A Microlensing Detection Algorithm for Wide-Field Surveys

Gravitational microlensing is a rare event in which the light from a foreground star (source star) is amplified temporarily as it goes around the Einstein radius of another star (lens star). This only occurs when the two stars align with the line of sight of the observer. The significance of microlensing is that it allows for the detection of planets, as when a planet orbiting the lensing star aligns within the Einstein radius, it acts as an additional lens that further amplifies the light. This results in a gaussian-like light curve with an additional deviation on the curve. Unlike transit events, microlensing allows for the detection of small, cold rocky worlds that are not detectable any other way due to their small size. Unfortunately, microlensing events are rare, and only occur once. A microlensing event can be expected for every million stars observed, thus in order to increase the probability of catching an event, we must look at large-field surveys that extract the photometry of many stars on a night-by-night basis. This research project aims to create an efficient machine learning algorithm for wide-field surveys that can distinguish between microlensing events and other type of stars that can often times yield false-alerts due to intrinsic variability. By detecting these events early on, we can point additional telescopes toward the star to measure a complete lightcurve, increasing our chance of detecting any exoplanets orbiting the lensing star