Abstract
Models in which the Higgs boson is a composite pseudo-Nambu-Goldstone boson offer attractive solutions to the Higgs mass naturalness problem. We consider three such models based on the minimal $SO(5) \to SO(4)$ symmetry breaking pattern and perform convergent global fits under a Bayesian framework to identify parameter regions that best satisfy experimental constraints, including recent Higgs measurements. We introduce a novel technique to analyze fine-tuning, quantifying it as the Kullback-Leibler (KL) divergence from the prior to the posterior probability. Each model is found to satisfy all constraints at the $3\sigma$ level. Furthermore, we analyze collider phenomenology, finding that the $gg \to H \to \gamma\gamma$ cross-section in two models is significantly lower than Standard Model predictions, offering a testable signature for the High-Luminosity LHC.
Introduction: The Naturalness Problem
In the Standard Model (SM), the Higgs boson is treated as an elementary scalar particle. This leads to the "naturalness problem": quantum corrections should theoretically drive the Higgs mass toward the Planck scale ($10^{19}$ GeV). Maintaining the observed mass of 125 GeV requires extreme, fine-tuned cancellations.
Composite Higgs Models (CHMs) propose that the Higgs is not elementary but a bound state of a new strong force, similar to how pions are bound states of quarks. While theoretically elegant, these models possess high-dimensional, non-linear parameter spaces that are notoriously difficult to explore using traditional scanning methods.
Methodology: Bayesian Global Fits and KL Divergence
To address these challenges, this study employs a rigorous Bayesian global fit framework. Rather than searching for isolated best-fit points, the researchers mapped the entire probability distribution of the models' parameters relative to experimental data.
Nested Sampling
The study utilized the PolyChord algorithm, a nested sampling technique designed to navigate "sheety" or multi-modal distributions in high-dimensional spaces (15–19 dimensions). This allowed for the identification of viable regions that previous "random walk" scans often missed.
Quantifying Fine-Tuning
A significant contribution of this work is the use of Kullback-Leibler (KL) divergence to measure fine-tuning. Traditionally a subjective concept, fine-tuning is here quantified as the information gain required to move from the initial model assumptions (prior) to the data-constrained reality (posterior). A higher KL divergence indicates a more fine-tuned, and thus less "natural," model.
Model Structure and Variants
The research focused on three variants of the Minimal 4D Composite Higgs Model (M4DCHM). These variants differ in how the top and bottom quarks are embedded into the $SO(5)$ symmetry group (specifically the 5, 10, and 14 representations).
The models were constrained using a comprehensive dataset, including SM masses, electroweak precision observables (S and T parameters), Z-boson decay widths, and Higgs signal strengths from the LHC.
Key Results and Collider Phenomenology
Model Viability and Naturalness
All three models were found to be viable at the $3\sigma$ level. The model using the "5" representation ($M4DCHM^{5-5-5}$) emerged as the most natural, showing the highest Bayesian evidence and requiring the least information gain (KL divergence) to match experimental data.
Mass Bounds for New Particles
The fits predict the existence of heavy fermion "partners" arising from the new strong dynamics. The results indicate these particles must generally lie above 1.1 TeV. This explains their current absence in LHC data while providing a clear target for future searches.
The Diphoton Signature
A critical finding for future experimental verification concerns the $gg \to H \to \gamma\gamma$ (diphoton) cross-section. In two of the three models, this cross-section is predicted to be approximately 10% lower than the Standard Model prediction. This discrepancy is already in slight tension with current data and could be definitively tested during the High-Luminosity run of the LHC (HL-LHC).
Conclusion
By applying advanced statistical techniques to 4D Composite Higgs Models, this study demonstrates that these theories remain robust candidates for physics beyond the Standard Model. The introduction of KL divergence provides a rigorous mathematical footing for naturalness, while the predicted diphoton suppression and TeV-scale fermion partners offer clear pathways for discovery or exclusion in the next decade of collider physics.
Summary updated