Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Cute position - difficult for null-movers

Author: Uri Blass

Date: 22:29:35 06/14/02

Go up one level in this thread


On June 15, 2002 at 00:44:19, John Merlino wrote:

>On June 14, 2002 at 21:24:02, Slater Wold wrote:
>
>I'm a little confused by this one....
>
>>Analysis by Shredder Paderborn:
>...
>>1.g7 f2 2.Be7 f1R 3.gxh8Q+ Kxh8 4.e6 Kg8 5.Kd7 Rd1 6.Bd6 Rxd3 7.e7 Rf3 8.e8Q+
>>Rf8 9.Qxf8#
>>  ±  (1.28)   Depth: 14/26   00:00:02  733kN, tb=935
>
>How does a PV that ends in checkmate get an eval of only 1.28???
>
>jm

I know this behvaiour from some engines(I remember Junior and Crafty)

The reason for this behaviour is that the engine does not know that Qxf8 is a
checkmate when it evaluates it but when the gui reads the pv it calculates and
find it.

I believe that it is not a good decision not to have checkmate detection in the
evaluation function espacially for engines that use most of their time in their
evaluation(and shredder is considered to be a slow searcher).

It cannot happen to Movei.
Movei is using the number of legal moves as part of it's evaluation and in the
case that it is 0 and the king is in check Movei knows that it is a checkmate.

Uri



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.