New Installment of my Project on Chess Evaluation

[Lyudmil Tsvetkov]

Hi there!
2 years ago I released a small project on chess evaluation,
the LittleChessEvaluationCompendium, still available here: http://www.winboardengines.de/doc/Littl ... endium.pdf
Now, since I have come up with some new, and possibly not uninteresting ideas, I have decided to release a small addendum, available for download at: http://www.winboardengines.de/doc/addendumlcec_2012.pdf
New things in the addendum:
- x-ray attacks
- complementarity (optimal spread)
- split ratio for control of black and white squares
- groups of pawns; scoring of groups of pawns
- more on double, isolated and backward pawns
- values for pawns on files and ranks (apart from space advantage)
There are also some suggestions on pruning, but since I am not a programmer, I do not know if they are sensible at all.

Best regards,
Ludmil

[mcostalba]

Thanks Ludmil, interesting stuff. You may also want to check what is already normally done:

http://chessprogramming.wikispaces.com/Evaluation

and you will see many of your ideas are what is already commonly used in the engines. Neverthless there are some new hints in your paper that deserve some test...and this is what I am going to do.

I also suggest to (re) post in the programmers subforum where it seems more appropiate.

A quick comment I would like to add is that for engine evaluation are more powerful terms that weights strategic/long term advantage than terms that weight tactical shots of 1-2 plies deep, becuase the latter ones are normally already take in consideration by the search, for instance pawn structure is a long term advantage while x-ray is a shallow tactical shot.

[lucasart]

A quick comment I would like to add is that for engine evaluation are more powerful terms that weights strategic/long term advantage than terms that weight tactical shots of 1-2 plies deep, becuase the latter ones are normally already take in consideration by the search, for instance pawn structure is a long term advantage while x-ray is a shallow tactical shot.

+1

[Lyudmil Tsvetkov]

Thanks, Marco!
I will try to repost in the programmers section,
but I am afraid I might become a laughing stock.
Many thanks for your wiki link - when I have more time I will try to read carefully everything.
I am basically a chess player, and maybe many suggestions are not feasible, sorry for that.
I am flattered by the fact you will try some ideas in Stockfish;
by the way, it is one of my favourable engines.
Sometimes I try my hand on it on a Dell xps laptop,
and it is very difficult to score a win, I might score 1 win in 20 games,
but at the same time I am able to draw almost half of the games.

Many thanks for your reply,
Best regards,
Ludmil

[Lyudmil Tsvetkov]

And I forgot to mention that many thanks are due to Thomas Mayer,
the programmer of Quark, for kindly hosting the files.

[mcostalba]

I am basically a chess player, and maybe many suggestions are not feasible, sorry for that.

Most of your suggestion are easily feasible. I am impressed that a non-programmer is able to come up with such a detailed quantitative explanation of his ideas: this is exactly what programmers need to know to implement a concept in software. But it is not a natural and human way to express ideas so it is amazing that a non-programmer wrote a specification like that. I'd say you have a good attitude at programming, if you decide to learn it my guess is that you'd do it fast.

P.S: 1 win out of 20 games against Stockfish at skill level 20 is a huge result ! I guess you are a very good chess player.

[Mincho Georgiev]

И предния път го казах, поздравявам те за добрата работа и продължавай все така!
Най-добри пожелания!

[JuLieN]

[moderation]
Simple relocation of this topic where it belongs: the programming sub-forum, where it will reach its natural audience, and will do it more broadly.
I left a shadow topic in the main forum so people can follow its track.
[/moderation]

[Kempelen]

I usually like to look into this link for new ideas and concepts to experiment with. Not all are mathematically implementable, but there are lot of them:
http://danheisman.home.comcast.net/~dan ... yings.html

[Joerg Oster]

Thank you for sharing. Quite interesting.