Dark Matter - Dark Energy coupling biasing parameter estimates from CMB data

TL;DR

This work investigates how misparametrizing the dark sector—via assuming a constant $w$ or neglecting DM--DE coupling—can bias cosmological parameter estimates from CMB data. Using MCMC on artificial CMB datasets generated with a SUGRA-based dynamical dark energy model and DM--DE coupling, the authors show that coupling can cause significant biases in $\omega_{o,c}$, $\Omega_{o,m}$, and $H_0$, even when CMB spectra appear nearly degenerate with constant-$w$ fits. They find that dynamical DE and coupling introduce degeneracies that limit CMB-only constraints, and that observable growth history and BAO signals, particularly through weak-lensing tomography, are needed to break these degeneracies. The results warn against overreliance on $\Lambda$CDM–based error bars and emphasize incorporating DM--DE coupling flexibility or richer $w(a)$ descriptions in parameter analyses.

Abstract

When CMB data are used to derive cosmological parameters, their very choice does matter: some parameter values can be biased if the parameter space does not cover the "true" model. This is a problem, because of the difficulty to parametrize Dark Energy (DE) physics. We test this risk through numerical experiments. We create artificial data for dynamical or coupled DE models and then use MCMC techniques to recover model parameters, by assuming a constant DE state parameter w and no DM--DE coupling. For the DE potential considered, no serious bias arises when coupling is absent. On the contrary, ω_{o,c}, and thence H_o and Ω_{o,m}, suffer a serious bias when the "true" cosmology includes even just a mild DM--DE coupling. Until the dark components keep an unknown nature, therefore, it can be important to allow for a degree of freedom accounting for DM--DE coupling, even more than increasing the number of parameters accounting for the w(a) behavior.

Figures (7)

• Figure 1: Marginalized a posteriori likelihood distributions (solid lines), when artificial data derive from a $\Lambda$CDM cosmology. Vertical lines yield the input parameter values. Derived parameter panels are marked by an asterisk. Dotted lines show the average likelihood distributions.
• Figure 2: Marginalized likelihood distributions on parameter values. Data built with an uncoupled SUGRA model. The parameter space for the fit includes either $\log(\Lambda/{\rm GeV})$ (solid lines) or a constant $w$ (dashed lines). Vertical dotted lines show the input parameter values. Derived parameters plots are with asterisk.
• Figure 3: Marginalized likelihood distributions on parameter values. Data built with coupled SUGRA models with $\beta = 0.05$, $\beta = 0.1$ (upper, lower panels). The parameter space for fits includes $\log(\Lambda/{\rm GeV})$ and $\beta$ (solid lines) or just a constant $w$ (dashed lines). Vertical dotted lines show the input parameter values. Pay attention to the different abscissa units in the two panel sets.
• Figure 4: Anisotropy spectra comparison. The 3 values of $\beta$ considered are clearly distinguished thanks to their different input normalization and to the line type indicated inside the upper frame. The dotted lines (hardly visible at low $l$'s in the upper panel) are the spectra of the corresponding best--fit model when assuming no coupling and $w = const.\,$. In the lower panel the relative differences between input and best--fit models are compared with WMAP5 error size.
• Figure 5: The comparison made in the upper panel of the previous Figure is extended to ET and E--polarization spectra. Model differences are hardly visible at very low $l$ and on the $l$ values where the ET spectrum changes sign.