Constraints on the SMEFT

We compare the constraints on \(C_{tu}^{(8)}\) and \(C_{tG}\) at 95% CL obtained from a binned analysis to those obtained from the unbinned ML model.

In the upper plots we make use of two features: the invariant mass \(m_{t \bar{t}}\) and rapidity \(y_{t \bar{t}}\) of the top quark pair. In the lower plots we use a single feature, \(m_{t \bar{t}}\).

Two sets of binnings are used:

• Binning 1:

\[\begin{split}m_{t \bar{t}} \in [1.45, 2.5, \infty) \textrm{ TeV}, \\ y_{t \bar{t}} \in \pm [0,1.5,3.0]\end{split}\]

• Binning 2:

\[\begin{split}m_{t \bar{t}} \in [1.45, 1.5, 1.55, 1.6, 1.7, 1.8, 1.9, 2.0, 2.1,\\ 2.2, 2.3, 2.4, 2.5, 2.6, 2.7, 2.8, 2.9, 3.0, \infty) \textrm{ TeV}, \\ y_{t \bar{t}} \in \pm [0, 0.3, 0.6, 0.9, 1.2, 3.0]\end{split}\]

In the left plots, we obtain constraints by working to linear order \(O(\Lambda^{-2})\) in the EFT calculation. In the right plots we work to quadratic order \(O(\Lambda^{-4})\).

In general, we find an improvement in the constraints obtained using the unbinned ML model compared to the binned analysis, as well as good agreement between the ML model and exact calculation.