Seminar - Connor Stone

Wednesday, August 16, 2023 3:00 pm - 3:00 pm EDT (GMT -04:00)

Connor Stone
Connor Stone is currently Postdoctoral Fellow at the Université de Montréal with an NSERC Postdoctoral Fellowship and a CITA National Fellowship after receiving his Ph.D from Queen's University in 2022. He is a University of Waterloo alumnus, class of 2016. As an undergraduate his interest wasn't yet in astrophysics but particle physics, spending the majority of his co-op time working on the DEAP-3600 dark matter detector at SNOLAB.  It was during his time at Queen's that he developed his expertise in astronomical image processing and galaxy structure under the supervision of Prof. Stéphane Courteau.

Title: AstroPhot: Fitting Everything Everywhere All at Once in Astronomical Images

Abstract: The latest generation of telescopes is producing multi-band and multi-epoch data at an unprecedented scale of quantity and quality. However, the pixels in these images are not of-themselves very informative; only when models are applied to these images do insights emerge. For this purpose we have developed AstroPhot, a python-based, GPU-accelerated, differentiable, object-oriented tool for fitting astronomical images. Everything: AstroPhot can fit models for sky, stars, galaxies, PSFs, and more in a principled chi^2 forward optimization, recovering Bayesian posterior information and covariance of all parameters. Everywhere: AstroPhot can optimize forward models on CPU or GPU; across images that are large, multi-band, multi-epoch, rotated, dithered, and more. All at once: The models are optimized together, thus handling overlapping objects and including the covariance between parameters. I will discuss how these capabilities compare with other PSF and galaxy fitting techniques. Finally, I will discuss how differentiable codes (like AstroPhot, though extending to machine learning) can enable new avenues for principled Bayesian analysis of complex data.