Shape analysis in Higgs boson pair production

TL;DR

This paper investigates how anomalous couplings affect the invariant mass distribution of Higgs boson pairs at NLO, using machine learning to classify shape types and infer underlying parameters.

Contribution

It introduces a clustering method based on unsupervised machine learning to categorize shape types in Higgs pair production, linking shapes to parameter space insights.

Findings

Shape classification into distinct types using machine learning

Correlation between shape types and coupling parameters

Framework applicable to non-linear EFT models

Abstract

We study the impact of anomalous couplings in the Higgs sector on the shape of the Higgs boson pair invariant mass distribution at NLO. Our analysis is based on a five-dimensional coupling parameter space relevant for Higgs boson pair production in gluon fusion, in the framework of a non-linear Effective Field Theory. In particular, we present a clustering procedure into certain shape types based on unsupervised machine learning, with the aim to infer information about the underlying parameter space from a given shape type.