Author: Dan Newman
Date: 02:26:15 10/16/99
Go up one level in this thread
On October 15, 1999 at 20:34:25, leonid wrote: >On October 15, 1999 at 13:56:07, Jari Huikari wrote: > >>On October 15, 1999 at 12:46:57, Antonio Dieguez wrote: >> >>>why did you say N passes over N moves? a pass is a round or a loop right? if I >>>understood what Jari told, there are only 7, or not? (what a coincidence I do 7 >>>rounds too!) >> >>My program is very different from the most of other programs. I have only >>written a primitive program, in my way. And done changes which have made >>_it_ better, but are done in a different (and more effective) way in usual >>programs. :-) In _my _program_ my movesorting doesn't make anything slower. >>(I generate _all_ the legal moves before I try any. It may be one difference.) >> >> Jari > >Are you saying that usually people find and do movesorting before recognizing >what move is legal? I am asking this question not because I contest what you >have said but for simple curiosity. > >Regards, >Leonid. Many of us generate "pseudo-legal" moves, sort them, and then filter out the illegal ones as we try them. (Pseudo-legal moves are moves that would be legal if we were allowed to put the king en prise.) Testing a move to see if it leaves the king en prise is fairly expensive, and since we are probably only going to try an average of say 6 out of the 36 or so moves that we generate (due to alpha-beta cutoffs), we delay testing for legality and so avoid doing a lot of work. Also, many do as I do and generate the captures and promotions separately and try them before trying the rest of the moves. -Dan.
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.