I am developing the hexxagon game...i am finding the ai very difficult...can anybody help me with this??? it is very urgent....does anyone have any algorithms or source code fr the same??? earliest reply will be appreciated

Standard game tree exploration techniques, possibly with conservative heuristic pruning of blatantly bad moves, of useless moves in one's "controlled" part of the board (they can wait until endgame) and of cycles of repeated positions.
Board evaluation can count safe and unsafe pieces and empty cells according to their position: behind one's lines (the same as a safe piece, they can be occupied in the endgame), reachable by either or both players or not reachable.

Quote:

Original post by annoynomous it is very urgent....does anyone have any algorithms or source code fr the same??? earliest reply will be appreciated

Smells like homework...

Quote:

Original post by annoynomous I am developing the hexxagon game...i am finding the ai very difficult...can anybody help me with this??? it is very urgent....does anyone have any algorithms or source code fr the same??? earliest reply will be appreciated

You need to be clear about what your requirements are for the game and the AI's function within it. If you want help from forum members, you'll need to provide some indication of what it is you're struggling with. Just saying that you find it difficult and can somebody help is far too broad a problem for people to take interest and offer advice. What advice do you expect them to offer?

Try this:

Sit down and look at your game. Ask yourself what you as a person would do if you took on the role of the AI within the game. Consider what information you have available, what decisions you need to make and what actions you can take as a result of your decisions. Then you need to ask yourself whether your AI should take a deliberative approach to mapping information to actions, or a reactive approach. If you're unsure, post your findings about your game analysis and we can offer advice. Once you've worked out your needs, we can also suggest techniques that should solve your problem for you, or at least get you close to a good solution.

Just asking for algorithms or source code, without specifying what the actual problem is, or what you want to focus on, is not going to receive much of a positive response (and certainly sounds like the occasionaly requests we get that are hiding the 'can you do my homework for me' question)!

Cheers,

Timkin

Quote:

Sit down and look at your game. Ask yourself what you as a person would do if you took on the role of the AI within the game. Consider what information you have available, what decisions you need to make and what actions you can take as a result of your decisions.

For example, as an exercise to suggest algorithms to evaluate positions, you could assess which of two slightly different positions is better for the same player, or assess which player has an advantage in a random position, and analyze what information and heuristic or exact criteria you used to make a decision.
Your AI can be based on a formal, simplified and generalized approximation of how you, as a player, make decisions.

A radically different approach is using Monte-Carlo simulations. Basically, from the current position run a bunch of games where each player plays randomly. Use the fraction of games you win as evaluation function. This approach combined with an algorithm called UCT -which is very popular in computer go these days- will probably give you a strong program with relatively little effort.

It's probably as hard to make a strong player using this approach than the more traditional one (hand-craft a evaluation function and add search), but I am writing a go program right now with MC methods and it's a lot of fun.