Introduction

Even though there is a strong evidence that the observed particle is the Higgs boson predicted by the SM, there is still space for BSM candidates. This uncertainty drives to a single characteristic of the SM Higgs boson that by itself can precisely determine its true nature, the Higgs self-couplings. Information on the couplings can be directly obtained from final states featuring at least two Higgs bosons. The possible existence of an extended scalar sector or the presence of new dynamics at higher scales should modify the Higgs self-couplings, getting different values that from the SM predictions.

However, the cross sections corresponding to di-Higgs processes are much smaller than those from single Higgs production due to the suppression induced by a heavier final state and an additional weak coupling. Even with the highest LHC centre-of-mass energy and the event statistic reached at the end of the LHC era, any meaningful extraction of the self-couplings are very challenging. However, the prospect to observe a di-Higgs production with a further determination of the triple Higgs coupling seems feasible under optimistic assumptions at the High Luminosity LHC (HL-LHC).

The most promising final state, due to the low background and acceptable branching fraction, includes a pair of b quarks and photons ($h\to\gamma\gamma b\bar{b}$) Current ATLAS cut-and-count $h\to\gamma\gamma b\bar{b}$ prospects analysis reached an expected significance for the di-Higgs production of 1.05/sigma at the HL-LHC. However, recent studies shows that the detector performance could improve, increasing the significance to 1.4\sigma. We propose to improve the sensitivity by introducing more variables that contains extra kinematic information to the fit, known as multivariate analysis (MVA).

People involved

• Marco Aurelio Diaz (Professor)
• Juan Pedro Ochoa (Professor)
• Sebastian Olivares (Post-doc)
• Danilo Quinteros (Master student)
• Matias Contreras (Bachelor student)
• Michael Haacke (Bachelor student)

Samples and repository

The package is located at:

 https://gitlab.com/saolivap/diHiggsbbyyAnalysis 

Samples are located in nyx:

 /data2/Higgs/diHiggs/samples/ntuples 

Analysis

The project is divided in two set of scripts:

• The main analysis that performs the full physics object reconstruction creating an input sample for the MVA analysis. This one is based entirely on Stephane's main analysis package.
• The tree analysis that takes the physics objects from the previous step to perform an improved MVA analysis

Instructions of how to run both are explained in the gitlab space.

Results

Legacy method

Cut-based Analysis

BDT Analysis

Weighted method

Cut-based Analysis

BDT Analysis

Presentations (slides)

• Higgs prospects meeting (Sebastian): https://indico.cern.ch/event/678625/contributions/2779959/attachments/1552762/2440327/diHiggs_bbyy_MVA_03.pdf
• Encuentro estudiantes HEP (Danilo): https://indico.cern.ch/event/666495/contributions/2824051/attachments/1576585/2489808/Presentacion-encuentro.pdf

Status

• legacy analysis with default b-tagging
• legacy analysis with all b-tagging
• weighted analysis with all b-tagging
• weighted analysis with all bkg samples
• weighted analysis using Nathan's input samples
• legacy analysis using more MVA methods
• legacy analysis with optimal MVA configuration