Measuring primordial non-gaussianity without cosmic variance

TL;DR

This work proposes a cross-tracer approach to measure primordial non-gaussianity $f_{nl}$ that eliminates cosmic variance by comparing the density fields of a highly biased tracer and an unbiased tracer. A two-tracer Fisher analysis shows that the relative-bias estimator can provide tighter constraints on $f_{nl}$ than the traditional power-spectrum method, especially when tracers are numerous and highly biased and well correlated on large scales. The results indicate potential error reductions up to ~7× for all-sky surveys to $z\sim 2$ and larger gains with dense high-bias tracers or photometric surveys, while highlighting caveats related to stochasticity and cross-tracer matching. If successful, this method could enable robust detection of local-type primordial non-gaussianity near $f_{nl}\sim 1$, providing a powerful discriminator among early-universe models.

Abstract

Non-gaussianity in the initial conditions of the universe is one of the most powerful mechanisms to discriminate among the competing theories of the early universe. Measurements using bispectrum of cosmic microwave background anisotropies are limited by the cosmic variance, i.e. available number of modes. Recent work has emphasized the possibility to probe non-gaussianity of local type using the scale dependence of large scale bias from highly biased tracers of large scale structure. However, this power spectrum method is also limited by cosmic variance, finite number of structures on the largest scales, and by the partial degeneracy with other cosmological parameters that can mimic the same effect. Here we propose an alternative method that solves both of these problems. It is based on the idea that on large scales halos are biased, but not stochastic, tracers of dark matter: by correlating a highly biased tracer of large scale structure against an unbiased tracer one eliminates the cosmic variance error, which can lead to a high signal to noise even from the structures comparable to the size of the survey. The square of error improvement on non-gaussianity parameter f_nl relative to the power spectrum method scales as Pn/2, where P and n is the power spectrum and the number density of the biased tracer, respectively. For an ideal survey out to z=2 the error reduction can be as large as a factor of seven, which should guarantee a detection of non-gaussianity from an all sky survey of this type. The improvements could be even larger if high density tracers that are sensitive to non-gaussianity can be identified and measured over a large volume.