Author: Tim Foden
Date: 07:49:24 10/04/01
Go up one level in this thread
On October 04, 2001 at 09:35:55, Steve Maughan wrote: >Tim, > >[D]5rk1/1pp2R1p/p1pb4/6q1/3P1p2/2P4r/PP1BQ1P1/5RKN w - - 2 0 > >>This can be evaluated in 2 ways... >> >>(1) BxP (+1000) BxB (-3500) R1xB (+3500) QxR (-5500) RxQ (+10000) RxR (-5500) >>.........+1000.......-2500........+1000.......-4500........+5500...........0 >> >>>>> value = 0 >> >>(2) BxP (+1000) BxB (-3500) R7xB (+3500) RxR (-5500) RxR (+5500) QxR (-5500) >>.........+1000.......-2500........+1000.......-4500.......+1000.......-4500 >> >>>>> value = -2500 > >Interesting position. > >I guess option 1 is better. I agree. >My 'old' SEE would 'work' on this position as it does a normal search linited to >capturing on one sqaure i.e. goes with option 1. The capturing *is* limited to only one square. Maybe you mean that it is limited to capturing *from* only one square? The difference here is that white has a choice of which R to capture with, which changes the predicted score. >My current SEE does not work >as it does not see the difference between R7xB and R1xB. I suspect that many >SEE routines would not work with this - Leen's routine which I based mine on >wouldn't. I'm not sure if it's worth worrying about as it's so rare - I guess a >simpler and faster routine would work in 99.9% of cases. Probably. But I would like to be more certain :) I had the thought that maybe it would be possible for the SEE to detect that there is/was a choice, and in that case, to return some kind of failure status (along with the value it calculated anyway). Then the code that called it has a choice of whether to look into it in more detail, or use a (different) simple approximation, or whatever. In the q-search, you would simply search the move (not prune it). In other places (e.g. for move ordering) you could just use the value, even though you know it is suspect. Cheers, Tim.
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.