Pizero Calibration Producer: Simple Clustering vs Island Basic Clusters

1
Pizero Calibration Producer Simple Clustering vs
Island Basic Clusters

• Vladimir Litvin
• Caltech, CMS

2
Intro

• Producer is ready, committed to CVS (HEAD
  revision). Two versions are available for quality
  checks
• Based on ECAL RecHits only
• Based on ECAL RecHits and Island Basic Clusters
• Special patch to reconstruct clusters in ECAL
  Barrel with Island Basic Clusters based on
  produced collections was done.
• First CMSSW MC QCD background events were
  analyzed to reproduce ORCA results. Statistics is
  too small.
• Official MC production request is the top
  priority for us
• Ecms 900 GeV
• Ecms 14 TeV

3
Software

• Data Producer to select pizero candidates based
  on RecHits either from special EBRecHits data
  collection or based on the whole ECAL Barrel
  RecHits collection if time will allow us to do
  this (Caltech)
• Input will be either RomaEBRecHits or full ECAL
  Barrel RecHits collection. We can also use Island
  Basic Clusters for quality checks
• Both Simple and Island Basic Clustering will be
  possible to apply.
• Simple clustering (XSxYS sliding window default
  is 3x3) means that each cluster will be treated
  as a photon candidate and invariant mass of two
  candidates will be calculated. If invariant mass
  is smaller than 200 MeV, all RecHits from input
  Collection in X1xY1 window (default is 21x21)
  around centers of two selected candidates will be
  stored in a separate file to transfer to
  calibration farm
• Output will be calibration pi0CaEBRecHits
  collection
• Timing and bandwidth measurements are important

4
Test setup

• There are two methods how to produce pi0
  calibration collection
• Based on ECAL Barrel RecHits only (Scalib),
  where clustering is done based on simple 3x3
  window
• Based on ECAL Barrel RecHits and Island Basic
  Cluster collection from RECO (BCcalib)
• There are also two methods to reconstruct
  clusters from the output data of the calibration
  stream
• Using simple clustering with 3x3 window
  (Sanalysis)
• Using patched CMSSW 1.2.0 version and default
  algorithm for Island Basic Clusters in ECAL
  Barrel (BCanalysis)
• There are four possible combinations here
• Scalib Sanalysis (worst quality)
• Scalib BCanalysis (most probably will be in
  use)
• BCcalib Sanalysis (least probably will be in
  use)
• BCcalib BCanalysis (best quality)

5
Calibration selection (BCcalib)

• Calibration selection
• Reading EcalRecHitsEB and islandBarrelBasicCluster
  s collections from RECO.
• Make a loop over Island Basic Clusters and
  identify Pi0 candidates. Those candidates are
  each pair of Island Basic Clusters, passed
  through next simplified selection
• Apply two cuts on Pt of each photon candidates
  Pt_gamma1 gt selePtGammaOne_ (default 1 GeV),
  Pt_gamma2 gt selePtGammaTwo_ (default 1 GeV)
  (where gamma1 is a Island Basic Cluster with
  highest Pt, gamma2 - Island Basic Cluster with
  second highest Pt)
• Apply cut on Pt of Pi0 candidate Pt_Pi0 gt
  selePtPi0_ (default 2.5 GeV)
• Apply cut on invariant mass of two photons
  (invariant mass of Pi0 candidate)
  seleMinvMinPi0_ lt Minv_Pi0 lt seleMinvMaxPi0_
  seleMinvMinPi0_ default value is 0.,
  seleMinvMaxPi0_ default vaue is 0.2 GeV
• Each pair, passed through this selection will be
  treated as good Pi0 candidate and will be marked
  to store in the separate file alCaPi0BC.root
• Most energetic crystals were identified in both
  Island Basic Clusters.
• All ECAL Barrel RecHits in gammaCandEtaSize_ x
  gammaCandPhiSize_ matrix (default is 21x21)
  around most energetic crystals in both clusters
  will be stored in the file as well.

6
Calibration selection (Scalib- I)

• Calibration Selection
• Reading EcalRecHitsEB collection from RECO.
• Make a loop over RecHits to prepare simple
  clusters. All RecHits with energies larger than
  clusSeedThr_ (default 0.5 GeV) will be
  considered as seeds and sorted with energy. After
  that all RecHits in the window clusEtaSize_ X
  clusPhiSize_ (default 3x3), will be considered as
  a cluster. If there is an overlap between
  clusters, RecHit will belong to the cluster with
  higher energy (it means, that there is no RecHit,
  which belongs to both clusters).

7
Calibration selection (Scalib - II)

• Calibration Selection (contd)
• Based on the vector of these simple clusters,
  further preselection will be done to identify the
  pi0 candidates. Those candidates are pairs of
  Simple clusters, passed through next simplified
  selection
• Apply two cuts on Pt of each photon candidates
  Pt_gamma1 gt selePtGammaOne_ (default 1 GeV),
  Pt_gamma2 gt selePtGammaTwo_ (default 1 GeV)
  (where gamma1 is a Island Basic Cluster with
  highest Pt, gamma2 - Island Basic Cluster with
  second highest Pt)
• Apply cut on Pt of Pi0 candidate Pt_Pi0 gt
  selePtPi0_ (default - 2.5 GeV)
• Apply cut on invariant mass of two photons
  (invariant mass of Pi0 candidate)
  seleMinvMinPi0_ lt Minv_Pi0 lt seleMinvMaxPi0_
  seleMinvMinPi0_ default value is 0.,
  seleMinvMaxPi0_ default vaue is 0.2 GeV
• Each pair, passed through this selection will be
  treated as good Pi0 candidate and will be marked
  to store in the separate file alCaPi0.root
• All ECAL Barrel RecHits in gammaCandEtaSize_ x
  gammaCandPhiSize_ matrix (default values are
  21x21) around most energetic crystals in both
  clusters will be stored in the file.

8
Analysis Selection (BCanalysis)

• Calibration RecHits collection was read
• Island Basic Clusters were reconstructed in the
  Barrel (patched version of CMSSW was used)
• Make a loop over Island Basic Clusters and
  identify Pi0 candidates. Those candidates are
  each pair of Island Basic Clusters, passed
  through next simplified selection
• Apply two cuts on Pt of each photon candidates
  Pt_gamma1 gt selePtGammaOne_ (default 1 GeV),
  Pt_gamma2 gt selePtGammaTwo_ (default 1 GeV)
  (where gamma1 is a Island Basic Cluster with
  highest Pt, gamma2 - Island Basic Cluster with
  second highest Pt)
• Apply cut on Pt of Pi0 candidate Pt_Pi0 gt
  selePtPi0_ (default 2.5 GeV)
• Each pair, passed through this selection will be
  treated as good Pi0 candidate

9
Simple Analysis Selection (Sanalysis)

• Calibration RecHits collection was read
• Simple Clusters were reconstructed in the Barrel.
  Same algorithm as in Scalib was used.
• Make a loop over Simple Clusters and identify Pi0
  candidates. Those candidates are each pair of
  Island Basic Clusters, passed through next
  simplified selection
• Apply two cuts on Pt of each photon candidates
  Pt_gamma1 gt selePtGammaOne_ (default 1 GeV),
  Pt_gamma2 gt selePtGammaTwo_ (default 1 GeV)
  (where gamma1 is a Island Basic Cluster with
  highest Pt, gamma2 - Island Basic Cluster with
  second highest Pt)
• Apply cut on Pt of Pi0 candidate Pt_Pi0 gt
  selePtPi0_ (default 2.5 GeV)
• Each pair, passed through this selection will be
  treated as good Pi0 candidate

10
Data rates for 21x21 window

• Rates are the same for both algorithms

11
QCD pthat 0-15 GeV noPU, Ptpi0 gt 2.5 GeV
2100 pi0s in the peak
1600 pi0s in the peak
2100 pi0s in the peak
1.0 of total events
ScalibBCanalysis Size 213MB/213k ev
ScalibSanalysis Size 213MB/213k ev
BCcalibBCanalysis Size 213MB/213k ev
Total number of events in this dataset - 213000
12
QCD pthat 15-20 GeV noPU, Ptpi0 gt 2.5 GeV
1600 pi0s in the peak
12500 events in the peak
13000 events in the peak
9200 events in the peak
6.6 of total events
ScalibBCanalysis Size 216MB/188k ev
ScalibSanalysis Size 216MB/188k ev
BCcalibBCanalysis Size 216MB/188k ev
Total number of events in this dataset - 188000
13
QCD pthat 20-30 GeV noPU, Ptpi0 gt 2.5 GeV
5600 events in the peak
6000 events in the peak
6000 events in the peak
8.7 of total events
ScalibBCanalysis Size 98MB/69k ev
ScalibSanalysis Size 98MB/69k ev
BCcalibBCanalysis Size 98MB/69k ev
Total number of events in this dataset - 69000
14
QCD pthat 30-50 GeV noPU, Ptpi0 gt 2.5 GeV
23000 events in the peak
20000 events in the peak
20000 events in the peak
9.1 of the total events
ScalibBCanalysis Size 418MB/220k ev
ScalibSanalysis Size 418MB/220k ev
BCcalibBCanalysis Size 418MB/220k ev
Total number of events in this dataset - 220000
15
QCD pthat 50-80 GeV noPU, Ptpi0 gt 2.5 GeV
40000 events in the peak
37000 events in the peak
40000 events in the peak
9.2 of the total events
ScalibBCanalysis Size 1.32GB/400k ev
ScalibSanalysis Size 1.32GB/400k ev
BCcalibBCanalysis Size 1.32GB/400k ev
Total number of events in this dataset - 400000
16
TODO

• Timing measurements
• Simple clustering algorithm performance vs the
  window size (right now default is 21x21)
• 10x10, 15x15
• Rectangular windows? 10(eta)x15(phi)
• Simple clustering algorithm optimization
• Boundary effects
• L1 trigger study right now L1 trigger is a
  mess, but will try anyway.
• Analysis selection optimization
• ECAL isolation
• Redo the same with PU 21033
• ECAL barrel digi producer.