MAGIC: Microlensing Analysis Guided by Intelligent Computation

Abstract The modeling of binary microlensing light curves via the standard sampling-based method can be challenging, because time-consuming light-curve computation and pathological likelihood landscape in high-dimensional parameter space. In this work, we present MAGIC, which is a machine-learning framework to efficiently accurately infer parameters events with realistic data quality. are divided into two groups inferred separately different neural networks. key feature MAGIC introduction controlled differential equation, provides capability handle irregular sampling large gaps. Based on simulated curves, show that achieve fractional uncertainties few percent mass ratio separation. We also test real event. able locate degenerate solutions even when gaps introduced. As samplings common astronomical surveys, our has implications for other studies involve time series.