Predicting stellar flares using hidden Markov models

Certain classes of stars in the universe emit flares – dramatic bursts of energy that appear across the electromagnetic spectrum. Flare activity for these stars is highly variable across time, and predicting this behaviour is an interesting problem both from an astronomical and from a statistical viewpoint. Within the field of astrostatistics, “changepoint detection” models seek to identify the previous times at which a flare star switches from a low-energy to a high-energy state by examining streams of photon emissions from the star. In this project, we aim to upgrade the astrostatistical toolbox by applying “hidden Markov models” to the same kinds of data, which will not only let us simulate the mechanisms that drive flare emissions, but also allow us to predict future flares.

Faculty Supervisor:

Radu Craiu

Student:

Partner:

Imperial College London

Discipline:

Physics

Sector:

Education

University:

University of Toronto

Program: