Search for double Higgs boson production in the four bottom quark final state using boosted large-radius jets with the CMS detector

The double Higgs production (HH) allows us to measure the Higgs self-interaction and is uniquely sensitive to the structure of the Higgs potential. This talk will cover the boosted production of HH→4b in the gluon-fusion (ggF) production mode with 138 fb^-1 of data collected at the CMS Experiment at √s = 13 TeV. This analysis uses dedicated jet identification algorithms developed to identify boosted H→bb jets, which are built using Dynamic Graph Convolutional Neural Networks (DGCNN), to analyze each particle inside the jet with the goal of identifying its specific jet substructure features [1][2]. A Boosted Decision Tree (BDT) is designed to distinguish the HH signal from QCD and top background processes. The QCD background is estimated from control regions using a data-driven fit and the top background is estimated from Monte Carlo simulations with corrections derived from the data. Corrections to the shape of the HH signal Monte Carlo simulations (due to known p_T dependent disagreement in the DGCNN efficiency) are derived using a data sample enriched in boosted g→bb (gluon splitting) proxy jets. I will show the final sensitivity of this analysis to the Higgs self-coupling (κ_λ) and discuss future directions and scope of improvements for measuring the HH process.

[1] Huilin Qu and Loukas Gouskos, Phys. Rev. D 101, 056019

[2] CMS Collaboration, CMS-DP-2020-002