Generative modelling of convergence maps based on predicted one-point statistics

¹ Université Paris-Saclay, Université Paris Cité, CEA, CNRS, AIM, 91191 Gif-sur-Yvette, France
² Institutes of Computer Science and Astrophysics, Foundation for Research and Technology Hellas (FORTH), Heraklion, Crete, Greece

^⋆ Corresponding author: vilasini.tinnanerisreekanth@cea.fr

Received: 14 February 2025
Accepted: 6 July 2025

Abstract

Context. Weak gravitational lensing is a key cosmological probe for current and future large-scale surveys. While power spectra are commonly used for analyses, they fail to capture non-Gaussian information from non-linear structure formation, which necessitates higher-order statistics and methods for an efficient map generation.

Aims. We develop an emulator that generates accurate convergence (κ) maps directly from an input power spectrum and wavelet ℓ₁-norm without relying on computationally intensive simulations.

Methods. We used either numerical or theoretical predictions to construct κ maps by iteratively adjusting the wavelet coefficients to match the marginal distributions of the target and their inter-scale dependences by incorporating higher-order statistical information.

Results. The resulting κ maps accurately reproduce the input power spectrum, and their higher-order statistical properties are consistent with the input predictions. They thus provide an efficient tool for weak-lensing analyses.