Peter Melchior

• Program

  Optimizing Subaru PFS

• Institution

  Princeton University

Peter Melchior is Assistant Professor of Statistical Astronomy at Princeton University’s Department of Astrophysical Sciences and the Center for Statistics and Machine Learning. He leads the Princeton Astro Data Lab, whose members develop new algorithms to change how astronomy is done. Prof. Melchior’s central research objective: how to optimally combine multiple data sets at the pixel level, in particular for the upcoming surveys Rubin, Euclid, and Roman. The Astro Data Lab develops techniques for source separation, data fusion, and fast inference, using generative modeling with and without neural networks. On an even larger scale, the Astro Data Lab optimizes the full scientific duty cycle, from observing strategy to data analysis, for maximum yield: precision measurements and discovery potential. Funded by the Schmidt Futures Foundation, Prof. Melchior and his lab build modern statistical and machine learning methods for target selection and physical modeling of the spectra from the Prime Focus Spectrograph, which is currently constructed at the Subaru Telescope in Hawai’i. Prof. Melchior also leads the method development of the NSF-funded Convergence Accelerator project “HydroGEN”, which builds a physics-based machine learning platform for hydrologic scenario generation to predict droughts, wildfire conditions, and the availability of drinking water for the entire continental United States.