vis-r: Quick radial profile modelling for radio interferometric visibilities

TL;DR

vis-r is a fast, efficient method for modeling radial profiles of radio interferometric visibilities, significantly reducing computational costs compared to traditional image-based approaches, suitable for ALMA debris disk observations.

Contribution

The paper introduces vis-r, a novel radial profile modeling technique that avoids image generation, leveraging u,v space averaging and the Discrete Hankel Transform for rapid analysis.

Findings

Method is 10-100 times faster than image-based modeling.

Suitable for most ALMA debris disk observations on standard laptops.

Compatible with gradient-based fitting methods like Hamiltonian Monte-Carlo.

Abstract

This paper describes a method for parametric radial profile modelling of radio interferometric visibility data. Image-based parametric modelling is common in the field of circumstellar debris disks, and high resolution ALMA observations make this method computationally expensive because many high resolution images must be generated and Fourier transformed. Most circumstellar disks are axisymmetric, so here a method that avoids generating images and instead models radial profiles is outlined. The two main elements are i) $u,v$ space visibility averaging, and ii) using the Discrete Hankel Transform to convert arbitrary parametric radial surface brightness profiles to visibilities. Vertical scale height can be included, but is fixed with radius; in principle this is a limitation but in practise radially-dependent vertical structure information is beyond even ALMA's capability for all but a few debris disks. Combined, these ingredients yield a method that is one to two orders of magnitude faster than image based methods and can be run on a laptop, is sufficient for most ALMA observations of debris disks, and can be used with gradient-based fitting methods such as Hamiltonian Monte-Carlo. A small software package is provided, which is dubbed vis-r.