Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Not too diffcult for computers

Author: Uri Blass

Date: 13:15:46 01/15/01

Go up one level in this thread


On January 15, 2001 at 16:06:00, David Rasmussen wrote:

>On January 15, 2001 at 10:07:42, Uri Blass wrote:
>
>>On January 15, 2001 at 08:33:46, David Rasmussen wrote:
>>
>>>On January 14, 2001 at 20:37:39, Djordje Vidanovic wrote:
>>>
>>>>On January 14, 2001 at 17:36:21, Dan Andersson wrote:
>>>>
>>>>>There is no problem. I suspect more than a few programs find Rxc6 in reasonable
>>>>>times. The first program I tried actually found it quickly, the second too,
>>>>>Crafty and Yace. Some others will find it later some sooner. Nothing tricky
>>>>>about this position at all.
>>>>>
>>>>>Regards Dan Andersson
>>>>
>>>>
>>>>Actually I believe that the position IS tough for some programs -- just look at
>>>>the commercials listed that did not find the winning move.  Of the programs that
>>>>I tested the quickest was Patzer 3.11b which found 1.Rxc6 in a second!  Also
>>>>quick were Deep Fritz (about 20 sec), Gandalf 4.32g (20 sec), Ikarus (around 25
>>>>sec), Fritz 5.32 (about 25 sec too), etc.  Some other programs could not find
>>>>the sequence at all.  The issue here is mobility, I suspect.  I had thought that
>>>>Phalanx would be fast, but after a minute passed and it had no clue, I gave it
>>>>up.
>>>>
>>>>***  Djordje
>>>
>>>I don't understand why this can be hard. My program Chezzz is definitely weaker
>>>than DF, Gandalf, Fritz 5.32, and probably some or all of the others you
>>>mention, but still, Rxc6 is the _only_ move it ever considers, even at depth=1.
>>>It never considers any other move at all. This takes 0.02 seconds on my Cel 464.
>>
>>I do not understand why your program considers Rxc6 even at depth 1 because this
>>move is losing material at depth 1.
>>I suspect that your program is not materialistic enough and can play sacrifices
>>that are not correct.
>>
>>Uri
>
>Chezzz is not very positional. It is fairly materialistic (unfortunately).
>Accumulated positional scores stay well below one pawn in most positions, and in
>this one too.
>Judge for yourself.
>
>Thinking for 11 seconds.
>time    depth   val     variation
>0.01    1/2     -2.32   32. Rxc6 dxc6 (0 kN)
>0.01    1/2     ----->  36 nodes        3600 nps        q 11%   mo 50%  h 0%
>0.02    2/3     -2.32   32. Rxc6 dxc6 (0 kN)
>0.03    2/3     ----->  176 nodes       5866 nps        q 14%   mo 93%  h 0%
>0.04    3/5     -2.32   32. Rxc6 dxc6 33. Rd3 (0 kN)
>0.06    3/5     ----->  1271 nodes      21183 nps       q 9%    mo 89%  h 0%
>0.06    4/7     -2.32   32. Rxc6 dxc6 33. Rd3 Rb4 (1 kN)
>0.10    4/7     ----->  4513 nodes      45130 nps       q 13%   mo 86%  h 6%


What is the score for Ra2 or R6d3 at depthes 1-4?

After these moves white is only a pawn down so the score should be 1 pawn for
black and not 2 pawns for black unless your program can see big positional
advantage for black and not only a pawn.

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.