Tools for btagging Validation and Soft Electron bjet Tagging in CMS

1
Tools for b-tagging Validation and Soft Electron
b-jet Tagging in CMS Jason Keller University of
Nebraska-Lincoln
Introduction In experimental high energy
physics, the capability to tag jets produced from
bottom (b) quarks has become a necessity. Many
standard model and new physics processes have a
b-jet signature, including top quarks, the Higgs
boson, and supersymmetry. It is important to
validate the performance of the b-tagging
algorithms between CMSSW releases to ensure all
tools are working optimally. It is also possible
to utilize the particle flow reconstruction in
CMSSW to develop simple taggers utilizing the
semileptonic decay of b quarks to electrons.
Both tools are described below in detail.
Tools for b-tagging Validation in CMS

• It is important to compare the performance of
  b-tagging algorithms between releases to note the
  effect of any changes made.
• The LPC bId group is very active in the
  development of these tools for CMSSW.
• The validation tools use the DQMOffline framework
  to produce plots for the performance of b-tagging
  algorithms in a current release.
• The validation framework can produce plots
  showing the mistag rate vs. b-efficiency, the
  b-efficiency/mistag rate vs. discriminator cut,
  and the discriminator for a given tagger.
• Python scripts (developed by F. Yumiceva,
  Fermilab) are used to overlay performance plots
  of two releases, and to publish these plots to
  the web for viewing.
• Results are reported to the CMS Release
  Validation group to monitor the effect changes
  between releases have on performance.
• I would like to thank F. Yumiceva (Fermilab) and
  P. Jindal (U. of Purdue, Calumet) at the LPC for
  their continued help and support in the
  development and use of these validation tools.

Figure 3 c-jet mistag rate vs. b-jet efficiency
for the various bTagging algorithms in
CMSSW_3_1_4.
Figure 1 Comparison of the simple secondary
vertex tagger performance between CMSSW_3_1_4 and
3_1_3.
Figure 2 Comparison of the simple secondary
vertex discriminator for various jet flavors
between CMSSW_3_1_4 and 3_1_3.
Figure 4 udsg-jet mistag rate vs. b-jet
efficiency for the various bTagging algorithms in
CMSSW_3_1_4.
Tools for Soft Electron b-jet Tagging in CMS

• Soft lepton tagging (for both muons and
  electrons) utilizes the large branching ratio of
  b quarks to leptons compared to lighter quarks.
• Soft electron tagging is made possible by the
  particle flow framework in CMSSW to reconstruct
  low pt, non-isolated electrons.
• Two variables are used to discriminate between b
  jets and clight jets pT of the electron and
  the signed impact parameter significance, both
  with respect to the jet axis.
• Additional cut-based electron ID is performed to
  suppress background due to pions/kaons, and
  electrons from conversions.
• The cuts are optimized using ROOTs TMVA package
  with distributions from non-isolated electrons in
  ttbar events.
• Variables used for soft electron ID are
• Supercluster Energy / Momentum at Vertex.
• DeltaR between track and supercluster
• Length of track.
• Transverse distance from beam line to first
  tracker hit.
• Longitudinal distance from center of detector to
  first tracker hit.
• The ratio of brem energy to momentum at vertex.

Figure 4 Discriminator for the soft electron
tagger based on the electron pT with respect to
the jet axis.
Figure 4 Discriminator for the soft electron
tagger based on the electron signed impact
parameter significance with respect to the jet
axis.
Figure 4 Performance of the soft electron
tagger based on the electron signed impact
parameter significance with respect to the jet
axis.
Figure 5 Performance of the soft electron
tagger based on the electron pT with respect to
the jet axis.