Identifying Hadronically Decaying W Bosons Merged into a Single Jet

Paper published as JHEP 12 (2014) 017 PAS-JME-13-006 in CDS

Abstract

In the new energy regime of the LHC, it is becoming increasingly important for new physics searches to identify single jet objects which originate from and contain a hadronically decaying massive W boson produced with high transverse momentum. A number of observables are explored for identifying these ”W-jets”. An algorithm is defined to identify jets originating from hadronically decaying W bosons for different signal scenarios. The efficiency for tagging W-jets is presented and comparisons are made between data and simulation. All the techniques discussed in identifying W-jets are applicable for other hadronic two-prong decays such as those from Z bosons and Higgs.

Plots from the paper JME-13-006 ( click on plot to get .pdf )

Simulation plots

Pileup and detector effects

	Pruned jet mass distribution in simulated samples of boosted W bosons and inclusive QCD jets in the W+jet topology. MG denotes the Madgraph5 generator. Thick dashed lines represent the generator predictions without pileup interactions and without CMS simulation. The histograms are the distributions after CMS simulation with two different pileup scenarios corresponding to an average number of interactions of 12 and 22.
	The mass drop distribution in simulated samples of boosted W bosons and inclusive QCD jets before a cut on the pruned jet mass. MG denotes the Madgraph5 generator. Thick dashed lines represent the generator predictions without pileup interactions and without CMS simulation. The histograms are the distributions after CMS simulation with two different pileup scenarios corresponding to an average number of interactions of 12 and 22.
	The mass drop distribution in simulated samples of boosted W bosons and inclusive QCD jets after a cut on the pruned jet mass. MG denotes the Madgraph5 generator. Thick dashed lines represent the generator predictions without pileup interactions and without CMS simulation. The histograms are the distributions after CMS simulation with two different pileup scenarios corresponding to an average number of interactions of 12 and 22.
	The C2 beta distribution in simulated samples of boosted W bosons and inclusive QCD jets before a cut on the pruned jet mass. MG denotes the Madgraph5 generator. Thick dashed lines represent the generator predictions without pileup interactions and without CMS simulation. The histograms are the distributions after CMS simulation with two different pileup scenarios corresponding to an average number of interactions of 12 and 22.
	The C2 beta distribution in simulated samples of boosted W bosons and inclusive QCD jets after a cut on the pruned jet mass. MG denotes the Madgraph5 generator. Thick dashed lines represent the generator predictions without pileup interactions and without CMS simulation. The histograms are the distributions after CMS simulation with two different pileup scenarios corresponding to an average number of interactions of 12 and 22.
	The QJets volatility distribution in simulated samples of boosted W bosons and inclusive QCD jets before a cut on the pruned jet mass. MG denotes the Madgraph5 generator. Thick dashed lines represent the generator predictions without pileup interactions and without CMS simulation. The histograms are the distributions after CMS simulation with two different pileup scenarios corresponding to an average number of interactions of 12 and 22.
	The QJets volatility distribution in simulated samples of boosted W bosons and inclusive QCD jets after a cut on the pruned jet mass. MG denotes the Madgraph5 generator. Thick dashed lines represent the generator predictions without pileup interactions and without CMS simulation. The histograms are the distributions after CMS simulation with two different pileup scenarios corresponding to an average number of interactions of 12 and 22.
	The n-subjettiness tau2/tau1 distribution in simulated samples of boosted W bosons and inclusive QCD jets before a cut on the pruned jet mass. MG denotes the Madgraph5 generator. Thick dashed lines represent the generator predictions without pileup interactions and without CMS simulation. The histograms are the distributions after CMS simulation with two different pileup scenarios corresponding to an average number of interactions of 12 and 22.
	The n-subjettiness tau2/tau1 distribution in simulated samples of boosted W bosons and inclusive QCD jets after a cut on the pruned jet mass. MG denotes the Madgraph5 generator. Thick dashed lines represent the generator predictions without pileup interactions and without CMS simulation. The histograms are the distributions after CMS simulation with two different pileup scenarios corresponding to an average number of interactions of 12 and 22.
	The n-subjettiness tau2/tau1 with exclusive kT axis distribution in simulated samples of boosted W bosons and inclusive QCD jets before a cut on the pruned jet mass. MG denotes the Madgraph5 generator. Thick dashed lines represent the generator predictions without pileup interactions and without CMS simulation. The histograms are the distributions after CMS simulation with two different pileup scenarios corresponding to an average number of interactions of 12 and 22.
	The n-subjettiness tau2/tau1 with exclusive kT axis distribution in simulated samples of boosted W bosons and inclusive QCD jets after a cut on the pruned jet mass. MG denotes the Madgraph5 generator. Thick dashed lines represent the generator predictions without pileup interactions and without CMS simulation. The histograms are the distributions after CMS simulation with two different pileup scenarios corresponding to an average number of interactions of 12 and 22.
	The n-subjettiness tau2/tau1 with pruning distribution in simulated samples of boosted W bosons and inclusive QCD jets before a cut on the pruned jet mass. MG denotes the Madgraph5 generator. Thick dashed lines represent the generator predictions without pileup interactions and without CMS simulation. The histograms are the distributions after CMS simulation with two different pileup scenarios corresponding to an average number of interactions of 12 and 22.
	The n-subjettiness tau2/tau1 with pruning distribution in simulated samples of boosted W bosons and inclusive QCD jets after a cut on the pruned jet mass. MG denotes the Madgraph5 generator. Thick dashed lines represent the generator predictions without pileup interactions and without CMS simulation. The histograms are the distributions after CMS simulation with two different pileup scenarios corresponding to an average number of interactions of 12 and 22.
	The jet charge distribution in simulated samples of boosted W bosons and inclusive QCD jets after a cut on the pruned jet mass. MG denotes the Madgraph5 generator. Thick dashed lines represent the generator predictions without pileup interactions and without CMS simulation. The histograms are the distributions after CMS simulation.
	Number of jet constituents distributions at generator level, with additional PU interactions, with additional PU interactions where charged particles from pileup are removed (CHS), and at detector level with CMS simulation.
	tau2/tau1 at generator level, with additional PU interactions, with additional PU interactions where charged particles from pileup are removed (CHS), and at detector level with CMS simulation.

W polarization

	Generator level cos Theta_J distributions for purely longitudinally and transversely polarized W bosons
	Pruned jet mass for pure WL and WT samples for jets with 250 < pT < 350 GeV for a resonance mass of 600 GeV
	Pruned jet mass for pure WL and WT samples for jets with 250 < pT < 350 GeV for a resonance mass of 1000 GeV
	Subjet angular observable, cos Theta_J, after a cut on pruned jet mass for pure WL and WT samples for jets with 250 < pT < 350 GeV for a resonance mass of 600 GeV
	Subjet angular observable, cos Theta_J, after a cut on pruned jet mass for pure WL and WT samples for jets with 250 < pT < 350 GeV for a resonance mass of 1000 GeV
	Subjet pT ratio observable after a cut on pruned jet mass for pure WL and WT samples for jets with 250 < pT < 350 GeV for a resonance mass of 600 GeV
	Subjet pT ratio observable after a cut on pruned jet mass for pure WL and WT samples for jets with 250 < pT < 350 GeV for a resonance mass of 1000 GeV
	Mass drop observable after a cut on pruned jet mass for pure WL and WT samples for jets with 250 < pT < 350 GeV for a resonance mass of 600 GeV
	Mass drop observable after a cut on pruned jet mass for pure WL and WT samples for jets with 250 < pT < 350 GeV for a resonance mass of 1000 GeV
	Qjet volatility observable after a cut on pruned jet mass for pure WL and WT samples for jets with 250 < pT < 350 GeV for a resonance mass of 600 GeV
	Qjet volatility observable after a cut on pruned jet mass for pure WL and WT samples for jets with 250 < pT < 350 GeV for a resonance mass of 1000 GeV
	Tau2/tau1 observable after a cut on pruned jet mass for pure WL and WT samples for jets with 250 < pT < 350 GeV for a resonance mass of 600 GeV
	Tau2/tau1 observable after a cut on pruned jet mass for pure WL and WT samples for jets with 250 < pT < 350 GeV for a resonance mass of 1000 GeV

Quark-gluon composition

	Pruned jet mass distribution for quark and gluon jet simulated with PYTHIA6.
	Pruned jet mass distribution for quark and gluon jet simulated with PYTHIA6.

High pT behavior

	Observable tau2/tau1 comparison for signal and background in the very high pT regime compared to more moderately boosted jets.
	Comparison of mass drop observable before a mass cut for signal and background in the very high pT regime compared to more moderately boosted jets.
	Comparison of mass drop observable after a mass cut for signal and background in the very high pT regime compared to more moderately boosted jets.
	Comparison of tau2/tau1 observable before a mass cut for signal and background in the very high pT regime compared to more moderately boosted jets.
	Comparison of tau2/tau1 observable after a mass cut for signal and background in the very high pT regime compared to more moderately boosted jets.

MVA and ROC curves

	Comparison of various discriminant observable performance for W+jet events in the low jet pT bin, 250-350 GeV.
	Comparison of jet substructure observable performance for W+jet events in the jet pT bin, 250-350 GeV, for longitudinally and transversely polarized signal W-jets.
	Comparison of jet substructure observable performance for W+jet events in the jet pT bin, 400-600 GeV, for longitudinally and transversely polarized signal W-jets.
	Correlation matrix for the input variables given to the MVA evaluated on signal sample (ggH at mH = 600 GeV)
	Correlation matrix for the input variables given to the MVA evaluated on background (W+jets Pythia pT >180 GeV)

Data/MC comparisons

dijet events

	Pruned jet mass distribution in data and simulation for dijet events. MG denotes the Madgraph5 generator.
	Mass drop distribution in data and simulation for dijet events without a mass cut. MG denotes the Madgraph5 generator.
	Mass drop distribution in data and simulation for dijet events with a mass cut 60-100 GeV. MG denotes the Madgraph5 generator.
	N-subjettiness tau2/tau1 distribution in data and simulation for dijet events without a mass cut. MG denotes the Madgraph5 generator.
	N-subjettiness tau2/tau1 distribution in data and simulation for dijet events with a mass cut 60-100 GeV. MG denotes the Madgraph5 generator.
	N-subjettiness tau2/tau1 with pruning distribution in data and simulation for dijet events without a mass cut. MG denotes the Madgraph5 generator.
	N-subjettiness tau2/tau1 with pruning distribution in data and simulation for dijet events with a mass cut 60-100 GeV. MG denotes the Madgraph5 generator.
	C2 beta distribution in data and simulation for dijet events after a mass cut 60-100 GeV. MG denotes the Madgraph5 generator.
	Jet charge distribution in data and simulation for dijet events after a mass cut 60-100 GeV. MG denotes the Madgraph5 generator.

W+jet events

	Pruned jet mass distribution in data and simulation for W+ events.
	Mass drop distributions in data and simulation for W+jet events without a mass cut.
	Mass drop distributions in data and simulation for W+jet events with a mass cut 60-100 GeV.
	N-subjettiness tau2/tau1 distribution in data and simulation for W+jet events without a mass cut.
	N-subjettiness tau2/tau1 distribution in data and simulation for W+jet events with a mass cut 60-100 GeV.
	Qjet volatility distributions in data and simulation for W+jet events without a mass cut.
	Qjet volatility distributions in data and simulation for W+jet events with a mass cut 60-100 GeV.
	Multi-layer perceptron discriminant distributions in data and simulation for W+jet events without a mass cut.
	Multi-layer perceptron discriminant distributions in data and simulation for W+jet events with a mass cut 60-100 GeV.

ttbar events

	pT of the W jet candidate in semileptonic ttbar events.
	Pruned jet mass of the W jet candidate in semileptonic ttbar events.
	N-subjettiness tau2/tau1 of the W jet candidate in semileptonic ttbar events.
	Jet charge distributions in ttbar events for W+ and W- jets in simulation and data. Simulated distributions are a sum of all processes.

Performance plots

	Efficiency of the pruned jet mass cut and the tau2/tau1 cut combined with the pruned jet mass cut on WW signal samples as a function of pT.
	Efficiency of the pruned jet mass cut and the tau2/tau1 cut combined with the pruned jet mass cut on WW signal samples as a function of number of reconstructed vertices.
	Efficiency of reconstructing either the simulated W or simulated decaying quarks from the W with a single generator level CA8 jet or two generator level AK5 jets. Matching of generator level jets to the simulated W and simulated quarks are required within deltaR < 0.1.
	Fraction of jets passing the pruned jet mass and tau2/tau1 cuts in data and simulation as a function of pT. The data over simulation ratio is shown for the combination of the pruned jet mass and tau2/tau1 cuts.
	Fraction of jets passing the pruned jet mass and tau2/tau1 cuts in data and simulation as a function of number of reconstructed vertices. The data over simulation ratio is shown for the combination of the pruned jet mass and tau2/tau1 cuts.
	Fraction of jets passing the pruned jet mass and tau2/tau1 cuts as a function pT for quark and gluon jets generated with Herwig++.
	Pruned jet mass distribution in the ttbar sample passing the tau2/tau1 cut for the muon channel.
	Pruned jet mass distribution in the ttbar sample passing the tau2/tau1 cut for the electron channel.
	Pruned jet mass distribution in the ttbar sample failing the tau2/tau1 cut for the muon channel.
	Pruned jet mass distribution in the ttbar sample failing the tau2/tau1 cut for the electron channel.
	Fit to the pruned jet mass distribution in ttbar simulation for events which pass the n-subjettiness cut and where the jet is matched to the generated W.
	Fit to the pruned jet mass distribution in ttbar simulation for events which fail the n-subjettiness cut and where the jet is matched to the generated W.
	Fit to the pruned jet mass distribution in ttbar simulation for events which pass the n-subjettiness cut and where the jet is not matched to the generated W.
	Fit to the pruned jet mass distribution in ttbar simulation for events which fail the n-subjettiness cut and where the jet is not matched to the generated W.