🔭 astrophysics

Non-parametric exploration of minimally coupled gravity with phantom crossing

This paper demonstrates that minimally coupled kinetic gravity braiding models can successfully realize phantom crossing and remain consistent with current cosmological observations, challenging the notion that such behavior necessarily requires non-minimal conformal coupling in Horndeski gravity.

Original authors: Matteo Cataneo, Kazuya Koyama

Published 2026-07-13

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Matteo Cataneo, Kazuya Koyama

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 universe as a giant, expanding balloon. For decades, scientists have been trying to figure out what's inside the balloon that's making it blow up faster and faster. The standard story, called $\Lambda$ CDM, says there's a mysterious "dark energy" pushing it, but it's a very boring, steady push.

Recently, however, some new data from a giant telescope instrument called DESI (along with other cosmic clues) suggested something wild: the dark energy might be changing its mind. It might be crossing a magical speed limit line called $w = -1$ . In physics-speak, this is called "phantom crossing." It's like a car that suddenly decides to drive faster than the speed limit without crashing.

The Big Question: How does the car break the speed limit?

For a long time, the leading theory was that the dark energy is connected to gravity in a very specific, "non-minimal" way. Think of it like a car with a special, heavy engine that is bolted directly to the road (the fabric of space-time). This "conformal coupling" was thought to be the only way to get that phantom crossing to happen without the car falling apart.

The New Discovery: A Different Engine

This paper asks: "Is that heavy engine the only way to do it?"

The authors, Matteo Cataneo and Kazuya Koyama, say: "Not necessarily!" They decided to test a different kind of engine called Kinetic Gravity Braiding (KGB).

Instead of bolting the engine to the road, imagine the engine has a complex system of gears and belts (derivative interactions) that twist and turn as it moves. These gears allow the car to speed up past the limit without needing that heavy, direct bolt to the road.

How They Tested It (The "Mochi" Machine)

To check if this "gears and belts" engine works, the authors used a super-smart computer program called mochi class. Think of this program as a physics simulator that can build millions of different universe models.

Usually, when scientists build these models, they have to guess the shape of the gears first. But this team did something clever: they let the gears be any shape they wanted, as long as the car didn't explode. They made sure the models were "stable" from the very start, meaning they didn't have any "ghosts" (invisible monsters that eat energy) or "gradient instabilities" (cars that shake apart).

They generated over 250,000 of these wild, gear-driven universe models. Then, they filtered them down to find the ones that:

Looked like our real universe (matching data from DESI, the Cosmic Microwave Background, and supernovae).
Actually managed to cross that $w = -1$ speed limit.
Didn't break any other rules of physics (like making the universe look too different from what we see in galaxy surveys).

What They Found

The result is exciting: Yes, it works.

They found a set of these "gears and belts" models (specifically six representative ones they named M1 through M6) that:

Successfully cross the phantom divide.
Do not need that heavy "conformal coupling" bolt to gravity.
Match all the current observations of the universe, including how galaxies are moving and how light bends around them.

In fact, these models predict that the "gravitational slip" (a difference between how gravity pulls on matter vs. how it bends light) is zero. This matches what we see in galaxy clusters, whereas some other theories struggle with this.

What They Ruled Out (Or At Least, Questioned)

The paper doesn't say the "heavy engine" (conformal coupling) is impossible. However, it argues that the recent data does not require it. The idea that we must have that non-minimal coupling to explain the phantom crossing is likely just a result of how previous scientists looked at the data or the specific tools they used. The "gears and belts" (minimally coupled KGB) are a perfectly valid, and perhaps simpler, alternative.

How Sure Are They?

The authors are careful not to say they have "solved" the mystery. They say their results demonstrate that viable solutions exist. They have simulated these models and shown they are consistent with current data.

They note that while these models look great now, future, more detailed statistical tests (called "full Bayesian analyses") are needed to see if they are the best fit compared to all other possibilities. They also mention a tiny catch: if the "braiding" (the gears) gets too strong (specifically if a value called $\alpha_B$ gets bigger than $10^{-2}$ ), the model might become unstable when gravitational waves are present. But all the models they found in this study stayed safely below that danger zone.

The Bottom Line

The universe might not need a heavy, bolted-down engine to explain why it's speeding up. It might just need a clever set of gears. This paper suggests that the "minimally coupled" path is a very real, very stable possibility that we should explore further, rather than assuming the "non-minimal" path is the only answer.

Technical Summary: Non-parametric exploration of minimally coupled gravity with phantom crossing

Problem Statement
Recent analyses of Dark Energy Spectroscopic Instrument (DESI) baryon acoustic oscillation (BAO) data, combined with Cosmic Microwave Background (CMB) and supernova (SNIa) observations, suggest a late-time dark energy equation of state that crosses the phantom divide ( $w = -1$ ). Within the framework of Horndeski gravity, such phantom crossing in minimally coupled single-scalar-field models typically leads to ghost instabilities. Consequently, recent literature has argued that this observational preference points toward non-minimal conformal coupling (where the scalar field couples to the metric, $\alpha_M \neq 0$ ). However, an alternative theoretical possibility exists: scalar-tensor theories with vanishing conformal coupling but non-trivial derivative interactions, specifically Kinetic Gravity Braiding (KGB) models. These models can stabilize phantom crossing via cubic derivative self-interactions. The central question addressed by this work is whether minimally coupled KGB theories can provide viable alternatives to conformally coupled scenarios while remaining consistent with current cosmological data, particularly regarding the growth of structure and galaxy-ISW cross-correlations.

Methodology
The authors employ a non-parametric approach to explore the space of viable KGB models, avoiding fixed Lagrangian ansätze or low-dimensional parameterizations. The study utilizes mochi_class, an extension of the Einstein-Boltzmann solver class, which implements a manifestly stable Effective Field Theory (EFT) basis.

Stable Basis Formulation: Instead of sampling arbitrary EFT functions that risk instability, the authors parameterize the stable basis functions: the de-mixed kinetic term ( $D_{\text{kin}}$ ) and the sound speed squared ( $c_s^2$ ). These are defined as smooth deformations of power laws in time ( $x = \ln a$ ), where the deviations are modeled using Gaussian Processes (GPs). This construction ensures by design that the models are free from ghost and gradient instabilities at the linear level.
Model Generation: Over 250,000 KGB models were generated by sampling the GP hyperparameters and power-law coefficients. The background expansion was fixed to specific dark energy equation of state parameters ( $w_0, w_a$ ), including the DESI DR2 best-fit values and "mirage" dark energy scenarios.
Selection Criteria: From the generated ensemble, models were selected based on two phenomenological criteria:
- The matter power spectrum at $z=0$ must deviate by less than 3% from $\Lambda$ CDM on sub-horizon scales.
- The late-time Integrated Sachs-Wolfe (ISW) effect must be positive (consistent with observations).
Observables: Six representative models (M1–M6) were analyzed against a suite of late-time observables:
- Background expansion (Hubble distance, transverse comoving distance).
- Linear growth of structure ( $f\sigma_8$ ).
- CMB temperature and lensing power spectra.
- Cosmic shear (weak lensing) auto-power spectra.
- Galaxy-ISW cross-power spectra.

Key Contributions

Non-parametric Exploration: The paper demonstrates a method to scan the KGB landscape without being constrained by specific Lagrangian forms, using a stable EFT basis to guarantee theoretical viability.
Viability of Minimally Coupled Models: It identifies a set of minimally coupled KGB models that successfully realize phantom crossing while satisfying stability conditions and observational constraints.
Comparison with Conformal Coupling: The work provides a direct counter-argument to the necessity of non-minimal conformal coupling for explaining phantom crossing, showing that derivative interactions alone can suffice.

Results

Background and CMB: The selected KGB models share identical or very similar expansion histories to the reference $w_0w_a$ CDM models. Consequently, they fit CMB temperature and lensing power spectra as well as non-minimally coupled Horndeski theories. The primary deviations are small shifts in acoustic peak positions (driven by background expansion) and a mild suppression of the low- $\ell$ ISW plateau.
Growth of Structure: While the effective gravitational coupling ( $\mu$ ) in KGB models can deviate by up to 10% from General Relativity at low redshift, the cumulative effect on the growth function $D(z)$ is sub-percent. The models predict enhanced structure growth relative to the fiducial cosmology due to an attractive fifth force, yet remain well within the observational uncertainties of DESI DR1 Redshift-Space Distortion (RSD) measurements ( $f\sigma_8$ ).
Weak Lensing and ISW:
- Cosmic Shear: The models predict an enhancement in the cosmic shear power spectrum (by 2–12%) due to the combined effect of enhanced structure growth and a stronger effective coupling to photons ( $\Sigma = \mu > 1$ ). These predictions lie within the 95% credibility region of joint DESI and KiDS-Legacy analyses.
- Galaxy-ISW: Crucially, all selected KGB models predict a positive galaxy-ISW cross-correlation. This is achieved because the models predict a decaying Weyl potential at late times ( $z \lesssim 1$ ), albeit with a slower decay rate than standard $\Lambda$ CDM. This result is consistent with Planck 2018 temperature maps cross-correlated with unWISE galaxy data.
Stability Constraints: The authors note that while the models satisfy linear stability, some KGB models with large braiding ( $\alpha_B \gtrsim 10^{-2}$ ) may suffer from instabilities in the presence of gravitational waves. The models studied here satisfy this constraint, but the authors acknowledge that identifying models with even smaller $\alpha_B$ that still fit data is a necessary future step.

Significance and Claims
The paper claims that viable phantom-crossing solutions exist within the minimally coupled KGB framework without requiring non-minimal conformal coupling. This finding challenges the interpretation that DESI data necessarily implies a non-zero conformal coupling ( $\alpha_M \neq 0$ ). The authors suggest that the apparent preference for conformal coupling in previous studies may stem from prior volume effects or limitations in the EFT bases used.

The significance of this work lies in demonstrating that derivative-coupled models offer a compelling, theoretically stable explanation for current observations. Furthermore, these models naturally satisfy constraints from local gravity tests (via Vainshtein screening) and gravitational wave speed constraints (luminal propagation), while predicting vanishing gravitational slip ( $\gamma=1$ ), which is consistent with current cluster mass measurements. The authors conclude that these results motivate future full Bayesian analyses of the KGB parameter space using non-parametric EFT priors to rigorously reconstruct the underlying physics.

Technical Summary: Non-parametric exploration of minimally coupled gravity with phantom crossing

More like this