Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: What I did. (oops, what a pity ...)

Author: Jason Kent

Date: 16:26:31 04/19/05

Go up one level in this thread


On April 19, 2005 at 14:54:40, Graham Banks wrote:

>On April 19, 2005 at 14:21:37, Kurt Utzinger wrote:
>
>>On April 19, 2005 at 13:22:01, Graham Banks wrote:
>>
>>>On April 19, 2005 at 02:08:43, Kurt Utzinger wrote:
>>>
>>>>Hi Graham
>>>>Have you ever read my comments on CM-Settings on our homepage at
>>>>http://www.utzingerk.com/how_to_use.htm
>>>>then you would have seen the following very
>>>>important information:
>>>>
>>>>"Furthermore we know from experience that The King is quite often
>>>>not implemented in the correct way and this of course leads to
>>>>completely wrong results of matches and unnecesary discussions
>>>>about a specific CM-setting. To create separate UCI-engines
>>>>for each CM-setting defined in the Wb2Uci.eng under [EXTRAS]
>>>>should be avoided at all cost."
>>>>
>>>>With such a set-up The King will not produce the same
>>>>moves as in its orginal GUI !!
>>>>
>>>>
>>>>Method 3 (do never use this one)
>>>>
>>>>[ENGINE]
>>>>Name=The King 3.23
>>>>Author=Johan de Koning
>>>>Filename=Wb2Uci.exe
>>>>
>>>>[OPTIONS]
>>>>Program=TheKing3.23.exe
>>>>InitString=cm_parm default\ncm_parm opk=xxx
>>>>HashCommand=cm_parm tts=%i004096
>>>>;ShowThinkingMove=true
>>>>;Logfile=true
>>>>Visible=Ponder,Hash,ShowThinkingMove,Logfile,Extras
>>>>
>>>>
>>>>[EXTRAS]
>>>>Opponents Pawn (opp)=cm_parm opp=%i|spin|100|0|4500
>>>>Opponents Knight (opn)=cm_parm opn=%i|spin|100|0|1500
>>>>Opponents Bishop (opb)=cm_parm opb=%i|spin|100|0|1500
>>>>Opponents Rook (opr)=cm_parm opr=%i|spin|100|0|900
>>>>Opponents Queen (opq)=cm_parm opq=%i|spin|100|0|498
>>>>Kings Pawn (myp)=cm_parm myp=%i|spin|100|0|4500
>>>>Kings Knight (myn)=cm_parm myn=%i|spin|100|0|1500
>>>>Kings Bishop (myb)=cm_parm myb=%i|spin|100|0|1500
>>>>Kings Rook (myr)=cm_parm myr=%i|spin|100|0|900
>>>>Kings Queen (myq)=cm_parm myq=%i|spin|100|0|498
>>>>Contempt (cfd)=cm_parm cfd=%i|spin|0|-500|500
>>>>Strength (sop)=cm_parm sop=%i|spin|100|0|100
>>>>Attack Defend (avd)=cm_parm avd=%i|spin|0|-100|100
>>>>Randomness (rnd)=cm_parm rnd=%i|spin|0|0|100
>>>>Selective Search (sel)=cm_parm sel=%i|spin|9|0|16
>>>>Max Depth (md)=cm_parm md=%i|spin|999|0|999
>>>>My Centre Control (mycc)=cm_parm mycc=%i|spin|100|0|600
>>>>My Mobility (mymob)=cm_parm mymob=%i|spin|100|0|600
>>>>My King Safety (myks)=cm_parm myks=%i|spin|100|0|600
>>>>My Passed Pawn (mypp)=cm_parm mypp=%i|spin|100|0|600
>>>>My Pawn Weakness (mypw)=cm_parm mypw=%i|spin|100|0|600
>>>>Opponent's Centre Control (opcc)=cm_parm opcc=%i|spin|100|0|600
>>>>Opponent's Mobility (opmob)=cm_parm opmob=%i|spin|100|0|600
>>>>Opponent's King Safety (opks)=cm_parm opks=%i|spin|100|0|600
>>>>Opponent's Passed Pawn (oppp)=cm_parm oppp=%i|spin|100|0|600
>>>>Opponent's Weakness (oppw)=cm_parm oppw=%i|spin|100|0|600
>>>>
>>>>
>>>>Kind regards
>>>>Kurt
>>>
>>>
>>>Hi Kurt,
>>>
>>>I disagree with you on this. I've always used this method and CM9000 has
>>>produced excellent results in my tournaments as shown. I would like to see some
>>>evidence to back up your claim!
>>
>>      Hi Graham
>>      Please compare moves/evaluation in CM-GUI and Fritz-GUI
>>      of a specific setting: they are not identical if the
>>      paramaters are defined under [EXTRAS] and that's the
>>      thing I do not like. If this does not bother you, you
>>      may nevertheless have fun with CM-Settings. Personally,
>>      I always want to know what I am testing and this is
>>      not guaranteed the way you have implemented The King
>>      under Fritz-GUI. Only the methods with textfiles or the
>>      the method with settings (parameters) defined in the
>>      InitString are therefore recommanded by me. We (Manfred
>>      Meiler and myself) have tried to find out why this is
>>      so with Odd Gunnar Malin some years ago, but did not
>>      found out the reason of the not identical behaviour
>>      of the engine.
>>      Kind regards
>>      Kurt
>
>
>Hi Kurt,
>
>I will try it out and see what happens. It is interesting to note that I have
>always used your preferred method with CM10th, but CM9000 still seems to do well
>in comparison!
>
>Regards, Graham.

The point isn't that one way is better strengthwise.  The point is that your way
may get slightly different results... I personally don't think it is a very big
deal.



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.