Where is the source code of ShashChess engine

[corres]

Corres , Dann Corbit has been recognized as one of the best computer chess experts alive today , What have you done for computer chess other than the blah-di-blah you posted here ? I googled you & found no contributions to computer chess at all. Maybe you can direct me to where I can find your many computer chess accomplishments , I really want to learn from you. Thanks , Karger

Such a good expert as Corbit is must have abilities to find the source codes of evaluation based on Shashin` method.
If he can not find these codes may be two case:
1, Corbit is not a great expert as you supposed
2, In the ShashChess there is no any original Shashin`idea.
Capisce?
That people who do not want to give straight answer to straight question that people talks beside the point: that is he makes bla-di-bla. What is not expected from such a great Expert...
And it is absolute independent from the situation of pandemic. Supposing he is not ill.
Maybe some logic you can learn from me...

[corres]

Corres , Dann Corbit has been recognized as one of the best computer chess experts alive today , What have you done for computer chess other than the blah-di-blah you posted here ? I googled you & found no contributions to computer chess at all. Maybe you can direct me to where I can find your many computer chess accomplishments , I really want to learn from you. Thanks , Karger

Why won't he use his real name?
I thought that this was a condition of joining Talkchess.

A people is characterized by his/her thoughts and sentences and not his/her name, Mister.

[amchess]

Shashchess personalities are bootstraping properties:
the engine will approach to the position with the corresponding player personality.
After, it adapts its search and the evaluation based on it.
In this manner, for example, ShashChess doesn't have the contempt notion.
ShashChess code is embedded in different files: search.cpp, evaluate.cpp, types.h etc but you're right:
I should do a refactoring as also made by stockfish community in the past.
For build problems, you can contact me in private.
Thanks for the questions and happy to help you.
Andrea Manzo

[Dann Corbit]

My actual answer is I run them both.
Some tiny fraction of the time ShashChess will find a tactical shot even on a very quiet looking position that SF will not see.
If it is a tactical problem set (e.g. a bunch of stuff I collected from Arves) then it will find a lot of solutions that SF will miss.

Pretty much any position I analyze today I will analyze with both SF and ShashChess.
If it is tactical, I will usually analyze with SF, ShashChess and Bluiefish.
I also use other engines occasionally.
For odd material imbalance, I like Komodo.
For certain classes of difficult chess studies, I like Sting.
The only way to know which engine will find the best answer is to try it.

[corres]

...
Where are the code lines what convert the ideas of Shashin to the program code becoming Stockfish 12 to a real "Shahsin" Engine?

This was my question and I am not curious to know what engines you used to and why.

[Cornfed]

My actual answer is I run them both.
Some tiny fraction of the time ShashChess will find a tactical shot even on a very quiet looking position that SF will not see.
If it is a tactical problem set (e.g. a bunch of stuff I collected from Arves) then it will find a lot of solutions that SF will miss.

Pretty much any position I analyze today I will analyze with both SF and ShashChess.
If it is tactical, I will usually analyze with SF, ShashChess and Bluiefish.
I also use other engines occasionally.
For odd material imbalance, I like Komodo.
For certain classes of difficult chess studies, I like Sting.
The only way to know which engine will find the best answer is to try it.

Sorry, I am beginning to think you...say a lot without actually saying a lot. Please don't take that bad...lot of 'snippy posts' on this thread, some directed toward you. I don't aim to be one of them. I do find it odd that you spend time solving 'chess studies' with engines. I guess some might find that a hobby.

Curious, do you enjoy actually 'playing chess'?

[Cornfed]

Shashchess personalities are bootstraping properties:
the engine will approach to the position with the corresponding player personality.
After, it adapts its search and the evaluation based on it.
In this manner, for example, ShashChess doesn't have the contempt notion.
ShashChess code is embedded in different files: search.cpp, evaluate.cpp, types.h etc but you're right:
I should do a refactoring as also made by stockfish community in the past.
For build problems, you can contact me in private.
Thanks for the questions and happy to help you.
Andrea Manzo

That's a pretty good answer. I do wonder how NNUE works with this personality approach....the 'bootstraping'. There is part of me that regards NNUE's as vague 'opening books' for engines to latch onto while calculating so they do not have to use their 'inferior' evaluations....rather similar to endgame tablebases, but less committal or 'exact'.

[amchess]

The nnue is used to determine a great part of evaluation, so that to determine the position's nature in terms of Shashin.
To simplify at max, not considering rollercoaster effect of contempt (to 0),

• -0.25 < sent to gui score < 0.25 => Capablanca type position

• sent to gui score < -0.25 => Petrosian type position

• sent to gui score > 0.25 => Tal type position

At the beginning, we don't have this score. So, we "bootstrap" it.
In Shashin book, this takes into account also the psychological aspect and different elo rating.
This is better, imho, than only contempt concept.
For example, if you play against a lower rated player, you'll have a Tal approach: privilege risk over solidity
and so on...
Andrea

[Dann Corbit]

For build problems, you can contact me in private.
Thanks for the questions and happy to help you.
Andrea Manzo

I am able to build ShashChess now on my system.
The problem was in the use of curl.
Another software package on my system had linked incorrectly to a debug version of libcurl.
This package had installed libraries referenced by ShashChess.
Now that the problem is corrected, I can link ShashChess and (more importantly) install my own changes.
I'm in hog heaven.

[Dann Corbit]

I can build all except the 32 bit versions, due to this header file problem:

Code: Select all

g++ -Wall -Wcast-qual -fexceptions -std=c++17 -fprofile-generate -pedantic -Wextra -Wshadow -m32 -DUSE_PTHREADS -DNDEBUG -O3 -mtune=native -msse -DUSE_MMX -mmmx -flto   -c -o search.o search.cpp
In file included from C:/msys64/mingw64/include/c++/10.2.0/random:49,
                 from search.cpp:27:
C:/msys64/mingw64/include/c++/10.2.0/bits/random.h:105:26: error: expected unqualified-id before '__int128'
  105 |       { typedef unsigned __int128 type; };
      |                          ^~~~~~~~
make[2]: *** [<builtin>: search.o] Error 1

But I rarely have use for 32 bit programs. The only 32 bit machine I have is ubuntu, so I guess that I can build on that one if I decide that I need it.

Here is the version I am using now:

My binaries are in the source folder and not the root folder.
If anyone wants to try my version, you will want to turn the UCI logging option off.
I use that to generate human readable analysis files, which I subsequently parse to write stored procedures to insert the data into my database.
Since you do not have my database or the parsing routines, it will not be useful for you.