Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: When to do a null move search - an experiment

Author: Josť Carlos

Date: 15:42:03 04/27/04

Go up one level in this thread


On April 27, 2004 at 15:06:29, Gerd Isenberg wrote:

>On April 27, 2004 at 03:33:26, Bas Hamstra wrote:
>
>>About double nullmove: I tested this in some pawnendgames to see if it could
>>handle zuzwang problems, but I don't see it perform any better than normal
>>nullmove. Can Vincent or you post a position where double null outperforms
>>normal null? I agree the idea is elegant, but I just don't see it work.
>>
>>Bas.
>
>
>Hi Bas,
>
>i don't use double nullmove, but a verification search with same R, if nullmove
>fails high. Some of my favorite zugzwang positions:
>
>CTT4-11 Crafty-IsiChess:
>[D] 8/8/1p1r1k2/p1pPN1p1/P3KnP1/1P6/8/3R4 b - - ; bm Nxd5

  I recall this position from past discussions here. It's really interesting.
Going one step further shows clearly a big hole in my search. In this position:
[D]8/8/1p1r1k2/p1pPN1p1/P3KnP1/1P6/8/3R4 b - -
this is what the latest public Averno thinks:

Averno v0.70a (070)  WB2 166 MB:
 7   00:00  0.77   d1d5
 7   00:00  0.93   d1d5
 7   00:00  1.14   d1d5
 7   00:00  1.45   d1d5
 7   00:00  1.96   d1d5
 7   00:00 -0.02   d1d5 d6e6 b3b4 a5b4 a4a5 b6a5 d5c5 a5a4
 8   00:00  0.33   d1d5
 8   00:00  0.49   d1d5
 8   00:00  0.70   d1d5
 8   00:00  1.01   d1d5
 8   00:00  1.52   d1d5
 8   00:00  2.53   d1d5
 8   00:00 -0.02   d1d5 d6e6 b3b4 a5b4 a4a5 b6a5 d5c5 a5a4 c5d5
 9   00:00  0.33   d1d5
 9   00:00  0.49   d1d5
 9   00:00  0.70   d1d5
 9   00:00  1.01   d1d5
 9   00:00  1.52   d1d5
 9   00:00  2.53   d1d5
 9   00:01 -0.69   d1d5 d6e6 b3b4 a5b4 a4a5 b6a5 d5c5 a5a4 e4d4 b4b3
10   00:01 -0.34   d1d5
10   00:01 -0.18   d1d5
10   00:01  0.03   d1d5
10   00:01  0.34   d1d5
10   00:01  0.85   d1d5
10   00:01  1.86   d1d5
10   00:01 -0.72   d1d5 d6e6 b3b4 a5b4 a4a5 b6a5 d5c5 a5a4 c5a5 b4b3 a5d5
11   00:02 -0.37   d1d5
11   00:02 -0.21   d1d5
11   00:02  0.00   d1d5
11   00:02  0.31   d1d5
11   00:02  0.82   d1d5
11   00:03  1.83   d1d5
11   00:04 -0.72   d1d5 d6e6 d5d8 e6e5 e4d3 e5e7 d8d5 e7e8 d5d6 e8e6 d6d5
                   f6g6 d5d8
12   00:07 -0.37   d1d5
12   00:07 -0.21   d1d5
12   00:07  0.00   d1d5
12   00:07  0.31   d1d5
12   00:07  0.82   d1d5
12   00:07  1.83   d1d5
12   00:10 -0.81   d1d5 d6e6 d5d8 e6e5 e4d3 e5e7 d8d5 e7e8 d5f5 f6g6 f5d5
                   e8e6 d5d7 g6f6
13   00:18 -0.46   d1d5
13   00:18 -0.30   d1d5
13   00:18 -0.09   d1d5
13   00:18  0.22   d1d5
13   00:19  0.73   d1d5
13   00:19  1.74   d1d5
13   00:30 -0.81   d1d5 d6e6 d5d8 e6e5 e4d3 e5e7 d8f8 f6g7 f8b8 e7d7 d3c2
                   d7d4 b8b6 d4g4 b6c6 g4g2 c2c3 g2g3 c3c4
14   01:01 -0.46   d1d5
14   01:01 -0.30   d1d5
14   01:01 -0.09   d1d5
14   01:01  0.22   d1d5
14   01:02  0.73   d1d5
14   01:02  1.74   d1d5
14   01:29 -0.82   d1d5 d6e6 d5d8 e6e5 e4d3 f6e7 d8b8 e5e6 b8g8 e7f6 g8f8
                   f6g7 f8a8 e6d6 d3c3 g7f7 a8h8

  I use small steps to open the window at the root. Look, for example, at depth
14. It starts off failing high from previous -0.81 to -0.46; then it fails high
several times while it opens the window. But with a full open window it comes
back to -0.82!!!
  I don't have this very behaviour in my latest developement version because
I've made some improvements on search, but anyway, the background effect must be
there somehow.
  Thanks for posting this very good test position again.

  Josť C.



>[D] 4KBkr/7p/6PP/4P3/8/3P1p2/8/8 w - - ; bm g7
>
>Cheers,
>Gerd



This page took 0.01 seconds to execute

Last modified: Thu, 07 Jul 11 08:48:38 -0700

Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.