The tree part of Monte Carlo Tree Search (or more precisely the UCT algorithm) is basically using the UCB1 mechanism not only at the root, but also at other nodes that you end up visiting many times.
The problem with this in the case of card games is that you will be dealing the cards you haven't seen randomly at each simulation, so there won't be many instances of visiting the same situation twice, because from the point of view of the next player, the situation is different after each plausible configuration is determined.
AI for a simple shedding card game
Hello,
We have this game with 4 players where each players take turns to discard cards that must be higher than the ones previously discarded. The card combinations are based on poker. Discarded cards are visible to everyone. The last player with remaining cards loses. I have code on the gameplay, and the AI has code that can list valid combinations from the cards that it has.
i'd like to toss out an idea, but before i can(as it might be moot depending), i'd like a bit of clarification of the rules, if i understand them correctly:
you can discard any number of cards, so long as it's a valid poker combination, and it must be higher than the last card discarded?, exactly how is that value calculated?, as well, how many cards are players given at start?
Author
I have more or less finished the game, it can be played here: [/url] . Java is needed.
I also put the rules here: [url="#"]
Everything on the website is still beta, the game doesn't even have drag-and-drop. I was too lazy to write it 
This topic is closed to new replies.
Advertisement
Popular Topics
Advertisement
