My research interests are broadly described as galaxy formation and evolution. Most of my undergraduate research focused on a subset of galaxies known as sub-mm galaxies, studying their properties, learning a lot about SED modeling, and how to think like a scientist. My current research is built upon the tools and techniques I learned then, and I am now working to improve our understanding of SED modeling and the assumptions we make when asking a simple question like “How do we measure the stellar mass of this galaxy?” I work with Dr. Desika Narayanan and utilize Hipergator, University of Florida’s supercomputer cluster, using cosmological simulations to push our understanding of galaxy formation and evolution.
Improving Galaxy SED Modeling With Cosmological Simulations
The primary method for inferring the stellar mass of a galaxy is through spectral energy distribution (SED) modeling. However, the technique rests on assumptions such as the galaxy star formation history (SFH) and dust attenuation law that can severely impact the accuracy of derived physical properties from SED modeling. I'm interested in examining the impact of various model assumptions on inferred galaxy properties, and the projects I'm working on are summarized below:
Nonparametric SFHs: First, in Lower et al. 2020, I examine the efficacy of non-parametric SFHs, that is, models for the star formation history of a galaxy that do not explicitly assume a shape, and the contribution to the uncertainty on inferred galaxy stellar masses from the SFH alone.
Mirkwood ML-Powered SED Modeling: In Gilda, Lower, & Narayanan 2021, we develop a machine learning framework, trained on a suite of 3 cosmological hydrodynamical simulations, to infer galaxy physical properties from broadband photometry. Our fiducial model outperforms state-of-the-art Bayesian inference based SED fitting, paving the way for next-generation machine learning based SED modeling.
Accounting for Geometry in Dust Attenuation Models: My current work expands on Lower et al. 2020 by incorporating the impact of diffuse dust and developing a model for dust attenuation that includes a term to describe the star-dust geometry of a galaxy. Namely, we show that allowing deviations from a uniform screen model, in which all stellar light is attenuated by equal optical depth, our ability to infer the true galaxy attenuation curves, and most importantly, dust-corrected star formation rates is greatly improved over traditional methods.
Infrared and Radio Properties of High Redshift Dusty Star Forming Galaxies
My first experiences with research in undergrad at UIUC were centered around galaxies discovered with the South Pole Telescope (SPT). These galaxies comprise a catalog of systematically selected gravitationally lensed dusty galaxies. Gravitational lensing results from massive foreground galaxies or galaxy clusters bending light emitted from background galaxies towards Earth, magnifying the brightness which allows fainter and higher redshift galaxies to be seen. As a consequence, these sources are selected to be at high redshift and are very luminous at sub-mm wavelengths. Under the supervision of Dr. Joaquin Vieira, my work broadly involved characterizing these sources in terms of their physical properties. I estimated the number density of the SPT source catalog, fit modified blackbody SEDs to their photometry to estimate dust properties, and analyzed radio data from ATCA to determine the far-infrared radio correlation at high redshift.
Publications & Press
First and Second Author
Galaxy Growth in a Massive Halo in the First Billion Years of Cosmic History, Marrone, D. P., J. S. Spilker, C. C. Hayward, J. D. Vieira, et al., including S. Lower, 2018, Nature, 553, 51
Young Scholars Program: local youth get hands-on with leading-edge research, https://npl.illinois.edu/news/story.asp?id=23389
Astronomers at Illinois shed light on how galaxies formed, https://las.illinois.edu/news/2017-12-06/astronomers-illinois-shed-light-how-galaxies-formed