Search found 1994 matches

by Michel
Sat Dec 22, 2018 11:35 am
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: UCI2WB 4.0
Replies: 82
Views: 4471

Re: UCI2WB 4.0

I now completely redesigned the master logic of UCI2WB: instead of having threads for all engine->GUI and all GUI->engine traffic, which would now and then block waiting for each other when there is a need to synchronize, it now has one of threads always listenining to the GUI, and instantly handli...
by Michel
Tue Dec 18, 2018 4:43 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: UCI pondering done right
Replies: 52
Views: 4018

Re: UCI pondering done right

I find the UCI pondering spec truly horrible. It violate both UCI key principles, to the point that it feels like a piece of Winboard grafted on UCI: (1) statelessness. The UCI protocol is not stateless: https://en.wikipedia.org/wiki/Stateless_protocol For example when the engine receives a "positi...
by Michel
Sun Dec 16, 2018 2:59 pm
Forum: Computer Chess Club: General Topics
Topic: Alphazero news
Replies: 381
Views: 42600

Re: Alphazero news

Yes, I only now realized that it would be a good check. I will do that today. Thanks Did it, interesting 1min + 1s TC (Lc0 on RTX 2070, SF10 on 4 i7 threads): Lc0 No Book vs SF10 No Book (Initial Board position): Score of lc0_v19_11261 vs SF10: 12 - 6 - 22 [0.575] 40 Elo difference: 52.51 +/- 73.05...
by Michel
Fri Dec 14, 2018 12:20 pm
Forum: Computer Chess Club: General Topics
Topic: Alphazero news
Replies: 381
Views: 42600

Re: Alphazero news

If AZ can always play into closed openings from start position no matter what the opponent does, why should its performance on open openings be reflected in its Elo rating? It is a question of philosophy. As 100% of the practical use of chess engines consists of analysis one can argue that a chess ...
by Michel
Tue Dec 11, 2018 10:27 pm
Forum: Computer Chess Club: General Topics
Topic: Alphazero news
Replies: 381
Views: 42600

Re: Alphazero news

Milos wrote:
Tue Dec 11, 2018 1:39 pm
BS. You obviously never worked in a multi billion dollar company research department. You never get outright rejection, never, not for Science not for Nature, not for any other journal.
You have no idea what you are talking about.
by Michel
Sat Dec 08, 2018 5:28 pm
Forum: Computer Chess Club: General Topics
Topic: Alphazero news
Replies: 381
Views: 42600

Re: Alphazero news

@mathew Congratulations on the achievement! But I am curious if using A/B search instead of MCTS was ever seriously tested? The rationale in the paper for using MCTS does not seem very strong. It is claimed that NN's make larger errors than traditional heuristic evaluation functions and that in A/B ...
by Michel
Thu Nov 29, 2018 12:52 am
Forum: Computer Chess Club: General Topics
Topic: Drawkiller Openings Project
Replies: 33
Views: 3888

Re: Drawkiller Openings Project

Re:"It should also be pointed out that there is nothing inherently wrong with unbalanced openings as long as one plays games in pairs with reversed colors and furthermore the pentanomial model is used to compute the variance." Consider the extreme example of an opening book that contains: 1. f3 e5 ...
by Michel
Wed Nov 28, 2018 7:58 pm
Forum: Computer Chess Club: General Topics
Topic: Drawkiller Openings Project
Replies: 33
Views: 3888

Re: Drawkiller Openings Project

It should also be pointed out that there is nothing inherently wrong with unbalanced openings as long as one plays games in pairs with reversed colors and furthermore the pentanomial model is used to compute the variance. Under the BayesElo model (using the advantage parameter to model bias) unbalan...
by Michel
Wed Nov 28, 2018 3:47 pm
Forum: Computer Chess Club: General Topics
Topic: Drawkiller Openings Project
Replies: 33
Views: 3888

Re: Drawkiller Openings Project

IMHO the correct way to compare opening books (for not too large elo differences and assuming balanced openings so that the pentanomial model is not needed) is to calculate (w-l)/sqrt(w+l) (*) where w,l are respectively the win,loss ratio. This quantity has a standard deviation of 1/sqrt(N) so one c...
by Michel
Fri Oct 26, 2018 7:49 pm
Forum: Computer Chess Club: General Topics
Topic: Chess Engines with Opening Book
Replies: 18
Views: 2052

Re: Chess Engines with Opening Book

Actually this is more or less just a gimmick, which has only disadvantages. The main problem of course is that modifying the book requires a recompile so that it can't be replaced easily. For me, the reason is that the embedded target hardware doesn't have a file system, which leaves a bytewise C a...