Denoising weak lensing mass maps with deep learning
Abstract
Weak gravitational lensing is a powerful probe of the largescale cosmic matter distribution. Widefield galaxy surveys allow us to generate the socalled weak lensing maps, but actual observations suffer from noise due to imperfect measurement of galaxy shape distortions and to the limited number density of the source galaxies. In this paper, we explore a deeplearning approach to reduce the noise. We develop an imagetoimage translation method with conditional adversarial networks (CANs), which learn efficient mapping from an input noisy weak lensing map to the underlying noise field. We train the CANs using 30000 image pairs obtained from 1000 raytracing simulations of weak gravitational lensing. We show that the trained CANs reproduce the true onepoint probability distribution function (PDF) of the noiseless lensing map with a bias less than
 Publication:

Physical Review D
 Pub Date:
 August 2019
 DOI:
 10.1103/PhysRevD.100.043527
 arXiv:
 arXiv:1812.05781
 Bibcode:
 2019PhRvD.100d3527S
 Keywords:

 Astrophysics  Cosmology and Nongalactic Astrophysics;
 Computer Science  Machine Learning
 EPrint:
 15 pages, 12 figures, accepted for publication in Phys. Rev. D