RE: Random Number Generators
Alex Popelyukhin ( (no email) )
Wed, 16 Jun 1999 15:33:16 -0400
>>>
1. According to Intel, "true" RNG is coming:
they built into the processor an instruction for
random number generation. They plan to use one of the known
"uniform" physical processes (in this case heat of the processor
itself) to get a "true" random number.
("True", because number of unique combinations of N bits to represent
ANY number in the computer is limited).
2. In fact (if I remember correctly) Excel tries to use an external
process for selecting a random seed as well: it asks for time and uses
number
of milliseconds as a seed - so, to a degree, it automatically resets
the seed.
3. As for pseudo-random RNG R(k+1) = A*R(k) + B the trick is in selecting
good A and B. My understanding is that the "best" pair is "discovered" by
Sun Microsystems - and many serious simulation packages use Sun's A and B.
4. The approach I personally would use is based on the fact that Euler
constant
(number e) is transcendent, as a corollary it's decimal digits are randomly
distributed. Having written a very fast algorithm for calculation of decimal
representation of e, I got all my random numbers right there.
Aleksey Popelyukhin, Ph.D.
Visit the CAS Web Site at http://www.casact.org
===============================================
To subscribe or unsubscribe from CASNET:
Send an e-mail to caslists@lists.casact.org
Type in the body join casnet to subscribe
or leave casnet to unsubscribe.