Search found 643 matches

by Desperado
Sun Jul 12, 2020 11:27 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Questions on Razoring and Code Structure for Pruning
Replies: 23
Views: 4715

Re: Questions on Razoring and Code Structure for Pruning

The castling move is from d=1. The score returned from QS. The static score is from d=1. That there are only bad captures for black is just one more reason to check the position. It cannot be a reason to check anything, as you did not know it before checking. A normal situation would be that stand-...
by Desperado
Sun Jul 12, 2020 8:03 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Questions on Razoring and Code Structure for Pruning
Replies: 23
Views: 4715

Re: Questions on Razoring and Code Structure for Pruning

Another thing i would like to mention... ... You argue from the assumption that the in-loop futility pruning is done on the basis of the full evaluation after the move. Because that is where these positional criteria are summed up.... That is what usually happens when entering the QS, a static eval ...
by Desperado
Sun Jul 12, 2020 7:16 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Questions on Razoring and Code Structure for Pruning
Replies: 23
Views: 4715

Re: Questions on Razoring and Code Structure for Pruning

You do not need to know that! The reason why a move is strong and why you decide to search the subtree might be different. If you exclude a killer move because you expect a killer to be strong for any reason it might be because of a double attack or anything else. Of course killer move is only one ...
by Desperado
Sun Jul 12, 2020 11:59 am
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Questions on Razoring and Code Structure for Pruning
Replies: 23
Views: 4715

Re: Questions on Razoring and Code Structure for Pruning

The problem is, how do you know you know you attack two pieces?... You do not need to know that! The reason why a move is strong and why you decide to search the subtree might be different. If you exclude a killer move because you expect a killer to be strong for any reason it might be because of a...
by Desperado
Sat Jul 11, 2020 12:28 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Questions on Razoring and Code Structure for Pruning
Replies: 23
Views: 4715

Re: Questions on Razoring and Code Structure for Pruning

Well, it is very simple. It is not about different margins, it is about that quiet moves can change the static score drastically. e.g: if a pawn push from d2 to d4 attacks a Knight on c5 and that frees a bishop on c1 to attack the opponent queen on g5 ... that's it. To ignore them in the test "outs...
by Desperado
Sat Jul 11, 2020 11:55 am
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Questions on Razoring and Code Structure for Pruning
Replies: 23
Views: 4715

Re: Questions on Razoring and Code Structure for Pruning

Outside of the loop, the current state of the art is to consider a position for a test, which evaluation is so bad, that it will most likely not get above alpha, the lower bound. For this purpose, a quiescence search is performed, since the idea is that "only" capture moves can achieve a big differ...
by Desperado
Sat Jul 11, 2020 10:59 am
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Questions on Razoring and Code Structure for Pruning
Replies: 23
Views: 4715

Re: Questions on Razoring and Code Structure for Pruning

Hi, I have recently added in extended futility pruning (up to 3 plys) and am having some positive success. I am reading about Razoring, have tried laying in some flavors of code, tests are not positive. I also decided to look at some other code and now wonder about more than razoring but also code ...
by Desperado
Fri Jul 10, 2020 7:08 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Question on Texel's Tuning Method
Replies: 14
Views: 3296

Re: Question on Texel's Tuning Method

You should first convert your scores to win probabilities first so the big scores won’t dominate. It is the mean squares of the win probabilities you want to minimise. Yeah, texel tuning does this via a sigmoid function. It isn't only to get the probabilities but the absolute values will have a dif...
by Desperado
Fri Jul 10, 2020 3:03 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Question on Texel's Tuning Method
Replies: 14
Views: 3296

Re: Question on Texel's Tuning Method

You should first convert your scores to win probabilities first so the big scores won’t dominate. It is the mean squares of the win probabilities you want to minimise. Yeah, texel tuning does this via a sigmoid function. It isn't only to get the probabilities but the absolute values will have a dif...
by Desperado
Mon Dec 10, 2018 8:21 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Gradient Descent Introduction
Replies: 4
Views: 1618

Re: Gradient Descent Introduction

Thanks both of you so far. Unfortunately my time is very limited, so there might be a delay for further posts. Now,first, i would like to follow Matthew's description and i would appreciate if you can guide me through. Let's start very simple, and say each training example has an input x (let's say ...