Pileup suppression methods
Methods to suppress the pileup in Etmiss Soft Term (Soft Jets + CellOutEflow) have been developed and they are documented in
the 2011 Etmiss CONF Note:
CDS link.
Track-based methods (STVF) and
jet area method have been tested with NTUP_JETMET D3PDs with special setup and now implemented in Athena.
Both the STVF method and the jet area method are ready to be released for physics analysis.
Run I Instructions to run MET_RefFinal with PU suppression from AOD/ESD to D3PD from AtlasPhysics 17.2.7.4.1 onwards
From cache
AtlasPhysics 17.2.7.5.2 onwards the tags
MissingET-03-03-60 and
MissingETEvent-00-41-23 are included.
Running the default MET reconstruction users will get a new default MET_RefFinal and MET_RefFinal_STVF and new additional cases for MET_RefJet and MET_CellOut_Eflow terms where the PU suppression is applied.
Users do not have to set any special flag to get MET_RefFinal_STVF and the PU suppressed terms.
In particular the new MET tags contain following changes for MET_RefFinal and MET_RefFinal_STVF:
- NEW default MET_RefFinal:
- MET_RefJet term is calculated from the 4-momentum of jets (including offset subtraction)
- MET_SoftJets is no more calculated, the contribution of the soft jets is included in the CellOut_Eflow term.
- NEW PU suppressed MET_RefFinal_STVF:
- obtained after PU suppression in the jet and soft terms, e.g. from RefJet_JVFCut and MET_CellOut_Eflow_STVF (see explanation below).
- obtained applying the PU correction to the default MET_RefFinal magnitude and also to its direction. This is different with respect to the previous version of MET_RefFinal_STVF in which only the MET magnitute was corrected. Users can eventually choose to use the default MET_RefFinal direction at analysis level.
- MET_RefFinal_STVF is calculated as follows: MET_RefFinal_STVF_x(y) = MET_RefEle_x(y) + MET_RefGamma_x(y)+ MET_RefTau_x(y) + MET_RefJet_JVFCut _x(y) + MET_CellOut_Eflow_STVF_x(y) + MET_MuonBoy_x(y)
In parallel the following PU suppressed MET terms are available:
- PU suppressed objects for MET_CellOutEflow term:
- MET_CellOut_Eflow_STVF: STVF PU suppression is applied
- MET_CellOut_Eflow_JetArea: Jet Area PU suppression is applied
- MET_CellOut_Eflow_JetAreaJVF: Jet Area PU suppression is applied plus jvf cut
- MET_CellOut_Eflow_JetAreaRhoEta5JVF: Jet Area PU suppression with a different rho is applied plus jvf cut
- PU suppressed objects for MET_RefJet term:
- MET_RefJet_JVF: obtained weighting each jet contributing to MET_RefJet with its |JVF|
- MET_RefJet_JVFCut: obtained applying a JVF cut (|JVF| >0) to each jet contributing to MET_RefJet
The same tags have to be used when analysis is performed from AODs .
For users who like to run a
customized MET_RefFinal and MET_RefFinal_STVF, they have to include in their jobOption a new Getter called:
METRefGetter_plup.py.
They just need to configure the MET_RefFinal as they like and they will get also the correspondig MET_RefFinal_STVF plus the PU suppressed objects without setting any special flag for STVF configuration.
In your customized MET_RefFinal you have to set for MET_RefJet : jet_calibType = 'ExclRefCalib' to have the MET_ReJet calculated from the 4-momentum of jets as in the new default. If you like to keep the old configuration you have to set for MET_RefJet : jet_calibType ='LocHad' and jet_ApplyJetScale ='Yes'.
Users don't have to dump anymore in their D3PDs the composition map of STVF configuration, because new status words have been added to allocate in the composition map of MET_RefFinal the weights for the new PU suppressed objects. So, they should save only the composition map of the default MET_RefFinal or of the customized MET_RefFinal.
The METUtility package has been updated to take into account those changes in the composition map, you can find all info in the
METUtility wiki.
In cache
AtlasPhysics 17.2.7.5.1 the tags
MissingET-03-03-45 and
MissingETEvent-00-41-22 are included. It contains some configuration changes to make the other MET objects like MET_RefFinal_em, _muid,_muons more
coherent with the new default MET_RefFinal with MET_RefJet term calculated from the 4-momentum of jets and MET_CellOut_Eflow including also the soft jets.
In cache
AtlasPhysics 17.2.7.4.1 the tags
MissingET-03-03-44 and
MissingETEvent-00-41-22 are included.
In caches
AtlasPhysics 17.2.7.4.1 and
AtlasPhysics 17.2.7.5.1 the same MET objects are provided as in the
AtlasPhysics 17.2.7.5.2 except for the MET PU terms: MET_CellOut_Eflow_JetAreaJVF and
MET_CellOut_Eflow_JetAreaRhoEta5JVF.
OLD Instructions to run MET_RefFinal with STVF pileup-suppression from AOD/ESD to D3PD
This STVF (Softerm Vertex Fraction) pileup correction is available since branch tag:
MissingET-03-03-31-04 with specific changes in configuration for MissingETD3PDmaker, included in
AtlasPhysics 17.2.2.4.2.
Do NOT use tag: MissingET-03-03-31-06 ( in which a wrong gamma identification in RefGamma term has been ported due to frozen T0 policy).
For NTUP_JETMET D3PD production (p1051,p1052) a previous tag is used:
MissingET-03-03-31 and included in
AtlasPhysics 17.2.2.3.1.
WARNING: a small problem in the direction of RefFinal_STVF found in this tag. It can be recalculated from the new objects dumped in D3PDs as explained below.
This is fixed in the tag:
MissingET-03-03-31-04
To run the pileup correction the following jobOption should be used:
metAlgs.py.
With this configuration you will get in D3PDs the
MET_RefFinal_STVF quantity with all Ref*_STVF terms. Note that the SoftJets term has been switched off.
With this configuration 3 more objects should be dumped in D3PDs with respect to the usual RefFinal terms, called:
MET_CellOutCorr, MET_CellOut_Track, MET_CellOut_TrackPV.
The MET_CellOutCorr is the corrected CellOutEflow term with STVF.
The MET_CellOut_Track is calculated from all tracks unmatched to physics objects
The MET_CellOut_TrackPV is calculated from all tracks unmatched to physics objects and from the primary vertex.
With the above configuration those objects will appear in D3PDs with the names: MET_CellOutCorr_STVF, MET_CellOut_Track_STVF,
MET_CellOut_TrackPV_STVF.
The MET_RefFinal_STVF provides also the
MET_RefJet_STVF term corrected with JVF , the MET_SoftJets term is no more calculated and its contribution is included
in the MET_CellOutCorr_STVF term. The other terms are calculated as in the defaul MET_RefFinal.
The new calculation for RefFinal_STVF is the following:
MET_Ex(y)MissRefFinal_STVF =MET_Ex(y)MissRefEle_STVF + MET_Ex(y)MissRefGamma_STVF + MET_Ex(y)MissRefTau_STVF + MET_Ex(y)MissRefJet_STVF +MET_Ex(y)MissRefMuon_STVF + MET_Ex(y)MissCellOutCorr_STVF
+ MET_Ex(y)MissMuonBoy_STVF
The STVF/JVF correction is applied only to the default MET_RefFinal magnitude and not to its direction that is kept the same.
For users who would like to correct also the MET_RefFinal direction, they can use the new dumped objects
and calculate the correction directly from those objects like that:
STVF_weight = MET_CellOut_TrackPV ->sumet()/ MET_CellOut_Track->sumet()
and rescale the MET_CellOut_STVF term (that is NOT corrected) with this weight.
To provide all needed information for METRefFinal_STVF recalculation and systematics studies with METUtility package
also the STVF composition
map should be saved in D3PDs .
--
SilviaResconi - 09-May-2012