[Paper Review] Galaxy shapes of Light (GaLight): a 2D modeling of galaxy images

Xuheng Ding, Simon Birrer+2 more|arXiv (Cornell University)|Nov 16, 2021

Galaxies: Formation, Evolution, Phenomena17 references4 citations

TL;DR

galight is an open-source Python package for automated 2D modeling of galaxy images, leveraging lenstronomy's fitting engine to decompose light profiles into components like disks, bulges, bars, and quasars. It enables efficient, user-friendly analysis of large extragalactic datasets with automatic PSF-star detection, noise map estimation, and cutout sizing, achieving robust photometric decomposition with minimal user intervention.

ABSTRACT

Galight is a Python-based open-source package that can be used to perform two-dimensional model fitting of optical and near-infrared images to characterize the light distribution of galaxies with components including a disk, bulge, bar, and quasar. The decomposition of stellar components has been demonstrated in published studies of inactive galaxies and quasar host galaxies observed by the Hubble Space Telescope and Subaru's Hyper Suprime-Cam. Galight utilizes the image modeling capabilities of lenstronomy while redesigning the user interface for the analysis of large samples of extragalactic sources. The package is user-friendly with some automatic features such as determining the cutout size of the modeling frame, searching for PSF-stars in field-of-view, estimating the noise map of the data, identifying all the objects to set the initial model, and associated parameters to fit them simultaneously. These features minimize the manpower and allow the automatic fitting tasks. The software is distributed under the MIT license. The source code, installation guidelines, and example notebooks code can be found at https://galight.readthedocs.io/en/latest/

Motivation & Objective

• To enable high-throughput, automated decomposition of galaxy light profiles in large imaging surveys.
• To reduce manual effort in fitting galaxy components (disk, bulge, bar, AGN) across vast datasets.
• To integrate robust PSF and noise estimation directly into the modeling pipeline for improved photometric accuracy.
• To provide a user-friendly, extensible interface for researchers analyzing optical and near-infrared galaxy images.
• To support the analysis of quasar host galaxies and dual AGN systems in deep imaging surveys.

Proposed method

• Uses lenstronomy’s forward modeling framework with semi-linear inversion and PSO/MCMC for parameter estimation.
• Automatically detects and masks objects to identify sky regions for noise map estimation.
• Employs a FWHM-based algorithm to identify suitable PSF-stars from the field-of-view using local maxima and profile sharpness.
• Estimates Poisson and Gaussian noise components from pixel counts and exposure time when no error map is provided.
• Applies adaptive cutout sizing and automatically sets initial model parameters for multiple sources in a single image.
• Supports composite models (e.g., disk + bulge) and includes built-in priors and mask handling for complex source configurations.

Experimental results

Research questions

• RQ1 How can galaxy image decomposition be automated to handle large-scale surveys efficiently and accurately?
• RQ2 What is the optimal method for automatically identifying PSF stars in crowded fields without user input?
• RQ3 How can noise maps be reliably estimated from sky regions when no error map is provided?
• RQ4 To what extent can a unified modeling framework handle diverse galaxy components including AGNs and dual AGN systems?
• RQ5 Can a modular, open-source Python package achieve performance comparable to specialized tools like Galfit while supporting advanced statistical inference?

Key findings

• galight successfully models ~1.5 million galaxies and ~5,000 quasars from HST and Subaru HSC surveys using automated workflows.
• The package automatically determines optimal cutout sizes, identifies PSF-stars via FWHM filtering, and estimates noise maps with high accuracy.
• Noise estimation based on sky region statistics yields median values near zero when background light is removed, confirming reliability.
• PSF-star candidates are ranked by FWHM, with the sharpest profile selected by default, improving PSF modeling fidelity.
• The integration of lenstronomy enables robust MCMC error analysis and PSF uncertainty handling, enhancing photometric accuracy.
• Example notebooks demonstrate successful fitting of quasars, dual AGN candidates, and composite galaxies, validating the tool’s versatility and scalability.