Author: Christophe Theron
Date: 10:57:16 03/06/00
Go up one level in this thread
On March 06, 2000 at 10:43:44, Wayne Lowrance wrote:
>On March 06, 2000 at 04:06:54, Christophe Theron wrote:
>
>>On March 06, 2000 at 03:22:32, robert michelena wrote:
>>
>>>I meant to say, "click on analysis brain"
>>>
>>>Sorry, I was typing too fast.
>>
>>
>>OK, so let's state it clearly. The correct settingq are:
>>
>>* enable permanent brain
>>* disable analysis brain
>>
>>
>>
>> Christophe
>
>Christophe, please tell me your thoughts on modifying contempt as suggested. Is
>there any merit in it at all, is it unique to his time control. It sounds like
>he has played a lot of games with it. But I wonder if he would have got the same
>result with default values. Thanks, I just got your program and i am impressed !
Thanks!
I have experimented with the "score for a draw" recently, and I have found that
the value -0.10 seems to be optimal for the Rebel-Tiger 12.0e engine. I have
noticed that this value seems to be harmless, and from time to time Tiger will
turn a draw into a win.
Looks like a little gain in strength can be achieved by this. Maybe 10 to 15 elo
points (of course you'll notice only if you play many games).
I think that my evaluation is too pessimistic. Very often Tiger evaluates an
equal position has a slightly negative score for him, and this can be balanced
in part by changing the "score for a draw".
I think a value of -0.30 is too risky. -0.15 looks OK too, but -0.30 is too
much.
Christophe
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.