Recent measurements from the Dark Energy Spectroscopic Instrument (DESI) have been interpreted as a sign of unusual, time-changing dark energy. This paper tests a different idea: the odd signal could come not from dark energy but from dark matter whose pressure changes over time. The authors build a simple model of “dynamical dark matter” and compare it to standard cosmological observations to see if it fits the data.

The model replaces the usual assumption that dark matter is pressureless with an equation of state (EoS) that smoothly changes from one constant value in the early universe to another value today. The transition happens around a scale factor a_t and can be either gradual or sharp. Physically, a non-zero EoS means dark matter carries pressure. Positive pressure at early times speeds up the early expansion, which reduces the sound horizon at recombination and can shift the angular scales seen in the cosmic microwave background (CMB). That in turn influences inferred values of other parameters such as the Hubble constant.

To test the idea the authors fit the model to a set of cosmological data: CMB measurements, the DESI DR2 baryon acoustic oscillation (BAO) results, three different compilations of Type Ia supernovae (PantheonPlus, Union3, and DESY5), and measurements of the growth rate of structure. They also include the effects of dark matter pressure on density and velocity perturbations. In their analysis the dark-matter shear was set to zero and adiabatic initial conditions were assumed. Model comparisons use chi-squared and a deviance information criterion.

The main numerical findings are that the early-time dark matter EoS is consistent with zero, so the model does not require exotic early behavior. The present-day EoS is mildly negative in some fits. Depending on the data combination the significance ranges from 0.42σ up to 3.02σ. The strongest signal comes from the combination of CMB, DESI, and DESY5, which yields a present-day EoS of w_dm,0 = −0.060 (+0.013, −0.028) and a transition scale factor a_t = 0.41 (+0.088, −0.13) at 68% confidence. The model also shifts the matter density Ω_m slightly higher than in the standard ΛCDM model and predicts lower values of σ8 and S8, bringing them closer to weak-lensing measurements. The inferred H0 (the Hubble constant) stays essentially unchanged and therefore remains in tension with local distance-ladder measurements. Statistically, the dynamical dark matter model is preferred over ΛCDM for one data combination (Δχ2_MAP = −14.093, ΔDIC = −7.838 for Planck+DESI+DESY5) but is disfavored compared with a commonly used dynamical dark energy parameterization (the CPL model).