Simone Dovetta (CNR-IMATI, Pavia)
Abstract:
The talk overviews some recent developments about nonlinear Schroedinger equations (NLS) on metric graphs. Precisely, we concentrate on variational problems for the NLS energy functional subject to the mass constraint. After a brief recap of the well-known behaviour of such model on the real line, we address the existence of NLS ground states on noncompact metric graphs, with a specific focus on periodic graphs and infinite trees. The emergence of threshold phenomena rooted in the nature of these graphs is discussed. Finally, we provide some insights on the uniqueness of ground states at fixed mass. On the one hand, uniqueness is shown to hold for two classes of graphs with halflines. On the other hand, a counterexample to uniqueness in full generality is exhibited. The matter we discuss is part of a wider research line, developed in collaboration with several authors. The results explicitly covered by the talk refer to a series of papers, some of which are joint works with Riccardo Adami, Enrico Serra and Paolo Tilli.
14:15 • Universität Basel, Zoom Seminar
Dr. Ulrike Riess (ETH-ITS)
On the Kähler cone of irreducible symplectic orbifolds abstract
Abstract:
In this talk I report on recent joint work with G. Menet: Wegeneralize a series of classical results on irreducible symplecticmanifolds to the orbifold setting. In particular we prove acharacterization of the Kähler cone using wall divisors. Thisgeneralizes results of Mongardi for the smooth case. I will finish thetalk by applying these results to study a concrete example.
15:00 • Zoom
Antionio Auffinger (Northwestern University)
Abstract:
I will discuss the complexity of minima and saddles in two classic examples of random landscapes: the spherical p-spin glass model and the model of a single particle in a random confining potential. In these models, the number of critical points diverge exponentially as the dimension of the ambient space grows. By combining Kac-Rice’s formula and random matrix tools, we are able to estimate this exponential growth. I will also explain some consequences of these results to statistical physics and optimization theory.
16:00 • Universität Basel, Zoom Seminar
Prof. Dr. Richard Nickl (University of Cambridge)
Bayesian inverse problems, Gaussian processes, and partial differential equations. abstract
Abstract:
The Bayesian approach to inverse problems has become very popular in the last decade after seminal work by Andrew Stuart (2010) and collaborators. Particularly in non-linear applications with PDEs and when using Gaussian process priors, this can leverage powerful MCMC methodology to tackle difficult high-dimensional and non-convex inference problems. Little is known in terms of rigorous performance guarantees for such algorithms. After laying out the main ideas behind Bayesian inversion, we will discuss recent progress providing both statistical and computational guarantees for these methods. We will touch upon issues such as how to prove posterior consistency and how to objectively validate posterior uncertainty quantification methods. A main focus will be on very recent results about mixing times of high-dimensional Langevin dynamics that establish the polynomial time computability of posterior measures in some non-linear model examples arising with PDEs.
Dr. Patrick Orson (ETH Zürich, Switzerland)
Topologically embedding spheres in knot traces abstract
Abstract:
The problem of when a given 2-dimensional homology class in a 4-manifold can be represented by an embedded 2-sphere has been studied by many authors. In this talk we will consider the question for knot traces; these are particularly simple 4-manifolds, described by taking the closed 4-ball and glueing a 2-handle to it along a framed knot in the boundary 3-sphere. The resulting manifold is homotopy equivalent to the 2-sphere, and we consider the embedding problem for a generator of the second homology group. I will describe recent joint work in which we completely characterise when this homology class can be represented by an embedded 2-sphere whose complement has abelian fundamental group. The answer is in terms of classical and easily computable invariants of the knot in the 3-sphere. Given time I will describe some knot-theoretic corollaries, relating to the study of slice knots, and a uniqueness result for oursphere embeddings.
16:15 • ETH Zentrum, Rämistrasse 101, Zürich, Building HG, Room G 19.1
Navid Nouri (EPF Lausanne)
Kernel Density Estimation through Density Constrained Near Neighbor Search abstract
Abstract:
In the kernel density estimation problem (KDE) the task is to preprocess the input dataset of points into a data structure that can quickly output, given a query, q, a multiplicative approximation to the average of K(q, p) over all data points in the dataset, for a kernel function K (e.g., the Gaussian kernel). Several recent works have proposed data structures for KDE that essentially use nearest neighbour search data structures as a sampling primitive.
In this talk we will show how to obtain an efficient data structure for KDE by directly implementing a natural importance sampling approach using locality sensitive hashing. The resulting data structure is easy to describe and analyse, and at the same time improves upon or essentially matches the best known query time and space complexity for all radial kernels considered in the literature so far. We will also outline an improved data structure based on recent data-dependent nearest neighbour search techniques. Joint work with Moses Charikar, Michael Kapralov and Paris Siminelakis.
16:15 • EPF Lausanne, Salle CO1