Author: Gerd Isenberg
Date: 23:28:14 03/26/03
Go up one level in this thread
On March 26, 2003 at 18:08:59, Dieter Buerssner wrote: >On March 26, 2003 at 13:45:53, Gerd Isenberg wrote: > >>After a closer look, i found an static eval bug in this PV-position. >>In pawn-endings "King in square" implementation should be rather accurate. >>Mine was not, even i had implemented the accurate routines already, with the >>result of a huge black winning bonus. >> >>[D]8/7p/2pp1p2/4k3/P7/8/6K1/8 w - - > >Interesting thought, Gerd. But don't we fail in similar situations sooner or >later. For example compare Hi Dieter, of course, but these pawn-endings may be evaluated more accurate eg. with flood fill algotithms if the king is on the critical boarder of a passers square. One could also consider possible pawn checks - and eval should trigger search in some way to look further, may be a flag or value of probability of unclearness. Regards, Gerd > >[D] 8/2pp3p/2p1k3/8/P7/8/6K1/8 w - - 0 1 > >with > >[D] 8/2pp3p/2p1k3/8/P1K5/8/8/8 w - - 0 1 > >In the first position, the black K cannot reach the a-pawn, in the second, he >can, because black can move his pawn with check. > >Regards, >Dieter
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.