Title: The scientific utility of software development: a case study in radio interferometry

Abstract: Data is one of the fundamental quantities on which progress in astronomy is predicated. Yet there is often a disproportionate emphasis on immediate science results over the development of robust methods to process and even model the underlying data. In this talk I will present a case study demonstrating the benefits of investing in modern software development for data modeling – the challenge of imaging in radio interferometry. I will first overview the standard approach to imaging, wherein telescopes such as ALMA are purpose-built to observe at generational resolution and sensitivity, while the software that images the data uses an empirical, decades-old modeling framework that does not leverage the highest resolution and sensitivity information. Then I will discuss two modern, open-source software packages designed to perform imaging more accurately, quickly and autonomously, within a simple and well-documented software architecture. Comparing their performance on ALMA observations of protoplanetary disks to the standard imaging approach, I will show how these algorithms employ statistical techniques to simultaneously achieve higher resolution and sensitivity in their reconstructed images. Finally I will preview the new science these tools are driving in the protoplanetary disk field, including the Bayesian inference made possible by self-consistently imaging hundreds of datasets at high fidelity.

Astro Colloquium and 'coffee & cookies' department gathering (3:45-4:00pm)

Please click the link to join virtually: https://psu.zoom.us/j/92637070419