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 

Search found 16668 matches
TalkChess.com Forum Index
Author Message
  Topic: Announcing lczero
H.G.Muller

Replies: 66
Views: 13048

PostForum: Computer Chess Club: General Topics   Posted: Thu Jan 18, 2018 8:49 pm   Subject: Re: Announcing lczero
Well, it should converge to alpha-beta, not minimax. Minimax is no good.
  Topic: Announcing lczero
H.G.Muller

Replies: 66
Views: 13048

PostForum: Computer Chess Club: General Topics   Posted: Thu Jan 18, 2018 4:37 pm   Subject: Re: Announcing lczero
OK, I have given this some more thought. The basic problem is that in Chess, width is no substitute for depth. It frequently happens in the tree that most of your 30 moves look good at 4 ply, and ever ...
  Topic: On-line engine blitz tourney January
H.G.Muller

Replies: 2
Views: 171

PostForum: Computer Chess Club: Programming and Technical Discussions   Posted: Thu Jan 18, 2018 9:26 am   Subject: On-line engine blitz tourney January
The monthly on-line engine blitz tourney for January will take place on:

Saturday January 20, 3pm Boston time (21:00 Amsterdam time)

To connect:

winboard -zp -ics -icshost winboard.nl -icshel ...
  Topic: Announcing lczero
H.G.Muller

Replies: 66
Views: 13048

PostForum: Computer Chess Club: General Topics   Posted: Wed Jan 17, 2018 6:03 pm   Subject: Re: Announcing lczero
That is why I do not use averaging as a backup operator for MCTS in chess -- I return the minimax value of the best bandit instead.
Indeed, that seems an improvement, as long as the deteriorated sta ...
  Topic: Announcing lczero
H.G.Muller

Replies: 66
Views: 13048

PostForum: Computer Chess Club: General Topics   Posted: Wed Jan 17, 2018 1:23 pm   Subject: Re: Announcing lczero
I have no experience with MCTS, but since you seem to need so many nodes to see so little, I wonder if it makes proper use of beta cutoffs. E.g. in a hypothetical situation where you have the choice b ...
  Topic: Announcing lczero
H.G.Muller

Replies: 66
Views: 13048

PostForum: Computer Chess Club: General Topics   Posted: Tue Jan 16, 2018 9:39 pm   Subject: Re: Announcing lczero
In the eighties, when there was no hope to do a brute-force search of more than 4 ply, I used an extension scheme to see deeper tactics that tried to follow the lines that consisted of logical combina ...
  Topic: A new design for Xianqgi (Chinese chess) Endgame Tablebase
H.G.Muller

Replies: 2
Views: 397

PostForum: Computer Chess Club: Programming and Technical Discussions   Posted: Tue Jan 16, 2018 10:34 am   Subject: Re: A new design for Xianqgi (Chinese chess) Endgame Tableba
I am not sure what exactly you mean by "omitting all positions of Elephants of the attacking side which are not in the King's palace". In general there are 21 positions with 2 Elephants (15 ...
  Topic: Chess World to Google Deep Mind..Prove You beat Stockfish 8!
H.G.Muller

Replies: 67
Views: 6729

PostForum: Computer Chess Club: General Topics   Posted: Mon Jan 15, 2018 2:20 pm   Subject: Re: Chess World to Google Deep Mind..Prove You beat Stockfis
You could store the hash table on disk, right? Wouldn't delay you much more than emulating a TPU with an x86 CPU! Wink
  Topic: Chess World to Google Deep Mind..Prove You beat Stockfish 8!
H.G.Muller

Replies: 67
Views: 6729

PostForum: Computer Chess Club: General Topics   Posted: Mon Jan 15, 2018 9:04 am   Subject: Re: Chess World to Google Deep Mind..Prove You beat Stockfis
CCRL approximately fixes the hardware, and AlphaZero does not run on that hardware.
Ofc it does don't spread misinformation.
You can run any kind of NN with TensorFlow framework on CPU only.
It wo ...
  Topic: multi-pv analysis
H.G.Muller

Replies: 19
Views: 2120

PostForum: Computer Chess Club: Programming and Technical Discussions   Posted: Sun Jan 14, 2018 9:12 am   Subject: Re: multi-pv analysis
I find the 'best' move and get an evaluation. Then, to find the second PV, I perform a search with the first move excluded from consideration giving me another move with its associated evaluation.
T ...
 
Page 1 of 1667 Goto page 1, 2, 3 ... 1665, 1666, 1667  Next
All times are GMT
Jump to:  




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