New dark energy constraints from supernovae, microwave background and galaxy clustering

TL;DR

Using supernova, cosmic microwave background, and galaxy clustering data, the most accurate measurements to date of the dark energy density rho(X) as a function of cosmic time are made, constraining it in a rather model-independent way, assuming a flat universe.

Abstract

Using the spectacular new high redshift supernova observations from the HST/GOODS program and previous supernova, CMB and galaxy clustering data, we make the most accurate measurements to date of the dark energy density rho_X as a function of cosmic time, constraining it in a rather model-independent way, assuming a flat universe. We find that Einstein's vanilla scenario where rho_X(z) is constant remains consistent with these new tight constraints, and that a Big Crunch or Big Rip is more than 50 gigayears away for a broader class of models allowing such cataclysmic events. We discuss popular pitfalls and hidden priors: parametrizing the equation-of-state w_X(z) assumes positive dark energy density and no Big Crunch, and the popular parametrization w_X(z)=w_0 +w_0' z has nominally strong constraints from CMB merely because w_0' > 0 implies an unphysical exponential blow-up rho_X ~ e^{3 w_0' z}.

Figures (2)

• Figure 1: $1\sigma$ constraints on the density of matter and dark energy from SN Ia (Riess sample, flux-averaged with $\Delta z=0.05$), CMB and LSS data, all in units of the current dark energy density. From inside out, the four nested dark energy constraints are for models making increasingly strong assumptions, corresponding, respectively, to the 4-parameter spline, the 3-parameter spline, the 2-parameter $(f_\infty,w_i)$ case and the 1-parameter constant $w$ case (hatched). The Universe starts accelerating when the total density slope $d\ln\rho/d\ln(1+z)>-2$, which roughly corresponds to when dark energy begins to dominate, i.e., to where the matter and dark energy bands cross. In the distant future, the Universe recollapses if the dark energy density $\rho_X$ goes negative and ends in a "Big Rip" if it keeps growing ($d\ln\rho_X/d\ln(1+z)<0$).
• Figure 2: How constraints on $w_0$ and $w_0'$ depend on assumptions and data used. Darker shaded regions are ruled out at 95% confidence by SNe Ia alone; lighter shaded regions are ruled out when adding other information as indicated. 68% contours are dotted. Models above the dotted line end in a Big Rip. The 157 SNe Ia (Riess sample) have been flux-averaged with $\Delta z=0.05$.