Fitting a Sum of Exponentials to Numerical Data

[crowlogic]

I've found a paper at http://arxiv.org/abs/physics/0305019

Essentially the paper describes a method to fit a sum of exponentials via a multi-linear fit instead of the old estimate/iterate procedure. This has huge advantages imho, but I am not a mathematican and it is not straight forward how to turn this into code.

Does anyone have any insight into this, or could break it down into plain english speak? I have no problem coding it myself if I could just get a grasp on the steps involved.

Thanks in advance!
--Stephen

Authors: Bernhard Kaufmann
Subj-class: Data Analysis, Statistics and Probability

A finite sum of exponential functions may be expressed by a linear combination of powers of the independent variable and by successive integrals of the sum. This is proved for the general case and the connection between the parameters in the sum and the coefficients in the linear combination is highlighted. The fitting of exponential functions to a given data- set is therefore reduced to a multilinear approximation procedure. The results of this approximation do not only provide the necessary information to compute the factors in the exponents and the weights of the exponential terms but also they are used to estimate the errors in the factors.