ths
04-30-2003, 06:55 PM
I've created a library of the NR f-90 routines using the Absoft 7.0 compiler (Intel / RedHat Linux 8.0).
Both ran1 and ran2 seem to work fine (they return uniform deviates between 0 and 1) when called with a vector (i.e., ran1_v and ran2_v). But when called with a scalar (ran1_s and ran2_s), they don't return correct numbers (the numbers are non-uniformly distributed between -1 and 1).
I've observed this with the NR demo programs, so it is not a problem with improper calling of the subroutine.
Anybody have any ideas about what is going on?
Both ran1 and ran2 seem to work fine (they return uniform deviates between 0 and 1) when called with a vector (i.e., ran1_v and ran2_v). But when called with a scalar (ran1_s and ran2_s), they don't return correct numbers (the numbers are non-uniformly distributed between -1 and 1).
I've observed this with the NR demo programs, so it is not a problem with improper calling of the subroutine.
Anybody have any ideas about what is going on?