help with an intro screen

[[Gandalf]]

You don't really need 3d models for a good chess program

Quote:

also about pocess's during the opponent (player's) move.. wouldn't that not work right, since the compter wouldn't be taking into account, the move that the player is making.. since if it decided on it's move, before the player moved, then the computer moved.. wouldn't it never take the last move into event.. so wouldn't it be like.. really REALLY easy computer.. since if you put it into check, it doens't know that Razz so it'll just lose...???

No, the computer would be thinking both during the players move and its own move. Think about it; thinking while the person makes their move is still advantageous, although not as much as during the move. If the ai thinks of a bunch of possibilities while the human is moving then those possibilities are narrowed down when the move is made - so now the computer can focus on those 'branches'. The computer thinks more than one move ahead, just like a good chess player would. It can predict that you will check/checkmate it so it would avoid that outcome.

[c0bra54]

ooo ok.. also i was thiking (while working on other programming hwk :S

what about if to do it, you did kinda program for every outcome.. but not really.. like if you said

for every piece where can it move, for every spot it can move how many pieces of mine, can get taken, how many pieces can i take.. and those are also not only how many but what.. also it has immediate moves, like

if he can take King then Queen, move not ok... and it has it be a certain value ... positive you tkae, negative you don't take, and then it taqkes numbers and branches that do make the grade.

then it assumes it did that move, then it says where can the human move... takes those values, and then makes it's final decision, based on not only the highest value, but then also what piece s taking what.. like it's kinda dumb for him to move his bishop if then the player can take his, but a rook (castle) to bishop would be ok, and a knight to bishop perhaps better.. like then instead of checking every branch, it see it's best move, then it finds the worst move the human has comparitivly.. and then decides...

ny input (and i dun just wanna goto google and type, how to make chess AI)


thx - might seriously consider this

and usin a function would be an excellent way to return the value

[[Gandalf]]

Well, ya. Most heavy duty chess engines (programs) do some heavy duty calculations. Those calculations being checking the advantages and disadvantages of every move possible. Which is how we get search depth. If the computer is looking ahead towards its next move, then players move, then its move again, that's a search depth of 2. The higher the search depth the better the program will be able to make its decision. What has to be kept in mind is the time that will be spent making the move. The larger the search depth the more possibilities, exponentially, so the longer it will take to make the neccessary calculations. The thing is, to decrease this time you should take into consideration some things which are excluded as soon as they are noticed by the program, obvious "don't go there" situations.

The computer doesn't only take the values of the peaces into consideration, it takes positioning, checkmate oppurtunities, control of the center of the board, attack/defend, king safety, and a large number of other variables. The best program has a good balance of each of these.

Finally, just as a note of interest. Computers don't "think" like humans do. Humans look at the board, then they immediately dismiss the bad positions, and focus on places where they can develop. Computers on the other hand, are much better at doing brute force calculations, and so they use this method more. It's still surprising that even the best supercomputer chess boxes in the world are about even to the best humans in the world.

[c0bra54]

yeh.. i am still thinking of maybe working on it during the summer, becuase onceyou develop a good system of what the AI will look for, and hwo to tell it what and where everything is, it probly wouldn't be impossible to set one up, and let it take lik 3 or 4 mins / move to a depth of 2 moves... be interresting to see if it would a descent opponent (prolly not.. but still)

[[Gandalf]]

Well, that matters... on how well it is coded and optimized. Seeing as how Turing marmolously (yep ) increases requirements for doing things, 2-3 minutes could probably do a depth of 2 or maybe 3, ya. The thing is, well I don't know, but a fairly decent played could crush a program with a depth of 2. Still, if you get it up to 3 maybe it'll be a pretty good challenge.

The two best commercial programs in the world go to search depths of 8-12, even 14 sometimes (longer games).

[c0bra54]

yeh.. but this is mainly just for testing, to see what it would be like..