Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: deep blue versus diep

Author: Uri Blass

Date: 15:34:26 04/14/03

Go up one level in this thread


I admit that I did not care much about the problem because it is my design
decision not to do tranposition in the near future for different reasons but
after thinking about it I think that Omid did not give full details of the way
that he is using hash tables:


Basically search with verification is better than search without verification
so if we decide to be conservative we cannot use result of search without
verification to avoid search to the same depth with verification.

I guess that David Omid also agree that search with verification to depth X may
be in some cases even better than search without verification to depth X+1.

The question is if we have a rule never to avoid search with verification
based on result of search without verification from the same position.

A conservative solution may be never to use hash tables for pruning when
verification=true but the question is if genesis does it.


David Omid wrote:
"the depth stored in the hash table is the depth after
reduction"
It only answer the question what is the depth that is stored but we do not know
when the stored depth is used to prune the tree.

If the stored depth is used to prune the tree only in cases of search without
verifications(most of the searches at big depth) then there is no problem but
if the stored depth is used to prune the tree also when verification=true then
we may get results that are not correct.

Uri



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.