Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: About random numbers and hashing

Author: Sven Reichard

Date: 06:40:40 12/10/01

Go up one level in this thread


On December 09, 2001 at 06:57:28, David Rasmussen wrote:

>On December 08, 2001 at 12:25:59, Sven Reichard wrote:
>
>>b) We especially don't want that to happen close to the root.
>
>Well, sure, but for me it is unacceptable to have it happening at all, if it
>means a collision (which it might not necesarily).

You can't avoid collisions once you search more nodes than your ttable has
entries.

>>f) The Hamming distance does depend on the choice of the vector space base.
>
>I disagree, but maybe I don't understand what you mean. Hamming distance does
>not depend on the choice of the vector space base.

See the example I gave before (in 2 dimensions). The Hamming distances of a set
of vectors changes if you apply a linear transformation.

>>g) Given a set of vectors with minimal weight and minimal distance 17, there is
>>a linear transformation moving it to a set of minimal weight 1 and minimal
>>distance 2 (which in terms of linear independence is just as good as the one we
>>started with).
>
>You lost me there.

Give me a set of vectors. I pick 64 linearly independent elements from that set
(provided that the set spans the whole space) and pick those as the basis
vectors. I then express every vector in the set in terms of that basis. That is,
the 64 vectors I picked are represented as 0-1 vectors which contain exactly one
1 and 63 0's. Hence, they have weight 1, and their pairwise Hamming distance is
two. Thus, my set has minimal weight 1, minimal distance <= 2, and is as good as
yours.

I'll try and write a program which checks Hamming distance and linear
independence for a small dimension. (I don't know if an exhaustive search is
feasible.) I'll let you know if I find anything.

Sven.

Sven



This page took 0.01 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.