These are all good tests, BUT all they can tell you is a negative. Ie, if some tests fail, the RNG is certainly BAD. If they do not fail, we still can't say that the RNG is good. igor wichita@cyberstation.net wrote:
You are also bringing a good point that Chi-squared tests are not sufficient to make any conclusions about usefulness of this particular pseudo random number generator.
- Igor.
Chi Squares alone are not sufficient but we are only talking about the seed algorithm, and at our web sites, you will find Standard Deviations, Chi Squares, Delta ICs, autocorrelations, cross correlations, and a variety of other tests done on single characters, couplets - pairs, first differences, second differences, offset differences and all kinds of other tests.
- Igor.