Exploiting scale dependence in cosmological averaging

Abstract

We study the role of scale dependence in the Buchert averaging method, using the flat Lemaitre-Tolman-Bondi model as a testing ground. Within this model, a single averaging scale gives too coarse predictions, but by replacing it with the distance of the objects R(z) for each redshift z, we find an O(1%) precision at z<2 in the averaged luminosity and angular diameter distances compared to their exact expressions. At low redshifts, we show the improvement for generic inhomogeneity profiles, and our numerical computations further verify it up to redshifts z~2. At higher redshifts, the method breaks down due to its inability to capture the time evolution of the inhomogeneities. We also demonstrate that the running smoothing scale R(z) can mimic acceleration, suggesting it could be at least as important as the backreaction in explaining dark energy as an inhomogeneity induced illusion.