Search for anomalies in vector-boson fusion production of the Higgs boson in H(to gamma gamma) jj events using 164 fb-1 of pp collision data collected at sqrt s =13.6 TeV with the ATLAS detector

arXiv:2603.20087 · hep-ex · Submitted 2026-03-20 · Read on arXiv

Listen

Radio episode about this paper

Transcript

Introduction to the show: ident: High Energy Physics: Experiment. The day in high-energy experiment: detectors, searches and limits.

Kai: We're looking at the paper: "Search for anomalies in vector-boson fusion production of the Higgs boson in H(to gamma gamma) jj events using 164 fb-1 of pp collision data collected at sqrt s =13.6 TeV with the ATLAS detector".

Mira: Here's the gist of the paper. This article presents two distinct studies investigating Higgs boson properties using vector-boson fusion (VBF) production in association with a gamma gamma jj final state, utilizing 164 fb-1 of proton–proton collisions at sqrt s = 13.6 TeV recorded by the ATLAS detector.

Kai: The first study focuses on constraining CP-odd couplings beyond the Standard Model within the Standard Model Effective Field Theory (SMEFT) framework, employing matrix element-based optimal observables expressed in the Warsaw basis.

Mira: The second study leverages angular distributions to probe Higgs boson couplings to longitudinally and transversely polarized W and Z bosons produced in association with the Higgs boson.

Kai: A key methodological advancement across both analyses is the implementation of a new neural network-based classification algorithm, which significantly improves the separation of the VBF signal from background processes by employing a multi-class classifier based on a neural network.

Paper summary: Kai: So, to wrap up this paper on Search for anomalies in vector-boson fusion production of the Higgs boson in H(to gamma gamma) jj events using one hundred sixty-four fb negative one of proton-proton collision data collected at sqrt s equals thirteen point six TeV with the ATLAS detector <ref:2603.20087#pg1,Search for anomalies in vector-boson fusion production of the Higgs boson>.

Mira: It’s a search for anything that deviates from what we expect in how the Higgs boson interacts with these vector bosons, both in terms of CP violation and polarization dependencies.

Lev: Essentially, they used a significant amount of data to check if the Standard Model holds up when you look at these specific production channels.

Kai: The main implication is that while their constraints are consistent with the Standard Model, they’ve pushed those constraints quite far, especially by combining Run two and Run three results for the CP odd Wilson coefficient c H W f.

Mira: That fifty percent improvement over run two is substantial because it means we can rule out a wide range of new physics effects in that specific coupling.

Lev: For the polarization studies, they tightened those scale factors a L and a T considerably by adding more information to their fits.

Kai: So, while nothing dramatic has been found yet, this work sets the bar for how sensitive we can get when looking at Higgs couplings through vector boson fusion in these specific channels.

Conclusion: Kai: So, we're wrapping up this look at vector boson fusion production of the Higgs boson in those specific gamma gamma jj events using that big dataset from ATLAS at thirteen point six TeV.

Mira: The paper is titled "Search for anomalies in vector-boson fusion production of the Higgs boson in H(to gamma gamma jj) events using one hundred sixty-four fb negative one of proton-proton collisions recorded by the ATLAS detector."

Lev: It’s basically checking if any weird physics shows up when we look at how the Higgs couples to these vector bosons, both through CP violation and polarization.

Kai: The authors found that their results line up with what we expect from the Standard Model in both cases. No significant CP odd contributions or deviations in those polarization couplings were seen.

Mira: But they did get tighter constraints on a specific thing called the CP odd Wilson coefficient c H W f by combining data from two different runs, run two and run three.

Lev: That combination actually improved their confidence level interval for that coefficient c H W f by fifty percent compared to just looking at run two alone.

Kai: And for those polarization scale factors a L and a T, they also got tighter bounds when they included both the shape information and the rate information in their fits.

Mira: So what does this mean for us? It means we've gotten much better at setting limits on new physics effects in these specific Higgs interactions using this kind of data.

Lev: It tells us that if there is new physics, it’s not showing up in these specific channels with the level of effect they observed.

Kai: The next thing to consider is how sensitive this method actually is when we move to even higher energy colliders or look at different production mechanisms for the Higgs.

The ATLAS Collaboration

EUROPEAN ORGANISATION FOR NUCLEAR RESEARCH (CERN) · ATLAS Collaboration

hep-ex

Submitted: 2026-03-20

Updated: 2026-11-03

Comments: 52 pages in total, author list starting page 36, 10 figures, 2 tables, published in JHEP. All figures including auxiliary figures are available at: https://atlas.web.cern.ch/Atlas/GROUPS/PHYSICS/PAPERS/HIGP-2024-23

License: http://creativecommons.org/licenses/by/4.0/

Importance score: 83/100

The gist: This article presents two distinct studies investigating Higgs boson properties using vector-boson fusion (VBF) production in association with a gamma gamma jj final state…

Key concepts

Vector-Boson Fusion (VBF)
This is a specific production mechanism where two vector bosons (like W or Z bosons) fuse to create the Higgs boson. Researchers use this process to study the Higgs boson's properties and search for new physics beyond the Standard Model.
CP-odd Couplings
These couplings represent hypothetical interactions that violate Charge-Parity symmetry in particle physics. The study constrained these couplings using kinematic observables to ensure they align with predictions from the Standard Model, finding no significant deviations.
Polarization-Dependent Couplings
This refers to how the Higgs boson interacts differently depending on whether the W or Z bosons it couples to are longitudinally or transversely polarized. The analysis used angular distributions to measure these factors, confirming they are consistent with Standard Model predictions.

Terminology

Summary

This article presents two distinct studies investigating Higgs boson properties using vector-boson fusion (VBF) production in association with a gamma gamma jj final state, utilizing 164 fb-1 of proton–proton collisions at sqrt s = 13. p. 0

Where the paper says it

EUROPEAN ORGANISATION FOR NUCLEAR RESEARCH (CERN) Submitted to: JHEP CERN-EP-2026-075 23rd March 2026 Search for anomalies in vector-boson fusion production of the Higgs boson in 𝑯(→ 𝜸𝜸) 𝒋 𝒋 events using 164 fb−1 of 𝒑 𝒑 collision data collected at √ 𝒔 = 13. 6TeV with the ATLAS detector The ATLAS Collaboration This article details two studies of Higgs boson properties using the vector-boson fusion production mode and the 𝛾𝛾 𝑗 𝑗 final state.

6 TeV recorded by the ATLAS detector. p. 7
Where the paper says it

4 Data and simulated event samples The 𝑝 𝑝 collision data at √ 𝑠 = 13. 6 TeV used in these analyses were recorded with the ATLAS detector between 2022 and 2024, corresponding to a total integrated luminosity of 163.

The first study focuses on constraining CP-odd couplings beyond the Standard Model within the Standard Model Effective Field Theory (SMEFT) framework, employing matrix element-based optimal observables expressed in the Warsaw basis. p. 0

Where the paper says it

6 TeV proton–proton collisions recorded by the ATLAS experiment at the Large Hadron Collider. The first study employs matrix element-based optimal observables to constrain CP-odd couplings beyond the Standard Model within the Standard Model Effective Field Theory framework, expressed in the Warsaw basis.

The second study leverages angular distributions to probe Higgs boson couplings to longitudinally and transversely polarized W and Z bosons produced in association with the Higgs boson. p. 0
Where the paper says it

The first study employs matrix element-based optimal observables to constrain CP-odd couplings beyond the Standard Model within the Standard Model Effective Field Theory framework, expressed in the Warsaw basis. The second study exploits angular distributions to probe the Higgs boson’s couplings to longitudinally and transversely polarised 𝑊 and 𝑍 bosons in the production of the Higgs boson.

A key methodological advancement across both analyses is the implementation of a new neural network-based classification algorithm, which significantly improves the separation of the VBF signal from background processes by employing a multi-class classifier based on a neural network. p. 2

Where the paper says it

Analysis regions are defined based on requirements on the kinematic properties of the final-state particles and eventlevel quantities. To enhance the separation of the VBF signal from the background processes, a multi-class classifier based on a neural network is employed.

Furthermore, for all Monte Carlo samples, this work represents a significant step forward as it relies exclusively on the AF3 detector simulation for all samples, validating this fast simulation tool. p. 7
Where the paper says it

Monte Carlo (MC) simulation samples were used to model the signal processes, assist in background estimates, and derive modelling uncertainties. All MC simulation samples were processed using a hybrid approach: the geometry and response of the ATLAS Inner Detector and muon spectrometer were simulated using Geant4 [48] (G4), while the calorimeter response was modelled using AtlFast3 (AF3) [49].

The results from both studies are consistent with Standard Model expectations; no significant CP-odd contributions or deviations in polarization-dependent couplings were observed. p. 2

Where the paper says it

To date, studies of Higgs-boson couplings to longitudinally and transversely polarised vector bosons have been performed only in 𝐻(→ 𝑊𝑊∗ → 𝑒𝜈𝜇𝜈) 𝑗 𝑗 events [18]. The two studies presented in this article probe CP- and polarisation-dependent observables constructed from the final-state particles in selected 𝛾𝛾 𝑗 𝑗 events to test for deviations from the SM expectations.

Specifically concerning the CP studies, constraints on the CP-odd Wilson coefficient c H W f were determined using kinematic observables, yielding 68% (95%) confidence level intervals of [-0. p. 6
Where the paper says it

6 TeV VBF H → γ γ (b) Figure 2: Distributions of (a) the OO observable shown for various configurations of the Wilson coefficient 𝑐𝐻𝑊f, and (b) the ΔΦ𝑗 𝑗 observable for various configurations of the 𝑎L and 𝑎T parameters. These comparisons are performed in VBF events at the reconstruction level using the predictions of the MadGraph5_aMC@NLO + Pythia 8 [39, 41] generators.

06, 0.54] (or [-0.35, 0.88]). Comparing these Run-3 constraints to a similar analysis using 140 fb-1 of data at sqrt s = 13 TeV, the Run-3 results show a substantial 38% improvement, with 29% of this gain attributed to refined analysis techniques, most notably the new NN-based classification algorithm. p. 0
Where the paper says it

To maximise the sensitivity, the constraints of the CP-odd couplings are combined with those from a previous analysis performed in 𝛾𝛾 𝑗 𝑗 events in a data sample of proton–proton collisions at √ 𝑠 = 13 TeV, corresponding to an integrated luminosity of 140 fb−1 . A significant improvement with respect to the previous analysis is achieved through the implementation of a new neural network-based classification algorithm.

Combining Run-2 and Run-3 results further narrows the confidence level intervals for c H W f to [0. p. 1
Where the paper says it

2 Theoretical systematic uncertainties 17 9 Results 18 9. 1 Run-3 results 18 9.

01, 0.49] (or [-0.23, 0.75]), representing a total improvement of 50% over the Run-2 results. p. 1
Where the paper says it

2 Theoretical systematic uncertainties 17 9 Results 18 9. 1 Run-3 results 18 9.

For the polarization studies, two fit configurations were utilized to determine the polarization-dependent scale factors (a L and a T). p. 2

Where the paper says it

The results of the CP studies are interpreted within the Standard Model Effective Field Theory (SMEFT) framework [32] using the Warsaw basis [33, 34]. Complementing these CP studies, the results from the polarisation studies are interpreted as polarisation-dependent coupling scale factors between the Higgs boson and the vector bosons [35].

Using shape information from the delta phi jj distribution alone, the 68% (95%) confidence level intervals were found to be a L in [0. p. 5
Where the paper says it

At the reconstruction level, the momentum fractions of the initial-state partons are derived in the reference frame of the ATLAS detector as: 𝑥 reco 1,2 = 𝑚𝐻 𝑗 𝑗 · 𝑒 ±𝑦𝐻 𝑗 𝑗 √ 𝑠 (5) by exploiting energy and momentum conservation, where 𝑚𝐻 𝑗 𝑗 and 𝑦𝐻 𝑗 𝑗 are the invariant mass and rapidity of the Higgs boson + dijet system, and √ 𝑠 is the centre-of-mass energy. A detailed description of this calculation can be found in Ref.

94, 1.36] (or [0.84, 2.02]) and a T in [0.74, 1.06] (or [0.50, 1.20]). When both shape and rate information were incorporated into the fits, these constraints were significantly tightened to a L in [0.98, 1.11] (or [0.93, 1.17]) and a T in [0.80, 1.03] (or [0.70, 1.14]). Overall, all derived results are consistent with SM expectations regarding CP-odd contributions and polarization-dependent couplings. p. 2
Where the paper says it

To date, studies of Higgs-boson couplings to longitudinally and transversely polarised vector bosons have been performed only in 𝐻(→ 𝑊𝑊∗ → 𝑒𝜈𝜇𝜈) 𝑗 𝑗 events [18]. The two studies presented in this article probe CP- and polarisation-dependent observables constructed from the final-state particles in selected 𝛾𝛾 𝑗 𝑗 events to test for deviations from the SM expectations.

Green — the paper says it, and the page is printed. Red — no passage carries it, so treat that sentence as unverified. The same check runs on your own text: check a summary against its source, $5.

Improvements for AI systems

1

Improve event classification accuracy by implementing a new neural network-based classification algorithm that is trained using a data-driven approach based on a flow-matching method to generate high-statistics, data-derived continuum background samples. This allows the system to accurately model non-resonant diphoton and photon plus jet backgrounds in VBF events, which are typically poorly described by standard Monte Carlo simulations.

2

Enhance sensitivity to CP violation by optimizing the optimal observable (OO) defined as the ratio of the interference term to the SM-only squared matrix element for a specific CP-odd coupling c H W f. This observable is constructed to be CP-odd, meaning any observed asymmetry would provide direct evidence of CP violation in the H V V vertex.

3

Improve sensitivity to polarisation-dependent couplings by using the absolute azimuthal angular difference between the two VBF-tagged jets (Delta Phi jj). This variable is employed because it is highly sensitive to the polarisation states of the vector bosons in the VBF production mode and provides sensitivity to these polarisation-dependent scale factors.

4

Develop a combined fitting procedure for Run-2 and Run-3 data that treats theoretical uncertainties in the parton shower model as fully correlated across both datasets, while treating QCD scale uncertainty as uncorrelated for the VBF signal. This combination is shown to result in a ∼ 5% broadening of the c H W f constraint relative to the original Run-2 publication.

5

Implement a two-fold cross-validation scheme when training neural networks by randomly dividing events into two equal subsamples and training independent networks on each. This procedure helps protect against potential biases in the analysis, as the output distributions from both NNs are merged for both simulated and collision data.

Abstract

This article details two studies of Higgs boson properties using the vector-boson fusion production mode and the γγjj final state. Both efforts are based on a data sample corresponding to 164 fb-1 of sqrt s =13.6 TeV proton--proton collisions recorded by the ATLAS experiment at the Large Hadron Collider. The first study employs matrix-element-based optimal observables to constrain CP-odd couplings beyond the Standard Model within the Standard Model Effective Field Theory framework, expressed in the Warsaw basis. The second study exploits angular distributions to probe the Higgs boson's couplings to longitudinally and transversely polarised W and Z bosons in the production of the Higgs boson. To maximise the sensitivity, the constraints of the CP-odd couplings are combined with those from a previous analysis performed in γγjj events in a data sample of proton--proton collisions at sqrt s =13 TeV, corresponding to an integrated luminosity of 140 fb-1. A significant improvement with respect to the previous analysis is achieved through the implementation of a new neural-network-based classification algorithm. All measurements are in agreement with the Standard Model prediction of a CP-even Higgs boson with the expected relative coupling strengths to longitudinally and transversely polarised vector bosons.

Sources

Related papers