Resolving the Vicinity of Supermassive Black Holes with Gravitational Microlensing

TL;DR

This paper presents simulations and machine learning methods to analyze caustic-crossing microlensing events in quasars, aiming to probe the vicinity of supermassive black holes and measure the innermost stable circular orbit (ISCO).

Contribution

It introduces realistic accretion disk models with relativistic effects and demonstrates neural networks can effectively estimate ISCO size and orientation from microlensing light curves.

Findings

Inflection points in light curves can reveal ISCO size.

Neural networks outperform traditional methods in estimating black hole parameters.

Simulations show potential for probing supermassive black hole environments.

Abstract

In the near future, wide field surveys will discover 1000's of new strongly lensed quasars, and these will be monitored with unprecedented cadence by the Legacy Survey of Space and Time (LSST). Many of these will undergo caustic-crossing microlensing events over the 10-year LSST survey, in which a sharp caustic feature from a stellar body in the lensing galaxy crosses the inner accretion disk. Caustic-crossing events offer the unique opportunity to probe the vicinity of the central supermassive black hole for 100s of quasars with multi-platform follow-up triggered by LSST monitoring. To prepare for these observations, we have developed detailed simulations of caustic-crossing light curves. These employ a realistic analytic model of the inner accretion disk that reveals the strong surface brightness asymmetries introduced when fully accounting for both special- and general-relativistic effects. We demonstrate that an inflection in the caustic-crossing light curve due to the innermost stable circular orbit (ISCO) can be detected in reasonable follow-up observations and can be analyzed to constrain ISCO size. We also demonstrate that a convolutional neural network can be trained to predict ISCO size more reliably than traditional approaches and can also recover source orientation with high accuracy.