Add Chest to the elite software chess engines (good or bad?)

[jplchess]

Since Komodo has an attacking style with better results, why not have Chest use and combine with the elite chess engines?

Dpes anyone think maximum efficiency in software is the way to go here?

Jonathan Lee

[Frank Quisinsky]

Hi Johanthan,

never I understand it if I asked programmers for a attacking style modul in the past. Different try to explain but again, never I have understand it as non programmer. It must be possible that engines can unsing moduls, we can switched on or off with an UCI option. In the past TheKing comes with special knowledge about mate search. Komodo is clearly improved around since the TECE version.

With Chest we have such a program with knowledge. We have the Stockfish development and can try out such special mate knowledge in Stockfish.

Best
Frank

[hgm]

Chest is not a Chess player, but a mate finder, right? So what exactly do you propose? That instead of thinking what is the best move to do, a top engine should in stead waste time on looking for a mate (which of course almost never would be there to find), and only after that spent whatever time is left on finding the best move? That would surely cost it a couple of hundred Elo...

[Frank Quisinsky]

Hi,

a second search switched on or off by engine automacitally if a attack way is possible. Must be a very intelligent programming. I know that the programmer of Gandalf made some experiments with a second search if I understand him in the time of development.

If a program find out, mate could be possible (open king) the second search will get 10% or 20% calculating time with Special knowledge about mate search. Position is more open, mate is more possible more calculating time.

In sources a special modul for the second search could be added without to changes the main sources. I am sure that the program lost ELO yes, but such an Option (on / off as UCI option) could be very interesting for analyzes.

Try to explain it in bad English as non programmer.

Since more as 20 years I asked different programmers for it, but it seems it's not possible what I have in my brain.

Best
Frank

[Dirt]

If a program find out, mate could be possible (open king) the second search will get 10% or 20% calculating time with Special knowledge about mate search. Position is more open, mate is more possible more calculating time.

That is quite difficult to do. On the other hand, each additional core in the traditional search gives you less benefit, so at some point (64?) it might make sense.

[Matthias Gemuh]

Hi,

a second search switched on or off by engine automacitally if a attack way is possible. Must be a very intelligent programming. I know that the programmer of Gandalf made some experiments with a second search if I understand him in the time of development.

If a program find out, mate could be possible (open king) the second search will get 10% or 20% calculating time with Special knowledge about mate search. Position is more open, mate is more possible more calculating time.

In sources a special modul for the second search could be added without to changes the main sources. I am sure that the program lost ELO yes, but such an Option (on / off as UCI option) could be very interesting for analyzes.

Try to explain it in bad English as non programmer.

Since more as 20 years I asked different programmers for it, but it seems it's not possible what I have in my brain.

Best
Frank

Speculative mate searches only waste engine time.
Singular extensions are the best way to go for some engines, depending on their search. TheKing seems to use them very successfully.

[bob]

Since Komodo has an attacking style with better results, why not have Chest use and combine with the elite chess engines?

Dpes anyone think maximum efficiency in software is the way to go here?

Jonathan Lee

Searching for a tactical solution, particularly mate, is a far different thing from trying to play the best move in any position. That's why win at chess type positions are easier for humans, because they already know there is a tactical threat, and that the key move is generally unexpected as opposed to being obvious. But in normal positions, this type of strategy doesn't work well at all...

[Frank Quisinsky]

Hi Bob,

thats whats different other programmers tell me in the latest years! This explantion isn't to accept.

Explain: Why different engines have so many short win games, can see exactly the complete and right win combination without mistakes in mate way (mate 15, mate 14, mate 13 up to mate 1). Others don't find the right mate way, don't see such combinations.

Good examples are Spark, Junior ...

A second and intelligent search, starting by engines itself can help. Such knowledge can be added as modul. I know that's very difficult and not esay ... but nothing is impossible.

I saw thousands of comp-comp games in all the years. All of this games up to mate, without to use resing=on option. Do you know what is really boring in computer chess ... if engine can't set mate and believe me the most have here very big problems. Give us ... mate in 17, mate in 22, mate in 15, mate in 20 ... and so on.

That is really bad!

Do you know what is really boring ...
If engine are thinking for mate in one move with time control 40 moves in 10 minutes repeaded ... 1 minute.

That are in my opinion mistakes in a good programming. A goold example is the Fruit clone Rykba.

I believe that endgame databases are not good for chess programming. Most lost interest for programming a clean endgame! A second and special search can do and can help.

Tell me ...
Why TheKing have it for so many years!
What is the secret?

That must be important in chess programming today. I think programmers should working on a clean programming for endgames (without to say ... I have endgames databases and nothing more is to do). But the most don't used resign = off and cant see the problems I think.

Best
Frank

One important topic for eng-eng matches!
The next topic is a clear ponder = on programming. Time Management for so many engines with ponder = on is bad, different engines can't set mate and crashes with ponder = on. That is the reason for me to use at the moment ponder = off.

Two bad points from my view in programming engines I dislike so many years. We have 2015 and such problems should be history.