Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Mate to solve for good program.

Author: leonid

Date: 15:13:00 06/09/01

Go up one level in this thread


On June 09, 2001 at 12:39:22, Heiner Marxen wrote:

>On June 08, 2001 at 18:21:53, leonid wrote:
>
>>Hi!
>>
>>If you want to solve this position, it will be easy only if your program is good
>>in mates.
>>
>>[D]1r1qkq1r/Q2nqn1Q/NQ2b1QN/qqQ1bQqq/BQ4QB/q6q/3RQR2/4K3 w - -
>>
>>Please indicate your result.
>>
>>Cheers,
>>Leonid.
>
>Hello Leonid,
>
>Chest finds this one to be a mate in 10:
>
>PV: Qxd7+ Qbxd7 Bxd7+ Qdxd7 Qhxf7+ Qfxf7 Qfxf7+ Bxf7 Qxb8+ Qad8 Qxe7+ Qgxe7
>Qxd8+ Qdxd8 Rxd8+ Qxd8 Qc6+ Qd7 Nc7#
>
>No other move forces a mate in 10.  This was done on a K7/600 with 350MB hash,
>and needed 45 minutes to complete.  The effective branching factor (EBF)
>became better and better this time:
>
>depth  seconds     EBF  moves/nodes  hash-speed
>#  3      0.04s                 1kN  0.96        107-         0
>#  4      0.27s [  6.75]       12kN  1.03       1158-         0
>#  5      2.07s [  7.67]       91kN  1.20       9021-         0
>#  6     14.33s [  6.92]      632kN  1.40      63459-         0
>#  7     60.02s [  4.19]     2644kN  1.72     286055-         0
>#  8    215.44s [  3.59]     9438kN  2.11    1080774-         2
>#  9    952.95s [  4.42]    41520kN  2.39    4979356-     29851
># 10   2703.48s [  2.84]   117175kN  2.87   14318287-   5603847
>
>Cheers,
>Heiner

Hi, Heiner!

As I see your program is well before every other program in performance. Even if
we have few other mate solvers on the Web they rarely come to work, at least to
make some search by brute force.

I can't say how far is mine, from your for this position, since mine don't use
any hash. Only tendency that you found look like what I could see with my
solver. By brute force I went here only 8 moves deep. Position was solved by
selective in 10 moves. For brute force branching factor is the next:


Moves depth      Time       Branching factor

4                0.76 sec
                            6.87
5                5.659
                            5.8
6                32,85
                            5.78
7                3 min 9 sec
                            5.69
8                18 min 1 sec

Heiner, what actually version of your mate solver you use for all these
positions? It is your original Unix based solver, or solver done by Dann Corbit
after your logic? And if you use other version, that what you have on your site,
on what language it was written and when?

Cheers,
Leonid.




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.