Recent results on searches with boosted Higgs bosons at CMS

* · 2025

Read Recent results on searches with boosted Higgs bosons at CMS by * (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-2507-11977.txt.

Introduction

Farouk Mokhtar summarizes recent CMS results on Higgs bosons produced with a big kick of momentum, called boosted Higgs bosons, in 2025 conference proceedings. At the Large Hadron Collider, a fast-moving Higgs decays into a single wide spray of particles, which takes special reconstruction and tagging tricks to read. That regime probes Higgs couplings at high energy and hunts signs of physics beyond the standard model. Mokhtar highlights the newest searches and the tagging methods that sharpen their sensitivity.

Catalog record

CREATOR*
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE10 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2507-11977

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Recent results on searches with boosted Higgs bosons at CMS * Abstract The study of boosted Higgs bosons at the LHC provides a unique window to probe Higgs boson couplings at high energy scales and search for signs of physics beyond the standard model. In these proceedings, we present recent results on boosted Higgs boson searches at the CMS experiment, highlighting innovative reconstruction and tagging techniques that enhance sensitivity in this challenging regime. Farouk Mokhtar 1 Introduction The discovery of the Higgs boson ( \mathrm{H} ) by ATLAS [ 1 ] and CMS [ 2 ] at the LHC completed the standard model (SM) and opened new avenues to study electroweak symmetry breaking. Since then, a core focus has been precision measurements of \mathrm{H} properties and couplings to test the SM and search for beyond the SM (BSM) physics. At high transverse momentum ( p_{\mathrm{T}} ), \mathrm{H} production probes higher energy scales and enhances sensitivity to anomalous couplings. The boosted regime, while challenging due to small SM cross sections, results in decay products that merge into large-radius jets, enabling jet substructure and machine learning (ML)-based taggers to separate signal from background. In these proceedings, we summarize innovative methods for boosted \mathrm{H} topologies and recent results from the CMS experiment [ 3 ] at the LHC. 2 ML-based jet tagging for boosted \mathrm{H} topologies Notable advances in CMS include the development of state-of-the-art large-radius jet taggers such as ParticleNet-MD [ 4 , 5 ] , with a graph neural network (GNN) architecture, and Global Particle Transformer (GloParT) [ 6 , 7 ] , with an attention-based transformer architecture. These architectures efficiently exploit particle-level pairwise features within jets to infer their origin. ParticleNet-MD classifies jets into 8 categories, specifically targeting \mathrm{X}(\mathrm{q}\mathrm{q}) decays, while GloParT targets a broader set of 37 classes, including \mathrm{X}(\mathrm{V}\mathrm{V}) decays and top quark-antiquark pairs ( \mathrm{t\overline{t}} ). The tagger performance is quantified via receiver operating characteristic (ROC) curves. Figure 1 (left) shows the performance of ParticleNet-MD for \mathrm{H}\to\mathrm{b\overline{b}} tagging, achieving strong QCD multijet background rejection and outperforming DeepAK8-MD [ 8 ] , DeepDoubleX [ 9 ] , and the double-b tagger [ 10 ] . Figure 1 (middle) highlights GloParT’s performance for \mathrm{H}\to\mathrm{W}\mathrm{W} tagging, with results shown split by final state. Moreover, large ML models like GloParT can be fine-tuned on specific signal topologies using transfer learning, where the model is further trained on a smaller, targeted dataset. In the one-lepton ( 1\ell ) channel of the boosted \mathrm{H}\to\mathrm{W}\mathrm{W} analysis [ 11 ] , a fine-tuned GloParT model ( P(\mathrm{H}_{1\ell}) ) was developed to capture the signature of isolated leptons overlapping with large-radius jets that was not well represented in the original training, leading to a nearly 70% increase in expected significance for that channel. Figure 1 (right) shows the performance of P(\mathrm{H}_{1\ell}) against a background consisting of \mathrm{t\overline{t}} , \mathrm{W}\text{+jets} , and QCD multijets before and after fine-tuning. Figure 1: Left: Performance of ParticleNet-MD for \mathrm{H}\to\mathrm{b\overline{b}} tagging [ 5 ] . Middle: GloParT for \mathrm{H}\to\mathrm{W}\mathrm{W} tagging [ 11 ] . Right: Fine-tuned version of GloParT targeting 1\ell final state topologies [ 11 ] . 3 Boosted \mathrm{H}\to\mathrm{W}\mathrm{W} In the analysis in Ref. [ 11 ] , \mathrm{H} boson decay products are merged into a single high- p_{\mathrm{T}} large-radius jet. Two categories are defined: the 0 \ell channel (no isolated leptons, QCD dominated, \mathrm{H} candidate identified using GloParT) and the 1 \ell channel (one isolated e / \mu , dominated by \mathrm{W}\text{+jets} and \mathrm{t\overline{t}} , \mathrm{H} candidate identified using P(\mathrm{H_{1\ell}}) ). The 1 \ell category is further split by production mode, gluon fusion (ggF) and vector boson fusion (VBF). The 0 \ell category is categorized by a GloParT tagger cut and a selection on missing transverse momentum: p_{\mathrm{T}}^{\text{miss}}>20 GeV . The QCD multijets background in 0 \ell is estimated using extrapolation from data in control region. Finally, since GloParT is trained in Monte Carlo (MC) simulation, the Lund jet plane reweighting technique [ 12 , 13 ] is used to calibrate it in data. This method provides a generalized calibration procedure that can be applied to jets with any number of prongs, in particular where no SM analogue for the signal exists. The signal is extracted from a binned maximum likelihood fit to the H candidate softdrop mass ( m_{\mathrm{SD}} ) distribution, where m_{\mathrm{SD}} is the jet mass after softdrop grooming [ 14 ] . The results are shown in Fig. 2 (left). The observed significance for the combined result is 0, while the expected significance is 1.76\sigma . This is the first dedicated study of boosted \mathrm{H}\to\mathrm{W}\mathrm{W} decays in CMS. 4 Boosted \mathrm{H}\mathrm{H}\to\mathrm{b\overline{b}}\mathrm{V}\mathrm{V} In the analysis in Ref. [ 7 ] , large-radius jets are identified using ParticleNet-MD for \mathrm{H}\to\mathrm{b\overline{b}} and GloParT for \mathrm{H}\to\mathrm{W}\mathrm{W} . The mass observable for signal extraction is the H candidate mass after a regression is applied to improve the mass resolution. The signal region is categorized by production mode (ggF/VBF), and the QCD multijets background estimation is performed using extrapolation from data in control region. Similar to the boosted \mathrm{H}\to\mathrm{W}\mathrm{W} analysis, GloParT is calibrated using Lund jet plane reweighting. The results are shown in Fig. 2 (right) and delivers the second-strongest CMS constraints on the the quartic \mathrm{V}\mathrm{V}\mathrm{H}\mathrm{H} coupling modifier \kappa_{2\mathrm{V}} [ 15 ] , the most sensitive one being boosted \mathrm{H}\mathrm{H}\to 4\mathrm{b} [ 16 ] . Figure 2: Left: Fitted signal strength and significance for the \mathrm{H}\to\mathrm{W}\mathrm{W} analysis [ 11 ] . Right: Upper limits on the inclusive \mathrm{H}\mathrm{H} production cross section as a function of \kappa_{2\mathrm{V}} [ 7 ] . 5 Boosted \mathrm{V}\mathrm{H}\to\mathrm{b\overline{b}} At high p_{\mathrm{T}} , the \mathrm{V}\mathrm{H} production mode with the \mathrm{H}\to\mathrm{b\overline{b}} final state has the largest branching fraction and SM cross section. This analysis [ 17 ] is motivated by the VBF excess observed in Ref. [ 18 ] , which could be a sign of an enhanced coupling to vector bosons. Events are selected using jet triggers, a lepton veto, and p_{\mathrm{T}}^{\text{miss}}<140\,\text{Ge\hskip-0.80002ptV} to suppress \mathrm{t\overline{t}} background. ParticleNet-MD identifies both \mathrm{V}\to\mathrm{q\overline{q}} and \mathrm{H}\to\mathrm{b\overline{b}} jets. Backgrounds are estimated from MC, except QCD multijets, which is estimated using extrapolation from data in control region. The \mathrm{V} candidate jet m_{\mathrm{SD}} distribution is used for signal extraction (Fig. 3 , left). The fitted signal strengths for \mathrm{V}\mathrm{H} and \mathrm{V}\mathrm{Z} production, denoted \mu_{\mathrm{V}\mathrm{H}} and \mu_{\mathrm{V}\mathrm{Z}} respectively, are extracted following Ref. [ 19 ] , and a combined fit over the full data set is performed for the final result. The ggF, VBF, and \mathrm{t\bar{t}H} processes are normalized and fixed to their SM predictions. The combined \mu_{\mathrm{V}\mathrm{H}} is measured (expected) to be \mu_{\mathrm{V}\mathrm{H}}=0.72^{+0.75}_{-0.71}\;(1.00^{+0.67}_{-0.62}) , and the combined \mu_{\mathrm{V}\mathrm{Z}} is measured (expected) to be \mu_{\mathrm{V}\mathrm{Z}}=0.09\pm 0.63\;(1.00\pm 0.58) . This corresponds to an observed (expected) SM significance of 1.00(1.64)\sigma for \mathrm{V}(\mathrm{q\overline{q}})\mathrm{H}(\mathrm{b\overline{b}}) and 0.15(1.76)\sigma for \mathrm{V}(\mathrm{q\overline{q}})\mathrm{Z}(\mathrm{b\overline{b}}) , marking the first dedicated CMS search for boosted \mathrm{V}\mathrm{H}\to\mathrm{b\overline{b}} . 6 \mathrm{H}\to\mathrm{Z}\mathrm{Z}/\mathrm{b\overline{b}}+\gamma Finally, the first \gamma H results at the LHC are presented [ 20 ] . Although the SM cross section is small, the observation of such a signal would provide clear evidence of BSM physics. The analysis is split into two categories: (i) \gamma H \to ZZ \to 4 \ell , with four isolated leptons, and (ii) \gamma\mathrm{H}\to\mathrm{b\overline{b}} , where a single large-radius jet is identified using

Read the complete text