Reply by robert bristow-johnson●January 17, 20042004-01-17
In article bu9i7q$tvp$1@blue.rahul.net, Ronald H. Nicholson, Jr. at
rhn@nojunk.rahul.net wrote on 01/16/2004 15:40:
> I asked this before under another subject line, but perhaps it got lost.
>
> Assume I want the autocorrelation peak or best least squares fit of
> sampled data for non-integer sample offsets (e.g. the periods of the
> frequencies of interest are not a multiple of the sample interval).
>
> Should I interpolate the data for the autocorrelation/ASDF? Or should
> I do the autocorrelation for integer sample offsets and interpolate the
> result? The former is a lot more computationally intensive, but does
> it give any better result than the latter?
ooops. sometimes i forget to watch threads that i have posted in. i'm
posting a response on that thread "Re: Detecting the period of a signal",
just to keep it easier for anyone following it or any progeny googling in
some years later.
> Thanks.
r b-j
Reply by ●January 16, 20042004-01-16
I asked this before under another subject line, but perhaps it got lost.
Assume I want the autocorrelation peak or best least squares fit of
sampled data for non-integer sample offsets (e.g. the periods of the
frequencies of interest are not a multiple of the sample interval).
Should I interpolate the data for the autocorrelation/ASDF? Or should
I do the autocorrelation for integer sample offsets and interpolate the
result? The former is a lot more computationally intensive, but does
it give any better result than the latter?
Thanks.
--
Ron Nicholson rhn AT nicholson DOT com http://www.nicholson.com/rhn/
#include <canonical.disclaimer> // only my own opinions, etc.