Potential of di-Higgs observation via a calibratable jet-free ๐ฏ โ ๐ฏ โ ๐ โ ๐ HH\to 4b framework
Tianyi Yang, Congqiao Li ยท 2025
Read Potential of di-Higgs observation via a calibratable jet-free ๐ฏ โ ๐ฏ โ ๐ โ ๐ HH\to 4b framework by Tianyi Yang, Congqiao Li (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2508-15048.txt.
Introduction
Tianyi Yang and Congqiao Li propose a new way to hunt for pairs of Higgs bosons at the Large Hadron Collider, in the channel where each Higgs decays to bottom quarks. Seeing Higgs pairs would test how the Higgs interacts with itself, a key open question. The usual search sorts spray-like jets, which adds noise and calibration trouble.
Their method skips jet building and works from all particles at once. A classifier spots the signal while ignoring the exact Higgs masses, and a second tool estimates both masses together. A related process with Z bosons can calibrate the response. Tests on realistic simulations suggest a large gain in sensitivity.
Catalog record
| CREATOR | Tianyi Yang, Congqiao Li |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 69 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2508-15048
Browse the full library ยท About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Potential of di-Higgs observation via a calibratable jet-free ๐ฏ โ ๐ฏ โ ๐ โ ๐ HH\to 4b framework Preprint: APS/123-QED Tianyi Yang, Congqiao Li Abstract We present a calibratable, jet-free framework that enhances the search significance of the flagship LHC channel H โ H โ 4 โ b HH\to 4b by more than a factor of five compared to existing approaches. The method employs a mass-decorrelated discriminant to identify h 1 โ h 2 โ 4 โ b h_{1}h_{2}\to 4b with variable h 1 , 2 h_{1,2} masses and a simultaneous estimator of ( m h 1 , m h 2 ) (m_{h_{1}},m_{h_{2}}) , both derived from multiclass classification on all-particle inputs. The H โ H HH signal response can be calibrated using Z โ Z โ 4 โ b ZZ\to 4b . Using a highly realistic simulation framework validated through multiple tests, we demonstrate the methodโs robustness and identify two prerequisites essential for achieving this level of sensitivity. Results indicate that with LHC Run 2 and 3 data, observation-level sensitivity to H โ H HH appears within reach, enabling constraints on ฮบ ฮป \kappa_{\lambda} comparable to HL-LHC projections and offering an accelerated path to precision measurements of the Higgs trilinear coupling. Introduction.โ The observation of the Higgs boson by ATLAS and CMS in 2012 Aad et al. 2012 ; Chatrchyan et al. 2012 completed the particle content of the Standard Model (SM), yet left the shape of the BroutโEnglertโHiggs potential largely untested. Expanding the scalar potential about its minimum through V(\phi)\to V(v+h) yields V=V_{0}+\frac{1}{2}m_{h}^{2}h^{2}+\frac{m_{h}^{2}}{2v^{2}}vh^{3}+\frac{1}{4}\frac{m_{h}^{2}}{2v^{2}}h^{4}, (1) where the cubic term encodes the trilinear self-coupling \lambda_{HHH}=m_{h}^{2}/2v^{2} . This relation makes Higgs-boson-pair production, pp\!\to\!HH , a unique probe of the Higgs potential and the mechanism of electroweak symmetry breaking Degrassi et al. 2012 ; McCullough 2014 . However, the most precise calculations predict SM cross sections of only \sigma_{\text{SM}}(pp\to HH)=31.1^{+2.1}_{-7.2}\,\text{fb} for gluon-gluon fusion ( gg\text{F} ) and 1.73\pm 0.04\,\text{fb} for vector-boson fusion (VBF) at \sqrt{s}=13\,\text{TeV} Dawson et al. 1998 ; Borowka et al. 2016 ; Baglio et al. 2019 ; de Florian and Mazzitelli 2013 ; Shao et al. 2013 ; de Florian and Mazzitelli 2015 ; Grazzini et al. 2018 ; Baglio et al. 2021 ; Dreyer and Karlberg 2018 , making inclusive HH production exceptionally rare yet a key physics target for the HL-LHC era. The destructive interference between the triangle diagram (proportional to \lambda_{HHH} ) and box diagram contributions in gg\text{F} further suppresses the total rate while creating strong sensitivity to deviations from the SM prediction, parameterized by the coupling modifier \kappa_{\lambda}=\lambda_{HHH}/\lambda_{HHH}^{\text{SM}} . With the largest branching ratio, HH\to b\overline{b}b\overline{b} (denoted 4b ) is the LHCโs flagship channel for di-Higgs studies. Recent ATLAS and CMS searches in the HH\to 4b final state using \approx 140 \mathrm{fb^{-1}} highlight both challenges and opportunities. The โresolved-channelโ analysis sets a 95% CL upper limit of 5.4 and 3.9\,\sigma_{HH}^{\rm SM} , constraining \kappa_{\lambda}\in[-3.9,\,11.3] and [-2.3,\,9.4] , respectively Aad et al. 2023 ; Tumasyan et al. 2022 . CMS also reports a โboosted-channelโ analysis in the gg\text{F} channel, excluding \kappa_{2V}=0 at 6.3\,\sigma while achieving a compatible \kappa_{\lambda} constraint: [-9.9,\,16.9] (expected [-5.1,\,12.2] ), despite targeting largely different phase space Tumasyan et al. 2023 . The power of the boosted regime is largely attributed to advanced deep learning-based H\to b\overline{b} jet tagging using jet constituent particles as input CMS Collaboration 2020 ; Qu and Gouskos 2020 , which dramatically suppresses multijet backgrounds Tumasyan et al. 2023 . This raises an intriguing question: can the superior background-rejection capability of the boosted channel be extended across the broader resolved-channel phase space? In this Letter, we extend advanced deep-learning techniques from boosted HH searches to the full kinematic regime through a jet-free HH\to 4b framework. Using a highly realistic simulation setup, we input all pileup-mitigated reconstructed particles into a transformer network to distinguish generic h_{1}h_{2}\to 4b signals from multijet backgrounds in the resolved-channel phase space. The flat mass prior of the signal sample permits calibration of the HH response with ZZ\to 4b events. Robustness is established through multiple validation tests, which also illuminate the origin of performance gains over conventional approaches. For a projected 450 \mathrm{fb^{-1}} dataset (Run 2+3), the method attains an expected 3.3\,\sigma sensitivity under realistic systematic uncertaintiesโexceeding existing resolved analyses by a factor of > 5 and approaching observation-level sensitivity. For comparison, the latest ATLASโCMS HL-LHC combination at 3\,\mathrm{ab^{-1}} quotes 2.8\,\sigma in HH\to 4b ATLAS and CMS Collaborations 2025 , with earlier projections of 1.0\,\sigma in 2019 Dainese et al. 2019 ; ATLAS and CMS Collaborations 2019 . This work thus represents a major methodological advance, offering a fast-track path to precision measurements of the trilinear Higgs coupling years ahead of schedule. Simulation setup.โ Our analysis is based on a dedicated simulation framework that emulates CMS conditions. All generated events are processed through Pythia 8.3 Sjรถstrand et al. 2015 for parton showering and hadronization and Delphes 3.5 de Favereau et al. 2014 for fast detector simulation, using the JetClass-II configuration Li et al. 2024a . This setup yields the full set of pileup-mitigated reconstructed particle candidates, which is essential for our fully particle-based strategy. As in CMS, these particles are clustered into anti- k_{\text{T}} jets Cacciari et al. 2008 with distance parameters R=0.4 and 0.8 (denoted as jets and fatjets). Their fidelity to the corresponding CMS particle-flow objects Sirunyan et al. 2017 ; Sirunyan et al. 2020a has been validated: state-of-the-art transformer-based jet taggers ( SophonAK4 Zhao et al. 2025 for R=0.4 and Sophon Li et al. 2024a for R=0.8 ), trained on jets built from these particles, achieve performance consistent with leading CMS taggers Zhao et al. 2025 . Trigger emulation follows CMS menusโ requirements. The resolved channel Tumasyan et al. 2022 requires \geq 4 jets with p_{\text{T}} thresholds [75, 60, 45, 40] GeV, scalar H_{\text{T}}>330 GeV , and \geq 3 jets passing a loose SophonAK4 b tag. The boosted channel Tumasyan et al. 2023 requires fatjet-based scalar H_{\text{T}}>800 GeV and a leading fatjet with trimmed mass m_{\text{trim}}>50\,\text{GeV} . Unless explicitly stated otherwise, event yields are scaled to 450 \mathrm{fb^{-1}} , representative of the combined Run 2 and Run 3 LHC data, to illustrate the discovery potential with currently available statistics. The signal comprises HH production via gg\text{F} and VBF. Backgrounds include inclusive QCD multijet production, top-quark processes ( t\overline{t} and single top), electroweak diboson production ( WW , WZ , ZZ ), single-Higgs production across all channels, and Z +jets process. The gg\text{F} HH and single-Higgs samples are generated with Powheg 2.0 at next-to-leading-order accuracy in QCD Alioli et al. 2008 ; Nason 2004 ; Frixione et al. 2007 . The QCD multijet samples are produced via Pythia . The remaining processes are produced with MG5_aMC@NLO 2.9.18 Alwall et al. 2014 at leading-order (LO) accuracy, using NNPDF 3.1 next-to-next-to-LO (NNLO) PDF set Ball et al. 2017 , with additional partons and interfaced to the parton shower. For the computationally dominant backgrounds, in particular QCD multijet, dedicated generator-level phase-space preselection is applied to enhance the population of analysis-relevant regions and keep event generation tractable. Signal grids with varying \kappa_{\lambda} are produced to enable profiling and limit setting across non-SM hypotheses. Cross sections are taken at the highest available perturbative accuracy Czakon and Mitov 2014 ; Kidonakis