Hi all,

I have a question which for sure has been answered a 1000 times before, but I didn’t find a satisfactory and convenient solution, albeit I have browsed the web and the Wolfram- (and other ) sites for hours.

I have a list of experimental data. It is known that there is some periodicity (or even more than one). Now I apply “Fourier” which produces a nice other list of complex numbers. When plotting the absolute values of this list (a spectrum) I even find some “peaks” and I think I translate them correctly to frequencies. Now I try to construct a function which approximates my original data, and fail drastically. So I set out to get insights with a “known” example. But I failed as well. With some hints of Daniel Lichtblau (thanks, Dan) I could construct a procedure which worked for my simple example, but it failed already for a slightly more complicated case ( and for sure for my original data ). See the attached notebook.

I am looking for a fool-proofed (the fool am I) method to construct a good approximation for the original function, or better the values given.

If anybody is interested in the original problem I will of course send all data I have.

Kind regards Hans

P.S. I am using version 7