Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: a mate to solve (GLC2.15c -- 5 minutes)

Author: Tim Foden

Date: 13:10:29 10/10/01

Go up one level in this thread


On October 10, 2001 at 15:54:52, Dann Corbit wrote:

>On October 10, 2001 at 15:31:17, Dann Corbit wrote:
>
>>On October 10, 2001 at 09:23:21, Tim Foden wrote:
>>
>>>Green Light Chess (96MB Hash, 920MHz Duron, no tablebases) claims that Rg3 is a
>>>mate in 11 after 2 mins 30 secs, and that it is a mate in 10 after 4 mins 59
>>>secs.  Finally it finds a different line for mate in 10 after 7 mins 56 secs.
>>>
>>> 14   2:30 +Mate11 118131k  Rg3 2. Kc5 h3 3. b8=Q Rxb8 4. Rd1 g1=Q+ 5. Rxg1
>>>                              Rxg1 6. Kd4 h2 7. Kc4 h1=Q 8. Kd4 Rd1+ 9. Ke3
>>>                              Re8+ 10. Kf2 Rd2+ 11. Kg3 Rg8#
>>> 15   4:59 +Mate10 236044k  Rg3 2. b8=Q Rxb8 3. Kc5 h3 4. Kd6 h2 5. Ra1 g1=Q 6.
>>>                              Ra5+ Kf6 7. Ra6 Rb6+ 8. Kc7 Rxa6 9. Kd7 Rg7+ 10.
>>>                              Kd8 Ra8#
>>> 16   7:56 +Mate10 388591k  Rg3 2. Kc5 h3 3. Kc6 h2 4. Rd1 g1=Q 5. Rd5+ Ke4 6.
>>>                              Rb5 Rxb5 7. Kxb5 Qc1 8. Kb4 Qc2 9. b8=Q Rb3+ 10.
>>>                              Ka4 Qa2#
>>>
>>
>>That's a truly astonishing result!  Finding the closest mate faster than a
>>dedicated mate solver.  It seems if GLC can work itself into an advantageous
>>position, it can really exploit it.
>
>It seems that hash helps this problem a lot.  Increasing the hash lead to
>correct solution in about 4 minutes on a machine about the same speed as yours.
>It seems that you cannot increase the hash above 192 Megs.

No.  It's another thing to go on my todo list!  It is one of the easiest ones,
though.  Perhaps I should add 384MB, 768MB, and 1536MB to the options and
release version 2.15d??

>I also notice that there is no way to add in tablebase files.  The reason seems
>clear -- you don't need 'em!
>;-)

Also on my todo list :)

>>hash 300
>
>Hash table size set to: 192.0MB
>>epd 999 \m10.epd m10.out
> Line 001: Test [Rg3]...
>>
> 14  47.02      ++  39879k  Rg3     (a=-321.00 b=21.31 e=21.31)
> 14   2:07 +Mate11 114305k  Rg3 2. Kc5 h3 3. b8=Q Rxb8 4. Rd1 g1=Q+ 5. Rxg1
>                              Rxg1 6. Kd4 h2 7. Kc4 h1=Q 8. Kd4 Rd1+ 9. Ke3
>                              Re8+ 10. Kf2 Rd2+ 11. Kg3 Rg8#
> 14   2:07 +Mate11 114305k  Rg3 2. Kc5 h3 3. b8=Q Rxb8 4. Rd1 g1=Q+ 5. Rxg1
>                              Rxg1 6. Kd4 h2 7. Kc4 h1=Q 8. Kd4 Rd1+ 9. Ke3
>                              Re8+ 10. Kf2 Rd2+ 11. Kg3 Rg8#
> 15   2:09      -- 117087k  Rg3     (a=320.98 b=321.00 e=320.98)
> 15   2:34      ++ 139315k  Rg3     (a=-321.00 b=320.98 e=320.98)
> 15   4:09 +Mate10 224669k  Rg3 2. b8=Q Rxb8 3. Kc5 h3 4. Kd6 h2 5. Ra1 g1=Q 6.
>                              Ra5+ Kf6 7. Ra6 Rb6+ 8. Kc7 Rxa6 9. Kd7 Rg7+ 10.
>                              Kd8 Ra8#
> 15   4:09 +Mate10 224670k  Rg3 2. b8=Q Rxb8 3. Kc5 h3 4. Kd6 h2 5. Ra1 g1=Q 6.
>                              Ra5+ Kf6 7. Ra6 Rb6+ 8. Kc7 Rxa6 9. Kd7 Rg7+ 10.
>                              Kd8 Ra8#
> 16   4:21      -- 236288k  Rg3     (a=320.98 b=321.00 e=320.98)
> 16   5:43      ++ 307438k  Rg3     (a=-321.00 b=320.98 e=320.98)
> 16   6:31 +Mate10 363016k  Rg3 2. Kc5 h3 3. Kc6 h2 4. Rd1 g1=Q 5. Rd5+ Ke4 6.
>                              Rb5 Rxb5 7. Kxb5 Qc1 8. Kb4 Qc2 9. b8=Q Rb3+ 10.
>                              Ka4 Qa2#
> 16   6:31 +Mate10 363016k  Rg3 2. Kc5 h3 3. Kc6 h2 4. Rd1 g1=Q 5. Rd5+ Ke4 6.
>                              Rb5 Rxb5 7. Kxb5 Qc1 8. Kb4 Qc2 9. b8=Q Rb3+ 10.
>                              Ka4 Qa2#

Yep, large hash seems to help a lot for this problem.

Cheers, Tim.



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.