Tuesday, April 2, 2019 at 4:00pm to 5:00pm
Mohler Laboratory, 453
200 W Packer Ave., Bethlehem, PA 18015
The advent of computational science has unveiled large classes of nonlinear optimization problems where derivatives of the objective and/or constraint functions are unavailable. Often, these problems are posed as black-box optimization problems, but rarely is this by necessity. We report on our experience extracting additional structure on problems consisting of both black-box and algebraic or otherwise known components. We provide diverse examples of such local optimization problems that are being solved at Argonne National Laboratory, including nonlinear least squares, robust optimization, and other forms of composite nonsmooth optimization. In each case, we use quadratic or RBF surrogates to locally model both the black-box and algebraic components and obtain new, globally convergent “grey-box” optimization methods. Joint work with Jeff Larson, Matt Menickelly, and others.
Stefan Wild is a Computational Mathematician and Deputy Division Director of the Mathematics and Computer Science Division at Argonne National Laboratory and a Senior Fellow in the Northwestern Argonne Institute for Science and Engineering at Northwestern University. Wild joined Argonne as a Director’s Postdoctoral Fellow in September 2008. Prior to this, he obtained his Ph.D. in operations research from Cornell University and his M.S. and B.S. in applied mathematics from the University of Colorado. Wild’s primary research focus is developing model-based algorithms and software for challenging numerical optimization problems. He applies these techniques for data analysis, machine learning, and the solution of nonlinear inverse problems. At Argonne he leads a number of multidisciplinary computational science projects and shapes strategy for applied mathematics, numerical software, and statistics.