We present a pair of new methods for solving a broad class of bound-constrained nonsmooth composite minimization problems. These methods are specially designed for objectives that are some algebraically specified (glass-box) mapping of a vector of outputs from a computationally expensive (black-box) function. This problem class covers many interesting problems from a wide range of applications relevant to the Department of Energy Office of Science, and has motivated a lot of work by myself and Jeff Larson (ANL/MCS). We provide rigorous convergence analysis and guarantees, and test the implementations on synthetic problems. For this particular presentation, I will also provide an introduction to the larger field of (model-based) black-box optimization.Short
Speaker Bio: Matt Menickelly (they/he) is a Computational Mathematician in the Mathematics and Computer Science Division at Argonne National Laboratory. Their research focuses primarily on optimization methods for computationally expensive black box oracles, but they have also worked in stochastic optimization, robust optimization, and optimization for machine learning. Matt earned a BM (Bachelor’s in Music, wise guy) in piano performance and a MS in mathematics from Miami University before earning a PhD in Industrial Engineering from Lehigh University. Matt worked at IBM TJ Watson Research Center for a year before becoming a postdoc at Argonne, and has not found a good enough reason to leave the lab since then. Outside of research, Matt still plays piano as a hobby (lately chewing through Samuel Barber’s entire piano works) and took on an over-dependent research assistant about 18 months ago.
See all upcoming talks at https://www.anl.gov/mcs/lans-seminars