Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: What approach do you use to handle castling/en passant for repetition?

Author: Robert Hyatt

Date: 10:53:36 02/28/00

Go up one level in this thread


On February 28, 2000 at 13:51:25, Robert Hyatt wrote:

>On February 28, 2000 at 09:17:48, Tom Kerrigan wrote:
>
>>On February 28, 2000 at 08:49:06, Robert Hyatt wrote:
>>
>>>On February 28, 2000 at 03:25:09, Tom Kerrigan wrote:
>>>
>>>>On February 27, 2000 at 18:00:32, Robert Hyatt wrote:
>>>>
>>>>>I don't believe this can be safely overlooked, because the wrong repetition
>>>>>value is a tiny part of a much bigger problem, that of not correctly hashing
>>>>>positions...  the former may not happen often.  The latter happens thousands
>>>>>of times every search.  Often enough that it will _definitely_ affect the
>>>>>root score enough to make a difference.
>>>>
>>>>I would like to see some numbers to back up your argument.
>>>>
>>>>-Tom
>>>
>>>
>>>3, 47.5, 2.71828, 3.14159265358979323846264, and -.2
>>>
>>>All you have to do is run with and without EP hashed into your signature to
>>
>>Exactly. I'd like you to do this [again]. And post some test suite scores
>>comparing the two, or something.
>>
>>I know for a fact that some of the best programs don't hash this stuff, and they
>>don't seem to be hurting _nearly_ as much as you suggest.
>>
>>-Tom
>
>
>Who isn't including EP status in the hash signature?  I can't imagine _anyone_
>not doing this.  It leads to simple failures that are easy to predict.


As far as doing this again, I remove bugs, I don't knowingly add them.  This is
a really ugly thing to do wrong...  And test suites are not the way to test this
particularly well.  It takes games where you carry the hash table from move to
move.  I don't use test suites for testing anything, other than comparing NPS
when optimizing a piece of code...



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.