Understanding Global Optimization VI

[DSPGuy]

If you can use Lev-Mar then I would definitely recommend doing so.  You could use the global optimization code to generate the initial guess, then use Lev-Mar to quickly improve the initial guess.  This will probably be the most robust solution. 

-Jim

[jeanlucpicard]

Wow, yesterday I tried it to get the initial guess with the Lev-Mar and then use the global optimization, however it gave almost identical to my old program's value by doing it the other way round. So now I tried with the global optimization and then use a Lev-Mar (by typecasting the complex data to "Y" input and then using only the variant for the fitting) in the end to get the final fit.

Compared to the data of my program I achieved 0,999989-1,00044 accuracy which I find impressive. Now I only have to check whether it is reproducible with other data sets as well.

Thanks for your invaluable help, Jim.

-Gergely