Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Hashtable: is a LOWER bound more important than an UPPER bound?

Author: Ulrich Tuerke

Date: 05:30:23 03/05/01

Go up one level in this thread


Based on bit testing, I do think that a lower bound is more valuable than an
upper bound at the same depth. Perhaps, this is true, because in case of a lower
bound you have in addition to the score also a best move stored in the hash
entry. This is not possible for an upper bound (because there is no cutting
move).
So, we have more information in case of a lower bound.

Uli


On March 05, 2001 at 07:26:28, Leen Ammeraal wrote:

>I understand that it is usual to have at most
>one entry in a hashtable for a given position.
>Clearly, an entry with a higher (depth) draft
>should replace one with a lower draft,
>while a new candidate with a lower draft
>should be ignored.
>Therefore, in the rest of this discussion,
>let us assume that the draft
>of the existing entry is equal to that of
>the entry that may replace the older one.
>It seems clear to me that a tighter bound (higher
>value in the case of LOWER and a lower value
>in the case of UPPER) should have precedence.
>If a lower bound has been stored and
>the new candidate is an upper bound
>with the same score, then I replace the LOWER
>flag with the EXACT flag.
>The situation is similar with lower and upper
>exchanged in the last sentence.
>But what if these scores are unequal?
>Is a lower bound more valuable than an
>upper bound, or is it the other way round?
>Remember, we are talking about equal draft,
>unequal scores and one flag being LOWER and
>the other UPPER.
>Leen



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.