Current approach is extremely long, a 10-fold MC move calculation takes over a minute and still results in pseudo random moves. Want to find a way to do 100-fold under 10 seconds.
Maybe nested dictionary for the board isn't needed and should try a simple numpy array? We could implement a graph storing every piece and every piece they see, but updating it in each move might take even more time.
Current approach is extremely long, a 10-fold MC move calculation takes over a minute and still results in pseudo random moves. Want to find a way to do 100-fold under 10 seconds. Maybe nested dictionary for the board isn't needed and should try a simple numpy array? We could implement a graph storing every piece and every piece they see, but updating it in each move might take even more time.