Constraints on compensated isocurvature perturbations from BOSS DR12 galaxy data

Abstract

Abstract We use the BOSS DR12 galaxy power spectrum to constrain compensated isocurvature perturbations (CIP), which are opposite-sign primordial baryon and dark matter perturbations that leave the total matter density unchanged. Long-wavelength CIP σ(x) enter the galaxy density contrast as δg (x) ⊃ bσσ(x), with bσ the linear CIP galaxy bias parameter. We parameterize the CIP spectra as Pσσ = A 2 Pℛℛ and Pσℛ = ξ√PσσPℛℛ , where A is the CIP amplitude and ξ is the correlation with the curvature perturbations ℛ. We find a significance of detection of Ab σ ≠ 0 of 1.8σ for correlated CIP (ξ = 1), consistent with no detection. For uncorrelated CIP (ξ = 0), the constraints are instead more significantly shifted away from zero, although this may be due to large-scale data systematics which have a bigger impact on these type of CIP. The constraints on A depend on the assumed priors for the bσ parameter, which we estimate using separate universe simulations. Assuming bσ values representative of all halos we find σA = 145 for correlated CIP and σ |A| = 475 for uncorrelated CIP. Our strongest uncorrelated CIP constraint is for bσ representative of the 33% most concentrated halos, σ |A| = 197, which is better than the current CMB bounds |A| ≲ 360. We also discuss the impact of the local primordial non-Gaussianity parameter f NL in CIP constraints. Our results demonstrate the power of galaxy data to place tight constraints on CIP, and motivate works to understand better the impact of data systematics, as well as to determine theory priors for bσ .