Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: About random numbers and hashing

Author: Sune Fischer

Date: 14:34:37 12/07/01

Go up one level in this thread


On December 07, 2001 at 15:59:29, David Rasmussen wrote:

>>>There is as much randomness in my keys with minimum hamming distance 17 as there
>>>are in anyone who has just taken the first keys that came and has a minimum
>>>hamming distance of 11. And my keys are less likely to collide.
>>
>>If they are less likely to collide I think you have increased randomness, so in
>>essence you have improved the PRNG.
>
>Why do you think that? Why do you insist that increased randomness is the only
>way to minimize collisions?

Because I am trying to consider the field F2^64, and a set of vectors can
"become closer by the XOR metric" if you _increase_ their Hamming distance,
because they get a more similar number of bits that are off-set. This could,
IMHO, increase the chance of them XOR out, since this seems (too me at least)
harder to do if you have a different number of bits off-set for each vector.

Enforcing special features upon the set of vectors definitely destroys
randomness to some degree, it could be you improve on the table thereby, I just
don't think it is obvious in any way.

But okay if this is a widely accepted technique in the world of computer
science, then who am I to say it shouldn't work ;)

-S.



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.