A Julia package to provide RML imaging methods for high-resolution radio interferometry
Author EHTJulia
5 Stars
Updated Last
1 Year Ago
Started In
February 2023


Dev Build Status Coverage

This module provides radio astornomy tool kits of the Regularized Maximum Likelihood (RML) imaging methods of radio interferometry. The current focus is the implementation of RML imaging methods for high-angular-resolution imaging within a modest field of view where the direction-dependent effects are ignorable. The tool kits and algorithms are designed to meet the need for multi-dimensional imaging with Very Long Baseline Interferometry (e.g. Event Horizon Telescope) and millimeter interferometry (e.g. ALMA). The library aims to provide the following features:

  • Julia native implementation of the end-to-end RML imaging process: all the codes involved in RML imaging methods are written in Julia. This allows a significant accerelation from popular python implementations of RML imaging methods, with keeping the easiness to install the package and its dependencies. The Julia-native codes allow the utilization of various powerful packages in Julia's ecosystem --- this also offers the potential to scale the entire code to GPU and clusters easily in the future.
  • A set of data types and functions for the sky intensity distributions, non-uniform fast Fourier transformations (NUFFT), regularization functions, data likelihood functions, and optimization to solve images. Julia's efficient way of abstracting data types and their function allows interested users to implement their imaging methods.
  • AD-friendly implementation: the relevant functions for imaging (e.g., forward/adjoint transforms from sky images to Fourier-domain visibilities, regularization functions) are compatible with Automatic differentiation (AD) packages in the Julia's ChainRules ecosystem, for instance Zygote.jl. This allows the easier implementation of algorithms using complicated sky models or sets of data products.


Assuming that you already have Julia correctly installed, it suffices to import RMLImaging.jl in the standard way:

using Pkg


The documentation is in preparation, but docstrings of available functions are listed for the latest version. The stable version has not been released.


The development of this package has been finantially supported by the following programs.

The National Radio Astronomy Observatory is a facility of the National Science Foundation operated under cooperative agreement by Associated Universities, Inc.