PT challenge: Validation of ShapeFit on large-volume, high-resolution mocks

TL;DR

This paper evaluates the ShapeFit method's effectiveness on large-volume, high-resolution mock data, demonstrating its potential for precise cosmological parameter estimation in future galaxy surveys.

Contribution

It validates ShapeFit on large mock datasets, analyzing redshift evolution and setup variations to assess robustness and accuracy in cosmological inference.

Findings

ShapeFit achieves sub-2σ deviations on large mock volumes.

Redshift evolution mapping improves parameter constraints.

Results are consistent across different analysis setups.

Abstract

The ShapeFit compression method has been shown to be a powerful tool to gain cosmological information from galaxy power spectra in an effective, model-independent way. Here we present its performance on the blind PT challenge mock products presented in [1]. Choosing a set-up similar to that of other participants to the blind challenge we obtained $\Delta \ln\left(10^{10} A_s\right) = -0.018 \pm 0.014$, $\Delta \Omega_\mathrm{m} = 0.0039 \pm 0.0021$ and $\Delta h =-0.0009 \pm 0.0034$, remaining below $2\sigma$ deviations for a volume of $566 \left[ h^{-1}\mathrm{Gpc}\right]^3$. This corresponds to a volume 10 times larger than the volume probed by future galaxy surveys. We also present an analysis of these mocks oriented towards an actual data analysis using the full redshift evolution, using all three redshift bins $z_1 = 0.38$, $z_2=0.51$, and $z_3 = 0.61$, and exploring different set-ups to quantify the impact of choices or assumptions on noise, bias, scale range, etc. We find consistency across reasonable changes in set-up and across redshifts and that, as expected, mapping the redshift evolution of clustering helps constraining cosmological parameters within a given model.