class AliAnalysisMuMuFnorm: public TObject

 ctor

 dtor

 Compute the REF to CINT ratio(s)

 Using offline method
   - in one go CINT/REF
   - in two steps CINT/CMSL and CMSL/REF

 Using scaler method
   - bare scaler values
   - scaler values corrected for pile-up
   - scaler values corrected for pile-up and physics selection

 Compute individual graphs for the correction factors (PS_REF, PS_CINT,
 F_pile-up,PS_CINT/PS_REF) used in the computation of (some) Fnorm factors

 Compute MB to REF ratio using offline method, either in 1 or 2 steps

 Compute the MB to REF ratio using the scalers method (from OCDB)

 i.e. Fnorm = L0B(MB) x PS(MB) x Fpile-up / ( L0B(REF) x PS(REF) )

 where MB is the minbias trigger
 REF is the fReferenceTrigger
 and PS is the fraction of events selected by the physics selection

 The correction factor (the two PS and one Fpile-up) are
 taken from graphs computed in other methods

 compute dispersion of b versus a

 computed differences graphs are put into the GRAPHS/ directory
 computed differences results are put into the HISTOS/ directory

 Compute the per-run graph of pile-up factor

 Compute the per-run graph of physics selection accept fraction
 for the given trigger

 Compute one single value for each graph, by weighting by the fraction
 of events in each run
 do this for certain groups of graphs

 Compute trigger fractions

 Compute trigger fractions

 Compute trigger L0B

 Creates a graph and adds it to our mergeable collection

 let go the ownership of our mergeable collection

 shortcut method to give access to one graph

 Get the physics selection accept fraction for a given trigger

 shortcut method to give access to one result

 get one value +- error for this graph (weighting the run-by-run values
 by the fraction of the number of triggers in that run)
 also the rms is computed.

 get the triggerclass to for a given trigger type and run number

 get the name of the trigger type

 get (value,error) corresponding to run=runNumber.
 Works both for compact and non-compact graphs

 get our mergeable collection

 FIXME : find a better way ?

 FIXME : find a better way ?

 FIXME : find a better way ?

 Make a new graph = g1*g2

 FIXME : find a better way ?

 Extract the run numbers from our counter collection

 Compute the MB to REF ratio and its associated error

 Print and plot Fnorm values

 Check if we have counts for that trigger,run combination

 Sum the graphs which name matches pattern
 Sum is made using a weighted mean (each element is weighted by the inverse
 of its error squared)