• b-tagging
• boosted objects
• CMS
• data-driven background estimate
• di-Higgs
• jet substructure

The discovery of the Higgs boson at the LHC in 2012 represents a major step in
our understanding of fundamental interactions. Moreover, the newly-discovered
particle can be seen as a promising "tool" to look for new phenomena. Specifically,
this thesis focuses on high mass resonances decaying to Higgs pairs.
The search presented in this thesis aims to be as much as possible model inde-
pendent. Nevertheless, the results can be interpreted within several New Physics
models. Among those, two possible scenarios are the Randall-Sundrum Radion
and massive KK-Graviton productions in Warped Extra Dimensions. Other well
motivated scenarios featuring a heavy CP-even scalar decaying to Higgs pairs are
the Next-to-Minimal Supersymmetric Standard Model and the Twin Higgs model.
The analysis target is a narrow di-Higgs resonance with both Higgs bosons
decaying to b quarks, in the highest possible mass range detectable at the LHC
(above ∼ 1 TeV). Despite the large QCD multi-jet background, the b ̄ bb ̄ b final
state has showed the best sensitivity overall for resonance masses 400 GeV in
the LHC Run1 searches, but no evidence of signal.
A high mass resonance is expected to produce two energetic Higgs bosons, so
that the b quarks produced from their decay are collimated along the direction
of motion. As a consequence, the hadronization products of a pair of narrowly
separated b quarks can be reconstructed as a single large jet. Boosted Higgs
bosons decaying to b-quark pairs can be distinguished from the QCD multi-jet
background exploiting the jet substructure and through a dedicated b-tagging
algorithm.
The transition region between the aforementioned boosted topology and the resolved
topology, where four unmerged b-jets are reconstructed, has been studied
for the first time in this thesis and it is addressed matching one large Higgs jet
to a pair of unmerged b-jets.
The analysis has been carried out on 2.7/fb of proton-proton collision data
recorded by the CMS experiment at the LHC at a center of mass energy of 13
TeV during 2015. The benchmark signal process used to design the analysis
strategy is the spin-2 Bulk Graviton with negligible natural width. A data-driven
estimation of the background, which minimizes systematic uncertainties that
might arise from poorly-understood QCD multi-jet backgrounds in the simulation,
has been employed.
As no signal evidence has been detected, upper limits on the production cross
section of the benchmark resonance in the mass range 600-3000 GeV are presented.
The event reconstruction addressing the transition region has been found
to improve the sensitivity of ∼15-20%.