Using SHA1 as RNG

[Klaus Alexander Seistrup]

Irmen de Jong wrote:

>>> Even if the hash doesn't repeat itself, you still get a 100%
>>> predictible sequence of numbers because the SHA algorithm is
>>> repeatable.
>> 
>> But the same goes for the original Wichmann-Hill generator,
>> doesn't it?
> 
> Umm.. perhaps. I don't know that much about strong random number
> generators. Can anybody enlighten me (us)?

The Wichmann-Hill PRNG is basically multiplication and modulus with
2×3 known constants (snipped from whrandom.py):

#v+

    def random(self):
        """Get the next random number in the range [0.0, 1.0)."""
        # This part is thread-unsafe:
        # BEGIN CRITICAL SECTION
        x, y, z = self._seed
        #
        x = (171 * x) % 30269
        y = (172 * y) % 30307
        z = (170 * z) % 30323
        #
        self._seed = x, y, z
        # END CRITICAL SECTION
        #
        return (x/30269.0 + y/30307.0 + z/30323.0) % 1.0

#v-

Start with the same seeds, and you end up with the same sequence
of numbers.  Oh, and it seems the cycle for the Wichmann-Hill RNG
is not 30268*30306*30322-1, but (30268*30306*30322-1)/4.  See (¹).
I believe the strength of the W-H generator is that it is portable
and computational inexpensive.


  // Klaus

 (¹)  <http://www.math.montana.edu/Rweb/Rhelp/Random.html>
-- 
 ><> 	unselfish actions pay back better