Search found 769 matches

by PK
Sun Jul 07, 2019 9:31 am
Forum: Computer Chess Club: General Topics
Topic: UCI_Elo
Replies: 36
Views: 1130

Re: UCI_Elo

Granting that the wishes of UCI designers is for the UCI_Elo to be equivalent to FIDE Elo... There's still a problem with time control. Node budget approach works for a single time control, so in order to approximate FIDE rating you would need at least to tune for two time controls (let's say 2+1 a...
by PK
Sun Jul 07, 2019 7:36 am
Forum: Computer Chess Club: General Topics
Topic: UCI_Elo
Replies: 36
Views: 1130

Re: UCI_Elo

This thread shows perfectly why UCI_Elo ratings for different engines will never converge. There are several methods described already: - strict node budget - limiting nodes per second - limiting nodes per second + random noise in eval function Furthermore, a couple of other methods can be added: - ...
by PK
Fri Jul 05, 2019 6:50 am
Forum: Computer Chess Club: General Topics
Topic: Wasp 3.60 time forfeit on UCI_LimitStrength
Replies: 1
Views: 186

Re: Wasp 3.60 time forfeit on UCI_LimitStrength

I remember having similar problem with Rodent. It turned out that once nodes per second are limited, one has to check for timeout a bit more often (assuming that checking for timeout is done at constant node intervals).
by PK
Thu Jul 04, 2019 8:11 am
Forum: Computer Chess Club: General Topics
Topic: Fireworks for the Fourth
Replies: 17
Views: 612

Re: Fireworks for the Fourth

Whit kind of magic is this? You have bested OpenTal.
by PK
Mon Jul 01, 2019 10:07 am
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Exchange
Replies: 15
Views: 528

Re: Exchange

My Texel tuning also consistently gets significantly lower value for a knight in the endgame and slightly lower for a bishop. Rook and queen behave as expected. The observation is consistent across two engines, with different k factor but the same set of positions. I guess it might be fixed by incre...
by PK
Sun Jun 30, 2019 5:46 pm
Forum: Computer Chess Club: Tournaments and Matches
Topic: The Cripple And The Cucumber
Replies: 9
Views: 713

Re: The Cripple And The Cucumber

Thank you, Graham. I expected more improvement from my recent changes, but it's good to see even a little!
by PK
Tue Jun 25, 2019 8:15 am
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: How many elo gain a good Opening book have on Stockfish?
Replies: 5
Views: 726

Re: How many elo gain a good Opening book have on Stockfish?

There is an awful lot of difference between opening books. I participated once in Sedat's tournament, and Rodent's book has been badly beaten. But the real question is, what book is used against Allie - general one, or anti-Allie.
by PK
Fri Jun 21, 2019 7:40 pm
Forum: Computer Chess Club: Tournaments and Matches
Topic: CCRL scaling versus human player
Replies: 24
Views: 1431

Re: CCRL scaling versus human player

I'm no master, but I'd gladly take on an engine in 1800-2000 range.
by PK
Wed Jun 12, 2019 8:15 am
Forum: Computer Chess Club: General Topics
Topic: OKE - Opening Knowledge Engines
Replies: 94
Views: 5547

Re: OKE - Opening Knowledge Engines

Rodent's score might well be an artifact caused by the fact that it likes fianchetto and finds/completes it more often than other engines. Other opening-related bits and pieces are penalty for developing queen before minor pieces (but it is quite low and unlikely to influence the result), some code ...
by PK
Fri May 31, 2019 10:16 pm
Forum: Computer Chess Club: General Topics
Topic: Rodent III personalities <2000 Elo?
Replies: 5
Views: 937

Re: Rodent III personalities <2000 Elo?

EvalBlur is pseudo-random factor added to eval. In general, I do not use it for personalities above 1500 Elo, with one funny exception: "Drunk" personality with EvalBlur of 500 and unrestricted search. NPS to Elo ratio for Rodent will never be exact, because eval tweaks also change Elo, often in a s...