Map Reconstruction of radio observations with Conditional Invertible Neural Networks

Abstract

In radio astronomy, the challenge of reconstructing a sky map from time ordered data (TOD) is known as an inverse problem. Standard map-making techniques and gridding algorithms are commonly employed to address this problem, each offering its own benefits such as producing minimum-variance maps. However, these approaches also carry limitations such as computational inefficiency and numerical instability in map-making and the inability to remove beam effects in grid-based methods. To overcome these challenges, this study proposes a novel solution through the use of the conditional invertible neural network (cINN) for efficient sky map reconstruction. With the aid of forward modeling, where the simulated TODs are generated from a given sky model with a specific observation, the trained neural network can produce accurate reconstructed sky maps. Using the five-hundred-meter aperture spherical radio telescope (FAST) as an example, cINN demonstrates remarkable performance in map reconstruction from simulated TODs, achieving a mean squared error of $2.29\pm 2.14\times 10^{-4}{\mathrm{K}}^2$, a structural similarity index of $0.968\pm 0.002$, and a peak signal-to-noise ratio of $26.13\pm 5.22$ at the $1\sigma$ level. Furthermore, by sampling in the latent space of cINN, the reconstruction errors for each pixel can be accurately quantified.

Accepted for publication in Research in Astronomy and Astrophysics (RAA); 20 pages, 10 figures

References in corpus (13)

• The Five-Hundred-Meter Aperture Spherical Radio Telescope (FAST) Project
• The CAMELS project: Cosmology and Astrophysics with MachinE Learning Simulations
• The Five-hundred-meter Aperture Spherical Radio Telescope (FAST) Project
• Simulation-Based Inference of Reionization Parameters From 3D Tomographic 21 cm Lightcone Images
• deep21: a Deep Learning Method for 21cm Foreground Removal
• Implicit Likelihood Inference of Reionization Parameters from the 21 cm Power Spectrum
• AI-assisted super-resolution cosmological simulations II: Halo substructures, velocities and higher order statistics
• Inference of cosmic-ray source properties by conditional invertible neural networks
• Reconstructing Cosmological Initial Conditions from Late-Time Structure with Convolutional Neural Networks
• Cosmic Velocity Field Reconstruction Using AI
• Single frequency CMB B-mode inference with realistic foregrounds from a single training image
• Emission-line diagnostics of HII regions using conditional Invertible Neural Networks
• HCGrid: A Convolution-based Gridding Framework for RadioAstronomy in Hybrid Computing Environments

Cited by in corpus (1)

• Deep learning for cosmological parameter inference from a dark matter halo density field