The new study, published in The European Physical Journal C by Shahnawaz A. Adil, Sonej Alam, Somasri Sen, and J. Alberto Vazquez, explores what happens when dark energy is treated as an imperfect fluid with bulk viscosity, a form of internal friction that arises whenever a medium expands and departs from perfect thermodynamic equilibrium. In a homogeneous and isotropic universe, shear viscosity vanishes by symmetry, leaving bulk viscosity as the only relevant dissipative effect. Crucially, the viscous pressure generated in an expanding cosmos is negative, which means it can act as an effective source of acceleration without invoking a cosmological constant, without adding new scalar fields, and without modifying Einstein’s theory of gravity.

The physics here is rooted in relativistic thermodynamics going back to Carl Eckart’s first-order theory of irreversible processes from 1940. In the researchers’ framework, the total pressure of the dark energy fluid has two parts: a thermodynamic pressure proportional to the energy density through an equation-of-state parameter w, and a dissipative bulk viscous pressure proportional to the viscosity coefficient times the expansion rate of the Universe. The viscosity coefficient itself is modeled as a power of the fluid’s energy density, characterized by an amplitude and a scaling index, giving the model enough flexibility to mimic a genuinely dynamical form of dark energy. For one special choice of the scaling index, the evolution equations even admit exact analytical solutions that map directly onto well-studied constant-equation-of-state models, providing a useful consistency check on the numerics.

The authors considered two scenarios. In the minimal case, dark matter and viscous dark energy evolve independently, each obeying its own conservation law. In the non-minimal case, an interaction term allows energy to flow between the two dark components, with a coupling parameter determining the direction and strength of the exchange. Such dark-sector interactions have long been discussed as possible remedies for cosmological tensions, and the DESI collaboration’s recent baryon acoustic oscillation measurements, which hint mildly at deviations from a pure cosmological constant, have only sharpened interest in models of this kind.

To confront these ideas with reality, the team performed a full Bayesian analysis using the latest and most powerful cosmological datasets available: Type Ia supernova distances from both the Union3 compilation of 2,087 supernovae and the Pantheon+ and SH0ES sample of 1,701 distance measurements, baryon acoustic oscillation data from the second data release of the Dark Energy Spectroscopic Instrument, and a compressed likelihood built from the Planck 2018 cosmic microwave background measurements. The DESI DR2 dataset alone spans tracers from bright galaxies at low redshift out to the Lyman-alpha forest beyond redshift two, making it an exceptionally sharp probe of the expansion history. Parameter estimation was carried out with dynamic nested sampling, and model comparison relied on the Akaike and Bayesian information criteria, which penalize models for their extra complexity.

The results are nuanced. In the minimal scenario, the viscous model actually achieves a lower minimum chi-squared than Lambda-CDM for both supernova datasets, but once the penalty for additional parameters is applied, the evidence evaporates. The Bayesian information criterion, which punishes complexity most harshly, clearly favors the simpler Lambda-CDM. The inferred viscosity amplitude in this case is tiny, corresponding to a present-day physical viscosity for the dark energy fluid of order one thousand to a few thousand pascal-seconds, a value constrained so tightly that its posterior presses against the physical boundary of zero viscosity.

The non-minimal, interacting scenario tells a more interesting story. Allowing dark matter and viscous dark energy to exchange energy enlarges the allowed parameter space dramatically, and the equation-of-state parameter drifts far from the cosmological constant value of minus one, landing near minus 0.5 in the Union3 analysis. The viscosity amplitude jumps by several orders of magnitude, implying a present-day dark energy viscosity of roughly twenty million pascal-seconds, about four orders of magnitude larger than in the minimal case. The interaction parameter itself is mildly favored to be nonzero at about the one-sigma level. For the Union3 combination, the Akaike criterion gives a marginal preference of 1.41 points over Lambda-CDM, which the authors describe as only a weak improvement below the conventional threshold for a meaningful detection. The Bayesian information criterion, however, still prefers Lambda-CDM, and for the Pantheon+ combination neither criterion supports the extended model.

One of the most visually compelling results is the reconstruction of the dark energy density and effective pressure as functions of redshift. In both scenarios, the reconstructed viscous dark energy density rises with redshift and departs clearly from the constant behavior expected of a cosmological constant, while the combination of density plus pressure deviates noticeably from the vacuum-energy limit. Importantly, the analysis confirms that the viscous component remains negligible all the way back to recombination, so the model leaves the well-tested early Universe essentially untouched, modifying only the late-time expansion history. The team also ran a control analysis with dark matter forced to be strictly pressureless, showing that the improved fit of the interacting model is not merely an artifact of degeneracies between the dark matter equation of state and the viscous parameters.

The authors are careful about the limits of their conclusions. The constraints on the equation-of-state parameter and especially the viscosity scaling index turn out to be sensitive to the prior ranges assumed, meaning they should be read as conditional on physically motivated choices rather than as prior-independent measurements. Moreover, the entire analysis operates at the level of the homogeneous background expansion, using a first-order Eckart description of dissipation that lacks a finite relaxation time and is known to suffer from acausal and unstable behavior when perturbations are included. A proper test of the framework will require reformulating it within the causal Israel-Stewart theory and studying how viscosity affects the growth of cosmic structure and the full cosmic microwave background anisotropy spectrum.

Even with those caveats, the study establishes something important: bulk-viscous dark energy is observationally viable against the sharpest background data currently available, and the interacting version produces a marginal, tantalizing preference over the standard model by some measures of evidence. Whether that preference hardens into a genuine detection will depend on the next generation of surveys. The Euclid satellite, future DESI data releases, and the Vera C. Rubin Observatory will measure the expansion history and the growth of structure with unprecedented precision, and any viscous friction in the dark sector should leave fingerprints in both. The deeper challenge is theoretical: explaining microphysically where such an effective viscosity comes from, whether through non-equilibrium thermodynamics, particle production, or dark-sector interactions. If the Universe’s acceleration really is the consequence of a sticky cosmic fluid, the answer to one of physics’ greatest mysteries may turn out to be not a new particle or a modified law of gravity, but something as ordinary as friction, written on cosmological scales.

Subject of Research: Bulk-viscous dark energy models constrained by DESI DR2, supernova, and CMB observations as an alternative to the cosmological constant

Article Title: Dissipative cosmology and the nature of dark energy: insights from bulk viscosity with DESI DR2 observations

Article References: Adil, S. A., Alam, S., Sen, S., & Vazquez, J. A. (2026). Dissipative cosmology and the nature of dark energy: insights from bulk viscosity with DESI DR2 observations. The European Physical Journal C, 86(10), Article 1157. https://doi.org/10.1140/epjc/s10052-026-16412-6

Image Credits: AI Generated

DOI: 10.1140/epjc/s10052-026-16412-6

Keywords: dark energy, bulk viscosity, DESI DR2, Lambda-CDM, cosmology, baryon acoustic oscillations, Type Ia supernovae, Planck 2018, Hubble tension, Eckart theory, interacting dark sector, Bayesian inference