Robust Preference for Dark Sector Interactions

Technical Summary: Robust Preference for Dark Sector Interactions

Problem Statement
Recent high-precision Baryon Acoustic Oscillation (BAO) data from the Dark Energy Spectroscopic Instrument (DESI) collaboration, when combined with Cosmic Microwave Background (CMB) and Type Ia supernova (SN) observations, indicate deviations from the standard $\Lambda$CDM cosmology. The prevailing interpretation of these deviations is a preference for dynamical dark energy (DE), often modeled via the Chevallier-Polarski-Linder (CPL) parametrization, with statistical significance ranging from $2.8\sigma$ to $3.8\sigma$. However, late-time observations constrain the expansion history and distance-redshift relations rather than the DE equation of state (EoS) directly. These deviations could alternatively arise from physical mechanisms modifying the evolution of dark sector energy densities, specifically interactions between Dark Matter (DM) and DE, without requiring an intrinsic dynamical DE component or a phantom-crossing EoS.

Methodology
The authors employ a unified computational framework to investigate Interacting Dark Energy (IDE) scenarios as an alternative to dynamical DE. They analyze two representative models:

1. Coupled Quintessence (CQ): A theoretically motivated model where DE is a scalar field $\phi$ interacting with DM via a Lagrangian-defined coupling. The interaction arises naturally in scalar-tensor theories and can be reformulated as modified gravity (e.g., $f(R)$). The model assumes an inverse power-law potential $U(\phi) \propto \phi^{-\alpha}$ and introduces two free parameters: the potential slope $\alpha$ and the coupling constant $\beta$.
2. Coupled Fluid (CF): A phenomenological framework where DE is treated as a fluid with a constant EoS $w$. The interaction is specified at the level of energy-momentum transfer, assuming a rate $Q = \beta H_0 \rho_{de}$. This model introduces two free parameters: the EoS $w$ and the coupling $\beta$.

To handle perturbations consistently across the parameter space, the authors utilize an extended parameterized post-Friedmann (PPF) framework. They parameterize the perturbations of energy and momentum transfer rates as linear combinations of density ($\delta$) and velocity ($\theta$) perturbations. The analysis utilizes Bayesian MCMC inference (via Cobaya) with the IDECAMB package (a modified CAMB) to compute theoretical predictions.

The datasets used include:

• CMB: Planck 2018, ACT DR6, SPT-3G, and joint CMB lensing spectra.
• BAO: DESI DR2.
• SN: PantheonPlus, DESY5, and the recalibrated DES-Dovekie dataset.

Key Results

• Preference for Interaction: Both IDE realizations (CQ and CF) show a robust preference for non-vanishing interactions. Marginalized constraints on the coupling parameter $\beta$ deviate from the non-interacting $\Lambda$CDM limit ($\beta=0$) at the $3\text{--}5\sigma$ level across all dataset combinations. This preference persists even when using the DES-Dovekie recalibration, which significantly weakens the evidence for dynamical DE in the CPL framework.
• Model Comparison:
  • Coupled Quintessence (CQ): The data favors a positive coupling ($\beta > 0$) at $>5\sigma$, implying energy transfer from DM to DE. The potential slope $\alpha$ is constrained to positive values. This interaction causes DM to dilute faster than in $\Lambda$CDM while sustaining DE density at intermediate redshifts ($0.5 \lesssim z \lesssim 2$). Crucially, the CQ model achieves a fit comparable to or better than CPL without requiring a phantom EoS ($w < -1$); the scalar field remains quintessence-like at all redshifts.
  • Coupled Fluid (CF): The data strongly favors a negative coupling ($\beta < 0$) and a phantom EoS ($w < -1$), implying energy transfer from DE to DM. This leads to suppressed DE density at high redshifts and enhanced DM density today ($\Omega_m \sim 0.6$). While this scenario yields the best statistical fit ($\Delta \chi^2$ and $\Delta$DIC) among all models, it results in strongly suppressed structure growth (low $\sigma_8$ and $S_8$).
• Goodness of Fit: Table II demonstrates that the CF model provides the largest improvement over $\Lambda$CDM, followed by CPL and then CQ. However, the CQ model still exhibits a statistically significant improvement over $\Lambda$CDM and remains highly competitive with CPL, despite being defined by a well-posed Lagrangian theory rather than a purely phenomenological parametrization.
• CMB Spectra: All models (CPL, CQ, CF) provide excellent fits to the CMB temperature power spectrum. The CF model shows mild oscillatory residuals at intermediate multipoles ($1000 \lesssim \ell \lesssim 2000$) due to reduced CMB lensing smoothing caused by its specific late-time clustering properties, but these remain within observational uncertainties.

Significance and Claims
The paper claims to identify dark-sector interactions as a viable and physically motivated alternative to dynamical DE for explaining recent cosmological tensions. The authors argue that deviations from $\Lambda$CDM need not be interpreted as evidence for an evolving DE EoS but can instead arise from interactions within the dark sector.

• Physical Motivation: The CQ scenario offers a theoretical foundation rooted in scalar-tensor theories, avoiding the need for a phantom-crossing EoS which is often required in phenomenological dynamical DE models to fit the data.
• Robustness: The preference for interaction is robust against recent supernova recalibrations that challenge the dynamical DE interpretation.
• Future Discrimination: The authors emphasize that while IDE and dynamical DE models are degenerate regarding background expansion and CMB observables, they predict distinct signatures in the evolution of cosmological perturbations. Consequently, future probes of weak lensing and galaxy clustering are highlighted as essential for discriminating between these scenarios.