| autocorrelation of RANDOMU (is RANDOMU white?) [message #2526] |
Sat, 30 July 1994 21:32 |
at913
Messages: 9
Registered: January 1993
|
Junior Member |
|
|
Hello!
I just computed the autocorrelation sequence of
x=randomu(seed,1024) ;seed is undefined
I get the autocorrelation to be nonzero for all delays. As the matter
of fact it looks like the autocorrelation of white noise, offset by some
positive value, i.e. it peaks at 0. and then has a constant nonzero value for
other delays. SOmehow, I was expecting pure white noise behavior from
randomu. Were my expectations wrong?
randomn does not suffer from the same problem.
(Note: i calculated the autocorrelation two ways
one:
ft=fft(x,-1)
ps=ft*conj(ft)
co=fft(ps,1)
two:
x3=[x,x,x]
co=convol(x3,x,/center,/edge_wrap) ; this way I get around the requirement
; of dim(x)<dim(x3) needed for convol.
)
Mirko
--
Mirko Vukovic
University of Wisconsin -- Madison
VUKOVIC@UWMFE.NEEP.WISC.EDU
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
