Author: Robert Hyatt
Date: 14:32:34 01/09/99
Go up one level in this thread
On January 09, 1999 at 00:57:18, Horace Ho wrote: >On January 08, 1999 at 18:01:17, Robert Hyatt wrote: > >>On January 08, 1999 at 15:01:58, Horace Ho wrote: >> >>>After a check-escape-check-escape-capture, my program >>>is satisfied after the quiesce search. However, my opponent >>>can mate-in-1 after that capture. >>> >>>Besides, the capture in fact leads my side to a bad position >>>It thinks that's ok after this capture at end of quiesce search, >>>no re-capture is found. However, it does not check one more ply >>>to find the mate-in-1. >>> >>>Here comes my question: >>> >>>1. Should I generate check moves in quiesce search? >>> (I consider those costly, comparing with captures.) >>>2. Is there a way to prevent the checkmate-in-1 after >>> this kind of check-check-capture? >>> >>>Thanks >>>horace >> >>one solution is to simply hope you search deep enough that it doesn't >>cause problems. I see the same thing, but I don't find myself losing games >>because of this, although I have seen an occasional PV from the search where >>crafty was winning material but the last move was the opponent mating it with >>a capture. Fortunately it has a chance to vary before getting to that move, if >>you go deeply enough... > >How about the question if quiesce should include checks? > >Thanks again I don't do that for one simple reason. If you notice you are in check, and try all moves to see if you are mated, there are other places in the q-search where you can 'stand pat' and avoid mate. So it can waste a lot of time. If you want to 'generate' check moves in the q-search, that really sounds like a waste... and it would be better to rely on search extensions instead...
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.