Speakers
Description
Measurements of the Higgs boson’s decays to heavy quarks—especially the dominant H→bb̄ mode and the much rarer H→cc̄—pose formidable challenges at the LHC. In H→bb̄ searches, overwhelming hadronic backgrounds dilute the signal, while the small H→cc̄ branching fraction and the difficulty of distinguishing c‑jets from both b‑ and light‑flavour jets further complicate the measurement. Limited energy resolution for heavy‑flavour jets exacerbates background contamination.
To address these issues, CMS has introduced a new generation of heavy‑flavour tagging and jet‑energy regression techniques based on graph neural networks, for both small‑ and large‑radius jets. These substantial improvements in jet identification have enabled a complete redesign of the Level‑1 and High‑Level trigger strategies for Run 3, resulting in significantly higher signal‑event yields.
In this talk, we will review the latest developments in graph‑based heavy‑flavour tagging and energy regression, and describe how these algorithms have been integrated into the Run 3 trigger menu. We will illustrate their impact on physics analyses by presenting Run 2 results: searches for tt̄H production in the H→bb̄ and H→cc̄ channels, as well as novel probes of the Higgs self‑coupling via triple‑Higgs production HHH→bb̄bb̄bb̄. Finally, to demonstrate the performance gains in Run 3, we will present results from a novel search for Higgs‑boson pair production in the HH→bb̄bb̄ final state.
These combined advances in detector performance, trigger strategy, and multivariate analysis mark a significant step forward in CMS’s ability to isolate heavy‑flavour Higgs signatures. They pave the way for precision measurements of Higgs–quark couplings and open new avenues for rare‑process discoveries as the LHC moves toward its high‑luminosity era.