Author: Robert Hyatt
Date: 09:29:40 04/14/04
Go up one level in this thread
On April 14, 2004 at 02:26:05, Christophe Theron wrote: >On April 14, 2004 at 02:21:41, Christophe Theron wrote: > >>On April 14, 2004 at 00:26:34, Eric Oldre wrote: >> >>>After you find the 1st "good" move don't you narrow the alpha beta window so >>>that you don't know how much worse the 2nd move is, only that it is not as good >>>as alpha? >>> >>>Or do you not narrow the window at the root node? that seems like it would >>>greatly expand your search tree. >>> >>>or am i missing something else? >>> >>> >>>On April 14, 2004 at 00:09:24, Robert Hyatt wrote: >>> >>>>Simple idea: >>>> >>>>a move is "easy" and can be made after using less than the planned time limit if >>>>and only if >>>> >>>>1. estimated score for first root move is way higher than the second move. IE >>>>say 2.00 better. >>>> >>>>2. This is a recapture. IE opponent just captured a piece of ours and we are >>>>recapturing on the same square. >>>> >>>>Other types of "easy" moves have higher risk to stop the search early... >>>> >>>> >>>>> >>>>>Thanks, >>>>>Eric Oldre (new chess programmer) >> >> >> >>I think that by "estimated score", Bob means the score returned by a SEE (Static >>Exchange Evaluator), not by a real search. >> >> >> >> Christophe > > > >I would add that you need some kind of logic to tell your search to not stop >early if you suddenly discover that the real score (the score found by a real >search) is much worse than what the SEE expected. > >So if you have an obvious recapture you allocate a small amount of time for the >search. But if the search does not confirm that the move is obvious (the score >for that move is not that good, or the best move found by the search is not the >supposed obvious move) you extend the time by falling back to a normal (much >longer) target time in order to see what's going on. > > > > Christophe Correct. If I "fail low" at any point, "easy" is turned off...
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.