Unveiling the cosmological information beyond linear scales: forecasts for sufficient statistics

TL;DR

This paper introduces the A* non-linear transform to extract more cosmological information from galaxy survey data beyond linear scales, significantly improving parameter constraints and validated with real survey data.

Contribution

It presents a novel A* transform optimized for non-linear regimes, enhancing cosmological parameter estimation from galaxy surveys beyond traditional power spectrum analysis.

Findings

A* transform can double the information content at low redshifts.

Forecasts show significant improvements in parameter constraints for upcoming surveys.

Validation with CFHTLS data confirms the robustness of the method.

Abstract

Beyond the linear regime, Fourier modes of cosmological random fields become correlated, and the power spectrum of density fluctuations contains only a fraction of the available cosmological information. To unveil this formerly hidden information, the A* non-linear transform was introduced; it is optimized both for the nonlinearities induced by gravity and observational noise. Quantifying the resulting increase of our knowledge of cosmological parameters, we forecast the constraints from the angular power spectrum and that of A* from l ~ 200 to 3000 for upcoming galaxy surveys such as: the Wide-Field Infrared Survey Telescope (WFIRST), the Large Synoptic Survey Telescope (LSST), Euclid, the Hyper Suprime-Cam (HSC) and the Dark Energy Survey (DES). We find that at low redshifts this new data analysis strategy can double the extracted information, effectively doubling the survey area. To test the accuracy of our forecasting and the power of our data analysis methods, we apply the A* transformation to the latest release of the Canada-France-Hawaii-Telescope Legacy Survey (CFHTLS) Wide. While this data set is too sparse to allow for more than modest gains (~1.1-1.2), the realized gain from our method is in excellent agreement with our forecast, thus verifying the robustness of our analysis and prediction pipelines.