Re: Random Number Replication

[landsman(at)mpb.gsfc.nasa.gov]

In article <37aaad97.5297206@trog.dera.gov.uk>,
  dsheerin@dera.gov.uk (David Sheerin) wrote:
> Hi
> I have tried to use the following code to generate a sequence of
> random numbers:
>
> FOR k = 1,  4 DO BEGIN
> 	p = RANDOMU (S)
> 	PRINT, p
> ENDFOR
>
> Why, when I repeat the operation, does the code (nearly) give me the
> same list of random numbers shifted by 1?  I thought that each time
> RANDOMU was called it took a new value for the SEED due to S being
> undefined.  How do I 'undefine' S after it has been called?

David,

    This is due to a bug in RSI's implementation of the RANDOMU function
in V5.1.1 and V5.2.   I believe it has been fixed in V5.2.1 (see below)
but I haven't been able to check V5.2.1 out yet.    The bug is that only
during the *first* call to RANDOMU is a seed initialized inside a
program.   Thus, if a program test.pro contains multiple RANDOMU calls,
then on a second call to test.pro, SEED will take the value it had after
the first call to RANDOMU during the first call to test.pro.   The net
effect is that you see a shift by 1, in the output of the random
numbers.

Another problem in V5.1 and V5.2 -- I'm not sure if it is related -- is
that the seed variable is initialized to the same value at the start of
each session rather than the system clock.  The bug fix notes in V5.2.1
specifically say that this problem is now fixed:
************************************************

RANDOMN and RANDOMU Function Initialization:
--------------------------------------------
The RANDOMN and RANDOMU functions are now properly initialized with the
system time the first time they are used.
************************************************

but I am not sure if your problem is also fixed.

--Wayne Landsman                              landsman@mpb.gsfc.nasa.gov



Sent via Deja.com http://www.deja.com/
Share what you know. Learn what you don't.