Harnessing Deep Learning for Quasar Light Curve Modeling with QNPy

Abstract

Quasar light curves exhibit intrinsic stochastic variability, which in combination with observational technical limitations, such as frequent observational gaps and irregular cadences, creates significant challenges for their analysis. To effectively address these challenges, in order to explore quasar underlying physical processes, the common incorporation of deep learning stands out as a key method for efficiently modeling quasar light curves. Here, we present our Python package, now available as ”QNPy” on the PyPI platform, which represents a groundbreaking tool for modeling quasar light curves using meta-learning algorithms which are called conditional neural processes. We demonstrate the first application of the QNPy Python package on two case-study samples sourced from the Data Challenge of the LSST AGN Science Collaboration [ 1] and the GAIA space mission.