Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Kasparov's and the computers

Author: Dann Corbit

Date: 15:08:55 02/02/99

Go up one level in this thread


Actually, you could say that chess is already reduced to a mathematical
equation.  Unfortunately, the algorithm that implements that equation is:
ouch = O(exp(n))
Which is not good for solving things.  Even with Alpha/Beta and all the other
clever searching techniques applied, the algorithm is still:
ouch = O(exp(n))
but with a smaller multiplier.  In other words, the difficulty is not in
reducing the problem to a mathematical model.  The problem is in finding an
algorithm which is faster than exponential.  If you could discover a linear
equation to solve chess positions, you could find a solution for practically any
position almost instantly.  Even a quadratic equation in the number of plies
would be solvable since the longest game is about 6000 moves that would give
k*144,000,000, which would take a few seconds or perhaps minutes to solve,
depending on k and how advanced the hardware you are using is.

The Hakmem page lists a simple way to solve the game of chess.  Ennumerate all
of the board positions.  Any position generator that produces legal moves could
do that.  Unfortunately, there are more than 10^40 of them.  So it would take
longer than it hurts even to imagine.




This page took 0 seconds to execute

Last modified: Thu, 15 Apr 21 08:11:13 -0700

Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.