Can dark matter and dark energy share an origin? A new cosmology test keeps the idea in contention

The standard cosmological model, known as ΛCDM, combines cold dark matter with a cosmological constant: dark energy whose density does not change as the universe expands. It explains a broad range of observations, but does not establish the physical identity of either dark component.

In a November 10 preprint, Jose Agustin Lozano Torres and Axel de la Macorra test a different proposal against measurements of cosmic expansion and structure. Their unified bound dark energy–bound dark matter model remains competitive with ΛCDM under their Bayesian comparison. That is not evidence that its proposed particles exist, but it identifies predictions that further observations can challenge.

One transition, two dark components

The model proposes a hidden particle sector that interacts with ordinary matter only through gravity. As the universe cools, its particles bind into composite objects, in an analogy with the way quarks form bound particles through the strong nuclear interaction.

The lightest composite state is described by a field that eventually drives accelerated expansion. Heavier states supply the dark matter. Their motion at formation determines how efficiently they escape small concentrations of matter, smoothing structures that would otherwise grow.

The underlying theory links the binding-energy scale to an assumed unification of particle-interaction strengths at very high energies. In the cosmological fit, that scale replaces a standard background parameter rather than simply disappearing from the calculation. The dark-energy history then follows from the model instead of requiring two independently adjustable parameters describing its evolution. The dark-matter sector adds one parameter: the particles’ formation velocity.

Testing expansion and the early universe together

The researchers combine DESI’s second-release baryon acoustic oscillation measurements, which use a characteristic separation in cosmic structure as a distance ruler, with observations of the cosmic microwave background from Planck, ACT and SPT-3G. They also include gravitational lensing of that ancient radiation and 2,884 supernovae from the Unite compilation.

Using modified versions of the CAMB cosmology software and statistical sampling with Cobaya, they compare four models: ΛCDM, a flexible evolving-dark-energy model, bound dark energy with ordinary cold dark matter, and the unified proposal.

Within the unified model, the inferred binding scale is 43.74 ± 0.11 electronvolts, inside the broader theoretical prediction. This is a parameter estimate conditional on the model—not a laboratory measurement of a new interaction.

The predicted dark energy changes at late times. Its equation-of-state parameter, which relates pressure to energy density, reaches about −0.927 today, rather than the cosmological constant’s −1. It approaches −1 earlier without crossing below it. The very small quoted uncertainty reflects the restrictions imposed by this particular theory, not an equally precise model-independent measurement of dark energy.

A better distance fit is not an overall victory

At the joint best fits, the unified model gives the lowest mismatch statistic for DESI’s distance measurements among the four alternatives: 10.97, compared with 13.75 for ΛCDM. The paper describes this as a 20% reduction. It is a reduction in that statistic, not a 20% increase in the probability that the theory is correct.

Across the complete dataset, its best-fit improvement over ΛCDM is only 0.60 in the same statistic. Bayesian evidence, which accounts for parameter freedom and the adopted priors, does not meaningfully distinguish the two. The flexible evolving-dark-energy model achieves a substantially better best fit, but its additional freedom also prevents a decisive Bayesian preference.

A narrower comparison favors bound dark matter over cold dark matter when both accompany bound dark energy. The authors report moderate Bayesian evidence, driven mainly by the microwave-background measurements. That result should not be read as a general detection of non-cold dark matter.

The strongest test lies in smaller structures

The inferred formation speed is 0.37 ± 0.14 times the speed of light. These initially moving particles suppress small-scale clustering, giving a present-day clustering measure, S8, of 0.807 ± 0.007, versus about 0.822 in the ΛCDM fit.

A dedicated analysis of the Lyman-alpha forest—absorption features in distant quasar light that trace intervening gas—could test that suppression. The paper uses DESI’s Lyman-alpha distance-ruler measurements, but does not fit the forest’s small-scale power. Existing reconstructed power-spectrum points assume a different model and cannot settle the issue. The predicted cutoff also lies where gravitational evolution is nonlinear, complicating the comparison.

The proposal leaves another major problem unresolved. Without the supernova sample, the inferred expansion rate remains 5.5 standard deviations below the local SH0ES measurement. This preprint therefore offers a testable connection between two unknown cosmic ingredients, not an established replacement for standard cosmology.