Primary Dual Hybrid Gradient Algorithm for Convex Optimization Problems

The Primary Dual Hybrid Gradient Algorithm (PDHG) is demonstrated by solving a two player game for advantageous positions for each player. This same algorithm would work with two different scoring matrices as long as a move has a loss for one and a gain for another player (zero-sum game). This is a central idea behind game theory.

http://www.optimization-online.org/DB_FILE/2014/08/4498.pdf

https://hal.inria.fr/hal-00490826/