Splittable random numbers

[Ian Lynagh]

On Thu, Nov 04, 2010 at 05:38:12PM +0000, Simon Peyton-Jones wrote:
> 
> The generator uses crypto functions,

Does that mean it couldn't be used in some countries?

> so it's probably more computationally expensive than common linear-sequence generators, but in exchange you get robust splitting.

I wonder if you can make a splittable generator that uses crypto
functions when you split it, but is a common linear-sequence generator
otherwise?


Thanks
Ian