Information on calculator Random Number Generators from PPC Journal articles

[Paul Dale]

I'll check some more test results for this function tomorrow if I remember -- as I mentioned it fails TestU01 and is marginal for the dieharder suite.

I never claimed this generator was unbiased, just that it passed the current (at the time) NIST randomness tests with a high entropy result (as close to 8 bits per byte as was possible). It was designed to be presented to the CMVP conference earlier this year as a simple deterministic function that passed the tests but showed they weren't great. It was based on the first draft tests, but the function passes the second draft also, and at the same level of apparent entropy

I'll also try to come up with something similarly simple but less biassed. Entropy is my day job currently...


Pauli