Hi ROOTers,
I found that the bounded parameters undergo non-linear transformation during the fit, and it is a good practice to set the bounds far from the final fit value.
My question is: if the parameter is bounded (0,1), and the parameter value after the fit is 0.001 or 0.999, should the bounds be changed to keep the fit as centred as possible?
Thank you for your time.
Regards,
ATD
What you can try is to fit your data enforcing some “limits” on some variables. Once the fit is successful, make these variables “free” and redo the fit (so the new “starting point” will be the already found “best fit”). If the new fit is successful and it stays close to the old “best fit”, you are done.
Hello
Thank you very much, I will do this way.
Regards,
ATD
This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.