AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |
Back to Blog
Stockfish chess make ai faster11/9/2022 I should always assume my opponent makes their best possible move if I want to win, right? Chess bots can’t actually look from a position to every terminal position that could arise from that position most of the time, and so rely on evaluation algorithms that do the best they can with many positions that result a certain depth from the original position, but they still use this minimax collapsing strategy (alternating between choosing the maximal and minimal child evaluations as the evaluation of each parent in your decision tree). I should make the first move because if I make the second move I’m allowing my opponent to win on the spot, and after all, this is chess. If I make the second move, my opponent has a move that wins and a move that loses (or allows me to win next move). If I make the first move, my opponent has 2 legal moves that both immediately result in a draw. To give a very simple example: imagine I have a position where I have 2 legal moves. What do I mean by that? The driving idea behind a chess bot is an algorithm called the minimax algorithm, which essentially involves evaluating a position by evaluating all of the terminal positions (positions where the game is over) that derive from a position (as a win, draw, or loss) and collapsing evaluations up from those terminal positions via alternating optimization by player. There’s one variable that no one has touched on yet regarding exactly how chess and pokemon differ in this sense: minimax just isn’t a valid solution to pokemon. I definitely would like to try and start a project like this, maybe name it ‘Stunfisk’ as a play on ‘Stockfish’ and this sub’s name.Įdit: The only thing I can think of being a bit annoying is constantly adapting to the new meta, but that just involves retraining with new mons and slightly different sets for preexisting mons. #Stockfish chess make ai faster how to#I can imagine this might be somewhat helpful when testing a newly built team and understanding its flaws more, but I haven’t thought any further than that on how to actually make it. Given a team comp, and whatever mons you know about the other team, it can make some educated guesses on what they might have, recommend moves/plays, and give a rating for the current position, which side the game is more favorable to, and whatnot. People might have thought of this already and I haven’t fully fleshed out the idea or details so it’s possible there’s a major error I’m overlooking, but what does everyone think about a Pokemon AI, kind of like Stockfish for Chess? Chess and Pokemon are somewhat similar in that both games require you to think multiple moves ahead and make a game plan based on what your opponent does/brings to the table.
0 Comments
Read More
Leave a Reply. |