Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: difficult position for most amateurs

Author: Uri Blass

Date: 03:56:42 11/30/02

Go up one level in this thread


On November 30, 2002 at 05:21:45, Johan Melin wrote:

>On November 29, 2002 at 17:22:20, Will Singleton wrote:
>
>>[D]2rr3k/1p4p1/1P2b2p/p1Bnpp1q/8/Q1P2PP1/P6P/RB3RK1 b - - 0 26
>>
>>Solution times (amd 1.6ghz):
>>
>>Yace       not in 5 min
>>Crafty     not in 5 min
>>Aristarch  3:48 min
>>Gromit     3:12 min
>>Ruffian    44 sec
>>CM9000     5 sec
>>
>>Seems like the strong amateurs are using pruning techniques that cause the
>>solution to be missed.  Ruffian and Yace have no problem, perhaps they don't use
>>null move in the way others do?
>>
>>How does Chess Tiger do on this?
>>
>>Will
>
>KnightDreamer on 733MHz PIII takes 14 seconds:
>
>  9   50  758 2494348  h8g8 b1c2 d5c3 a3c3 d8d5 c2b3 c8c5 c3e3
>                       f5f4 g3f4 (329.1k)
>Failhigh Nf4 4164545
>  9   61 1412 4735765  ++ d5f4 (335.4k)
>  9   77 1686 5665106  ++ d5f4 (336.0k)
>  9  109 1917 6423024  ++ d5f4 (335.1k)
>  9  173 2367 7929781  ++ d5f4 (335.0k)
>  9  232 3125 10500363  d5f4 h2h4 d8d2 f1f2 f4h3 g1g2 h3f2 c5f2
>                        e6d5 c3c4 (336.0k)
>CV(0.24): Nf4 gxf4? Rd2! Bf2 Qxf3! null none
>
>I think some engines have problems with the line:
>  Nf4 gxf4 Rd2 Bf2 Qxf3 null Bd5 null
>After that line most engines will hit q-search. If it can't see Qg2# in that
>q-search, it will need a large search depth.
>
>/Johan Melin

It is even worse than that

If you cannot see mates in the qsearch you may have the line
Nf4 gxf4 Rd2 Bf2 Qxf3 null Bd5 Be4 Bxe4 null

Movei can see the mate at depth 12 thanks to checks in the qsearch so that line
is not a problem for it.

The line Nf4 gxf4 Rd2 Bf2 Qxf3 null Bd5 Be4 Bxe4 is still 11 plies and when
movei call qsearch for white it cannot see the mate.


With verification search it is depth 11

The problem with verification search is that
Nf4 gxf4 Rd2 Bf2 Qxf3 Be5 Bd5 Be4 Bxe4 is 10 plies because there is negative
reduction after Qxf3.

I guess that knightdreamer has some extension rules(not check extensions or
recapture exensions) that help it to see the problem in 9 plies.

Avoiding null move pruning+checks in the qsearch does not seem to be enough to
find Nf4 at depth 9.

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.