Rejuvenating the Matter Power Spectrum III: The Cosmology Sensitivity of Gaussianized Power Spectra

TL;DR

Applying Gaussianizing transforms like the logarithm to nonlinear matter density fields significantly enhances the precision of cosmological parameter estimation from power spectra, outperforming conventional methods across multiple parameters.

Contribution

This study demonstrates that Gaussianized power spectra, especially the log-density, substantially improve cosmological parameter constraints compared to traditional power spectra.

Findings

Log-density power spectrum yields 2-3 times tighter parameter constraints.

Improvement reaches a factor of 5 for the spectral index n_s.

Combining Gaussianized and conventional spectra offers similar benefits.

Abstract

It was recently shown that applying a Gaussianizing transform, such as a logarithm, to the nonlinear matter density field extends the range of useful applicability of the power spectrum by a factor of a few smaller. Such a transform dramatically reduces nonlinearities in both the covariance and the shape of the power spectrum. Here, analyzing Coyote Universe real-space dark matter density fields, we investigate the consequences of these transforms for cosmological parameter estimation. The power spectrum of the log-density provides the tightest cosmological parameter error bars (marginalized or not), giving a factor of 2-3 improvement over the conventional power spectrum in all five parameters tested. For the tilt, n_s, the improvement reaches a factor of 5. Similar constraints are achieved if the log-density power spectrum and conventional power spectrum are analyzed together. Rank-order Gaussianization seems just as useful as a log transform to constrain n_s, but not other parameters. Dividing the overdensity by its dispersion in few-Mpc cells, while it diagonalizes the covariance matrix, does not seem to help with parameter constraints. We also provide a code that emulates these power spectra over a range of concordance cosmological models.