⚛️ high-energy theory

Observational constraints on scalar-vector-tensor dark energy with phantom-divide crossing

Using observational data from DES, DESI, CMB, and RSD measurements, this study finds that a scalar-vector-tensor dark energy model featuring a stable crossing of the phantom divide provides a better fit and stronger Bayesian support than the standard $\Lambda$ CDM model.

Original authors: Nandan Roy, Shinji Tsujikawa, Ying-li Zhang, Zejun Zhang

Published 2026-09-28

📖 4 min read🧠 Deep dive

CC0 1.0

Original authors: Nandan Roy, Shinji Tsujikawa, Ying-li Zhang, Zejun Zhang

Original paper dedicated to the public domain under CC0 1.0 (http://creativecommons.org/publicdomain/zero/1.0/). ✨ 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

The universe is expanding, and for decades, astronomers have watched this expansion accelerate, driven by a mysterious force known as dark energy. This invisible agent makes up most of the cosmos, yet its true nature remains one of the greatest puzzles in physics. The standard model of cosmology treats dark energy as a constant, unchanging pressure that has been the same since the beginning of time. However, recent observations have hinted that this pressure might not be so static. Some data suggests that the strength of dark energy has shifted over the history of the universe, perhaps even crossing a critical threshold where its behavior changes fundamentally. This possibility challenges the simplest explanation and invites scientists to look for more complex, dynamic models that can evolve alongside the universe itself.

A team of researchers has now put a specific, complex theory to the test against the most recent and precise astronomical data available. They examined a model where dark energy is not a single, unchanging entity, but a mixture of two different fields: a scalar field, which behaves like a standard energy source, and a vector field, which acts more like a directional force. In this scenario, the vector field dominated the early universe, pushing the expansion rate into a regime where dark energy was stronger than the standard constant allows. As the universe aged, the scalar field began to take over, gently pulling the expansion rate back across that critical threshold to a value slightly weaker than the constant. The researchers wanted to know if the universe actually followed this path, or if the simpler, unchanging model was still the best fit.

To find the answer, the team combined three massive datasets that map the history of the cosmos. They used measurements of distant exploding stars to trace the brightness and distance of objects across billions of years, data on the clustering of galaxies to measure the expansion rate at different times, and a compressed summary of the cosmic microwave background, which is the afterglow of the Big Bang. They then ran sophisticated computer simulations to see how their two-field model behaved under these conditions. The results showed that the model was not only possible but actually provided a better fit to the data than the standard, unchanging model. The analysis revealed that the universe likely did cross that critical threshold, with dark energy being stronger in the past and slightly weaker today.

When the researchers added a fourth layer of data—measurements of how fast cosmic structures like galaxy clusters are growing over time—the picture became even clearer. This growth data acts as a stress test for the theory, checking if the forces driving expansion are consistent with the forces pulling matter together. Even with this stricter test, the two-field model held up. In fact, the probability that the universe followed a path where dark energy crossed this threshold increased significantly when the growth data was included. The study calculated that there is a very high likelihood, roughly 93 percent, that the universe's history includes this crossing event, compared to about 79 percent when only the expansion data was considered.

The researchers also compared their complex model directly against the standard, simpler model using statistical tools designed to weigh the value of new information against the cost of adding extra variables. While the standard model is elegant in its simplicity, the data consistently favored the more complex, evolving description. The statistical evidence indicated that the improvement in how well the model matched the observations was substantial enough to justify the additional complexity. The team found that the universe's expansion history and the growth of its structures are both consistent with a stable, dynamic dark energy that has changed its character over time.

This work does not prove that the standard model is wrong, but it demonstrates that a dynamic alternative is a very strong contender that fits the current evidence well. The study confirms that the universe is compatible with a scenario where dark energy was once more intense than it is now, having crossed a dividing line in its behavior. By showing that such a transition can happen without breaking the laws of physics or creating instabilities, the researchers have opened a new window for understanding the cosmos. The findings suggest that the force driving the universe apart is not a static backdrop, but a living, evolving component of the universe that has shifted its nature as the cosmos aged.

Technical Summary: Observational Constraints on Scalar-Vector-Tensor Dark Energy with Phantom-Divide Crossing

Problem and Motivation
The standard $\Lambda$ -cold-dark-matter ( $\Lambda$ CDM) model, while observationally successful, leaves unresolved the theoretical origin of the cosmological constant and the coincidence problem. Recent observational hints, particularly from Dark Energy Spectroscopic Instrument (DESI) data, suggest a time-dependent dark energy (DE) equation of state ( $w_{DE}$ ) that may cross the "phantom divide" ( $w_{DE} = -1$ ) from $w_{DE} < -1$ at earlier times to $w_{DE} > -1$ at low redshift. Realizing such a crossing in a theoretically consistent framework is nontrivial. Standard canonical scalar fields satisfy $w_{DE} \geq -1$ , while phantom fields introduce ghost instabilities. Previous models, such as scalar Galileon theories, often suffer from instabilities or require non-minimal couplings that induce fifth forces, which are tightly constrained by local gravity experiments.

This paper investigates a specific Scalar-Vector-Tensor (SVT) model proposed in Ref. [58]. This model utilizes a vector field to drive $w_{DE} < -1$ at intermediate epochs and a canonical scalar field with a potential to drive the crossing toward $w_{DE} > -1$ at late times. Crucially, the model avoids non-minimal couplings to curvature (avoiding fifth-force constraints) and restricts vector interactions to cubic order to ensure luminal tensor propagation. The authors aim to constrain this model against current cosmological data to determine if a stable phantom-divide crossing is consistent with observations.

Methodology
The authors perform a statistical analysis using the Metropolis Markov Chain Monte Carlo (MCMC) sampler (Cobaya) and nested sampling (PolyChord) for Bayesian evidence. The analysis is divided into two stages:

Background Analysis (No-RSD): The model is constrained using geometric probes: DES Year 5 Supernovae (SN Ia), DESI DR2 Baryon Acoustic Oscillations (BAO), and a compressed Cosmic Microwave Background (CMB) likelihood based on distance priors (shift parameters). The background evolution is solved via forward shooting from the radiation era, with the vector density reconstructed algebraically to avoid numerical instabilities.
Growth Analysis (With-RSD): The baseline is extended by including Redshift-Space Distortion (RSD) measurements (Gold-2017 compilation of $f\sigma_8$ ). This stage incorporates linear perturbation equations specific to the SVT model, including the effective gravitational coupling ( $\mu$ ) and the sound speed of the longitudinal vector mode ( $c_\psi^2$ ). Perturbative stability conditions (absence of ghosts and Laplacian instabilities) are explicitly imposed as priors in this analysis.

The authors compare the SVT model against the flat $\Lambda$ CDM model using the Akaike Information Criterion (AIC), Bayesian Information Criterion (BIC), and matched Bayesian evidence calculations under two prior-volume conventions:

Convention I: The prior volume includes the full rectangular parameter space, penalizing points that fail to yield regular physical solutions.
Convention II: The prior is conditioned on the existence of a regular radiation-era solution, treating the existence of the branch as part of the model definition rather than an Occam penalty.

Key Results

Background Constraints: The background-only analysis favors a present-day equation of state $w_{DE,0} > -1$ (posterior mean $\approx -0.92$ ) but assigns substantial weight to histories crossing the phantom divide. Approximately 79.1% of the posterior weight corresponds to histories where $w_{DE}(0) > -1$ and $w_{DE}(z) < -1$ at some $0 < z \leq 4$ . The SVT model provides a better best fit than $\Lambda$ CDM ( $\Delta\chi^2_{min} \approx -12.1$ ), overcoming the parameter penalty in AIC but not in BIC.
Impact of RSD Data: Including RSD data preserves the qualitative background evolution (phantom crossing) but tightens constraints on the clustering amplitude ( $\sigma_{8,0} \approx 0.75$ ). The inclusion of perturbative support conditions removes a secondary low-vector-fraction peak present in the background-only posterior. The probability of directional crossing increases to $\approx 93.3\%$ .
Model Comparison: The SVT model improves the fit over $\Lambda$ $Λ$ CDM in the RSD analysis ( $\Delta\chi^2_{min} \approx -11.6$ $Δ χ_{min}^{2} \approx - 11.6$ ).
- Bayesian Evidence: Under Convention I, the Bayes factor favors SVT ( $B \approx 5.2$ ). Under Convention II, the support for SVT is stronger ( $B \approx 12.6$ ).
- Information Criteria: The AIC difference ( $\Delta AIC \approx -1.6$ ) suggests SVT nearly compensates for its additional parameters, while the BIC difference ( $\Delta BIC \approx +26$ ) strongly favors the simpler $\Lambda$ CDM model.

Significance and Claims
The paper claims that current expansion and structure-growth data are consistent with a stable dynamical crossing of the phantom divide within the SVT framework. The primary significance lies in demonstrating that such a crossing can be realized without introducing fundamental phantom fields or non-minimal curvature couplings, thereby avoiding associated instabilities and fifth-force constraints.

The authors emphasize that the preference for the SVT model over $\Lambda$ CDM is sensitive to the choice of prior normalization and the inclusion of growth data. While the Bayesian evidence favors SVT under both conventions, the BIC penalizes the model's complexity. The paper concludes that the SVT mechanism is a viable theoretical target for explaining potential phantom-crossing phenomenology, but notes that a full Boltzmann CMB analysis and future high-precision growth measurements (e.g., from Euclid) are necessary to further test the perturbation sector and the specific signatures of the vector field.

More like this