This was the problem I had asked in the last post...
"Given a particular chess game position and the number of moves played, can we precisely determine what moves were played by the players to arrive at the current configuration on the set?"
Now, there can be 'x' number of ways in which a particular chess position can be obtained by 'n' moves. Let us construct a tree for chess game positions. Let us denote each possible chess board position by a node. (I don't know the exact number, but it may well be in billions, trillions or much more ;)). The problem of constructing a game tree based on moves is that a node may have multiple parents. For example, lets assume white plays knight, black plays pawn and white plays a pawn. The same board position can be obtained by white playing pawn, black playing pawn and white playing knight.
So, it seems impossible to determine exact moves played to arrive at a chess game position. But, maybe given some snapshots of the game at different intervals of time (say 3 game positions after every n/3 moves), it "may" very well be possible to predict exactly the moves.
What is the practical use of this problem? Well, I believe, if we can do this for game of chess, we can do it for other things too. For example, stock markets, marketing a new product, etc. I know there will be many people not agreeing with me on this. (Rules of chess are fixed, stock market is based on market sentiments, etc etc.) But even the stock market is governed by some rules, like the recession comes every 10 years, depression comes every 60 years etc. (The years mentioned are not verified numbers) So, we may be very close to predicting stock market if we are able to predict chess moves :D.
"Given a particular chess game position and the number of moves played, can we precisely determine what moves were played by the players to arrive at the current configuration on the set?"
Now, there can be 'x' number of ways in which a particular chess position can be obtained by 'n' moves. Let us construct a tree for chess game positions. Let us denote each possible chess board position by a node. (I don't know the exact number, but it may well be in billions, trillions or much more ;)). The problem of constructing a game tree based on moves is that a node may have multiple parents. For example, lets assume white plays knight, black plays pawn and white plays a pawn. The same board position can be obtained by white playing pawn, black playing pawn and white playing knight.
So, it seems impossible to determine exact moves played to arrive at a chess game position. But, maybe given some snapshots of the game at different intervals of time (say 3 game positions after every n/3 moves), it "may" very well be possible to predict exactly the moves.
What is the practical use of this problem? Well, I believe, if we can do this for game of chess, we can do it for other things too. For example, stock markets, marketing a new product, etc. I know there will be many people not agreeing with me on this. (Rules of chess are fixed, stock market is based on market sentiments, etc etc.) But even the stock market is governed by some rules, like the recession comes every 10 years, depression comes every 60 years etc. (The years mentioned are not verified numbers) So, we may be very close to predicting stock market if we are able to predict chess moves :D.
3 comments:
You can possibly use Hidden Markov Models...
To undertand in Maths :
http://en.wikipedia.org/wiki/Stochastic_process
Post a Comment