Search for top squark production in fully-hadronic final states in proton-proton collisions at 13 TeV (SUS-19-010)

Further information

This analysis is documented in arXiv:2103.01290 and the public webpage for this analysis is here.

Technical plots for CMS speakers

DeepResolved Top Tagger

Show Details Hide Details

Figure	Caption
	The trijet candidate mass variable which is used as input to the DeepResolved neural network. The distribution is shown for 4 different categories: the candidates passing/failing the NN discriminator selection (choosen to give a ~2% fakerate in QCD MC) are shown in solid blue/orange respectively, while those matched to a generator-level top quark vs unmatched are shown in dotted green/brown respectively. The results are produced from a ttbar MC sample generated with one leptonic and one hadronic decaying top with the following requirements: at least 4 jets with pT > 30 GeV and the HT > 300 GeV.
	The invariant mass between the second and third jet sorted highest to lowest momentum in the trijet rest frame. The distribution is shown for 4 different categories: the candidates passing/failing the NN discriminator selection (choosen to give a ~2% fakerate in QCD MC) are shown in solid blue/orange respectively, while those matched to a generator-level top quark vs unmatched are shown in dotted green/brown respectively. The results are produced from a ttbar MC sample generated with one leptonic and one hadronic decaying top with the following requirements: at least 4 jets with pT > 30 GeV and the HT > 300 GeV.
	The angular separation between the first and second jet sorted highest to lowest momentum in the trijet rest frame. The distribution is shown for 4 different categories: the candidates passing/failing the NN discriminator selection (choosen to give a ~2% fakerate in QCD MC) are shown in solid blue/orange respectively, while those matched to a generator-level top quark vs unmatched are shown in dotted green/brown respectively. The results are produced from a ttbar MC sample generated with one leptonic and one hadronic decaying top with the following requirements: at least 4 jets with pT > 30 GeV and the HT > 300 GeV.
	The deepCSV b discriminator for the first jet as sorted from highest to lowest momentum in the trijet rest frame. The distribution is shown for 4 different categories: the candidates passing/failing the NN discriminator selection (choosen to give a ~2% fakerate in QCD MC) are shown in solid blue/orange respectively, while those matched to a generator-level top quark vs unmatched are shown in dotted green/brown respectively. The results are produced from a ttbar MC sample generated with one leptonic and one hadronic decaying top with the following requirements: at least 4 jets with pT > 30 GeV and the HT > 300 GeV.
	The semimajor axis for the first jet as sorted from highest to lowest momentum in the trijet rest frame. The jet is approximated as an ellipse in the eta-phi plane. The semimajor axis of this is ellipse is shown in the plot. The distribution is shown for 4 different categories: the candidates passing/failing the NN discriminator selection (choosen to give a ~2% fakerate in QCD MC) are shown in solid blue/orange respectively, while those matched to a generator-level top quark vs unmatched are shown in dotted green/brown respectively. The results are produced from a ttbar MC sample generated with one leptonic and one hadronic decaying top with the following requirements: at least 4 jets with pT > 30 GeV and the HT > 300 GeV.
	The number of charged hadron particle flow candidates in the first jet as sorted from highest to lowest momentum in the trijet rest frame. The distribution is shown for 4 different categories: the candidates passing/failing the NN discriminator selection (choosen to give a ~2% fakerate in QCD MC) are shown in solid blue/orange respectively, while those matched to a generator-level top quark vs unmatched are shown in dotted green/brown respectively. The results are produced from a ttbar MC sample generated with one leptonic and one hadronic decaying top with the following requirements: at least 4 jets with pT > 30 GeV and the HT > 300 GeV.

Background prediction from analysis

Show Details Hide Details

Figure	Caption
	Background composition in 0-lepton search region in bins of the number of b-tagged jets and the number of top-tagged candidates from background predictions.
Figure	Caption
	MTb distribution with the baseline selection and an additional high DM Dphi cut. Signal is normalized to the total background yield. The background uncertainty only includes the statistical contribution from the simulation of the backgrounds. There is a large peak for the ttbar background in the region of MTb below 175 GeV, which is known as the ttbar enhanced region and motivates the definition of the low MTb cut. Additionally, we use a high MTb region of MTb > 175 GeV, where the ttbar background is depleted. The overflow contributions are not shown to avoid the large overflow contributions in the last bins of some of the distributions.
	Comparison of the prediction for the Lost Lepton background, in the high DM search region with Nb=1 or Nb=2 (bins 69 to 152), using the extrapolation and without extrapolation method. The prediction with extrapolation (histogram) does not have selections on the number of top/W tags in the lost lepton CR and the prediction without extrapolation (points) has these selections included in all regions. For both methods, only the statistical uncertainties are shown. The bottom panel shows the ratio of the statistical uncertainty of the prediction with extrapolation over that without extrapolation. The prediction between these two methods are in good agreement in bins with a large number of events. In bins with low event yields a larger variance in the predictions is observed especially for the prediction without the extrapolation. The statistical uncertainty for the extrapolation method is much smaller than the method without extrapolation.

Figure	Caption
	Comparison of the data to the simulation in the low DM electron control region with the selection Nb=0, Nsv=0 as a function of the dielectron mass. The ratio of data over simulation is shown with the statistical uncertainty. The normalization measured for Z to dielectron processes for this selection (low DM, Nb=0, Nsv=0) is 0.870. This is combined with the Z to dimuon measurement using a weighted average to give the value 0.858 that is applied to the portion of the search region that has the corresponding selection.
	Comparison of the data to the simulation in the low DM muon control region with the selection Nb=0, Nsv=0 as a function of the dimuon mass. The ratio of data over simulation is shown with the statistical uncertainty. The normalization measured for Z to dimuon processes for this selection (low DM, Nb=0, Nsv=0) is 0.849. This is combined with the Z to dielectron measurement using a weighted average to give the value 0.858 that is applied to the portion of the search region that has the corresponding selection.
	A comparison of the modified MET distributions for the low DM lepton and photon control regions. For each control region, the simulation is normalized such that the simulation has the same number of events as data. The modified MET includes the four-vector of the reconstructed Z or photon for the respective control region to mimic the Z to neutrinos decay. The upper panel shows the ratio of Z and photon data (black points) and the ratio of Z and photon simulation (blue histogram). The bottom panel shows the ratio of the two distributions in the upper panel, the data ratio divided by the simulation ratio.
	A comparison of the modified MET distributions for the high DM lepton and photon control regions. For each control region, the simulation is normalized such that the simulation has the same number of events as data. The modified MET includes the four-vector of the reconstructed Z or photon for the respective control region to mimic the Z to neutrinos decay. The upper panel shows the ratio of Z and photon data (black points) and the ratio of Z and photon simulation (blue histogram). The bottom panel shows the ratio of the two distributions in the upper panel, the data ratio divided by the simulation ratio. For the statistical uncertainty on the leftmost bin (250 to 350 GeV), the largest contribution comes from the statistical uncertainty of simulated QCD events that are selected in the photon control region.
	A comparison of the HT distributions for the high DM lepton and photon control regions. For each control region, the simulation is normalized such that the simulation has the same number of events as data. The upper panel shows the ratio of Z and photon data (black points) and the ratio of Z and photon simulation (blue histogram). The bottom panel shows the ratio of the two distributions in the upper panel, the data ratio divided by the simulation ratio. For the statistical uncertainty on the leftmost bin (300 to 500 GeV), the largest contribution comes from the statistical uncertainty of simulated QCD events that are selected in the photon control region.

Figure	Caption
	Comparison of the data to the simulation in the high DM QCD control region as a function of missing pT. The hatched region indicates the total systematic uncertainty.
	Comparison of the data to the simulation in the low DM QCD control region as a function of missing pT. The hatched region indicates the total systematic uncertainty.
	Distribution of r_pseudo,jet in the QCD control region before application of the jet response correction.
	Distribution of r_pseudo,jet in the QCD control region after application of the jet response correction.
	Predicted yield of QCD in the search bins. The total uncertainty is broken down into statistical uncertainty from data in the QCD control region (blue), statistical uncertainty from the Monte Carlo (green), and systematic uncertainties (orange).

Event Display

Show Details Hide Details

Figure	Caption
	Event display for a SUSY candidate event (from Run 316060, Event number 621760811) with two b-tagged jets, one W-tagged jet, and some extra jets which passed the search region selection, in view of the CMS detector. The W-tagged jet is marked in orange with its pT, eta, and phi. The two b-tagged jets are marked in magenta with each of their pT, eta, and phi. The extra jets are jets that are colored yellow. The missing transverse momentum is marked in purple.
	Event display for a SUSY candidate event (from Run 304062, Event number 1887015358) with two b-tagged jets, one initial state radiation (ISR) jet, and some extra jets which passed the search region selection, in view of the CMS detector. The ISR jet is marked in vermilion with its pT, eta, and phi. The two b-tagged jets are marked in magenta with each of their pT, eta, and phi. The extra jets are jets that are colored yellow. The missing transverse momentum is marked in purple.
	Event display for a SUSY candidate event (from Run 316469, Event number 371464374) with one merged top, one resolved top, and extra jets, in view of the CMS detector. The merged top jet is marked in light blue with its pT, eta, and phi. The resolved top jets are marked in dark blue with each of their pT, eta, and phi. The extra jets are jets that are colored yellow. The missing transverse momentum is marked in purple.
	Event display for a SUSY candidate event (from Run 323841, Event number 471815433) with two merged tops and extra jets, in view of the CMS detector. The two merged tops jets are marked in light blue with its pT, eta, and phi. The extra jets are jets that are colored yellow. The missing transverse momentum is marked in purple.
	Event display for a SUSY candidate event (from Run 322106, Event number 448246125) with two resolved tops and two extra jets, in view of the CMS detector. One resolved top jet is marked in dark blue and the other is marked in light blue with each of their pT, eta, and phi. The extra jets are jets that are colored yellow. The missing transverse momentum is marked in purple.