Search found 481 matches

by jhellis3
Mon Dec 01, 2014 6:15 pm
Forum: Computer Chess Club: General Topics
Topic: Stockfish Mate Finder
Replies: 23
Views: 7150

Re: Stockfish Mate Finder

Are you able to provide a short outline of the heuristics and adjustments you made? Hi Steve. By far the most important changes are the checks for no legal king moves in a position or a small number of total legal moves (potential zugzwang). In this case, it does a full depth verification search fo...
by jhellis3
Sun Nov 30, 2014 9:39 pm
Forum: Computer Chess Club: General Topics
Topic: Stockfish Mate Finder
Replies: 23
Views: 7150

Stockfish Mate Finder

I noticed a number of people were interested in a SF that is better at finding mates. To this end, I created the following branch: https://github.com/jhellis3/Stockfish/tree/mate_finder Commit: https://github.com/jhellis3/Stockfish/commit/c6c152e94a2bb9c92e71d1bb4803b24d67876302 In my limited testin...
by jhellis3
Fri Nov 14, 2014 7:40 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Solving a fail low situation at the root
Replies: 47
Views: 10455

Re: Solving a fail low situation at the root

Yes, you misread Lucas's patch. The patch shifts the non-failing bound in the direction of the failing bound, not away from it.

EX:

SF original (-16, 16) -> (-16, 24)

SF new (-16, 16) -> (0, 24)
by jhellis3
Wed Oct 01, 2014 2:13 pm
Forum: Computer Chess Club: General Topics
Topic: Chess Engine - Gopher Check
Replies: 31
Views: 8056

Re: Chess Engine - Gopher Check

LoL, that has to be the best name for a chess engine I've seen in a while...
by jhellis3
Sat Sep 13, 2014 4:24 pm
Forum: Computer Chess Club: General Topics
Topic: Hard testposition from master Om
Replies: 20
Views: 4915

Re: Hard testposition from master Om

Ok, I created a branch which should help finding threats against the king earlier with as little ELO loss as I could manage. It also contains some cosmetic score smoothing, which Master Om previously mentioned he would appreciate. By all means, disregard this if you don't want it. Branch: https://gi...
by jhellis3
Thu Sep 11, 2014 9:07 pm
Forum: Computer Chess Club: General Topics
Topic: Hard testposition from master Om
Replies: 20
Views: 4915

Re: Hard testposition from master Om

Even without looking at the position I guessed it would either be: Rook sac, Queen sac, or subtle Queen move. SF is really good about finding minor piece sacrifices but tends to struggle with major pieces. I doubt Rxf5 is really the problem here, but rather Qxh6+ in the followup. I've tried various ...
by jhellis3
Fri Sep 05, 2014 2:43 pm
Forum: Computer Chess Club: General Topics
Topic: My Humble Engine's First Game Against Stockfish
Replies: 11
Views: 2996

Re: My Humble Engine's First Game Against Stockfish

It's hard to tell much from one game and without seeing the source, but just as a wild guess, I'd suggest try reducing Rook material value by ~8%.
by jhellis3
Tue Aug 12, 2014 10:25 pm
Forum: Computer Chess Club: General Topics
Topic: Linux/Mac engine gauntlet, cluster testing
Replies: 32
Views: 5300

Re: Linux/Mac engine gauntlet, cluster testing

One can compile the library into cutechess by adding -config static to the qmake command.
by jhellis3
Sun Jun 22, 2014 10:14 pm
Forum: Computer Chess Club: General Topics
Topic: Who is who ?
Replies: 18
Views: 3196

Re: Who is who ?

I'll guess:

W - Komodo

B - Houdini
by jhellis3
Mon Jun 09, 2014 9:17 pm
Forum: Computer Chess Club: General Topics
Topic: Stockfish 5 prefers a KBNK ending to a mate in 9 ??
Replies: 22
Views: 7235

Re: Stockfish 5 prefers a KBNK ending to a mate in 9 ??

This is fairly easily and safely fixed by: https://github.com/jhellis3/Stockfish/commit/1b270d5b5e59437ffea7b33ce7f3c96227c15ecf It is safe because opposition does not have pawns, which means there is no fortress which can avoid exchange of material leading a won endgame. The only potential danger i...