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 

Singular extensions
Post new topic    TalkChess.com Forum Index -> Computer Chess Club: Programming and Technical Discussions Flat
View previous topic :: View next topic  
Author Message
Shawn Chidester



Joined: 11 Mar 2015
Posts: 193
Location: United States

PostPost subject: Singular extensions    Posted: Sat Mar 21, 2015 11:05 pm Reply to topic Reply with quote

I found this thread started by Mr Hyatt back in 2006:

http://talkchess.com/forum/viewtopic.php?t=54281&postdays=0&postorder=asc&highlight=singular&topic_view=flat&start=0

The thread seemed to evolve into more of a discussion on best practices for determining ELO. But the main point Robert makes in the very beginning is pretty similar to what I'm experiencing: SE is a wash at best.

So today I'm trying something a little different.

Code:
if (tt_move.score >= beta) and (tt_move.depth >= depth - N)
   mark tt_move as candidate for SE
   of course, if tt_move.depth >= depth just return beta as usual

... perform normal pruning stuff ...
... perform normal search on tt_move, if it causes beta cutoff, great ...

if (tt_move is candidate) and (depth not extended) and (pc_count > 1)
  do depth+1 search on tt_move
  if (causes beta cutoff) awesome
  if (increases alpha), great, increase alpha (can only happen in PV nodes)
  if (depth+1 score is much worse than previous score)
    possible threat in this position, do depth+1 on remaining moves


I think this may be something between PV extensions and typical SE. The number of moves extended is pretty small due to the candidate pre-requisite. In my tests I'm seeing typically anywhere from 0 to a couple hundred extensions in the 10 to 15 ply search range. And only a small fraction of those cause a depth+1 search on remaining moves.

I'd appreciate any feedback on this idea. I'd also appreciate reports on success with other SE techniques.

Thanks,
STC
Back to top
View user's profile Send private message Visit poster's website
Display posts from previous:   
Subject Author Date/Time
Singular extensions Shawn Chidester Sat Mar 21, 2015 11:05 pm
      Re: Singular extensions Robert Hyatt Mon Mar 23, 2015 12:42 am
            Re: Singular extensions Shawn Chidester Tue Mar 24, 2015 6:34 am
                  Re: Singular extensions Robert Hyatt Tue Mar 24, 2015 8:47 pm
                        Re: Singular extensions Robert Hyatt Tue Mar 24, 2015 9:25 pm
                        Re: Singular extensions Shawn Chidester Thu Mar 26, 2015 3:02 am
                              Re: Singular extensions Robert Hyatt Thu Mar 26, 2015 7:27 pm
      Re: Singular extensions Peter Österlund Tue Mar 24, 2015 7:28 pm
            Re: Singular extensions Shawn Chidester Thu Mar 26, 2015 2:50 am
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