Dortmund 2021 – scientific programme
Parts | Days | Selection | Search | Updates | Downloads | Help
T: Fachverband Teilchenphysik
T 30: Higgs decay in fermions II
T 30.7: Talk
Tuesday, March 16, 2021, 17:30–17:45, Te
Search for Higgs-boson pair production in the bbℓℓ+ETmiss final state with the ATLAS detector — •Benjamin Rottler, Benoit Roland, and Markus Schumacher — Albert-Ludwigs-Universität Freiburg
The determination of the triple Higgs-boson self-coupling λ is one of the key goals of the physics program at current and future colliders. It will allow to reconstruct the Higgs potential. The self-coupling can be accessed via non-resonant Higgs-boson pair production, which can happen at the LHC via the destructively interfering top-loop and Higgs self-interaction diagrams. The data can also be analyzed to probe resonant Higgs-boson pair production in a search for new heavy particles.
The goal of this analysis is to measure the cross-section of the non-resonant Higgs-boson pair production σHH using the full Run-2 dataset collected by the ATLAS experiment, corresponding to an integrated luminosity of ∼140 fb−1 at √s = 13 TeV. This is done by considering the bbℓℓ+ETmiss final state, which combines the high branching ratio of the H→ bb decay and the good efficiency of lepton triggers. Our focus is on a combined search for the HH→ bb(WW → 2ℓ 2ν), HH→ bb(ττ→ 2ℓ 4ν) and HH→ bb(ZZ → 2ℓ 2ν) processes.
A multi-class deep neural network (NN) is used to separate signal and background processes on top of a loose preselection. In this talk, I will focus on modern technologies used to optimize the NN architecture, like Bayesian hyperparameter optimization and input feature ranking algorithms, as well as on the statistical analysis which makes use of the shape of the NN output distribution to extract σHH.