Re: Non-linear combination
From: Bryan Olson (fakeaddress_at_nowhere.org)
Date: 11/19/03
- Next message: Paul Rubin: "Re: Database encryption."
- Previous message: Bryan Olson: "Re: Database encryption."
- In reply to: Mok-Kong Shen: "Re: Non-linear combination"
- Next in thread: Mok-Kong Shen: "Re: Non-linear combination"
- Reply: Mok-Kong Shen: "Re: Non-linear combination"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ] [ attachment ]
Date: Wed, 19 Nov 2003 04:32:55 GMT
Mok-Kong Shen wrote:
> I see that's a consequence of my not having been
> precise/clear enough in the original post (and a few
> later additions).
I think it's a consequence of not having anything precise to
express. For example, you had thought it bad that the product
of the more- significant halves of the inputs get shifted out of
the output. But your current favorite...
> The fact that 2xy+x+y mod 2^n is,
> for a fixed value of y, a permutation easily leads
> one to think that that forms a cheap/simple S-box.
...does the same thing. So was the objection right or wrong?
How can you tell?
> But
> what I have originally in mind is more in the direction
> of combining two streams from PRNGs of n-bit units.
> Commonly one employs xor or addition mod 2^n for that.
> Having a non-linear term 2xy, which is not overtly
> expensive, seems to be worthwhile.
So how would you know if it's worthwhile? Can you prove it?
Can you state a hypothesis and what evidence would refute that
hypothesis? If you cannot, you are not doing science.
-- --Bryan
- Next message: Paul Rubin: "Re: Database encryption."
- Previous message: Bryan Olson: "Re: Database encryption."
- In reply to: Mok-Kong Shen: "Re: Non-linear combination"
- Next in thread: Mok-Kong Shen: "Re: Non-linear combination"
- Reply: Mok-Kong Shen: "Re: Non-linear combination"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ] [ attachment ]
Relevant Pages
|