Author: Brian Richardson
Date: 11:43:33 08/02/02
Go up one level in this thread
On August 02, 2002 at 12:24:33, Uri Blass wrote: >On August 02, 2002 at 10:07:47, Brian Richardson wrote: > >>This is a pretty interesting position. As it happens, Tinker finds Rxc4 in >>about 1 second at ply 7, BUT only with a special version of the evaluation >>function that I have been testing (with little success so far) that only uses >>material and piece square values. It sticks with this move with a slowly rising >>score until at least ply 12 (+0.44). Funny thing is that Gerbil (also material >>and piece square only eval version) does not like Rxc4, so there must be >>something else going on here with extensions or something else search-related. > > >What is your piece square table for pawn at b3 or b2? > >I suspect that if you use big values there you can get Rxc4 by piece square >table evaluation but I think that using big values there is a mistake. > >of course b3 and b2 should be evaluated as more than b7 but the difference >should not be very big. > >Uri I use the values from the ICCA article some years ago that were developed from one of the TD learning experiments. pawn b3 = 120, b2 = 192
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.