",
            "Epoch: 1  batch: 539  final cost: 0.00012834207\n",
            "\n",
            "Epoch: 1  batch: 540  final cost: 0.00019049886\n",
            "\n",
            "Epoch: 1  batch: 541  final cost: 0.00019476928\n",
            "\n",
            "Epoch: 1  batch: 542  final cost: 0.00029368344\n",
            "\n",
            "Epoch: 1  batch: 543  final cost: 0.00022146429\n",
            "\n",
            "Epoch: 1  batch: 544  final cost: 0.00025386713\n",
            "\n",
            "Epoch: 1  batch: 545  final cost: 0.0003854812\n",
            "\n",
            "Epoch: 1  batch: 546  final cost: 0.00035977957\n",
            "\n",
            "Epoch: 1  batch: 547  final cost: 0.00035224468\n",
            "\n",
            "Epoch: 1  batch: 548  final cost: 0.00036450173\n",
            "\n",
            "Epoch: 1  batch: 549  final cost: 0.0001338023\n",
            "\n",
            "Epoch: 1  batch: 550  final cost: 0.00035363156\n",
            "\n",
            "Epoch: 1  batch: 551  final cost: 0.00026891593\n",
            "\n",
            "Epoch: 1  batch: 552  final cost: 0.00025320397\n",
            "\n",
            "Epoch: 1  batch: 553  final cost: 0.0001633007\n",
            "\n",
            "Epoch: 1  batch: 554  final cost: 0.0002284475\n",
            "\n",
            "Epoch: 1  batch: 555  final cost: 0.00016446358\n",
            "\n",
            "Epoch: 1  batch: 556  final cost: 0.00024513234\n",
            "\n",
            "Epoch: 1  batch: 557  final cost: 0.00023724693\n",
            "\n",
            "Epoch: 1  batch: 558  final cost: 0.00023725665\n",
            "\n",
            "Epoch: 1  batch: 559  final cost: 0.00023461471\n",
            "\n",
            "Epoch: 1  batch: 560  final cost: 0.00022387598\n",
            "\n",
            "Epoch: 1  batch: 561  final cost: 0.00015402123\n",
            "\n",
            "Epoch: 1  batch: 562  final cost: 0.0001268884\n",
            "\n",
            "Epoch: 1  batch: 563  final cost: 0.00038591385\n",
            "\n",
            "Epoch: 1  batch: 564  final cost: 0.0002489775\n",
            "\n",
            "Epoch: 1  batch: 565  final cost: 0.0002870089\n",
            "\n",
            "Epoch: 1  batch: 566  final cost: 0.00036232287\n",
            "\n",
            "Epoch: 1  batch: 567  final cost: 0.0001897576\n",
            "\n",
            "Epoch: 1  batch: 568  final cost: 0.00019420216\n",
            "\n",
            "Epoch: 1  batch: 569  final cost: 0.00024116959\n",
            "\n",
            "Epoch: 1  batch: 570  final cost: 0.00022940146\n",
            "\n",
            "Epoch: 1  batch: 571  final cost: 0.0003812352\n",
            "\n",
            "Epoch: 1  batch: 572  final cost: 0.0003901524\n",
            "\n",
            "Epoch: 1  batch: 573  final cost: 0.00023582103\n",
            "\n",
            "Epoch: 1  batch: 574  final cost: 0.00024625924\n",
            "\n",
            "Epoch: 1  batch: 575  final cost: 0.00036544807\n",
            "\n",
            "Epoch: 1  batch: 576  final cost: 0.00026213218\n",
            "\n",
            "Epoch: 1  batch: 577  final cost: 0.00016724979\n",
            "\n",
            "Epoch: 1  batch: 578  final cost: 0.0002999008\n",
            "\n",
            "Epoch: 1  batch: 579  final cost: 0.0003723902\n",
            "\n",
            "Epoch: 1  batch: 580  final cost: 0.00024247274\n",
            "\n",
            "Epoch: 1  batch: 581  final cost: 0.00023356325\n",
            "\n",
            "Epoch: 1  batch: 582  final cost: 0.00015701642\n",
            "\n",
            "Epoch: 1  batch: 583  final cost: 0.00028674153\n",
            "\n",
            "Epoch: 1  batch: 584  final cost: 0.00015132228\n",
            "\n",
            "Epoch: 1  batch: 585  final cost: 0.00013489513\n",
            "\n",
            "Epoch: 1  batch: 586  final cost: 0.0003341612\n",
            "\n",
            "Epoch: 1  batch: 587  final cost: 0.00022403974\n",
            "\n",
            "Epoch: 1  batch: 588  final cost: 0.000239135\n",
            "\n",
            "Epoch: 1  batch: 589  final cost: 0.00012510591\n",
            "\n",
            "Epoch: 1  batch: 590  final cost: 0.00022382593\n",
            "\n",
            "Epoch: 1  batch: 591  final cost: 0.00038971717\n",
            "\n",
            "Epoch: 1  batch: 592  final cost: 0.0003778327\n",
            "\n",
            "Epoch: 1  batch: 593  final cost: 0.0002617588\n",
            "\n",
            "Epoch: 1  batch: 594  final cost: 0.00016455495\n",
            "\n",
            "Epoch: 1  batch: 595  final cost: 0.0002670227\n",
            "\n",
            "Epoch: 1  batch: 596  final cost: 0.00025995926\n",
            "\n",
            "Epoch: 1  batch: 597  final cost: 0.00013381436\n",
            "\n",
            "Epoch: 1  batch: 598  final cost: 0.00036460717\n",
            "\n",
            "Epoch: 1  batch: 599  final cost: 0.00016352514\n",
            "\n",
            "Epoch: 1  batch: 600  final cost: 0.00027545713\n",
            "\n",
            "Epoch: 1  batch: 601  final cost: 0.00036323402\n",
            "\n",
            "Epoch: 1  batch: 602  final cost: 0.000265039\n",
            "\n",
            "Epoch: 1  batch: 603  final cost: 0.00014913203\n",
            "\n",
            "Epoch: 1  batch: 604  final cost: 0.00037213924\n",
            "\n",
            "Epoch: 1  batch: 605  final cost: 0.00027251206\n",
            "\n",
            "Epoch: 1  batch: 606  final cost: 0.00023751275\n",
            "\n",
            "Epoch: 1  batch: 607  final cost: 0.0001315425\n",
            "\n",
            "Epoch: 1  batch: 608  final cost: 0.0002851684\n",
            "\n",
            "Epoch: 1  batch: 609  final cost: 0.00035837374\n",
            "\n",
            "Epoch: 1  batch: 610  final cost: 0.00021703019\n",
            "\n",
            "Epoch: 1  batch: 611  final cost: 0.00015990323\n",
            "\n",
            "Epoch: 1  batch: 612  final cost: 0.00023406943\n",
            "\n",
            "Epoch: 1  batch: 613  final cost: 0.0002161834\n",
            "\n",
            "Epoch: 1  batch: 614  final cost: 0.00036859798\n",
            "\n",
            "Epoch: 1  batch: 615  final cost: 0.00037824726\n",
            "\n",
            "Epoch: 1  batch: 616  final cost: 0.00012960646\n",
            "\n",
            "Epoch: 1  batch: 617  final cost: 0.00014978644\n",
            "\n",
            "Epoch: 1  batch: 618  final cost: 0.00025112025\n",
            "\n",
            "Epoch: 1  batch: 619  final cost: 0.00015766843\n",
            "\n",
            "Epoch: 1  batch: 620  final cost: 0.0003557443\n",
            "\n",
            "Epoch: 1  batch: 621  final cost: 0.00015913213\n",
            "\n",
            "Epoch: 1  batch: 622  final cost: 0.0003181668\n",
            "\n",
            "Epoch: 1  batch: 623  final cost: 0.00012225368\n",
            "\n",
            "Epoch: 1  batch: 624  final cost: 0.0002219351\n",
            "\n",
            "Epoch: 1  batch: 625  final cost: 0.00027178111\n",
            "\n",
            "Epoch: 1  batch: 626  final cost: 0.00024155536\n",
            "\n",
            "Epoch: 1  batch: 627  final cost: 0.00034693227\n",
            "\n",
            "Epoch: 1  batch: 628  final cost: 0.00011859604\n",
            "\n",
            "Epoch: 1  batch: 629  final cost: 0.00022630928\n",
            "\n",
            "Epoch: 1  batch: 630  final cost: 0.00023211021\n",
            "\n",
            "Epoch: 1  batch: 631  final cost: 0.00024576578\n",
            "\n",
            "Epoch: 1  batch: 632  final cost: 0.00022398347\n",
            "\n",
            "Epoch: 1  batch: 633  final cost: 0.00024693584\n",
            "\n",
            "Epoch: 1  batch: 634  final cost: 0.00018005335\n",
            "\n",
            "Epoch: 1  batch: 635  final cost: 0.00036788176\n",
            "\n",
            "Epoch: 1  batch: 636  final cost: 0.0002844657\n",
            "\n",
            "Epoch: 1  batch: 637  final cost: 0.00022041716\n",
            "\n",
            "Epoch: 1  batch: 638  final cost: 0.00037768678\n",
            "\n",
            "Epoch: 1  batch: 639  final cost: 0.00014834931\n",
            "\n",
            "Epoch: 1  batch: 640  final cost: 0.00018436619\n",
            "\n",
            "Epoch: 1  batch: 641  final cost: 0.00015591379\n",
            "\n",
            "Epoch: 1  batch: 642  final cost: 0.00027803393\n",
            "\n",
            "Epoch: 1  batch: 643  final cost: 0.00022276008\n",
            "\n",
            "Epoch: 1  batch: 644  final cost: 0.0003632622\n",
            "\n",
            "Epoch: 1  batch: 645  final cost: 0.0002710879\n",
            "\n",
            "Epoch: 1  batch: 646  final cost: 0.00018405844\n",
            "\n",
            "Epoch: 1  batch: 647  final cost: 0.00028497697\n",
            "\n",
            "Epoch: 1  batch: 648  final cost: 0.00012696662\n",
            "\n",
            "Epoch: 1  batch: 649  final cost: 0.00024175787\n",
            "\n",
            "Epoch: 1  batch: 650  final cost: 0.00012092722\n",
            "\n",
            "Epoch: 1  batch: 651  final cost: 0.000261825\n",
            "\n",
            "Epoch: 1  batch: 652  final cost: 0.00015518848\n",
            "\n",
            "Epoch: 1  batch: 653  final cost: 0.00035155716\n",
            "\n",
            "Epoch: 1  batch: 654  final cost: 0.00024642752\n",
            "\n",
            "Epoch: 1  batch: 655  final cost: 0.00034659795\n",
            "\n",
            "Epoch: 1  batch: 656  final cost: 0.00014591958\n",
            "\n",
            "Epoch: 1  batch: 657  final cost: 0.00012586445\n",
            "\n",
            "Epoch: 1  batch: 658  final cost: 0.00035298226\n",
            "\n",
            "Epoch: 1  batch: 659  final cost: 0.00022331456\n",
            "\n",
            "Epoch: 1  batch: 660  final cost: 0.00034184835\n",
            "\n",
            "Epoch: 1  batch: 661  final cost: 0.0003112785\n",
            "\n",
            "Epoch: 1  batch: 662  final cost: 0.00037133312\n",
            "\n",
            "Epoch: 1  batch: 663  final cost: 0.00021691644\n",
            "\n",
            "Epoch: 1  batch: 664  final cost: 0.00017650293\n",
            "\n",
            "Epoch: 1  batch: 665  final cost: 0.00024461388\n",
            "\n",
            "Epoch: 1  batch: 666  final cost: 0.00020816847\n",
            "\n",
            "Epoch: 1  batch: 667  final cost: 0.00020767823\n",
            "\n",
            "Epoch: 1  batch: 668  final cost: 0.00022501174\n",
            "\n",
            "Epoch: 1  batch: 669  final cost: 0.00014525546\n",
            "\n",
            "Epoch: 1  batch: 670  final cost: 0.000116285664\n",
            "\n",
            "Epoch: 1  batch: 671  final cost: 0.00022529198\n",
            "\n",
            "Epoch: 1  batch: 672  final cost: 0.00036996772\n",
            "\n",
            "Epoch: 1  batch: 673  final cost: 0.00024466554\n",
            "\n",
            "Epoch: 1  batch: 674  final cost: 0.00022904143\n",
            "\n",
            "Epoch: 1  batch: 675  final cost: 0.00014063407\n",
            "\n",
            "Epoch: 1  batch: 676  final cost: 0.00025220623\n",
            "\n",
            "Epoch: 1  batch: 677  final cost: 0.00015526447\n",
            "\n",
            "Epoch: 1  batch: 678  final cost: 0.0003478746\n",
            "\n",
            "Epoch: 1  batch: 679  final cost: 0.00022476133\n",
            "\n",
            "Epoch: 1  batch: 680  final cost: 0.0003691389\n",
            "\n",
            "Epoch: 1  batch: 681  final cost: 0.00037751338\n",
            "\n",
            "Epoch: 1  batch: 682  final cost: 0.000125258\n",
            "\n",
            "Epoch: 1  batch: 683  final cost: 0.00035851877\n",
            "\n",
            "Epoch: 1  batch: 684  final cost: 0.00020925471\n",
            "\n",
            "Epoch: 1  batch: 685  final cost: 0.00036804492\n",
            "\n",
            "Epoch: 1  batch: 686  final cost: 0.00022539006\n",
            "\n",
            "Epoch: 1  batch: 687  final cost: 0.00022529093\n",
            "\n",
            "Epoch: 1  batch: 688  final cost: 0.00020835554\n",
            "\n",
            "Epoch: 1  batch: 689  final cost: 0.00022290603\n",
            "\n",
            "Epoch: 1  batch: 690  final cost: 0.00022032535\n",
            "\n",
            "Epoch: 1  batch: 691  final cost: 0.0003484127\n",
            "\n",
            "Epoch: 1  batch: 692  final cost: 0.00024732965\n",
            "\n",
            "Epoch: 1  batch: 693  final cost: 0.00022953929\n",
            "\n",
            "Epoch: 1  batch: 694  final cost: 0.00018099384\n",
            "\n",
            "Epoch: 1  batch: 695  final cost: 0.0003153212\n",
            "\n",
            "Epoch: 1  batch: 696  final cost: 0.00022390144\n",
            "\n",
            "Epoch: 1  batch: 697  final cost: 0.00011978496\n",
            "\n",
            "Epoch: 1  batch: 698  final cost: 0.00014545849\n",
            "\n",
            "Epoch: 1  batch: 699  final cost: 0.00025625905\n",
            "\n",
            "Epoch: 1  batch: 700  final cost: 0.0001259404\n",
            "\n",
            "Epoch: 1  batch: 701  final cost: 0.00035693377\n",
            "\n",
            "Epoch: 1  batch: 702  final cost: 0.00025139097\n",
            "\n",
            "Epoch: 1  batch: 703  final cost: 0.00014122843\n",
            "\n",
            "Epoch: 1  batch: 704  final cost: 0.0002454232\n",
            "\n",
            "Epoch: 1  batch: 705  final cost: 0.00038138952\n",
            "\n",
            "Epoch: 1  batch: 706  final cost: 0.0001553594\n",
            "\n",
            "Epoch: 1  batch: 707  final cost: 0.00014611377\n",
            "\n",
            "Epoch: 1  batch: 708  final cost: 0.00017730109\n",
            "\n",
            "Epoch: 1  batch: 709  final cost: 0.00015367662\n",
            "\n",
            "Epoch: 1  batch: 710  final cost: 0.0003559305\n",
            "\n",
            "Epoch: 1  batch: 711  final cost: 0.0003549081\n",
            "\n",
            "Epoch: 1  batch: 712  final cost: 0.00011678655\n",
            "\n",
            "Epoch: 1  batch: 713  final cost: 0.00028715943\n",
            "\n",
            "Epoch: 1  batch: 714  final cost: 0.00036107737\n",
            "\n",
            "Epoch: 1  batch: 715  final cost: 0.0002650223\n",
            "\n",
            "Epoch: 1  batch: 716  final cost: 0.00022743602\n",
            "\n",
            "Epoch: 1  batch: 717  final cost: 0.00027805133\n",
            "\n",
            "Epoch: 1  batch: 718  final cost: 0.00015220125\n",
            "\n",
            "Epoch: 1  batch: 719  final cost: 0.00024944165\n",
            "\n",
            "Epoch: 1  batch: 720  final cost: 0.00026509925\n",
            "\n",
            "Epoch: 1  batch: 721  final cost: 0.0002830393\n",
            "\n",
            "Epoch: 1  batch: 722  final cost: 0.00034652263\n",
            "\n",
            "Epoch: 1  batch: 723  final cost: 0.00015035519\n",
            "\n",
            "Epoch: 1  batch: 724  final cost: 0.00022189289\n",
            "\n",
            "Epoch: 1  batch: 725  final cost: 0.0002241984\n",
            "\n",
            "Epoch: 1  batch: 726  final cost: 0.0002602735\n",
            "\n",
            "Epoch: 1  batch: 727  final cost: 0.00034392677\n",
            "\n",
            "Epoch: 1  batch: 728  final cost: 0.00021978006\n",
            "\n",
            "Epoch: 1  batch: 729  final cost: 0.00013592368\n",
            "\n",
            "Epoch: 1  batch: 730  final cost: 0.000347992\n",
            "\n",
            "Epoch: 1  batch: 731  final cost: 0.0002403117\n",
            "\n",
            "Epoch: 1  batch: 732  final cost: 0.00021553243\n",
            "\n",
            "Epoch: 1  batch: 733  final cost: 0.00023384161\n",
            "\n",
            "Epoch: 1  batch: 734  final cost: 0.00021353972\n",
            "\n",
            "Epoch: 1  batch: 735  final cost: 0.00020164305\n",
            "\n",
            "Epoch: 1  batch: 736  final cost: 0.00023860601\n",
            "\n",
            "Epoch: 1  batch: 737  final cost: 0.00024464756\n",
            "\n",
            "Epoch: 1  batch: 738  final cost: 0.00020139509\n",
            "\n",
            "Epoch: 1  batch: 739  final cost: 0.000120189\n",
            "\n",
            "Epoch: 1  batch: 740  final cost: 0.00034077797\n",
            "\n",
            "Epoch: 1  batch: 741  final cost: 0.000112119924\n",
            "\n",
            "Epoch: 1  batch: 742  final cost: 0.00030305068\n",
            "\n",
            "Epoch: 1  batch: 743  final cost: 0.0003448088\n",
            "\n",
            "Epoch: 1  batch: 744  final cost: 0.00035898096\n",
            "\n",
            "Epoch: 1  batch: 745  final cost: 0.00023941626\n",
            "\n",
            "Epoch: 1  batch: 746  final cost: 0.00025177898\n",
            "\n",
            "Epoch: 1  batch: 747  final cost: 0.00021549576\n",
            "\n",
            "Epoch: 1  batch: 748  final cost: 0.00014564667\n",
            "\n",
            "Epoch: 1  batch: 749  final cost: 0.00033355382\n",
            "\n",
            "Epoch: 1  batch: 750  final cost: 0.000113061\n",
            "\n",
            "Epoch: 1  batch: 751  final cost: 0.00014605797\n",
            "\n",
            "Epoch: 1  batch: 752  final cost: 0.00013681436\n",
            "\n",
            "Epoch: 1  batch: 753  final cost: 0.00011798313\n",
            "\n",
            "Epoch: 1  batch: 754  final cost: 0.00026209166\n",
            "\n",
            "Epoch: 1  batch: 755  final cost: 0.00013706577\n",
            "\n",
            "Epoch: 1  batch: 756  final cost: 0.00017007493\n",
            "\n",
            "Epoch: 1  batch: 757  final cost: 0.00021062995\n",
            "\n",
            "Epoch: 1  batch: 758  final cost: 0.00016631756\n",
            "\n",
            "Epoch: 1  batch: 759  final cost: 0.00035951278\n",
            "\n",
            "Epoch: 1  batch: 760  final cost: 0.0003565485\n",
            "\n",
            "Epoch: 1  batch: 761  final cost: 0.00026239845\n",
            "\n",
            "Epoch: 1  batch: 762  final cost: 0.00033422705\n",
            "\n",
            "Epoch: 1  batch: 763  final cost: 0.0002120156\n",
            "\n",
            "Epoch: 1  batch: 764  final cost: 0.00014585923\n",
            "\n",
            "Epoch: 1  batch: 765  final cost: 0.00033632346\n",
            "\n",
            "Epoch: 1  batch: 766  final cost: 0.00011283836\n",
            "\n",
            "Epoch: 1  batch: 767  final cost: 0.000110221634\n",
            "\n",
            "Epoch: 1  batch: 768  final cost: 0.00035974412\n",
            "\n",
            "Epoch: 1  batch: 769  final cost: 0.00023128586\n",
            "\n",
            "Epoch: 1  batch: 770  final cost: 0.00014442285\n",
            "\n",
            "Epoch: 1  batch: 771  final cost: 0.00033564342\n",
            "\n",
            "Epoch: 1  batch: 772  final cost: 0.00036339057\n",
            "\n",
            "Epoch: 1  batch: 773  final cost: 0.00021412478\n",
            "\n",
            "Epoch: 1  batch: 774  final cost: 0.00013246699\n",
            "\n",
            "Epoch: 1  batch: 775  final cost: 0.00023743659\n",
            "\n",
            "Epoch: 1  batch: 776  final cost: 0.00025142464\n",
            "\n",
            "Epoch: 1  batch: 777  final cost: 0.000118080185\n",
            "\n",
            "Epoch: 1  batch: 778  final cost: 0.00021175088\n",
            "\n",
            "Epoch: 1  batch: 779  final cost: 0.00019889657\n",
            "\n",
            "Epoch: 1  batch: 780  final cost: 0.00019857266\n",
            "\n",
            "Epoch: 1  batch: 781  final cost: 0.00016696259\n",
            "\n",
            "Epoch: 1  batch: 782  final cost: 0.00027234523\n",
            "\n",
            "Epoch: 1  batch: 783  final cost: 0.00024377991\n",
            "\n",
            "Epoch: 1  batch: 784  final cost: 0.00023920197\n",
            "\n",
            "Epoch: 1  batch: 785  final cost: 0.00034285494\n",
            "\n",
            "Epoch: 1  batch: 786  final cost: 0.00025247646\n",
            "\n",
            "Epoch: 1  batch: 787  final cost: 0.00014434665\n",
            "\n",
            "Epoch: 1  batch: 788  final cost: 0.0001174911\n",
            "\n",
            "Epoch: 1  batch: 789  final cost: 0.0003315906\n",
            "\n",
            "Epoch: 1  batch: 790  final cost: 0.00021471121\n",
            "\n",
            "Epoch: 1  batch: 791  final cost: 0.0002952077\n",
            "\n",
            "Epoch: 1  batch: 792  final cost: 0.00016780668\n",
            "\n",
            "Epoch: 1  batch: 793  final cost: 0.00021049284\n",
            "\n",
            "Epoch: 1  batch: 794  final cost: 0.00023254594\n",
            "\n",
            "Epoch: 1  batch: 795  final cost: 0.00013410996\n",
            "\n",
            "Epoch: 1  batch: 796  final cost: 0.00020841122\n",
            "\n",
            "Epoch: 1  batch: 797  final cost: 0.00013477363\n",
            "\n",
            "Epoch: 1  batch: 798  final cost: 0.00035101958\n",
            "\n",
            "Epoch: 1  batch: 799  final cost: 0.0003348542\n",
            "\n",
            "Epoch: 1  batch: 800  final cost: 0.00021227749\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "colab_type": "text",
        "id": "U54F8NuW3EBr"
      },
      "source": [
        "##### LSTM V2"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "DTHSBdO7V0Er",
        "outputId": "9d094808-33ad-478e-c68c-568dd1614f7c",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 493
        }
      },
      "source": [
        "## Run me\n",
        "\n",
        "gateDimensions = [[3], [4], [5], [6], [7], [8,8], [9,9], [10], [11,11], [12,12], [13,13], [14,14], [15,15], [16,16]]\n",
        "registerDimensions = [17, 17, 17]\n",
        "\n",
        "nramController2 = NRAMControllerLSTM2(hidden_dimensions=256, gateControllerDimensions=gateDimensions, registerControllerDimensions=registerDimensions)\n",
        "\n",
        "gates = [zeroModule, oneModule, twoModule, incModule, addModule, subModule, decModule, lessModule, lessOrEqualThenModule, equalModule, minimumModule, maximumModule]\n",
        "\n",
        "loss_history2, model2 = trainModel(inputTrainSet, targetTrainSet, gates, nramController2, readMemory, writeMemory, min_error=1e-5)\n"
      ],
      "execution_count": 125,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Epoch: 0  batch: 1  final cost: 184.42093\n",
            "\n",
            "Epoch: 0  batch: 2  final cost: 139.9801\n",
            "\n",
            "Epoch: 0  batch: 3  final cost: 72.17754\n",
            "\n",
            "Epoch: 0  batch: 4  final cost: 22.548386\n",
            "\n",
            "Epoch: 0  batch: 5  final cost: 4.364687\n",
            "\n",
            "Epoch: 0  batch: 6  final cost: 1.0808675\n",
            "\n",
            "Epoch: 0  batch: 7  final cost: 0.19303645\n",
            "\n",
            "Epoch: 0  batch: 8  final cost: 0.037795674\n",
            "\n",
            "Epoch: 0  batch: 9  final cost: 0.0045655123\n",
            "\n",
            "Epoch: 0  batch: 10  final cost: 0.00067967334\n",
            "\n",
            "Epoch: 0  batch: 11  final cost: 8.916114e-05\n",
            "\n",
            "Epoch: 0  batch: 12  final cost: 2.4955685e-05\n",
            "\n",
            "Epoch: 0  batch: 13  final cost: 3.3829494e-06\n",
            "\n",
            "Exit: Epoch: 0  batch: 13  final cost: 3.3829494e-06\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "colab_type": "text",
        "id": "zzDVr5bL3HPD"
      },
      "source": [
        "##### MLP Model"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "TNi_n42GWhfl",
        "outputId": "8b4f80a2-6e21-4f97-d7d1-a734d9959676",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 969
        }
      },
      "source": [
        "## Run me\n",
        "\n",
        "gateDimensions = [[3], [4], [5], [6], [7], [8,8], [9,9], [10], [11,11], [12,12], [13,13], [14,14], [15,15], [16,16]]\n",
        "registerDimensions = [17, 17, 17]\n",
        "nramController3 = NRAMController(hidden_dimensions=400, gateControllerDimensions=gateDimensions, registerControllerDimensions=registerDimensions);\n",
        "\n",
        "gates = [zeroModule, oneModule, twoModule, incModule, addModule, subModule, decModule, lessModule, lessOrEqualThenModule, equalModule, minimumModule, maximumModule]\n",
        "\n",
        "loss_history3, model3 = trainModel(inputTrainSet, targetTrainSet, gates, nramController3, readMemory, writeMemory, min_error=1e-5)\n"
      ],
      "execution_count": 92,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Epoch: 0  batch: 1  final cost: 183.07669\n",
            "\n",
            "Epoch: 0  batch: 2  final cost: 147.57573\n",
            "\n",
            "Epoch: 0  batch: 3  final cost: 114.1943\n",
            "\n",
            "Epoch: 0  batch: 4  final cost: 76.894554\n",
            "\n",
            "Epoch: 0  batch: 5  final cost: 26.695028\n",
            "\n",
            "Epoch: 0  batch: 6  final cost: 15.366084\n",
            "\n",
            "Epoch: 0  batch: 7  final cost: 11.696941\n",
            "\n",
            "Epoch: 0  batch: 8  final cost: 22.958727\n",
            "\n",
            "Epoch: 0  batch: 9  final cost: 15.510916\n",
            "\n",
            "Epoch: 0  batch: 10  final cost: 9.109247\n",
            "\n",
            "Epoch: 0  batch: 11  final cost: 1.8692589\n",
            "\n",
            "Epoch: 0  batch: 12  final cost: 3.276905\n",
            "\n",
            "Epoch: 0  batch: 13  final cost: 0.25586993\n",
            "\n",
            "Epoch: 0  batch: 14  final cost: 0.3107514\n",
            "\n",
            "Epoch: 0  batch: 15  final cost: 0.04565944\n",
            "\n",
            "Epoch: 0  batch: 16  final cost: 0.021887556\n",
            "\n",
            "Epoch: 0  batch: 17  final cost: 0.016588349\n",
            "\n",
            "Epoch: 0  batch: 18  final cost: 0.0092329485\n",
            "\n",
            "Epoch: 0  batch: 19  final cost: 0.0038904476\n",
            "\n",
            "Epoch: 0  batch: 20  final cost: 0.0010529716\n",
            "\n",
            "Epoch: 0  batch: 21  final cost: 0.00019495122\n",
            "\n",
            "Epoch: 0  batch: 22  final cost: 0.0002838562\n",
            "\n",
            "Epoch: 0  batch: 23  final cost: 0.00011486539\n",
            "\n",
            "Epoch: 0  batch: 24  final cost: 3.2422176e-05\n",
            "\n",
            "Epoch: 0  batch: 25  final cost: 4.167106e-05\n",
            "\n",
            "Epoch: 0  batch: 26  final cost: 1.3751036e-05\n",
            "\n",
            "Epoch: 0  batch: 27  final cost: 8.6002065e-06\n",
            "\n",
            "Exit: Epoch: 0  batch: 27  final cost: 8.6002065e-06\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "rWILQepYA-jf",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 5389
        },
        "outputId": "70b4c58e-6704-48d0-ae62-5f1d237ee03d"
      },
      "source": [
        "## Run me\n",
        "\n",
        "gateDimensions = [[3], [4], [5], [6], [7], [8,8], [9,9], [10], [11,11], [12,12], [13,13], [14,14], [15,15], [16,16]]\n",
        "registerDimensions = [17, 17, 17]\n",
        "nramController3 = NRAMController(hidden_dimensions=100, gateControllerDimensions=gateDimensions, registerControllerDimensions=registerDimensions);\n",
        "\n",
        "# gates = [zeroModule, oneModule, addModule]\n",
        "\n",
        "gates = [zeroModule, oneModule, twoModule, incModule, addModule, subModule, decModule, lessModule, lessOrEqualThenModule, equalModule, minimumModule, maximumModule]\n",
        "\n",
        "loss_history3_2, model3_2 = trainModel(inputTrainSet, targetTrainSet, gates, nramController3, readMemory, writeMemory, min_error=1e-5)\n"
      ],
      "execution_count": 104,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Epoch: 0  batch: 1  final cost: 204.02846\n",
            "\n",
            "Epoch: 0  batch: 2  final cost: 181.59215\n",
            "\n",
            "Epoch: 0  batch: 3  final cost: 186.86281\n",
            "\n",
            "Epoch: 0  batch: 4  final cost: 169.34142\n",
            "\n",
            "Epoch: 0  batch: 5  final cost: 97.7998\n",
            "\n",
            "Epoch: 0  batch: 6  final cost: 85.44443\n",
            "\n",
            "Epoch: 0  batch: 7  final cost: 70.48814\n",
            "\n",
            "Epoch: 0  batch: 8  final cost: 87.16077\n",
            "\n",
            "Epoch: 0  batch: 9  final cost: 51.928284\n",
            "\n",
            "Epoch: 0  batch: 10  final cost: 49.1476\n",
            "\n",
            "Epoch: 0  batch: 11  final cost: 26.456648\n",
            "\n",
            "Epoch: 0  batch: 12  final cost: 63.160385\n",
            "\n",
            "Epoch: 0  batch: 13  final cost: 22.069487\n",
            "\n",
            "Epoch: 0  batch: 14  final cost: 32.578815\n",
            "\n",
            "Epoch: 0  batch: 15  final cost: 12.798198\n",
            "\n",
            "Epoch: 0  batch: 16  final cost: 11.754823\n",
            "\n",
            "Epoch: 0  batch: 17  final cost: 21.87871\n",
            "\n",
            "Epoch: 0  batch: 18  final cost: 17.821373\n",
            "\n",
            "Epoch: 0  batch: 19  final cost: 14.355574\n",
            "\n",
            "Epoch: 0  batch: 20  final cost: 10.671577\n",
            "\n",
            "Epoch: 0  batch: 21  final cost: 2.9967742\n",
            "\n",
            "Epoch: 0  batch: 22  final cost: 7.7291036\n",
            "\n",
            "Epoch: 0  batch: 23  final cost: 5.3452845\n",
            "\n",
            "Epoch: 0  batch: 24  final cost: 1.3414637\n",
            "\n",
            "Epoch: 0  batch: 25  final cost: 5.310838\n",
            "\n",
            "Epoch: 0  batch: 26  final cost: 1.235021\n",
            "\n",
            "Epoch: 0  batch: 27  final cost: 1.9434938\n",
            "\n",
            "Epoch: 0  batch: 28  final cost: 0.6268354\n",
            "\n",
            "Epoch: 0  batch: 29  final cost: 1.7945462\n",
            "\n",
            "Epoch: 0  batch: 30  final cost: 0.28000003\n",
            "\n",
            "Epoch: 0  batch: 31  final cost: 0.7068134\n",
            "\n",
            "Epoch: 0  batch: 32  final cost: 0.9678565\n",
            "\n",
            "Epoch: 0  batch: 33  final cost: 0.13587229\n",
            "\n",
            "Epoch: 0  batch: 34  final cost: 0.13582823\n",
            "\n",
            "Epoch: 0  batch: 35  final cost: 0.09632691\n",
            "\n",
            "Epoch: 0  batch: 36  final cost: 0.27025008\n",
            "\n",
            "Epoch: 0  batch: 37  final cost: 0.03997059\n",
            "\n",
            "Epoch: 0  batch: 38  final cost: 0.027836185\n",
            "\n",
            "Epoch: 0  batch: 39  final cost: 0.036287777\n",
            "\n",
            "Epoch: 0  batch: 40  final cost: 0.03182613\n",
            "\n",
            "Epoch: 0  batch: 41  final cost: 0.019674193\n",
            "\n",
            "Epoch: 0  batch: 42  final cost: 0.015693843\n",
            "\n",
            "Epoch: 0  batch: 43  final cost: 0.0208826\n",
            "\n",
            "Epoch: 0  batch: 44  final cost: 0.010245494\n",
            "\n",
            "Epoch: 0  batch: 45  final cost: 0.0121530965\n",
            "\n",
            "Epoch: 0  batch: 46  final cost: 0.009369274\n",
            "\n",
            "Epoch: 0  batch: 47  final cost: 0.009432655\n",
            "\n",
            "Epoch: 0  batch: 48  final cost: 0.009097817\n",
            "\n",
            "Epoch: 0  batch: 49  final cost: 0.01100786\n",
            "\n",
            "Epoch: 0  batch: 50  final cost: 0.007683811\n",
            "\n",
            "Epoch: 0  batch: 51  final cost: 0.0059631197\n",
            "\n",
            "Epoch: 0  batch: 52  final cost: 0.006550072\n",
            "\n",
            "Epoch: 0  batch: 53  final cost: 0.0049851784\n",
            "\n",
            "Epoch: 0  batch: 54  final cost: 0.0066451174\n",
            "\n",
            "Epoch: 0  batch: 55  final cost: 0.0049999845\n",
            "\n",
            "Epoch: 0  batch: 56  final cost: 0.0039775753\n",
            "\n",
            "Epoch: 0  batch: 57  final cost: 0.0038313544\n",
            "\n",
            "Epoch: 0  batch: 58  final cost: 0.0032366703\n",
            "\n",
            "Epoch: 0  batch: 59  final cost: 0.0040184483\n",
            "\n",
            "Epoch: 0  batch: 60  final cost: 0.0026037905\n",
            "\n",
            "Epoch: 0  batch: 61  final cost: 0.003920512\n",
            "\n",
            "Epoch: 0  batch: 62  final cost: 0.0039712773\n",
            "\n",
            "Epoch: 0  batch: 63  final cost: 0.0025793107\n",
            "\n",
            "Epoch: 0  batch: 64  final cost: 0.0024085394\n",
            "\n",
            "Epoch: 0  batch: 65  final cost: 0.0027971736\n",
            "\n",
            "Epoch: 0  batch: 66  final cost: 0.0017877561\n",
            "\n",
            "Epoch: 0  batch: 67  final cost: 0.0018678831\n",
            "\n",
            "Epoch: 0  batch: 68  final cost: 0.0020474622\n",
            "\n",
            "Epoch: 0  batch: 69  final cost: 0.0020433024\n",
            "\n",
            "Epoch: 0  batch: 70  final cost: 0.0010128049\n",
            "\n",
            "Epoch: 0  batch: 71  final cost: 0.0011395921\n",
            "\n",
            "Epoch: 0  batch: 72  final cost: 0.0012543485\n",
            "\n",
            "Epoch: 0  batch: 73  final cost: 0.001075503\n",
            "\n",
            "Epoch: 0  batch: 74  final cost: 0.0010130225\n",
            "\n",
            "Epoch: 0  batch: 75  final cost: 0.0012615386\n",
            "\n",
            "Epoch: 0  batch: 76  final cost: 0.000992955\n",
            "\n",
            "Epoch: 0  batch: 77  final cost: 0.0009770697\n",
            "\n",
            "Epoch: 0  batch: 78  final cost: 0.0010414285\n",
            "\n",
            "Epoch: 0  batch: 79  final cost: 0.0017678423\n",
            "\n",
            "Epoch: 0  batch: 80  final cost: 0.0010590325\n",
            "\n",
            "Epoch: 0  batch: 81  final cost: 0.0007073082\n",
            "\n",
            "Epoch: 0  batch: 82  final cost: 0.0012024678\n",
            "\n",
            "Epoch: 0  batch: 83  final cost: 0.0004617093\n",
            "\n",
            "Epoch: 0  batch: 84  final cost: 0.00042127978\n",
            "\n",
            "Epoch: 0  batch: 85  final cost: 0.00035836006\n",
            "\n",
            "Epoch: 0  batch: 86  final cost: 0.00035391367\n",
            "\n",
            "Epoch: 0  batch: 87  final cost: 0.0005737521\n",
            "\n",
            "Epoch: 0  batch: 88  final cost: 0.00034568014\n",
            "\n",
            "Epoch: 0  batch: 89  final cost: 0.00025877455\n",
            "\n",
            "Epoch: 0  batch: 90  final cost: 0.00027281768\n",
            "\n",
            "Epoch: 0  batch: 91  final cost: 0.00032055497\n",
            "\n",
            "Epoch: 0  batch: 92  final cost: 0.0002492698\n",
            "\n",
            "Epoch: 0  batch: 93  final cost: 0.00024286385\n",
            "\n",
            "Epoch: 0  batch: 94  final cost: 0.00033061148\n",
            "\n",
            "Epoch: 0  batch: 95  final cost: 9.3574636e-05\n",
            "\n",
            "Epoch: 0  batch: 96  final cost: 4.9936076e-05\n",
            "\n",
            "Epoch: 0  batch: 97  final cost: 5.5081975e-05\n",
            "\n",
            "Epoch: 0  batch: 98  final cost: 3.480322e-05\n",
            "\n",
            "Epoch: 0  batch: 99  final cost: 7.6519034e-05\n",
            "\n",
            "Epoch: 0  batch: 100  final cost: 6.3823645e-05\n",
            "\n",
            "Epoch: 0  batch: 101  final cost: 6.831599e-05\n",
            "\n",
            "Epoch: 0  batch: 102  final cost: 9.735998e-05\n",
            "\n",
            "Epoch: 0  batch: 103  final cost: 4.339333e-05\n",
            "\n",
            "Epoch: 0  batch: 104  final cost: 5.482213e-05\n",
            "\n",
            "Epoch: 0  batch: 105  final cost: 0.00017465932\n",
            "\n",
            "Epoch: 0  batch: 106  final cost: 0.00016825358\n",
            "\n",
            "Epoch: 0  batch: 107  final cost: 4.4362845e-05\n",
            "\n",
            "Epoch: 0  batch: 108  final cost: 0.00011839381\n",
            "\n",
            "Epoch: 0  batch: 109  final cost: 0.00010655199\n",
            "\n",
            "Epoch: 0  batch: 110  final cost: 6.985762e-05\n",
            "\n",
            "Epoch: 0  batch: 111  final cost: 6.578691e-05\n",
            "\n",
            "Epoch: 0  batch: 112  final cost: 0.00011266745\n",
            "\n",
            "Epoch: 0  batch: 113  final cost: 0.00010360238\n",
            "\n",
            "Epoch: 0  batch: 114  final cost: 6.1874765e-05\n",
            "\n",
            "Epoch: 0  batch: 115  final cost: 9.9561934e-05\n",
            "\n",
            "Epoch: 0  batch: 116  final cost: 4.4729823e-05\n",
            "\n",
            "Epoch: 0  batch: 117  final cost: 8.449541e-05\n",
            "\n",
            "Epoch: 0  batch: 118  final cost: 4.3112363e-05\n",
            "\n",
            "Epoch: 0  batch: 119  final cost: 4.6236626e-05\n",
            "\n",
            "Epoch: 0  batch: 120  final cost: 4.570465e-05\n",
            "\n",
            "Epoch: 0  batch: 121  final cost: 3.3101263e-05\n",
            "\n",
            "Epoch: 0  batch: 122  final cost: 6.43744e-05\n",
            "\n",
            "Epoch: 0  batch: 123  final cost: 2.9791101e-05\n",
            "\n",
            "Epoch: 0  batch: 124  final cost: 5.9195863e-05\n",
            "\n",
            "Epoch: 0  batch: 125  final cost: 5.7609308e-05\n",
            "\n",
            "Epoch: 0  batch: 126  final cost: 3.709604e-05\n",
            "\n",
            "Epoch: 0  batch: 127  final cost: 3.2485128e-05\n",
            "\n",
            "Epoch: 0  batch: 128  final cost: 2.7236663e-05\n",
            "\n",
            "Epoch: 0  batch: 129  final cost: 5.1608426e-05\n",
            "\n",
            "Epoch: 0  batch: 130  final cost: 1.9979787e-05\n",
            "\n",
            "Epoch: 0  batch: 131  final cost: 4.8637765e-05\n",
            "\n",
            "Epoch: 0  batch: 132  final cost: 2.6261765e-05\n",
            "\n",
            "Epoch: 0  batch: 133  final cost: 3.8686187e-05\n",
            "\n",
            "Epoch: 0  batch: 134  final cost: 2.1743848e-05\n",
            "\n",
            "Epoch: 0  batch: 135  final cost: 1.8707386e-05\n",
            "\n",
            "Epoch: 0  batch: 136  final cost: 1.736047e-05\n",
            "\n",
            "Epoch: 0  batch: 137  final cost: 3.332631e-05\n",
            "\n",
            "Epoch: 0  batch: 138  final cost: 3.4887777e-05\n",
            "\n",
            "Epoch: 0  batch: 139  final cost: 3.2390984e-05\n",
            "\n",
            "Epoch: 0  batch: 140  final cost: 4.823637e-05\n",
            "\n",
            "Epoch: 0  batch: 141  final cost: 1.3473221e-05\n",
            "\n",
            "Epoch: 0  batch: 142  final cost: 2.8923527e-05\n",
            "\n",
            "Epoch: 0  batch: 143  final cost: 1.7531009e-05\n",
            "\n",
            "Epoch: 0  batch: 144  final cost: 1.7708937e-05\n",
            "\n",
            "Epoch: 0  batch: 145  final cost: 1.2723476e-05\n",
            "\n",
            "Epoch: 0  batch: 146  final cost: 1.4537953e-05\n",
            "\n",
            "Epoch: 0  batch: 147  final cost: 1.642809e-05\n",
            "\n",
            "Epoch: 0  batch: 148  final cost: 1.6668311e-05\n",
            "\n",
            "Epoch: 0  batch: 149  final cost: 2.3430752e-05\n",
            "\n",
            "Epoch: 0  batch: 150  final cost: 1.6078762e-05\n",
            "\n",
            "Epoch: 0  batch: 151  final cost: 1.4887337e-05\n",
            "\n",
            "Epoch: 0  batch: 152  final cost: 1.2812639e-05\n",
            "\n",
            "Epoch: 0  batch: 153  final cost: 2.342726e-05\n",
            "\n",
            "Epoch: 0  batch: 154  final cost: 2.3481727e-05\n",
            "\n",
            "Epoch: 0  batch: 155  final cost: 3.219587e-05\n",
            "\n",
            "Epoch: 0  batch: 156  final cost: 1.0602773e-05\n",
            "\n",
            "Epoch: 0  batch: 157  final cost: 8.908961e-06\n",
            "\n",
            "Exit: Epoch: 0  batch: 157  final cost: 8.908961e-06\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "CqegnGN0CgIb",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 459
        },
        "outputId": "a9c227d7-591d-4e95-c339-242ab6f1a7de"
      },
      "source": [
        "optimizer = tf.train.AdamOptimizer(\n",
        "    learning_rate=0.01,\n",
        "    beta1=0.9,\n",
        "    beta2=0.999,\n",
        "    epsilon=1e-08\n",
        ")\n",
        "\n",
        "## Run me\n",
        "\n",
        "gateDimensions = [[3], [4], [5], [6], [7], [8,8], [9,9], [10], [11,11], [12,12], [13,13], [14,14], [15,15], [16,16]]\n",
        "registerDimensions = [17, 17, 17]\n",
        "nramController3 = NRAMController(hidden_dimensions=400, gateControllerDimensions=gateDimensions, registerControllerDimensions=registerDimensions);\n",
        "\n",
        "# gates = [zeroModule, oneModule, addModule]\n",
        "\n",
        "gates = [zeroModule, oneModule, twoModule, incModule, addModule, subModule, decModule, lessModule, lessOrEqualThenModule, equalModule, minimumModule, maximumModule]\n",
        "\n",
        "loss_history3_3, model3_3 = trainModel(inputTrainSet, targetTrainSet, gates, nramController3, readMemory, writeMemory, min_error=1e-5)\n"
      ],
      "execution_count": 107,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Epoch: 0  batch: 1  final cost: 177.59534\n",
            "\n",
            "Epoch: 0  batch: 2  final cost: 73.0798\n",
            "\n",
            "Epoch: 0  batch: 3  final cost: 31.641598\n",
            "\n",
            "Epoch: 0  batch: 4  final cost: 24.414001\n",
            "\n",
            "Epoch: 0  batch: 5  final cost: 3.3741121\n",
            "\n",
            "Epoch: 0  batch: 6  final cost: 0.8009982\n",
            "\n",
            "Epoch: 0  batch: 7  final cost: 0.17930831\n",
            "\n",
            "Epoch: 0  batch: 8  final cost: 0.03978226\n",
            "\n",
            "Epoch: 0  batch: 9  final cost: 0.008012021\n",
            "\n",
            "Epoch: 0  batch: 10  final cost: 0.00093557243\n",
            "\n",
            "Epoch: 0  batch: 11  final cost: 9.7647906e-05\n",
            "\n",
            "Epoch: 0  batch: 12  final cost: 8.510944e-06\n",
            "\n",
            "Exit: Epoch: 0  batch: 12  final cost: 8.510944e-06\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "colab_type": "text",
        "id": "3mkEhLYs3Lhz"
      },
      "source": [
        "### Plots "
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "by3g9kxxBYq9",
        "colab_type": "text"
      },
      "source": [
        "#### Plot of MLP with different dimensions"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ad4KzlE8BBfv",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 295
        },
        "outputId": "a39bd6b5-b195-4964-c7e6-e7363d7cfbbc"
      },
      "source": [
        "# plt.plot(loss_history, label='LSTM V1')\n",
        "# plt.plot(loss_history2, label='LSTM V2')\n",
        "plt.plot(loss_history3, label='MLP-Dimension-400')\n",
        "plt.plot(loss_history3_2, label='MLP-Dimension-100')\n",
        "\n",
        "plt.legend(loc='upper right')\n",
        "plt.ylabel('loss value')\n",
        "plt.xlabel('batches')\n",
        "plt.title(\"Training Loss\")\n",
        "# plt.ylim(0, 10)\n",
        "\n",
        "plt.show()"
      ],
      "execution_count": 126,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYgAAAEWCAYAAAB8LwAVAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "515TghboBrR4",
        "colab_type": "text"
      },
      "source": [
        "Higher the dimensions in MLP Controller faster it converges to a lower error rate"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "IaY87MG1DTrV",
        "colab_type": "text"
      },
      "source": [
        "#### Plot of MLP with different learning rate"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "oUjZmVzaC30T",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 295
        },
        "outputId": "4c6229fd-f659-43c7-95c1-805b827b34fc"
      },
      "source": [
        "# plt.plot(loss_history, label='LSTM V1')\n",
        "# plt.plot(loss_history2, label='LSTM V2')\n",
        "plt.plot(loss_history3, label='MLP Learning Rate=0.001')\n",
        "plt.plot(loss_history3_3, label='MLP Learning Rate=0.01')\n",
        "\n",
        "plt.legend(loc='upper right')\n",
        "plt.ylabel('loss value')\n",
        "plt.xlabel('batches')\n",
        "plt.title(\"Training Loss\")\n",
        "# plt.ylim(0, 10)\n",
        "\n",
        "plt.show()"
      ],
      "execution_count": 109,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYgAAAEWCAYAAAB8LwAVAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "MIQGZo1NBSuW",
        "colab_type": "text"
      },
      "source": [
        "#### Plot of different type of controller"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "KEASgZnQhfk7",
        "outputId": "e4069b44-493e-4a5c-ce29-8bfbd162415c",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 295
        }
      },
      "source": [
        "plt.plot(loss_history, label='LSTM V1')\n",
        "plt.plot(loss_history2, label='LSTM V2')\n",
        "plt.plot(loss_history3, label='MLP')\n",
        "\n",
        "plt.legend(loc='upper right')\n",
        "plt.ylabel('loss value')\n",
        "plt.xlabel('batches')\n",
        "plt.title(\"Training Loss\")\n",
        "# plt.ylim(0, 10)\n",
        "plt.xlim(0, 100)\n",
        "\n",
        "plt.show()"
      ],
      "execution_count": 131,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAY4AAAEWCAYAAABxMXBSAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nF3RR/F5vAAAAAElFTkSuQmCC\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "4hz_Cv5jB0FI",
        "colab_type": "text"
      },
      "source": [
        "It takes longer for LSTM model to converge than compared to MLP. Also LSTM V2 converges faster than V1 because of higher trainable parameters"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "colab_type": "text",
        "id": "AEX1zI9rT4E9"
      },
      "source": [
        "## Test Model"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "ZoxIdVzC0cDl",
        "colab_type": "text"
      },
      "source": [
        "#### Test Model"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "kLBDPqwDlWdP",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "def testModel(inputSet, targetSet, gates, nramController, readMemory, writeMemory, epochs=2, max_timesteps=5, debug=False):\n",
        "  loss_history = []\n",
        "\n",
        "  for (input_data, target_data) in zip(inputSet, targetSet):\n",
        "        cost = 0\n",
        "        cum_prob_complete = 0\n",
        "        prob_incomplete = 1\n",
        "\n",
        "        registers = tf.concat([registerValueModule(0), registerValueModule(0), registerValueModule(0)], axis=0)\n",
        "        registers = tf.reshape(registers, shape=[1, registers.shape[0], registers.shape[1]])\n",
        "        memory = input_data\n",
        "\n",
        "        for timestep in range(max_timesteps):\n",
        "          register_data = []\n",
        "          registersNumpy = registers.numpy()\n",
        "          for i in range(registersNumpy.shape[1]):\n",
        "            register_data.append(registersNumpy[0][i][0])\n",
        "          register_data = np.array([register_data], dtype=\"float32\")\n",
        "\n",
        "          gateControllerOutput, registerControllerOutput, status = nramController(register_data)\n",
        "          output_registers, memory = runTimeStepNN(gates, registers, memory, readMemory, writeMemory, gateControllerOutput, registerControllerOutput)\n",
        "\n",
        "          if debug:\n",
        "            print(\"\\n\\noutputs\")\n",
        "            print(\"Timestep: \",timestep)\n",
        "\n",
        "            for values in gateControllerOutput:\n",
        "              print(\"\\n Gate Controller vals:\")\n",
        "              for v in values:\n",
        "                print(v.numpy())\n",
        "            \n",
        "            for i in range(len(registerControllerOutput)):\n",
        "              print(\"\\nRegister controller \", i)\n",
        "              print(registerControllerOutput[i].numpy())\n",
        "\n",
        "            print(\"\\nOutput registers\")\n",
        "            print(output_registers.numpy())\n",
        "            \n",
        "            print(\"\\nOutput memory\")\n",
        "            print(memory.numpy())\n",
        "\n",
        "          registers = output_registers\n",
        "          if timestep == max_timesteps:\n",
        "              prob_complete = 1 - cum_prob_complete\n",
        "          else:\n",
        "              prob_complete = status * prob_incomplete\n",
        "\n",
        "          # Update the probability that the computation isn't complete\n",
        "          prob_incomplete *= 1 - status\n",
        "\n",
        "          # Accumulate the probability that a result has been produced.\n",
        "          cum_prob_complete += prob_complete\n",
        "\n",
        "          # Cost for this timestep.\n",
        "          unscaled_cost = 0\n",
        "\n",
        "          # Add epsilon to every log to avoid having inf in costs.\n",
        "          epsilon = tf.constant(1e-35, shape=[output_registers.shape[0], output_registers.shape[2]])\n",
        "\n",
        "          for r in range(memory.shape[0]):\n",
        "              y = target_data[r]\n",
        "              yPredicted = memory[r]\n",
        "              loss = tf.math.reduce_sum((y * tf.math.log(tf.math.add(yPredicted, epsilon))), axis=1)\n",
        "              unscaled_cost += tf.reshape(loss, [loss.shape[0], 1])\n",
        "\n",
        "          scaled_cost = prob_complete * unscaled_cost\n",
        "          cost -= scaled_cost\n",
        "        final_cost = tf.math.reduce_sum(cost)\n",
        "\n",
        "        # store the loss history \n",
        "        loss_history.append(final_cost.numpy())\n",
        "        if debug:\n",
        "          print(\"final cost:\", final_cost.numpy())\n",
        "  return loss_history, memory"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "610N-4rUDwcG",
        "colab_type": "text"
      },
      "source": [
        "#### Testing - MLP - Dimension-400, Learning rate = 0.001"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "-D7DThMRmnmh",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "## Test\n",
        "test_loss_history3, outputMemory = testModel(inputTestSet, targetTestSet, gates, model3, readMemory, writeMemory)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "PMOYgIBwoXoD",
        "colab_type": "code",
        "outputId": "a7c6b394-fda8-4f4b-9525-c0daa4bff5ad",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 295
        }
      },
      "source": [
        "# plt.plot(loss_history, label='LSTM V1')\n",
        "# plt.plot(loss_history2, label='LSTM V2')\n",
        "plt.plot(test_loss_history3, label='MLP - Dimension-400, Learning rate = 0.001')\n",
        "\n",
        "plt.legend(loc='upper right')\n",
        "plt.ylabel('loss value')\n",
        "plt.xlabel('batches')\n",
        "plt.title(\"Testing Loss\")\n",
        "# plt.ylim(0, 10)\n",
        "\n",
        "plt.show()"
      ],
      "execution_count": 116,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAaQAAAEWCAYAAAApTuNLAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n/4Q8XQAAAABJRU5ErkJggg==\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "8rWTAnGnEAHN",
        "colab_type": "text"
      },
      "source": [
        "#### Testing - MLP - Dimension-100, Learning rate = 0.001"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "tECL_1XBD_Yx",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 295
        },
        "outputId": "98af75b8-ebe3-499f-af5d-7bb87015921e"
      },
      "source": [
        "## Test\n",
        "test_loss_history3_2, outputMemory = testModel(inputTestSet, targetTestSet, gates, model3_2, readMemory, writeMemory)\n",
        "\n",
        "# plt.plot(loss_history, label='LSTM V1')\n",
        "# plt.plot(loss_history2, label='LSTM V2')\n",
        "plt.plot(test_loss_history3_2, label='MLP - Dimension-100, Learning rate = 0.001')\n",
        "\n",
        "plt.legend(loc='upper right')\n",
        "plt.ylabel('loss value')\n",
        "plt.xlabel('batches')\n",
        "plt.title(\"Testing Loss\")\n",
        "# plt.ylim(0, 10)\n",
        "\n",
        "plt.show()"
      ],
      "execution_count": 110,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAaQAAAEWCAYAAAApTuNLAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n/wF4iFL/WLviVwAAAABJRU5ErkJggg==\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "VKimU7zLESpo",
        "colab_type": "text"
      },
      "source": [
        "#### Testing - MLP - Dimension-400, Learning rate = 0.01"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "FobmcDLcERwB",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 295
        },
        "outputId": "8f68f06b-e9a4-4d73-ccad-5eb2380287b7"
      },
      "source": [
        "## Test\n",
        "test_loss_history3_3, outputMemory = testModel(inputTestSet, targetTestSet, gates, model3_3, readMemory, writeMemory)\n",
        "\n",
        "# plt.plot(loss_history, label='LSTM V1')\n",
        "# plt.plot(loss_history2, label='LSTM V2')\n",
        "plt.plot(test_loss_history3_3, label='MLP - Dimension-400, Learning rate = 0.01')\n",
        "\n",
        "plt.legend(loc='upper right')\n",
        "plt.ylabel('loss value')\n",
        "plt.xlabel('batches')\n",
        "plt.title(\"Testing Loss\")\n",
        "# plt.ylim(0, 10)\n",
        "\n",
        "plt.show()"
      ],
      "execution_count": 117,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYwAAAEWCAYAAAB1xKBvAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "FRPVv0ExFk37",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 312
        },
        "outputId": "d22b3a33-520c-4fd3-a0bd-827355645680"
      },
      "source": [
        "## Test\n",
        "test_loss_history1, outputMemory = testModel(inputTestSet, targetTestSet, gates, model, readMemory, writeMemory)\n",
        "\n",
        "plt.plot(test_loss_history1)\n",
        "\n",
        "plt.legend(loc='upper right')\n",
        "plt.ylabel('loss value')\n",
        "plt.xlabel('batches')\n",
        "plt.title(\"Testing Loss - LSTM V1\")\n",
        "# plt.ylim(0, 10)\n",
        "\n",
        "plt.show()"
      ],
      "execution_count": 137,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "No handles with labels found to put in legend.\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAZgAAAEWCAYAAABbgYH9AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nIK+NAAAAAElFTkSuQmCC\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "RUKgJyzfFY_-",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 312
        },
        "outputId": "41352deb-5bab-4d32-8b1a-dc3c8b2b40f0"
      },
      "source": [
        "## Test\n",
        "test_loss_history2, outputMemory = testModel(inputTestSet, targetTestSet, gates, model2, readMemory, writeMemory)\n",
        "\n",
        "plt.plot(test_loss_history2)\n",
        "\n",
        "plt.legend(loc='upper right')\n",
        "plt.ylabel('loss value')\n",
        "plt.xlabel('batches')\n",
        "plt.title(\"Testing Loss - LSTM V2\")\n",
        "# plt.ylim(0, 10)\n",
        "\n",
        "plt.show()"
      ],
      "execution_count": 138,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "No handles with labels found to put in legend.\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAasAAAEWCAYAAADYRbjGAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz\nAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDMuMC4zLCBo\ndHRwOi8vbWF0cGxvdGxpYi5vcmcvnQurowAAIABJREFUeJzsfXmcJEWV//dlVffcDDCAnDKoeKAo\nrojrgat4wYLiesIeuiuru66uIuuu+BNFXe9jwWN1ZcUVvABFdJT7PuQcbgaGmWGYYe6jmbunu6sq\n3++PzMiM40VmVHV1V890fj8fmOrMyIjIzMh48d77vhfEzKhQoUKFChUmMqJed6BChQoVKlQoQyWs\nKlSoUKHChEclrCpUqFChwoRHJawqVKhQocKERyWsKlSoUKHChEclrCpUqFChwoRHJawqVOgARDSF\niLYT0YG97kuFCpMBlbCqsFshFSDqv5iIdmp//80o6r2TiP5W/c3Mw8w8k5lXd6fnRltfI6Ifd7ve\nbsB+Dta5DxPRovRZryWiPxLRNCK6QXsHDSIa1v4+l4iOJyImol9Z9b0iPX6V0NbM9PpXCed+SEQ/\nJ6LpRPR/RPQUEW0lonuJ6E3dexoVxhOVsKqwWyEVIDOZeSaApwC8VTv2i173b3cFEb0FwFkA3pk+\n+xcB+C0AMPNx2ju5FMB/au/k9LSKNQCOI6I9tGrfD2CR1B4zb0/rep/Vj34A7wFwAYApAJ4A8BoA\newL4MoBLK21410QlrCpMKhBRjYg+S0RLiWgjEf2CiPZMz80goouI6Gki2kxEdxHRXkT0bQAvB/Dj\ndDX/bSKamq76D06vvSjVEq4mom1E9CciOlRr90QiWpzWe26RhlLS/yOJ6Na0noeI6ATt3MlEtDBt\nfwURfSw9vj8RXZVeM0BEN4z2OQp4OYBbmflhAGDmjcz8E2beGXj9IIArAbw77XM/gHcA+FXBNRcA\neA8RTdGOnQRgJ4DrmXkTM3+JmZ9i5piZfwtgLYCXtnVnFSYEKmFVYbLhkwDejGS1fTCABoBz0nP/\nCKAO4CAA+wD4KIARZv43APcA+MdUG/g3T91/DeDTAPZGoil8AQCI6AAAFwP4BIB9AawG8LJ2O05E\nUwFcDuB3aT3/DuDXRHRYWuQnAN7HzLMAHAXg1vT4pwA8nt7TAQA+327bAbgTwNuI6HNE9MpU2LSL\nC5FrSicBuAvAxoLyNwLYBuCt2rG/A/BzZo7twunCYi6ARzvoW4UeoxJWFSYb/hnAmcy8mpmHkAiU\n9xIRIRFc+wJ4NjM3mfkeZt7RRt2XMPN9zNwA8EskAgNIJtN7mPmP6blvAdjUQd+PBcAA/ouZG8x8\nNYBrAbw3Pd8C8EIimsXMA8x8f3q8AeBAAM9k5hFmvqWDtgvBzNcBOAXAKwBcDWAjEX2diNqZY24E\ncCgRzUUitC4saZMB/CwtCyKaA+AvkWhcBlLt65cAfsjMT7bRpwoTBJWwqjBpkAqkQwBckZrENgO4\nH8l3MAfA+QBuBvAbIlpJRF8holobTazVfg8CmJn+PhDACnUiXfWv6uAWDgTwFJvZp5cj0QQB4GQA\n7wTwVEpqeHl6/MtItLkbiWgJEZ0hVU5EX9CJD+12jpnnMfOJSPxD7wbwYSSaTuj1jESgnA7gzwH8\nIeCyCwEcT0T7IhHaDzHzY3oBIqoDuAjAAADx3itMfFTCqsKkQToZrgJwHDPvqf03NfWxDDPz55j5\n+QBei2TCPUVdPoqm1yAxOQIAUm3jIH9xL1YDeKZ17JlIBR8z38HMJwF4BoBrkEz8YOYtzPxxZj4U\niTA7i4hebVfOzGcLxIe2kfqHrgZwCxKiRTu4EMDHAFzGzCMBbS0CMB/AqUgEo6FVpc/6QgDTAbyX\nmVtt9qfCBEElrCpMNvwPgK8R0SEAQET7EdFb099vJKIj0gluK4AmAOX7WAfgWR22OQ/AK4joL9NV\n/hkA9iq5ppaSONR//Uh8UBERnU5E9ZSG/WYAl6TkkFNSNl0DiS8nTu/rbUT0rFSz3ILEXOj4dNpA\nn9W3OhG9i4jeTUR7UoJXAXg1El9WMJh5IYDXIfX3BeICJM/0z6ARMtL7PR/JQuGvQoRfhYmLSlhV\nmGz4BoDrANxARNsA3I5kkgMSbef3SCb6RwBcgYQYASQkjPcR0SYi+kY7DTLzGiQr/+8iIQwcDOBh\nAMMFl/09Elab+u/R1Md2EoB3ITFp/RcSbWFpes0HkJgFtyDx4yiywguQkxFuAfAtZr6jnXuw8BOr\nb/+DxAf3L0io4lvTMl9g5kvbrZyZb2HmteUlM1yMRJu8gpkHtOPPRfIcXw5gvWbifGe7farQe1C1\n+WKFCuOLVLtaiyQGbDRCo0KFSYNKs6pQYRxARCcQ0eyUfn42EgLGvT3uVoUKuwwqYVWhwvjgtQCe\nBLAewBtQ+VAqVGgLlRmwQoUKFSpMeFSaVYUKFSpUmPCo97oDuwv22Wcfnjt3bq+7UaFChQq7DO69\n996NzLxvSNlKWHUJc+fOxfz583vdjQoVKlTYZUBEy0PLVmbAChUqVKgw4VEJqwoVKlSoMOFRCasK\nFSpUqDDhUfmsKlSoUKHCqNFoNLBy5UoMDQ0556ZOnYqDDz4YfX19HddfCasKFSpUqDBqrFy5ErNm\nzcLcuXOR5BBOwMwYGBjAypUrcdhhhxXUUIzKDFihQoUKFUaNoaEhzJkzxxBUAEBEmDNnjqhxtYNK\nWFWoUKFCha7AFlRlx9tBJawqVKhQYTfB0ztGcOXDa3rdjTFBJawqVKhQYTfBhy6cjw//4j5s3F60\nVdquiUpYVahQocJughWbBgEAjdZoNoLuHL7E6N1ImF4JqwoVKlTYTUBIfEO92Exj6tSpGBgYcAST\nYgNOnTp1VPVX1PUKwfjGVQtx2D4z8O6jD+l1VypUqCCgCzyGjnHwwQdj5cqV2LBhg3NOxVmNBpWw\nqhCMH9z0BABUwqpChQmOXuxS2NfXN6o4qjJUZsAKFSpU2M1ww8L1GGq0et2NrqISVhUqVKiwm0BZ\nAT/7u0dw9u8X9LQv3UYlrCpUqDCmeOVXr8ePbn6i193Y5XD9Y+vw+wdWdXz90o3bu9ib3qMSVhWC\n0A3qaYXJiTVbhvDVKxf2uhu7HE67YD4+ftEDbV2jZ4og9JBtMQaohFWFIAw387iNVlwJrgoVKowv\nKmE1wTHUaGHeg6t7rtnowmr7ULOHPalQocJkRCWsJji+cdXj+Niv7seflgz0tB/DzZxZtHWo0cOe\nVNiVEFdaeIUuoRJWExxrtuwEAGzZ2VsBMaJpVr3uS4VdB63K11mhS6iE1QSH8pdyT8L8cuhmwK2T\nUFjduHA9Fqze0utu7HKo/JvjCyODxe7Fr6gyWEx09DLXl47hhuazGp58Pqt/+Ok9AIBlXzuxxz3Z\ntVAJq/FFL9MtjTUqzWqiI9OsegvdZ9XrvlTYdVCZAXuH3U1uVcJqgkMNuInEBqzmnwqhaLWqwTKe\n2N1iq3RUwmqCI5oger0urCrdqkIoKs2qQrdQCasJjoxg0XOflWYG3MXnn5FmjBVPD/a6G5MCFXW9\nd5gg69yuoRJWExyZGXACsQFH05NmKzb8X73Amb99CMd+40bsmIREkfFGsxJW44rdTUDpqITVBIfK\n9RX3ZpfqDN3yWf31j+/C8866qgs96hy3LNoIAJWwAnDto+twynl3jJlPtGIDji92Y1k1tsKKiI4n\noseJaAkRnSmcn0JEF6fn7yKiudq5T6fHHyeit5TVSUSHpXUsSevsH0UbHyeiR4hoARGd3u3n0g5y\nzaq3MNmAnffm7ief7kZ3RoW+WvJUR1o9XgFMAHzwwvm4c+nTaIwREaISVhW6hTETVkRUA/DfAE4A\ncASAU4noCKvYaQA2MfNzAJwD4OvptUcAOAXACwEcD+AHRFQrqfPrAM5J69qU1t1JGy8C8EEAxwB4\nCYCTiOg53XsybSLzWfX2ox/ZhdiAtz+xEcefe4vX3NhXS4Z9s2KqZRgrM3NFsBhf0G5sBxxLzeoY\nAEuYeSkzjwC4CMDJVpmTAVyQ/v4NgDdQ8rRPBnARMw8z85MAlqT1iXWm1xyX1oG0zrd32MYLANzF\nzIPM3ARwM4B3dOmZtI0sKLhXHUjRLZ/VeODxtduwcO02bPMk3K2nmlWz17bVCYSxkimVZtU77G40\n9rEUVgcBWKH9vTI9JpZJBcMWAHMKrvUdnwNgc1qH3Va7bTwC4FgimkNE0wH8JYBDpBskog8R0Xwi\nmr9hwwbvgxgNaILYAfUMFr3W8sqg5kdfN/uiZNiPlelrV0Rc+awqTHBUBAsLzPwYElPhNQCuAvAA\nANGexMznMfPRzHz0vvvuOyb9mShswJFWbxl87UDRpX3PTGlWI81Ks1KQZNW2oQaGGqN775WwGl8Y\nqQF3L8VqTIXVKpgaycHpMbEMEdUBzAYwUHCt7/gAgD3TOuy22m0DzHw+M7+MmV+LxP+1qI377ipU\nUHCvlRmdgtzrvpRBaQm+ftZTn1VFsMghaVZHfv4a/OV3bh1VvZWwGmfsZgJKx1gKq3sAHJ6y9PqR\nkBnmWWXmAXh/+vtdAG7gxMY0D8ApKZPvMACHA7jbV2d6zY1pHUjr/H2HbYCI9kv/fSYSf9Uvu/JE\nOkCedb230NPm9FrLU1i0bhv+8YL5DpGi3AxYaVY2fDJl6cYdo6q3Ilj0DrubZjVmWdeZuUlEHwVw\nNYAagJ8w8wIi+iKA+cw8D8D5AH5GREsAPI1E+CAtdwmARwE0AXyEmVsAINWZNvkpABcR0ZcA3J/W\njU7aAHApEc0B0EiPbx6LZxSCiZLBQp90et0XhXuXb8J1j63Dhm3DOHiv6dnxTLPyCFXFBux1cPJE\nQhVntXtgN5NPBsZ0ixBmvgLAFdaxz2m/hwC823PtlwF8OaTO9PhSJGw++3gnbRwrle8NkuHX6xVq\nPAHNgMo0afdH9VWfJxutGJ+69CH863GHVz4rAWMlUyph1TtMlO+0W6j2s5rgUJpVr3OsGZpVF+pj\n5lHHhDRTn5P9UbYyn1V+4oEVm/Hb+1bhqYFB7DGtD4CdnHf3xlCjheUDg3je/rPE8xUbcPfD7vbs\nKzbgBIeaznudY03nInTDZNSND6mVaVBmXZLPSpWNiFDfxXxWf3hwNX5+5/JR1fFvlzyIt5x7C7YO\n5bs86wugSljtHtAXgLvbs6+E1QTHhNGs4u4GBXfDrNn0CSvBPKiORVHusxpPNuDWoQY+P29BR1Tw\nf/3V/Tjrd4+Mqv27lyVproZG8vaHmmOfSX88zNcX3L4Mc8+8HFt2NsoL7+bQ54leL3C7jUpYTXCo\nKPReD7xuz+vdSB7REnxTyd8uwUKVqUWU5wYcR83qe9cvxk9vX4aL7n5q3NrUodbb+rPaMZwLqzHT\nrMYh8Ppnqda5buvQmLfVC3zy1w/igxfODyqrLw52N82q8llNcGSaVa8JFsyoRZR8AF3oSjdW3I3M\nZ2XWlfus8mPq+UVEGhuwu8LqyY07cNg+M8RzarHRq0VHHgKRtz84kqejGjOCxTiM22iCMGbHCr+5\nd2Vw2ValWVUYS1z1yBp89/rF4rnMZzVGK9Tf3LsS2z1bZdyyaAMeXrkFQPIR1CKVp3Ci+azM42rS\n0gW8mjSJKA8K7qKwuv6xdXj9t27CFQ+vEc/3Orhbal9PNzVWZuaxXN0PN1sJUSf9Snq9oJsI0J93\nr10H3UYlrCYA/vnn9+G/ri1OkjEWK9T7n9qET/76QZx12cPi+ff95G689fu3Ze0rYkI3utKND8nn\ns1IfrH5UaV81ylfi3RRWj6/bBgB4eNUW8XwkaDbjidwMqIcgjH04wlgJqy07G3jeWVfh+zcsGVUs\n4tahBk787q1YnL6/XR2mZrVrEIhCUQmrCQ419MZilTSYOtvXbxsuLdtqacKqC213Q/i2BCIFIKdb\nUj63iCjTxEIJFiHsx1o6Y/om52wTzR4tdknQrGLjd3c7tnlwBMsHdoyZsNq4PRmzl92/Kr+3Dkbm\nrYs2YsHqrTjnup5lVOsqDGtCpVlVGE+0uuTraLRirHh60DjWzvzUYs7MZxNGs2qVsQHdDzeKKDse\nqlmF3K8ykfrMtb3ORJImmjfa1yf3R9dsxXeuk03RneCN/3UL/uKbN43ZhJm9W9KSPXfQ1ERJHdYt\nVD6rCj2DGm+jXfmePW8Bjv3Gjdg8OOKcC4nNjbvts+oKdV0OCs7irLRjanKLKH+WoQSLkJ4qrdP3\nnnrtV8n3RdN9Gvn5f/nFfTjnukVd0+CV5jNWBAtNVnVlIbC77P2kC6tKs6owrlCT7GgJFjc/nuy3\n5duQsAy6z+qRVVuwavPOUfWnGx+S12elmQG3DjWwbaiRCTDdDBiaGzBEwNRSrdN3X9EYzoXb0nss\ngmq/zPTX7eltrCZM/X1GozAD7m6cDP1xV5pVhXGFL0tDu1BCr9MMRzob8Fd3r8Crv3bDKPszqsuT\nPnnMgOpeY2a8+PPX4MjPX5MJsCiirHxXzYBUHA+nJtR2NZeQyf7I9B6LkPvMikkV3db8xoqdpgRT\nRKSFd3St+l0WOqmi0qwqjCvUeHvq6UF87Ff3d5wpXA1bPR1L6Eq00YrRijmLTwpFoxV7J6jualZy\n3YZ/JmMDUnY8VFgFaVbpo/Hdb6cTarcYi7lfRxMekmbV5fltrHwoak4mkgXxZIVu2m3uZvu1VcJq\ngkNNLjcsXI95D67GLYs2dlhP8m8nitWO4aahWYXi8M9ciQ/97F7xXHd9VhbBIvNZCQQLzWfV6OBj\n9jEDaymDwTchd8pY69Y2JkpYGjkehXJjqlmV1P342m3Y4Yn5s6EHeY+GYJFh93BZ7dYZLCphNcFh\nT+qdTiZqkuzEDLhtqIlY81m1g+seWyce726clVy3RNNOzIDJ79AehNCBlWbV8sS2SOmOQjDU6JJm\nJWgfRWOpW/tbhWpWrZjxlnNvwT95Fjc2OHufuj+wA59V21dMbFRswAo9g5udoUNhpTmk7WNlTKht\nQ8Wa1c6RVttaSlfirFpKKMkC3ZduqWzbextGvFaJZuXjwUQdUta6pVlFmWal+6zcvrT7bMoQutJX\n4+fuJ59uq14CZYJ4NFavdpdh24YaOOPiB4KS5zLzmG1uqcNeAFaaVYVxhW/7i3aR+aw6uHZ7agas\ne3xWL/jcVTjlvDvbqnMsfVaSGTDWzIA6ASMEhjbimRCV1unVrDr0WXUrf2EkalZuOSlV1WgQSqVW\nwkptjBlaL2lxVuM5OV94x3L89v5V+NHNT5SW/Zsf34XDPu3sF9t12AupXm/Y2m1UwmqCw14tdWwG\nbNP0pWP7cMOgrku4d/mmtursRiYYn89KSmSrjtUialtr0Iv7JoAoW937zqu62mu8ky1FiqB3r4gN\n2K15P1xY5e8nBKqfpFHXx5NgkS9Oytu8/YmBse4OgLwvJx91IE4+6kAwd4+BuXH7MB7xpBIbL1RZ\n1yc4XFp2pzV1bt7ZORKjFYdPJCHoZroln6lU0iJ0M2Do5MaaYPVPTlx4vtN0S93SrEgQpkVxVj3T\nrALHmIo7jAiZatWJZtWpeS7LWDKBTG3q/o84YA8jy39/F77bk757G9ZuHcKyr5046ro6RaVZTXC4\nJq7RaladOKEZcdwZwcIH38Syc6QVvImeL91SUSJbPSg42GclmBNtqMNlE2a7r2+4SwSLfBuNEmEV\nq3JdadYUVgWVKoq+8v2F1quzAUezAKI2mUftaFbjBd16kAvT7oyftRNgr7BKWE1w2JNjp9+juszM\nDZeg7DtlTj6EbmpWPqH7hm/fhJd8oTjAVaGzbe078Vnlv30raVVXWVBwr8yAJGgf0u2r/nVNs9IJ\nFgVZWJRm1Rfos1KTcESdB1yPBsp/20n4w1gh1gR4WWLlXRGVsJpAKGJn+f5W2D7cLPxYVd2dDN2Y\nuaOg4CL4PqLVW8JXcN7cgEIiW4m6Hvodl2kjeh/8ZkCzXCi6TbBoldxLt3JRKoRqVu36rHKCBWVJ\nejszAyb/trsMm5CaVZw/wyhyzc43LFyH+cvC2JYTEZWwmkCQHd7lZbYNNfCis6/GN6953F93dn1n\nH1cnQcFFaGcV/PSOETEaX2kxtrYSC4JZX3VmE3FgF/Su+iYnVWcpwaLN5z/S6pZm5W4RImpWAsFi\n21ADH7pwPtZ3YAoyfVZ+wav7rOKYsWzjjsJ61btPXFbjLzhsn9WPb12Ke5f3VhDoKcWyXJDaM/nA\nT+fjXf9zx6ja6OWGjpWwmkCQV7rlmtXmwcTHM++B1f66hUDZUDAnE8FofFZbhxpG8ttQ/8JQo4U/\n+89r8aXLH3POZT4raw5U1iYfG7Bd7UHKhOGDV7OCu9INgTS/P7JqS9vsS4neXaRZ6UL10ntX4ppH\n1+H7Ny5pq027vSKLmdpbrBYRfnjzE3jdt27C42v9GyLGmWalmTg97/PqBWtx3Ldv6jj90GNrtuIX\ndy03jimKvarzS5c/hnf+cHSCYLRQY6Wu+ay6TV/vJR2+YgNOIEgTWQgbMAt4LVh6jGaIxcyIR+Gz\n+sFNS/CNq0ytL3QVvGwgWWHfuniDc87ns5LZgPnk1q5JVAoutqGOb97ZwPbhJmZOMT+tTs2AUnsn\nfS/ZvbkdZlYkTOjFPqv8mBIk/R2YgUN3rm00lc8qwj2pqWr15p143v6zxPJNTVOWmI46PnXpQ9g8\n2MC2oSb2mtEvliny257wnVsBAH/zikOzY2XptXoB9XxrRGi1See/44kBHHPY3qXfeLPF6KuNrp+d\notKsJhCkgWV/gEVlakVfnKBthII5+Sg71axsQQWEf0TLNiYbRh46Z4ZzTn2c9nwhJbKVzIDBmlWI\nGTCdh5es344XnX01lm7YbpzvNBaoWwvZ3AxYrFlJQcGKkTilrwNhpbcXqFkF1SuwATvxs3W6N1uv\nCAyNVoxF62SNUz3fxAxIxrEi/GnJRpz6v3fihzeVa87dYhd2gkpY9RhcstL1ZWcwj+W2am872b96\ne2EfWswpdb2rBIuwcstTzeqZe093zmU+K4+pVL/XpibQ4zbp2Xrmc9/kZB9daxFFqEOfVbd2ss3N\ngPmxorGkT/y5ZtX+klpnAPomuhVPD+Lvzr8bQGLCCnlETc0MmKeSksuG1NfuMky9l/HWrL70x0fx\n5nNucfaTG2nGmqlby1UZcPOr07qWlvgJgd4SSiph1WM0S3wI9uQmTV56wKsPmelLGGtlMSaMlLre\n6WZYAto1A+4x1bVY53FW5nFJGEm+mhDB8fjabXjtN2/M6ykxA/rK5dT19tCtuUHS7KT7z7gnuhkw\nFdb99fanixWbBvO6PWW+eXWueRuaVcFwa2XU9dwMWOb8H+3w1etX42m8J++7lyW+Sn3H743bh/Hc\ns67E+bctBWBuSBlCiFDvOuT77qXZsxJWPYa+A3CYGdCtI8QMyNa/bYFTNmBgDEwIQk0267YOp+Xd\nc/44K1cw54HC3FbqqQdXbhbbdOAxRSrkuQHbewNF5duKwRLYYaGa1XDmT2r//d+5dABHHjQbgH9x\nMFUzL9ZrUdB70TWr0KDg0ZpU9Yla1TXecVbSM1zxdLIguOy+VQDMoOCQ8ZaxCEOE1Sh3LB8NKmHV\nY+imEclK4kslpCOz3xeZAbMVc/uDLTMDCvWPhgofgsGRZtYHG6Xb2gssPmZ5QvbBvmOfyb6MCCNR\nx0NQVH574N5PgI9g4desJILFlA486zEDxx6+DwDgM5c9gpseX++UmarVm5gBU0FUVG8bBAsFSZi1\n8z6Mb7UkVGGsoe+U0GjlzwJInmFZrkodIQQthcpnNYlRplk5SVpj929lay5yKUksr9DPjKEIFm4D\nnX6roRrGzpGWtx3fhyMF/eoZ2jMtM8SXYa02/WZA829Hs/KUK0PRYmDbUDvCyqXOS33JhJV2UpkB\np6QDbNOOEbz3R3c4fjkfZqTMyIVrt+Hv/+8e57wurHQzoGSe/vy8Bfjkrx8046wCtdai8yHplnTN\nSr3fXpnF9O4q7U4di0gPCg4RVqrO8mfQS59VRV3vMRrCak2Hk/bfGizfuW4RvntDwuIp9lllv7Rj\nYQNPUdel7Rs6zXQQrlklwkry1Snn/b3LN+FNRzwD0/uT4SxlsNDNgO1kFrfv2E+wkLW7rB7KS7aD\noj5uGwrLoai3b5oBhcWRkPBYmQHVBLh4/Xbc9eTTWLh2K/afPbW07TLzki6s+mpRoWntp7cvAwA8\nf/8XAFBZ15NzZSYqkarfxuvQySLquvGevKX+quel3k8tytMtFXXv3OsW4YDZU7PvpPJZVSiE6fh3\nz9vKgz1Y7lmWB4cWCiv1r76yDtTomf0ZLPzph0pMMu0KK1GzSg5eeMdyfOLiB5w+6Zdk7bEc+OqD\n/Uj9GSysv60DucZS2qR1XXc0KynLg1SzFDCtYqDy/Iex93oJvpCHVZt3YqjRMn1WoVnXMzOgSx7Z\nNtTA7Us2OteMNoWUtLDsNNC4U0g7fishrYRNFJG2c7X/ns+9bjE+denDxl5vZajYgJMYTWO15g4E\nX0ZxBV3bKYxRYeMf53cRmBkxyxOJb/ItmxeCzYApiSCOGT++damRhkd/Fo+s2qrVrTqR15OnZipm\nRtoIFVZ2ZT5fY7tU9GLNqg1hJZjKZJ8VO+WUz0qdy/1/Yfci+VKZGa/+2g348M/vxTTdZ1XLqetF\nc6cRZ2Ul6T3jkgfx1z++C+vS9FBSCqlOIOU57LamsW2ogV/ctbwtX/BIZgZMNas2/HgAsGxg0Li+\nCL1M3DumwoqIjieix4loCRGdKZyfQkQXp+fvIqK52rlPp8cfJ6K3lNVJRIeldSxJ6+wfRRufIKIF\nRPQIEf2KiMptHR3CpK675+1J3TZ16AKkmA3oTtBZVofAPralWZXUGTrmFcFi21ATX7r8Mbz3vDyl\nje6z0m9drRT1vrXifMJtj2Bh3rM/g4X5t20GlDTbEOjtfe/6xcYk1p4Z0PVhFCeyzY+NZJpV8rcv\nzZUPEolQjakbH99g+azyKalo7lR9SNiAZmohFZBtbzVTROMO0edM/3Lybyvu7pb1X7niMXzmskdw\nm6AZAtAEed7j4WayoFPalGkGlPuma4QPp5sqhgRk75aaFRHVAPw3gBMAHAHgVCI6wip2GoBNzPwc\nAOcA+Hp67REATgHwQgDHA/gDoffgAAAgAElEQVQBEdVK6vw6gHPSujaldXfSxkEAPgbgaGZ+EYBa\nWm5MoA+aoslDwU4GqgfqFqZbyjSr4pW13MeknJR1vWMzYIjjN2YMNZTJSZl4mtk5/dnoJlCZuq7O\naXFYpT1oxwxoaVY+M+AovvVvX7sIdz2ZJ0ttj2CR/GsEBYvCxtWa1GRob4MSqh3XhHGjBGBEpvkp\novxd7xhu4pJ7VngYsO5CRT3zaf2J8LOp/aLPKugOVJuasEp/N1px12LhAGTjXYVs2FBN3bl0AMd9\n+yYMNVrZNeobMKnrcjtbtbGT+bwCJPbu6rM6BsASZl7KzCMALgJwslXmZAAXpL9/A+ANlCwBTwZw\nETMPM/OTAJak9Yl1ptccl9aBtM63d9gGkBBPphFRHcB0AP4MsaNEo4QN6GhW1mDRY1+KfFbSBG43\n9+dfuR5//b93Otfqedjcet22mgEfcEiw4k5tssmp92afFPQPLaeua8e07UTaTbekwydk7cO2UOvc\nDGiWD915d+2WIVyQkhGA/N0tH9iB8297Uqw7OWb+C0ialZzmygdJ41d11muRd0xe99h6/MelD2Hl\npp2w0dLGc27ySs4ps6Jikiro9/vRX96Hr125MD8ZNFHL1PWicdSuJjJ7Wh8AM+hXhxpHX/zjo1i6\nYQee3Lgju09dWEn7l+nQtc6mRX2XUFbfeGAshdVBAFZof69Mj4llmLkJYAuAOQXX+o7PAbA5rcNu\nq602mHkVgG8BeArAGgBbmFncDZCIPkRE84lo/oYNbqLVEJirNfe8Pak7PitNnSpS4yUzVE5ZTf5d\nu3UItz8x4FzbzPK2lfcPAJ7zmSvx8Yvu9/YFCBv0g9pk07IEjH29/qHlWkxeRt9OJDsc8N051HUP\n46wsg0WnZsCi8kWT5AcvnI+z5y3AyjSDhLqLH92yFP/5x0exdagh3r707BQbUL3rfKEQdjOSrzML\nNNay4Ns1KoEmrebVMWNzx/S3MiuqxY4qod/THx9ag/+5+Ymg/tttJnXlx4reUZGPh5kx98zL8S0t\ng4cSVls9u2WrpnQzn7pPnbpeFhSsCyt9bzAf1PdVBQVPIBDRXki0rsMAHAhgBhH9rVSWmc9j5qOZ\n+eh99923o/bKqOv2d2p/uPpEEEJdN7ZoD5w5G4Iz21eHmmCufGRtYZ1lbTOzsTK2tzhxYqx0zcoq\nq7fXrmZlP1Fp4ly4disuTbMH6P03/zb7EYqi+K0ieb9554hR3p6IIiLRvFaUG9BmA4YusiWChaFZ\nWWNSNa2Ej7SwUYsG5txnpMaIElbKPGbfm452/E0PrdyCX971lNFWsxWuWbk5LJN/9a1XMs3KI6wy\nkKrTNXcaPqsAzaoRl5sBs/CA3TQoeBWAQ7S/D06PiWVSk9tsAAMF1/qODwDYM63DbqvdNt4I4Elm\n3sDMDQC/BfCqNu67LZTvL1SiWYWyAVMYJpfgPuYb4zkZHaxK1gVu0FemWTEDg43crq4WdDll2Lxe\n71f+zDTNSjC3hty/LZylj/X4c2/FY2sSNuLlH3tN0l+rWKealf3+GyU+TgX1pjKHvHUfiuFpwza3\nAgUEi1CflTDLKD9YX83SrDRhJcXLKRhB3pZwU8JKtRESkO2ObBf/8ZuH8P8uezitK9cyix6DTcrY\nPtzEf9+4xGs+VGZ9mxySwbpEF1bqPdW0oGCf2VoyAxbNH2qxs7v6rO4BcHjK0utHQlKYZ5WZB+D9\n6e93AbiBk5E5D8ApKZPvMACHA7jbV2d6zY1pHUjr/H2HbTwF4M+JaHrq23oDAHfnvy7BnHzc8/bK\nyJ4sDYJFWIhKhqJVpX5OH8z2Ct2uw84G7UMZwaLFbJoBY3t1b15vECy0iSy/Pp9g29Fy7EmszAwy\na0pf1n8d6jk9uHIz5p55OZ4aGHSuDYE+XnStwka2M3H6t7TIKFocGZpVU6auB/usBOaPMgPWo8iQ\njPrwbngSFevXx8yZZqb6NTVNuFvks1LodOrN2YBxoR/Stpx87veP4JtXP45bF2+QNcb0mNpQ1egr\nuy0xcjPgSBYcrMWeeRQhU1iZ1HcJGUFndzQDpv6hjwK4GslkfwkzLyCiLxLR29Ji5wOYQ0RLAJwB\n4Mz02gUALgHwKICrAHyEmVu+OtO6PgXgjLSuOWndnbRxFxIixn0AHkbyjM7r+gNKUR5nZf5tD/C+\nQDNg3ob8u6hcRrAQNCt7Ul6zJUxYlREsYssMaE+QvkSxehm9a7pDXiKb+GA/0kYrxtIN2zHvQZlz\nozRd3/0tT4XUjUKOPABYsHqLQUm36xmxsij4HqNNVbfHhq7BmMeTf29dnFOnG5YZUJmFH04F7/xl\nxdu5SwSLTFjZmhV04WO2q0NpEzq7U40Jmw2ori6yYKku3rl0AIs9+0XpyJ5FS9ZQFZrW+1J7Ue0x\nrU98/qq4pFn5NOGdI8mNjWgLgLItQrYaZsB0MRris+qhZjWm6ZaY+QoAV1jHPqf9HgLwbs+1Xwbw\n5ZA60+NLkbP59OOdtHE2gLOla7qN0gwW1mBz4qxqYQQLBclnJV2lt5IRLALYgKs3J2bAZEdefz/K\n4qyYLYKFk8nDPKBPxpmvQ3e+awQLSZj54AjnmHH8ubdipBXjbS850Clf9zi27bakVTUz48Tv3oZj\n5u6NS/75lWk9ZpmmpYn7zKmq3z4zYGI+kzQNxp+WbDS27ci3w0j/Tn/cvCghFV2/cD2Onru32A9A\nNgNm247UIqMf+u0UUeQzQcS5OS0TVqkZcLBRrlnZqsop5yVs2LJdmNUiYvtwE9sLQghss+2a9Pvo\niyJPYt3k2FYhhk7SpGNmDKXmTrUAqOmalWeQ636uZgB1XZ3aXX1WFQIwap+Vrlm16bMqWiTpH0W2\n8hIzWJiVbNg27LQjQX2oj6zagmsfXefWy5wFBANufFmROS43V+XXNDOHfC6wy5zrTw0MuppVzE5G\nBx1q8eBQ1+FOMjaU2evep/IUWvZ1tu/Na8rM+i1rVonPxL0sZmD9tiHnmN5n3V8ElL9ryQyonqGj\nWYnmR7fOwSzBcf6ERloxVjw9mMUDDg6HxFkVdz4kvdaS9dvFMoCriQzsGMn6rb+7O1IWbmZBENr1\nva+h9FnoWT0yNqCn/yOaENUtJz60k8V9rFAJqx7D2PY7wAzosAED46wUnh4cyYRDoc9K76PHASuZ\nkkY8KtPLDt0Lc+fku/2qj+ik792GD1443ynfinMzIFFuHlEo8lnZmgBgxl4pGVb03T24YjNe+80b\n8fM7nzLb1SqV7tVHGbafk2ROkfYls4vpPhCfKQ8o16zYI+iY2fHT2emVfDs0+yASLBoq60K+f9Uz\n955uECaUViJNkHkarrwfv7zrKRz7jRszgZBp5paw1VF2Cz7quf7dLl7vNxv6gv5jZkOQnPq/d+Ke\nZU8XLyBF0cpGPCKQjMEy4aIvenLGqL9tJcgq6vokhmECEfezKtOsNDNgAMHin352Lz544XwMbB8O\njuFRE2QtIuNjeXrHCN75P7cb1/kSe771xQcYwq6MYBFzPtnM6K+XxptJPitpZ1eD6lzQhyfTHIT3\nLt9kHNc/1uGme69K0/WxAX39B3ITi34vtkBoNC0zoOcesv2zrL/1a30rdRt5jFvaT6VZii27kBZR\nStD31XIKvdrLStUbYgZssashKp/fzoZpnitaDNo9VHX4Fl9ZpvKIsGhdmGal91N6/rcvGSgck7q/\nVWc4jljj0NgpWHrHMRvfqR0UvHrzTsdnlmdAmeDCiogOJaI3pr+nEdGsse3W5IE5gP2rbQWHDdgm\nwUIN7JFWnA984TqDnBCbg1nh1/euzMx+ef/kwRxpKWCAcoIFc75inDGlVvocpNyA0v5DOikh5LNz\nzYB5u9JOvRnBokSzku5fohAXaWRFZkBVRU4mMcvFgv8jbdH1b1lCo6k9y+SK4icp7YM23MjDITIS\niBUgXMQGzLPxu/ehnt+O4RA2oNx3xWptCAsSVVdfjbDntD5s2iFnmwD8BCpmFnYBd48ZfRVOsbBg\nKdspuBkzRlqMKSlrsmnFWb3qazfgbd+/zbhml6CuE9EHkbDjfpQeOhjA78ayU5MJZT6rMse8sQpv\no93EdxMGXxyGqB14zASkrfZ81+o46ovXYuHabahFhKl9rrAqymCROdzZnSj01WknCUibrXyrlOGG\nYAb0BGOW7XcFaAmDhXtRaFhxO+zxd9tbgtit6RR+87i7eMk0nMy82t7zk8zHuc8qT7eUCa7070aB\nb1DflNMeSqr4oENdd/uWaVaWgF6thJUwnhNBk47piAqtBAZ1Xc/LyPICokizkt6ZJOCSnYKT39J3\nFnOiWfWnwkoV+coVC/HVK5IoneVWaMWuEhT8EQCvBrAVAJh5MYD9xrJTkwmmHbv4PJBMHOu2DmVx\nOmWamQ9cUl4/pSYNe9KRBJNPCOn7Dqn2y3Dn0gFM76shIndCsCcRvWeZ2UrXrDKfVRgb0Lf7bLMV\nZ9qsZAZUz8iNszLLiZqVYAZ0fFZWnJXfDKjaycva/WnXh6P6UqTxSLDHzUgrznxWelBwklUjL2fn\nINTvQWk+0gSvxqCt+YravOeG1UJE8lkpQaOS8BbtaaV/I/pE34oZK62YRN2PJeu8rg9SElZRVLxT\ncDNmNGPGlHrNOfejW5aK9zER0i2FUNeHmXlErbbSLBC96/FuBn0wydml3Q/xFV+5HkBCry0Tdj74\nMhhI/ZJW/Mlx9yP1OaSTrbbzv0Ns39uHmpg1tZ4QLEp9Vi4pwTCZaaSKMp/VOdcuwneuX2zUpdCM\nGf21CMPNOMuQIPXDzbrualbXP7YOT27cgX889llJ3ZIZ0PrUbCp02QIl2yhR0NB9iyOfMTmniCuf\nlZIihV1wCBaNFms+qzzdUuITzWkESigWBSFLx9Q9u2a2sOuTOpIT85e7MWTqudcoSWtUZBprCqw7\nAPj5nctx+cNrjLKtGKCCTRN91Hu7rOqX75pWi9FoxZkZUMI+M/uNv3cVNuDNRPT/kGQhfxOAXwP4\nw9h2a/LANg3YYDb9UvbKRh+L7eadK1pG62d8+1lJJpIizUoXdiEmpJ2NFqb3p5pVgM9qy2ADl92/\nMvexCCZWwwzoaVcJKqmfzZgzv5Sde06hJpiGXGEBnHbBfHzp8se0Y0mhTYMN/P6BVeJ1DcMH4l/s\n2EHBkvnIp1n5XJ+2ZhU63Gzq+kgzNgJYVT1RRCm7T7VjmgGlxZHEaszYoM47kO5XvglV5ycuftA5\nt2nHCBau3ZbmynTHpo6Gdk4XVncsdRNG65qyKJdEApZgEtd8VtLascXlwurAPacZf6sxMaF9Vkgy\nPmxAks3hn5AE5J41lp2aTLDprDZazMY+Ur6tJ+zfZUgmuqLz+Um1krbjMOzYJ2Y2Pk4dBBK1nzJM\n66+DUK5ZRUT4+MX34xMXPyhm5M4FVK5RhGxT4gqLOIulkjQrIBHKjh/FKmM/O8CcjD9+0QNp+5aw\ntIKC9Wr0ovb+VW6cl9h1kbqun0vqdBcDRbA18pFmnJlQk+DxpB6dbAG48Vz2O3/WPjNS05nZXmam\nbEOzaiet1onfuw23Lt4IomRhEqxZWSZcG3q+wFAyCMM1BetbhPjMzc0WZz4rCfvMnGL8nZsBe+ez\nKjUDMnMM4H/T/yp0GWXCKuZ0JZ8ySe3VpX5Fu77PQp+V9lt9/K4Z0NUemq040SwEIojBBgwUrNP7\na9g+5NrlpUS2KjuAgpgdJECz0iFlEOkr8FkBSW423+aLed/c66RJr8hnZWtHMTOidNK1zTZ2+8ws\nTmQMv2al6spMi9o1RbDJgLpmpZuja6nPStXX0PL/Je1awmrfmVi1eacbOK3MgNZ7bic3oLrXGf01\n7LCIGooBW0uJDEWalU4r1/svXREzQBaJxT4vZW6332OtZIuQOE40viLNysYuoVkR0ZNEtNT+bzw6\nNxlgECSsCezWxRvADMyckq8pJLqr9DsEqvjCNVuzrcCzc1pfmhrBQkpwm/WNE8dtnxDwlcR+6P0O\n6+P0/loaFGxPSK4wLIpJ0x3XqljI43J9Vppm5TEDSmZLhw0oaVbCat6+pxGLCu3zWdoEEbvm2KNZ\n+1iCev3NNs2ANnV9pBVriWjzeqLIfE6KSZf5yqznU4tk32vTq1kJwspzE2p89dUjvOOl9jZ8aX9T\nNmDRBG5kirBMuDYSsoT67Z6XEtnGKVlCh75FiMw6jdFoxoWalWvBkY+PJ0JE69EAXp7+dyyA7wL4\n+Vh2ajKhSNhcdPcK7DNzCv7mFc/Mjtl+ItNnFd6u7u9YvWUIx337ZvO8vr2G5rPSm7A/klacUGL7\nhJQFeibopP2wzk5L2YBuULDts3IFpBQWoGsj24eb2f5EoWi0cmFcZAYs81kVbShYdN1Is8AMqL0d\n9TR8mpXPZxWzP9DYTbcU9g7t4aCbAfV+1NI4q8xHZQlFKWuJFC/m91m5ffPdQpZfssWY0idPkyoc\no8g0pmvfNpPThv4sfO/G1dBdDbmmswGz968v3NJFVxR58wHa7fu02/FEqbBi5gHtv1XMfC6A4iyP\nFYIhkQAUGq0Y+8zsxxzNfuwOovx3Oz6rwniO2Fxd5yw1s5wtMBSNtl8SVlacVTtmwCDNCq728PCq\nLXlfNce13vb/u+xhbB/2JyK1oQtjr2YVmRTspRu2O32zsw4AsrZlv1NdQNoCxyhqraylCd1HOJD6\noa75xV3LjRgnqW4bdjB5M86ZlK04F7GZ8MnKcVZG/xdGef9CwJYhRQw7e62jkvX6KN7qmloJwUJ/\nz8ZGjEJZ3WclCUCdKZldw67PSg8Tyd9Rfr4Zx2i0EveClO9TMuPbTNBeoNRnRUR/pv0ZIdG0xjRb\n+2SCKWzsc5wkpSwIpjUzVrcnrHzlm7H5SfgyWDhmwJhTzcO3EtXbD+vntP66zAZsSROXeeyGhfk2\nHFm8EWT/TSgSM2cJwUL72H//wCp8/KIH8OKDZxtlpDQ+ErvSfk6OZuURVlkGC838qUP5hiJyx6BP\nUfjtfSsN/01ogKhtBnzb9/9k9oOTrBmUCh97/sxX9Xl7M6fUM9OvG4uYam0BZkDfONQFpU+zUv0u\nNAMaPitds3LLxgxQpllJ7QlzRGz6c6P0OWZmQOH9x5ywAftrUWqRyM++/5WH4rG127y0f2mMjhdC\nhM63td9NAMsAvGdMejMJUZTBohWzodIDgmlD+92Ohu7zWah29b4o34E96djMv5iTa/vqss/KFFbh\nmpXkxJZ8VkWe/jyDheuTKdp0zkazlVPXvQQLzQy4cG2S5HTB6q1GGVmzEjQdhzygCyeXYKHgmgHd\niVsthgyBB79mZRMNQgNEawVJK1upFp/5NJkdNUfdsv58jjhwjyyI2Oezsr8VUQDA1KwSdmLeViNW\nk7q0mFT+Sb/Q9vusBMEZM9Sbk2n6+e8sQwmbwkppSupTlcyKzZizcWwvDJSgs9u3s5f0AiFswNeP\nR0cmK4qCelucRqPrk7wtIARh9/0bFqPRYnziTc/1tluUsbsRx6IZ0GZ1OWbAmNGIPT4rIoMeHKrM\nJGZAV7Oy25Y0K6O8ElYQFgUtxubBEew5vV+40kRTuz8pN2DSl/y9KOag3X9JWPmyJUhlcq1CL6sJ\nKzvOyqmXs8lWPxvHfs3KRr5VSnG5ok39lNAk5GY98mi+SlCf9OID8OW3H4mz5z1imA0V7PRQdj3m\nMfNvtdBoMWfm8HoUoR6Ro1WoxaQvEB5olw3IgCJYiDFV7r1+57rF2KTtKqysH0Vs0Fb6ndZrkfNu\n1PYiw01ZK+1luiWvsCKiM4ouZOb/6n53Jh/YM9kAycdWsyjfkmY1c0odz9t/VlbXt65ZBACFwso2\nIelotUzbuM4G1OF8vJys2Kb3uzb+xDyhtx8mraYpn1WAZlVUo+5fsdv+1KUP4aoFa7HkyycYm1na\nUBOWegw+zUo3A/pMopIZ0Lchow61cOhLg2lNrShHrlmpesx6s/dvyRGGn2Bho6H5dYog5LHN22Nl\njqTUJJnT7/W+AvnzOeFFB2D29L7MPOyLRbP9dZKPzb42IqCF5BtQ91WvEepRhEbLTYzbMXVduER/\n7qJmJVy32NpLSyUQyKnreV+zduLcDGib9iNCmu/QbEfV4/PTjgeKCBazSv6r0AVIyVYVlJnGICZY\nYyVO7ebqQ9chbWqoX1dI22Xrb+RxMFnfBS2vFcs+K9vUFmpO8OUGlCd2fz1mIlvz3FUL1gIon3T7\n6xGacZy9gyIz4MK12zD3zMuxYtOgWEa/VsoSn/fb/FtNYmrTQsNpr3VHPW471ihrk5X5zT3eClSt\nVN/K3qWUdV2hpRYPBABuMLXqE+Cmo1JEFtdnJZuspGweeZGkTj27eLa/WERGFhkF5nRiLxJWvqBg\nYWkVxzlZQnwOAd+MchkU7RTcUmbAiBxiSRQRauS2peoZ8vhpxwNezYqZvzCeHZmsMFc85rlWzM7W\nGlL8A0E5p81zH7xwPm7+99fh0DkznHaVr0BCM44Nk11GsHA0K7PDKo2LFGelHOh6v0Mwvb8url5t\nrS4iKtyqIs8v5y9T1qe+WoRGK59Qhj1mwFpEGRPx+sfWi2XsFXd/REFxVtlqX+39pJ02qet+6rI6\nzuliSAcznFV1GcpMQ2VmQLBisKX9FBZtQP4O69mEnAtdHb4MG5L/xnnnmgDOFgYRiX43tZgsWuTo\nY8TOmO/W55pAXRQXqFmaldcM2OLEDGh90yp4X1qIAv4UY+OBEDbgVACnAXghgKnqODN/YAz7NWlQ\naAZkTmMhCsyAzJlfS/oAfINL0jAUEudr/rcSSvbq0vbZtAo0q8RnlaMtMyBC4qyKhU1uSvKXLzN/\n9dcjNFtxVldRbkCFuodcIFGapUnfRx6o1yKHCaeXzTQrYbICgK9cuRCH7j1dEFZ+goUPZUSLIjNg\nHCufFTlZ17Mylr9ECQ7lp3T9eqlwc74VVZ9bt3oMSuDrxAWfZtWKk2S2RXFWumZlCADhPluaz0qC\n6ZOUy6iFgb2fmU2wSMyAJJgBSfT/qq7vHOmdZhUSFPwzAPsDeAuAm5HsZ+Xfx7lCWwhhAxZtWhin\nmlXyobtfQFHQn9dnZWldauVoD2zbDMacCDYpMr7MZ+UzSYbGWfkmuqw9S7PyTT5F6K9FaMZaLIyn\nvH6fPhOYTntX9yZN+vY96QsH2wwoTWb61ig6HlyxGfMeXO1ud49wgkXW/5LnJsXyKMSpIpX4Sjz0\n8thsR707RXV3Y8hSn5UWrqBfL/n5bEG2dMN2rNs6nLUnvUdmiP5UHcM+6rqU56/gm1TtqdO+YpH2\nbADg3OsW48bH11tp2ZLg/XpGXdeutzSr+57aZPSrl2bAEGH1HGb+LIAdzHwBkoDgV4xttyYPikwS\nig2oKyouwYK11ZBbv4+WLTGLFJpxbGaq8Gy+aDtb29Os7P7IfZnmybpup95JWM/+D71lCRhpV+Uy\nn0BfLTHVlZkUdbOXV7PSV9zWVhg6pKzvSV8iR6vQi7pmQLEbTsrauAPNype82O6LBHUPRJSZsu3a\n7MVBTTMDShN8Rl0XgtYB81nkz5eNv69esA5vOfcWABDNZao+O5Gt/b4Mc29JuiU7ZERqz3dWDTnJ\n5Hr+rU+6mlWsgoLNsiqFVIsZty3eiHf84Hb85E/LcmE1QQkWCooXuZmIXgRgNqrNF7uGogwWig2o\nCxwpWI/SGCZpoPvMBdKKVKEZmylcMvOLLaysVVaLk48gzGfFuO+pTVp/fJpVHVFkajFPbNiOb1+7\nyKpfFtb6PQH58xY1K8skZKOvlhAsfBkVFAwzoEer0OdR9XxFgoXdR23CZusdmluEpOWFCVqH7Ydk\nljWrolA0Kft++LUJ85RILTjcvqq/c80qmbb0DBbT+nK7tU2wIKseaYGo73dmoxaRuOhoCT6rohAF\nU7NyYWvK0nnfNytljVGwM6oo37LEBlRxVnHMGBxJMrvctniDxgac2JrVeUS0F4DPApgH4FEAXx/T\nXk0imPbzZDJ9eGXinM/MgAXEBBVF79WsPO0WxVnZ5ihVb5kZUJkXQjWrd/zgdu1vmZ04PfVZ6ac+\nc9nDTrmEul7woVsTmD1Jqz4UIfFZ5c/N5+OStDYb+qSUa32Sz8psIzMD1ggqvZWCXlJ1Ic9gEdbX\nmOXdh4smQ9948bWhQwknZR2QemlrxbpmpczZ07RwiVzzTdvI+uVqxHkcmp8dW2+DDWjfu65BmwQL\nt63Ef2cee+GBe+BzJx2hWvSOcJVBXSpRI1PINZpxFj8mUdfVfmwq7nD15qGsvzt7KKxCMlj8HzO3\nkPirnjXG/Zl00Aft5sERfOnyx/CTPz2JS/7plRnbqIwNaDCpLPgGd5HJoRnLgsz+YG1h1Uw/Nmli\no1KflbyqndZXc+m1wuSnVuU+2HtciWadEgtHXy3C1riRCTrf5Ka7N3x90gVCRrAISLeUESwiN3uD\n/kzLtgjJywl9Ex7ElHrkperbGqH9bgtcVhl1XQ+/cPfeUgsNk+ijgsWZgaman7Rsp2CJ5yBpdAo1\nj89Kujf73r25AYW29KBghbcfdRD222NK1m91nf0J9NdrAJpivTVLsxpq5gse+7Yyl0Kcj+9Vm3dm\n1/sC4ccDIZrVk0R0HhG9gdrJS1MhCPqkpQQVAKzZsjNPt1TABtSZVJLw8e+E6vcTteJYXKHZ2og9\ncNWH6dOsdD3P9VnJwnPGlLojnCRB41uV21CahmTbz8yAnmuVZlWWeqZW8L7sfgC5APLthKujGeua\ntKkNSN1R8o8B/Pmz9sZX33Gkcd7+pBNtza1nSp+czFX1ya6jqA27rK5ZiRpHesj2nWamUAB7zeh3\nykvfCgDrmSnNyq9Z1yPZZwW446jIDHj2vAViHdm1LD27fOFRtBibmuYvlMak/VzVd+sLCq5FajGb\nHNMTPU90n9XzAVwH4CMAlhHR94noNWPbrcmDotVci904K2lTP5W8UtIMfMwuFlawCo2Wa45QfdJh\nr7SVyUPKDWhrVtImctEPI3EAACAASURBVNKzUGzAon6o+kuseADKNCvGAys2e4V4vxVn5Xu2UYEm\nbPcDcAkEOux7asZxNrE3Y8a//up+rWxeOEu3FOcTNIFw5EFmQl2HYBHLmlWxGdBlheoo0qxy6nrS\nGWkM2zsU1zPqer7IOXy/mTj9jYc7dUv1GIeVMGe/FaJek31WgCzsdUiZSnxItrUX2tDq9i0+56TC\nepuwg0C9Zi7klN+pLlDXKbXktIS2iBI2YDuJn7uJkC1CBpn5EmZ+B4CjAOyBxCRYoQvwMdCUkzPZ\n9TM/LiXnTPbV8djBCzQrr4kqlj+K0FVkiM9Kyi4grSqn1N3Vn6QVJWac8o+okGARM97+339yjiuo\nDBbtsAF9XTIJLOlk7OxVJjPdapSYb1Zt2omBHSNiW6oHOsFCX6UrOHFWkIVw0a6y0o7ROoo1K6S7\nE5NXg7AzfOSxRJSNGyLCiUceYFxnk0tyM6CgWXk0e8DvswLcbXNsYTvSjL3X2pC+O8WSTProv1Zt\nQy/lnHQ1K2UGdPezihQrM3Z9lzP662BuTwB3E0H7GhPRXxDRDwDciyQwuMq63iX4zERqI7rImmDs\n4jrBQqrKt7KPWd7WHHC3CNH7VISL7lkBQF6F21nXbf8Mx+69Te+rZYJYhzT5+fVEE+p5d0Kw6KsR\nWgFmwBDNSj+cZ/gu1zYbrdjZyFLBEFaKDZgRLFJh5fgo7H7Jk3Y7u8qGBnyrsvkYzs2CZhmzHTUO\n8zgrl20KuAsYaZGhkzA681lZCzirkuFmjKkFJlQdNmEGUNlp8vO+Jztnpj8Jc5Swj7Q+KTOgFBSc\nLpTZta4oEsvQyAQVVkS0DMDpAG4FcCQzv4eZLx3rjk0W+D6Qeo3EdEvO9VCxEfIk4V3ZFwz8ZisW\nrysTVn94cDUAWbNKqMn59Q0hBsbu/7T+enqt7bNy2y6i9epQi0JptVsurCI04ljL3+YRVrpQ9pkB\nBbqzFBckpVuqpatfO3RAIljEMeN71y/Gvcs3Zb5NHY7gZ9l3VqhZtdx+h0JlNleLsiLrQJ5qSlHX\nc1arTURSkIKm9SZU34vYsUU+K4lNqWOkGWf+pDIkJljXumC04enjPtoGrTbUwldBme9lNiBlQcH2\nvcxQwqpHgcEhbMAXM/PW8mIVOkErdZi7dv7kw62RO8HoUEwqKTcgUODgLzB7JJOCey7QmuExA5o+\nqxAzoMrebjfrC54NmSKVViRpVmXWjTxrRN5nCfrENuL5sEU2oPhM7D6qIHCXRqwX1c2A5163ODkm\nmQFtMxbLBIuibPRuv71FHSREn8SfRpSbBe0+AbkwN9Mtqes9zMb0+wDkOCvV92TE+9+nz2dljyPb\nWjHcjL07DduI2c04r9OSYmHMKJQJK/3eRgw2oL140ePX5MVjrxiBIT6rSlCNIWJmbwyHxAaUymUx\nKu2YAQWzm36NfZkicYRAJliYZkA7ES3DneRe+aw5adtmfTuG3Y+F2e//09G06M86ytIGJcGVrE2e\nPs1KE1YeCSj5rBzTKAPioiFKdFSbmeWL3TH7Zp63s0swZIJFIf3cfg5tCKsW55oVecawql6NGbVH\nWER5ajB7d4Ls2tgVUpIJltkvZPtqfp+Va0Y1/x5ptoI1K2mLeqKcXs6aILff9expfd567XszNSuz\nbJT6yFsxO/63TLPqESMw7ClWGDPELOePUznoisyAauLU7f1SGQlP7xjBtiGXOQQkKy/7sjIToA6v\nz8rI5O5OtHZfv/6uFyfXWtXtGHH7XWTW1PHEhh1Zf6Q6imDn4/PFZenPyhebZK6S5bggadGg6o+I\nnBWuRMnWJz9pQpd8VnIGC//7t7Pvt2MGZM7TLeVmPVfDBHKTndLylFaQsGblMWqyLtN/hYVCEdMu\nyc8pT5UO6ciqo9HiNAaqHEpL1KGbz3Xrgz0ufJpf0ofYuDe1YLP94dmxlA3omuWT+7j7yQHMPfNy\nPLBic9B9dQtjKqyI6HgiepyIlhDRmcL5KUR0cXr+LiKaq537dHr8cSJ6S1mdRHRYWseStM7+Ttog\noucR0QPaf1uJ6PSxeD5A8uHIq/w4y+rss8AkDuQ8ka0YZ+Nh9v3HpQ/h0vtWivWu3TrkmETaEVY+\nM6C+iJe0CN/K1p4otwtC1qOEeDEoZI8um2QT2z+LgsAoF8IGNFb3yb/SpC/1SU0yUiJhu369yiA2\nII9es2pHWD29YwRrtuw0YsdsqEP6Zoh635P2Zc2qzGeVZ/jwjz+1U7AEh7puVdJoxYX+Pvta+3qd\nYKGfsZ950ffZaMk7fydam20WztMt2c9jRmoGvOLhZP+3GxbK29+MFUIIFh8noj0owflEdB8RvTng\nuhqA/wZwAoAjAJxKREdYxU4DsImZnwPgHKRpnNJypyDZluR4AD8golpJnV8HcE5a16a07rbbYObH\nmfkoZj4KwMsADAK4rOx+O0XMLK6K1OZvRWZAZWdXTDtfQGUbcwcA4L7lm7B1pykQivYkcvolNEiW\nwUn2z8gdtVveKggraVVahMP3m+kcC8kerpsbfeXbjZ3fkcbGSBkXpFuymZV6+fw3G/8C5sSX99X8\n2xfrU2SKHq3P6vYnBrLcgNK16nnnZsCcYAEoP56rgatr1cKLhWei+n75Q2tw+5KNYh+LfFYOdV2r\nmznZbbiISWlf6yyADEq/nwRSI8LeM/rx7H3dveuufGQt/u3XD2Z/58mcBbMw5XFWPh9yrwgWIU/x\nA6nf6s0A9gLwdwC+FnDdMQCWMPNSZh4BcBGAk60yJwO4IP39GwAqS8bJAC5i5mFmfhLAkrQ+sc70\nmuPSOpDW+fYO29DxBgBPMPPygPvtCDHLzmtlAvKxnJJrE7tyEXW9SAj4cM2j6/CeH91hHJMICTou\n+5dXYe80MHG7EJhoU4tdBlkRu87SrIaT3Mp6gGsowQIAjj18H7z++W4u5lLNKu2Hopj7CRaBHUnx\nDz+9B4vWbXMnfYGRldQvL2D0J5BpI5rkIWEsSXFWku+vUFgJ8WEh0Pui4nukCfnb1y7Cd69frJkB\nk+vUeMpJJ4FmQA9R4cO/uE/sZ73QZ2WZAbX6lHAN1awkP5G+wChaBEQRYf5n3ojrzvgL8fzdTz6d\n/VbPUXpekTaX2IsnZQZUe1q1tyQbPUKeourTXwL4GTMvQFg/DwKwQvt7ZXpMLMPMTQBbAMwpuNZ3\nfA6AzWkddlvttqHjFAC/8t0gEX2IiOYT0fwNGzb4ihXCR7BQW1HUPDE16tp8ixCfZiX7PdpFmRnw\n2fvNxLtedjAAOYreDgqWJmbfHGc3PdSIcdprDsMf/jVPpKJMoiGQsmIA5WxA5bdQH3sIGzAUC9du\nC6aAq1AFG/pEpwSXrVm51HXgxk++Dld+/Ni0Dnao68qP4YMTFOwtaUKfxFXf2LNo+enty7JFQr5T\nMKXtxyCStX9pVwO9u2W7HAPFPiuHDSj4hoqyf+iQ7l033SYEC/np1qPEpBei1avnmIS7mOf0xbH9\nbGZMcdmAmwdHMF4IeYr3EtE1SITV1UQ0Cyjaz3L3QOrzehuAX/vKMPN5zHw0Mx+97777dtSOzwzY\nijlLt+SbKJSfR0X/i9R1z+q8DHaTZWbAiAgvPWRPAMChe7umiCgyTU5S1gOvGVBoe5oVaKkEdwi8\nq/BSM6BqS5X3198u+iJ3e3SftukLCjaCXa0NCwE18bl9PWyfGThgdrIJOMO9r0ST8/fdntRCx5vu\n29QXXD5fYLOVZINQ40H1KY6VAJc1q4wNmDH/ZM3K288Cn1URG1BpVqFmQMn0ppvPpYDprB9tLJAa\nzVyzsrXgmiasbMauMgOqkImbF23AUV+8Frc/IZtPu42Qp3gagDMBvJyZBwH0AfiHgOtWAThE+/vg\n9JhYhojqSPbKGii41nd8AMCeaR12W+22oXACgPuYeV3AvXaMVsxeNiBzSrAo0qyYU1OBTLAoEgIS\nvvT2F6G/FmXbAyiUaQs1Ipxw5AG46vRjceKLD3DOO5qVoEX4uindvr4lBNCeZhV5Jt8y85W9uvZR\n5X3Cqsgc1FeLhK02XLMQgCwouAgyCcS9Tv2dM85cggV5hLuCvQlnqCZvCABCHmflub4Zmws71adG\nqllJ77TZchl0pmZV3tlaQW5AKQXZso07cMcTAxlhpi2fVep/U9DN50XDsx2fsr4BaZEmZ4/H6RZ1\nXbEBH1m1Jbjt0SDkKb4SwOPMvJmI/hbAWUhMaWW4B8DhKUuvH4lJbZ5VZh6A96e/3wXgBk5mjHkA\nTkmZfIcBOBzA3b4602tuTOtAWufvO2xD4VQUmAC7BWZPMGPKBvSZfID8w05W2vJk66M/+zBrah1H\nHjzbYaaVCSv1rTx//z3E+7HjtOxJ4rYlG3HxPSvsy9Jr3Qpn9AuaVaiwIpvukcC3olewrTlFqbIk\nKDOKhL56JLIBfdnvpSbUxHP/U5uyiUQXqJJDXf1NWSxPQrB4ySF74it/lWRor1GxZrWzYSY3DYl3\nA8zxkO9n5bcENFpxRq5Irld9ToStbzdf21dVFFwroSg3oC38WzHjdd+6Caf+753ZZC+xY23UIspM\n4boPW/dZFfllpf69+YhniGV1YWVr0SrOKilnnlRBwXYwemjQ82gRIqx+CGCQiF4C4N8APAHgwrKL\nUv/QRwFcDeAxAJcw8wIi+iIRvS0tdj6AOUS0BMAZSDQ4pH6xS5Bs9HgVgI8wc8tXZ1rXpwCckdY1\nJ6277TYAgIhmAHgTgN8GPJ9RIWYWzSwq87mikvquVdH/Pur6nUsH8BffuDG4P1GqydmrqjLTVmRN\nPC5szcqs/6zfPYLvXL/YU7d7zJ74GX57vo0aydpauRlQ1qzsunzPamrBCrsekZgbUIyzIknU5ivv\nv/rB7Rmt3TUDygQL3S+iwinUcw8JCNdp9KGLBn0O160DkrAiJGNG13BMgob83HUzuESwkPYQc/tZ\nEGdlDU696+1oVrX03lvMWdAzYPmssv+5kMyA573v6CywXkdOsHAXuLqJ2TEDpqZ3O1luKIFktAhJ\nt9RkZiaikwF8n5nPJ6LTSq8CwMxXALjCOvY57fcQgHd7rv0ygC+H1JkeXwqXzddpGzuQCLwxh85k\n0j8iNdB1G7LTz1jlBvQHBf/4tifb6g8RstxgOso0K9t0IZ7Xjtu5AQv7JEzNM21h1Y5m5XFEh7IB\nFfT9r/QrfQtpIsIz956Oeo2wNA1OzuqKOfMlZP3xxMh5szVImrVBsJDT66h7UHU04zg1NebtlZmZ\ndo60soSt7fgO9d9q8vQNjWYcG1qHcb2HIalr3OpZ6o+pTJsGkoVEny/dknW4JXzDIQSLKMpTryX3\nqBh3+Xso8sv6vk9bWwfMOCt7zBgEC+vaeo3QX48cYRWaqHe0CBGJ24jo00go65cTUYTEb1WhC1D0\ndPtDUwOiFvmdp8rEQZSn9bdh719UhogS+7wtTMq2ObBpyFK9utBph/MhLWqnO8IqnIWmJkYbpbkB\nrQlLfei24CsS7Lf8x+tx2msOE9p2WXhqonXo5pawPeNNz03Lu+3FjmZlns/JCvnqPY6TNjN/Fsnv\nAMiJLrppKNTsrI8T0nxWfjOgqXVYLi/xnUpbybdtBqyFJ7LV61PmthDNo0Y5bd8QjNo7K/LLtiOs\nGloGC/v29fAG+9qISLQOhPrkRouQVt4LYBhJvNVaJESEb45pryYRmOVUMWqglJkBy7YIaReRp64y\ntlGZmcgXyBoCqe6ZU1yfVSiRpObpS0hQsNFmm2xAdVh6n82YMeIQLFIzsSMM8wlsen8NR6UsTElc\n61XKZkCzb5lmFeVLiyJK9MypyaJBzwgS+h50AagzWr0JllumZkWGsJMtEPqE2ynBoshnZX8XLGlW\n1mQuWh7SQFybcEXa/4ueq2+OsE15QK5ZRZGkWeVj5Fd3r7DOkahFdcI27gQhiWzXAvgFgNlEdBKA\nIWYu9VlVCEOLWQz8HdbMgP44Kz2RrTxopJVVEYjkD7MdtpFUlKjzIELpOttn1U6mDv9+UG2aAZXP\nyiq313R5byHdrGajFceuGZCV39Lth6pjSj0qDBo1NCshJZH6Wx3+xlWPY93WYUNAFY3BWVPd2JvQ\noGBDswKyRLbrtg47ZYmS+CCJDZj8VhaG5O96ph24mpWerktKLWWjMM7KeixbhxrZb2Udsc2AfUJd\nahsPKZQl06zgtx60pVkZPivznD4Xrdq80+mHzcIFwvx+3UBIuqX3IGHJvRvJpot3EdG7iq+qEApf\nlorcDOjXSLJEtlCrJLdMu8LKlzGjnTgOaWJTZp5OILMBk0lyWuYnGV19QAgb0BJWmRnQLLffHvJ2\nDUpXkZ5lUzIDxomHwhaSSU635He/tpOy1H1DWxTeAWX/5idWbd5pEiwK4qxmTXEZYsEEC8PnJJvx\ndDQtNqDkJ1XPQk34ut8ljhnLB3bgb8+/S6szgGBBMnX9dc/b13k3H/jp/Oy3onjbmpVUV56Pjw32\noDLxA3KSXwXfmJay/ittksj1TUfkX5hGEWGqwPwL0U67gRAz4GeQxFi9n5nfh4TE8Nmx7dbkQZzG\nVdjazIgWuKfO2XvWMHLNijyalRpIN37ydUH9iUhepbWTQkga6j66eGifbCjN6r7PvgnHHLZ38Goe\n8Kcrat8MqDQr8/h+szzCSpkBhWfZitkx2STaIjvCTfcnTann2Tje86M7MGhlpDcJFkVsQDjHcwHg\nn8CUGXDnSPs+K71KH2lEh80GlBioqp9KqJmaFURiSxEiSiZp+/tc8IW34Cfvf3mh+XtnI3kXjrAS\nBvRRh+yZxVnp5wmmz8oHXxyYra0DuQBXOwLrkJLbKiRmQCk13PjkiAiZgiJm1tPrDgReVyEAMatY\nKr9mVa9FePDsN+Pr7zzSuTbfIkT2WanBascl+eDTrHxmEF8d0rFu+qxmpD6raf01zJxSb4+w4emL\nzXKyoT+XekQY2DGCr17xmCOd9501VbxeFZOeT7PF3qzrdnndLNdfjwxh+cR6czI246zcd6uqtp9x\nPcoXFypvnwTFyuzEZ/XsffNkwoRyM3FiBtQ0K4PibZoz1eStE4XUt6KjTFip9uznNqUele7irZ6J\nq1mZfy/72ok4ZO/pWUyk6ZczA7Z9vfVrVpJrgLNrZDOgrw1giuCzmkia1VVEdDUR/T0R/T2AyyFQ\nxyt0hjhdOavV1LtedjBm9Ncy9V19kLOn9TnBhcpPkwgreZIYaSVxXKFKDQlaHpDEJoVCdCCTPjGG\n1+Ur328FToZMkGpi8a3iyz46Q1ilD+RHtyx1JjyfZpWt/oXnm2hWsXOM2X0f6n0D6aSpnbYzYhua\nlXad3Sf7uG5+9rEnAWDW1IQYPNRoT1j9+p9fiaPn7qX1zb+iV0jMgLo2Bee3erZqwm9ampX93sve\nuXr29jvwPTcdStu0fVbSNUR5jJvOBtTN50WB7z6hOXfOdOeYMjcTucJaJ1jYiIicNGdAGKOyGwgh\nWPw7gPMAvDj97zxm/tRYd2yyIMlrlg82FVelx1kpSGaoPJGtnBuw0UpjZgKllS/Hmu9jeOML9sPb\nXnKgcUxahSfH0tVvUE/MPsn15b9DFvO1TFjIfbDjSnzXA2ZWAvtj3dcjrFSjPjagb48v0QyY/u6v\nR8azsFMfGUHBcN+NUpjt4301LQdfQTLlmR36rJ673ywnVKBsEdNoxV4zoHoi6tkq4aK/U+5As7KF\nX9a2WvgUSCv1TFw2oLQYzLflMM2A+hYh/n76unHe+47GEQfsYRxrappVUZyV00YkmwElxuFYIMi2\nw8yXMvMZ6X9jtrfTZITKYFHTBn+9Fmk+q7ysPWGoySxZffkJFnYS2SL4NCvfZPXWlxyI75760qB6\nfSanMpQ53n1apQ1ds5L6UPbR6ZcUBXqWBUlKbbfi2HGGq9AESetR76OvZmlWViocXVsTFxGepcO0\n/lpOsCA35ZcaI4oN+OnfPozlA4kJMkRYJUGwmiCBf4wpNFo2+UCrz9LalVAbaZlmQFezchcoH339\nc7LfPs0qb7fcDGjHWUlVKRp5i9k1A6bl2/HLKuw9ox9vfqGZdilLtyRkXSfy+yf3nNYnju2e+6yI\naBslu+Ta/20joq3j0rtJAPUB5RNpMhmp9DX6ys3RrLLJLDfT2AO62WJveh4JSV9cxpVvtRW6HUak\n9aEbmpUOZUIpQ11bEEjdLtsuQu9HWb63+z/7Jvz9q+aaB9M+Ss9M0qyU2cdOdKwLW10DAkzqNGAS\nH6Sn6Hu0U/tqhs/Kfgfq/vVMIrelmxeGLBzqUWSa4FJWqx+UZLAwzIDab2WZSP9VBAvbDGi3IWlW\nRxy4R/bu1LfgHf9FBAuPz8rn043jpI92tgwj3ZIHRQqiPVb1/aycdEsF5th9Z03pKRvQm26JmWeN\nSw8mOVqcBjRq/ox6RFi3dSj52zADmtfqGSzUgLY/vpFWjCl99WBthshsR5kKfB+rlDFeQrs+q7cf\npZkWtfIfft2zRf9ByAQZaQuCTjQr/Vb76sU3sdeMfmcVqmr3swE9BAurfOJXSH7Xo8h4npsGTWGl\nEx8kaeBbCEzry1mGkh+jr0bY2TDjbm54bD1eeODsIA1ApRdSkPxJNhI2oJxuyT6WUdctgoX9DKSJ\nlvR6lPDzLE6KuqzMgPa1cr/T/lmLEz12zB7jRx+6FzZuH8aygcHCZ25/L9l+VoJFoii11pyZ/WKc\nVWuczIAhuQErjCGSbUBc5/+iddsBFKcxUtTm5ONKjkkfn+7jKENClbcnhCJhFSoEc78ZpXX68L1T\nX4oTXrS/1YcEp73mMIfCLzmKi/rqyyJeFJO2z8z+tjQrwBVKakKRJqtGi8X9rKQ4K9MMaGo9m3aY\nm+ENGpqVPElKmNZnmQGtckpb0MfK9QvX4/qF63Hph18lV6rfA5n7dzFcf5KNRis2tA7JRK4WJHWB\nui6lK5Im2iQbRvJbCb1OLAsqjMA2GUv3qbaSB0zNSv+2JebeKw6bg2UDg1kIgQTb36bGuTfOynNP\nU+o1TBF8VhOJDVhhDKES2aqPIiIyVoNFZkC1ZYCuKUiTti+9kATb5KPmIt9qqxZIEzQ0q5Kye03v\nt1bQZv9sELm5DBVOPeYQoxyANDuDW9YXILrfrCmYf9abDG1MykJgw35muWaVHz/9jYcDcH1NALIg\nUdsso/uQ6lFkPM8BR1g1tevcPvq0mUQrzMekS8BQwkq6Psx/qI9VFS9YhKaViigyFnJpvZoQB8wF\nSCtm5/sQNSvShZ75r42iPu8ckYOCfWNYCVNTswK86ZYI+OLbX4hrPvFa7OcJlwBcs6KqpkYuMamM\nlSmZAScMG7DC2EKZ8XSmmr7dghHlL2hWSexILmCkjy90u+ukDTPAsKZN8BJCJm3Vd1VDWVfspnSN\nQKT9wi9oXnt4voNzbtbyECw8Ai9LSaQdKzMDJu3Jk4SkoUnCSu1yK2lWqu56O5qV9Ow8tzG1XzMD\nRq7PSk3AtYhw1enHGudC5i4icxsaFYLhL18cFKybLIFcm7h1cb6LrQq61SFNtPqCLWPpFjDkfPAF\nBUuX6O+4bmhW/rCBiBJt57nPKPbY+Ez1MhuwmNAkESza2UFhNKjMgD0GM6OmJbKNIjImLl2Dtz+Y\nPzy4Otu8UZ2SMq+3o1nZCUHtFaZTd7AZUAvcLHOlW53Vm5BNWeT1NxkU9/Rfn1nUJ/B0c5hC6IZ6\nOmLBDFhPtTxJWDGr8eGuzFXVfTXTZ/X0oCms9IWPdNe+RUxiBswXUPbtZppVjTC9z8rTGLjSbln+\npLIVur35YrEZMPn3hoXr8fz9Z2FwpAVmdzEnpdgi6BT4pD3f+y4a/n42oDSG898249FnNQk17vuy\nWxC5C4soIjRG/MJnmpTBovJZTQ5kZsD0o4iIjAlGSimj8P0blwAADtxzWqFm1Z7PStbmfCtI34fg\n1uuugov6oMMQOMKcERGwcbub/NRuS9es5CwS8keab6ORHwsyAzpmW/d48u7J2X0V0BLZ2sJb03SU\nsFPwPYfkPoQ+FggrdUZiiKkJtRaRo2WGWoXssVrm+2jGbLRljikyjukT/iF7T8fygR3hmpWWp7BU\nsyoYzFkGi5qpjUiX6M93ukZiIOR9cYRV4Eft3YtL85Nlx8jN5PKjv3sZDtkrCS6WNKueswErjA+U\nT0L3DenjR/9IvJv6IR+4vo8vnA1IVpvFmlU4wcLsbxEkH03RterepvfXTPabVT6nYssfupSaJqnf\nbAcIMwP6JjJjh1xKnvFQwxWUapdbe6KsUX5f9VpktGPnvtPhc+xLmNZvMtJskkB/LR8XtpkpNB7I\n9llJeex0NFqx4M9Jf6f/5kG8uhacB82H+Ff0TSr7asXjXxf2B86eitVbhrK/lbZsj5UiFiNghgMQ\n5ePWJUMEalYFGePtd0XWYhkA3vLCnOwkx1lVPqtJAeVYzjQre6IuYAPqx9W5O5cOOOfbi7MyP0x7\nhWkjnLqep5Qq+8hczcqsx4Y6JG00aWqmyb/+RLY+zcrtV7fYgOrdS5qVcrhLwtuMs5LbdwOX3YK+\n9zq1XjOEuyOsMp9V5Drw5e44sNmAigzxiTc+F8/ad4bT82aLLTag+24zn5U2LhN6fxI0XxZLB6QE\nC8tnZTPq9LIKv/voq/H8/XP/0eBIC/XIpIIfMHuqOPb06s3tb/L3a2tBo9asyE0kYFt2bEgZLCo2\n4CRBK81QIE2IQHG6JYXk40p+n37xA855H/tNgm3yycwqHqHUjhnwX99wOE495pn46z9/ZmHZIp+V\nKKzgrqbzutx6ySO8/T4rV8iGCGkn40j6b81aDNQsP6WCCvq2b0s3A/ZZmpUOm2bcjmY1tb+WybYa\nkUMS0NmATtBpsM/KjLNSmm1f3V1MMIq3tXfNgNq7qlGmRQTF42nWBfWeZ06RjVD685s1pQ8H7Tkt\n+3vnSCsZk9qt3PHpN8hmwCLNSgkrxww4Os2KSM4N6E0XBp9mNXGyrlcYQ9g7wdo+hCKflX7cXn1/\n7qQjjHyDoboVLickJQAAIABJREFUWZpVVodHKAWbAZEk4/3qO47M9qIqKmv2SZ+U3PK5xuQOZ9lP\nI3/o0t4/en/06vs7MANKbECVxFgJK/15JmxPV6DocWL1mv/N2hOL9KqKgoJ1Ie3LHJ7sCmDWMSwI\nXglnnvD87Ddzrln1WymkgMSfWLStvZ3BQu9vXxRh9rQ+rN06FLR/lS4gVH1HHLgHfvg3f+aUtX2x\n+rva2WihL3IXE22ZAbVzdt8D158Wg7K8H69/3n74W8+CUvRZTaTcgBXGDnEMg3rumgHz30Wa1Y5h\nc4LQiRJFGzja0FeV6m8AxiShox02oN63sj74rpXuQx2T+igmDfU8jzLNyvBZdcAGtOtTddY0goWd\nIFclKjaujzSfShR5V9h2hmxJrPkWGzrBgsic/GsRZdpePSJH6/YJfRvP2GMqzn3vUQBSzSm9Tt+e\nJKszNU35tgipWe9IZ+DVa4SXHboXHluzDduGzP2+JOhZHPSJ/hXPmuOWtckyWvnBkSb66oKw8pCE\nFGYYmpU/O024GVB/Ftp7LDCpv/rZ+4h19ZJgUQmrHiN2zIDuKjr77RVWhBWbBo1jBmkjUKCo9k3q\nuqpDHiohk7bqo/Tb1wff3z7qetJH4eMz+pC3L7IBfeYM4d2E3Le98ODUEGibAesawUI3X3HKBrTv\nS/9TmbgkSP6Fsj4qTNPirGqRSbDQfX61yNXqdwyXCwSF7JFyLuT66pEzESth7ouzqkXmv4ZmVYvw\nskP3QitmfPGPj5b3CfKYkjVT8150k9tQI04Fr31NcuTlc/fCHZ8+zmlHz0ZByJ+RHc8UTrAg8XdE\nhJccsqdYp29cyJsvVsJqUsBme9lzYFG6pfw4sGGbSVnWJ+T2NCvLDCj4AHz9C0XZNfaeTKbpwi2f\nkw0kM6BpplHtS/X4YrVyc1h+LEiz8pgB7dyLtRplSU/1etV+ZY4ZMMqzPyRxVmGaFQvUB99tTO2r\nGWZLffKfOSXPNSn5CTdb+QmLoO4tZsZIM2mwX/DDqfmw3/BZQfudjvVMs8rvvR4Rjjls7+A+6Vkc\n7Dx9Tv8tM72tqUo+RVXP3jP6ccDsaU7dM6do1HVtIXv5Q2vMegLvp24tNLLrI+DCDxyDP3z0NcY9\n2PelQ9rPKoS00g1UwqrHaNlmQMGxrODVrACcdeILjGO6GTBqy2clEyxGmxvQ7lsR7MlOcqTrsP0L\nvrbKqOu+j04V1fvRH0AscdiAQj1RuhofbgpmQC2rvo6IcmFV9whewJywATkzvW9SmtoXZWQEIlNI\nXPAPxxT6CTfv9Aurt73kQOw1vc9pXxEoADfQWcf0ftNEpmCbavsNM2CE6f11fPakI7z90kGkswrd\nxY5Z1nyXjhYcwZEqOXNRX6Tmv2d4qOsLVm/Fc58xM7u3UIKFQTaxvu3Z0/pw5MGztWNuf3TYZsB6\n5OYXHCtUwqrHUBkspAkRMAP0fBNLRIRD58zA+195qHFMn8Q71azUdT7qru94cRtyZz7y+meDCHjJ\nIS4FXe+f75gkOHUhbZpa3bJlmpXe7ZD7Dkm3pMyuG7ePpPWSVt4TZxXlaXLqtci7ELEfszSn+My7\n/bXICGLOE9cSjjx4dj4uhGe+2cqioeO7p74U93/uzdnfamEUawSLIoajHjCrN20H79ajXMtRk3XI\nAiOpNzeF1zxmR7tdIHnftqYpJU2WFqaGz0oXyFa6pYP2nJb9HfpNGzR+j2aq3wNQYAa0FkDT+msT\na/PFCv+/vXOPs6O47vzv9L3z0MxIM3qM3m9phJDQA0noYRAyEggJORZ2FBsMBD6AiR0MZGHXQOzF\nibEdSDYQJ2vI8glssFc2tsE2MotJCJCPN3YskMPDYJAzNoolLMJLwgihx8zU/tFV3dXVVd19X3N7\n5p7v5yPNvd3VVdV1u+vUOXXqVO0wN4QzOyZdWLm8pW1mMNLycq2z+otti6152TqwSh0sbPU1WTlr\nLF7+sy2xoJypmlVW13X511bnokfOqOu2+cQsLvvxwYVtziq6p1DUG1BYNyX0KJzU1tdZFT2KOBbE\nmyqbGfClmzdBj8itmwH7tGNA5WZA/d7WyjiO8yZ0OLXFqLDS2lFeoOZUSDuWFjLJxDe9hUIvLM9W\nf/NaU7OKz48GmpXDxD9Sj6BO0Tw7WpvC3QvSbwVAsmZlEmhWjnfUXA7R3lxkzapRUOGWSHuA/9+n\nzwjWaxzt0+MEJo+gddOHvlDYFcHi91ZMw+ROUzBEO7BgdO3onF1zWUm4OiL34sVs+Vld1/XPkRFt\n9AUrFsjpDWiu4QGSdwpWuMItRTopL7pBYueI0ETWP+DH2TOL8jwK4u8VPS/itq2babJMwNtG0CoP\nPZaheb+epTNXHEjQrExUtkIA550yDc/cdBZ6Jox0qg26icymmag5FX0tnRKoWYWVbkJPmzM2289M\nob/bQRrLHKiet26+JSNdR0vRakZMwjlnZRW+YZ9ho8VwfmlrLvCcVaMQbo0QPsDTxrThzz68CACw\naGrorZPkYAHEA2DqnZirwze7Z8/QrFSH5VoUXI5m5boPZ7DQlDL0aA6uc4C+Xiq+NUKT57k1K3Wd\no3MB/E7kGx9fnVjvYFGwYQZ8Vwqr686ah2lj2oJzaqdg2zorpeHoJuSiRxFvLdc6L52kOceBwGwZ\njxwempjjv1kW93CzjkLuPtDV1hyUaUPf/C+yrEPm0yKFVdRzUz4fxayalRYo2BHeyay/iwIlaFYO\ni4p+X/o+cICvdYWL2xOLDojM36VqVhTU2wYRRUyBI5oLrFk1CmYEC/WQnD6vG3tu2RJZEe/qWNTD\nHNesws+l7BRsLkwFKg+3pOPqiNz3l4w+N+c6p2dUIIoJ6UKBnOtF1C0mLU6+fO0srJkTXYcT9wYU\nkfxUnkqzmj9pVCT9M3sP4p2jfVYHi4FAWIXzO8WCh5ljwzBFZh1LcbAAopqVOZCwmckUh472ZXKb\n9+uvhFU87p2tavp8js0MqMyg+julntGsc1ak5Zc6Z2UcM5PYBlrqnvW0kzQLR8QDEYgEb/Y9Me1l\nuYhG88i2zippEKqbYtubi7zOqlFQgWz1Ub8L16S+erYjcdM8uynDxLb7qP5SBppVFV3XXYLTvXV4\nchlJJimLrIJH8fs+ePg4el87ZC9f03rNY67vZnrApVmF64tGthYjQvTrO3/tp495mIXRsgue3ikT\n7rxwOc48cYLM29TsbEGO3W3bHwhEsmxzAWvdAF+zsi0etaF7A+oQ/PtZYyzEdc1ZqXxUuf47FdW4\n4wLXXif9Hcg6x+Oi4NnmHCnyFwCWzRitXRMdFHVoArqjpagt1s727ukm3DTvRvU8JD0XX5JWH4A1\nq4binktOwfkrpzsXBWchcNc14qYFo8MEs4HZgZlrRUKPsOo9Kq57dI3G09pEnbUJcy/ycoYvYtbI\n4H756fWwTr675qzI3gGOdGxNHtt8kaLOD6FnIGFMezO2LZ8KwO0NuHHBhOBYshlQc1137Mlk1ayO\n9Fl3lLWhfhPboKngEb5xxWrconWOLm9A9XiqcknXrAp2B4s2h0Al0hcZa8+PJW2bETPQHLTYvAGt\nW84UPCye2onZ48wAvtGlJB26GdBa+zjNRjQPvw52YZfmYAFEo7A3FTz2BmwU3jdnHGaNaw8e8jI8\nwYOHtiniCRYP4WR7/OKdRNT0IbSOUGfq6BEoF1f/6BqNp8nvpA0i7ZpV3AyYiGUknKWObjNgVLNS\njGptgg1T6BW8UJAUvNAxRJmPio7fW/3Wd/3+Ciyb3mXN25beXBQMJJuLjvUPlGwGNAPM+uZoPw/d\nqcIMRaRQv40qV98AMnRhj9ZphCNGpUuzsv3G4zqarXkEeXlxE7xLK/3uH56Kx65bFzmmLlXtP7IM\nB4umiIOFMpPar01bVxnPmwYtkC3vZ5UTKtGs1DVJrut+GXH7l6nBE0W3NFDnTUHwvy5ajsmd5Qks\nt2ZlF1ZZLY1pc1bhi2ifv3EhNA3Dlq//PV52zBvQcly/ziWszLr2DYT7MkWdLdTgJFpeWL6+JQeC\n613oAtHtDWgXSpnNgI66EoW/++SucD6nJTIvq+ejhJVf7pHjA+EgThZiBh/W9+wy0ecBwzrF22pc\nRzRCuZkkaZ2VmV9SuLD25gKO9Q1ENauM74Ut3JLrncraD6m94wqee6632tRUsyKiTUS0m4h6iegG\ny/kWIvqmPL+TiGZq526Ux3cT0dlpeRLRLJlHr8yzuYIyuojofiJ6iYheJKI11W4bk8rMgP7fFsPB\nwhz9ZsnZo/haHyBuYmspehjdnjyqdJbheFPcmlVyzfXQQ0nXNheTR5UulJkjst4rVk78uth9KpOq\nwwzY4TADvnc86l239633oBwXCx5hUmcrRrc1BVFM1D0n9SG2XYtN1PU2M6C6yrWkoSWjsNK9Ac3j\n6pmbrDkZ2bQp/bN6B4729YdOII45q7am0jQrG+Yza6a2OTeFmlFi1pEMVeSOyJxVhsuBaJuFzh2V\naVaPXbcOD3xyzfCIYEFEBQBfAbAZwAIA5xORGe/kMgAHhBBzAdwO4FZ57QIA5wFYCGATgDuIqJCS\n560Abpd5HZB5l1yGvObLAB4RQswHsATAi9VpFTfZt0eMY9OsPIo7WNifT9P8ErWRuzSrrK/KnRcs\nw20fWWLU15621eFanCZcdA3ARD8STB4TWZ0NXCiX9qRpuyyeYkG4JS0fm3ZgoiLqtzUXcMPm+bhm\nQ09gelFrq56+aSM2yrmEUAAY9yjiHxOFlWVRcFjv5M7c9VuahO0WnztVdTMXiStsa6BsmpVrUbDu\nBq/jmrPKQnxRsDtNlkGT6hfUXF1LsVDS9SZpmpVufUhiUucILJ8xBgXPGxZbhKwE0CuE+JUQ4hiA\n+wBsNdJsBXCv/Hw/gA3k/xJbAdwnhDgqhHgZQK/Mz5qnvGa9zAMyz3PLKYOIOgGcDuBuABBCHBNC\nHKxSm7iRD0/az95uecHUM9tkmCzMUZJNINrmrKIOFmEMOluZaSye1oUPL5tqlGG/2Lkba0oZ4QJZ\ni7DSy9KFmnHfW5dOduYfCCtdszKKstXR1SG4NCsXKoq5R4RPrJuD6WPboPoHm5aqDpm/rT4vJCwC\n/muXrcRDV4VBTc9aMAE94zvwyffPia2zS4von1WzCuesosdJew7TFsPraZRmdeR4f/CjuLwBTQGs\n0LWhcmJf6iTFq8wkrGSSj63y95fqHtkS3ncZVVPvWNqazcwLjj0aFouCpwDYq33fJ49Z0wgh+gC8\nDWBswrWu42MBHJR5mGWVWsYsAK8D+N9E9DQR/R0RRV10akAwvkyZTLFNCtvXWUU1iUghGvG5gugW\nIQMOB4us74ntXS91QJjmiNjvEKhm+eEi17iDhdpXycbxPiWsEupo86yKeQNazIkE3Lh5Pj64xC0s\nD0lhpec2oM1ZuepiOi3o32yeiWt7unHSlDAu4+j2Zjx67TrM6e6IR2pQAyHHj2m6urtwaYHmc5h0\nLRBqAkqz0rdmD+asNGH1g2vWRpwwovnqVonKusikhbfZNCufS943Ey/dvMkXVgl5pxEMABzXBr9r\nRiHd0uSVtdayHNjBIk4RwDIAVwkhdhLRlwHcAOC/mwmJ6AoAVwDA9OnJW7WnkXXNRHtLAW8Yy4HU\nc9VkLGAMvY5kGZb84osxow+qOms+kJkXGSfsP5WVtA4jmL/ROqN7LlkBQnT1vzL9eRReE9TTqNPa\nnnE4Z9Ek3PidnwXbretp0ibW/XpHDw5YtCGPCH+wbk6YyNJ/vnssHhFCjWaTPCBjwkr7OpAg4LNA\niA6IFGr7EpfWYhIsCjaOm8/hvZeujISl8tPoQt//rLwBjxzvj4db0hwsTpw0CoeP2Xc0JgoHSFna\n58wTx+OfXnzNes4W3igUVqlZa84UYSitwEEj/fIYgaeo4+JSNavPbz0JnzftZTWilsLqFQDTtO9T\n5TFbmn1EVATQCeDNlGttx98E0EVERak96elLLWMfgH1CiJ3y+P3whVUMIcRdAO4CgBUrVlRkuA01\nq+R0bTbNSl7cHDEDat6ACQ+oPdySplk5TGxZ+zir40GJwipt/sNWx/Xz/bVEz7/ydnBMdyoYNzIq\nbEzGtDdjtAz9YzMDnnvyFIxsbcIPnt+Ph57b71izYmhWtkW5GdrikCV8kbK82MyArrVL+leb4CwF\nfz1e/HdpLnh4b6C/BM1K1s00AyLa0a+b1+28FoiHWzpi0axMM+CR4y5hpTlYZIh6cddFK5zme1Wv\nL31oEU6ZOVrmL+ufoe1tj0dgKCnLDEiJZZfquj6Y1FJ/ewpAj/TSa4bvzLDDSLMDwMXy8zYAjwt/\nqL8DwHnSk28WgB4AT7rylNc8IfOAzPPBcsoQQrwKYC8RnSCv2QAgfXvRClEPXtrEv23OSj1gzkC2\nwUgsfABvknv7WL2wtE5InTfnkzLvj2U5Vup7kOYGHURzsLy9likrEBHmdHfE1rToCBFGmO7rj5sB\nCx5hy+JJQQdouyVXIFudJKVxllwgesiy825/klOJQwAI25xVOT0e4pqPQmn35l5aznyCOavkxek2\nbF5ualHwkeOaN6A8Zwqr91zCCukOJOY9uAaE6v4+tmq6H6AXoXdjFi86+/uT3YxoUu11VoNJzYSV\n1HA+BeAf4HvTfUsI8QIRfZ6IPiiT3Q1gLBH1ArgWUoMRQrwA4FvwhcQjAK4UQvS78pR5XQ/gWpnX\nWJl3yWXIa64CsJ2IngOwFMCXqts6cbJqVmobhci1Fm9AorBzDdbfyO/nr5yOS0+bBQCYMCpuztJH\nky5Pu8zviXVkWKJmlSKsAq81q0nMfS96HD0TAaBFtqfNdT1Yuya/295th+e6kcbeFn+wbja+94en\n+p9Pl2ZCLanSXOzzdHYBYCs/i+ag+PDJU3Dz1oUA/E7PFoKruai81jKaAcNRWgRdu3Gh37pKOrrd\nX6s2YVRrzPHIrJPLDKhbFyqfs4ofWyBjQO55493U65OaoCzNKsUb0DRV5omazlkJIR4G8LBx7Cbt\n8xEAv+e49osAvpglT3n8V/C9Bc3j5ZTxDIAVtmtqhct0Y3LV+rlYNqMLF939pHat/zfquk6akArt\n3np6APjqpavw41++gc/teAHvHOnDgBAR4aAGf+VsBQK45qxKy8O2lbZOKIQs5Vs0K1PQ2BBCBJqV\nit2n5xVEBTHaVif2wts0K0ensGhKJzrbmrDnli3oHxB4+Pn9+G8bTwjO3/aRpdi+8z+wdFpX7FrV\nvqaw0svXg9Rm5TbNCeXC1dMjW8U/dNVp2P3qO/jLf9wNIB466tqz5uGpPW/F8nRFsLhozQy8nbDj\nsFl3lc/8iaNwxwXLsLZnHNb++RMA3OuslHm3tcnDkeOh2ZAo/K1LndOLhVuyXH+iFFYvvfpOphxj\nRwIzYOnvZNFzP69AvjUrdrDICYFmlZLO83wTVuSYvNiMDRibszLKAoCJna348LKpWDi5Ew//bD/a\nmouRF7dSzaoqc1YpoXsCN+40MyCikSjSqtFciApJ24scjkTj18cjWIS/rkf+QCBLn1DwCI9f9/7I\nsYmdrbhOE146wcDHOB6JYJFhUXASs7s7MFt7Dk+a0omTpnTilkdeAuAvXtW5ekOPNR8l9M262uao\nXNcCUXPmOYsmAQifc2XWNu91++Wr8OAzv8H3n/0NjhwP9+DyiDIFgc6C7Zmc3e1r9FsWT0q9Pun9\nKadmwZyV42Iz6k2eYGGVF5Q1JEMcIPMFcLmuqw4jeOESVPwTJo7ECRN9m7ouHESgWZXrDRinFNOT\nX59sZkD7mpbwmNlBJ92DQHxX1EBYadcFRVrySpqzGtPejDcOHYvVoZTFyi5UluaUiF5+sEC5yp2S\n2tnaFY3DxOVgUcq1QLKjicsqcPL00Th5+mh8/9nfxM6p367UZ9VsTvu8noeXbt6UyVRqK70iB4vU\nOSv5N4dRY3NYpcZECZws76z5/Ntc16EJK3MklvaQ6xGznZpVhnr6ZcVTjm1P9sQzMYWGiR6BPFa+\n9jkp0kUMEd8N2PYiJ20xHvcGDBkjQ1W5fotqRDQxJUBEWFXouu5C7WxtalYuwi1CSpdWruj1ikCz\nSgnHEHPu8PRFwaV1kaZwc2korU2FTAM+Wxp1pDwHC6VZ2a8tdZ3VYMKaVU4o5bmLPcAOb8CRMjBq\nsM4qoxdRNNyS/yKbEQyy1tf2zJtOHWmkzVn1a/Mv2y9fFel8omZAd51MBOJrhWwLYcONGeN5xDWr\nsF7KLd7lPl2JhhVoK8bxaAQLex0rRS3GHdXahPXzxwcmORehN2DpZdkcXnSCudyUe0xa45W6oafB\nlWfMxbtH+/DCb36Lp399sOylAUFZlmOVmAHVgDbNgtNwDhZMdoJHI8NL69KsdHdhjygwxSTtZ5WG\nqo4ZsDTrw2zTEMZ2lCasSjEDnjp3XLR8ixkwy4h27viRzg0Hox2QevnjeSR5AyrN6q13j6HaLJzc\niYvXzMClp83Cur/4Z2v5geCusrBS7dDRWsQ9l5ySmt7lDZiFqMNLUsLkfMxIFgTCKPnudI6IRsL/\n3pWnYnzCGr1RrU34wrmLcNHd/jLNSud+rJcHZsDS81Z9xPGU0UEeNSs2A+aE0+WE8onG1uY2TEER\nBrKNzqUoU0wQgcFxfRLBnFWZZkBbwlJfBNcOwopEb0CtzGCrj5Ty/vr8k3HNhh5n8NZoAFX/r22k\nGjMDakk+sW4OmgseVs4cAxuVmAELHuFPt56EGUmu+TUyAyqymgGDcEsVmgFtz1TWOsTz9R1IHvmj\ntVg9O/r7LJ3WFYkCn1a3cud+wj3J4vflVTD4bDXWDir+79Wn4YbN84Pv7GDBOPmdJZNx+rzu2EjO\nRtzBQv7VjhNRYD57T4brsbmuZyVm98+YR9ayrnF4i2WhP0Fj0ifZszoVrJk9Vm7lbvcGtM2V2Lra\nJKG8ZFoXfvHFzYn1qDbRcEv+31qZe1y7HpsEe29V6mBhuY97L12Jh57bj+4UTT5WtMxq/sT0gaOL\ntLmhNDzpLmq7PHjfyxjQqGfajJS+cHInFk4O40JWW+OuBiysckQWQQUAZMgNV+Tt9hb/wVSLH0k7\nVypxB4usZsB0bvvIklhk9lJIDurq/23yvMg27UlYHVYQtntUWEXroJM/U4o2Z4USnE3KYGRLtmfZ\nFRuwlGsBu1CYMbYdV54xNzUfWxSXSkma88pC0SO4DMRJyyXSMNcODiVYWA1BzJfpt5bFkx5REKH9\nsJzEr2QxYazjzqxZ2RNesGo6tu/8dcn1sNGf4LquhGpT0bNGGrdhLqIO84IsRzuWoFnlbZI6olkF\nmzfWpiw1UErDufdWBmzhltK4ev1cdBtzTrGI7yXXJE7BooWXc33iOqtyzIBKsxqkDROrCQurYUDv\na4dix4jCOIKHg9hypP1fGmW7rjuOf/FDi3D4WD+++/QrZZmAdHRvwFj5mpaUtZzUIJ8Rc6v/1/bu\nlzPyrbQtkrDvZ1UbaZXmLq5IMqOWQta2vtaykNr87fKgWSmHpiQLRin1VIvQlWY1WLv7VhN2sBiC\nmA/aO5ZApx5REKE9MANWolmVu0VIQrJq6R22jQTDc/7foueFDhYZzYDx42rS3DJnZZEylZjYaqGU\n2bwBqz2RvvmkiSWlD8yAFfad1ZxjqUaTeJUKqwTtKXj3Ssi66IiPOJRgzWoIMqq1iI+umIbzVk7D\nrj0HcPbCeAdBFG6Fbc5ZlfIynj6vGz/8xevxzfcyXj8YprDQDBg/d1zau5qLXrCmKa1OaTvTRsIt\nyb+xOHzI3yR1dFGw/7faitUdFywrSfCoJlrbMy45YQqVCF3TE7GcwZyJspqXbQb03MIqDEiTPe+/\nOf9kfOWJXrSX6SGZB4ZuzRsYIsKt2xYD8EPG2PCI0DOhHaPbmnDtxnnBMf9v9rLuuGCZdT+lPM3H\nJMUGVMFKmwqEw8eyalbRBDPHtkWORxcFuzWDvLn/RjUr5bpeXWlFRCUvcH/8unWY2NlaUbmVaLFx\nM2BFVQGgb8VR7vVuM2A5DhZnL5yIsxdOtAYTHiqwsBqmKDPg0zdtDI5RGSOy1qKHDltHktnBIj1N\nloF4R0vRuq8TAFy0egae3XvQuuWHctEd0VzEwcO+I0qaoNVPP/u5jYHpxLYoONSs4vnkwRuQKBSk\nuqlS1bdWDhalMNsIzFwOFY0LjN+uKpqVp/7WQrMq38HCDCE2lGBhNUyxvSOUcM6F68XN7A2YJNVK\nqMdP/ngD+vvtYm3b8qnYttzu+r5g0ih8fO0sXPy+mdi15wA+/cBzQagjF7pGpC8nsDtYKAeB9EXB\n9YBgHwxk9YwcKlSixcY3fqy0NvY1eaVg7pSgE77HpedtLnQfSrCwGqYkT8xmf8hdKTN7A2bRrDJM\ncpQdjcAjfGaLvyvy1NFtOPfkKc60S6Z24tl9b6cK6IhmJT9mCbeUhWp7A5KmWkXzro0ZsF5UosWa\nTV5J9JAgD8v8ZikkaVbqYDk5s4MFkxtU35Rkysjy/my/fBUe+Ok+d1TwrN6AiefyNar/6qWr0Pv6\nOwkOFv58TDRygv+36t6AZV8ZxSNAhcrVtb+BGjlY1ItKTHexdVZVfCwrWRQMALa1u0nb0qRRqmZV\nqndnLWFhNcxoKng41jeQaCLIIiROnTsuFhQ2mkc2qmH/Hyw625qwfIY9Vp9C39QSCNvSNmdVUQda\n9pVGHTRDoG2LkDzMq9WbmGZVxSYp3wwoY/gNuCNNlPPTlSKs9tyypfQCasgwGVcxCjWBanuQRbB4\ntvJyss9ZDS88sodbquVi3orQfgDb5ot581isB7UIt6QavnxvQP9v0uLdsmIDFrJFFskjLKyGGU3B\nttUJmlU1hFXW2IBV8gbMC2RoVuoGbeusyuHS02YBAFbMtC9JKBW9qnodk3ZXbnSqIqoqnbMiZQZ0\nP1e11qzyBpsBhxnqYUyMHFEFaVVpbMBS8sgTHtm3ta+WwF05a0xVzS+uQUWgWbGwilFND8lKvQET\nNasGE1bN41Q1AAAMN0lEQVRDt+aMFbX3k22grw5VRbOqZh83hFQrj8i6rX05gVgHA10WRRYFBw4W\nLKxGGp6meXCwyCasSs97KA9OWFgNM9TIKWkLgKq45lYlj6GH6WCRZc4qy4aatSLSoVkcLIbLOqtK\n+N6nTsXNWxcG36vpFFRpBItqa1ZDGRZWwwzlYHHcIqzChaCVl9NoL4qCDAeLMOq6vVPZ9dkz8fXL\nVw1G1azoP5M+D6b6wAb9GSPM6e7ARWtmVjXPoF3LfFFu3HwiesZ3YMm0rti5IMhtg/16LKyGGddv\nno9RrUVMH9PmTFMdB4vGJOa6nrLFxbiOFoxorp8Hlvqtv/7xVbhO2x7jLz+yBHPHdwS7STPAD65Z\niz8+Z356wlIo0zy8ZFoXHr12nTXwbJeMwNJoA0Z2sBhmnHHCeDz3J2dbz6lFodUw/VTTVGILVZRX\nTAeL8XIjv3EJW6fX09Smfqe53R0RIXvOokk4Z9GkelUrl5w4aVRdTbZZGdfhC6shPP1UFiysmLKo\npntvTn0TrPgOFuHd/+6yqWhtKiR2/PXsVMIIWw3Ws9WZ4NmuQd5qYPTesaG3NX0lsLBqQKqjWVWh\nIkMQIoqGW/IIv7NkcuI19dSsKonQzZRPLeeTxrT7mtVb7x4t6/ovnHtSYBEYSrCwakCq47pePY/C\nIaRY+WbAElWlegoKVfRQ0l6ZZMZKM+Ab7x4r6/oLV8+oZnUGDXawaCAEe4BVjEdUsqZUTxNc0hYm\nzNBEmQHfPFSeZjVUYc2qAeGFoCH/cv0ZJa3qL0ezqidDcV6QSWZy1wgAQNeI5H3ZhhusWTUgeelq\nJ3X5OxCnbYZYS6aObsP4kdm3VC8UaEjtAfWBxb7jh80FmqkdtRwkzBrXjr+9cDlu/d3F1c88x/AT\n3ECE4ZbyIa6uPGMu5o7vwNkLJ9S7Kpn53AcWBkJ2KPDZLQtw9fqesjevZMojnCusjUq7KUf7TA0W\n/AQ3IDmRVWgqePjA4mRPurxx5oKhI1gB32Q5ur2xzEXM8KSm9gwi2kREu4mol4husJxvIaJvyvM7\niWimdu5GeXw3EZ2dlicRzZJ59Mo8mysoYw8R/YyIniGiXdVul3oROliUL60+c86J6GprqlKNGIZh\nslEzYUVEBQBfAbAZwAIA5xPRAiPZZQAOCCHmArgdwK3y2gUAzgOwEMAmAHcQUSElz1sB3C7zOiDz\nLrkMrW5nCCGWCiFWVKVBcsDBw76rayXC5uOnz8YzN22sVpUahuUzRuNLH1pU72owgwz7tVSPWmpW\nKwH0CiF+JYQ4BuA+AFuNNFsB3Cs/3w9gA/kTKlsB3CeEOCqEeBlAr8zPmqe8Zr3MAzLPc8ssY9jS\nJ6OXTuwcOnMuw4UHPvk+fGzV9HpXg2GGLLUUVlMA7NW+75PHrGmEEH0A3gYwNuFa1/GxAA7KPMyy\nSi0D8AdE/0hEPyWiK1w3SERXENEuItr1+uuvu5LljkksrBimpuTFiWk4MXR8cAeX04QQy+CbG68k\notNtiYQQdwkhVgghVnR3dw9uDSugOyHoKsMw1YPXt1WPWnoDvgJgmvZ9qjxmS7OPiIoAOgG8mXKt\n7fibALqIqCi1Jz19yWUIIdTf14jou/DNgz8s5ebzTLFQ+zHKw1evxS/+852al8M0Nt/+xBoMJGxQ\nWC/Gj/IHhGPYE7Nq1LLXegpAj/TSa4bvzLDDSLMDwMXy8zYAjwt/YcIOAOdJT75ZAHoAPOnKU17z\nhMwDMs8HyymDiNqJaCQAEFE7gI0Anq9SmzQMCyaPwrknm1Zfhqkup8wcg1Wzx9a7GjGuWDsbf/XR\npdi6dGgtzcgzNdOshBB9RPQpAP8AoADgHiHEC0T0eQC7hBA7ANwN4GtE1AvgLfjCBzLdtwD8HEAf\ngCuFEP0AYMtTFnk9gPuI6AsAnpZ5o9QyiGgCgO9Km3MRwNeFEI/UqJkGlTsvWDYoWhXDNDrFgseD\ntSpDtVph3WisWLFC7No1bJZkMQzD1Bwi+mnW5UE8zGYYhmFyDwsrhmEYJvewsGIYhmFyDwsrhmEY\nJvewsGIYhmFyDwsrhmEYJvewsGIYhmFyDwsrhmEYJvfwouAqQUSvA/iPMi8fB+CNKlZnqMLt4MPt\nEMJt4TNc22GGECJTFHAWVjmAiHYNp00ey4XbwYfbIYTbwofbgc2ADMMwzBCAhRXDMAyTe1hY5YO7\n6l2BnMDt4MPtEMJt4dPw7cBzVgzDMEzuYc2KYRiGyT0srBiGYZjcw8KqjhDRJiLaTUS9RHRDvetT\na4joHiJ6jYie146NIaJHiejf5d/R8jgR0V/LtnmOiJbVr+bVhYimEdETRPRzInqBiK6RxxuqLYio\nlYieJKJnZTv8qTw+i4h2yvv9JhE1y+Mt8nuvPD+znvWvNkRUIKKniegh+b0h28EFC6s6QUQFAF8B\nsBnAAgDnE9GC+taq5vw9gE3GsRsAPCaE6AHwmPwO+O3SI/9dAeDOQarjYNAH4DohxAIAqwFcKX/7\nRmuLowDWCyGWAFgKYBMRrQZwK4DbhRBzARwAcJlMfxmAA/L47TLdcOIaAC9q3xu1HaywsKofKwH0\nCiF+JYQ4BuA+AFvrXKeaIoT4IYC3jMNbAdwrP98L4Fzt+FeFz08AdBHRpMGpaW0RQuwXQvyb/PwO\n/A5qChqsLeT9HJJfm+Q/AWA9gPvlcbMdVPvcD2ADEdEgVbemENFUAFsA/J38TmjAdkiChVX9mAJg\nr/Z9nzzWaEwQQuyXn18FMEF+boj2kSackwHsRAO2hTR9PQPgNQCPAvglgINCiD6ZRL/XoB3k+bcB\njB3cGteMvwLwaQAD8vtYNGY7OGFhxeQG4a+jaJi1FETUAeABAH8khPitfq5R2kII0S+EWApgKnxr\nw/w6V2nQIaIPAHhNCPHTetclz7Cwqh+vAJimfZ8qjzUa/6lMWvLva/L4sG4fImqCL6i2CyG+Iw83\nZFsAgBDiIIAnAKyBb+YsylP6vQbtIM93AnhzkKtaC04F8EEi2gN/OmA9gC+j8dohERZW9eMpAD3S\n46cZwHkAdtS5TvVgB4CL5eeLATyoHf996Qm3GsDbmolsSCPnF+4G8KIQ4jbtVEO1BRF1E1GX/DwC\nwFnw5++eALBNJjPbQbXPNgCPi2EQ1UAIcaMQYqoQYib8fuBxIcQFaLB2SEUIwf/q9A/AOQB+Ad9O\n/5l612cQ7vcbAPYDOA7fBn8ZfFv7YwD+HcA/ARgj0xJ8b8lfAvgZgBX1rn8V2+E0+Ca+5wA8I/+d\n02htAWAxgKdlOzwP4CZ5fDaAJwH0Avg2gBZ5vFV+75XnZ9f7HmrQJu8H8FCjt4PtH4dbYhiGYXIP\nmwEZhmGY3MPCimEYhsk9LKwYhmGY3MPCimEYhsk9LKwYhmGY3MPCimFyCBHN1KPTZ0h/CRFNzpDm\nf1ZeO4YZfFhYMczw4BIAicKKYYYyLKwYJr8UiWg7Eb1IRPcTURsR3URETxHR80R0l4xqsQ3ACgDb\niegZIhpBRKcQ0Y/lXlFPEtFImedkInpE7pn156ogItpIRP9KRP9GRN+WcQtBRLfIfbeeI6L/UYc2\nYBgA4EXBDJNHZDT2lwGcJoT4ERHdA+DnAO4RQrwl03wNwLeEEN8non8G8F+FELtk+K6XAHxUCPEU\nEY0CcBjAhQBugh/l/SiA3fCjabwH4DsANgsh3iWi6wG0wI+a8WMA84UQgoi6hB/Dj2EGnWJ6EoZh\n6sReIcSP5Of/A+BqAC8T0acBtAEYA+AFAN83rjsBwH4hxFMAIGREd7nl0WNCiLfl958DmAGgC/4G\noD+SaZoB/Cv8rSeOALhb7l77UG1uk2HSYWHFMPnFNHsIAHfAjw24l4j+BH6cuFI4qn3uh98HEIBH\nhRDnm4mJaCWADfADpn4KfkRwhhl0eM6KYfLLdCJaIz9/DMC/yM9vyDmlbVradwCoeandACYR0SkA\nQEQjta0mbPwEwKlENFembyeiebKMTiHEwwD+C4AlVbkrhikD1qwYJr/sBnClNl91J4DR8COUvwp/\nmxnF3wP4WyJ6D/6eUB8F8Ddy6433AJzpKkQI8ToRXQLgG0TUIg9/Fr4AfJCIWuFrX9dW79YYpjTY\nwYJhGIbJPWwGZBiGYXIPCyuGYRgm97CwYhiGYXIPCyuGYRgm97CwYhiGYXIPCyuGYRgm97CwYhiG\nYXLP/wcwj8Ma10QW0wAAAABJRU5ErkJggg==\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "OxNAhayeP7B-",
        "colab_type": "text"
      },
      "source": [
        "#### Debug"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "OoKj0c8CV88j",
        "colab_type": "code",
        "outputId": "7e60cf95-5c89-475c-b8f9-b0a34a30a426",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 51
        }
      },
      "source": [
        "np.argmax(inputTestSet[0], axis=1)"
      ],
      "execution_count": 133,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "array([ 9, 12, 15,  0,  2, 11, 16, 13,  0,  0,  0,  0,  0,  0,  0,  0,  0,\n",
              "        0,  0,  0])"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 133
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "GIjmJSBlV_Nk",
        "colab_type": "code",
        "outputId": "42351557-23e1-4820-9f19-2f36983a997d",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 51
        }
      },
      "source": [
        "np.argmax(targetTestSet[0], axis=1)"
      ],
      "execution_count": 134,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "array([ 9, 12, 15,  0,  2, 11, 16, 13,  0,  2,  0,  0, 11, 13,  0, 16,  0,\n",
              "        0,  0,  0])"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 134
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "cCh2vI9eo8DY",
        "colab_type": "code",
        "outputId": "ae65870b-9890-41f7-f678-5da5d4fcf745",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 17680
        }
      },
      "source": [
        "## Test\n",
        "test_loss_history2_, outputMemory = testModel([inputTestSet[0]], [targetTestSet[0]], gates, model2, readMemory, writeMemory, debug=True)"
      ],
      "execution_count": 135,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "\n",
            "\n",
            "outputs\n",
            "Timestep:  0\n",
            "\n",
            " Gate Controller vals:\n",
            "[[0.12939098 0.27711508 0.59349394]]\n",
            "\n",
            " Gate Controller vals:\n",
            "[[0.2631662  0.24550663 0.23931772 0.25200948]]\n",
            "\n",
            " Gate Controller vals:\n",
            "[[0.16784763 0.19264847 0.21338378 0.22956415 0.19655594]]\n",
            "\n",
            " Gate Controller vals:\n",
            "[[0.18549913 0.19599164 0.1518271  0.16398153 0.14852326 0.15417746]]\n",
            "\n",
            " Gate Controller vals:\n",
            "[[0.065294   0.065294   0.065294   0.29949227 0.065294   0.13983937\n",
            "  0.29949227]]\n",
            "\n",
            " Gate Controller vals:\n",
            "[[0.14922522 0.10906383 0.11728441 0.10716963 0.15904632 0.12333504\n",
            "  0.1120802  0.1227954 ]]\n",
            "[[0.11999992 0.11066956 0.12215625 0.11564429 0.11309431 0.14789505\n",
            "  0.11817537 0.15236527]]\n",
            "\n",
            " Gate Controller vals:\n",
            "[[0.12440063 0.1074911  0.12180342 0.10052332 0.10995676 0.0977047\n",
            "  0.11574163 0.10516009 0.11721843]]\n",
            "[[0.10909406 0.09161556 0.08154092 0.11635105 0.13675518 0.11910682\n",
            "  0.13568763 0.10948407 0.10036466]]\n",
            "\n",
            " Gate Controller vals:\n",
            "[[0.12589358 0.12589358 0.12589358 0.26962474 0.05878242 0.05878242\n",
            "  0.05878242 0.05878242 0.05878242 0.05878242]]\n",
            "\n",
            " Gate Controller vals:\n",
            "[[0.08971894 0.09889004 0.1066973  0.10353603 0.0767145  0.0793992\n",
            "  0.08632424 0.10636942 0.08049607 0.08961201 0.08224219]]\n",
            "[[0.0876548  0.09282148 0.10204001 0.09278339 0.0788233  0.09098488\n",
            "  0.08929776 0.09499869 0.09396973 0.08265966 0.09396626]]\n",
            "\n",
            " Gate Controller vals:\n",
            "[[0.07782298 0.07423282 0.08939794 0.08693857 0.08114914 0.09122683\n",
            "  0.08196668 0.08426211 0.07254578 0.08266234 0.09329425 0.08450064]]\n",
            "[[0.0758699  0.08320087 0.09795909 0.08320547 0.084737   0.08610249\n",
            "  0.08447292 0.09139789 0.08534751 0.07773978 0.07551478 0.07445231]]\n",
            "\n",
            " Gate Controller vals:\n",
            "[[0.07060593 0.06867059 0.07823165 0.08978839 0.09337639 0.0855656\n",
            "  0.07755401 0.07515675 0.06232182 0.07593726 0.06466901 0.08289076\n",
            "  0.07523178]]\n",
            "[[0.08132195 0.08035218 0.06909917 0.07213109 0.07971601 0.08288151\n",
            "  0.06768163 0.0701554  0.07185324 0.07914716 0.08385336 0.08628993\n",
            "  0.07551739]]\n",
            "\n",
            " Gate Controller vals:\n",
            "[[0.06855766 0.08505335 0.07149657 0.06272615 0.06364758 0.09310973\n",
            "  0.07013394 0.07862148 0.07732263 0.06549948 0.06872837 0.06076142\n",
            "  0.06700742 0.06733422]]\n",
            "[[0.0725544  0.07280798 0.08088204 0.07681216 0.06180235 0.06769092\n",
            "  0.07615747 0.09118621 0.06662131 0.06676576 0.07644101 0.06905313\n",
            "  0.06160033 0.05962498]]\n",
            "\n",
            " Gate Controller vals:\n",
            "[[0.06591091 0.06524961 0.06333394 0.07174416 0.06133398 0.07679569\n",
            "  0.057627   0.07586572 0.06957373 0.06027362 0.07279118 0.07064705\n",
            "  0.07287309 0.06020939 0.05577089]]\n",
            "[[0.06993041 0.07002646 0.07092936 0.07094271 0.06118596 0.07114229\n",
            "  0.05517527 0.0539999  0.06987579 0.06276546 0.06749446 0.06809233\n",
            "  0.07086487 0.07136325 0.06621143]]\n",
            "\n",
            " Gate Controller vals:\n",
            "[[0.03736854 0.03736854 0.03736854 0.17140299 0.03736854 0.03736854\n",
            "  0.03736854 0.17140299 0.03736854 0.03736854 0.17140299 0.03736854\n",
            "  0.03736854 0.03736854 0.03736854 0.03736854]]\n",
            "[[0.03583951 0.03583951 0.03583951 0.16438961 0.03583951 0.03583951\n",
            "  0.07675704 0.16438961 0.03583951 0.03583951 0.16438961 0.03583951\n",
            "  0.03583951 0.03583951 0.03583951 0.03583951]]\n",
            "\n",
            "Register controller  0\n",
            "[[0.02987847 0.02987847 0.02987847 0.06399035 0.02987847 0.06399035\n",
            "  0.13704735 0.13704735 0.02987847 0.02987847 0.13704735 0.02987847\n",
            "  0.06399035 0.06399035 0.02987847 0.02987847 0.06399035]]\n",
            "\n",
            "Register controller  1\n",
            "[[0.02987847 0.02987847 0.02987847 0.06399035 0.02987847 0.06399035\n",
            "  0.13704735 0.13704735 0.02987847 0.02987847 0.13704735 0.02987847\n",
            "  0.06399035 0.06399035 0.02987847 0.02987847 0.06399035]]\n",
            "\n",
            "Register controller  2\n",
            "[[0.03177928 0.03177928 0.03177928 0.06806128 0.03177928 0.06806128\n",
            "  0.14576602 0.14576602 0.03177928 0.03177928 0.08214805 0.03177928\n",
            "  0.06806128 0.06806128 0.03177928 0.03177928 0.06806128]]\n",
            "\n",
            "Output registers\n",
            "[[[0.3014004  0.24772024 0.16274278 0.04432949 0.         0.\n",
            "   0.         0.         0.03990864 0.07756428 0.04432949 0.\n",
            "   0.         0.         0.         0.         0.         0.\n",
            "   0.         0.08200473]\n",
            "  [0.3014004  0.24772024 0.16274278 0.04432949 0.         0.\n",
            "   0.         0.         0.03990864 0.07756428 0.04432949 0.\n",
            "   0.         0.         0.         0.         0.         0.\n",
            "   0.         0.08200473]\n",
            "  [0.31585854 0.2592171  0.17197612 0.04714965 0.         0.\n",
            "   0.         0.         0.02529456 0.08137878 0.04714965 0.\n",
            "   0.         0.         0.         0.         0.         0.\n",
            "   0.         0.05197556]]]\n",
            "\n",
            "Output memory\n",
            "[[0.11683315 0.05635607 0.03582354 0.01718364 0.         0.\n",
            "  0.         0.         0.01546996 0.70936215 0.01718364 0.\n",
            "  0.         0.         0.         0.         0.         0.\n",
            "  0.         0.03178785]\n",
            " [0.05635606 0.02718412 0.01727997 0.00828876 0.         0.\n",
            "  0.         0.         0.00746215 0.02816328 0.00828876 0.\n",
            "  0.8316436  0.         0.         0.         0.         0.\n",
            "  0.         0.0153333 ]\n",
            " [0.02154235 0.01039125 0.00660534 0.00316841 0.         0.\n",
            "  0.         0.         0.00285244 0.01076553 0.00316841 0.\n",
            "  0.         0.         0.         0.93564504 0.         0.\n",
            "  0.         0.00586122]\n",
            " [0.01718364 0.00828876 0.00526887 0.00252734 0.         0.\n",
            "  0.         0.         0.0022753  0.00858732 0.00252734 0.\n",
            "  0.         0.         0.         0.         0.         0.\n",
            "  0.         0.00467531]\n",
            " [0.         0.         1.         0.         0.         0.\n",
            "  0.         0.         0.         0.         0.         0.\n",
            "  0.         0.         0.         0.         0.         0.\n",
            "  0.         0.        ]\n",
            " [0.         0.         0.         0.         0.         0.\n",
            "  0.         0.         0.         0.         0.         1.\n",
            "  0.         0.         0.         0.         0.         0.\n",
            "  0.         0.        ]\n",
            " [0.         0.         0.         0.         0.         0.\n",
            "  0.         0.         0.         0.         0.         0.\n",
            "  0.         0.         0.         0.         1.         0.\n",
            "  0.         0.        ]\n",
            " [0.         0.         0.         0.         0.         0.\n",
            "  0.         0.         0.         0.         0.         0.\n",
            "  0.         1.         0.         0.         0.         0.\n",
            "  0.         0.        ]\n",
            " [0.01546996 0.00746215 0.00474342 0.0022753  0.         0.\n",
            "  0.         0.         0.00204839 0.00773093 0.0022753  0.\n",
            "  0.         0.         0.         0.         0.         0.\n",
            "  0.         0.00420905]\n",
            " [0.05838598 0.02816328 0.01790239 0.00858732 0.         0.\n",
            "  0.         0.         0.00773093 0.0291777  0.00858732 0.\n",
            "  0.         0.         0.         0.         0.         0.\n",
            "  0.         0.0158856 ]\n",
            " [0.01718364 0.00828876 0.00526887 0.00252734 0.         0.\n",
            "  0.         0.         0.0022753  0.00858732 0.00252734 0.\n",
            "  0.         0.         0.         0.         0.         0.\n",
            "  0.         0.00467531]\n",
            " [0.         0.         0.         0.         0.         0.\n",
            "  0.         0.         0.         0.         0.         0.\n",
            "  0.         0.         0.         0.         0.         0.\n",
            "  0.         0.        ]\n",
            " [0.         0.         0.         0.         0.         0.\n",
            "  0.         0.         0.         0.         0.         0.\n",
            "  0.         0.         0.         0.         0.         0.\n",
            "  0.         0.        ]\n",
            " [0.         0.         0.         0.         0.         0.\n",
            "  0.         0.         0.         0.         0.         0.\n",
            "  0.         0.         0.         0.         0.         0.\n",
            "  0.         0.        ]\n",
            " [0.         0.         0.         0.         0.         0.\n",
            "  0.         0.         0.         0.         0.         0.\n",
            "  0.         0.         0.         0.         0.         0.\n",
            "  0.         0.        ]\n",
            " [0.         0.         0.         0.         0.         0.\n",
            "  0.         0.         0.         0.         0.         0.\n",
            "  0.         0.         0.         0.         0.         0.\n",
            "  0.         0.        ]\n",
            " [0.         0.         0.         0.         0.         0.\n",
            "  0.         0.         0.         0.         0.         0.\n",
            "  0.         0.         0.         0.         0.         0.\n",
            "  0.         0.        ]\n",
            " [0.         0.         0.         0.         0.         0.\n",
            "  0.         0.         0.         0.         0.         0.\n",
            "  0.         0.         0.         0.         0.         0.\n",
            "  0.         0.        ]\n",
            " [0.         0.         0.         0.         0.         0.\n",
            "  0.         0.         0.         0.         0.         0.\n",
            "  0.         0.         0.         0.         0.         0.\n",
            "  0.         0.        ]\n",
            " [0.03178785 0.0153333  0.00974683 0.00467531 0.         0.\n",
            "  0.         0.         0.00420905 0.0158856  0.00467531 0.\n",
            "  0.         0.         0.         0.         0.         0.\n",
            "  0.         0.00864881]]\n",
            "\n",
            "\n",
            "outputs\n",
            "Timestep:  1\n",
            "\n",
            " Gate Controller vals:\n",
            "[[0.12939098 0.27711508 0.59349394]]\n",
            "\n",
            " Gate Controller vals:\n",
            "[[0.2558368  0.24764998 0.24626298 0.25025022]]\n",
            "\n",
            " Gate Controller vals:\n",
            "[[0.18258627 0.19586183 0.20958553 0.21748894 0.19447735]]\n",
            "\n",
            " Gate Controller vals:\n",
            "[[0.17447357 0.18423955 0.1622207  0.16612358 0.15586638 0.15707627]]\n",
            "\n",
            " Gate Controller vals:\n",
            "[[0.06529437 0.06529437 0.06529437 0.29949394 0.06529437 0.13984016\n",
            "  0.29948837]]\n",
            "\n",
            " Gate Controller vals:\n",
            "[[0.13924262 0.11831887 0.12154431 0.11413009 0.14262639 0.1239147\n",
            "  0.11745641 0.12276659]]\n",
            "[[0.12157916 0.11833699 0.12129699 0.11866491 0.12010201 0.14002475\n",
            "  0.12011316 0.13988201]]\n",
            "\n",
            " Gate Controller vals:\n",
            "[[0.1206134  0.10969709 0.11761013 0.10387305 0.1107465  0.10275798\n",
            "  0.11312961 0.10729887 0.11427339]]\n",
            "[[0.11004606 0.10015415 0.09256568 0.11251201 0.13098545 0.11540382\n",
            "  0.12407875 0.10904531 0.10520877]]\n",
            "\n",
            " Gate Controller vals:\n",
            "[[0.12589358 0.12589358 0.12589358 0.26962474 0.05878242 0.05878242\n",
            "  0.05878242 0.05878242 0.05878242 0.05878242]]\n",
            "\n",
            " Gate Controller vals:\n",
            "[[0.09043793 0.09548497 0.10125713 0.09647685 0.08504583 0.08340758\n",
            "  0.08844835 0.09891181 0.08436575 0.09007116 0.08609267]]\n",
            "[[0.08938475 0.09077056 0.09713762 0.09280413 0.08379717 0.09106476\n",
            "  0.09005231 0.09408909 0.09247568 0.08689532 0.09152862]]\n",
            "\n",
            " Gate Controller vals:\n",
            "[[0.07862488 0.07701647 0.08664188 0.08590685 0.0832361  0.08713894\n",
            "  0.08233447 0.08383643 0.07872797 0.08340102 0.08851557 0.08461945]]\n",
            "[[0.07721212 0.08222564 0.09310756 0.08297201 0.08435009 0.08409546\n",
            "  0.08467897 0.08785923 0.0863077  0.07939035 0.0802012  0.0775997 ]]\n",
            "\n",
            " Gate Controller vals:\n",
            "[[0.07369573 0.07215073 0.07922295 0.08273377 0.08479172 0.08052418\n",
            "  0.07723457 0.07564794 0.06863218 0.07652713 0.07152015 0.08052517\n",
            "  0.07679379]]\n",
            "[[0.07992709 0.07896143 0.07193691 0.07360608 0.07960574 0.08025058\n",
            "  0.06877965 0.0744943  0.0720326  0.07830232 0.08193745 0.08366014\n",
            "  0.07650573]]\n",
            "\n",
            " Gate Controller vals:\n",
            "[[0.0699094  0.07844982 0.07246014 0.06581876 0.06637301 0.08286405\n",
            "  0.07089508 0.07609079 0.07655696 0.0674642  0.06918014 0.06492677\n",
            "  0.06851463 0.07049624]]\n",
            "[[0.07200444 0.07205956 0.07643921 0.07484894 0.06668057 0.06971238\n",
            "  0.07449663 0.08241363 0.06645306 0.06822176 0.07405166 0.06967258\n",
            "  0.06703816 0.06590737]]\n",
            "\n",
            " Gate Controller vals:\n",
            "[[0.06654433 0.06599724 0.06648845 0.06937303 0.06284907 0.07199989\n",
            "  0.06157716 0.07161652 0.06874435 0.06110669 0.07056414 0.06879401\n",
            "  0.07028924 0.06254165 0.06151417]]\n",
            "[[0.06819955 0.0692827  0.06972012 0.0685743  0.063514   0.07121502\n",
            "  0.05968576 0.05798149 0.0683898  0.0644357  0.06781677 0.06757411\n",
            "  0.07035393 0.06788153 0.06537522]]\n",
            "\n",
            " Gate Controller vals:\n",
            "[[0.03736854 0.03736854 0.03736854 0.17140299 0.03736854 0.03736854\n",
            "  0.03736854 0.17140299 0.03736854 0.03736854 0.17140299 0.03736854\n",
            "  0.03736854 0.03736854 0.03736854 0.03736854]]\n",
            "[[0.03583951 0.03583951 0.03583951 0.16438961 0.03583951 0.03583951\n",
            "  0.07675704 0.16438961 0.03583951 0.03583951 0.16438961 0.03583951\n",
            "  0.03583951 0.03583951 0.03583951 0.03583951]]\n",
            "\n",
            "Register controller  0\n",
            "[[0.02987847 0.02987847 0.02987847 0.06399035 0.02987847 0.06399035\n",
            "  0.13704735 0.13704735 0.02987847 0.02987847 0.13704735 0.02987847\n",
            "  0.06399035 0.06399035 0.02987847 0.02987847 0.06399035]]\n",
            "\n",
            "Register controller  1\n",
            "[[0.03009833 0.03009833 0.03009833 0.06446123 0.03009833 0.06446123\n",
            "  0.13805583 0.13805583 0.03009833 0.03009833 0.13069722 0.03009833\n",
            "  0.06446123 0.06446123 0.03009833 0.03009833 0.06446123]]\n",
            "\n",
            "Register controller  2\n",
            "[[0.03157115 0.03157115 0.03157115 0.06761554 0.03157115 0.06761554\n",
            "  0.14481136 0.14481136 0.03157115 0.03157115 0.08815933 0.03157115\n",
            "  0.06761554 0.06761554 0.03157115 0.03157115 0.06761554]]\n",
            "\n",
            "Output registers\n",
            "[[[0.16794899 0.3643815  0.19211523 0.05517012 0.00172297 0.\n",
            "   0.         0.00229133 0.01781394 0.03014373 0.01780341 0.01017121\n",
            "   0.01641955 0.00938414 0.00628149 0.01220837 0.00697736 0.\n",
            "   0.00470824 0.04545312]\n",
            "  [0.16749805 0.3660454  0.19323865 0.05557106 0.00173564 0.\n",
            "   0.         0.00219427 0.0172591  0.03018447 0.01792938 0.00982605\n",
            "   0.01651295 0.0094532  0.00601543 0.01227781 0.0070287  0.\n",
            "   0.00450882 0.04393821]\n",
            "  [0.16447724 0.37719145 0.20076418 0.05825685 0.00182058 0.\n",
            "   0.         0.00154414 0.01354234 0.03045735 0.01877325 0.00751391\n",
            "   0.01713862 0.00991577 0.00423315 0.01274302 0.00737264 0.\n",
            "   0.00317292 0.03379025]]]\n",
            "\n",
            "Output memory\n",
            "[[1.31779641e-01 1.03739880e-01 5.60052544e-02 2.61433721e-02\n",
            "  3.82353493e-04 0.00000000e+00 0.00000000e+00 5.08481346e-04\n",
            "  1.64672118e-02 5.78094184e-01 1.78511087e-02 2.25715153e-03\n",
            "  7.07577588e-03 2.08248966e-03 1.39396195e-03 5.26102539e-03\n",
            "  1.54838606e-03 0.00000000e+00 1.04483345e-03 3.58007103e-02]\n",
            " [9.77544039e-02 1.08681545e-01 5.38637899e-02 2.47329678e-02\n",
            "  5.85589907e-04 0.00000000e+00 0.00000000e+00 7.78759830e-04\n",
            "  1.14400275e-02 3.55814025e-02 1.20330304e-02 3.45691945e-03\n",
            "  5.87980986e-01 3.18941777e-03 2.13490939e-03 8.05747416e-03\n",
            "  2.37141596e-03 0.00000000e+00 1.60020485e-03 2.65145767e-02]\n",
            " [3.85519564e-02 3.87023613e-02 1.96277145e-02 9.05410945e-03\n",
            "  1.91359621e-04 0.00000000e+00 0.00000000e+00 2.54483864e-04\n",
            "  4.61331103e-03 1.49294343e-02 4.90401220e-03 1.12965528e-03\n",
            "  3.54127190e-03 1.04224088e-03 6.97647629e-04 8.44712198e-01\n",
            "  7.74933549e-04 0.00000000e+00 5.22916438e-04 1.04622701e-02]\n",
            " [2.84655839e-02 2.67334562e-02 1.37768332e-02 6.37506414e-03\n",
            "  1.23751757e-04 0.00000000e+00 0.00000000e+00 1.64574041e-04\n",
            "  3.45137343e-03 1.14209689e-02 3.69120692e-03 7.30545144e-04\n",
            "  2.29013129e-03 6.74014445e-04 4.51166910e-04 1.70277280e-03\n",
            "  5.01147471e-04 0.00000000e+00 3.38168669e-04 7.72748701e-03]\n",
            " [3.82353493e-04 5.85589965e-04 9.98277783e-01 1.23751757e-04\n",
            "  3.81796553e-06 0.00000000e+00 0.00000000e+00 5.07740697e-06\n",
            "  4.08237684e-05 1.04556719e-04 4.09499153e-05 2.25386393e-05\n",
            "  7.06546925e-05 2.07945650e-05 1.39193153e-05 5.25336218e-05\n",
            "  1.54613062e-05 0.00000000e+00 1.04331157e-05 1.03493483e-04]\n",
            " [0.00000000e+00 0.00000000e+00 0.00000000e+00 0.00000000e+00\n",
            "  0.00000000e+00 0.00000000e+00 0.00000000e+00 0.00000000e+00\n",
            "  0.00000000e+00 0.00000000e+00 0.00000000e+00 1.00000000e+00\n",
            "  0.00000000e+00 0.00000000e+00 0.00000000e+00 0.00000000e+00\n",
            "  0.00000000e+00 0.00000000e+00 0.00000000e+00 0.00000000e+00]\n",
            " [0.00000000e+00 0.00000000e+00 0.00000000e+00 0.00000000e+00\n",
            "  0.00000000e+00 0.00000000e+00 0.00000000e+00 0.00000000e+00\n",
            "  0.00000000e+00 0.00000000e+00 0.00000000e+00 0.00000000e+00\n",
            "  0.00000000e+00 0.00000000e+00 0.00000000e+00 0.00000000e+00\n",
            "  1.00000000e+00 0.00000000e+00 0.00000000e+00 0.00000000e+00]\n",
            " [5.08481346e-04 7.78759946e-04 3.63053259e-04 1.64574056e-04\n",
            "  5.07740742e-06 0.00000000e+00 0.00000000e+00 6.75230331e-06\n",
            "  5.42904054e-05 1.39047101e-04 5.44581599e-05 2.99735129e-05\n",
            "  9.39617239e-05 9.97374296e-01 1.85109129e-05 6.98630101e-05\n",
            "  2.05615652e-05 0.00000000e+00 1.38747118e-05 1.37633120e-04]\n",
            " [1.92282572e-02 1.35643966e-02 7.56127387e-03 3.54997767e-03\n",
            "  4.08237684e-05 0.00000000e+00 0.00000000e+00 5.42903945e-05\n",
            "  2.44119810e-03 8.68397579e-03 2.66461517e-03 2.40995389e-04\n",
            "  7.55478395e-04 2.22346789e-04 1.48832885e-04 5.61718072e-04\n",
            "  1.65320700e-04 0.00000000e+00 1.11556546e-04 5.22586517e-03]\n",
            " [6.56667203e-02 4.26610932e-02 2.44004000e-02 1.15071088e-02\n",
            "  1.04556711e-04 0.00000000e+00 0.00000000e+00 1.39047072e-04\n",
            "  8.42649117e-03 3.04467790e-02 9.23954230e-03 6.17230777e-04\n",
            "  1.93491054e-03 5.69468481e-04 3.81186692e-04 1.43865682e-03\n",
            "  4.23414836e-04 0.00000000e+00 2.85715563e-04 1.78518351e-02]\n",
            " [2.09168755e-02 1.43921804e-02 8.08418635e-03 3.80056771e-03\n",
            "  4.09499189e-05 0.00000000e+00 0.00000000e+00 5.44581599e-05\n",
            "  2.66446522e-03 9.52498429e-03 2.91246921e-03 2.41740097e-04\n",
            "  7.57812930e-04 2.23033872e-04 1.49292799e-04 5.63453825e-04\n",
            "  1.65831560e-04 0.00000000e+00 1.11901274e-04 5.68528473e-03]\n",
            " [2.25715199e-03 3.45692015e-03 1.61159562e-03 7.30545260e-04\n",
            "  2.25386411e-05 0.00000000e+00 0.00000000e+00 2.99735148e-05\n",
            "  2.40995432e-04 6.17230893e-04 2.41740112e-04 1.33052614e-04\n",
            "  4.17096686e-04 1.22756799e-04 8.21700523e-05 3.10122326e-04\n",
            "  9.12729083e-05 0.00000000e+00 6.15899335e-05 6.10954303e-04]\n",
            " [7.07577588e-03 1.08368387e-02 5.05207013e-03 2.29013129e-03\n",
            "  7.06546925e-05 0.00000000e+00 0.00000000e+00 9.39617166e-05\n",
            "  7.55478453e-04 1.93491078e-03 7.57812872e-04 4.17096628e-04\n",
            "  1.30752497e-03 3.84821033e-04 2.57588719e-04 9.72179230e-04\n",
            "  2.86124588e-04 0.00000000e+00 1.93073647e-04 1.91523461e-03]\n",
            " [2.08248966e-03 3.18941777e-03 1.48688769e-03 6.74014445e-04\n",
            "  2.07945650e-05 0.00000000e+00 0.00000000e+00 2.76541141e-05\n",
            "  2.22346818e-04 5.69468539e-04 2.23033858e-04 1.22756785e-04\n",
            "  3.84821033e-04 1.13257673e-04 7.58116003e-05 2.86124559e-04\n",
            "  8.42100635e-05 0.00000000e+00 5.68239993e-05 5.63677633e-04]\n",
            " [1.39396195e-03 2.13490962e-03 9.95282200e-04 4.51166939e-04\n",
            "  1.39193153e-05 0.00000000e+00 0.00000000e+00 1.85109129e-05\n",
            "  1.48832914e-04 3.81186750e-04 1.49292799e-04 8.21700451e-05\n",
            "  2.57588748e-04 7.58116075e-05 5.07462209e-05 1.91523999e-04\n",
            "  5.63679241e-05 0.00000000e+00 3.80364399e-05 3.77310469e-04]\n",
            " [5.26102586e-03 8.05747509e-03 3.75634711e-03 1.70277292e-03\n",
            "  5.25336254e-05 0.00000000e+00 0.00000000e+00 6.98630101e-05\n",
            "  5.61718130e-04 1.43865705e-03 5.63453825e-04 3.10122326e-04\n",
            "  9.72179289e-04 2.86124588e-04 1.91523999e-04 7.22840836e-04\n",
            "  2.12741172e-04 0.00000000e+00 1.43555342e-04 1.42402737e-03]\n",
            " [1.54838606e-03 2.37141643e-03 1.10554020e-03 5.01147530e-04\n",
            "  1.54613081e-05 0.00000000e+00 0.00000000e+00 2.05615634e-05\n",
            "  1.65320715e-04 4.23414895e-04 1.65831560e-04 9.12729010e-05\n",
            "  2.86124588e-04 8.42100635e-05 5.63679241e-05 2.12741157e-04\n",
            "  6.26124020e-05 0.00000000e+00 4.22501435e-05 4.19109187e-04]\n",
            " [0.00000000e+00 0.00000000e+00 0.00000000e+00 0.00000000e+00\n",
            "  0.00000000e+00 0.00000000e+00 0.00000000e+00 0.00000000e+00\n",
            "  0.00000000e+00 0.00000000e+00 0.00000000e+00 0.00000000e+00\n",
            "  0.00000000e+00 0.00000000e+00 0.00000000e+00 0.00000000e+00\n",
            "  0.00000000e+00 0.00000000e+00 0.00000000e+00 0.00000000e+00]\n",
            " [1.04483345e-03 1.60020497e-03 7.46006088e-04 3.38168698e-04\n",
            "  1.04331157e-05 0.00000000e+00 0.00000000e+00 1.38747118e-05\n",
            "  1.11556561e-04 2.85715592e-04 1.11901267e-04 6.15899262e-05\n",
            "  1.93073662e-04 5.68240030e-05 3.80364399e-05 1.43555342e-04\n",
            "  4.22501435e-05 0.00000000e+00 2.85099213e-05 2.82810157e-04]\n",
            " [4.04330827e-02 3.03775892e-02 1.66198500e-02 7.77698308e-03\n",
            "  1.03493490e-04 0.00000000e+00 0.00000000e+00 1.37633135e-04\n",
            "  5.08801825e-03 1.78606622e-02 5.53247612e-03 6.10954245e-04\n",
            "  1.91523484e-03 5.63677691e-04 3.77310498e-04 1.42402749e-03\n",
            "  4.19109216e-04 0.00000000e+00 2.82810186e-04 1.09864408e-02]]\n",
            "\n",
            "\n",
            "outputs\n",
            "Timestep:  2\n",
            "\n",
            " Gate Controller vals:\n",
            "[[0.12939098 0.27711508 0.59349394]]\n",
            "\n",
            " Gate Controller vals:\n",
            "[[0.25420687 0.24831821 0.24754326 0.24993174]]\n",
            "\n",
            " Gate Controller vals:\n",
            "[[0.18637046 0.1967938  0.20819552 0.21380189 0.1948383 ]]\n",
            "\n",
            " Gate Controller vals:\n",
            "[[0.17232893 0.18061621 0.16397531 0.1665209  0.15801455 0.15854414]]\n",
            "\n",
            " Gate Controller vals:\n",
            "[[0.06529929 0.06529929 0.06529929 0.29951647 0.06529929 0.13985068\n",
            "  0.2994357 ]]\n",
            "\n",
            " Gate Controller vals:\n",
            "[[0.13615867 0.12032738 0.12240006 0.11635831 0.13839614 0.12412936\n",
            "  0.11904359 0.12318649]]\n",
            "[[0.12202393 0.11995302 0.12179861 0.11991245 0.12166571 0.13731566\n",
            "  0.12088069 0.13645001]]\n",
            "\n",
            " Gate Controller vals:\n",
            "[[0.11902973 0.11011365 0.11637647 0.10514344 0.11092419 0.10429602\n",
            "  0.11257858 0.10795508 0.1135828 ]]\n",
            "[[0.11032166 0.10244606 0.09622249 0.11185706 0.1281986  0.11442005\n",
            "  0.12097365 0.10919318 0.10636724]]\n",
            "\n",
            " Gate Controller vals:\n",
            "[[0.12589358 0.12589358 0.12589358 0.26962474 0.05878242 0.05878242\n",
            "  0.05878242 0.05878242 0.05878242 0.05878242]]\n",
            "\n",
            " Gate Controller vals:\n",
            "[[0.09062097 0.09442373 0.0992891  0.09502895 0.08676202 0.0847469\n",
            "  0.08897337 0.09708362 0.08572184 0.09022255 0.08712696]]\n",
            "[[0.08977523 0.09047807 0.09568485 0.09262362 0.08530968 0.09112865\n",
            "  0.09024172 0.09358754 0.09212839 0.08786284 0.09117937]]\n",
            "\n",
            " Gate Controller vals:\n",
            "[[0.07932984 0.07806862 0.08593918 0.08549682 0.08348114 0.08618236\n",
            "  0.08247791 0.08370069 0.080066   0.0834782  0.08728035 0.08449893]]\n",
            "[[0.07816845 0.08235306 0.09133852 0.08287953 0.08419119 0.08371843\n",
            "  0.08450982 0.08682126 0.08609722 0.08011732 0.08109812 0.07870711]]\n",
            "\n",
            " Gate Controller vals:\n",
            "[[0.07444879 0.0731226  0.07902487 0.08117409 0.08291099 0.0795453\n",
            "  0.07716925 0.07578839 0.07042095 0.07663095 0.07302739 0.07975107\n",
            "  0.07698536]]\n",
            "[[0.07939654 0.078536   0.07288313 0.07419266 0.07930298 0.07954035\n",
            "  0.0700159  0.07526474 0.07250531 0.0780029  0.08113655 0.08252279\n",
            "  0.07670018]]\n",
            "\n",
            " Gate Controller vals:\n",
            "[[0.07023057 0.07670838 0.07248078 0.06679799 0.0673591  0.08032241\n",
            "  0.07110048 0.07525383 0.07588927 0.06817672 0.06941406 0.06618232\n",
            "  0.06906564 0.07101839]]\n",
            "[[0.07188328 0.07189906 0.07529177 0.07421084 0.06791767 0.07016765\n",
            "  0.07392985 0.08006081 0.06692635 0.0687829  0.07344497 0.06995393\n",
            "  0.06819031 0.06734069]]\n",
            "\n",
            " Gate Controller vals:\n",
            "[[0.06663158 0.06612389 0.06686375 0.06873026 0.06353373 0.07085064\n",
            "  0.06258459 0.07051752 0.06839781 0.0618946  0.06987116 0.0683298\n",
            "  0.06954112 0.06327098 0.06285857]]\n",
            "[[0.06778747 0.06894968 0.06920704 0.06810285 0.06416582 0.07066458\n",
            "  0.06104062 0.05948531 0.06804444 0.06486322 0.06770072 0.06739264\n",
            "  0.06983315 0.0673513  0.06541119]]\n",
            "\n",
            " Gate Controller vals:\n",
            "[[0.03736854 0.03736854 0.03736854 0.17140299 0.03736854 0.03736854\n",
            "  0.03736854 0.17140299 0.03736854 0.03736854 0.17140299 0.03736854\n",
            "  0.03736854 0.03736854 0.03736854 0.03736854]]\n",
            "[[0.03583951 0.03583951 0.03583951 0.16438961 0.03583951 0.03583951\n",
            "  0.07675704 0.16438961 0.03583951 0.03583951 0.16438961 0.03583951\n",
            "  0.03583951 0.03583951 0.03583951 0.03583951]]\n",
            "\n",
            "Register controller  0\n",
            "[[0.02987847 0.02987847 0.02987847 0.06399035 0.02987847 0.06399035\n",
            "  0.13704735 0.13704735 0.02987847 0.02987847 0.13704735 0.02987847\n",
            "  0.06399035 0.06399035 0.02987847 0.02987847 0.06399035]]\n",
            "\n",
            "Register controller  1\n",
            "[[0.03040177 0.03040177 0.03040177 0.06511109 0.03040177 0.06511109\n",
            "  0.13944763 0.13944763 0.03040177 0.03040177 0.12193342 0.03040177\n",
            "  0.06511109 0.06511109 0.03040177 0.03040177 0.06511109]]\n",
            "\n",
            "Register controller  2\n",
            "[[0.0315548  0.0315548  0.0315548  0.06758051 0.0315548  0.06758051\n",
            "  0.14473636 0.14473636 0.0315548  0.0315548  0.0886315  0.0315548\n",
            "  0.06758051 0.06758051 0.0315548  0.0315548  0.06758051]]\n",
            "\n",
            "Output registers\n",
            "[[[1.53570876e-01 3.54224950e-01 2.34802246e-01 5.85981458e-02\n",
            "   2.56655202e-03 6.59795041e-05 1.30262240e-04 1.48435892e-03\n",
            "   8.61402228e-03 1.41127184e-02 9.05876327e-03 1.17546720e-02\n",
            "   1.98370386e-02 1.18749626e-02 8.59627780e-03 1.51534127e-02\n",
            "   8.74887407e-03 5.34856226e-04 3.55313509e-03 3.53090428e-02]\n",
            "  [1.52334526e-01 3.56659859e-01 2.38104582e-01 5.95947579e-02\n",
            "   2.61104386e-03 6.71350863e-05 1.17130156e-04 1.36171933e-03\n",
            "   8.01732112e-03 1.41546521e-02 9.14084632e-03 1.08031295e-02\n",
            "   2.00322438e-02 1.20376293e-02 7.85573851e-03 1.53099913e-02\n",
            "   8.90024565e-03 5.12551924e-04 3.25041963e-03 3.24120186e-02]\n",
            "  [1.49610326e-01 3.62024933e-01 2.45380893e-01 6.17906563e-02\n",
            "   2.70907744e-03 6.96812713e-05 8.81950255e-05 1.09149644e-03\n",
            "   6.70255255e-03 1.42470496e-02 9.32170451e-03 8.70650634e-03\n",
            "   2.04623528e-02 1.23960441e-02 6.22403994e-03 1.56549904e-02\n",
            "   9.23377462e-03 4.63406759e-04 2.58341827e-03 2.60287412e-02]]]\n",
            "\n",
            "Output memory\n",
            "[[1.39985949e-01 1.37159944e-01 7.98631012e-02 3.35517749e-02\n",
            "  8.41332716e-04 1.34776192e-05 2.66086408e-05 7.24863959e-04\n",
            "  1.53273437e-02 4.76363063e-01 1.65560767e-02 4.27285023e-03\n",
            "  1.31274806e-02 4.17968770e-03 2.91189225e-03 9.94070899e-03\n",
            "  3.07111815e-03 1.09254972e-04 1.59221841e-03 3.67201380e-02]\n",
            " [1.21160559e-01 1.61291048e-01 9.31078494e-02 3.71409953e-02\n",
            "  1.26782083e-03 2.18094501e-05 4.30580367e-05 1.04846817e-03\n",
            "  1.10917073e-02 3.24253663e-02 1.16690164e-02 6.36163820e-03\n",
            "  4.24882323e-01 6.25364948e-03 4.37069638e-03 1.47978356e-02\n",
            "  4.59054345e-03 1.76796122e-04 2.32068961e-03 3.07661686e-02]\n",
            " [5.93641475e-02 7.59561956e-02 4.47048396e-02 1.77157950e-02\n",
            "  5.96969854e-04 1.10637075e-05 2.18428941e-05 4.70460218e-04\n",
            "  5.46732172e-03 1.64464805e-02 5.79570327e-03 2.95456313e-03\n",
            "  9.04272962e-03 2.92087835e-03 2.04883935e-03 7.21811950e-01\n",
            "  2.14171363e-03 8.96868296e-05 1.05106179e-03 1.50426636e-02]\n",
            " [3.87109220e-02 4.46511358e-02 2.56852061e-02 1.05338190e-02\n",
            "  3.14492034e-04 5.07504683e-06 1.00195812e-05 2.69871700e-04\n",
            "  3.92358983e-03 1.23598408e-02 4.18423070e-03 1.59529212e-03\n",
            "  4.90036234e-03 1.56126893e-03 1.08804426e-03 3.71135864e-03\n",
            "  1.14706648e-03 4.11403598e-05 5.93229954e-04 1.00179678e-02]\n",
            " [9.10509145e-04 1.44034962e-03 9.95839655e-01 3.22691398e-04\n",
            "  1.24418366e-05 2.19880974e-07 4.34107363e-07 1.01181122e-05\n",
            "  7.15167407e-05 1.79190480e-04 7.32636690e-05 6.21290383e-05\n",
            "  1.90405117e-04 6.11957294e-05 4.28246130e-05 1.44509162e-04\n",
            "  4.49036888e-05 1.78244306e-06 2.24672876e-05 2.25607364e-04]\n",
            " [1.34776192e-05 2.18094501e-05 1.41900000e-05 5.07504728e-06\n",
            "  2.19880988e-07 5.59882052e-09 1.10536584e-08 1.28739202e-07\n",
            "  7.84653139e-07 1.90838500e-06 8.25932489e-07 9.99924600e-01\n",
            "  3.05459821e-06 1.03032198e-06 7.37077983e-07 2.34598338e-06\n",
            "  7.50871493e-07 4.53862761e-08 3.07222734e-07 3.11728809e-06]\n",
            " [2.66086408e-05 4.30580367e-05 2.80150816e-05 1.00195812e-05\n",
            "  4.34107363e-07 1.10536575e-08 2.18230536e-08 2.54167674e-07\n",
            "  1.54912766e-06 3.76769253e-06 1.63062475e-06 1.99365263e-06\n",
            "  6.03064245e-06 2.03414743e-06 1.45520073e-06 4.63163587e-06\n",
            "  9.99850631e-01 8.96053720e-08 6.06544745e-07 6.15440968e-06]\n",
            " [8.17492255e-04 1.27887807e-03 6.88700122e-04 2.80980486e-04\n",
            "  1.01244277e-05 1.28739217e-07 2.54167702e-07 9.70066867e-06\n",
            "  7.22373297e-05 1.82684191e-04 7.33539637e-05 5.31404330e-05\n",
            "  1.64034049e-04 9.95645702e-01 3.54267540e-05 1.23683785e-04\n",
            "  3.77909746e-05 1.04361152e-06 2.09146128e-05 2.09070218e-04]\n",
            " [2.09112000e-02 1.64756626e-02 9.46898293e-03 4.22321353e-03\n",
            "  7.12020992e-05 7.84653082e-07 1.54912755e-06 7.17513612e-05\n",
            "  2.52502435e-03 8.85844138e-03 2.75183422e-03 3.79935926e-04\n",
            "  1.17547903e-03 3.64361564e-04 2.50537822e-04 8.84483801e-04\n",
            "  2.68782198e-04 6.36071127e-06 1.53418106e-04 5.60678355e-03]\n",
            " [6.85504749e-02 4.89839315e-02 2.86016725e-02 1.29372841e-02\n",
            "  1.76781235e-04 1.90838477e-06 3.76769253e-06 1.79307244e-04\n",
            "  8.47449340e-03 3.03043369e-02 9.28044040e-03 9.45355161e-04\n",
            "  2.92569445e-03 9.05828085e-04 6.22496009e-04 2.20082956e-03\n",
            "  6.68326044e-04 1.54701302e-05 3.82993050e-04 1.84494630e-02]\n",
            " [2.26693228e-02 1.74472723e-02 1.00863632e-02 4.50639706e-03\n",
            "  7.29249950e-05 8.25932545e-07 1.63062498e-06 7.28358355e-05\n",
            "  2.75018485e-03 9.69914906e-03 3.00148292e-03 3.87981709e-04\n",
            "  1.19988283e-03 3.72512091e-04 2.56343104e-04 9.03180218e-04\n",
            "  2.74730264e-04 6.69533938e-06 1.55961214e-04 6.08106423e-03]\n",
            " [4.65688948e-03 7.34286383e-03 4.14871471e-03 1.63582037e-03\n",
            "  6.18860795e-05 1.00981083e-06 1.99365263e-06 5.27800439e-05\n",
            "  3.79195495e-04 9.52924078e-04 3.87375127e-04 3.13349941e-04\n",
            "  9.62280377e-04 3.06895410e-04 2.13978085e-04 7.28973420e-04\n",
            "  2.25443320e-04 8.18592980e-06 1.16152311e-04 1.16477336e-03]\n",
            " [1.41339283e-02 2.22838838e-02 1.25832409e-02 4.96350648e-03\n",
            "  1.87671889e-04 3.05459844e-06 6.03064291e-06 1.60282332e-04\n",
            "  1.15207664e-03 2.89542903e-03 1.17683492e-03 9.50641523e-04\n",
            "  2.91954540e-03 9.30901850e-04 6.48985384e-04 2.21157307e-03\n",
            "  6.83857186e-04 2.47617954e-05 3.52639647e-04 3.53612518e-03]\n",
            " [4.53342590e-03 7.15805404e-03 4.07729391e-03 1.59847224e-03\n",
            "  6.09657473e-05 1.03032198e-06 2.03414766e-06 5.09564998e-05\n",
            "  3.63616215e-04 9.12651827e-04 3.71890026e-04 3.06861039e-04\n",
            "  9.41532198e-04 3.01270047e-04 2.10386293e-04 7.13820686e-04\n",
            "  2.21205031e-04 8.35220180e-06 1.12561647e-04 1.12941046e-03]\n",
            " [3.15425312e-03 4.98462329e-03 2.85336748e-03 1.11475261e-03\n",
            "  4.27263731e-05 7.37077926e-07 1.45520073e-06 3.52730785e-05\n",
            "  2.50634504e-04 6.28589129e-04 2.56524130e-04 2.14283908e-04\n",
            "  6.57131954e-04 2.10689701e-04 1.47271203e-04 4.98443376e-04\n",
            "  1.54652254e-04 5.97504823e-06 7.80993505e-05 7.83902477e-04]\n",
            " [1.07399048e-02 1.69379786e-02 9.58189555e-03 3.77477426e-03\n",
            "  1.42984951e-04 2.34598315e-06 4.63163587e-06 1.21569872e-04\n",
            "  8.72515491e-04 2.19223858e-03 8.91492178e-04 7.23318546e-04\n",
            "  2.22097430e-03 7.08683510e-04 4.94238280e-04 1.68269873e-03\n",
            "  5.20555885e-04 1.90174778e-05 2.67690048e-04 2.68462440e-03]\n",
            " [3.34003661e-03 5.27203595e-03 2.99726753e-03 1.17663946e-03\n",
            "  4.47916573e-05 7.50871493e-07 1.48243316e-06 3.76164026e-05\n",
            "  2.68858450e-04 6.75014453e-04 2.74900143e-04 2.25765834e-04\n",
            "  6.92852424e-04 2.21526192e-04 1.54641661e-04 5.25186770e-04\n",
            "  1.62672033e-04 6.08686469e-06 8.30196077e-05 8.32881778e-04]\n",
            " [1.09254972e-04 1.76796122e-04 1.15029812e-04 4.11403598e-05\n",
            "  1.78244318e-06 4.53862761e-08 8.96053720e-08 1.04361152e-06\n",
            "  6.36071218e-06 1.54701320e-05 6.69533893e-06 8.18593071e-06\n",
            "  2.47617954e-05 8.35220180e-06 5.97504868e-06 1.90174796e-05\n",
            "  6.08686423e-06 3.67919313e-07 2.49047048e-06 2.52699829e-05]\n",
            " [1.78000727e-03 2.79024104e-03 1.52152288e-03 6.15232741e-04\n",
            "  2.24548530e-05 3.07222734e-07 6.06544745e-07 2.08808178e-05\n",
            "  1.54144946e-04 3.89236055e-04 1.56753318e-04 1.16742820e-04\n",
            "  3.59878468e-04 1.13122383e-04 7.83224823e-05 2.71684083e-04\n",
            "  8.32754013e-05 2.49047025e-06 4.52485547e-05 4.52678680e-04]\n",
            " [4.62170579e-02 4.12282906e-02 2.38134861e-02 1.02718119e-02\n",
            "  2.21515249e-04 3.11728763e-06 6.15440922e-06 2.03457050e-04\n",
            "  5.30844834e-03 1.81634054e-02 5.75698214e-03 1.14720222e-03\n",
            "  3.53448652e-03 1.11335702e-03 7.71646737e-04 2.66963569e-03\n",
            "  8.19347217e-04 2.52699792e-05 4.41833545e-04 1.22547047e-02]]\n",
            "\n",
            "\n",
            "outputs\n",
            "Timestep:  3\n",
            "\n",
            " Gate Controller vals:\n",
            "[[0.12939098 0.27711508 0.59349394]]\n",
            "\n",
            " Gate Controller vals:\n",
            "[[0.25403523 0.24839364 0.2476711  0.24990003]]\n",
            "\n",
            " Gate Controller vals:\n",
            "[[0.18678495 0.1968984  0.20803162 0.21338739 0.19489764]]\n",
            "\n",
            " Gate Controller vals:\n",
            "[[0.1721055  0.18020655 0.16414687 0.16656178 0.15825595 0.15872338]]\n",
            "\n",
            " Gate Controller vals:\n",
            "[[0.06530103 0.06530103 0.06530103 0.2995245  0.06530103 0.13985442\n",
            "  0.29941693]]\n",
            "\n",
            " Gate Controller vals:\n",
            "[[0.13581176 0.1205367  0.12248702 0.11661132 0.13793746 0.12415256\n",
            "  0.11922055 0.12324262]]\n",
            "[[0.12207459 0.12012078 0.12186955 0.12005895 0.12182717 0.13700472\n",
            "  0.12097277 0.13607149]]\n",
            "\n",
            " Gate Controller vals:\n",
            "[[0.118842   0.11015778 0.11623632 0.10529186 0.11094214 0.10447118\n",
            "  0.11251938 0.10803022 0.11350909]]\n",
            "[[0.11035254 0.10269983 0.09664281 0.11179464 0.12784666 0.11431535\n",
            "  0.12064008 0.10921545 0.10649264]]\n",
            "\n",
            " Gate Controller vals:\n",
            "[[0.12589358 0.12589358 0.12589358 0.26962474 0.05878242 0.05878242\n",
            "  0.05878242 0.05878242 0.05878242 0.05878242]]\n",
            "\n",
            " Gate Controller vals:\n",
            "[[0.09063969 0.09430636 0.09906258 0.09487855 0.08693893 0.0848995\n",
            "  0.08903015 0.09688474 0.08587881 0.09024043 0.08724032]]\n",
            "[[0.08981574 0.09045229 0.09552354 0.09259911 0.08548027 0.09113608\n",
            "  0.09026273 0.09352567 0.09209178 0.08796607 0.09114677]]\n",
            "\n",
            " Gate Controller vals:\n",
            "[[0.07941807 0.07819288 0.08586159 0.085449   0.08350078 0.0860785\n",
            "  0.08249482 0.08368468 0.08020455 0.08348487 0.08714644 0.08448378]]\n",
            "[[0.07829133 0.08237572 0.09113434 0.08286893 0.08417078 0.0836804\n",
            "  0.08448671 0.08670388 0.08606214 0.08020292 0.08118793 0.07883495]]\n",
            "\n",
            " Gate Controller vals:\n",
            "[[0.07453006 0.07323083 0.0789937  0.08101042 0.08271264 0.0794453\n",
            "  0.07716263 0.07580514 0.07061874 0.07664195 0.07318369 0.07966363\n",
            "  0.07700119]]\n",
            "[[0.07933395 0.07848749 0.07299035 0.07426198 0.07926316 0.07945948\n",
            "  0.07017661 0.07533909 0.07257145 0.07796864 0.08104146 0.08238729\n",
            "  0.07671909]]\n",
            "\n",
            " Gate Controller vals:\n",
            "[[0.07026441 0.07651974 0.07247678 0.06690972 0.06747409 0.08004715\n",
            "  0.07112396 0.0751588  0.07580322 0.06825884 0.06944185 0.06632388\n",
            "  0.06912888 0.07106867]]\n",
            "[[0.07187039 0.07188242 0.07516694 0.07413784 0.06804994 0.0702156\n",
            "  0.07386471 0.0798027  0.06699345 0.06884994 0.07337803 0.06998653\n",
            "  0.06830973 0.06749178]]\n",
            "\n",
            " Gate Controller vals:\n",
            "[[0.06663919 0.06613631 0.06689504 0.06865849 0.06361579 0.07072661\n",
            "  0.06269444 0.07039723 0.0683565  0.0619959  0.06979219 0.06827888\n",
            "  0.06945805 0.06335411 0.06300128]]\n",
            "[[0.06774348 0.06890978 0.0691466  0.06805427 0.06423839 0.07058942\n",
            "  0.06119312 0.0596633  0.06800751 0.0649103  0.06768348 0.06737205\n",
            "  0.06976621 0.06730152 0.06542054]]\n",
            "\n",
            " Gate Controller vals:\n",
            "[[0.03736854 0.03736854 0.03736854 0.17140299 0.03736854 0.03736854\n",
            "  0.03736854 0.17140299 0.03736854 0.03736854 0.17140299 0.03736854\n",
            "  0.03736854 0.03736854 0.03736854 0.03736854]]\n",
            "[[0.03583951 0.03583951 0.03583951 0.16438961 0.03583951 0.03583951\n",
            "  0.07675704 0.16438961 0.03583951 0.03583951 0.16438961 0.03583951\n",
            "  0.03583951 0.03583951 0.03583951 0.03583951]]\n",
            "\n",
            "Register controller  0\n",
            "[[0.02987847 0.02987847 0.02987847 0.06399035 0.02987847 0.06399035\n",
            "  0.13704735 0.13704735 0.02987847 0.02987847 0.13704734 0.02987847\n",
            "  0.06399035 0.06399035 0.02987847 0.02987847 0.06399035]]\n",
            "\n",
            "Register controller  1\n",
            "[[0.03044314 0.03044314 0.03044314 0.0651997  0.03044314 0.0651997\n",
            "  0.1396374  0.1396374  0.03044314 0.03044314 0.12073843 0.03044314\n",
            "  0.0651997  0.0651997  0.03044314 0.03044314 0.0651997 ]]\n",
            "\n",
            "Register controller  2\n",
            "[[0.03155391 0.03155391 0.03155391 0.06757862 0.03155391 0.06757862\n",
            "  0.1447323  0.1447323  0.03155391 0.03155391 0.08865717 0.03155391\n",
            "  0.06757862 0.06757862 0.03155391 0.03155391 0.06757862]]\n",
            "\n",
            "Output registers\n",
            "[[[1.5357432e-01 3.5921389e-01 2.4268739e-01 6.1983798e-02 3.1312597e-03\n",
            "   1.2509394e-04 1.1421436e-04 9.4605889e-04 5.8346968e-03 9.7903367e-03\n",
            "   6.6995220e-03 9.3939481e-03 1.5549634e-02 9.6731195e-03 9.6414294e-03\n",
            "   1.6578164e-02 9.6498327e-03 6.7582040e-04 3.0499639e-03 3.4185458e-02]\n",
            "  [1.5215483e-01 3.6148658e-01 2.4631734e-01 6.3111477e-02 3.1890613e-03\n",
            "   1.2650850e-04 1.0374504e-04 8.6598803e-04 5.3833318e-03 9.8324437e-03\n",
            "   6.7269355e-03 8.6217336e-03 1.5674928e-02 9.7842524e-03 8.7842904e-03\n",
            "   1.6748996e-02 9.8251607e-03 6.5897417e-04 2.7727326e-03 3.1080306e-02]\n",
            "  [1.4936261e-01 3.6595723e-01 2.5345784e-01 6.5329760e-02 3.3027630e-03\n",
            "   1.2929112e-04 8.3150830e-05 7.0848083e-04 4.4954522e-03 9.9152727e-03\n",
            "   6.7808609e-03 7.1027102e-03 1.5921399e-02 1.0002864e-02 7.0982114e-03\n",
            "   1.7085042e-02 1.0170053e-02 6.2583585e-04 2.2273897e-03 2.4972161e-02]]]\n",
            "\n",
            "Output memory\n",
            "[[1.46914229e-01 1.66110650e-01 1.01439223e-01 4.04423662e-02\n",
            "  1.34054339e-03 3.68844085e-05 4.55426489e-05 7.94652791e-04\n",
            "  1.38141103e-02 3.91588598e-01 1.50009906e-02 5.47118066e-03\n",
            "  1.62583031e-02 5.46879135e-03 4.39378712e-03 1.42390933e-02\n",
            "  4.52880375e-03 2.29501005e-04 1.94536173e-03 3.72768752e-02]\n",
            " [1.38390347e-01 2.01814681e-01 1.24760032e-01 4.74456474e-02\n",
            "  1.96340960e-03 5.79955049e-05 6.93564798e-05 1.06488599e-03\n",
            "  9.88422334e-03 2.80500855e-02 1.05917165e-02 7.70528149e-03\n",
            "  3.03995073e-01 7.75205810e-03 6.37650955e-03 2.04243176e-02\n",
            "  6.53773034e-03 3.55117838e-04 2.68208492e-03 3.35104875e-02]\n",
            " [7.85161033e-02 1.11267060e-01 6.93251044e-02 2.63259821e-02\n",
            "  1.07430120e-03 3.19776773e-05 3.92167385e-05 5.72915247e-04\n",
            "  5.73992776e-03 1.67204011e-02 6.17765682e-03 4.21906123e-03\n",
            "  1.24503234e-02 4.25531622e-03 3.48811643e-03 6.10110044e-01\n",
            "  3.57052754e-03 1.98289257e-04 1.44846737e-03 1.90470368e-02]\n",
            " [4.89512123e-02 6.30313531e-02 3.82934399e-02 1.49823204e-02\n",
            "  5.52568934e-04 1.50346950e-05 1.87988808e-05 3.31304152e-04\n",
            "  4.16160282e-03 1.27861639e-02 4.47700638e-03 2.27108132e-03\n",
            "  6.75687799e-03 2.26891134e-03 1.81395770e-03 5.88626461e-03\n",
            "  1.87071692e-03 9.41470062e-05 8.07878328e-04 1.22111989e-02]\n",
            " [1.56201015e-03 2.51879636e-03 9.92889464e-01 5.82792447e-04\n",
            "  2.54599072e-05 7.35759158e-07 9.08429115e-07 1.41482669e-05\n",
            "  9.76337033e-05 2.45460862e-04 1.03074155e-04 1.01575381e-04\n",
            "  3.00609681e-04 1.02112535e-04 8.29761993e-05 2.66413612e-04\n",
            "  8.51522927e-05 4.57791066e-06 3.53057258e-05 3.71404574e-04]\n",
            " [3.93758637e-05 6.46698027e-05 4.28692438e-05 1.54085810e-05\n",
            "  7.36538595e-07 2.60334314e-08 2.98751601e-08 2.89592691e-07\n",
            "  1.82792178e-06 4.55530153e-06 2.01553098e-06 9.99780178e-01\n",
            "  7.44061026e-06 2.65739163e-06 2.33124365e-06 7.18813817e-06\n",
            "  2.34917616e-06 1.56202461e-07 8.18226340e-07 8.91614400e-06]\n",
            " [5.04618038e-05 8.25339375e-05 5.44297363e-05 1.95371558e-05\n",
            "  9.09972925e-07 2.98752880e-08 3.91581274e-08 4.02313276e-07\n",
            "  2.50998164e-06 6.20552237e-06 2.72625812e-06 3.43877628e-06\n",
            "  1.00702073e-05 3.53270980e-06 2.92349341e-06 9.09144728e-06\n",
            "  9.99717653e-01 1.91672456e-07 1.07719097e-06 1.14953391e-05]\n",
            " [1.02044514e-03 1.61483989e-03 9.13694501e-04 3.62011342e-04\n",
            "  1.41802902e-05 2.89463571e-07 4.02055434e-07 1.09559442e-05\n",
            "  8.03680668e-05 2.03313582e-04 8.26354080e-05 6.54324540e-05\n",
            "  1.98376743e-04 9.94514406e-01 4.79365226e-05 1.61662887e-04\n",
            "  5.03306219e-05 1.91483423e-06 2.49136683e-05 2.54483428e-04]\n",
            " [2.20777523e-02 1.85413789e-02 1.08628105e-02 4.71938122e-03\n",
            "  9.70523542e-05 1.82218832e-06 2.49864433e-06 7.94303560e-05\n",
            "  2.55947816e-03 8.92757904e-03 2.79206922e-03 4.57223738e-04\n",
            "  1.39068021e-03 4.44727892e-04 3.30070296e-04 1.12513802e-03\n",
            "  3.48390109e-04 1.19716487e-05 1.78367583e-04 5.86109515e-03]\n",
            " [7.06093609e-02 5.36102206e-02 3.17762084e-02 1.40313506e-02\n",
            "  2.40370224e-04 4.51949427e-06 6.13478642e-06 1.96756053e-04\n",
            "  8.44940916e-03 3.00742313e-02 9.25907213e-03 1.13075273e-03\n",
            "  3.43859242e-03 1.09948916e-03 8.17246619e-04 2.78650899e-03\n",
            "  8.62746499e-04 2.95335376e-05 4.41952026e-04 1.88518446e-02]\n",
            " [2.39845458e-02 1.97944529e-02 1.16704302e-02 5.06974664e-03\n",
            "  1.02386148e-04 2.00863292e-06 2.71262047e-06 8.15826061e-05\n",
            "  2.78755743e-03 9.77083948e-03 3.04523949e-03 4.76040907e-04\n",
            "  1.44505885e-03 4.64080716e-04 3.46980814e-04 1.17742759e-03\n",
            "  3.65452812e-04 1.30903754e-05 1.84388307e-04 6.36701332e-03]\n",
            " [6.59352075e-03 1.05519705e-02 6.30459748e-03 2.41103442e-03\n",
            "  1.00868048e-04 2.56769044e-06 3.41669784e-06 6.45415275e-05\n",
            "  4.55064612e-04 1.14551606e-03 4.74389439e-04 4.30333486e-04\n",
            "  1.28832227e-03 4.28406696e-04 3.34000913e-04 1.09264883e-03\n",
            "  3.45655630e-04 1.66040518e-05 1.54092850e-04 1.59703603e-03]\n",
            " [1.92505512e-02 3.07863206e-02 1.83454137e-02 7.02623650e-03\n",
            "  2.92698969e-04 7.34534933e-06 9.88206375e-06 1.89791477e-04\n",
            "  1.33994117e-03 3.37292766e-03 1.39533461e-03 1.25768362e-03\n",
            "  3.76962102e-03 1.25101674e-03 9.70861409e-04 3.18171969e-03\n",
            "  1.00552943e-03 4.77740323e-05 4.51286993e-04 4.67015011e-03]\n",
            " [6.54309988e-03 1.04879672e-02 6.31372165e-03 2.40278430e-03\n",
            "  1.01399797e-04 2.64556661e-06 3.50934033e-06 6.31740695e-05\n",
            "  4.42471734e-04 1.11283315e-03 4.62301716e-04 4.28245519e-04\n",
            "  1.27987086e-03 4.27339663e-04 3.34888784e-04 1.09111937e-03\n",
            "  3.45911511e-04 1.70796557e-05 1.51946791e-04 1.57806720e-03]\n",
            " [5.13894204e-03 8.27186555e-03 5.05845156e-03 1.90828810e-03\n",
            "  8.25494426e-05 2.32295747e-06 2.90711750e-06 4.74218687e-05\n",
            "  3.29180039e-04 8.27952172e-04 3.46418994e-04 3.34264943e-04\n",
            "  9.91858891e-04 3.35238234e-04 2.69971613e-04 8.70555406e-04\n",
            "  2.77591636e-04 1.45528657e-05 1.17079500e-04 1.22740970e-03]\n",
            " [1.65058766e-02 2.65092887e-02 1.60469487e-02 6.09307457e-03\n",
            "  2.60476896e-04 7.10733275e-06 8.93186188e-06 1.55486006e-04\n",
            "  1.08958140e-03 2.74370401e-03 1.14257832e-03 1.07014528e-03\n",
            "  3.18359165e-03 1.06976577e-03 8.54930840e-04 2.77200062e-03\n",
            "  8.81319400e-04 4.46208796e-05 3.79531702e-04 3.96600179e-03]\n",
            " [5.32775978e-03 8.56453460e-03 5.20643592e-03 1.97152956e-03\n",
            "  8.46945841e-05 2.34071285e-06 2.93781022e-06 4.97699948e-05\n",
            "  3.47399910e-04 8.74365272e-04 3.64818727e-04 3.45927430e-04\n",
            "  1.02804089e-03 3.46274523e-04 2.77576619e-04 8.97959922e-04\n",
            "  2.85839196e-04 1.46856855e-05 1.22044825e-04 1.27698178e-03]\n",
            " [2.49627221e-04 4.09108266e-04 2.70479650e-04 9.71513218e-05\n",
            "  4.58307841e-06 1.56173172e-07 1.91613566e-07 1.91520758e-06\n",
            "  1.20140194e-05 2.98138184e-05 1.31419729e-05 1.66893296e-05\n",
            "  4.85302771e-05 1.71701249e-05 1.46161801e-05 4.52636450e-05\n",
            "  1.47503633e-05 9.68628342e-07 5.25995256e-06 5.67000679e-05]\n",
            " [2.42122123e-03 3.85198556e-03 2.23323004e-03 8.71426077e-04\n",
            "  3.52942443e-05 8.17149783e-07 1.07505844e-06 2.48276810e-05\n",
            "  1.79675146e-04 4.54016146e-04 1.85933575e-04 1.55559130e-04\n",
            "  4.68261162e-04 1.53402914e-04 1.17906107e-04 3.91793321e-04\n",
            "  1.22944504e-04 5.25283212e-06 5.78636646e-05 5.96055179e-04]\n",
            " [5.16526885e-02 5.16845509e-02 3.09664123e-02 1.27790412e-02\n",
            "  3.58968973e-04 8.78746687e-06 1.12412226e-05 2.40682435e-04\n",
            "  5.38464729e-03 1.81622356e-02 5.85612888e-03 1.54499826e-03\n",
            "  4.63289255e-03 1.52902084e-03 1.19211595e-03 3.93328536e-03\n",
            "  1.23901549e-03 5.56732666e-05 5.68559801e-04 1.33928303e-02]]\n",
            "\n",
            "\n",
            "outputs\n",
            "Timestep:  4\n",
            "\n",
            " Gate Controller vals:\n",
            "[[0.12939098 0.27711508 0.59349394]]\n",
            "\n",
            " Gate Controller vals:\n",
            "[[0.25403345 0.24839449 0.24767226 0.24989977]]\n",
            "\n",
            " Gate Controller vals:\n",
            "[[0.18678878 0.19689952 0.20803    0.21338339 0.19489828]]\n",
            "\n",
            " Gate Controller vals:\n",
            "[[0.17210346 0.18020266 0.16414852 0.16656213 0.1582581  0.1587251 ]]\n",
            "\n",
            " Gate Controller vals:\n",
            "[[0.06530105 0.06530105 0.06530105 0.29952458 0.06530105 0.13985446\n",
            "  0.29941675]]\n",
            "\n",
            " Gate Controller vals:\n",
            "[[0.13580835 0.12053856 0.12248784 0.11661372 0.13793316 0.12415297\n",
            "  0.11922225 0.12324315]]\n",
            "[[0.1220751  0.12012243 0.12187024 0.12006038 0.12182867 0.1370017\n",
            "  0.12097368 0.13606784]]\n",
            "\n",
            " Gate Controller vals:\n",
            "[[0.11884018 0.11015812 0.11623492 0.10529339 0.11094232 0.10447279\n",
            "  0.11251887 0.10803093 0.11350845]]\n",
            "[[0.11035282 0.10270229 0.09664688 0.11179405 0.12784332 0.1143142\n",
            "  0.12063697 0.10921576 0.10649372]]\n",
            "\n",
            " Gate Controller vals:\n",
            "[[0.12589358 0.12589358 0.12589358 0.26962474 0.05878242 0.05878242\n",
            "  0.05878242 0.05878242 0.05878242 0.05878242]]\n",
            "\n",
            " Gate Controller vals:\n",
            "[[0.09063999 0.0943051  0.09906048 0.09487706 0.08694056 0.08490095\n",
            "  0.08903072 0.09688284 0.08588025 0.09024055 0.08724149]]\n",
            "[[0.08981617 0.090452   0.09552204 0.09259898 0.08548187 0.09113622\n",
            "  0.09026286 0.09352496 0.09209139 0.08796702 0.09114645]]\n",
            "\n",
            " Gate Controller vals:\n",
            "[[0.07941893 0.07819408 0.08586086 0.08544841 0.08350091 0.08607759\n",
            "  0.08249506 0.08368462 0.08020592 0.08348492 0.08714519 0.08448356]]\n",
            "[[0.07829259 0.08237593 0.09113239 0.08286878 0.08417056 0.08368005\n",
            "  0.08448642 0.08670282 0.08606175 0.08020381 0.08118873 0.0788362 ]]\n",
            "\n",
            " Gate Controller vals:\n",
            "[[0.07453088 0.07323189 0.07899338 0.08100889 0.08271071 0.07944432\n",
            "  0.07716249 0.0758053  0.07062066 0.07664209 0.07318527 0.07966274\n",
            "  0.07700129]]\n",
            "[[0.07933332 0.07848702 0.07299142 0.07426263 0.07926279 0.0794588\n",
            "  0.07017818 0.07533978 0.07257207 0.07796831 0.08104055 0.08238601\n",
            "  0.0767192 ]]\n",
            "\n",
            " Gate Controller vals:\n",
            "[[0.07026477 0.07651787 0.07247674 0.06691073 0.06747521 0.08004453\n",
            "  0.07112418 0.07515787 0.0758024  0.06825967 0.06944218 0.06632528\n",
            "  0.06912949 0.07106907]]\n",
            "[[0.0718703  0.07188234 0.0751657  0.0741371  0.06805118 0.07021603\n",
            "  0.07386403 0.0798003  0.06699413 0.0688506  0.07337739 0.06998686\n",
            "  0.06831086 0.06749324]]\n",
            "\n",
            " Gate Controller vals:\n",
            "[[0.06663924 0.06613638 0.06689534 0.06865785 0.06361661 0.07072544\n",
            "  0.0626955  0.07039607 0.06835606 0.06199686 0.06979146 0.06827842\n",
            "  0.06945724 0.06335483 0.06300267]]\n",
            "[[0.06774311 0.06890943 0.06914601 0.06805382 0.06423908 0.07058871\n",
            "  0.06119458 0.05966502 0.0680072  0.06491076 0.06768331 0.06737182\n",
            "  0.06976554 0.06730108 0.06542055]]\n",
            "\n",
            " Gate Controller vals:\n",
            "[[0.03736854 0.03736854 0.03736854 0.17140299 0.03736854 0.03736854\n",
            "  0.03736854 0.17140299 0.03736854 0.03736854 0.17140299 0.03736854\n",
            "  0.03736854 0.03736854 0.03736854 0.03736854]]\n",
            "[[0.03583951 0.03583951 0.03583951 0.16438961 0.03583951 0.03583951\n",
            "  0.07675704 0.16438961 0.03583951 0.03583951 0.16438961 0.03583951\n",
            "  0.03583951 0.03583951 0.03583951 0.03583951]]\n",
            "\n",
            "Register controller  0\n",
            "[[0.02987847 0.02987847 0.02987847 0.06399035 0.02987847 0.06399035\n",
            "  0.13704735 0.13704735 0.02987847 0.02987847 0.13704735 0.02987847\n",
            "  0.06399035 0.06399035 0.02987847 0.02987847 0.06399035]]\n",
            "\n",
            "Register controller  1\n",
            "[[0.03044354 0.03044354 0.03044354 0.06520054 0.03044354 0.06520054\n",
            "  0.13963921 0.13963921 0.03044354 0.03044354 0.120727   0.03044354\n",
            "  0.06520054 0.06520054 0.03044354 0.03044354 0.06520054]]\n",
            "\n",
            "Register controller  2\n",
            "[[0.0315539  0.0315539  0.0315539  0.06757861 0.0315539  0.06757861\n",
            "  0.14473227 0.14473227 0.0315539  0.0315539  0.08865737 0.0315539\n",
            "  0.06757861 0.06757861 0.0315539  0.0315539  0.06757861]]\n",
            "\n",
            "Output registers\n",
            "[[[1.5619522e-01 3.6405510e-01 2.4730040e-01 6.4242825e-02 3.6694934e-03\n",
            "   1.7096556e-04 9.3055110e-05 7.4118964e-04 4.7222888e-03 8.0085741e-03\n",
            "   5.5994405e-03 7.3264851e-03 1.1864418e-02 7.6907198e-03 9.1296053e-03\n",
            "   1.5433380e-02 9.0548936e-03 7.7580486e-04 3.0735845e-03 3.4805343e-02]\n",
            "  [1.5462944e-01 3.6623263e-01 2.5094676e-01 6.5404132e-02 3.7366862e-03\n",
            "   1.7321817e-04 8.5263397e-05 6.7811733e-04 4.3537766e-03 8.0424296e-03\n",
            "   5.6154565e-03 6.7529748e-03 1.1947004e-02 7.7506555e-03 8.3487825e-03\n",
            "   1.5575092e-02 9.2167640e-03 7.6230161e-04 2.7923877e-03 3.1639837e-02]\n",
            "  [1.5155266e-01 3.7051165e-01 2.5811192e-01 6.7686111e-02 3.8687203e-03\n",
            "   1.7764457e-04 6.9952614e-05 5.5417971e-04 3.6296474e-03 8.1089567e-03\n",
            "   5.6469273e-03 5.6260219e-03 1.2109288e-02 7.8684315e-03 6.8144603e-03\n",
            "   1.5853561e-02 9.5348414e-03 7.3576777e-04 2.2398341e-03 2.5419593e-02]]]\n",
            "\n",
            "Output memory\n",
            "[[1.53451517e-01 1.91733003e-01 1.20752007e-01 4.67776954e-02\n",
            "  1.87714119e-03 6.63887040e-05 5.71547171e-05 8.10630212e-04\n",
            "  1.23109296e-02 3.20335627e-01 1.34666609e-02 6.05338952e-03\n",
            "  1.74705721e-02 6.13835640e-03 5.54322032e-03 1.72877163e-02\n",
            "  5.64005319e-03 3.52535397e-04 2.24963203e-03 3.79090160e-02]\n",
            " [1.51932836e-01 2.33164728e-01 1.49519026e-01 5.57709113e-02\n",
            "  2.65921769e-03 1.00383884e-04 8.11646460e-05 1.00647495e-03\n",
            "  8.62200744e-03 2.39476450e-02 9.42846201e-03 7.97031447e-03\n",
            "  2.15546563e-01 8.14678892e-03 7.66309956e-03 2.35454366e-02\n",
            "  7.75263458e-03 5.20063739e-04 2.95605185e-03 3.57450098e-02]\n",
            " [9.59691256e-02 1.43522516e-01 9.20737982e-02 3.43956351e-02\n",
            "  1.60023221e-03 5.92455399e-05 5.06936558e-05 6.24822453e-04\n",
            "  5.76188974e-03 1.64289754e-02 6.29871432e-03 4.95299604e-03\n",
            "  1.41820936e-02 5.06097171e-03 4.67155408e-03 5.11823237e-01\n",
            "  4.72867070e-03 3.13820143e-04 1.80533435e-03 2.27039568e-02]\n",
            " [5.91653734e-02 8.14148188e-02 5.10284156e-02 1.95256863e-02\n",
            "  8.31907033e-04 2.87384337e-05 2.55908526e-05 3.73527932e-04\n",
            "  4.29324526e-03 1.29569126e-02 4.66313306e-03 2.75492668e-03\n",
            "  7.98310526e-03 2.78832740e-03 2.48107081e-03 7.77205825e-03\n",
            "  2.52830656e-03 1.54702255e-04 1.01951009e-03 1.44007821e-02]\n",
            " [2.34544603e-03 3.82092083e-03 9.89451945e-01 9.02184518e-04\n",
            "  4.35370457e-05 1.57224463e-06 1.37006236e-06 1.79149138e-05\n",
            "  1.22768703e-04 3.09719413e-04 1.32637157e-04 1.38358781e-04\n",
            "  3.98003089e-04 1.40893768e-04 1.28203959e-04 3.97427997e-04\n",
            "  1.30079716e-04 8.38913911e-06 5.06266151e-05 5.46668423e-04]\n",
            " [7.58510287e-05 1.25278646e-04 8.37760381e-05 3.02680928e-05\n",
            "  1.57608019e-06 6.47943210e-08 5.13641076e-08 4.66274514e-07\n",
            "  3.00759575e-06 7.57037378e-06 3.40068982e-06 9.99580920e-01\n",
            "  1.19959332e-05 4.46780814e-06 4.43549652e-06 1.32888226e-05\n",
            "  4.43999670e-06 3.33045250e-07 1.53248266e-06 1.70804651e-05]\n",
            " [7.06852443e-05 1.16138210e-04 7.71103587e-05 2.77759664e-05\n",
            "  1.37546090e-06 5.13669178e-08 5.10722380e-08 5.00266879e-07\n",
            "  3.16400883e-06 7.87712906e-06 3.49421884e-06 4.39132327e-06\n",
            "  1.25957395e-05 4.53644816e-06 4.09018912e-06 1.24739299e-05\n",
            "  9.99607384e-01 2.89723914e-07 1.47320134e-06 1.60219861e-05]\n",
            " [1.18584104e-03 1.88974303e-03 1.09939591e-03 4.29440173e-04\n",
            "  1.79955514e-05 4.65938285e-07 4.99685370e-07 1.17516829e-05\n",
            "  8.56743718e-05 2.16877525e-04 8.88766372e-05 7.32078479e-05\n",
            "  2.18970075e-04 9.93611097e-01 5.74882943e-05 1.89335144e-04\n",
            "  5.98189763e-05 2.71947260e-06 2.81479734e-05 2.91480566e-04]\n",
            " [2.30531842e-02 2.02732366e-02 1.20417969e-02 5.14291739e-03\n",
            "  1.22018573e-04 2.99107842e-06 3.13765963e-06 8.43242160e-05\n",
            "  2.57973489e-03 8.96474998e-03 2.81715859e-03 5.06740063e-04\n",
            "  1.52087898e-03 4.97131550e-04 3.92118061e-04 1.30400131e-03\n",
            "  4.09917004e-04 1.72824366e-05 1.99022965e-04 6.07380224e-03]\n",
            " [7.23432973e-02 5.74880391e-02 3.44623849e-02 1.49681857e-02\n",
            "  3.01938009e-04 7.46479145e-06 7.71112809e-06 2.07428820e-04\n",
            "  8.40905588e-03 2.98384707e-02 9.22204368e-03 1.24677317e-03\n",
            "  3.73932673e-03 1.22318277e-03 9.68210225e-04 3.21766711e-03\n",
            "  1.01195369e-03 4.28323292e-05 4.90620907e-04 1.91922951e-02]\n",
            " [2.51161642e-02 2.18189359e-02 1.30489524e-02 5.56453038e-03\n",
            "  1.31662295e-04 3.37976371e-06 3.46139268e-06 8.73133467e-05\n",
            "  2.80970335e-03 9.80842486e-03 3.07287951e-03 5.34046034e-04\n",
            "  1.59754301e-03 5.25472220e-04 4.19713644e-04 1.38706691e-03\n",
            "  4.37572948e-04 1.93180804e-05 2.08597892e-04 6.61313161e-03]\n",
            " [8.15214776e-03 1.31454524e-02 8.06229655e-03 3.04844487e-03\n",
            "  1.37194758e-04 4.26341285e-06 4.33917467e-06 7.18016381e-05\n",
            "  5.03323099e-04 1.26901304e-03 5.31587051e-04 5.02688519e-04\n",
            "  1.47885573e-03 5.04872703e-04 4.24332015e-04 1.35343417e-03\n",
            "  4.35287540e-04 2.42973765e-05 1.84391654e-04 1.94482785e-03]\n",
            " [2.30840705e-02 3.71846482e-02 2.27213278e-02 8.60723574e-03\n",
            "  3.84484272e-04 1.17285799e-05 1.21751818e-05 2.06081400e-04\n",
            "  1.44698482e-03 3.64773581e-03 1.52522756e-03 1.42997771e-03\n",
            "  4.21617366e-03 1.43432303e-03 1.19532016e-03 3.82381142e-03\n",
            "  1.22759037e-03 6.74359471e-05 5.24599571e-04 5.51968068e-03]\n",
            " [8.18594638e-03 1.32214008e-02 8.16577859e-03 3.07452120e-03\n",
            "  1.39673371e-04 4.43167028e-06 4.48051287e-06 7.08371153e-05\n",
            "  4.93441476e-04 1.24327268e-03 5.22686169e-04 5.04507858e-04\n",
            "  1.48072117e-03 5.07924298e-04 4.30065120e-04 1.36593124e-03\n",
            "  4.40357020e-04 2.51818819e-05 1.83893731e-04 1.94472424e-03]\n",
            " [7.06361700e-03 1.14729051e-02 7.22464919e-03 2.69439118e-03\n",
            "  1.27237479e-04 4.40232270e-06 4.04240654e-06 5.64999500e-05\n",
            "  3.89654655e-04 9.82661149e-04 4.17863834e-04 4.23924357e-04\n",
            "  1.22843485e-03 4.29900567e-04 3.81295977e-04 1.19235460e-03\n",
            "  3.88103770e-04 2.39970104e-05 1.54589434e-04 1.65737327e-03]\n",
            " [2.17262805e-02 3.52127701e-02 2.19795629e-02 8.23975541e-03\n",
            "  3.84408224e-04 1.29845394e-05 1.20326949e-05 1.78385846e-04\n",
            "  1.24084880e-03 3.13161826e-03 1.32451952e-03 1.30680285e-03\n",
            "  3.80012533e-03 1.32097083e-03 1.15917693e-03 3.64480494e-03\n",
            "  1.18261657e-03 7.10503373e-05 4.79976938e-04 5.12590911e-03]\n",
            " [7.23810680e-03 1.17419753e-02 7.35720713e-03 2.75193108e-03\n",
            "  1.29074237e-04 4.40661870e-06 4.06552272e-06 5.87646864e-05\n",
            "  4.07291082e-04 1.02758373e-03 4.35608148e-04 4.34888294e-04\n",
            "  1.26271497e-03 4.40213189e-04 3.88108223e-04 1.21740880e-03\n",
            "  3.95556795e-04 2.40681329e-05 1.59261574e-04 1.70366443e-03]\n",
            " [4.15855844e-04 6.85326057e-04 4.56912094e-04 1.64874305e-04\n",
            "  8.41003930e-06 3.32904108e-07 2.89510609e-07 2.71984413e-06\n",
            "  1.73870521e-05 4.35472066e-05 1.94516870e-05 2.45153369e-05\n",
            "  6.92766844e-05 2.54170118e-05 2.42057686e-05 7.30636821e-05\n",
            "  2.42785591e-05 1.77474533e-06 8.51474124e-06 9.39067395e-05]\n",
            " [3.08459718e-03 4.95491363e-03 2.97898566e-03 1.14210031e-03\n",
            "  5.06381148e-05 1.52854295e-06 1.46720572e-06 2.79928354e-05\n",
            "  2.00756418e-04 5.07915625e-04 2.10779224e-04 1.86655001e-04\n",
            "  5.50503843e-04 1.86208548e-04 1.56257156e-04 5.02797833e-04\n",
            "  1.61029413e-04 8.49298340e-06 7.08153311e-05 7.44340010e-04]\n",
            " [5.71503490e-02 6.22657090e-02 3.82738262e-02 1.53686600e-02\n",
            "  5.20647620e-04 1.65814254e-05 1.52930770e-05 2.67719763e-04\n",
            "  5.40517969e-03 1.80283710e-02 5.89998718e-03 1.84159260e-03\n",
            "  5.39658358e-03 1.84574386e-03 1.58499018e-03 5.05212601e-03\n",
            "  1.62707479e-03 9.05741472e-05 6.94974151e-04 1.45458318e-02]]\n",
            "final cost: 6.1672915e-07\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "N8KIX7K1522T",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 51
        },
        "outputId": "5318c9fa-58ea-491e-bd4b-2bac9f48236f"
      },
      "source": [
        "np.argmax(outputMemory, axis=1)"
      ],
      "execution_count": 136,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "array([ 9,  1, 15,  1,  2, 11, 16, 13,  0,  0,  0,  1,  1,  1,  1,  1,  1,\n",
              "        1,  1,  1])"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 136
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "edz6c1hoZJ43",
        "colab_type": "text"
      },
      "source": [
        "## Bucket Sort"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "89QSMdkFZMxS",
        "colab_type": "text"
      },
      "source": [
        "https://www.geeksforgeeks.org/bucket-sort-2/"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "46mVnczkaSMN",
        "colab_type": "code",
        "outputId": "5aef80ae-4ead-4984-c404-f92fa8abbd96",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 105
        }
      },
      "source": [
        "# Python3 program to sort an array \n",
        "# using bucket sort \n",
        "def insertionSort(b): \n",
        "\tfor i in range(1, len(b)): \n",
        "\t\tup = b[i] \n",
        "\t\tj = i - 1\n",
        "\t\twhile j >=0 and b[j] > up: \n",
        "\t\t\tb[j + 1] = b[j] \n",
        "\t\t\tj -= 1\n",
        "\t\tb[j + 1] = up\t \n",
        "\treturn b\t \n",
        "\t\t\t\n",
        "def bucketSort(x, M): \n",
        "\tarr = [] \n",
        "\tslot_num = 3 # 10 means 10 slots, each \n",
        "\t\t\t\t# slot's size is 0.1 \n",
        "\tfor i in range(slot_num): \n",
        "\t\tarr.append([]) \n",
        "\t\t\n",
        "\t# Put array elements in different buckets \n",
        "\tfor j in x: \n",
        "\t\tindex_b = int((slot_num * j)/M) \n",
        "\t\tarr[index_b].append(j) \n",
        "\t\n",
        "\t# Sort individual buckets \n",
        "\tfor i in range(slot_num): \n",
        "\t\tarr[i] = insertionSort(arr[i]) \n",
        "\t\n",
        "\t# concatenate the result \n",
        "\tk = 0\n",
        "\tfor i in range(slot_num): \n",
        "\t\tfor j in range(len(arr[i])): \n",
        "\t\t\tx[k] = arr[i][j] \n",
        "\t\t\tk += 1\n",
        "\treturn x, arr\n",
        "\n",
        "# Driver Code \n",
        "x = [1, 4, 3, 6, \n",
        "\t2, 5, 7, 9, 8] \n",
        "print(\"Sorted Array is\") \n",
        "print(bucketSort(x, 10))\n",
        "\n",
        "\n",
        "# Driver Code \n",
        "x = [1, 14, 13, 6, \n",
        "\t12, 15, 7, 19, 8] \n",
        "print(\"Sorted Array is\") \n",
        "print(bucketSort(x, 20))"
      ],
      "execution_count": 101,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Sorted Array is\n",
            "([1, 2, 3, 4, 5, 6, 7, 8, 9], [[1, 2, 3], [4, 5, 6], [7, 8, 9]])\n",
            "Sorted Array is\n",
            "([1, 6, 7, 8, 12, 13, 14, 15, 19], [[1, 6], [7, 8, 12, 13], [14, 15, 19]])\n"
          ],
          "name": "stdout"
        }
      ]
    }
  ]
}