Reduction of Cosmological Data for the Detection of Time-varying Dark Energy Density
We present a method for reducing cosmological data to constraints on the amplitudes of modes of the dark energy density as a function of redshift. The modes are chosen so that (1) one of them has constant density and (2) the others are non-zero only if there is time-variation in the dark energy density and (3) the amplitude errors for the time-varying modes are uncorrelated with each other. We apply our method to various combinations of three-year WMAP data, baryon acoustic oscillation data, the 'Gold' supernova data set, and the Supernova Legacy Survey data set. We find no significant evidence for a time-varying dark energy density or for non-zero mean curvature. Although by some measure the limits on four of the time-varying mode amplitudes are quite tight, they are consistent with the expectation that the dark energy density does not vary on timescales shorter than a Hubble time. Since we do not expect detectable time variation in these modes, our results should be viewed as a systematic error test which the data have passed. We discuss a procedure to identify modes with maximal signal-to-noise ratio.