Author: scott farrell
Date: 08:01:53 11/04/03
Go up one level in this thread
On November 04, 2003 at 10:18:48, Mridul Muralidharan wrote: >Hi, > >In this specific position , proper evaluation that an unstoppable pawn duo at >seventh rank gets a high value should be sufficient. >But that is not the main problem - this would be the end result of a variation - >and if this position is not given the sufficient bonus it requires then it could >have bad instabilities. >example : even if the white rook is absent - still white is up - but definitely >by more than atleast 2 pawns here. > This is indeed one of the problems that I faced ,and I did mention it in one >of my previous posts on blind dependence on stand-pat. >Since I also do "threatening moves" in qsearch - this is indeed an even major >problem for me. >I dont have a solution yet - though lot of hacks to get this reduced and in most >positions , eliminated : (like if previous was a threatening move , then see how >good the responses are before trying to see whether the standpat should be used >, etc). yeh, this is the issue I was trying to get at. I think we should change its name - something like "blinkered blind stand-pat". My guess that CM uses something very different. They just extend off of a single node, and hardly any full width brute force. >But most of these have to be seriously contained , or they will blow up the >qsearch. my experience is: good ideas = exploding trees. but I am going to keep trying. I am thinking of a specific null-move-qsearch, if a null-move-qsearch shows huge problems, dont allow stand pat. Scott >I hope you have tried out the idea of null-move in qsearch - I found it in the >CCC archives. Interesting idea - but was just too bad for mess. Maybe it will be >better for you. > >The only decently generic way to eliminate this problem was better evaluation - >especially due to threats like hung / pinned pieces , king safety , etc. >But in this position , esp more for position with the modification that i >mentioned - i think it will be tough to get it right. > > >Regards >Mridul
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.