[Paper Review] Interaction in the dark sector: a Bayesian analysis with latest observations

Tassia Ferreira, C. Pigozzo+2 more|arXiv (Cornell University)|Dec 14, 2017

Cosmology and Gravitation Theories5 citations

TL;DR

This Bayesian analysis investigates interacting dark sector models—specifically the non-adiabatic generalized Chaplygin gas (GCG) and Λ(t)CDM—using latest cosmological data. It finds moderate to positive evidence favoring interacting models over standard ΛCDM, particularly when combining H₀, SNe Ia, BAO, and CMB acoustic scale data.

ABSTRACT

By combining cosmological probes at low, intermediate and high redshifts, we investigate the observational viability of a class of models with interaction in the dark sector. We perform a Bayesian analysis using the latest data sets of type Ia supernovae, baryon acoustic oscillations, the angular acoustic scale of the cosmic microwave background, and measurements of the expansion rate. When combined with the current measurement of the local expansion rate obtained by the Hubble Space Telescope, we find that these observations provide evidence in favour of interacting models with respect to the standard cosmology.

Motivation & Objective

• To assess the observational viability of interacting dark energy and dark matter models within the generalized Chaplygin gas (GCG) framework.
• To test whether current cosmological data favor interacting models over the standard ΛCDM cosmology.
• To evaluate the impact of including local H₀ measurements and cosmological clock data on model selection.
• To perform a robust Bayesian model comparison using Bayes factors and Jeffreys' scale to quantify evidence strength.

Proposed method

• Employs a non-adiabatic generalized Chaplygin gas (GCG) model with a dimensionless parameter α to describe energy transfer between dark energy and dark matter.
• Uses a Bayesian framework with the MultiNest algorithm to explore the full parameter space and compute Bayesian evidence.
• Combines multiple observational probes: 740 SNe Ia (JLA), 4 BAO (D_V/r_s), 14 BAO (θ_BAO), Planck 2015 CMB acoustic scale (ℓ_A), and 25 H(z) measurements from cosmic chronometers.
• Applies a model-independent Gaussian prior on the sound horizon r_s to avoid double-counting, and validates results with a flat prior for robustness.
• Computes Bayes factors using Jeffreys' scale to quantify the strength of evidence for one model over another.
• Performs model comparisons across ΛCDM, Λ(t)CDM, and GCG models under different data combinations.

Experimental results

Research questions

• RQ1 Does the inclusion of energy transfer between dark energy and dark matter improve the fit to current cosmological data compared to ΛCDM?
• RQ2 What is the strength of evidence for interacting dark sector models (GCG and Λ(t)CDM) relative to ΛCDM using the latest observational datasets?
• RQ3 How do local H₀ measurements and cosmological clock data influence the Bayesian model selection outcome?
• RQ4 Is the evidence for interacting models robust under alternative priors on the sound horizon r_s?
• RQ5 What are the posterior constraints on the interaction parameter α in the GCG model?

Key findings

• The GCG model shows moderate evidence (ln B = 3.314 ± 0.513) in favor of interacting models over ΛCDM when combining H₀, SNe Ia, BAO, and CMB acoustic scale data.
• The Λ(t)CDM model yields positive evidence (ln B = 1.627 ± 0.518) over ΛCDM under the same data combination.
• When including cosmological clock data (OHD), the GCG model shows positive evidence (ln B = 1.362 ± 0.513), while Λ(t)CDM is disfavored relative to ΛCDM (ln B = 1.247 ± 0.514).
• The 2σ credible interval for the GCG interaction parameter is α = −0.24⁺⁰.³⁵₋₀.³⁰, indicating a non-zero interaction rate.
• The inclusion of cosmological clocks shifts model preference toward ΛCDM, suggesting that this dataset may be pulling the evidence in the opposite direction.
• Robustness checks with a flat prior on r_s confirm that the evidence for interacting models remains, albeit inconclusive (|ln B| < 1), validating the results against prior double-counting concerns.