Author: Don Dailey
Date: 14:47:04 06/10/98
Go up one level in this thread
On June 10, 1998 at 13:10:23, Carlos Adan Bonilla wrote: >Hi! > >When using a computer to analyze or to play chess, you may find that a >lot of programs have a "Next best move" option in the menu. >As a matter of fact, if you wanna know the second best move in a certain >position alpha-beta pruning is not enough because that second best move >could have been cut off. > >So my first question is: >Do those programs use alpha-beta cut offs? Or do they use an special >version of that algorythm? > >I mean programs which move that second best move immediately after >pushing the "Next best" move... (they don't need to search for it >again). > >Second question: when the user selects "Next best" is because he doesn't >like the best move... but the new selected move may have been selected >because it's preparing to move that former best move again but in the >next move (I mean two plies after, the next move of that colour). >But this new line is likely to unsatisfy the user again because it >contains the same rejected move... >How can this things be handled? >Do the existing programs take this into account? >(I don't think so) > >Thx in advance for your answers... I had a version of my program that would give you the best n moves within half a point of best! I found this a very useful feature because I was not interested in blunders, only alternatives that might be reasonable. To get this, you have to relax your alpha/beta bounds after searching each root move. It slows down the search somewhat, but since you have a goal in mind (within half a pawn), you can still use a reasonable bound and do it much faster than Bob's approach. But Bob's approach is probably best if you want the best n moves and do not put scoring constraints on it. - Don
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.