Dear Roottalk,
I use a chi square minimization routin minuit. I call it in the following way:
TGraph2DErrors *dte = new TGraph2DErrors(data_length);
for(int i=0;i<file_length;i++){
dte->SetPoint(i,x_b[i],Q2[i],F2[i]);
dte->SetPointError(i,er_x[i],er_y[i],er_z[i]);
};
TF2 *func = new TF2("fit", fitted_function, 0,0.013, 0,150, 3);
func->SetParName(0,"mu0"); func->SetParName(1,"A"); func->SetParName(2,"Omega_0");
func->SetParameters(param);
dte->Fit("fit");
where "fitted_function" is my function with three parameters and two variables (x_b,Q2)
The problem is: the result of minuit or a value of chi square depends on the initial values of my parameters, i.e. a little change in the initial values affects considerably on the result of minimization (a value of chi squre). I think that the reason for this is that my function has many local minima and one global minimum, but I can not achieve it.
Can I do something with it?
-- Sincerely, Andrey K. *************************************************************Received on Sun Apr 22 2007 - 12:09:31 CEST
* Andrey Kormilitsin * *
* * *
* Department of * *
* Particle Physics * *
* School of Physics * Tel: ++ 972- 3 - 640 7954 (o) *
* Tel Aviv University * *
* Ramat Aviv,Tel Aviv * E-mail: andreyk1_at_post.tau.ac.il *
* 69978, ISRAEL * *
* * *
************************************************************* ---------------------------------------------------------------- This message was sent using IMP, the Internet Messaging Program.
This archive was generated by hypermail 2.2.0 : Mon Apr 23 2007 - 11:50:01 CEST