Show Posts
-----
Quote from: Nimda
If we call SHA256 a pseudorandom generator G, we can test the bias of its outputs using any efficient algorithm A such that A(G(k <--R-- ??)) (the algorithm run on pseudorandom numbers) returns 0 with sufficiently higher probability (called the 'advantage') than A(r <--R-- {0, 1}n) (the algorithm run on truly random numbers).
http://www.fourmilab.ch/random/ looks promising. I'll be back.
http://www.fourmilab.ch/random/ looks promising. I'll be back.
There are a bunch of non-intuitive issues with randomness/entropy testing, which is a well-studied field, and there is no one (practical) single test that you can run that should satisfy you. The gold standard for many years was the Die Hard battery of tests by George Marsaglia (who discovered, among many other things in the field, that Linear Congruential Generators are of poor quality "randomness", which is basically why fast-but-decent PRNGs use Mersenne Twisters these days). The Die Hard battery was recently rewritten, added to and GPLed by Robert Brown, see here: http://www.phy.duke.edu/~rgb/General/dieharder.php
That said, John Walker (Fourmilab that you linked to) is great, and I used his HotBits about 15 years ago when it first came out, and I wanted "true" randomness (and afaik, nothing else online existed - or for a long time afterwards).
