Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: a little deeper

Author: Uri Blass

Date: 01:45:02 07/18/02

Go up one level in this thread


On July 17, 2002 at 21:49:14, Joshua Lee wrote:

>On July 17, 2002 at 18:52:47, Slater Wold wrote:
>
>>W,S - Deep Junior 7
>>1br1r1k1/1b1q1pp1/p1p2n1p/1p2n3/1P6/PNN1P3/2QBBPPP/4RRK1 b - - 0 1
>>
>>Analysis by Deep Junior 7:
>>
>>1...Neg4 2.h3 Nh2
>>  ³  (-0.36)   Depth: 3   00:00:00
>>1...Neg4!
>>  ³  (-0.66)   Depth: 6   00:00:00  3kN
>>1...Neg4!
>>  ³  (-0.66)   Depth: 6   00:00:00  3kN
>>1...Neg4 2.Nc5 Bxh2+ 3.Kh1 Qc7 4.Nxb7 Qxb7 5.f3 Nd5 6.fxg4 Nxe3 7.Bxe3
>>  =  (-0.15)   Depth: 9   00:00:00  150kN
>>1...c5 2.bxc5 Qc6 3.e4 Nc4 4.Bf3 Nxd2 5.Qxd2
>>  ³  (-0.34)   Depth: 9   00:00:00  180kN
>>1...Nc4 2.Bxc4 bxc4
>>  ³  (-0.36)   Depth: 9   00:00:00  266kN
>>1...Nc4 2.Bxc4
>>  =  (-0.06)   Depth: 12   00:00:00  775kN
>>1...c5 2.bxc5 Qc6 3.e4 Nc4 4.Bf3 Qc7 5.g3 Nxa3
>>  ³  (-0.44)   Depth: 12   00:00:00  1177kN
>>1...c5 2.bxc5 Nc4 3.Bc1 Qc6 4.e4 Nxe4 5.Nxe4 Qxe4 6.Qxe4 Bxe4 7.Bh5
>>  µ  (-0.72)   Depth: 15   00:00:04  10707kN
>>1...c5 2.Nxc5 Nf3+ 3.Bxf3 Qd6 4.g3 Bxf3 5.e4 Qe5 6.Nxa6 Nxe4 7.Re3 Red8 8.Nxb8
>>Qxb8 9.Be1 Ng5 10.Re7
>>  ³  (-0.51)   Depth: 18   00:01:17  170263kN
>>1...c5 2.Nxc5 Nf3+ 3.Bxf3 Qd6 4.g3 Bxf3 5.e4 Ng4 6.Nd5 Qe5 7.Bc3 Qh5 8.h4 Be5
>>9.Bxe5
>>  ³  (-0.61)   Depth: 20   00:04:14  539289kN
>>1...c5 2.bxc5 Nc4 3.Bc1 Qc7 4.g3 Ba7 5.a4 Bxc5 6.axb5 axb5 7.Nxb5 Qc6 8.e4 Bxf2+
>>9.Rxf2 Qxb5
>>  µ  (-0.71)   Depth: 21   00:21:03  2726068kN
>>
>>(W,  17.07.2002)
>
>While not 22 Nxc5 RxN! 23 pxR Nf3+ etc. it's clear that The Move order would
>work itself out but how about the position with Black to Play RxN! will DJ7 find
>that? I am very impressed DJ7 might be the next program i get


I think that if you want positions that programs cannot find you should not use
positions when the difference between the evaluation before the move and the
evaluation after the move is less than 0.5 pawn.

It is even better to take only positions when the difference between the
evaluation of the root position and the evaluation after the right move is at
least 2 pawns so programs canot solve it by slightly different evaluation and
really need search.

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.