-1

I am implementing an artificial intelligence that plays chess using the minimax algorithm. The initial board I have is white rook and white king against black rook and black king. All the evaluations I've thought of, do not provide enough intelligence to my IA.

What could be a good evaluation of this especific configuration board?

Sergio
  • 81
  • 4
  • What did you try? How do you know it wasn't good enough? – myrtlecat Oct 31 '22 at 10:35
  • Unless the initial position allows some obvious win or capture of material, this is a draw. What is your current evaluation function? And for which position is it a problem? – trincot Oct 31 '22 at 10:47

1 Answers1

1

Do you just want an evaluation function that works in this specific case, and not a general one? There are two options that you can implement, if you want it to work for other cases you can just expand on these:

  1. More advanced: Read up on https://www.chessprogramming.org/Endgame where you find links to some concepts that you can implement in your evaluation function, such as King Centralization, Draw Evaluation, Mop-up Evaluation etc.

  2. Easier approach: Use piece square tables only. Usually you have a set of tables for the middle game and one for the end game. They will differ since e.g. a king in the middle game wants to be safe behind own pawns, but in the end it needs to be active. Here you can basically copy/paste these ones and try for yourself: https://www.chessprogramming.org/PeSTO%27s_Evaluation_Function.

eligolf
  • 1,682
  • 1
  • 6
  • 22