Randomu Period [message #39256] |
Fri, 30 April 2004 16:59 |
jargoogle
Messages: 11 Registered: April 2004
|
Junior Member |
|
|
Hello,
Does anyone know the actual period of the randomu uniform
pseudorandom number generator? I'm using IDL 5.3 and IDL5.6 and the
online documentation references a paper in an ACM journal. I'm on my
way to the library to check that now.
Does anyone know if RSI implemented the algorithm described in that
paper exactly? That is, can I expect the article in the journal to
describe an algorithm which takes a 36 element long array as seed? If
not, where can I get information about the exact period (repetition
cycle) after which the pseudorandom numbers will begin to repeat?
The documentation also notes the Numerical Recipes routine. That
routine uses a simple 2 byte seed to get a period around 2^16-1. Does
that mean IDL gets something like 2^1152-1 ?!!! That'd be nice, but
I'm interested in something more than wishful thinking. What's the
real period?
Anybody know?
I'll probably give RSI an email or call.
Thanks,
JG.
|
|
|