Author: John Merlino
Date: 11:18:28 01/24/05
Go up one level in this thread
On January 24, 2005 at 12:34:10, Wilhelm Hudetz wrote: >On January 24, 2005 at 12:20:05, John Merlino wrote: > >>On January 24, 2005 at 11:21:14, Wilhelm Hudetz wrote: >> >>>On January 24, 2005 at 09:54:04, Wilhelm Hudetz wrote: >>> >>>>Hi Jason, >>>> >>>>I think your right. >>>>I read a german translation of the manual from CM4000-Turbo, >>>>and there stands, if you set minus CM will more look to find a winning >>>>way in drawing positions. >>>>I think there is a false translation! >>>>Ok, I check this, more later. >>>>Thank you >>>> >>>>Wilhelm >>>>I better should read the english manual.... >>> >>>I make some tests with different positions >>>and cfd has big influence of the search. >>>So I think it can not only be a parameter for accept draw!? >>> >>>here a testposition as example: >>> >>>8/5BK1/8/2kn4/1p2b3/8/1R3P2/8 w - - 0 62 >>> >>>try different values for cfd and look how long it take to find Rd2. >>> >>>In many games I have controlled I coudn´t see that CM accept a >>>draw when he is in plus (C3PO has cfd=-10). >>>So I don´t know whats realy right. >>> >>>regards >>>Wilhelm >> >>Indeed, the Contempt for Draw value affects the search VERY MUCH! Johan could >>probably give a more complete description of how this works, but, in a nutshell: >> >>A low (negative) contempt for draw means that the engine will look for and >>attempt to steer the game towards drawish lines, even at the cost of improving >>its position. It will consistently attempt to get draws by repetition and will >>very likely trade material to simplify. This can result in boring play, and is >>likely to reduce the quality of play. >> >>A high (positive) contempt for draw means that the engine will attempt to avoid >>drawish positions, even at the cost of losing half a point. You can easily see >>an example of how this can reduce the quality of play by looking at the first >>game in the CM vs. Christiansen match in which the Alekhine personality was >>used. This personality has a +3 cfd value, which means that it won't accept a >>draw unless it is down by at least 3 points! >> >>The engine made some very poor moves attempting to attack and avoiding >>simplification. >> >>Johan, as far as I know, ALWAYS uses a cfd value of 0 in tournaments with The >>King. >> >>jm > >Hi John, > >thank you for explaining! >In my tests I got some better results with a little negative value. >To clear this, a value of -10 for use outside of CM-GUI is a vlue of -1 in >CM-GUI. > >Wilhelm Not enough testing has been done with "large" cfd values (greater than +1 or less than -1), so it is certainly possible that there might be some combination of settings that can benefit from them. Hopefully, you will lead the way in this testing! jm
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.