Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: would someone test this position with Tiger15 and latest crafty version

Author: Ulrich Tuerke

Date: 05:23:32 11/08/02

Go up one level in this thread


On November 08, 2002 at 06:16:39, Uri Blass wrote:

>On November 08, 2002 at 03:59:49, Ulrich Tuerke wrote:
>
>>On November 07, 2002 at 17:09:58, Gerd Isenberg wrote:
>>
>>>On November 07, 2002 at 12:09:49, Ulrich Tuerke wrote:
>>>
>>>>The DOS-Comet (on AMD/700) selects Bxg6 immediately and sticks to it although
>>>>the score keeps on oscillating a bit:
>>>>
>>>>Comet-B.50    Thu Nov  7 18:01:04 2002
>>>>
>>>>depth	score	  sec	  nodes		  pv
>>>> 1.	+2.29	    0	      920	 c2c6
>>>> 2.	+1.85	    0	     1495	 c2c6  e8d6
>>>> 3.	+2.53	    0	     3989	 c2c6  e8d6  h5g4
>>>> 4.	+2.27	    0	    16545	 h5g6  h7g6  c2g6  e8g7
>>>> 4.	+2.27	    0	    18567	 h5g6  h7g6  c2g6  e8g7
>>>> 5.	+2.81	    0	    53927	 h5g6  h7g6  c2g6  e8g7  a1c1  a8c8  c1c2
>>>> 5.	+2.81	    0	    56939	 h5g6  h7g6  c2g6  e8g7  a1c1  a8c8  c1c2
>>>> 6?	+2.67	    1	   372685	 h5g6  h7g6  c2g6  e8g7  g6g7  e7g7  g5g6
>>>> 6?	+2.40	    3	   625143	 h5g6  h7g6  c2g6  e8g7  g6g7  e7g7  a1c1
>>>> 6.	+2.39	    3	   687663	 h5g6  h7g6  c2g6  e8g7  a1c1  e7d6
>>>> 6.	+2.39	    3	   698427	 h5g6  h7g6  c2g6  e8g7  a1c1  e7d6
>>>> 7?	+2.25	    4	   796132	 h5g6  h7g6  c2g6  e8g7  a1c1
>>>> 7?	+1.98	    5	  1015341	 h5g6  h7g6  c2g6  e8g7  a1c1
>>>> 7!	+1.98	    7	  1211842	 h5g6  h7g6  c2g6  e8g7  g6h7  g8f7  h1h6
>>>> 7.	+1.98	    7	  1211842	 h5g6  h7g6  c2g6  e8g7  g6h7  g8f7  h1h6
>>>> 8!	+2.11	    7	  1285067	 h5g6  h7g6  c2g6  e8g7  g6h7  g8f7  h1h6
>>>> 8?	+2.11	   11	  2002215	 h5g6  h7g6  c2g6  e8g7  g6h7  g8f7  h1h6
>>>> 8.	+2.11	   14	  2362916	 h5g6  h7g6  c2g6  e8g7  g6h7  g8f7  h1h6
>>>> 9?	+1.97	   33	  6194490	 h5g6  h7g6  c2g6  e8g7  g6h7  g8f7  h1h6
>>>> 9!	+1.97	   43	  7604098	 h5g6  h7g6  c2g6  e8g7  a1c1  f8c8  g6h7
>>>> 9.	+1.97	   43	  7604098	 h5g6  h7g6  c2g6  e8g7  a1c1  f8c8  g6h7
>>>>10!	+2.10	   47	  8511407	 h5g6  h7g6  c2g6  e8g7  a1c1  f8c8  g6h7
>>>>10?	+2.10	  134	 27955236	 h5g6  h7g6  c2g6  e8g7  a1c1  f8c8  g6h7
>>>>10.	+2.10	  148	 30110201	 h5g6  h7g6  c2g6  e8g7  a1c1  f8c8  g6h7
>>>>11.	+2.10	  200	 40603934	 h5g6  h7g6  c2g6  e8g7  a1c1  f8c8  g6h7
>>>>
>>>>==> My Move: h5g6 in 3:20
>>>
>>>Hi Uli,
>>>
>>>Interesting position, thanks kburcham. Nice for tuning king safety.
>>>
>>>IsiChessMMX found it instantly too, at Depth 0 with +2 score!
>>>Guess blacks king safety penalty after Bxg6 hxg Qxg6+ Ng7 is a bit too huge or
>>>speculative (new paradigm :-)
>>>Open h-file with is rather expensive here, two heavy pieces controlling adjacent
>>>king squares, the white queen near the black king, a pinned knight, domination
>>>of an adjacent king square (h7), the ability to castle queenside, the g5 passer
>>>(pawnstorm)...
>>>
>>>Oscillating like Comet from +2.48 at depth 5 via +1.75 at 9 and +2.04 at 11.
>>>
>>>Regards,
>>>Gerd
>>
>>Hi Gerd,
>>
>>agreed. Finding this move is purely a matter of static evaluation. Search and
>>tactical extensions can't find a quick win here, it seems.
>>
>>See you in PB (hopefully),
>>Uli
>
>I disagree
>
>I believe that tactical extensions can find that
>Bxg6 hxg6 Qxg6+ Ng7 Rh7 is good for white.
>
>Movei says +0.29 for white at depth 12 for Bg4
>
>depth=12
>+0.29 h5g4 e8d6 c2d3 a8c8 a1c1 c6c5 d4c5
>b6c5 d3c3 d5d4 e3d4 c5a3
>
>After Bxg6 it shows:
>
>depth=8 +0.70 h7g6 c2g6 e8g7 g6h7 g8f7 a3a4 f8g8 b2a3 c6c5 g5g6 f7f8 h7h6
>depth=9 +0.65 h7g6 c2g6 e8g7 a1c1 f8c8 g6h7 g8f7 g5g6 f7e8 h7h8 d7f8 e3e4 a5a4
>b3a4 a8a4
>depth=10 +0.29 h7g6 c2g6 e8g7 a3a4 f8f5 g6h7 g8f7 b2a3 e7e8 a1c1 a8c8 g5g6 f7f6
>e3e4
>depth=11 +0.56 h7g6 c2g6 e8g7 a3a4 f8f5 b2a3 c6c5 g6h6 g8f7 e3e4 d5e4 d2e4
>depth=12 +0.44 h7g6 c2g6 e8g7 a3a4 f8c8 b2a3 c6c5 g6h7 g8f7 g5g6 f7e8 h7h8 e7f8
>a1c1 f8h8 h1h8 e8e7 h8h7 e7f6
>depth=13 +0.44 h7g6 c2g6 e8g7 a3a4 f8c8 b2a3 c6c5 d2f3 a8b8 g6h6 g8f7 g5g6 f7e8
>h6h8 e7f8 h8f8 d7f8 d4c5 b6c5 a3c5 c8c5
>
>After Bxg6 hxg6 is shows
>
>depth=8 +0.75 e8g7 g6h7 g8f7 a3a4 a8c8 b2a3 c6c5 a1c1 f7e8 h7g6 e8d8 a3c5 d7c5
>d4c5 b6c5 c1c5 e7c5 g6g7
>depth=9 +0.29 e8g7 a3a4 f8f5 g6h7 g8f7 b2a3 e7e8 a1c1 a8c8 g5g6 f7f6 e3e4
>depth=10 +0.56 e8g7 a3a4 f8f5 b2a3 c6c5 g6h6 g8f7 e3e4 d5e4 d2e4
>depth=11 +0.44 e8g7 a3a4 f8c8 b2a3 c6c5 g6h7 g8f7 g5g6 f7e8 h7h8 e7f8 a1c1 f8h8
>h1h8 e8e7 h8h7 e7f6
>depth=12 +0.44 e8g7 a3a4 f8c8 b2a3 c6c5 d2f3 a8b8 g6h6 g8f7 g5g6 f7e8 h6h8 e7f8
>h8f8 d7f8 d4c5 b6c5 a3c5 c8c5
>
>After Ng7 Rh7 it shows
>
>depth=7 +0.80 a8a7 a1c1 a7c7 h7h2 c7c8 g6h7 g8f7 e3e4
>Nodes: 875216 NPS: 129469
>Time: 00:00:06.76
>depth=8 -0.90 a8c8 e1e2 d7f6 g5f6 f8f6 g6h5 g7h5 h7e7
>Nodes: 6906157 NPS: 130823
>Time: 00:00:52.79
>depth=9 -2.07 a8c8 g6h6 b6d8 g5g6 e7h4 h6h4 d8h4 h7h4 f8f6 h4h6 g7f5 h6h7
>Nodes: 11746058 NPS: 131756
>Time: 00:01:29.15
>depth=9 -1.36 a5a4 g6h6 b6d8 g5g6 e7h4 h6h4 d8h4 h7h4 a4b3 e3e4 d7f6 d2b3 f6e4
>Nodes: 15916290 NPS: 131833
>Time: 00:02:00.73
>depth=9 -1.04 d7f6 g5f6 e7f6 g6f6 f8f6 h7h1 g7f5 a1c1 a8c8 h1g1 g8h8
>Nodes: 17558847 NPS: 131991
>Time: 00:02:13.03
>depth=9 -1.03 f8f5 g6h6 d7f8 g5g6 f8g6 h6g6 f5f6 g6c2 a8e8
>Nodes: 21498826 NPS: 131918
>Time: 00:02:42.97
>depth=10 -1.07 f8f5 g6h6 d7f8 g5g6 f8g6 h6g6 e7f6 g6f6 f5f6 h7h1
>Nodes: 24662748 NPS: 132183
>Time: 00:03:06.58
>
>I will need to check what happened to the line
>Ng7 Rh7 in the previous analysis
>
>I guess that it got lost in some bug.
>You can see that movei can see tactical problems
>at depth 8 after Ng7 Rh7
>
>It means that programs with the right extensions
>can see the tactical win for Bxg6 in a short time.

So, you have manually inputted 6 half moves; the subsequent search returned a
score of +1.07 for white.

Neither do I think that 1.07 is a convincing win score, nor do I see how to
extend the search in order to close the gap of 6 half moves quickly.

Uli

>
>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.