Search found 19371 matches

by hgm
Wed Aug 08, 2007 12:21 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: multithreading questions
Replies: 49
Views: 26397

Re: multithreading questions

Well, the idea of lockless hashing is very simple, and based on the philosophy that you don't care so much if a hash entry gets lost, as long as it produces a recognizably invalid hash entry rather than an erroneous one that might send your engine off track. To this end you XOR all independently sto...
by hgm
Tue Aug 07, 2007 4:13 pm
Forum: Computer Chess Club: General Topics
Topic: possible positions
Replies: 5
Views: 2176

Re: possible positions

H.G. if I'm reading you correctly the number of possible positions is comparable to chess but significantly fewer (by perhaps 100 times). let me know if this is roughly the case. Well, I never calculated this number for Chess, but it seems intuitively correct. The total number of pieces is equal. T...
by hgm
Tue Aug 07, 2007 11:51 am
Forum: Computer Chess Club: General Topics
Topic: possible positions
Replies: 5
Views: 2176

Re: possible positions

In Xiangqi it is not so easy to construct positions like that, because the line of Pawns is moved two ranks forwards compared to FIDE Chess, and already has a number of holes in it. (There is only a Pawn in every other file.) I, at least, could not think of one; every other piece seems to be alble t...
by hgm
Tue Aug 07, 2007 11:09 am
Forum: Computer Chess Club: General Topics
Topic: possible positions
Replies: 5
Views: 2176

Re: possible positions

According to http://en.wikipedia.org/wiki/Game-tree_complexity the state-space comlplexity would be 10^48. Actually for Xiangqi the exact number is reasonably easy to calculate (compared to FIDE Chess) as in-check positions are not necessarily illegal (in Xiangqi one has to capture the King), and th...
by hgm
Tue Aug 07, 2007 7:06 am
Forum: Computer Chess Club: General Topics
Topic: Trusting chess engines
Replies: 8
Views: 4718

Re: Trusting chess engines

Note that the given position is not about mate, but bout winning a rook. Most engines do implement an ad-hoc solution for preferring fast mates (making the evaluation of a checkmate position dependent on its level in the tree). That engines do score fast wins equal to slow wins des of course not mea...
by hgm
Thu Aug 02, 2007 1:52 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: tactical play or positional play for chess engine
Replies: 21
Views: 26909

Re: The Art of Evaluation (long)

Great post. I fell exactly into the trap with the open-line bonus you describe, in the early development of Joker. :oops: I increased the open-line bonus, and suddenly it started hopelessly losing many games by unprovoked exchanging of 2 minor pieces for R+P. Joker also suffers from discontinuity, i...
by hgm
Thu Aug 02, 2007 12:44 pm
Forum: Computer Chess Club: General Topics
Topic: Trusting chess engines
Replies: 8
Views: 4718

Re: Trusting chess engines

Because most engines are not programmed to prefer quick wins over slow wins. They use the minimax algorithm, which assigns scores to the moves only based on the final position the will lead to (against best opponent play), and ignore completely how you get there. If it were not for the explicit prog...
by hgm
Thu Aug 02, 2007 11:37 am
Forum: Computer Chess Club: General Topics
Topic: Regarding Strekla and CCT10+...
Replies: 18
Views: 10003

Re: Regarding Strekla and CCT10+...

Well, I am an author. And I probably would not participate in events were authors were allowed to 'vote out' programs of other authors.

I prefer justice over democracy...
by hgm
Thu Aug 02, 2007 11:04 am
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Mate handling with transposition tables
Replies: 7
Views: 3894

Re: Mate handling with transposition tables

It is clear that Joker 1.1.07 did not use one of these idea otherwise it could avoid repetition draw against Neurosis 2.3 in round 14 of the 4th CCRL Amateur Championship Division 5. OK, I looked at the Neurosis-Joker game, and there is indeed something fishy there. From the zero score of the move ...
by hgm
Wed Aug 01, 2007 10:48 pm
Forum: Computer Chess Club: General Topics
Topic: Joker 2.0 memory useage
Replies: 5
Views: 2457

Re: Joker 2.0 memory useage

Yes, NEG typically strips Pos bare. This is actually a game where Pos last unusually long, and which NEG probably would not have won without adjudication. As NEG has no clue what checkmate is, it is unlikely to win KQK, and needs at least 2 or 3 queens to have a good chance on an accidental checkmat...