Ttbar analysis skeleton
The analysis of the Ntuples for the Rome data is done using ROOT. On this page the Analysis Skeleton is presented and instructions on how to use it.
The 3 files that make up the Analysis Skeleton
The Analysis of the Root ntuple proceeds in a very straightforward way. Inside ROOT we used MakeClass() to get the .h and .C files. These were extended and together with an additional steering file these three files form the TTbar analysis skeleton:
- The standard header file extended with the implementation of a few routines the user is not expected to change often. Also a new class is introduced: TLorentzVectorWIdx, which is a TLorentzVector with an additional index that points back to it's position in the original list so we can access it's additional characteristics if we wish to do so.
- The user analysis-file where the user:
- o reads in the event
- o decides what (b-)jet algorith he wants to use
- o fills vectors with pseudo-particles (leptons, jets, photons etc.)
- . Here actually two vectors are filled: one for 'all' and one for 'good' pseudo-particles that pass the user's quality requirements
- o performs the top-analysis. The resulting histograms will be written out in a separate root file
- The small steering-file containing the location of the ntuples and a function that decides:
- o what ntuples to process (fast/full/background simulation)
- o how many events to process
- o name of the output file
The code itself & How to run the Analysis Skeleton
The skeleton analysis (compiling and running) can be done completely in ROOT or standalone using a Makefile. The code to follow these two lines is presented below and it is left to the user which path he would like to follow.
1) Compile and run Analysis in Root
The code to do the analysis completely in ROOT can be found in: Analysis Skeleton Code (RootBased, Version 6)
The first step is to correctly set the location of your ntuples in the file simpleNtupleTTbarAnalysis.C. Then, to analyse the first 100 events in ntuple-set 1 (full simulation) and to store the output histograms in a file called summary_ttbar_full_test.root simply open a root session and type:
Root> .L simpleTTbarAnalysis.C++; Root> .x simpleNtupleTTbarAnalysis.C(1,100,"test");
When setting the number of events to "-1", all events are processed. If you want to analyze all events in ntuple-set 2 (W+4jets background) and to store the output histograms in a file called summary_wjets_full_allevents.root simly execute the following line:
Root> .x simpleNtupleTTbarAnalysis.C(2,-1,"allevents");
2) Compile and run Analysis StandAlone
You can of course also compile the code and run the analysis outside root. The code and the additional Makefile (from SACLAY) can be found in Analysis Skeleton Code (StandAlone, Version 6)
The first step is to correctly set the location of your ntuples in the file simpleNtupleTTbarAnalysis.C. Then, to analyse the first 100 events in ntuple-set 1 (full simulation) and to store the output histograms in a file called summary_ttbar_full_test.root type on the command line:
your_pc> gmake; your_pc> ./simpleNtupleTTbarAnalysis.C 1 100 test
When setting the number of events to "-1", all events are processed. If you want to analyze all events in ntuple-set 2 (W+4jets background) and to store the output histograms in a file called summary_wjets_full_test.root simly execute the following line:
your_pc> ./simpleNtupleTTbarAnalysis.C 2 -1 "test"
Structure of the program
The '.C'-file is documented and should speak for itself. Nevertheless it is always nice to get a little bit more information on the logic behind the program:
Step 1) Reading the full event information into memory
Step 2) fillAllVectors(): This routine fills for each type of 'pseudo particle' (electrons, muons, photons, jets) a vector of TLorentzvectors-with-index to make the analysis more abstract and easily readable. There are actually TWO vectors filled, one for 'all' and one for 'good' pseudo-particles. For electrons we have for example ElectronVec and GoodElectronVec and it is up to the user to decide for each electron if it is good enough to be included in the 'good' electron list.
The routine that decides that is also contained in this file and is called isGoodElec. For muons, photons and jets we have similar routines. In these routines the 'all'-verctors are filled and for muons you can ask for example that it is isolated from all jets thay you have selected.
Standard quality plots: Finally some standard plots are produced for the various pseudo particles. These are the routines you can use to compute efficiencies or simply check what are the characteristics of the objects you have selected. For electrons we have electronQA(i_event), with smiilar routines for other psudo-particles.
If Print_Info is set to 1, information on the MC truth record and the characteristics of ALL jets before selection is print to screen.
Step 3) Computes the missing transverse energy. Used both in the event selection and estimation of the leptonic top mass candidate.
Step 4) The event selection. At this moment at lease one good lepton, exactly 4 good jets and large missing transverse energy.
If Print_Info is set to 1, information on the characteristics of good electrons and muons after selection is print to screen.
Step 5) The selected events are treated as ttbar candidates. Here a few plots are produced that were also shown during the Rome meeting. The normalisation of the plots should be done by you of course (some numbers on this later here).
- a) Plot the 3-jet mass for all combinations and the define the combination with the highest vector sum Pt as the top candidate. Millions of other algorithms are possible and we will investigate this to see if we can come up with something a bit more sophisticated.
- b) For the top-candidate you have defined in a), plot the top mass if you (don't) find a 2-jet mass combination close to the W mass.
- c) Decide on the W combination and plot the both 1) W mass in top signal and sideband window ands 2) the top mass in the W signal and side-band region.
Step 6) Write all histograms to a file.
Location of the Ntuples & AOD content
The location of the Ntuples for all samples can be found at the: Overview Rome Monte Carlo data samples
For information on the content of the AOD (Ntuple) read: AOD (Ntuple) content (Marcello Barisonzi)
A prescription on how to producing the Ntuple from the AOD: producing the Ntuple from the AOD. (Version 6) Ntuples .
Questions or remarks ?
If you have questions or remarks please let us know.
Wouter Verkerke & Ivo van Vulpen