Author: F. Huber
Date: 09:44:12 05/13/03
Go up one level in this thread
On May 12, 2003 at 18:31:13, John Merlino wrote:
>On May 12, 2003 at 16:00:04, F. Huber wrote:
>
>>On May 12, 2003 at 14:35:21, John Merlino wrote:
>>
>>Hello John,
>>
>>>The "ana=" parameter is the one that is used in CM9000 for the "Mentor Lines"
>>>window to request a certain number of "best lines". So, "ana=3" will produce at
>>>least 3 PVs for each depth.
>>
>>many thanks for your answer - I knew you would know it. :-)
>>Seems interesting, this "ana" - if I understand this right, it should also
>>work when using TheKing outside CM9k, e.g. in Arena. (Must try it!)
>>
>>>From what I know about the sel parameter, this should work, as "sel=0" pretty
>>>much is the same thing as using the "Solve for Mate" feature in CM9000. Could
>>>you post the position so I can test it?
>>>jm
>>
>>Here´s one (of a lot of similar) examples - running TheKing under Arena
>>with "sel=0" (better: cm_parm sel=0), no difference if running as UCI
>>or WinBoard-engine:
>>
>>8/3p2N1/8/N7/1B1k1K2/8/2Bn4/8 w - - (is ´Mate in 5´)
>>
>>TheKing UCI:
>>D3 00:00 11,34 Nf5 Kd5 Bxd2 Kc5 Nb7 Kd5 Bb3 Kc6
>>D4 00:00 11,48 Nf5 Kd5 Bxd2 Kc5 Ke5 Kb6 Nb3
>>D5 00:01 11,48 Nf5 Kd5 Bxd2 Kc5 Ke5 Kb6 Nc4 Kc5 Ncd6
>>D6 00:04 12,11 Nf5 Kd5 Bxd2 Kc5 Ba4 Kb6 Bxd7 Kc7
>>D6 00:05 99,87 Bg6 Kd5 Bh7 d6 Bg6 Kd4 Ne6 Kd5 Nc7 Kd4 Nb5 Kd5 Bf7
>>D7 00:07 99,87 Bg6 Kd5 Bh7 d6 Bg6 Kd4 Ne6 Kd5 Nc7 Kd4 Nb5 Kd5 Bf7
>>D8 00:16 99,89 Bg6 Kd5 Bh7 d6 Bc2 Kd4 Ne6 Kd5 Kf5 Nf1 Bb3
>>D8 00:23 99,91 Bh7 Kd5 Ne8 Kd4 Nc7 Ne4 Bxe4 d6 Nb5
>>D9 00:24 99,91 Bh7 Kd5 Ne8 Kd4 Nc7 Ne4 Bxe4 d6 Nb5
>>
>>
>>WB-King WB1:
>>D3003 00:00 11,34 Nf5+ Kd5 Bxd2 Kc5 Nb7+ Kd5 Bb3+ Kc6
>>D4004 00:00 11,48 Nf5+ Kd5 Bxd2 Kc5 Ke5 Kb6 Nb3
>>D5005 00:01 11,48 Nf5+ Kd5 Bxd2 Kc5 Ke5 Kb6 Nc4+ Kc5 Ncd6
>>D6006 00:04 12,11 Nf5+ Kd5 Bxd2 Kc5 Ba4 Kb6 Bxd7 Kc7
>>D6006 00:06 M6 Bg6 Kd5 Bh7 d6 Bg6 Kd4 Ne6+ Kd5 Nc7+ Kd4 Nb5+ Kd5 Bf7+
>>D7007 00:07 M6 Bg6 Kd5 Bh7 d6 Bg6 Kd4 Ne6+ Kd5 Nc7+ Kd4 Nb5+ Kd5 Bf7+
>>D8008 00:17 M5 Bg6 Kd5 Bh7 d6 Bc2 Kd4 Ne6+ Kd5 Kf5 Nf1 Be4+
>>D8008 00:24 M4 Bh7 Kd5 Ne8 Kd4 Nc7 Ne4 Bxe4 d6 Nb5+
>>D9009 00:25 M4 Bh7 Kd5 Ne8 Kd4 Nc7 Ne4 Bxe4 d6 Nb5+
>>
>>In both cases TheKing first finds #7, #6 before the shortest one (#5)
>>(numbers in WB-mode are 1 too small, maybe an Arena problem?).
>>
>>As I understand ´selectivity´, TheKing shouldn´t search deeper than
>>its current search depth (with sel=0) - strange?
>>
>>But certainly you´ll have the right explanation for it ;-)
>>
>>Best regards,
>>Franz.
>
>I don't see this behavior happen with Chessmaster 9000 using the CM GUI. In both
>analysis and "solve for mate" modes, the first mate that is announced is the
>Mate in 5. In analysis mode, this takes about 6 seconds, and in the "solve for
>mate" mode it takes about one second.
>
>jm
Very strange behaviour!
I´ve tested the same position a few more times (with the same and with
other settings of "sel"), and got different results.
Sometimes immediately the shortest mate, sometimes first some longer ones,
and I can´t see any system behind it - even when I changed nothing the
next run gave sometimes another result.
Seems as there are some ´random´ conditions when analyzing a position,
although random is switched off ("rnd=0") of course.
Nevertheless thanks for your testing,
Franz.
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.