My research has focused primarily on using large-scale survey projects to study transient events, particularly supernovae, and the computation involved with analyzing massive data sets. I am especially interested in type-Ia supernovae and their use for probing the expansion history of the universe and the nature of dark energy.
For more on the specific projects I have worked on, see below.
Since coming to Ohio State I have been involved with the All-Sky Automated Survey for SuperNovae, or ASAS-SN. The goal of the ASAS-SN project is ultimately to use robotic 14-cm telescopes to survey the entire visible sky nightly down to about 17th magnitude. The transient, variable, and violent events that will be discovered by such a survey will provide insight that may prove transformative to the field of astrophysics.
As a member of the ASAS-SN team, I work with Professors Kris Stanek and Chris Kochanek to identify and classify potentially interesting events, and to follow up on such events with larger telescopes. My collaborators and I have already authored multiple papers on objects discovered by ASAS-SN, and are collecting and analyzing follow-up data on numerous others. For more on discoveries made by the ASAS-SN project, please see the project website, our transients page, or our supernova page.
I have also worked on creating videos to highlight the capabilities and discoveries of the ASAS-SN project. To view these videos, please visit our YouTube channel.
I spent the summer of 2016 at SLAC National Accelerator Laboratory working with Professor Risa Wechsler and Dr. Phil Marshall on two new pieces of open-source Python software to be used for sampling realistic Type Ia supernova properties given a set of observed host galaxy properties. The ability to predict and simulate realistic supernova for host galaxies is important for testing the detection efficiency of survey projects, which is needed to calculate transient rates, and to developing automated data processing pipelines for future surveys like LSST, and our software was written with the goal of improving supernovae in simulated LSST data.
XDGMM is a new Python class for using extreme deconvolution (XD) Gaussian mixture models to do density estimation of noisy, heterogenous, and incomplete data. It allows the user to select from two existing XDGMM fitting methods (the astroML and Bovy et al. 2011 algorithms) and extends the scikit-learn BaseEstimator class so that it is compatible with scikit-learn cross validation methods. Most importantly, it allows the user to produce an XDGMM model that has been conditioned on known values of some of the parameters used to create the model. The user can then sample values for the remaining parameters from the conditioned model, allowing XDGMM to be used as a prediction tool.
empiriciSN is a tool for sampling realistic supernova parameters given photometric observations of a host, and is an application of the XDGMM class. We built empiriciSN to use only observed host measurements to avoid the need for using theoretical models to infer physical host properties and introduce further uncertainty. The class comes with a default XDGMM model that has been trained on a set of 1432 supernova and host galaxy observations from SNLS and SDSS, but can also be refit with new or additional datasets. Given host ugriz magnitudes, redshift, and surface brightness profiles in each filter, empiriciSN can sample a likely location within the host for a supernova to be placed, calculate the local surface brightness at the location of the host, and use all the host properties to condition the XDGMM model and sample realistic SALT2 supernova parameters for the supernova. empiriciSN is currently being implemented to improve the supernovae planted in the LSST Twinkles data simulations.
To download or get more information on XDGMM or empiriciSN, please see the github repositories linked above.
At Rutgers I worked as a member of the supernova search teams for the Cluster Lensing And Supernova survey with Hubble (CLASH) and Cosmic Assembly Near-infrared Deep Extragalactic Legacy Survey (CANDELS) survey projects. As part of these teams, I searched Hubble Space Telescope survey images of galaxy clusters for new supernovae, with the goal of finding new high-redshift Type Ia supernovae and using these supernovae to improve cosmological measurements. I also participated in searching images with planted fake supernovae in order to measure the supernovae rate and constrain Type Ia progenitor models. To learn more about our findings, please see the papers linked from my publications page.