Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: What is History Pruning?

Author: Vincent Diepeveen

Date: 09:57:44 07/03/05

Go up one level in this thread


On July 03, 2005 at 11:25:16, David Dahlem wrote:

>Are there any papers explaining this method? Perhaps someone could post some
>simple pseudo code.
>
>Thanks
>Dave

In general descriptions, you could see it like this:
  search the first n moves, all other moves search them to a reduced depth
  if they fail high, then research them to the full depth.

That's history pruning.

Now of course the exceptions when to ALWAYS search to the full depth or which
moves to search to the full depth.

First of all, you reduce captures always of course in search depth.
I suggest searching checks to full depth. Both check giving and all moves when
incheck. I suggest taking a maximum of n=3, not more. Fruit is using 3 there if
i saw correct.

Secondly you might not want to use this reduction algorithm when you have a PV
node. The best way is to use the Yngvi definition as he described it for his
multicut algorithm in his paper.

Results: The results are mixed. For some engine it works, for others it doesn't.

I suspect Shredder is doing something similar to this (not sure).

A problem with this algorithm is you will need to play games, and a lot, to
prove whether it works or not. We're soon speaking about hundreds of games at a
reasonable time control (30 0 will do if you ask me, if it isn't working at 30 0
it for sure won't work at 90 0. if it DOES work at 30 0 that's no garantuee it
works at 120 0 though).

I intended to try this old algorithm again for Diep as a test before world
champs 2005. However i didn't make it. Too late now.

My assumption is it doesn't work unless hard proven otherwise.




This page took 0.01 seconds to execute

Last modified: Thu, 15 Apr 21 08:11:13 -0700

Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.