Search found 10 matches

by Meni Rosenfeld
Thu Jan 25, 2018 5:23 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Is modern chess software lossless or lossy?
Replies: 22
Views: 5737

Re: Is modern chess software lossless or lossy?

Compared to vanilla Alpha-Beta, which only prunes irrelevant possibilities and is thus lossless. So, which is true? And is there a clear, concise, reliable reference which unambiguously answers this question? https://en.wikipedia.org/wiki/Alpha%E2%80%93beta_pruning#Heuristic_improvements "Further i...
by Meni Rosenfeld
Thu Jan 25, 2018 5:14 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Is modern chess software lossless or lossy?
Replies: 22
Views: 5737

Re: Is modern chess software lossless or lossy?

The pure Type B programs overlooked too many important moves, so Type A programs took over. See also the Chess (Program) wiki page. But then people started to add pruning and reduction heuristics like the null-move pruning and late-move reductions, which again made the search selective. Thank you! ...
by Meni Rosenfeld
Wed Jan 10, 2018 10:39 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Is modern chess software lossless or lossy?
Replies: 22
Views: 5737

Is modern chess software lossless or lossy?

Hello, I'll be happy for help in settling a dispute. In an online discussion I participated in, I mentioned the following: Modern chess software is "lossy" - it aggressively prunes possibilities that potentially could be relevant, but statistically are unlikely to be so, in order to reach such high ...
by Meni Rosenfeld
Mon Jan 25, 2016 10:23 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Best way to debug perft?
Replies: 26
Views: 6306

Re: Best way to debug perft?

In the past I've simply embedded qperft code in my engine, so that I have called the qperft function at any node, to compare with mine perft count. This was useful to find some kind of bug. There must be a thread in this forum, on the way I've converted qperft from C to C++. This solution could be ...
by Meni Rosenfeld
Mon Jan 25, 2016 3:00 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Best way to debug perft?
Replies: 26
Views: 6306

Re: Best way to debug perft?

Not sure what you mean by 'By the time I got to searching for the bad 5th move, "perft 2 -2 FEN" no longer gave me a list of results per move' I was referring to the divide feature, and my issue seems indeed specific to qperft - it does offer a divide, but for some reason divide is not working with...
by Meni Rosenfeld
Mon Jan 25, 2016 1:37 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Best way to debug perft?
Replies: 26
Views: 6306

Best way to debug perft?

An anecdote, and a question. I had a bug in my move generating function. It was supposed to handle castling by examining a king move of two horizontal spaces (only if king never moved, not in check, and the square stepped over neither blocked nor threatened); then it was supposed to register it as a...
by Meni Rosenfeld
Fri Jan 15, 2016 4:43 am
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Alpha-Beta woes, textbook-like resources, etc.
Replies: 9
Views: 2225

Re: Alpha-Beta woes, textbook-like resources, etc.

The first thing that seems wrong is your depth-2 node count. 421 means your alpha-beta procedure didn't produce a single cutoff, and that's just wrong. If you are only counting material, all nodes should have 0 evaluation and alpha-beta should visit something like 41 nodes. Hm, right. Well this par...
by Meni Rosenfeld
Fri Jan 15, 2016 1:22 am
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Alpha-Beta woes, textbook-like resources, etc.
Replies: 9
Views: 2225

Re: Alpha-Beta woes, textbook-like resources, etc.

Thanks for all the answers! Will take me a while to work through all of the information. You can measure how many positions you can visit per second, which should be of the order of a million in a modern CPU, unless you are doing something very wrong (and you probably are). A profiler can help you f...
by Meni Rosenfeld
Thu Jan 14, 2016 9:03 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Alpha-Beta woes, textbook-like resources, etc.
Replies: 9
Views: 2225

Alpha-Beta woes, textbook-like resources, etc.

Hi all, I've been interested in chess programming for a long time, but only recently started being more active about it. Both as a hobby, and with the hopes of one day making meaningful contributions to the field. My expertise is in machine learning and statistical algorithms, and I hope to be able ...