Searching for HWW Anomalous Couplings with Simulation-Based Inference

(Password: 041022 )

Understanding the universe’s asymmetry between matter and antimatter is one of the major open questions in particle physics. Explaining this imbalance requires a source of Charge-Parity (CP) violation beyond the Standard Model (BSM). Given the Higgs boson’s role in electroweak symmetry breaking, its interactions are a natural place to search for anomalous couplings. In this work, the $HWW$ interaction vertex is studied in the channel, where , within the Standard Model Effective Field Theory (SMEFT) framework.

The sensitivity to two HWW anomalous couplings, and , is explored using machine-learning-based inference techniques. These methods leverage simulator information to train neural networks that estimate likelihood ratios by capturing correlations between observables and avoiding the typical approximations in traditional methods. ALICES, a cross-entropy estimator functioning as an unbinned, high-dimensional surrogate model for simulation-based inference, is benchmarked against SALLY, a detector-level optimal observable, and against traditional histograms of kinematic and angular observables.

The ALICES and SALLY methods yielded tighter constraints than 1D summary observables, but SALLY's results closely matched those from 2D histograms. However, SALLY remains promising, given its potential for simultaneous probing of several couplings. Although ALICES sometimes struggled to capture the minima and likelihood shapes accurately, further refinement could enhance its sensitivity beyond both SALLY and 2D histograms. These results underscore the value of further exploring these methods with Run 3 data to potentially improve current ATLAS and CMS results.