The N5K Challenge: Non-Limber Integration for LSST Cosmology

• DOI: 10.21105/astro.2212.04291
• Publication Date: 2023-02-18T16:58:40Z
• Authors: C. Danielle Leonard; Tassia Ferreira; Xiao Fang; Robert Reischke; Nils Schoeneberg; Tilman Tröster; David Alonso; Jean-Eric Campagne; François Lanusse; Anže Slosar; Mustapha Ishak
• Keywords: Cosmology and Nongalactic Astrophysics (astro-ph.CO); Limber Approximation; LSST; Astronomy; FOS: Physical sciences; QB1-991; Angular Power Spectrum; Cosmology; Angular Power Spectrum; Limber Approximation; LSST; Statistical Methods; Astrophysics; Cosmology; QB460-466; 518; Statistical Methods; [PHYS.ASTR] Physics [physics]/Astrophysics [astro-ph]; [PHYS.ASTR]Physics [physics]/Astrophysics [astro-ph]; Astrophysics - Cosmology and Nongalactic Astrophysics

The rapidly increasing statistical power of cosmological imaging surveys requires us to reassess the regime validity for various approximations that accelerate calculation relevant theoretical predictions. In this paper, we present results 'N5K non-Limber integration challenge', goal which was quantify performance different approaches calculating angular spectrum galaxy number counts and cosmic shear data without invoking so-called 'Limber approximation', in context Rubin Observatory Legacy Survey Space Time (LSST). We performance, terms accuracy speed, three implementations: ${\tt FKEM (CosmoLike)}$, Levin}$, matter}$, themselves based on schemes approximations. find challenge's fiducial 3x2pt LSST Year 10 scenario, (CosmoLike)}$ produces fastest run time within required by a considerable margin, positioning it favourably use Bayesian parameter inference. This method, however, further development testing extend its certain analysis scenarios, particularly those involving scale-dependent growth rate. For other reasons discussed herein, alternative such as matter}$ Levin}$ may be necessary full exploration space. also usual first-order Limber approximation is insufficiently accurate $\ell=200-1000$, whereas second-order these scales (with method at smaller $\ell$) does suffice.