Aspiration Windows on the root search -- Determining margin

[AndrewGrant]

I've just now got around to adding a basic Aspiration window wrapper around my rootSearch function. in pseudo code, it is something like this..

for (margin = 30; margin < 250; margin *= 2)
alpha = lastScore - margin; // lastScore is from previous draft
beta = lastScore + margin;

I'm aware that I might want to try increasing only alpha, or only beta, depending on how the last window failed.

Anyway, setting a flat margin of 30 seems silly, and arbitrary. I'm now running a set of tests where my margin is calculated as follows...

margin = abs(previousValues[depth-1] - previousValues[depth-2])
margin = margin > Tempo ? margin : Tempo

Does anyone have any thoughts on calculating aspiration windows, or perhaps any good threads to read through?

Thanks, Andrew Grant

[Ajedrecista]

Hello Andrew:

Does anyone have any thoughts on calculating aspiration windows, or perhaps any good threads to read through?

Thanks, Andrew Grant

I am not a programmer so I can not help you directly. I guess that you already know CPW (Chess Programming Wikispace). Just in case:

chessprogramming - Aspiration Windows

There you have links to lots of threads on the subject, just look at the bottom of the page. Some of them might be interesting and/or useful. Good luck!

Regards from Spain.

Ajedrecista.

[Sergei S. Markoff]

See http://talkchess.com/forum/viewtopic.php?t=60285