Search found 570 matches

by petero2
Fri May 10, 2019 10:31 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Searching crazy positions
Replies: 11
Views: 464

Re: Searching crazy positions

This is absolutely normal; >99.9% of the positions conventional alpha-beta engines consider belong in this totally ridiculous category. I get completely different results with Texel. ... If we consider positions with a score >= 100cp "ridiculous", it means that about 1-0.3753 = 62.5% of the positio...
by petero2
Fri May 10, 2019 8:02 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Searching crazy positions
Replies: 11
Views: 464

Re: Searching crazy positions

When my engine searches from the starting position, I noticed that it considers board positions which are completely terrible ... Is it normal to search positions that are this unlikely to happen in a real game? Or do your engines find a reason not to go anywhere near them? It is probably likely to...
by petero2
Sat May 04, 2019 12:27 am
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Re-Pair compression questions
Replies: 11
Views: 2022

Re: Re-Pair compression questions

The decision tree is constructed by a greedy algorithm. First all possible predicates are evaluated for all positions in the tablebase, and the predicate that causes the largest reduction in entropy is chosen as the root of the decision tree. The procedure is repeated for a leaf node in the decisio...
by petero2
Sun Apr 28, 2019 10:52 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Re-Pair compression questions
Replies: 11
Views: 2022

Re: Re-Pair compression questions

For the complicated cases, such as KRNPvKQ which is the largest 6 man WDL table, the decision tree is a lot less efficient at reducing the compressed size. The KRNPvKQ table is reduced by around 30% by the decision tree prediction. Other "complicated" tables are reduced by around 30% to 50%. Do you...
by petero2
Sun Apr 21, 2019 9:03 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Re-Pair compression questions
Replies: 11
Views: 2022

Re: Re-Pair compression questions

I think it's worth to try using ANS with probabilities delivered by convolutional neural network fed with position. I tnink well-trained CNN is able to deliver almost exact evaluation which could dramatically reduce memory usage... How are you going to run an NN on all the positions in a 7-piece ta...
by petero2
Sun Mar 03, 2019 10:04 pm
Forum: Computer Chess Club: General Topics
Topic: Artificial intelligence 2019
Replies: 25
Views: 2127

Re: Please Help for fix motion in all games across movements.

Unfortunatelly I dont know how to put motions in al these games or examples here into the post. Here are the games surrounded by pgn tags to make them viewable directly in the web browser. [Event "Rated game, 5 min"] [Site "Engine Room"] [Date "2019.03.02"] [Round "?"] [White "iSoulTraiN"] [Black "...
by petero2
Sun Feb 10, 2019 9:57 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: unit testing polyglot probe code....
Replies: 7
Views: 977

Re: unit testing polyglot probe code....

I'm using perfect2017.bin as opening book. do you know some opening lines containing promotion in the early phases? do you know some opening lines containing en passant moves in the early phases? It is possible to find out by starting from the initial position and following all moves found in the b...
by petero2
Fri Feb 08, 2019 5:44 pm
Forum: Computer Chess Club: General Topics
Topic: Stockfish Cluster MPI
Replies: 16
Views: 2117

Re: Stockfish Cluster MPI

I would like to test the stockfish cluster version. https://github.com/official-stockfish/Stockfish/tree/cluster Can someone provide an .exe file? I have uploaded stockfish_cluster_win64.7z to this public dropbox folder: https://www.dropbox.com/sh/v16dye7ww33x7oe/AAAXmGY0phMkB7XlR-vaOxasa/test Stoc...
by petero2
Mon Dec 03, 2018 8:07 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: check extension
Replies: 12
Views: 1625

Re: check extension

Demolito extends checks, whose SEE >= 0. In practice, I've never seen search explosions as a result, but I wonder… Is it possible to have long sequences of checks, which involve no position repetition, and no SEE losing moves? If the checks can also be captures, it is possible to construct a positi...
by petero2
Sun Oct 28, 2018 9:30 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Technical reason why probing N-men syzygy will also probe N-1 men?
Replies: 2
Views: 653

Re: Technical reason why probing N-men syzygy will also probe N-1 men?

In a recent thread I asked why, and got the answer that Ronald, author of the Syzygy endgame tablebases, gave this explanation : The N-1-men will be probed while the N-men are getting probed because of how the compression scheme works. Does anyone know the technical details about it? It is explaine...