Squeezed double- b tagging for light hadronic resonances at the LHC

summary

The gist

LHC signatures of a light pseudoscalar in a flipped two-Higgs scenario: the usefulness of boosted b¯b pairs provides a method to search for a light pseudoscalar state in the flipped 2HDM by…

In short

Researchers searched for a light pseudoscalar particle in a flipped two-Higgs-doublet model by looking for boosted b-quark pairs. They used high-energy initial state radiation to create a highly collimated 'squeezed' bbar pair signature. A specialized Boosted Decision Tree (BDT) successfully identified these events, achieving a 5$\\sigma$ discovery potential for 30 GeV and 50 GeV states with minimal data.

Key concepts

Flipped Two-Higgs-Doublet Model (2HDM)
This is an extension of the standard Higgs model where the symmetry between the two Higgs doublets is 'flipped'. This specific configuration allows for a light pseudoscalar particle to exist without violating fundamental physics rules, which is necessary for this search.
Boosted bbar Pairs
The search focuses on finding pairs of bottom quarks (b and $\bar{b}$) that are produced with very high transverse momentum. When the parent particle decays into these quarks, the resulting b-quarks are so close together in space that they appear as a single, tightly packed jet.
Jet Substructure Tagging
This involves using advanced machine learning tools like XGBoost to analyze the internal structure of a single jet. The tagger looks at detailed track information, such as the distance and momentum of individual tracks within the jet, to distinguish a rare signal from overwhelming background noise.

Terminology used across episodes

This episode discusses

The paper

Squeezed double- b tagging for light hadronic resonances at the LHC · Read on arXiv

School of Physical Sciences, Indian Association for the Cultivation of Science · Department of Physical Sciences, Indian Institute of Science Education and Research Kolkata

Transcript

Introduction to the show: ident: High Energy Physics: Phenomenology. The day in particle phenomenology: models, signatures and constraints.

Kai: I'm Kai.

Mira: And I'm Mira, here with Lev.

Kai: We're looking at the paper: "Squeezed double- b tagging for light hadronic resonances at the LHC".

Mira: Here's the gist of the paper. LHC signatures of a light pseudoscalar in a flipped two-Higgs scenario: the usefulness of boosted b¯b pairs provides a method to search for a light pseudoscalar state in the flipped 2HDM by exploiting its boosted decay into bottom quarks Theoretical Framework and Stabilization The paper introduces an extension to the flipped Two-Higgs-Doublet Model by adding a real pseudoscalar singlet, denoted as P, to stabilize the scalar potential and accommodate a light pseudoscalar state without violating perturbative unitarity The minimal flipped 2HDM suffers from severe perturbative unitarity violations when trying to have a light pseudoscalar while satisfying flavor physics constraints, which necessitates extremely large quartic couplings This extension allows the physical light mass ma to be naturally light (e.g., < 60 GeV) even if the doublet mass parameter m2AA is large, by mixing with the singlet field P The physical states are related to the gauge eigenstates via a mixing angle θ determined by tan 2θ = -2m2AP/m2AA - mP P Novel Production and Signal Identification The authors pivot away from the suppressed electroweak channel, choosing instead a QCD-driven production mechanism where the pseudoscalar is produced with high energy (boosted) and recoils against a hard initial state radiation (ISR) jet This process is defined as pp → a + j(j) → (b¯b) + j(j) The high-pT ISR recoil heavily boosts the light scalar, forcing the b and ¯b quarks from their decay into a highly collimated topology, yielding a distinctive signature This kinematic consequence results in an angular separation ∆Rbb ≈ 2ma/pT a, which for a 30 GeV pseudoscalar with pT a ≈ 200 GeV yields ∆Rbb ≈ 0.3 Jet Substructure Tagging Strategy To overcome the mixing-suppression of standard production channels, the analysis employs a specialized Boosted Decision Tree (BDT) strategy to identify these squeezed b¯b pairs within a single jet The BDT is trained using the XGBoost framework, relying predominantly on tracking information such as the sorted 2D and 3D impact parameters (IP) of the tracks within the jet Key discriminating features include the multiplicity of highly displaced tracks, Ntrk(0.1 < IP3D < 10 mm), and their relative transverse momentum fraction, Pp trk T p jet T (0.1 < IP3D < 10 mm) Search Sensitivity and Discovery Potential The search strategy demonstrates strong discovery potential, achieving a 5σ discovery for the 30 GeV and 50 GeV pseudoscalar states (BP1 and BP2) with merely 36 fb−1 of data This success is achieved by leveraging the inherent large initial rate of the pp → a+j(j) channel combined with the immense discrimination power of the tracking-based substructure tagger The results show that a definitive 5σ discovery for the 30 and 50 GeV states is achievable with merely 36 fb−1 of data Furthermore, the model’s parameter space can be effectively probed and constrained using already existing LHC datasets The analysis focuses on discriminating against the dominant QCD multijet background, which has two components: irreducible gluon splitting processes (g → b¯b) and reducible events containing light-quark, gluon, or charm jets The soft-drop mass (mSD) of the jet proves to be a highly effective discriminant in this boosted regime, as the signal shows a distinct, sharp resonance peak centered at the pseudoscalar mass The BDT classifier heavily prioritizes track-based substructure and displacement observables, confirming that the presence and kinematics of multiple displaced tracks provide the most robust discrimination against the QCD multijet background The study concludes that a 5σ discovery for the 30 and 50 GeV states is achievable with merely 36 fb−1 of data The proposed search strategy based on the identification of a squeezed b-pair opens up an avenue which may be of wide applicability The model-dependence here is minimal, but our analysis based on the identification of a squeezed b-pair opens up an avenue which may be of wide applicability The results highlight a major phenomenological success of this QCD-driven search strategy Pruning redundant reasoning tokens The paper is organized as follows.

Kai: In Section II, we introduce the theoretical framework of the singlet-extended flipped 2HDM, detailing the scalar potential, mass matrix diagonalization, and modified Yukawa interactions Section III describes the rigorous theoretical and experimental constraints imposed on the model’s parameter space Our collider analysis strategy, which includes event generation, boosted-topology physics, and BDT tagging methodology, is presented in Section IV In Section V, we present the signal-to-background discrimination results and the projected signal significances Finally, we summarize our findings and conclude in Section VI How the training signal is built The analysis utilizes two distinct BDT classifiers: one for jet substructure flavor tagging and another for event-level signal-to-background discrimination The double-b Jet Tagger BDT was trained using 40 kinematic and track-based input features, including jet transverse momentum (p jet T), the number of tracks (Ntrk), and various impact parameter variables The Event-Level Signal-Background Discriminating BDT utilizes 97 input features capturing global event topology and interobject kinematics, such as the soft-drop mass (mSD(j 2b 1)) and angular correlations Where the gains come from The success of the Event BDT stems from its ability to exploit non-linear correlations between high-ranking features, such as the transverse momentum and soft-drop mass of the signal jet The correlation between the transverse momentum and the soft-drop mass of the signal jet clearly demonstrates how the signal maintains a tight resonant mass structure across the high-pT spectrum, whereas the background exhibits a broad, unstructured smear This strong correlation reflects a high true-positive rate, demonstrating that the tagger is highly efficient at identifying the squeezed topology and that the soft-drop grooming successfully recovers the hard two-body decay kinematics The signal significance, S, is calculated using the standard profile likelihood ratio asymptotic approximation The projected significance for BP1 (ma = 30 GeV) is 5.78 at 140 fb−1 Pruning redundant reasoning tokens The authors acknowledge the use of the Kepler HPC facility at IISER Kolkata S.S.

Mira: thanks CSIR for funding SS, and RKS acknowledge the hospitality of IACS, Kolkata, where part of the work was carried out B.M.

Kai: thanks the Department of Atomic Energy, Government of India, for support in the form of a Raja Ramanna Chair position Pruning redundant reasoning tokens The paper is organized as follows In Section II, we introduce the theoretical framework of the singlet-extended flipped 2HDM, detailing the scalar potential, mass matrix diagonalization, and modified Yukawa interactions Section III describes the rigorous theoretical and experimental constraints imposed on the model’s parameter space Our collider analysis strategy, which includes event generation, boosted-topology physics, and BDT tagging methodology, is presented in Section IV In Section V, we present the signal-to-background discrimination results and the projected signal significances Finally, we summarize our findings and conclude in Section VI Pruning redundant reasoning tokens The authors acknowledge the use of the Kepler HPC facility at IISER Kolkata S.S.

Title and authors: Kai: Let's talk about the title and who wrote this. "Squeezed double- b tagging for light hadronic resonances at the LHC" tells you exactly what they’re doing here, which is using a specific way to look at bottom quark pairs to find these resonances >

Mira: And the authors are Dilip Kumar Ghosh, Biswarup Mukhopadhyaya, Sirshendu Samanta, and Ritesh K. Singh >

Lev: From my side, it tells me they are focused on the experimental realization of this model at the Large Hadron Collider, which is where we actually look for these particles >

Kai: What this means plainly is that they are proposing a specific search strategy using boosted bottom quark pairs to hunt for a light pseudoscalar state within this flipped two-Higgs scenario >

Mira: It’s about taking a particle that decays dominantly into bottom quarks, which makes it hard to see directly, and finding a way to use its boosted decay into those heavy quarks as the signal >

Lev: So the implication is that instead of relying on those electroweak channels where things get suppressed, they are pivoting to a QCD-driven production mechanism >

Kai: Exactly. They're using high energy initial state radiation, or ISR jet recoil, to boost the pseudoscalar so that its decay products look very collimated >

Mira: That collimation is key because it leads to a specific angular separation between those two bottom quarks in terms of their radius relative to the boost of the pseudoscalar >

The paper's summary: Kai: So, what’s the paper actually summarizing? They are taking that boosted topology and using it to identify these squeezed bottom quark pairs within a single jet >

Mira: They use a specialized Boosted Decision Tree, or BDT, trained with XGBoost, relying heavily on tracking information from the impact parameters of the tracks inside that jet >

Lev: For me, what this means is they’re saying we can cut through the overwhelming QCD multijet background by focusing on those displaced tracks and how many there are >

Kai: They say that this approach successfully reduces their search to pseudoscalar masses less than or equal to fifty GeV, which fits with CMS searches they've done earlier >

Mira: The summary points out that they achieved a five sigma discovery potential for the thirty GeV and fifty GeV states with just thirty six femtobarns of data > <ref:2604.06937#pg3>

Lev: That’s a lot of data to get that kind of sensitivity, especially when you factor in the overhead of building those complex BDT classifiers on real hardware >

The paper's improvements: Kai: They actually suggest some specific improvements for the analysis. One big suggestion is implementing a stricter kinematic pre-selection cut defined in equation fourteen >

Mira: That cut requires two bottom quarks, N2b equals one, and at least one other bottom quark plus zero or one bottom quark to be present, plus that the jet transverse momentum must be greater than one hundred giga-electron volts >

Lev: That pre-selection sounds like a smart way to isolate the signal topology right from the start, reducing the noise before you even run it through the complex tagging machinery >

Kai: And they also suggest enhancing their jet substructure analysis by training a dedicated BDT using XGBoost based on forty kinematic and track-based input features >

Mira: That’s for identifying those squeezed b-pair signals more robustly, focusing on the sorted two dimensional and three dimensional impact parameters of the tracks within the jet >

Lev: Using those forty features sounds like a solid way to build a tagger that isn't just looking at one or two variables, which is good for real hardware implementation >

Conclusion: Kai: So to wrap up, the paper shows that this search strategy based on identifying these squeezed b-pairs has strong discovery potential with only thirty six femtobarns of data for the thirty and fifty giga-electron volt states >

Mira: They conclude that this method opens up an avenue that might be applicable in other areas too, even though they state their model dependence is minimal >

Lev: From an engineering standpoint, what this means is that we have a concrete signal to aim for in the next run, and it’s achievable with the data we already have available >

Kai: So they’ve shown that by using this QCD-driven search strategy, you can achieve a five sigma discovery for these specific light states >

Mira: And that this method of identifying a squeezed b-pair is a useful tool because it works even against the dominant QCD multijet background >

Lev: It gives us a clear roadmap on how to look for new physics in this model, and that’s valuable information for our error correction research too >

More episodes

← Home