Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Root Search Question

Author: Robert Hyatt

Date: 12:53:53 12/28/99

Go up one level in this thread


On December 28, 1999 at 14:10:37, Gian-Carlo Pascutto wrote:

>On December 27, 1999 at 13:38:29, Robert Hyatt wrote:
>
>>Here is what I think is the right way to do this, as was done in Cray Blitz
>>and Belle:
>>
>>search the first move with a tight aspiration window.  Search the rest with
>>a null window.  If one fails high, mark it as "best" but don't re-search it
>>yet.  If that is the only one that fails high, it is the best.  If a second
>>move fails high, re-search one with a wider window to get a real score, then
>>search the second to see if if it fails high.  If not, the first is the best,
>>otherwise the fail-high move is the best.
>>
>>The only problem with this is that you can start a new iteration without a PV.
>>If you use internal iterative deepening, this shouldn't be a problem.  If you
>>don't, it can cause problems...
>>
>>However, many hate getting a fail high without knowing whether the move is
>>a centipawn better or a rook better.  :)
>
>Well, thats seems quite logical to me. After all, how are you going to set the
>bounds on your next interation's aspiration search if you don't have a score to
>go with ? You'll almost certainly end up with a fail-high or fail-low again.
>
>--
>GCP


Yes... although a fail high or fail low are easier to compute than a true score
of course...  but you can get into trouble.  And most of the time you get the
score anyway because you get more than one fail-high and have to resolve it.  I
didn't like it in Cray Blitz, but kept it for efficiency.  The drawback is that
if you stop on a fail-high move, you have no move to ponder.  Crafty solves that
by doing a search, but it was too complex in Cray Blitz so at times it would be
idle when the opponent was thinking, something I didn't like much.



This page took 0 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.