Author: Andrew Shapira
Date: 20:23:40 12/10/05
Suppose that we have a program that uses null window searches, i.e., it always has beta = alpha + epsilon, for very small positive epsilon. Also suppose that we are searching some position P whose evaluation will eventually be found to be some value x, and we're about to search P using a null window search S(P,alpha) with some value alpha. Is anything known about the time to run S(p, alpha), as a function of alpha-x? Here alpha may be less than, equal to, or greater than x. The time will obviously depend on the previous state of the program, including hash tables, but maybe something useful can be said about this time anyway.
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.