The decade of heavy flavor: recent H-bb/cc results in CMS

• Marko Stamenkovic (Brown University (US))

Room: D-122 Measurements of the Higgs boson’s decay 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. For the first time at the LHC, I will introduce a high performance jet-charge tagger, built from track and secondary-vertex features, that infers if the jet originates from quarks or antiquarks and is expected to improve pairing, suppress combinatorial backgrounds, and thereby boost sensitivity in Hbb/Hcc, HH4b, and HHH6b analyses from Run 3 through the HL-LHC.