At this time, only Columbia University ID-holders are invited to in-person talks. All others are invited to attend remotely. You may also request the Zoom link for a seminar by sending an email to [email protected].

Speaker: Georg Stadler, Courant Institute of Mathematical Sciences, NYU

Title: Taming nonlinearity by lifting Newton methods, with application to sea-ice modeling

Abstract: Newton-type methods are the main tool for solving nonlinear equations, and how well they work depends on the degree of nonlinearity.  I will illustrate that for severe nonlinearities, it can be beneficial to lift nonlinear equations to a higher-dimensional space, linearize there and project the linearization back to the original space where they are solved.  The approach will be illustrated on simple examples, and connections to nonlinear preconditioning and primal-dual interior point methods will be made. The novel Newton solvers are illustrated for the solution of the most common sea-ice model in climate simulations.  This is joint work with Melody Shih, Carolin Mehlmann and Martin Losch.

Biography: Georg Stadler is a Professor of Mathematics and Computer Science at New York University's Courant Institute of Mathematical Sciences. Before that, he was a member of UT Austin's Oden Insitute. He received his PhD in Mathematics from the University of Graz (Austria). His research interests are in computational inverse problems and uncertainty quantification, PDE-constrained optimization, and large-scale (non)linear solvers. His research is driven by applications in sea and land ice modeling and inference, plasma fusion, and solid earth geoscience. He is a recipient of the Gordon Bell Prize in 2015 and the 2019 SIAM CSE Best Paper Prize, and is an editor for the SIAM Journal on Scientific Computing.

Host: Prof. Kui Ren