The ratio of data to the sum of the SM backgrounds. The uncertainties of simulated data are only the statistical unvertainty in the simulation predictions.

Distribution of $p_{T}^{miss}$ after the preselection from 2017 data (black points) and simulation (colored lines). The simulated distribution of two signal points are represented by colored lines, while not being stacked on the background distributions: $(m(\mathrm{\widetilde{t}}_{1}),m(\widetilde{\chi}^{0}_{1}))$ = (500, 490) and (500, 420) GeV. The last bin includes the overflow events.

The ratio of data to the sum of the SM backgrounds. The uncertainties of simulated data are only the statistical unvertainty in the simulation predictions.

Distribution of $p_{T}^{miss}$ after the preselection from 2018 data (black points) and simulation (colored lines). The simulated distribution of two signal points are represented by colored lines, while not being stacked on the background distributions: $(m(\mathrm{\widetilde{t}}_{1}),m(\widetilde{\chi}^{0}_{1}))$ = (500, 490) and (500, 420) GeV. The last bin includes the overflow events.

The ratio of data to the sum of the SM backgrounds. The uncertainties of simulated data are only the statistical unvertainty in the simulation predictions.

Distribution of $N_{Jet}$ after the preselection from 2017 data (black points) and simulation (colored lines). The simulated distribution of two signal points are represented by colored lines, while not being stacked on the background distributions: $(m(\mathrm{\widetilde{t}}_{1}),m(\widetilde{\chi}^{0}_{1}))$ = (500, 490) and (500, 420) GeV. The last bin includes the overflow events.

The ratio of data to the sum of the SM backgrounds. The uncertainties of simulated data are only the statistical unvertainty in the simulation predictions.

Distribution of $N_{Jet}$ after the preselection from 2018 data (black points) and simulation (colored lines). The simulated distribution of two signal points are represented by colored lines, while not being stacked on the background distributions: $(m(\mathrm{\widetilde{t}}_{1}),m(\widetilde{\chi}^{0}_{1}))$ = (500, 490) and (500, 420) GeV. The last bin includes the overflow events.

The ratio of data to the sum of the SM backgrounds. The uncertainties of simulated data are only the statistical unvertainty in the simulation predictions.

Simulated distribution of $p_{T}(l)$ after the preselection. The total background distribution and the signal distributions are all normalized to unit area. The signal distributions are given for a top squark mass of 300 GeV and $\Delta {m}=10$, 30,50, and 80 GeV.

Simulated distribution of $p_{T}(l)$ after the preselection. The total background distribution and the signal distributions are all normalized to unit area. The signal distributions are shown for four different $(m(\mathrm{\widetilde{t}}_{1}),m(\widetilde{\chi}^{0}_{1}))$ values, all corresponding to the same $\Delta {m} = 30$ GeV.

Simulated distribution of $p_{T}^{miss}$ after the preselection. The total background distribution and the signal distributions are all normalized to unit area. The signal distributions are given for a top squark mass of 300 GeV and $\Delta {m}=10$, 30,50, and 80 GeV.

Simulated distribution of $p_{T}^{miss}$ after the preselection. The total background distribution and the signal distributions are all normalized to unit area. The signal distributions are shown for four different $(m(\mathrm{\widetilde{t}}_{1}),m(\widetilde{\chi}^{0}_{1}))$ values, all corresponding to the same $\Delta {m} = 30$ GeV.

Simulated distribution of $N_{Jet}$ after the preselection. The total background distribution and the signal distributions are all normalized to unit area. The signal distributions are given for a top squark mass of 300 GeV and $\Delta {m}=10$, 30,50, and 80 GeV.

Simulated distribution of $N_{Jet}$ after the preselection. The total background distribution and the signal distributions are all normalized to unit area. The signal distributions are shown for four different $(m(\mathrm{\widetilde{t}}_{1}),m(\widetilde{\chi}^{0}_{1}))$ values, all corresponding to the same $\Delta {m} = 30$ GeV.

BDT output distributions from data (black points) and simulation (colored lines) after the preselection for $\Delta {m} = 10$ GeV for 2017 data. The last bin corresponds to the SR. A representative $(m(\mathrm{\widetilde{t}}_{1}),m(\widetilde{\chi}^{0}_{1}))$ signal point is also presented, while not added to the SM background.

The ratio of data to the sum of the SM backgrounds. Only the statistical uncertainty in the data and the simulated background are represented.

BDT output distributions from data (black points) and simulation (colored lines) after the preselection for $\Delta {m} = 20$ GeV for 2017 data. The last bin corresponds to the SR. A representative $(m(\mathrm{\widetilde{t}}_{1}),m(\widetilde{\chi}^{0}_{1}))$ signal point is also presented, while not added to the SM background.

The ratio of data to the sum of the SM backgrounds. Only the statistical uncertainty in the data and the simulated background are represented.

BDT output distributions from data (black points) and simulation (colored lines) after the preselection for $\Delta {m} = 30$ GeV for 2017 data. The last bin corresponds to the SR. A representative $(m(\mathrm{\widetilde{t}}_{1}),m(\widetilde{\chi}^{0}_{1}))$ signal point is also presented, while not added to the SM background.

The ratio of data to the sum of the SM backgrounds. Only the statistical uncertainty in the data and the simulated background are represented.

BDT output distributions from data (black points) and simulation (colored lines) after the preselection for $\Delta {m} = 40$ GeV for 2017 data. The last bin corresponds to the SR. A representative $(m(\mathrm{\widetilde{t}}_{1}),m(\widetilde{\chi}^{0}_{1}))$ signal point is also presented, while not added to the SM background.

The ratio of data to the sum of the SM backgrounds. Only the statistical uncertainty in the data and the simulated background are represented.

BDT output distributions from data (black points) and simulation (colored lines) after the preselection for $\Delta {m} = 50$ GeV for 2017 data. The last bin corresponds to the SR. A representative $(m(\mathrm{\widetilde{t}}_{1}),m(\widetilde{\chi}^{0}_{1}))$ signal point is also presented, while not added to the SM background.

The ratio of data to the sum of the SM backgrounds. Only the statistical uncertainty in the data and the simulated background are represented.

BDT output distributions from data (black points) and simulation (colored lines) after the preselection for $\Delta {m} = 60$ GeV for 2017 data. The last bin corresponds to the SR. A representative $(m(\mathrm{\widetilde{t}}_{1}),m(\widetilde{\chi}^{0}_{1}))$ signal point is also presented, while not added to the SM background.

The ratio of data to the sum of the SM backgrounds. Only the statistical uncertainty in the data and the simulated background are represented.

BDT output distributions from data (black points) and simulation (colored lines) after the preselection for $\Delta {m} = 70$ GeV for 2017 data. The last bin corresponds to the SR. A representative $(m(\mathrm{\widetilde{t}}_{1}),m(\widetilde{\chi}^{0}_{1}))$ signal point is also presented, while not added to the SM background.

The ratio of data to the sum of the SM backgrounds. Only the statistical uncertainty in the data and the simulated background are represented.

BDT output distributions from data (black points) and simulation (colored lines) after the preselection for $\Delta {m} = 80$ GeV for 2017 data. The last bin corresponds to the SR. A representative $(m(\mathrm{\widetilde{t}}_{1}),m(\widetilde{\chi}^{0}_{1}))$ signal point is also presented, while not added to the SM background.

The ratio of data to the sum of the SM backgrounds. Only the statistical uncertainty in the data and the simulated background are represented.

BDT output distributions from data (black points) and simulation (colored lines) after the preselection for $\Delta {m} = 10$ GeV for 2018 data. The last bin corresponds to the SR. A representative $(m(\mathrm{\widetilde{t}}_{1}),m(\widetilde{\chi}^{0}_{1}))$ signal point is also presented, while not added to the SM background.

The ratio of data to the sum of the SM backgrounds. Only the statistical uncertainty in the data and the simulated background are represented.

BDT output distributions from data (black points) and simulation (colored lines) after the preselection for $\Delta {m} = 20$ GeV for 2018 data. The last bin corresponds to the SR. A representative $(m(\mathrm{\widetilde{t}}_{1}),m(\widetilde{\chi}^{0}_{1}))$ signal point is also presented, while not added to the SM background.

The ratio of data to the sum of the SM backgrounds. Only the statistical uncertainty in the data and the simulated background are represented.

BDT output distributions from data (black points) and simulation (colored lines) after the preselection for $\Delta {m} = 30$ GeV for 2018 data. The last bin corresponds to the SR. A representative $(m(\mathrm{\widetilde{t}}_{1}),m(\widetilde{\chi}^{0}_{1}))$ signal point is also presented, while not added to the SM background.

The ratio of data to the sum of the SM backgrounds. Only the statistical uncertainty in the data and the simulated background are represented.

BDT output distributions from data (black points) and simulation (colored lines) after the preselection for $\Delta {m} = 40$ GeV for 2018 data. The last bin corresponds to the SR. A representative $(m(\mathrm{\widetilde{t}}_{1}),m(\widetilde{\chi}^{0}_{1}))$ signal point is also presented, while not added to the SM background.

The ratio of data to the sum of the SM backgrounds. Only the statistical uncertainty in the data and the simulated background are represented.

BDT output distributions from data (black points) and simulation (colored lines) after the preselection for $\Delta {m} = 50$ GeV for 2018 data. The last bin corresponds to the SR. A representative $(m(\mathrm{\widetilde{t}}_{1}),m(\widetilde{\chi}^{0}_{1}))$ signal point is also presented, while not added to the SM background.

The ratio of data to the sum of the SM backgrounds. Only the statistical uncertainty in the data and the simulated background are represented.

BDT output distributions from data (black points) and simulation (colored lines) after the preselection for $\Delta {m} = 60$ GeV for 2018 data. The last bin corresponds to the SR. A representative $(m(\mathrm{\widetilde{t}}_{1}),m(\widetilde{\chi}^{0}_{1}))$ signal point is also presented, while not added to the SM background.

The ratio of data to the sum of the SM backgrounds. Only the statistical uncertainty in the data and the simulated background are represented.

BDT output distributions from data (black points) and simulation (colored lines) after the preselection for $\Delta {m} = 70$ GeV for 2018 data. The last bin corresponds to the SR. A representative $(m(\mathrm{\widetilde{t}}_{1}),m(\widetilde{\chi}^{0}_{1}))$ signal point is also presented, while not added to the SM background.

The ratio of data to the sum of the SM backgrounds. Only the statistical uncertainty in the data and the simulated background are represented.

BDT output distributions from data (black points) and simulation (colored lines) after the preselection for $\Delta {m} = 80$ GeV for 2018 data. The last bin corresponds to the SR. A representative $(m(\mathrm{\widetilde{t}}_{1}),m(\widetilde{\chi}^{0}_{1}))$ signal point is also presented, while not added to the SM background.

The ratio of data to the sum of the SM backgrounds. Only the statistical uncertainty in the data and the simulated background are represented.

Distribution of $p_{T}(l)$ for $\Delta {m} = 10$ in the VR where $200 < p_{T}^{miss} < 280$ GeV from 2017 data (black points) and simulation (colored lines). The predicted signal distribution is shown for $(m(\mathrm{\widetilde{t}}_{1}),m(\widetilde{\chi}^{0}_{1}))$ = (475, 465).

The ratio of data to the sum of the SM backgrounds. Only the statistical uncertainty in the data and the simulated background are represented.

Distribution of $p_{T}(l)$ for $\Delta {m} = 10$ in the VR where $200 < p_{T}^{miss} < 280$ GeV from 2018 data (black points) and simulation (colored lines). The predicted signal distribution is shown for $(m(\mathrm{\widetilde{t}}_{1}),m(\widetilde{\chi}^{0}_{1}))$ = (475, 465).

The ratio of data to the sum of the SM backgrounds. Only the statistical uncertainty in the data and the simulated background are represented.

Distribution of BDT for $\Delta {m} = 10$ in the VR where $200 < p_{T}^{miss} < 280$ GeV from 2017 data (black points) and simulation (colored lines). The predicted signal distribution is shown for $(m(\mathrm{\widetilde{t}}_{1}),m(\widetilde{\chi}^{0}_{1}))$ = (475, 465).

The ratio of data to the sum of the SM backgrounds. Only the statistical uncertainty in the data and the simulated background are represented.

Distribution of BDT for $\Delta {m} = 10$ in the VR where $200 < p_{T}^{miss} < 280$ GeV from 2018 data (black points) and simulation (colored lines). The predicted signal distribution is shown for $(m(\mathrm{\widetilde{t}}_{1}),m(\widetilde{\chi}^{0}_{1}))$ = (475, 465).

The ratio of data to the sum of the SM backgrounds. Only the statistical uncertainty in the data and the simulated background are represented.

Distribution of $p_{T}^{miss}$ for $\Delta {m} = 60$ in the VR where $p_{T}(l) > 30$ GeV from 2017 data (black points) and simulation (colored lines). The predicted signal distribution is shown for $(m(\mathrm{\widetilde{t}}_{1}),m(\widetilde{\chi}^{0}_{1}))$ = (576, 516).

The ratio of data to the sum of the SM backgrounds. Only the statistical uncertainty in the data and the simulated background are represented.

Distribution of $p_{T}^{miss}$ for $\Delta {m} = 60$ in the VR where $p_{T}(l) > 30$ GeV from 2018 data (black points) and simulation (colored lines). The predicted signal distribution is shown for $(m(\mathrm{\widetilde{t}}_{1}),m(\widetilde{\chi}^{0}_{1}))$ = (576, 516).

The ratio of data to the sum of the SM backgrounds. Only the statistical uncertainty in the data and the simulated background are represented.

Distribution of BDT for $\Delta {m} = 60$ in the VR where $p_{T}(l) > 30$ GeV from 2017 data (black points) and simulation (colored lines). The predicted signal distribution is shown for $(m(\mathrm{\widetilde{t}}_{1}),m(\widetilde{\chi}^{0}_{1}))$ = (576, 516).

The ratio of data to the sum of the SM backgrounds. Only the statistical uncertainty in the data and the simulated background are represented.

Distribution of BDT for $\Delta {m} = 60$ in the VR where $p_{T}(l) > 30$ GeV from 2018 data (black points) and simulation (colored lines). The predicted signal distribution is shown for $(m(\mathrm{\widetilde{t}}_{1}),m(\widetilde{\chi}^{0}_{1}))$ = (576, 516).

The ratio of data to the sum of the SM backgrounds. Only the statistical uncertainty in the data and the simulated background are represented.

Data yield and prediction of the total background and its composition in the eight SRs as defined in for the 2017 analysis. The predictions of W + jets, ttbar, and nonprompt lepton processes are based on data, while those of rare backgrounds are based on simulation. The predictions and their associated uncertainties are pre-fit. The uncertainties are the quadratic sum of the statistical and systematic uncertainties for background processes predicted from data, and the uncertainties on the cross section for the backgrounds predicted from simulation. The yields for two signal points, with $\Delta {m} = 10$ GeV and $\Delta {m} = 80$ GeV, are also shown. The bins corresponding to different $\Delta {m}$ trainings are not statistically independent.

The ratio of the number of observed events over the predicted total background per SR.

Data yield and prediction of the total background and its composition in the eight SRs as defined in for the 2018 analysis. The predictions of W + jets, ttbar, and nonprompt lepton processes are based on data, while those of rare backgrounds are based on simulation. The predictions and their associated uncertainties are pre-fit. The uncertainties are the quadratic sum of the statistical and systematic uncertainties for background processes predicted from data, and the uncertainties on the cross section for the backgrounds predicted from simulation. The yields for two signal points, with $\Delta {m} = 10$ GeV and $\Delta {m} = 80$ GeV, are also shown. The bins corresponding to different $\Delta {m}$ trainings are not statistically independent.

The ratio of the number of observed events over the predicted total background per SR.

Exclusion limit at 95% CL for the four-body decay of the top squark as a function of $m(\mathrm{\widetilde{t}}_{1})$ and $\Delta {m}$, for the combined data of the years 2016, 2017, and 2018. The lines represent the observed limits. These limits are derived using the expected top squark pair production cross section. The green line represents the central values, and the other lines the variations due to the theoretical or experimental uncertainties.

Exclusion limit at 95% CL for the four-body decay of the top squark as a function of $m(\mathrm{\widetilde{t}}_{1})$ and $\Delta {m}$, for the combined data of the years 2016, 2017, and 2018. The lines represent the expected limits. These limits are derived using the expected top squark pair production cross section. The green line represents the central values, and the other lines the variations due to the theoretical or experimental uncertainties.

Exclusion limit at 95% CL for the four-body decay of the top squark as a function of $m(\mathrm{\widetilde{t}}_{1})$ and $\Delta {m}$, for the combined data of the years 2016, 2017, and 2018.

Expected and observed limits computed using asymptotic formulas as a function of the signal mass m_{G*} and k/Mpl for the spin-2 resonance search.

The expected and observed upper limits at 95\% CL on the fiducial cross-section times branching ratio to two photons of a spin-0 resonance as a function of its mass m_X. The decay width of the resonance is equal to 2% of m_X. For masses greater than 1000 GeV, pseudo-experiments are used to verify the expected and observed limits, and used in place of the asymptotic limit when differences are observed.

The expected and observed upper limits at 95\% CL on the fiducial cross-section times branching ratio to two photons of a spin-0 resonance as a function of its mass m_X. The decay width of the resonance is equal to 6% of m_X. For masses greater than 1000 GeV, pseudo-experiments are used to verify the expected and observed limits, and used in place of the asymptotic limit when differences are observed.

The expected and observed upper limits at 95\% CL on the fiducial cross-section times branching ratio to two photons of a spin-0 resonance as a function of its mass m_X. The decay width of the resonance is equal to 10% of m_X. For masses greater than 1000 GeV, pseudo-experiments are used to verify the expected and observed limits, and used in place of the asymptotic limit when differences are observed.

The expected and observed upper limits at 95\% CL on the production cross-section times branching ratio to two photons of the lightest KK graviton as a function of its mass for k/Mpl=0.01. For masses greater than 1000 GeV, pseudo-experiments are used to verify the expected and observed limits, and used in place of the asymptotic limit when differences are observed.

The expected and observed upper limits at 95\% CL on the production cross-section times branching ratio to two photons of the lightest KK graviton as a function of its mass for k/Mpl=0.05. For masses greater than 1000 GeV, pseudo-experiments are used to verify the expected and observed limits, and used in place of the asymptotic limit when differences are observed.

Diphoton invariant mass in the signal region using a binning based on the invariant mass resolution.

Diphoton invariant mass in the signal region binned in 1 GeV bins..

Effect of event selections on a NWA spin-0 and a spin-2 ($k/Mpl=0.01$) MC samples generated for m_X = 1 TeV and on the data. For the MC samples, the event yield is shown after applying event weights; for the data, the absolute yields are shown. The initial yields for data include a trigger preselection that is the OR of a list of single photon and diphoton triggers. The "2 $loose$ photons" step includes the kinematic acceptance cuts. The cross-section times branching ratio for the spin-0 sample is normalized to the MadGraph NLO prediction of 112~fb and the cross-section times branching ratio for the graviton corresponds to the Pythia8 prediction of 12~fb.

Parameterization for the CX and AX factors as function of mX for a scalar particle, as obtained from the narrow-width MC signal samples.

Parameterization for the $C_{X}\cdot A_{X}$ factor as function of $m_{G^*}$ for the RS graviton, as obtained from k/MPl = 0.05 MC signal samples.

Parameterization of the Double Sided Crystal Ball function parameters describing the scalar mass resolution model as a function of m_{X} [GeV].

Parameterization of the Double Sided Crystal Ball function parameters describing the graviton mass resolution model as a function of m_{X} [GeV].

Upper limits on the fiducial cross section times branching ratio to two photons at centre-of-mass energy of 13 TeV of a narrow-width (Γ_X = 4 MeV) spin-0 resonance as a function of its mass m_X.

Upper limits on the production cross section times branching ratio to two photons at centre-of-mass energy of 13 TeV of the lightest KK graviton as a function of its mass m_G* for k/MPl=0.01.

Upper limits on the production cross section times branching ratio to two photons at centre-of-mass energy of 13 TeV of the lightest KK graviton as a function of its mass m_G* for k/MPl=0.10. Only the limits derived with the asymptotic formulae are quoted here.

Upper limits on the production cross section times branching ratio to two photons at centre-of-mass energy of 13 TeV of the lightest KK graviton as a function of its mass m_G* for k/MPl=0.20.

Upper limits on the production cross section times branching ratio to two photons at centre-of-mass energy of 13 TeV of the lightest KK graviton as a function of its mass m_G* for k/MPl=0.30.

Distribution of the diphoton invariant mass for data events passing the spin-0 selection. The data points are plotted at the centre of each bin. The error bars indicate statistical uncertainties. The distribution consists of not only genuine diphoton events, but also events where at least one photon is faked by a jet. There is no data event above 2700 GeV.

Distribution of the diphoton invariant mass for data events passing the spin-2 selection. The data points are plotted at the centre of each bin. The error bars indicate statistical uncertainties. The distribution consists of not only genuine diphoton events, but also events where at least one photon is faked by a jet. There is no data event above 2700 GeV.

Fiducial acceptance and product of the fiducial acceptance times the combined reconstruction and identification efficiency as a function of the mass for a spin-0 resonant signal with narrow width under the spin-0 selection. Only MC statistical uncertainties are included.

Fiducial acceptance and product of the fiducial acceptance times the combined reconstruction and identification efficiency as a function of the mass for a spin-2 resonant signal with k/MPl=0.1 under the spin-2 selection. The definition of fiducial region is provided in JHEP 09 (2016) 001. Only MC statistical uncertainties are included.

Fiducial acceptance and product of the fiducial acceptance times the combined reconstruction and identification efficiency as a function of the mass for an ADD excess (GRW formalism), calculated as the difference between the sum of ADD and SM contributions (including their interference) and the SM contribution, under the spin-2 selection in the signal region M(gamma gamma)>2240 GeV. The definition of fiducial region is provided in JHEP 09 (2016) 001, and M(gamma gamma)>2240 GeV is required in addition. Only MC statistical uncertainties are included.