Introduction
The B mass fit is performed by the functions defined in MassFit.cpp. In this file, the main function massFit is defined. This function is rather complex with many parameters as we tried a lot of mass models.
The fit model is based on RooFit.
Before executing any part of the selection code involving the mass fit, it is needed to compile the files with the classes needed for the fit.
Table of Contents
- Introduction
- Mass Fit compilation
- Adding double crystal ball and ExpGauss to RooFit
- Fit model description
- Source files
Mass Fit compilation
Recompile mass fit
root
.L BmassShape/SignalType.cpp+
.L BmassShape/SignalPdf.cpp+
.L BmassShape/BackgroundType.cpp+
.L BmassShape/BackgroundPdf.cpp+
.L BmassShape/ParamValues.cpp+
.L MassFit.cpp+
.q
There might be some warning about unitialized variables, just ignore those.
Adding double crystal ball and ExpGauss to RooFit
Standard RooFit does not include a double crystal ball or ExpGaus function used to descibe partialy reconstructed background. Thanks go to Flavio for pointing me to the RooDoubleCB code and helping with the compilation!
RooDoubleCB compilation
Run
cd Code/Selection/RooFit/RooDoubleCB/
root
.L RooDoubleCB.cpp+
.q
If you do not compile the RooDoubleCB before running the MassFit for the first time, the code will compile but later throw an error it cannot find macro ./RooFit/RooDoubleCB/RooDoubleCB_cpp.so.
RooExpAndGauss compilation
This will fail if you already sourced ROOT or python described here. You might have to restart the enviroment (easiest is to log out and in)
cd Code/Selection/RooFit/RooExpAndGauss/
make
Recompile Mass Fit
Now we need to link the RooFit models to MassFit.cpp by recompiling it in ROOT
root
.L MassFit.cpp+
.q
Fit model description
The fit is defined in the massFit function in MassFit.cpp. The user can decide
- What year, polarity and Run to fit
- What sample (data, MC, PHSP, inclusive MC) to fit
- What selection to fit (stripped, preselected, set cut on MVA response, random subset of a selection, remove multiple candidates)
- What Q2 range to fit (below Jpsi/Jpsi/rare/all)
- Whether to constrain of fix the parameters to MC
- Whether to use LL or DD tracks
- Wheter to use sWeights or special weights calculated from efficiencis
- Reduce verbosity/saving
- What truth-matching to use
- Mass range for the yield calculation
- Apply additional cut on a variable (eg when scanning the B mass fit in different cos(thetak) bins)
Signal
The signal can fully turned off or be described by
- Left-sided Crystal Ball
- Right-sided Crystal Ball
- Two Crystal Balls
- Both-sided Crystal Ball
- Gaussian
- Double Gaussian
The fixed/starting parameters are saved in ParamValues.cpp.
Background
The signal can fully turned off or be described by
- Exponential
- Double Exponential
- Exponential + ExpGaus
- One Crystal Ball + single exponential
The fixed/starting parameters are saved in ParamValues.cpp.