Our universe may conceal far more than what we can see. Physicists have long theorized about the existence of a dark sector — an undiscovered set of particles and forces beyond the well-tested laws of known physics, described by the Standard Model (SM), and possibly connected with the existence of dark matter. But how do you search for something so elusive?
The CMS collaboration at the Large Hadron Collider (LHC) has recently conducted a search for lepton-enriched semi-visible jets, a unique signature predicted by extended dark sector model, known as Hidden Valleys. These jets could form when hypothetical dark quarks, bound by a dark strong force, create showers containing a mixture of invisible dark sector particles and visible leptons (electrons and muons). To help distinguish these jets from ordinary SM ones, a machine learning approach using a graph neural network was developed to trace the detailed formation history of jets. This physics-informed method aims to capture subtle differences in jet substructure, providing a new way to look for signs of a strongly-coupled dark sector. In addition, a novel neural network–based background estimation technique was employed, which aims to learn the background characteristics directly from data. This approach helps improve the accuracy of the search by reducing potential biases from standard background modeling.
Using the full Run 2 dataset, the CMS collaboration set the first constraints on these types of dark sector signatures, excluding portal mediators with masses up to 4.7 TeV, depending on the model parameters. This work explores a new direction in dark sector searches and illustrates how advanced machine learning techniques can aid in the ongoing quest to understand the universe’s hidden components.