ChessUSA.com TalkChess.com
Hosted by Your Move Chess & Games
 
 FAQFAQ   SearchSearch   MemberlistMemberlist   UsergroupsUsergroups   RegisterRegister 
 ProfileProfile   Log in to check your private messagesLog in to check your private messages   Log inLog in 

Reducing Strength
Post new topic    TalkChess.com Forum Index -> Computer Chess Club: Programming and Technical Discussions Flat
View previous topic :: View next topic  
Author Message
Robert Hyatt



Joined: 27 Feb 2006
Posts: 15866
Location: Birmingham, AL
PostPost subject: Re: Reducing Strength    Posted: Thu Mar 15, 2012 2:02 pm Reply to topic Reply with quote
kinderchocolate wrote:
Other than limiting the number of nodes and depths, any suggestions on how to reduce the playing strength as realistic as possible? I've found that limiting the number of nodes isn't working very well.

Also, what'd be the best way to tune the parameters?


I've spent a lot of time to find something that works well, as in the "skill" command in Crafty.

1. I use a skill level between 0 and 100, where 0 is bad, and 100 is optimal. In eval, there is something like this:

final score = skill * eval_score / 100 + (100 - skill) * random_score / 100;

Net effect is that skill 50 uses 50% of normal eval, 50% random eval, combined.

2. selective search parameters are reduced proportional to skill. For example, for skill 50, null-move R value is cut 50%, check extension is cut but since it is 1, it gets cut to zero for any skill reduction. LMR reduction values (upper and lower bound) are reduced in the same way.

3. A "time burner" loop in the eval is activated to slow the program down, without making it play unnaturally fast. The lower the skill setting, the larger the loop, which reduces the tactical skill at about the same rate as the positional knowledge is phased out.

Doing the above produces a pretty natural-feeling weak opponent. If you just whack the eval, you still get a tactical genius at times. If you just whack the search, you still get a positionally strong player. And a pure random eval, by itself, still plays way too strongly due to "the Beal effect". Doing the above, one can tune the program down to 1000 or less, if desired.
Back to top
View user's profile Send private message
Display posts from previous:   
Subject Author Date/Time
Reducing Strength Ted Wong Thu Mar 15, 2012 12:22 am
      Re: Reducing Strength Edmund Moshammer Thu Mar 15, 2012 2:43 am
            Re: Reducing Strength Ted Wong Thu Mar 15, 2012 6:21 am
                  Re: Reducing Strength Pawel Koziol Thu Mar 15, 2012 9:10 am
                        Re: Reducing Strength Ted Wong Thu Mar 15, 2012 10:04 am
      Re: Reducing Strength Ferdinand Mosca Thu Mar 15, 2012 10:34 am
      Re: Reducing Strength Robert Houdart Thu Mar 15, 2012 10:44 am
            Re: Reducing Strength Ed Schroder Thu Mar 15, 2012 12:17 pm
                  Re: Reducing Strength Mike Robinson Thu Mar 15, 2012 1:55 pm
      Re: Reducing Strength Robert Hyatt Thu Mar 15, 2012 2:02 pm
            Re: Reducing Strength Mike Robinson Thu Mar 15, 2012 2:40 pm
                  Re: Reducing Strength Ted Wong Thu Mar 15, 2012 4:35 pm
                  Re: Reducing Strength Edmund Moshammer Thu Mar 15, 2012 5:26 pm
            Re: Reducing Strength Jon Dart Sat Mar 17, 2012 2:53 pm
                  Re: Reducing Strength Robert Hyatt Sat Mar 17, 2012 4:07 pm
                        Re: Reducing Strength Pawel Koziol Sun Mar 25, 2012 7:36 pm
                              Re: Reducing Strength Ted Wong Mon Mar 26, 2012 12:47 am
      Re: Reducing Strength Ben-Hur Carlos Langoni Sat Mar 17, 2012 3:39 pm
Post new topic    TalkChess.com Forum Index -> Computer Chess Club: Programming and Technical Discussions

 
Jump to:  
You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot vote in polls in this forum




Powered by phpBB © 2001, 2005 phpBB Group
Enhanced with Moby Threads