Inclusive top pair production at the parton level

We present here the plots and animations related to Sections 4.2 and 5.1 of [Gomez Ambrosio et al., 2022].

We consider inclusive top quark pair production with stable tops, \(p p \rightarrow t \bar{t}\), at the LHC 14 TeV, benchmarking the neural network parametrisation of the likelihood ratio against the analytical calculation.

We consider the effects of two representative dimension-six SMEFT operators modifying inclusive top quark pair production as defined in [Brivio, 2021] in the topU3l flavour scheme:

\[\begin{split}\nonumber \mathcal{O}_{tu}^{(8)} &= (\bar{t}\gamma_\mu T^A t)(\bar{u}\gamma^\mu T^A u) \, ,\\ \mathcal{O}_{tG} &= (\bar{Q}\sigma^{\mu\nu}T^a t) \tilde{H}G_{\mu\nu}^a \, . \label{eq:tt_part_operators}\end{split}\]

Models and training reports

Trained models of top-quark pair production with two features, the invariant mass \(m_{t\bar{t}}\) and the rapidity \(y_{t\bar{t}}\), are available here. See Table 1 for the training reports corresponding to each Wilson coefficient, at both \(\mathcal{O}(\Lambda^{-2})\) and \(\mathcal{O}(\Lambda^{-4})\) in the EFT expansion.

Table 1 Training reports per EFT ratio function for top-quark pair production at the parton level with \(c_{tG}\) and \(c_{tu}^{(8)}\). Each ratio function is trained on two features, the invariant mass \(m_{t\bar{t}}\) and the rapidity \(y_{t\bar{t}}\) of the top-quark pair.

\(p p \rightarrow t \bar{t}\)	Training report \((m_{t\bar{t}}, y_{t\bar{t}})\)
ctu8	report
ctu8_ctu8	report
ctGRe	report
ctGRe_ctGRe	report

Trained models of top-quark pair production with a single feature, the invariant mass \(m_{t\bar{t}}\), are available here. See Table 2 for the corresponding training reports.

Table 2 Training reports per EFT ratio function for top-quark pair production at the parton level with \(c_{tG}\) and \(c_{tu}^{(8)}\), trained on a single feature, the invariant mass \(m_{t\bar{t}}\) of the top-quark pair.

\(p p \rightarrow t \bar{t}\)	Training report \((m_{t\bar{t}})\)
ctu8	report
ctu8_ctu8	report
ctGRe	report
ctGRe_ctGRe	report

Results

Plots and animations are available at the following links:

• Pseudo-data generation and benchmarking
• Neural network training
  • Overview
  • Animations of the progression of the exact/ML model comparison with training
  • Animation of the training of the decision boundary function
• Constraints on the SMEFT