Computer Chess Club Archives




Subject: Re: Very easy mate to solve.

Author: Heiner Marxen

Date: 09:18:14 12/26/01

Go up one level in this thread

On December 26, 2001 at 11:23:41, leonid wrote:

>>For Chest this cannot be (logically) correct.  The first thing a mate-in-N
>>does, is to compute the mate-in-(N-1).  So the former cannot be faster.
>>That must be a timing problem/inaccuracy.
>>Do I miss something important?
>Hi, Heiner!

Hi Leonid!

>Just look into fact that with new depth procedure of search will be changed.
>Specialized files, for instance, will be at new depth. Somewhere it can bring
>different search sequence and with it even better time for higher depth.

Aah!  Yes, you have a point, here.

>I will try, by precaution, to indicate you what I know as fact and where I am
>only guessing.
>When search is done for position that contain mate in 8 (8 only as example) that
>in some visible minority of mates, mate in 9 will take less time. For sure I
>know this for selective and "complex selective" search. Anyway, for identical
>selective search, in some instances, mate in 9 can take less time that in 8.
>This I have seen repeatedly for many positions, since very often I start finding
>mate by selective in 13 moves and only later move to much lower level. This is
>how I could compare.

I see.

>My presumption:
>The same is true for brute force search.

For Chest I'm not sure, yet.  I have to investigate that.
Mostly, what Chest does, does not depend on the top level depth.
Functionally that is forced (otherwise I cannot manage a hash table,
the entries of which shall be self contained).

But some speed up techniques are global in nature.
E.g. if the total depth is less than 4, the hash table is globally disabled
by a global flag.  That can make a visable difference, yes.
This could explain a timing anomaly when there is a mate-in-3 computed as
mate-in-4 or more.

I'm not (yet) sure whether there are other such effects.
I've never thought about such effects.

>I have impression that I even saw this in the past but now, when I must say
>something as sure fact, feel me not that certain. Reason for this is that brute
>force finding is executed for my positions when I already know that mate existe
>in it. I start searching by brute force from below (4, 5, 6...) ending my search
>exactly at shortest mate.
>Disgracefully, I don't keep all data from my positions that can't indicate you
>those positions right now. Will try to keep this data in future positions.

Usually, I append any analysis results to the file containing the job.
That is why Chest's native input language knows a "logical EOF" (".."),
used to separate those analysis results from the job input.
I find it useful to be able to look into the results & statistics I had
done years ago, with older versions.

Thanks for clarifying that depth dependancy point!


This page took 0.03 seconds to execute

Last modified: Thu, 07 Jul 11 08:48:38 -0700

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