The cutflow table for $N1$ and $N2$ signal samples for different mass points. The samples consist of both $ee$ and $\mu\mu$ final states in equal ratio. Each sample is normalized to a total cross-section of 0.1 pb. The abbreviations used in the table are, SC = Same Charge, SF = Same Flavor.

The cutflow table for $N3$ signal samples for different mass points. The samples consist of both $ee$ and $\mu\mu$ final states in equal ratio. Each sample is normalized to a total cross-section of 10 pb. The abbreviations used in the table are, SS = Same Charge, SF = Same Flavor.

Expected and observed upper limits on the strength of HNL mixing with electron neutrinos

Expected and observed upper limits on the strength of HNL mixing with muon neutrinos

Expected and observed upper limits on the strength of HNL mixing with tau neutrinos

Post-fit distribution of the GNN score evaluated with $m_{H/A}$ = 400 GeV in the validation region in the 2LOS region with $\geq 8$ jets. These regions do not enter the fit. The post-fit background prediction is obtained using the post-fit nuisance parameters from the background-only fit in the control and signal regions.

Observed and expected 95% CL upper limits on the cross-section times branching ratio,$ \sigma(pp \rightarrow t\bar{t}H/A ) \times B(H/A \rightarrow t\bar{t})$, as a function of $m_{H/A}$, obtained using the 1L/2LOS final states.

Observed and expected 95% CL lower limits on tanβ as a function of the $m_{H/A}$ mass obtained using the 1L/2LOS final states, assuming a Type-II 2HDM in the alignment limit. Values of tanβ below the observed limit are excluded. The scenario where both the scalar $H$ and pseudo-scalar $A$ contribute with $m_H = m_A$ is shown.

Observed and expected 95% CL lower limits on tanβ as a function of the $m_{H/A}$ mass obtained using the 1L/2LOS final states, assuming a Type-II 2HDM in the alignment limit. Values of tanβ below the observed limit are excluded. The scenario with the contribution from scalar $H$ only is shown.

Observed and expected 95% CL lower limits on tanβ as a function of the $m_{H/A}$ mass obtained using the 1L/2LOS final states, assuming a Type-II 2HDM in the alignment limit. Values of tanβ below the observed limit are excluded. The scenario with the contribution from pseudo-scalar $A$ only is shown.

Expected and observed 95% CL upper limits on the cross-section times branching ratio,$ \sigma(pp \rightarrow t\bar{t}H/A ) \times B(H/A \rightarrow t\bar{t})$, as a function of $m_{H/A}$, obtained from the combination of the 1L/2LOS and 2LSS/ML final states.

Expected and observed 95% CL lower limits on tanβ as a function of the $m_{H/A}$ mass obtained using the 1L/2LOS and 2LSS/ML final states, assuming a Type-II 2HDM in the alignment limit. Values of tanβ below the observed limit are excluded. The scenario where both the scalar $H$ and pseudo-scalar $A$ contribute with $m_H = m_A$ is shown.

Expected and observed 95% CL lower limits on tanβ as a function of the $m_{H/A}$ mass obtained using the 1L/2LOS and 2LSS/ML final states, assuming a Type-II 2HDM in the alignment limit. Values of tanβ below the observed limit are excluded. The scenario with the contribution from scalar $H$ only is shown.

Expected and observed 95% CL lower limits on tanβ as a function of the $m_{H/A}$ mass obtained using the 1L/2LOS and 2LSS/ML final states, assuming a Type-II 2HDM in the alignment limit. Values of tanβ below the observed limit are excluded. The scenario with the contribution from pseudo-scalar $A$ only is shown.

Expected and Observed 95% CL upper limits on the $ \sigma(pp \rightarrow t\bar{t})\times B(S_8\rightarrow t\bar{t})$ production cross-section as a function of $m_{S_8}$, obtained from the combination of the 1L/2LOS and 2LSS/ML final states. The expected limits from the individual 1L/2LOS and 2LSS/ML analyses are also shown.

Post-fit distribution of the GNN scores evaluated with $m_{H/A}$ = 700 GeV in the 1L region with $\geq 10$ jets and four $b$-tagged jets. The fit is performed under the background-only hypothesis.

Post-fit distribution of the GNN scores evaluated with $m_{H/A}$ = 700 GeV in the 2LOS region with $\geq 8$ jets and $\geq 4$ $b$-tagged jets. The fit is performed under the background-only hypothesis.

Post-fit distribution of the GNN scores evaluated with $m_{H/A}=$1000 GeV,in the 1L region with $\geq 10$ jets four $b$-tagged jets. The fit is performed under the background-only hypothesis.

Post-fit distribution of the GNN scores evaluated with $m_{H/A}=$1000 GeV in the 2LOS region with $\geq 8$ jets and $\geq 4$ $b$-tagged jets. The fit is performed under the background-only hypothesis.

Post-fit distribution of the most important global features used in the GNN training, Sum of the pcb scores of the six jets with the highest scores ($\sum_{i\in[1,6]}\mathrm{pcb}_i$), in the region where the training is performed, i.e., the region with $\geq 9$ jets and $\geq3$ $b$-tagged jets in the 1L channel. The fit is performed under the background-only hypothesis using the GNN scores evaluated with $m_{H/A} = 400$ GeV.

Post-fit distribution of the most important global features used in the GNN training, Sum of the pcb scores of the six jets with the highest scores ($\sum_{i\in[1,6]}\mathrm{pcb}_i$), in the region where the training is performed i.e.the region with $\geq 7$ jets and $\geq3$ $b$-tagged jets in the 2LOS channel. The fit is performed under the background-only hypothesis using the GNN scores evaluated with $m_{H/A} = 400$ GeV.

Post-fit distribution of the most important global features used in the GNN training, $p_{\mathrm{T}}$ sum of all reconstructed leptons and jets ($\mathrm{H}_{\mathrm{T}}^{\mathrm{all}}$), in the region where the training is performed, i.e., the region with $\geq 9$ jets and $\geq3$ $b$-tagged jets in the 1L channel. The fit is performed under the background-only hypothesis using the GNN scores evaluated with $m_{H/A} = 400$ GeV.

Post-fit distribution of the most important global features used in the GNN training, $p_{\mathrm{T}}$ sum of all reconstructed leptons and jets ($\mathrm{H}_{\mathrm{T}}^{\mathrm{all}}$), in the region where the training is performed, i.e. the region with $\geq 7$ jets and $\geq3$ $b$-tagged jets in the 2LOS channel. The fit is performed under the background-only hypothesis using the GNN scores evaluated with $m_{H/A} = 400$ GeV.

Post-fit distribution of the most important global features used in the GNN training, Number of jets, in the region where the training is performed, i.e., the region with $\geq 9$ jets and $\geq3$ $b$-tagged jets in the 1L channel. The fit is performed under the background-only hypothesis using the GNN scores evaluated with $m_{H/A} = 400$ GeV.

Post-fit distribution of the most important global features used in the GNN training, Number of jets, in the region where the training is performed, i.e., the region with $\geq 7$ jets and $\geq3$ $b$-tagged jets in the 2LOS channel. The fit is performed under the background-only hypothesis using the GNN scores evaluated with $m_{H/A} = 400$ GeV.

Pre-fit MC statistical uncertainties per bin. These values are used to quantify the MC statistical uncertainty contributions to the total uncertainty in $m_t$ based on elements from the covariance matrix obtained in the profile likelihood fit. Each MC statistical contribution is estimated by dividing the relevant covariance matrix entry (see Table 2) by the corresponding pre-fit MC statistical uncertainty for that bin.

The contribution of each source of systematic uncertainty to the total uncertainty in $m_t$ is provided. Each source's contribution is taken directly from the relevant element of the covariance matrix for the profile likelihood fit to $\overline{m_J}$, $m_{jj}$, and $m_{tj}$ using data. The sign of the effect of each uncertainty is included.

The fitted $m_t$ is provided for each replica generated using bootstrapping. Each replica is a 3D histogram of $m_{J}$, $m_{jj}$, and $m_{tj}$, where $m_{J}$ has 50 bins between 145.0 GeV and 205.0 GeV. This method uses the BootstrapGenerator class in ROOT to initialise a pseudo-random number generator for each event, which is seeded by the value of the eventNumber and runNumber variables for a given event in the input dataset. The pseudo-random numbers are drawn from a Poisson distribution with rate parameter equal to 1. These are used to fluctuate the amount by which each replica histogram is filled. There are 1000 replicas, numbered from 0 to 999.