Precision Prediction of the Log Power Spectrum

TL;DR

This paper introduces a new, accurate, cosmology-independent method to predict the log power spectrum, which captures more small-scale cosmological information than the standard power spectrum, especially at high wavenumbers.

Contribution

The authors develop a novel, simulation-independent prediction method for the log power spectrum that matches the accuracy of existing models across various conditions.

Findings

The method accurately predicts the log power spectrum up to $1.5h$ Mpc$^{-1}$.

It captures more cosmological information than the standard power spectrum.

The prediction is robust across different redshifts and smoothing scales.

Abstract

At translinear scales, the log power spectrum captures significantly more cosmological information than the standard power spectrum. At high wavenumbers $k$, the Fisher information in the standard power spectrum $P(k)$ fails to increase in proportion to $k$ in part due to correlations between large- and small-scale modes. As a result, $P(k)$ suffers from an information plateau on these translinear scales, so that analysis with the standard power spectrum cannot access the information contained in these small-scale modes. The log power spectrum $P_A(k)$, on the other hand, captures the majority of this otherwise lost information. Until now there has been no means of predicting the amplitude of the log power spectrum apart from cataloging the results of simulations. We here present a cosmology-independent prescription for the log power spectrum; this prescription displays accuracy comparable to that of Smith et al. (2003), over a range of redshifts and smoothing scales, and for wavenumbers up to $1.5h$ Mpc$^{-1}$.