[ITensors] Optimize directsum again (#1221)

src/itensor.jl

@@ -269,6 +269,14 @@ ITensor(x::RealOrComplex{Int}, is...) = ITensor(float(x), is...)
 # EmptyStorage ITensor constructors
 #
 
+# TODO: Replace with a simpler and more generic `zeros` constructor
+# when the new `UnallocatedZeros` type lands.
+# This is only used internally inside the implementation of `directsum`
+# right now.
+function zeros_itensor(elt::Type{<:Number}, inds::Index...)
+  return ITensor(elt, zero(elt), inds...)
+end
+
 # TODO: Deprecated!
 """
     emptyITensor([::Type{ElT} = NDTensors.EmptyNumber, ]inds)

src/qn/qnitensor.jl

@@ -11,6 +11,22 @@
   return setindex!!(T, x, I...)
 end
 
+# TODO: Replace with a simpler and more generic `zeros` constructor
+# when the new `UnallocatedZeros` type lands.
+# This is needed for now since there is some issue with calling
+# `setindex!` on `EmptyTensor`, it's not really worth investigating
+# right now since that type will be removed soon anyway in
+# https://github.com/ITensor/ITensors.jl/pull/1213.
+# This is only used internally inside the implementation of `directsum`
+# right now.
+function zeros_itensor(elt::Type{<:Number}, inds::QNIndex...)
+  return itensor(
+    tensor(
+      BlockSparse(elt, undef, NDTensors.Dictionary{Block{length(inds)},Int}(), 0), inds
+    ),
+  )
+end
+
 """
     ITensor([::Type{ElT} = Float64, ][flux::QN = QN(), ]inds)
     ITensor([::Type{ElT} = Float64, ][flux::QN = QN(), ]inds::Index...)

src/tensor_operations/tensor_algebra.jl

@@ -279,8 +279,15 @@ end
 function directsum_projectors(
   elt1::Type{<:Number}, elt2::Type{<:Number}, i::Index, j::Index, ij::Index
 )
-  D1 = ITensor(elt1, QN(), dag(i), ij)
-  D2 = ITensor(elt2, QN(), dag(j), ij)
+  # Ideally we would just use the following but it gives
+  # an error that `setindex!` isn't defined:
+  # D1 = ITensor(elt1, dag(i), ij)
+  # D2 = ITensor(elt1, dag(j), ij)
+  # Or with new notation:
+  # D1 = zeros(elt1, dag(i), ij)
+  # D2 = zeros(elt1, dag(j), ij)
+  D1 = zeros_itensor(elt1, dag(i), ij)
+  D2 = zeros_itensor(elt1, dag(j), ij)
   directsum_projectors!(tensor(D1), tensor(D2))
   return D1, D2
 end