Heavy Flavor Overlap Removal Tool in Run2


This page describes the use of the Heavy Flavor Overlap removal Tool to be used in the Alpgen+Pythia6 mc15 production. The present version of the tool is available in and is heavily based on the Run1 angular based overlap removal tool (link). In this process, the algorithm will first classify the heavy quarks (b,c) as comming either from Matrix Element (ME) or Parton Shower (PS) and then determine the angular distance (DeltaR) between pairs of opposite charge quarks. There are two strategies implemented in this tool :

DeltaR based overlap removal

If a pair of quark classifyed as ME occupies the PS phase space (DeltaR<0.4) the event is vetoed. Simillarly, events with opposite charge heavy quarks from PS that occupies the ME phase space (DeltaR>0.4) are vetoed. Please notice that the tool operates in the xAOD Truth containers, which must be available in the derivation you are using ( xAOD::TruthParticle , xAOD::TruthVertex and xAOD::TruthEvent ).

Jet Based overlap removal

The procedure is the same as in DRBased, except that b and c quarks are identified from the reconstructed TruthJets ( xAOD::TruthParticle , xAOD::TruthVertex, xAOD::TruthEvent and xAOD::AntiKt4TruthWZJets=).

Tool usage

It is very important that the TRUTH container in the derivation contains the full event tree (and the AntiKt4TruthWZJets container for Jet Based HFOR). Some versions of truth thinning service may not contain the c quarks which will lead to incorrect behavior of the tool . It is always recommended that you check against the table bellow if the yelds you obtained are consistent. STDM4 derivations with tag p3020 are known to work.

The tool itself (HFORSelectionTool) can be used in both Athena Based and RC analysis. Currently, the only user configuration for the tool accepts are the following :

HFORSelectionTool::setProperty("HFORStrategy", "DRBased") : HFOR using the b,c quarks from Truth MC event record (default)

HFORSelectionTool::setProperty("HFORStrategy", "JetBased") : HFOR using the b,c quarks from TruthJets MC event record

HFORSelectionTool::myTool.setProperty("MatchCone", 0.4) : HFOR cone size for removal (default)

HFORSelectionTool::setProperty("runConfigFile", "HFORTools/mc15_AlpgenPythia_2016.cfg") : Name of configuration file used to match the MC sample channel number/ run number, see bellow (default)

HFORSelectionTool::setProperty("TruthJetCollectionName", "AntiKt4TruthWZJets" ): Name of TruthJets collection to be used in the JetBased HFOR strategy

The HFOR tool needs to know the type of the current MC sample it will operate (light, V+bb, V+cc or V+c). This is done using a configuration file where the run number is matched to the specific process used in the event generation. This file is located in HFORTool/data/mc15_AlpgenPythia_2016.cfg and will be accessed by default unless the user points to another configuration file during the initialization process in RC or in his Athena jobOption file.

The tool automatically checks the file type at every file change to make sure the correct definition of the file is used (see configuration file definition above). If the file type is not a recognized Alpgen+Pythia6 mc15 file, the isSelected method will bypass any overlap procedure and return always true for all events (so, it's safe to call the tool regardless of the file type).

Checking out the tool

The HFORTool will soon be part of AnalysisBase. For now, it can be check out from svn :

svn co svn+ssh://svn.cern.ch/reps/atlasoff/PhysicsAnalysis/AnalysisCommon/PMGOverlapRemovalTools/HFORTools/tags/HFORTools-04-00-07  HFORTools

A companion test package with some usage examples is also available from svn :

svn co svn+ssh://svn.cern.ch/reps/atlasoff/PhysicsAnalysis/AnalysisCommon/PMGOverlapRemovalTools/Test_HFORTools/tags/Test_HFORTools-04-00-05 Test_HFORTools

How to use the tool in RC based analysis

See the example steering routine on how to use the tool in Test_HFORTools.cxx

If you do your analysis in python, an example is also available here selection.py

How to use the tool in Athena based analysis

For a detailed algorithm exemplifying the use of the tool, please see Test_HFORToolsAlg.cxx and the jobOption file Test_HFORToolsAlgJobOptions.py


HFOR effect on samples

The table bellow presents the percentage of events killed after applying HFOR procedure on each sample, for TRUTH0 derivations (T0), TRUTH1 derivation (T1), TRUTH6 derivation (T6) and STDM4 derivations (STDM4). In the STDM4, numbers are presented for DR based (STDM4_DR) and Jet Based (STDM4_Jet) strategies .


Since events are removed, the cross section from AMI needs to be recalculated taking into account the new number of events. The table bellow lists the modified cross sections fro the AlpgenPythia6 samples after the HFOR. The kFactor for the combination are :

- Z light+HF : k = 1.196

- W light+HF: k = 1.044 (to be updated)


-- MarcoALeite - 2016-07-04

Topic attachments
I Attachment History Action Size Date Who Comment
PNGpng AlpgenPythia6_xSections.png r2 r1 manage 765.5 K 2016-08-29 - 11:04 MarcoALeite AlpgenPythia6 xSection
PNGpng AlpgenPythia6_xSections_2.png r1 manage 765.5 K 2016-08-29 - 11:08 MarcoALeite AlpgenPythia6 xSection
PNGpng HFOR_selection_factor.png r1 manage 647.3 K 2017-03-28 - 21:21 MarcoALeite HFOR selection factors for each Alpgen+Pythia6 sample
PNGpng HFOR_selection_factor2.png r1 manage 625.0 K 2017-03-28 - 21:30 MarcoALeite  
PNGpng HFOR_xSection.png r1 manage 399.8 K 2017-03-31 - 12:46 MarcoALeite  
Edit | Attach | Watch | Print version | History: r9 < r8 < r7 < r6 < r5 | Backlinks | Raw View | WYSIWYG | More topic actions
Topic revision: r9 - 2017-04-10 - MarcoALeite
    • Cern Search Icon Cern Search
    • TWiki Search Icon TWiki Search
    • Google Search Icon Google Search

    Sandbox All webs login

This site is powered by the TWiki collaboration platform Powered by PerlCopyright & 2008-2023 by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
or Ideas, requests, problems regarding TWiki? use Discourse or Send feedback