ROOT::Fit vs RooFit

[atd]

Dear rooters,

I am using the recipe for unbinned likelihood fits given in this example; the example uses ROOT::Fit classes, and it was suggested that the fit function should be normalized.

I was told today that RooFit performs unbinned likelihood fits with fit functions that are not normalized. Is this correct? (I was not able to find any proper explanation in the RooFit documentation; I found some examples, but the fit function was normalized)

Thank you for your time.

Regards,
atd

---

Please read tips for efficient and successful posting and posting code

ROOT Version: 6.18
Platform: CC7
Compiler: gcc 4.8

---

[StephanH]

Hello @atd,

RooFit normalises every function. If the normalisation (= the integral) is known analytically, that is, you use built-in RooFit functions, it is applied automatically. For all other functions that you can implement using the RooGenericPdf, RooFit runs a numeric integration, and uses that to normalise the function.

If you want to know more in a hands-on style, you can have a look at the tutorials at
https://root.cern.ch/doc/master/group__tutorial__roofit.html

Especially about normalisation:
https://root.cern.ch/doc/master/rf110__normintegration_8C.html
And about the GenericPdf:
https://root.cern.ch/doc/master/rf103__interprfuncs_8C.html