Compared to the fit method in spectral analysis softwares( XPEC RMFIT ) and IDL, what is the differences between them? I am not experted in it. Thank you.
Nobody can help? I am curious about the answer.
well. your question is very general and not clear, so nobody feels
inclined to help …
For starters, all fit methods in ROOT are using a form of maximum
likelihood .
Are you asking about a linear fit (TLinearFit) , non-linear fit (TFitter,
TMinuit,TMinuit2,TFumili) ?
Eddy
[quote=“Eddy Offermann”]well. your question is very general and not clear, so nobody feels
inclined to help …
For starters, all fit methods in ROOT are using a form of maximum
likelihood .
Are you asking about a linear fit (TLinearFit) , non-linear fit (TFitter,
TMinuit,TMinuit2,TFumili) ?
Eddy[/quote]
Thank u for your answer.
I’ve checked out sth on some websites. There are various fitting methods including Levenberg-Marquardt(LM), Genetic algorithm(GA), Simulated Algorithm(SA), Simplex Method(SM), Powell method ,etc. Chi-square statistics, Cash statistics and other maximum likelihood functions are used as Fit Statistics Functions.
So I want to know which fitting methods are the fit class(TLinearFit, TFitter, TMinuit…) use. Can you show me the form of the maximum likelihood function?
Each class has a description, which can be browsed, see
for instance :
http://root.cern.ch/root/htmldoc/TFumili.html#TFumili:description
http://root.cern.ch/root/htmldoc/TMinuit.html#TMinuit:description
An extensive discussion of Minuit is found at for instance:
http://root.cern.ch/root/htmldoc/TMinuit.html#TMinuit:description