Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Another good underpromotion test position

Author: Tord Romstad

Date: 16:22:24 07/23/03

Go up one level in this thread


On July 23, 2003 at 18:56:29, Matthew White wrote:

>On July 23, 2003 at 16:09:50, Tord Romstad wrote:
>
>>On July 23, 2003 at 15:24:18, Matthew White wrote:
>>
>>>[D]8/5p1k/6pp/8/q2QB3/2K4P/1p3P2/8 b - - 0 45
>>>
>>>From Levin (2331) - IM Donaldson, Western States Open 2003 (analysis)
>>>The move actually played was 45... Qa1? 46. Bxg6+!! and draw by perpetual check
>>>Can your engine find the winning 45... b1=N+!! 46. Bxb1 Qa1+ and black enters
>>>the endgame a pawn up
>>>
>>>Every engine that I have tested finds it eventually. However, some do take
>>>longer than others :).
>>
>>Really?  Which programs need time to find this move?  Gothmog finds it at ply 1
>>after 0.00 seconds:
>>
>> 1      -0.62     0.00          6 a4d4 c3d4
>> 1    > -0.62     0.00         72 b2b1N
>> 1       1.74     0.00        134 b2b1N e4b1 a4a1 c3d3 a1b1 d3e3
>> 2       1.74     0.00        858 b2b1N e4b1 a4a1 c3d3 a1b1 d3e3
>> 3       2.07     0.01       1892 b2b1N e4b1 a4a1 c3d3 a1b1 d3e2
>> 4       2.07     0.02       5245 b2b1N e4b1 a4a1 c3d3 a1b1 d3e2
>> 5       2.31     0.13      25439 b2b1N e4b1 a4a1 c3d3 a1b1 d3e2 f7f5 d4d7
>>                                  h7h8
>> 6       2.12     0.24      49236 b2b1N e4b1 a4a1 c3d3 a1b1 d3e2 b1c2 e2f1
>>                                  c2c1 f1g2
>> 7       2.25     0.57     133399 b2b1N e4b1 a4a1 c3d3 a1b1 d3e2 b1h1 d4f6
>>                                  h1d5
>> 8       2.23     1.74     448236 b2b1N e4b1 a4a1 c3d3 a1b1 d3e2 b1c1 d4d2
>>                                  c1h1
>> 9       2.16     5.69    1564170 b2b1N e4b1 a4a1 c3d3 a1b1 d3e2 b1b3 h3h4
>>                                  b3c2 e2f1 c2c1 f1g2
>>10       2.12    13.98    3946148 b2b1N e4b1 a4a1 c3d3 a1b1 d3e2 b1b3 h3h4
>>                                  b3e6 e2f1 e6c6 f1g1
>>
>>Thanks for the test!  In fact I implemented underpromotions in Gothmog
>>yesterday, and now I know that it works!  :-)
>>
>>Tord
>
>Crafty 19.1 took a while on my machine at home. Something like 13 ply...

Yes, you are right. 12 plies with Crafty 19.3 on my machine.  Perhaps this
is related to checks in qsearch?  Last time I checked, Crafty didn't include
checks in its qsearch.  Gothmog does.  When I remove them, my program needs
much more time (well, it's only 0.02 seconds, but the search depth and
the number of nodes is much bigger than with checks in qsearch) to find the
correct move:

 1      -0.62     0.00          6 a4d4 c3d4
 1      -0.14     0.00         27 a4b5
 1    > -0.14     0.00         52 a4a3 c3d2
 1      -0.12     0.00         59 a4a3 c3d2
 1       0.20     0.00         62 a4a2
 2    >  0.70     0.00        135 a4a2 e4g6 f7g6
 2       0.70     0.00        239 a4a2 d4d3
 2    >  0.70     0.00        480 a4a1 e4g6 f7g6
 2       1.20     0.00        653 a4a1 c3c4
 3    >  1.70     0.01       1260 a4a1 c3c4 b2b1Q
 3       2.59     0.01       1934 a4a1 e4g6 f7g6 d4d7 h7g8 d7e8 g8h7
 4    <  2.09     0.02       3242 a4a1 e4g6 f7g6 d4d7 h7g8 d7e8 g8g7 e8e7
                                  g7g8
 4       0.23     0.02       4666 a4a1 e4g6 h7g6 d4b6 g6f5 b6b2 a1b2 c3b2
 4    >  0.23     0.02       5004 b2b1N e4b1 a4a1 c3c4 a1b1
 4       1.74     0.02       5092 b2b1N e4b1 a4a1 c3d3 a1b1 d3e3
 5       1.99     0.11      20689 b2b1N e4b1 a4a1 c3c4 a1b1 c4d5
 6       2.12     0.16      32948 b2b1N e4b1 a4a1 c3d3 a1b1 d3e2 b1c2 e2f1
                                  c2c1 f1g2
 7       2.25     0.23      50928 b2b1N e4b1 a4a1 c3d3 a1b1 d3e2 b1h1 d4f6
                                  h1d5
 8       2.23     0.49     123152 b2b1N e4b1 a4a1 c3d3 a1b1 d3e2 b1c1 d4d2
                                  c1h1
 9       2.12     1.08     309778 b2b1N e4b1 a4a1 c3d3 a1b1 d3e2 b1b5 e2e1
                                  b5f5 h3h4 f7f6
10       2.09     2.60     811155 b2b1N e4b1 a4a1 c3d3 a1b1 d3e2 b1b5 e2e1
                                  b5f5 h3h4 f5b1 e1e2 b1c2 e2e1 f7f5


Does Doctor? have checks in its qsearch?  How about BigLion, Bogo and T.rex?

Tord



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.