Nonlinear fitting Initial coefficiencies

[Camu]

Hi, everyone, currently I'm working on fitting nonlinear functions. My question is, how do I tell the program to fix some of the initial coefficiencies(I mean accept the value I put as known parameters; they may be treated as known parameter or not, depending on the results)? Any help will be highly appreciated!

Thank you!

Best

Camu 

[chilly_charly]

Trying to understand your question : in case the non-linear fit do not converge to a satisfactory solution (as estimated using the mse for instance), you would like the initial coefs to be automatically changed before running the fit again ? 

There is no definite receipe for such a problem. You could try generating random values as seeds, after deciding of a working range (average and deviation) for each/some parameter.

[Camu]

Hi, Chilly, sorry for the misunderstanding.

Actually, I just want my program to be more "fancy". Like fitting an bi-exponential decay, there are five parameters in the fitting: A1, t1, A2, t2, and background. However, just like in some other programs (e.g. Origin or FluoFit), not all of the fitting parameters need to be treated as variables. e.g. I knew for sure t2 must be 3ns, then I just want to tell the program, don't treat t2 as a variable this time, just by simply clicking or changing "some state of something" (that's what I'm after) in the program. 

Hope this is clear.

Cheers

Camu