Lomb normalized periodogram

Both functions calculate the Lomb-Scargle periodogram (aka Gauss-Vanicek/Least-squares spectrum)

Description

Both functions caculate the Lomb normalized periodogram (aka Lomb-Scargle, Gauss-Vanicek or Least-Squares spectrum) of a vector x with coordinates in t, which is essentially a generalization of the DFT for unevenly sampled data.

The codes are transcriptions from Fortran of the subroutines found in Section 13.8 (pp. 569-577) of "Numerical recipes in Fortran 77: the art of scientific computing", 2nd ed., vol. 1, Cambridge University Press, NY, USA, 2001 by WH Press, SA Teukolsky, WT Vetterling and BP Flannery,

However, Matlab's characteristics have been taken into account in order to make it fast for Matlab.

FASTLOMB is much faster than LOMB (especially when the length of the input increases) but even LOMB is faster than any other implementation I found in FileExchange. Also they both do not suffer from memory problems (I tested them both for inputs of 100,000 samples).

I'd also like to acknowledge file ID: 20004 (for some reason I can't get two file IDs in the acknowledgements)

Acknowledgements

The author wishes to acknowledge the following in the creation of this submission:
lombscargle.m

• MATLAB release: MATLAB 7.6 (R2008a)
• Other requirements: Example times on my 3-year old pc: Samples of input/LOMB time (sec)/FASTLOMB time (sec): 1,000 / 6 / 0.15, 5,000 / 12 / 0.5, 10,000 / 50 / 1., 50,000 / 20min / 5., 100,000 / 2hrs / 40

[Daniel Armyr]

Nice implelemtation of the Lomb method. The only reason I am not giving this a five is that the built-in plotting functionality is far to over-specified. Opening up a new figure and then drawing raw lines in it is not particularly flexible or helpful for integrating into larger applications.

[Benjamin Bayes]

Excellent function. However, in fastlomb.m, it seems like on line 212,

ck = 1 + mod(fix(t(j)*fac),2*nfreq);

should be

ck = 1 + mod(t(j)*fac,2*nfreq);

If not changed, it seems like the else statement on line 228 is never reached. Thanks!

[Benjamin Bayes]

 

[Christos Saragiotis]

Daniel, thank you for your commenting.

Regarding the plotting part, I am not sure I understand what you mean, but you can always set fig to 0 and then take the output of the m-files and plot them as you wish.

Sorry it took me so long to answer, I thought I was getting an automatic email every time someone commented but it seems this is not the case

[Felipe G. Nievinski]

Does the job.
You might want to replace line 198:

    if length(x)~=nt, disp(sprintf('WARNING %s: Double entries have been eliminated',mfilename)); end

for:

    if length(x)~=nt, warning('fastlomb:Duplicates','Double entries have been eliminated.'); end

so that it can be disabled issuing warning('off','fastlomb:Duplicates')