LHC signatures of a light pseudoscalar in a flipped two-Higgs scenario: the usefulness of boosted $b{\bar b}$ pairs

This paper proposes that boosted $b\bar{b}$-pair tagging combined with boosted decision tree analysis can achieve a 5–10$\sigma$ signal significance for a light pseudoscalar in a flipped two-Higgs doublet model at the LHC with 3 $ab^{-1}$ of integrated luminosity, effectively overcoming challenges posed by dominant $b\bar{b}$ decays and suppressed $Z$-associated production.

The Gist

Imagine the Standard Model of particle physics as a very successful, well-oiled machine that explains almost everything we see in the universe. But, like any machine, it has a few loose screws. Physicists suspect there's a hidden part—a "light pseudoscalar" particle—that the machine is missing.

This paper is a detective story about how to find this missing piece at the Large Hadron Collider (LHC), the world's biggest particle smasher. Here is the story of their hunt, explained simply.

The Mystery: A Ghost in the Machine

The scientists are looking for a specific type of particle (let's call it the "Ghost") that is very light (between 20 and 60 GeV) and decays almost instantly into a pair of bottom quarks (heavy particles).

The Problem:

1. The Camouflage: The Ghost loves to hide in a crowd. When it decays, it creates two bottom quarks that look exactly like the billions of "junk" particles (background noise) constantly created by the LHC. It's like trying to find a specific red marble in a pile of a billion red marbles.
2. The Theoretical Glitch: In the simplest version of their theory, making this Ghost light requires the math to break down. It's like trying to build a bridge with a span so wide that the laws of physics say the bridge must collapse. The forces holding it together become too strong, and the theory stops making sense.

The Solution: Adding a Secret Ingredient

To fix the bridge without making it collapse, the authors propose adding a new ingredient: a "Singlet" particle. Think of this as a secret agent that mixes with the Ghost.

• The Trade-off: This mix saves the math (the bridge stands!), but it makes the Ghost even harder to find. The new "mixed" Ghost is shy; it refuses to interact with the usual ways we try to catch it (like the "Z boson" channel). It's like the Ghost put on a disguise that makes it invisible to the standard police scanners.

The New Strategy: The "Squeezed" Pair

Since the standard police scanners don't work, the team had to invent a new way to catch the Ghost. They decided to look for it when it is produced in a very specific, high-energy crash.

The Analogy: The High-Speed Train
Imagine the Ghost is a passenger on a train.

• Normal Scenario: The train moves slowly. The passenger steps off and walks away. You can see them clearly, but they look just like everyone else in the crowd.
• The New Strategy: The team waits for the train to be moving at supersonic speeds (a "boosted" state). When the passenger steps off at this speed, they don't just walk; they are flung forward so fast that they and their twin (the other bottom quark) are squeezed together so tightly that they look like a single, blurry object.

This "squeezed" pair is the key. Because they are so close together, they merge into a single "jet" of particles in the detector. This is a rare signature that the background noise (the billions of junk particles) rarely mimics.

The Detective Work: The "Smart Eye" (BDT)

To spot this squeezed pair, the scientists couldn't just use a ruler. They needed a "Smart Eye"—a computer program called a Boosted Decision Tree (BDT).

Think of the BDT as a super-smart security guard at an airport who has seen a million faces.

• The Clues: The guard doesn't just look at the shape of the jet. They look at the "footprints" left behind.
  • The Footprints: Bottom quarks leave behind heavy, slow-moving particles (B-mesons) that travel a tiny bit before decaying. This leaves a "displaced track" (a footprint slightly off-center).
  • The Signal: A "squeezed" pair leaves two sets of these displaced footprints packed into one tiny space.
  • The Noise: Background junk usually leaves no footprints or just one.
• The Decision: The BDT analyzes thousands of these tiny details (how many footprints, how far they are, how much energy they carry) and gives a score: "Is this a squeezed Ghost pair, or just noise?"

The Results: A Clear Victory

The team ran simulations for the future High-Luminosity LHC (which will run for a long time, collecting a massive amount of data).

• The Outcome: Even with the Ghost being shy and the background being noisy, their new strategy worked.
• The Significance: They found that with enough data, they could spot the Ghost with a confidence level of 5 to 10 sigma. In the world of particle physics, "5 sigma" is the gold standard for discovery—it means there is less than a 1 in 3.5 million chance that the result is a fluke.

The Takeaway

This paper is a brilliant example of scientific problem-solving.

1. The Problem: The theory broke, and the particle was too hard to see.
2. The Fix: They added a new particle to save the theory, which made the target harder to find.
3. The Innovation: Instead of giving up, they changed their hunting ground. They looked for the particle when it was moving so fast that it squished its decay products together, creating a unique "fingerprint" that a smart computer could recognize.

They proved that even if a particle is light, shy, and hiding in a sea of noise, with the right "squeezed" strategy and a smart computer, we can still find it.

Technical Summary

1. Problem Statement

The paper addresses the challenge of detecting a light pseudoscalar ($A$) with a mass between 20 and 60 GeV within the context of a Flipped Two-Higgs-Doublet Model (2HDM).

• Theoretical Instability: In the minimal flipped 2HDM, accommodating a light pseudoscalar while satisfying experimental constraints (specifically a heavy charged Higgs $H^\pm \gtrsim 600$ GeV required by flavor physics like $b \to s\gamma$) forces the scalar quartic couplings ($\lambda_{3,4,5}$) to become excessively large. This leads to a violation of perturbative unitarity at scales well below 1 TeV, rendering the theory inconsistent.
• Detection Difficulty: The light pseudoscalar decays dominantly into a bottom-antibottom pair ($b\bar{b}$). At the Large Hadron Collider (LHC), this channel suffers from overwhelming QCD multijet backgrounds.
• Suppression of Standard Channels: Previous strategies relied on the associated production of the pseudoscalar with a $Z$ boson ($pp \to Z A$). However, the authors propose extending the model with a pseudoscalar singlet ($P$). While this solves the unitarity problem via mixing, it suppresses the couplings of the physical light state ($a$) to gauge bosons and fermions by a factor of $\sin \theta$ (mixing angle). Consequently, the $Z$-associated production rate is suppressed by $\sin^2 \theta$, making the standard electroweak search channel unviable.

2. Methodology

The authors propose a novel search strategy focusing on QCD-driven gluon fusion production with a hard recoil jet, utilizing advanced jet substructure and machine learning techniques.

A. Theoretical Framework

• Model Extension: The flipped 2HDM is extended by adding a real pseudoscalar singlet $P$.
• Mixing Mechanism: The singlet mixes with the 2HDM pseudoscalar ($A_{2HDM}$) via a trilinear coupling. This creates two physical mass eigenstates: a heavy $A$ and a light $a$.
• Resolution: The mixing allows the light mass $m_a$ to be small ($<60$ GeV) without requiring large quartic couplings, thereby preserving perturbative unitarity. However, the couplings of $a$ to SM particles are diluted by $\sin \theta$.

B. Production and Kinematics

• Production Mode: The study focuses on the gluon fusion process $gg \to a$, mediated primarily by a top-quark loop (due to the specific Yukawa structure of the flipped model and low $\tan \beta$).
• Recoil Strategy: To trigger on the event and boost the light pseudoscalar, the analysis requires a hard Initial State Radiation (ISR) jet. The process is $pp \to a + j$.
• Boosted Topology: The high transverse momentum ($p_T$) of the ISR jet boosts the light pseudoscalar. This forces the decay products ($b$ and $\bar{b}$) to be highly collimated, merging into a single jet cone (a "squeezed" $b\bar{b}$ pair) rather than two resolved jets.
  • Angular separation: $\Delta R_{b\bar{b}} \approx 2m_a / p_T^a$. For $m_a \sim 30-60$ GeV and $p_T \sim 200$ GeV, $\Delta R < 0.6$.

C. Analysis Strategy & Machine Learning

• Event Generation: Signal and background events were generated using MadGraph5_aMC@NLO (LO, 4-flavor scheme) and showered with PYTHIA8. Detector simulation was performed using Delphes.
• Jet Reconstruction: Jets were clustered using the anti-$k_t$ algorithm ($R=0.5$) and groomed using the Soft-Drop algorithm to remove soft radiation and improve mass resolution.
• Double-b Tagging (Jet Level): A Boosted Decision Tree (BDT) based on XGBoost was trained to distinguish "squeezed" double-$b$ jets from single-$b$ or light-flavor jets.
  • Key Features: The classifier relies heavily on track impact parameters (2D and 3D) and the multiplicity of displaced tracks. Since the signal contains two $B$-hadrons, it produces a higher density of displaced tracks compared to background jets.
• Event-Level Discrimination: A second BDT was employed to separate signal from the remaining QCD background using global event kinematics, including the soft-drop mass of the leading jet, $N$-subjettiness ratios ($\tau_{21}$), and angular correlations with recoil jets.

3. Key Contributions

1. Theoretical Stabilization: Demonstrated that adding a pseudoscalar singlet to the flipped 2HDM resolves the perturbative unitarity violation inherent in the minimal model for light pseudoscalars.
2. Novel Search Channel: Identified that while the singlet mixing suppresses electroweak production channels, the QCD-driven gluon fusion channel remains viable and dominant due to the large gluon luminosity.
3. Boosted $b\bar{b}$ Tagging: Developed a specialized BDT strategy to identify squeezed $b\bar{b}$ pairs within a single jet. This approach leverages the unique substructure of two overlapping $B$-hadron decays (high track multiplicity and displacement) to discriminate against the massive QCD background.
4. Complementarity: The method is specifically optimized for light masses ($m_a \le 50$ GeV), complementing existing CMS searches which are more sensitive to heavier resonances ($m_a \ge 50$ GeV).

4. Results

The analysis was performed for three benchmark points (BPs) with light pseudoscalar masses of 30, 50, and 60 GeV, assuming an integrated luminosity of 3000 fb$^{-1}$ at $\sqrt{s} = 14$ TeV (HL-LHC).

• Background Suppression: The combination of kinematic cuts (requiring a hard ISR jet and $p_T > 100$ GeV), soft-drop mass windows, and the BDT classifiers effectively suppresses the QCD multijet background by several orders of magnitude.
• Signal Significance:
  • With 10% systematic uncertainty, the expected signal significances are:
    • BP1 (30 GeV): $9.7\sigma$
    • BP2 (50 GeV): $6.1\sigma$
    • BP3 (60 GeV): $4.7\sigma$
  • With 20% systematic uncertainty, the significances drop to $4.8\sigma$, $3.0\sigma$, and $2.3\sigma$ respectively.
• Key Finding: The discovery potential is highly sensitive to the control of systematic uncertainties. However, under nominal assumptions, the model predicts a $5\sigma$ discovery for light pseudoscalars in the 30–60 GeV range.

5. Significance

This paper provides a robust theoretical and phenomenological framework for discovering light pseudoscalars in extended Higgs sectors.

• Theoretical Impact: It validates the "singlet-extended flipped 2HDM" as a theoretically consistent model that avoids unitarity violations while accommodating light new physics.
• Experimental Impact: It offers a concrete, optimized search strategy for the HL-LHC. By shifting focus from resolved di-jet topologies to boosted jet substructure, the authors unlock a kinematic regime previously difficult to probe due to background contamination.
• Methodological Advance: The successful application of track-based BDTs to identify "squeezed" double-$b$ jets demonstrates a powerful technique for future searches for light resonances decaying to heavy quarks in high-pileup environments.