Exploring Higgs EFT in $t\bar{t}hh$ at High Luminosity LHC

This paper presents a comprehensive study of the non-resonant $t\bar{t}hh$ production process at the High-Luminosity LHC within the Higgs Effective Field Theory framework, utilizing both cut-based and multivariate analysis techniques to establish the first sensitivity projections for key HEFT couplings and demonstrate the channel's unique potential to probe extended top-Higgs interactions beyond the Standard Model.

The Gist

Imagine the universe as a giant, high-stakes game of billiards played at the speed of light. In this game, the Standard Model is the rulebook that physicists have written down over the last 50 years. It predicts exactly how the balls (particles) should bounce off each other. But, just like any good rulebook, there might be hidden rules or "cheats" we haven't discovered yet. This paper is a detective story about hunting for those hidden rules in a very specific, rare, and chaotic corner of the game.

Here is the breakdown of the research in simple terms:

1. The Rare Event: The "Four-Ball" Collision

The researchers are looking at a specific event at the Large Hadron Collider (LHC), a massive machine that smashes protons together. They are interested in a collision that produces four heavy particles at once:

• Two Top Quarks (the heaviest particles in the universe, like the "bowling balls" of the particle world).
• Two Higgs Bosons (the particles that give other particles their mass, like the "glue" of the universe).

In the standard rulebook, this event is incredibly rare. It's like trying to hit four specific bowling balls with a single cue ball; it happens so rarely that you might wait a lifetime to see it. However, if there are "new physics" (hidden rules), this event might happen much more often, or the balls might fly off in weird directions.

2. The Detective's Toolkit: HEFT

The team uses a framework called Higgs Effective Field Theory (HEFT). Think of HEFT as a "flexible rulebook."

• The standard rulebook is rigid.
• HEFT allows the rules to bend slightly. It introduces "knobs" or couplings (like δκλ, c2, c2g, ctg) that represent how strongly the particles interact.
• If the universe follows the standard rules, these knobs are set to zero. If there is new physics, the knobs are turned to different numbers.

The goal of the paper is to figure out how far these knobs can be turned before the physics breaks, based on what we expect to see at the High-Luminosity LHC (HL-LHC). The HL-LHC is an upgraded version of the current collider that will run for many years, smashing billions more protons together to gather more data.

3. The Challenge: Finding a Needle in a Haystack

The problem is that the "haystack" (background noise) is huge.

• The Signal: The rare t¯thh event (Top-Top-Higgs-Higgs).
• The Noise: Common collisions that look almost the same, like a Top-Top pair plus some random junk (jets).

The researchers explain that if you just count the number of particles, the noise drowns out the signal. It's like trying to hear a whisper in a stadium full of screaming fans.

4. The Strategy: Two Ways to Listen

To find the signal, the team tested two different strategies:

Strategy A: The "Cut-Based" Approach (The Strict Bouncer)
Imagine a bouncer at a club with a very strict list of rules. "If you don't have exactly 6 tickets, you can't get in."

• They set hard rules: "We only want events with at least 6 jets (sprays of particles) and 5 of them must be 'b-jets' (heavy-flavored particles)."
• They also looked at how much energy was in the collision.
• Result: This method is good at filtering out the noise, but it's a bit blunt. It throws away some of the signal along with the noise.

Strategy B: The "Parametric BDT" (The Smart AI)
Instead of a bouncer with a checklist, imagine a super-smart AI detective (a Boosted Decision Tree, or BDT).

• This AI doesn't just look at one thing; it looks at everything at once: the angle of the particles, their speed, their mass, how they are spaced out, and even how the event "shapes" itself.
• It learns from millions of simulated examples to spot the subtle differences between the "whisper" (signal) and the "scream" (noise).
• Result: This method is much more sensitive. It can find the signal even when the bouncer would have missed it.

5. The Findings: What Did They Discover?

The team ran simulations for the future HL-LHC (which will have 3,000 times more data than current runs) to see what limits they could set on those "knobs" (couplings).

• The "Self-Coupling" Knob (δκλ): This knob controls how Higgs bosons interact with each other. The team found that with the t¯thh process, they could only constrain this knob to a range of roughly -16.5 to +12.9.

  • The Catch: Current experiments looking at other types of Higgs collisions have already set a much tighter rule (roughly -2.8 to +5.9). So, for this specific knob, the t¯thh process isn't the best detective yet.
  • The Twist: However, this knob is connected to the others. Even if we can't pin it down tightly on its own, knowing how it might move helps us understand the other knobs better. It's like knowing how a steering wheel moves helps you understand how the tires turn, even if you can't see the tires directly.

• The "New Physics" Knobs (c2, c2g, ctg): These knobs represent interactions that do not exist in the current Standard Model.

  • This is the paper's big win. There are currently no experimental limits on these specific knobs.
  • This paper provides the first-ever projections of how well the HL-LHC can measure them using the t¯thh process. They found that the t¯thh channel is very sensitive to these new interactions.

6. The Conclusion: Why This Matters

The paper concludes that while the t¯thh process is incredibly difficult to see (it's a rare, messy event), it is a powerful tool for the future.

• Multivariate Analysis Wins: The "Smart AI" (Parametric BDT) method is significantly better than the "Strict Bouncer" (Cut-based) method. It extracts much more information from the same amount of data.
• Combining Channels: Looking at both the "single-lepton" and "dilepton" (different decay patterns of the particles) together gives the best results.
• The Future: Even though we can't beat current limits on the Higgs self-coupling with this specific method yet, this process is the only way to probe certain new types of interactions (the c2, c2g, ctg knobs) that we haven't been able to measure before.

In a nutshell: This paper is a blueprint for how to use the future, super-powerful LHC to hunt for "ghosts" in the machine. It shows that by using advanced AI techniques to analyze a very rare, chaotic collision, we can finally start measuring parts of the universe's rulebook that have been completely invisible until now.

Technical Summary

Technical Summary: Exploring Higgs EFT in $t\bar{t}hh$ at High Luminosity LHC

Problem Statement
The non-resonant production of a Higgs boson pair in association with a top–antitop quark pair ($pp \to t\bar{t}hh$) is a rare process within the Standard Model (SM) with a next-to-leading order (NLO) cross section of approximately 0.981 fb at 14 TeV. While difficult to observe due to small cross sections and large backgrounds, this channel offers a unique probe of the top–Higgs sector. It provides sensitivity to the Higgs self-coupling and potential higher-dimensional interactions beyond the SM (BSM) that modify couplings between top quarks, Higgs bosons, and gluons. The High-Luminosity LHC (HL-LHC), with an integrated luminosity of 3 ab$^{-1}$, presents an opportunity to study this process, particularly within the framework of Higgs Effective Field Theory (HEFT), which allows for a model-independent description of electroweak symmetry breaking where the Higgs is treated as a singlet.

Methodology
The study employs a comprehensive analysis strategy combining traditional cut-based methods with advanced multivariate techniques to extract constraints on HEFT couplings.

• Theoretical Framework: The analysis is conducted within the HEFT framework, utilizing a specific subset of operators affecting the $t\bar{t}hh$ process. The Lagrangian includes modifications to the trilinear Higgs coupling ($\delta\kappa_\lambda$), the top Yukawa coupling ($\delta\kappa_t$), and new effective couplings: the double top-Higgs Yukawa ($c_2$), the Higgs-gluon coupling ($c_g$), the double Higgs-gluon coupling ($c_{2g}$), and the top-gluon dipole coupling ($c_{tg}$). Based on existing global constraints, $\delta\kappa_t$ and $c_g$ are fixed to their SM values (zero deviation), focusing the analysis on $\delta\kappa_\lambda$, $c_2$, $c_{2g}$, and $c_{tg}$.
• Simulation: Signal and background events are generated using MadGraph5_aMC@NLO at leading order (LO) in QCD, with heavy particle decays handled by MadSpin to preserve spin correlations. Parton showering and hadronization are performed with PYTHIA8, and a fast detector simulation is executed using Delphes with a modified CMS card to reflect HL-LHC conditions.
• Channels and Selection: The analysis targets two final states:
  • Single-Lepton (SL): One top decays leptonically, the other hadronically.
  • Dilepton (DL): Both tops decay leptonically.
    Basic selection requires at least four jets (three b-tagged), specific lepton kinematics, and missing transverse energy (MET).
• Analysis Strategies:
  1. Cut-Based Approach: Optimized hard cuts are applied to kinematic variables (e.g., $H_T$, jet multiplicity, reconstructed Higgs $p_T$, and $\chi^2$ minimization for boson reconstruction) in the 6J5B (six jets, five b-tags) category. The $H_T$ distribution is binned to construct a $\chi^2$ function for limit setting.
  2. Parametric BDT (PM:BDT): A multivariate approach using XGBoost is employed. The classifier is trained on nine physics classes (signal and various backgrounds) using a high-dimensional feature set (Table 5) including object multiplicities, kinematic variables, angular separations, and reconstructed boson observables. Crucially, the HEFT coupling values are provided as additional input features to the BDT, allowing the model to interpolate smoothly across the parameter space and extract continuous constraints from discrete benchmark points.

Key Contributions

• First Sensitivity Projections: The paper provides the first projected sensitivity limits for the HEFT couplings $c_2$, $c_{2g}$, and $c_{tg}$ specifically in the $t\bar{t}hh$ channel at the HL-LHC, as no direct experimental limits currently exist for these parameters in this process.
• Multivariate Optimization: The study demonstrates the superior performance of the PM:BDT method over the traditional cut-based approach. By exploiting multidimensional correlations among kinematic variables, the PM:BDT significantly tightens the confidence intervals for all couplings.
• Channel Combination: The work quantifies the benefit of combining the single-lepton and dilepton channels within the PM:BDT framework, showing a consistent improvement of 10–20% in sensitivity compared to individual channels.
• Two-Parameter Analysis: The study maps the allowed parameter space for simultaneous variations of two couplings, revealing complex correlations and degeneracy lifting, particularly between $\delta\kappa_\lambda$ and the other couplings.

Results
Using an integrated luminosity of 3 ab$^{-1}$ and assuming no systematic uncertainties, the study derives 95% confidence level (CL) limits:

• $\delta\kappa_\lambda$: The combined PM:BDT limit is $[-16.5, +12.9]$. This is weaker than current experimental constraints from dedicated Higgs-pair measurements (e.g., ATLAS $b\bar{b}\gamma\gamma$ channel), which is attributed to the small $t\bar{t}hh$ cross section and high background multiplicity. However, the inclusion of $\delta\kappa_\lambda$ is critical for shaping the multidimensional allowed space.
• $c_2$: Limits are $[-1.47, +1.68]$.
• $c_{2g}$: Limits are $[-25.6, +23.7]$.
• $c_{tg}$: Limits are $[-0.019, +0.018]$.

The two-parameter analysis reveals that deviations in $\delta\kappa_\lambda$ significantly shrink the allowed regions for $c_2$ and $c_{2g}$, indicating a strong interplay between the Higgs self-coupling and operators affecting top-Higgs and gluonic interactions. In contrast, $c_{tg}$ shows weaker correlations with other parameters, likely due to its distinct contribution to $t\bar{t}h$ production.

Significance
The paper claims that the $t\bar{t}hh$ production process holds strong potential for probing extended Higgs and top-quark interactions beyond the Standard Model at the HL-LHC. While the direct constraints on the Higgs self-coupling from this channel are currently less stringent than dedicated Higgs-pair searches, the study emphasizes that the channel is indispensable for constraining the multidimensional HEFT parameter space. The inclusion of $\delta\kappa_\lambda$ in the analysis is shown to be essential for correctly interpreting limits on other couplings due to non-trivial correlations. Furthermore, the results establish the first sensitivity benchmarks for the previously unexplored couplings $c_2$, $c_{2g}$, and $c_{tg}$ in this specific topology, demonstrating that multivariate parametric learning is a powerful tool for extracting BSM physics signals from complex, high-multiplicity final states.