ShOpt.jl: A Julia Package for Empirical Point Spread Function Characterization of JWST NIRCam Data

TL;DR

ShOpt.jl is a Julia-based software package designed for efficient and scalable point spread function characterization of JWST NIRCam data, leveraging advanced optimization techniques and flexible modeling options.

Contribution

This paper introduces ShOpt.jl, a novel Julia package that combines performance optimizations and versatile PSF modeling methods for astrophysical data analysis.

Findings

Achieves high performance through multithreading and advanced algorithms

Provides flexible PSF modeling with PCA, autoencoders, and analytic profiles

Demonstrates scalability and accuracy for JWST NIRCam data analysis

Abstract

As astronomical data grows in volume and complexity, the scalability of analysis software becomes increasingly important. At the same time, astrophysics analysis software relies heavily on open-source contributions, so languages and tools that prioritize both performance and readability are especially valuable. Julia, with its just-in-time compiler and high level syntax, offers a compelling alternative to traditional languages like Python or C. In this paper, we outline ShOpt.jl, a new software package for point spread function (PSF) characterization written in Julia. ShOpt.jl features a number of performance optimizations, such as multithreading, the use of preconditioners, and the implementation of the memory-limited Broyden-Fletcher-Goldfarb-Shanno algorithm, as well as the flexibility to choose between principal component analysis, an autoencoder, and analytic profiles for PSF characterization. As observatories like the James Webb Space Telescope bring astrophysics into a new era of wide-field, high-resolution imaging, the challenges of PSF modeling become more pronounced. Tools like ShOpt.jl provide the community with a scalable, efficient, and accurate solution to these challenges, while also demonstrating the potential of Julia as a language that meets the demands of modern astrophysical research.