Next: Discrete Fourier Transform, Previous: Prime Numbers, Up: Mathematical Packages [Contents][Index]

A pseudo-random number generator is only as good as the tests it passes.
George Marsaglia of Florida State University developed a battery of
tests named *DIEHARD* (http://stat.fsu.edu/~geo/diehard.html).
`diehard.c` has a bug which the patch
http://groups.csail.mit.edu/mac/ftpdir/users/jaffer/diehard.c.pat corrects.

SLIB’s PRNG generates 8 bits at a time. With the degenerate seed
‘`0`’, the numbers generated pass DIEHARD; but when bits are
combined from sequential bytes, tests fail. With the seed
‘`http://swissnet.ai.mit.edu/~jaffer/SLIB.html`’, all of those
tests pass.

• Exact Random Numbers: | ’random | |

• Inexact Random Numbers: | ’random-inexact |