How to improve fit result if having many parameters

ROOT
1

i have a fit of 30 parameter, do everyone know how to improve the resule?

i found in my combine fit function , the different sequence of the sub-function can get different result!
for example:

Double_t fit_f( Double_t *x, Double_t *par)
{
    Double_t fitval=fit_3+fit_1+fit_2;
       //different with  Double_t fitval=fit_2+it_3fit_1;
                  return fitval;
}

f->Fit("fit","LEV+"); is in use.

2

Could you post the shortest possible running version of your script explaining what you are doing. No way to give you more hints
without looking at your problem.

Rene

3

hi brun,
i attatch you the code .
i have describe the problem in the code file.
all_diff_data.txt (3.39 KB)
fit2Dg_of_diff.c (4.16 KB)

4

Two problems:
-you must include

#include <Math/SpecFuncMathCore.h> -the prototype for assoc_legendre is

double assoc_legendre(unsigned l, unsigned m, double x) ie you have a missing argument in your code

Rene

5

Thank you Rene ! :unamused: