⚛️ phenomenology

Scrutinizing the Mass Matrices in Three-Higgs-Doublet Models with Generalized CP Symmetries

This paper demonstrates that softly broken generalized CP-symmetric three-Higgs-doublet models can successfully reproduce all observed quark masses and mixings, identifying 22 phenomenologically viable, inequivalent models after eliminating redundant parameterizations and performing numerical fits to experimental data.

Original authors: Duarte D. Correia, Jorge C. Romao, Joao P. Silva

Published 2026-07-15

📖 6 min read🧠 Deep dive

CC BY 4.0

Original authors: Duarte D. Correia, Jorge C. Romao, Joao P. Silva

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine the Standard Model of physics as a giant, slightly messy Lego castle. It's a masterpiece that explains almost everything we see, but it has some missing bricks. It can't explain dark matter, why the universe is made of matter instead of antimatter, or how neutrinos get their weight. To fix this, physicists often try adding more "Higgs" pieces to the castle.

In this study, three researchers decided to try adding three Higgs doublets (let's call them three special Lego towers) instead of the usual one or two. But they didn't just throw them in randomly. They imposed a strict set of rules called Generalized CP (GCP) symmetry. Think of this symmetry like a magical mirror: if you look at the universe in the mirror, the laws of physics should look the same, even if time runs backward or left and right are swapped.

The Great Filter: Sorting the Chaos

Before this paper, a previous study had identified 40 different ways to arrange these three towers while obeying the mirror rules. It was like having 40 different blueprints for a house, and the researchers thought they were all unique.

The authors of this paper put on their detective hats and realized something surprising: not all 40 blueprints are actually different.

They found that many of these models were just the same house painted in different colors or viewed from a different angle. By using "basis transformations" (which is like rotating your head or rearranging the furniture without changing the room's structure), they showed that several models were identical twins. Some models even had "spurious parameters"—extra knobs and dials that looked important but didn't actually change the physics at all. It's like having a remote control with 50 buttons, but only 10 actually turn the TV on; the other 40 are just for show.

After cleaning up the duplicates, they were left with a smaller, more manageable list.

The Big Test: Can They Fit the Puzzle?

The real challenge was to see if any of these remaining models could explain the real world. The universe has a very specific "puzzle" to solve:

6 Quark Masses: The weights of the six types of quarks (the building blocks of protons and neutrons).
4 Mixing Angles: How these quarks switch identities when they interact (described by the Cabibbo–Kobayashi–Maskawa matrix, or CKM matrix).

In the simpler case of having only two Higgs towers (2HDM), the researchers found that the GCP-symmetric version failed. It couldn't fit the puzzle pieces together; the numbers just didn't add up. It was like trying to force a square peg into a round hole.

But here is the exciting twist: In the three-tower world (3HDM), it works!

The authors ran massive numerical simulations, tweaking the knobs on their models to see if they could match the 10 experimental values (the 6 masses and 4 mixing angles) perfectly. They used a method called $\chi^2$ minimization, which is basically a fancy way of saying "finding the best possible fit."

The Winners and the Losers

After crunching the numbers, they found 22 distinct models that successfully reproduced all the experimental data.

The "CPc" Models: These are the most complex ones, with 18 real parameters just for the quark interactions (plus 4 more from the vacuum setup). They fit the data beautifully. In fact, the authors found specific "benchmark points"—exact settings for the model—that predicted the quark masses and mixing angles with incredible precision. For example, one model predicted the top quark mass to be 172.569954 GeV, matching the experimental value of 172.57 ± 0.29 GeV almost perfectly.
The "Minimal" Models: The researchers also looked at models with fewer knobs (as few as 10 parameters). They hoped to find a "simple and elegant" solution. However, none of these minimal models could fit the data. They were too restrictive, like a puzzle with too few pieces to complete the picture.
The "14-Parameter" Models: There were some models with 14 parameters (the lowest count among the viable candidates), but the authors found that none of these could fit the data either. They were too tight.
The "16-Parameter" Models: However, some models with 16 parameters (12 for Yukawa interactions and 4 for the vacuum) did work. This is a significant finding because it shows you don't need the maximum complexity to get a good result, but you do need some flexibility.

What About the "Boundary" Models?

The authors introduced a clever concept called "boundary models." Imagine a model as a vast landscape. A "boundary model" is like a specific path or edge of that landscape. They showed that if a complex model (with many parameters) fits the data, then its "boundary" versions (where some parameters are set to zero or specific values) will also fit. This helped them quickly identify which models were viable without having to test every single one from scratch.

The Verdict

The paper concludes that Three-Higgs-Doublet Models with a softly broken GCP symmetry are a viable extension of the Standard Model.

Unlike the two-Higgs version, which failed to explain the quark world under these symmetry rules, the three-Higgs version succeeds. The authors found 22 physically distinct models that can reproduce the six quark masses and the four mixing angles we observe in nature.

They are careful to note that while these models fit the numbers, they still allow for "flavor-changing neutral couplings" (a fancy way of saying particles changing types in ways that are usually forbidden). The paper doesn't solve whether these specific interactions are safe from current experiments; that's a job for future studies. But for now, the door is open: the three-Higgs universe is a real possibility that fits the data we have today.

In short: The universe might be a bit more complex than we thought, needing three Higgs towers instead of one or two, and if it does, the rules of the "mirror symmetry" (GCP) can still hold true, provided we have enough knobs to turn to make the numbers match.

Technical Summary: Scrutinizing the Mass Matrices in Three-Higgs-Doublet Models with Generalized CP Symmetries

Problem Statement
The Standard Model (SM) is incomplete, failing to explain phenomena such as dark matter and baryogenesis. Multiple Higgs-doublet models (NHDM) offer a natural extension. While previous studies established that Generalized CP (GCP) symmetric Two-Higgs-Doublet Models (2HDM) with soft breaking struggle to accommodate all quark flavor observables (specifically, the unique non-trivial GCP 2HDM model fails to fit the 10 observables: six quark masses and four CKM parameters), the viability of GCP-symmetric Three-Higgs-Doublet Models (3HDM) remained an open question. A prior classification identified 40 unique GCP-symmetric 3HDM textures consistent with non-null quark masses, non-block-diagonal mixing, and a non-zero Jarlskog invariant. However, the physical distinctness of these 40 models and their phenomenological viability had not been fully scrutinized.

Methodology
The authors investigate the 3HDM Lagrangian, focusing on the scalar potential and the Yukawa sector under a softly broken GCP symmetry. The scalar fields transform as $\Phi_a \to X_{ab}\Phi^*_b$ , where $X \in SU(3)$ , and quark fields undergo corresponding transformations defined by angles $\theta, \alpha, \beta, \gamma$ .

The methodology proceeds in three stages:

Reduction of Model Space: The authors demonstrate that the 40 previously identified models are not all physically distinct. By constructing specific basis transformations (weak-basis transformations, WBT) and identifying spurious parameters, they relate apparently distinct models. They introduce the concept of "boundary models," where setting specific parameters to zero or equal values in a model with a continuous range of angles reduces it to a model with fixed angles.
Numerical Fitting: The authors perform $\chi^2$ minimizations using the CERN Minuit library. They fit the 10 observables (six quark masses and four independent CKM magnitudes) against the model parameters. The fit parameters include the vacuum expectation value (vev) phases and magnitudes ( $\beta_1, \beta_2, \delta_2, \delta_3$ ) and the real parameters within the Yukawa textures. The Jarlskog invariant ( $J_{CP}$ ) is derived from the CKM magnitudes to ensure CP violation is reproduced.
Classification: Models are categorized by the scalar potential symmetry classes ( $CP_a, CP_b, CP_c, CP_d$ ) and the number of free real parameters in their Yukawa matrices.

Key Contributions and Results

Identification of Redundancies: The study reveals that the 40 models from Ref. [2] contain redundancies. Specific basis transformations relate models that appear distinct, and some models with different parameter counts are physically equivalent due to spurious parameters.
Viability of GCP 3HDM: In contrast to the 2HDM case, the authors find that GCP-symmetric 3HDMs can successfully reproduce all quark masses and mixings.
Quantification of Viable Models: After removing redundancies and performing numerical fits, 22 physically distinct models are found to be compatible with current experimental data. These are listed in Table III of the paper.
Parameter Count Analysis:
- CPc Models: The models invariant under $CP_c$ (where $\theta = \pi/3$ ), which are the 3HDM analogs of the non-trivial GCP 2HDM, fit the data very well. These models typically possess 18 Yukawa parameters plus 4 vev parameters (22 total).
- Minimal Models: Models with as few as 10 real Yukawa parameters (totaling 14 parameters with vevs) were tested but found unable to fit the experimental values.
- Intermediate Models: The authors identify viable models with 16 Yukawa parameters (12 in matrices + 4 vevs) and 18 Yukawa parameters. Specifically, certain $CP_d$ models with 16 parameters are shown to provide good fits.
- Excluded Models: Models with 14 parameters (the minimum among the 40) were found to be too stringent to reproduce the data.
Benchmark Points: The paper provides specific numerical benchmark points for representative models across the viable classes (e.g., $(\pi/3, \pi/3, 0, \pi/3)$ , $(\theta, \theta, \theta, 2\theta)$ , etc.), demonstrating $\chi^2$ values consistent with experimental errors.

Significance
The paper establishes that softly broken GCP-symmetric 3HDMs are phenomenologically viable extensions of the Standard Model, capable of accommodating the precise experimental values for quark masses and mixings where their 2HDM counterparts fail. By reducing the set of 40 candidate models to 22 physically distinct, viable models, the work provides a refined framework for future studies of flavor phenomenology in these theories. The authors note that while these models fit the mass and mixing data, the detailed study of flavor-changing neutral couplings (FCNCs) and their constraints lies beyond the scope of this work and must be addressed on a model-by-model basis.

The Great Filter: Sorting the Chaos

The Big Test: Can They Fit the Puzzle?

The Winners and the Losers

What About the "Boundary" Models?

The Verdict

More like this