Conditional Gradient Methods are an important class of methods to minimize smooth convex functions over polytopes. In this talk, Pokutta will give a broad overview of these methods and their applications and present some recent results both in traditional optimization and learning as well as
Tree decompositions are a powerful tool in structural graph theory that is traditionally used in the context of forbidden graph minors. Connecting tree decompositions and forbidden induced subgraphs has so far remained out of reach. Recently, several results in this direction have been obtained. The
On 1 and 2 March 2021, the Berlin Mathematical School will host its first online BMS Days. On Tuesday, 2 March 2021 all events will be open to the MATH+ community. We are pleased to invite you to attend the following two talks: Gabriele Steidl
In this talk, Schwartz will explain the bounds proved by Halpern and Weaver and then show how to improve the lower bound. He will also explain an approach to getting the sharp bound, which amounts to approximating a paper Möbius band by certain kinds of
Large ensembles of points with Coulomb interactions arise in various settings of condensed matter physics, classical and quantum mechanics, statistical mechanics, random matrices and even approximation theory, and they give rise to a variety of questions pertaining to analysis, partial differential equations and probability. Sylvia Serfaty
Complex coupled systems arise in medicine, science and engineering. Mathematically, these can be nonlinear coupled systems of PDEs or the nonconvex minimization problems found in deep learning. In this talk, we show how multi-scale approaches give a mathematical technique for accurate and efficient simulation of
An expander graph may be defined in any of several ways: in terms of boundaries of vertex sets, or in terms of eigenvalues of the graph Laplacian, or by random walks
We have three more MATH+ Friday Talks coming up in the current winter semester: 15 January: Harald Helfgott (Göttingen) 22 January: Rolf Krause (Lugano) 05 February: Sylvia Serfaty (Courant Institute, NYU)
Multilevel Strategies for Non-Linear Problems and Machine Learning: On Non-Linear Preconditioning, Multilevel Optimization, and Multilevel Training
In classical mechanics, symmetry occurs for a reason: there is a conserved quantity such as angular momentum. This is Noether’s theorem, and it points to a broader theme in dynamics that symmetry is rare and meaningful. Wilkinson will discuss, in the contexts of modern dynamics