Eman 5.30

[corres]

I don't like it either.
I won't tell anyone else what to do.
Someone sent me a link, it contained just a binary.
All of my interest ended there.

No any check?
For virus, malware, worm or others?

[Dann Corbit]

I did not run it, so I did not check it. I never extracted it from the zip. There was some kind of scan because my machine has McAfee on it.
I deleted the file anyway.

[Leo]

I did not run it, so I did not check it. I never extracted it from the zip. There was some kind of scan because my machine has McAfee on it.
I deleted the file anyway.

Good move.

[Damir]

I don't like it either.
I won't tell anyone else what to do.
Someone sent me a link, it contained just a binary.
All of my interest ended there.

Hi Dann,

There is nothing wrong with the file I sent you. It is more likely your McAfee Anti Virus that is blocking it, or saying it contains suspicious content.

[Damir]

Eman engine contain no virus or any other malicious content

[corres]

I did not run it, so I did not check it. I never extracted it from the zip. There was some kind of scan because my machine has McAfee on it.
I deleted the file anyway.

O.K.

[Dann Corbit]

I don't like it either.
I won't tell anyone else what to do.
Someone sent me a link, it contained just a binary.
All of my interest ended there.

Hi Dann,

There is nothing wrong with the file I sent you. It is more likely your McAfee Anti Virus that is blocking it, or saying it contains suspicious content.

I agree that there was no virus, and in fact I never did suspect any.
I just like programs that share the code, or those that I have seen the code somehow.
Other people like Eaman a lot, but it is not my cup of tea because it comes with no software.

[Ovyron]

There is nothing special about Eman.

Here's your recipe to have Eman at home:

-Implement "Result Learning." That is, the engine should be aware of when a game ends, and what was its result (even if the GUI never tells the engine.) Save this to a file that the engine loads before the game. The engine will modify its score of positions it has already played and know its results. I don't think this code comes from its author, it was gotten from another Stockfish Derivative. It creates EXP files so if you have one that does it when "learning", you already have the code for this.

-Implement "Experience Book". If the position has been played before and the engine already had a good result in the experience file above, move instantly (use this before using a book.)

-Implement "BIN Book piling". This is being able to use multiple BIN books. The engine will check the first one, and if it runs out of moves, look into the second one, and so on. The Honey family of engines provides code for this.

-Implement "BIN Book learning". After a game is over the engine knows its result (even if the GUI didn't tell it) and modifies the contents of the bin book itself to make bad moves play less often and good moves play more often.

-Implement "Full Depth Search to depth N". To a specified by the user depth, the engine will search to this depth fully, and further depths will be searched normally.

-Implement "Full Depth Search to depth N when pondering". Use a different depth as above when it's the opponent's turn.

-Implement "Depth Pruning Threshold" and "Move Count Pruning". Presumably, this would allow the user to modify how the engine prunes in certain situations. Default is 31. Stockfish dev would already be doing this pruning, this would only allow users to change them without needing to recompile the engine.

-Implement "OutOfBook Thinkhard for N moves". The user specifies N. After the engine is out of book, for N moves pretend that you have more time than you actually do. This will make the engine have a consistent depth thorough the game, instead of the very first move out of book to be at lower depth, because the engine starts with an empty hash and has to fill it up.

And that's it. With any luck anybody implementing this will end with an open source Stockfish derivative that is better than Eman.

I abandoned Eman at version 3.72. The reason was that I managed to outperform Eman with 4CPU in my machine with CFishR with 3CPU, by just reducing the Slow Mover setting, apparently with some "E.L.T.O.N." code that helps the engine avoid blunders, though it makes it blind to some wins too. Unfortunately CFishR is yet another private derivative where the author refuses to provide the source - but you know my opinion on this, I'd rather have it like this than not at all...

[matejst]

Uly,

The features you proposed should be an imperative sine qua non in all engines, if we want them to be useful for chess. I am grateful Michael Byrne, Marco Zerbinatti, Andrea Manco, etc. tried to implement some of these options in their SF derivatives. Their good work is so often overlooked.

[carldaman]

Yes, these feature are indeed imperative, and it would be good if commercial authors, the few remaining, would focus just a little more on features in general.

I'm glad you brought it up, because one of the given reasons for the absence of such features from commercial engines is a supposed "lack of interest" among users.