Abstract

The wealth of precise data gathered from the Large Hadron Collider (LHC) at CERN presents both opportunities and challenges in the determination of fundamental parameters of the Standard Model (SM) of particle physics and in the search for physics beyond the SM. Central to this effort is a detailed understanding of the proton’s subnuclear structure, described in terms of quarks and gluons via the parton distribution functions (PDFs).

In this talk, I will explore how cutting-edge machine learning techniques provide a robust solution to the inverse problem of extracting PDFs from experimental data. I will discuss recent advancements in global PDF fits and the broader potential of machine learning in this context. Additionally, I will present new insights into how the parametrization of the proton structure interacts with signals of new physics at the LHC , using two complementary approaches. First, I will introduce a novel framework that simultaneously determines PDFs and the Wilson coefficients of an effective field theory (EFT), allowing for a model-independent exploration of heavy new physics. Second, I will outline a systematic methodology to investigate whether global PDF fits could inadvertently “wash out” hints of new physics in the high-energy tails of observed distributions.

This talk aims to shed light on the interplay between established proton structure models and emerging theories, demonstrating how modern computational tools can drive discovery in high-energy physics.