Levenberg-Marquardt Method for Complex functions

[KevinHall]

I'm familiar with the Levenberg-Marquardt Method for real functions and use it frequenly. Recently I've needed to fit some models to complex functions. Anyone know if the Levenberg-Marquardt Method works well for complex functions (I have been unable to find any examples using complex functions)?

Thanks,

Kevin

[galathaea]

I read your question and was quite stumped at first. I had never seen the L-M method used on complex, but I was curious. So I looked in my algorithms books to see the full implementation (I could not recall its entire justification or proof off the top of my head). I could see how it would still work with the appropriate use of the complex absolute and ordering, but was still unsure. So I did a little search and found this site (http://aips2.nrao.edu/docs/trial/implement/Fitting/NonLinearFitLM.html). I noticed that this implementation could be applied to complex values. I still would like to check into the implementation to check some things out, but I have to return back to work right now so I thought I'd drop off what I had learned. I hope it is helpful...

[KevinHall]

Galathaea,

Thanks for the help. I tried doing the same thing you suggested (complex magnitudes and ordering). I've gotten the code to compile, and am about to test it. I'll let you know if it works on my test case...

Thanks again!

- Kevin

[KevinHall]

I've missed something b/c it doesn't work well. Atleast I know its possible...