Nuclear Transients Monitor

Python DASH Plotly

Building on the results of my previous works, I developed a pipeline to query the nuclear transients found every day and retrieve their steady-state information from various sources in order to curate potentially interesting candidates (e.g., TDEs or highly variable AGNs) for spectroscopic classification. To facilitate the inspection on candidates, I also built a Dash-app with Python to help visualize hundreds of currently active nuclear transients and their properties.

My pipeline uses Gaussian process to resample the noisy, nonuniformly sampled light curve data. This allows a robust extraction of time-series features such as gradient, color gradient, and other statistics that can be used to generate a diagnostic or predictive model. For example, I used these information to separate the light curve shapes into different stages: flat, rising, peaked, and declining. The ones with flat light curves are more likely to be from AGNs with stochastic variability while the burst-like light curves tend to be stellar explosions or TDEs.