An apparent dark-energy reversal could require dark matter to lose mass

What needs explaining?

Combinations of measurements from the Dark Energy Spectroscopic Instrument’s second data release, the cosmic microwave background and type Ia supernovae prefer an evolving dark-energy equation of state, according to the supplied paper. This quantity, called w, relates pressure to energy density. The preferred histories place it below −1 in the past and above −1 today.

The boundary matters because a canonical scalar field—the simple dynamical dark-energy model considered here—cannot itself pass below −1 when its energy density is positive. Its kinetic energy cannot be negative.

But the crossing could be apparent. If dark-matter particles lose mass through an interaction with the field, an analysis that assumes separate, noninteracting components can misattribute the resulting expansion history to phantom dark energy. The field itself need never cross the boundary.

Deriving a minimum interaction

Sapone and Arratia consider a model in which dark-matter particle mass depends on the scalar field. Dark-matter particle number is conserved, while baryons and radiation are not coupled to the field. They normalize the dark-matter density at recombination, the early epoch used to anchor their comparison with the microwave background.

Their first step uses energy conservation and nonnegative field kinetic energy. For a specified expansion history, these requirements set a minimum fractional mass loss relative to the dark-matter mass at recombination. That minimum depends on how quickly the inferred dark-energy density grows relative to the reference dark-matter density—not simply on its total growth.

Their second step introduces the coupling, beta, which measures how strongly particle mass changes as the field moves. A weak coupling cannot change the mass quickly enough unless the field moves sufficiently far. The expansion history also constrains the field’s speed, producing a minimum allowed coupling regardless of the chosen potential—the function governing the field’s dynamics.

For a field-dependent interaction, the bound constrains the largest coupling reached since recombination, rather than requiring the same strength throughout that period.

What the DESI-based calculation finds

The authors apply the bound to public DESI DR2 parameter chains combining galaxy-distance measurements with microwave-background data and one of three supernova samples: Pantheon+, DESY5 or Union3. Their representative backgrounds use posterior-mean parameters, although the paper calls them ‘best fits.’

In the baseline calculation, dark matter must have lost at least 4.6–6.9% of its mass by redshift approximately 0.8. The corresponding minimum coupling is 0.047–0.058, in the paper’s reduced-Planck-unit convention.

This epoch is not the apparent crossing itself. The representative histories cross w = −1 later, at redshifts around 0.37–0.45. The strongest mass-loss requirement occurs earlier. On the minimum-coupling trajectory, the field momentarily comes to rest near that strongest requirement.

To propagate observational uncertainty, the researchers remove 30% of each chain as burn-in and calculate the bound for 4,000 samples. The resulting baseline 95% lower limits on coupling magnitude are 0.025, 0.033 and 0.038 for the three supernova combinations.

They also test forward-evolving models. Of 240 models explored, 111 show apparent phantom behavior in the observed redshift range; none violates the bounds. Six reconstructed published fits also respect them, though their constant-coupling examples produce shallower crossings than the DESI representative histories.

Possible observational consequences

For constant coupling, the baseline representative values imply an additional attraction between dark-matter particles of at least about 0.4–0.7% of ordinary gravity, on scales within the scalar field’s force range. Baryons do not experience this additional interaction in the model.

Mass loss also reduces gravitational lensing per unit dark-matter clustering. The predicted minimum deficit is approximately 3.8–5.8% near redshift 0.8 for the representative histories. This is a comparison between lensing and clustering, not a universal reduction in every lensing measurement. The paper does not derive a unique change in structure growth: different field trajectories can change its size and even its sign.

Assumptions and preprint status

Potential-independent does not mean assumption-free. The numerical coupling bound depends on the expansion history before the low-redshift observations. A late-acting potential raises the representative requirement to roughly 0.06–0.08. Allowing the largest early tracking component considered compatible with Planck lowers the 95% limits to 0.013–0.026.

Most importantly, these limits assume that the apparent crossing is real and arises through this particular mass-changing interaction. They are not a detection of coupling. Noncanonical fields, momentum exchange and modified gravity fall outside the analysis. This is an arXiv preprint; the supplied evidence does not establish peer-reviewed publication.