1
$\begingroup$

I'm looking for a simple chess algorithm that works well on a ~132MHz CPU.

I know that MiniMax would take too long to go deep in order to find good moves. Is Monte Carlo Tree Search working well on low CPUs? Are there other options? What algorithms did old chess computers use?

$\endgroup$
1
$\begingroup$

The given algorithm (Minimax and MCTS) are brute force searching strategies in the game tree. They will easily bring a CPU to it's limit if the search horizon is set to 8 moves and more into the future. To reduce the CPU consumption to a minimum, not a chess solver but a chess simulator needs to be realized first. Somebody may ask what the purpose of a chess simulator might be if the rules are known. But the normal chess rules are not describing the full game. What a good simulator contains also, are macro actions and an evaluation function. In such enhanced chess game, the user can make more than one move at once and the game prints out the score, before the game is over. A well known technique is to count the number of pieces and their strategic position on the board.

If a chess simulator was written and well tested, it make sense to build on top of the model a solver which can be a normal MCTS one who is traversing a bit in the game tree and recognizes unintentionally the winning move.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.