Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: An idea to generate a new test suite that is not only tactics

Author: Uri Blass

Date: 12:14:33 01/15/01

Go up one level in this thread


On January 15, 2001 at 15:03:01, Pete Galati wrote:

>On January 15, 2001 at 12:55:34, Uri Blass wrote:
>
>>I suggest to analyze the positions from the first match between kasparov and
>>deeper blue.
>>
>>Here is  one position that I suggest to analyze(it is from game 5 of this match)
>>
>>[D]3r2k1/p4bp1/5q1p/8/3Npp2/1PQ5/P2R1PPP/6K1 w - - 0 1
>>
>>Deep blue(1) played g3 when I suspect(but not sure) that all top programs of
>>today converge to Ne2 when part of them need a long search.
>>
>>The following programs are the candidates
>>
>>1)Deep Fritz
>>2)Tiger13
>>3)Gambittiger
>>4)Junior6
>>5)Nimzo8
>>6)Shredder5
>>7)Gandalf4.32g
>>8)Rebel Cnetury3
>>9)Chessmaster8000
>>10)Hiarcs7.01 or 7.32
>>11)The latest Crafty
>>
>>I will be happy if at least 6 of them will be used.
>>
>>Minimal Hardware:Everything with at least 450 Mhz and 64 Mbytes of Ram.
>>
>>Uri
>
>This particular position might not be tough enough.  Dos Comet saw Ne2 at about
>3 minutes, changed it's mind in less than 8 minutes (to DB's g3), then in less
>than 19 minutes went back to Ne2 and stayed there when I quit at 50 minutes.
>
>Pete
>
>Comet-B.28    Mon Jan 15 12:57:06 2001
>
>depth	score	  sec	  nodes		  pv
> 1.	+0.01	    0	      498	 c3c5
> 2.	-0.19	    0	     1066	 a2a3  g8h7
> 3.	-0.05	    0	     4434	 a2a3  a7a5  c3c6
> 4.	-0.42	    0	    20871	 h2h3  g8h7  g1h2  f6g5
> 4.	-0.25	    0	    22837	 a2a3  a7a5  c3c5  f6a6
> 4.	-0.25	    0	    28170	 a2a3  a7a5  c3c5  f6a6
> 5.	-0.32	    0	    47524	 a2a3  f6e5  g1h1  g8h7  g2g3
> 5.	-0.26	    0	    86079	 d2d1  f6e5  c3b2  e4e3  f2e3  f4e3
> 5.	-0.16	    0	   108087	 h2h3  g8h7  c3c7  a7a5  c7b7
> 5.	-0.16	    0	   115904	 h2h3  g8h7  c3c7  a7a5  c7b7
> 6?	-0.30	    0	   126678	 h2h3  g8h7  b3b4
> 6.	-0.34	    1	   377917	 h2h3  d8e8  d4b5  f6c3  b5c3  e4e3
> 6.	-0.33	    2	   441734	 d2d1  f4f3  g2g3  f6a6  c3c7  d8d5
> 6.	-0.33	    2	   512801	 d2d1  f4f3  g2g3  f6a6  c3c7  d8d5
> 7.	-0.32	    3	   732233	 d2d1  f6e5  g1h1  g7g6  d1d2  e4e3  f2e3
> 7.	-0.32	    5	  1237271	 d2d1  f6e5  g1h1  g7g6  d1d2  e4e3  f2e3
> 8.	-0.44	    9	  2014475	 d2d1  f6e5  g1h1  d8d5  c3c8  g8h7  c8c4
> 8.	-0.39	   18	  4133180	 g2g3  g8h7  g3f4  f6f4  c3e3  f4g4  e3g3
> 8.	-0.39	   19	  4315628	 g2g3  g8h7  g3f4  f6f4  c3e3  f4g4  e3g3
> 9?	-0.53	   36	  7711136	 g2g3  f6g5  d4c6  d8d2  c3d2  f4g3  c6e7
> 9.	-0.51	   56	 12310144	 g1h1  g8h7  h1g1  f6e5  g1h1
> 9.	-0.51	   68	 15189388	 g1h1  g8h7  h1g1  f6e5  g1h1
>10.	-0.62	   86	 19479349	 g1h1  g8h7  h2h3  f6e5  d2d1  g7g6  h1g1
>10.	-0.53	  102	 23331659	 g2g3  f6g5  a2a3  g8h7  d4c6  d8d2  c3d2
>10.	-0.50	  189	 43143485	 d4e2  d8d2  c3d2  f4f3  e2d4  f3g2  g1g2
>10.	-0.50	  193	 44038893	 d4e2  d8d2  c3d2  f4f3  e2d4  f3g2  g1g2
>11?	-0.64	  212	 48498714	 d4e2  d8d2  c3d2  f6a1  d2c1  a1a2  e2f4
>11.	-0.65	  468	107805592	 d4e2  d8d2  c3d2  f6a1  e2c1  a1e5  d2d8
>11!	-0.64	  474	109003974	 g2g3  f6g5
>11.	-0.64	  474	109003974	 g2g3  f6g5
>12.	-0.75	  677	158195979	 g2g3  f6g5  h2h4  g5f6  g1h2  d8e8  d2e2
>12.	-0.52	 1129	266062669	 d4e2  d8d2  c3d2  f6a1  e2c1  a1e5  c1e2
>12.	-0.52	 1229	289062412	 d4e2  d8d2  c3d2  f6a1  e2c1  a1e5  c1e2
>13?	-0.66	 1253	294016856	 d4e2  d8d2  c3d2  f6a1  e2c1  a1e5  c1e2
>13.	-0.66	 3026	723269583	 d4e2  d8d2  c3d2  f6a1  e2c1  a1e5  c1e2
>
>==> My Move: d4e2 in 50:26

Deep Fritz also showed a similiar behaviour.

I think that this position is tough enough because some top programs(including
Deep Fritz) do not converge to Ne2 in less than 3 minutes.

I know that Hiarcs7.32 does not converge to Ne2 even after an hour and if there
is a program that I am afraid that it will not converge to Ne2 in 24 hours then
it is Hiarcs because it has also a bad branching factor.

Some analysis with Hiarcs suggest that it is probably going to converge to Ne2
but I am not sure if 24 hours are enough for it.

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.