Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: About random numbers and hashing

Author: Sune Fischer

Date: 05:45:47 12/07/01

Go up one level in this thread


On December 07, 2001 at 07:30:10, David Rasmussen wrote:

>Hamming distance is not the only criterion that is important, and as you
>yourself point out, you can get some really bad stuff by just going by hamming
>distance (although it is unlikely. The example you give is improbable). On the
>other hand, the "connection between the quality of hash codes and their Hamming
>distance" is quite obvious. If you have very low hamming distance, you are much
>more like to get a low-dimensioned vector space.

Why is that?

If you create a set of vectors with mutually high hamming distances, then you
are actually giving them a feature they all have in common.
This is against randomness, some should have large distances and some small,
this is "more random", random means random in _every_ aspect.

If you use 64 bit keys there is probably enough randomness left in the keys so
that this is not a problem, but for 32 bit keys it could be a bigger problem.


-S.



This page took 0.02 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.