My astrophysical research is on the role of dust and gas in the evolution of galaxies over cosmic time. I study this role over a range of scales, with a focus on the flow of dust and gas between different phases. By examining the velocity field of dust and gas in our galaxy, we can learn about how the sites of star formation—cold, dense molecular clouds—arise out of and return to the warmer and more diffuse medium that pervades the galaxy. On the opposite end of the range of scales, by studying the composition and physical conditions of the large halos of gas surrounding galaxies, we can learn about how galaxies acquire (or, just as importantly, fail to acquire) fresh star formation fuel from the cosmic web.
Studying gas and dust in space gives rise to some interesting statistical problems. Measurements of dust and gas in our galaxy do not come with distance information. Determining the velocity field of dust and gas is a tomography problem that requires combining qualitatively different kinds of observations. I am interested in ways of incorporating physical priors such as hydrodynamical constraints into this problem and other tricky inverse problems.
Measurements of the cosmic web and gaseous halos of galaxies are generally done using absorption spectroscopy. Algorithmic identification of features in these absorption spectra is, in general, an unsolved problem. I am interested in applying different model selection frameworks to this identification problem, drawing on information from related identification problems—topic modeling, peak identification in other forms of spectroscopy—for guidance.