[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: your implementation of L'Ecuyer's MRG32k3a generator



> 
> On the other hand: It would be nice to hear what crypto people
> would like the interface to the RNG to be. Your earlier proposal
> for a method to obtain a stream of bytes rather than range-limited
> integers with variable range is a start. I am still thinking on
> how to solve that one nicely.

As a crypto person, we generally want to be able to generate random 
integers that can be represented in a fixed range of bits (for example, 
0 -> 1024 bits) for public/private key cryptosystems.  Block ciphers and 
protocol nonces require the same, though usually in smaller sizes 
(64,128,256 bits).  Generally floating point numbers are not required, 
as most of the algorithms operate on the natural numbers.

The nice thing about obtaining a stream of bytes is that if you have 
one, and your Scheme system has arithmetic shift, you can get any of the 
above with some simple routines and a stream of random bytes.  

	Scott

Attachment: pgpG27j4PIf0M.pgp
Description: PGP signature