Peter K. Friz (Technische Universität Berlin)
Rough Analysis and Applications (ONLINE) abstract
Abstract:
We plan to discuss: the algebra of iterated integrals, signatures and expected signatures, rough paths and models; Kolmogorov type criteria; Schauder estimates; p-variation, abstract Riemann-Stieltjes integration (Sewing); rough integration and reconstruction; a paracontrolled view; rough differential equations and flows; rough transport equations; second order rough partial differential equations; rough volatility; a regularity structure view; large deviations and precise asymptotics for rough volatility.A good part of the lecture will follow F-Hairer (2014,2020).
10:15 • ETH Zentrum, Rämistrasse 101, Zürich, Building HG, Room
Christian Brennecke (Harvard University)
Abstract:
In this talk I will present a variational formula for the thermodynamic limit of the free energy of the Sherrington-Kirkpatrick (SK) spin-glass model with transverse field, a quantum generalization of the classical SK model. Through its path integral representation, the quantum SK model can be translated to a classical vector-spin model with the spins taking values in the space of $ \\{ -1, 1 \\} $-valued cadlag paths. I will explain how to approximate the model by a sequence of classical finite-dimensional vector-spin models which enables us to use results of Panchenko to determine the free energy in the thermodynamic limit. The talk is based on joint work with Arka Adhikari.
11:00 • Universität Basel, Zoom Seminar
Dante Bonolis (IST Austria)
Title T.B.A.
15:15 • EPF Lausanne, Salle tba
Lisa Picrillo (Massachusetts Institute of Technology)
CANCELLED: Title T.B.A.
15:45 • ETH Zentrum, Rämistrasse 101, Zürich, Building HG, Room G 43
Dr. Bram Petri (Sorbonne Université)
Statistics of finite degree covers of torus knot complements abstract
Abstract:
Abstract: a classical theorem due to Hempel states that the fundamental group of a compact three manifold is residually finite. In other words, the manifold has many finite degree covers. This leads to the question how many such covers there are of a given degree, what their typical properties are and what data of the manifold is encoded in these statistics. In this talk, I will speak about joint work with Elizabeth Baker in which we studied these questions for torus knot complements.
Dr. Bram Petri (Université Sorbonne)
ONLINE Seminar - Statistics of finite degree covers of torus knot complements abstract
Abstract:
A classical theorem due to Hempel states that the fundamental group of a compact three manifold is residually finite. In other words, the manifold has many finite degree covers. This leads to the question how many such covers there are of a given degree, what their typical properties are and what data of the manifold is encoded in these statistics. In this talk, I will speak about joint work with Elizabeth Baker in which we studied these questions for torus knot complements.
15:45 • ONLINE Seminar
Prof. Dr. Francis Bach (INRIA/ENS, Paris, France)
On the effectiveness of Richardson Extrapolation in Machine Learning abstract
Abstract:
Richardson extrapolation is a classical technique from numerical analysis that can improve the approximation error of an estimation method by combining linearly several estimates obtained from different values of one of its hyperparameters, without the need to know in details the inner structure of the original estimation method. The main goal of this presentation is to study when Richardson extrapolation can be used within machine learning. We identify two situations where Richardson interpolation can be useful: (1) when the hyperparameter is the number of iterations of an existing iterative optimization algorithm, with applications to averaged gradient descent and Frank-Wolfe algorithms, and (2) when it is a regularization parameter, with applications to Nesterov smoothing techniques for minimizing non-smooth functions and ridge regression. In all these cases, I will show that extrapolation techniques come with no significant loss in performance, but with sometimes strong gains.
Onlline Seminar
Dr. Alain Couvreur (Ecole polytechnique)
eSeminar: On the code equivalence problem in rank metric abstract
Abstract:
(**This eSeminar will take place on Zoom, using same meeting details as previous seminars. If you do not have meeting details, please contact karan.khathuria@math.uzh.ch **)
The code equivalence problem can roughly be stated as follows : "Given two codes \\(C_1\\), \\(C_2\\), is there an isometry \\(\\phi\\) of the ambient space such that \\(\\phi(C_1) = C_2\\)?" In Hamming metric, this problem has been intensively studied in the last decades, with in particular the {\\it support splitting algorithm} by N. Sendrier which solves this problem in the generic case when the isometry \\(\\phi\\) is a permutation. On the rank metric side, the linear isometries of the ambient space are classified and various algebraic invariants related to a given matrix space have been identified. In this talk, we will focus on the algorithmic aspects of the code equivalence problem in rank metric by focusing on three versions: 1. \\(\\mathbb{F}_{q^m}\\)--linear codes with a vector representation 2. \\(\\mathbb{F}_{q^m}\\)--linear codes with a matrix representation 3. Non structured matrix spaces. We propose efficient algorithms to solve versions (1) and (2) of the problem. Then we prove that (3) is at least as hard as the monomial equivalence problem in Hamming metric. This is a work in progress in collaboration with Thomas Debris-Alazard (Royal Holloway, London) and Philippe Gaborit (University of Limoges).
16:00 • Onlline Seminar
Lionel Lang (Stockholm University)
Abstract:
In this joint work in progress with J. Forsgård, we study the topology of maps P:(\\C*)^2 \\to \\C given by Laurent polynomials P(z,w). For specific P, we observed that the topology of the corresponding map can be described in terms of the co-amoeba of a generic fiber. When the latter co-amoeba is maximal, it contains a dimer (a particularly nice graph) whose fundamental cycles corresponds to the vanishing cycles of the map P. For general P, the existence of maximal co-amoebas is widely open. In the meantime, we can bypass co-amoebas, going directly to dimers using a construction of Goncharov-Kenyon and obtain a virtual correspondence between fundamental cycles and vanishing cycles. In this talk, we will discuss how this (virtual) correspondence can be used to compute the monodromy of the map P.
https://agenda.unige.ch/events/view/28620
16:00 • Université de Genève, ZOOM Meeting
Prof. Dr. George Karniadakis (Brown University, Rhode Island)
Physics-Informed Neural Networks (PINNs): Algorithms & Applications abstract
Abstract:
We will present a new approach to develop a data-driven, learning-based framework for predicting outcomes of physical and biological systems and for discovering hidden physics from noisy data. We will introduce a deep learning approach based on neural networks (NNs) and generative adversarial networks (GANs). We also introduce new NNs that learn functionals and nonlinear operators from functions and corresponding responses for system identification. Unlike other approaches that rely on big data, here we learn from small data by exploiting the information provided by the physical conservation laws, which are used to obtain informative priors or regularize the neural networks. We will also make connections between Gauss Process Regression and NNs and discuss the new powerful concept of meta-learning. We will demonstrate the power of PINNs for several inverse problems in fluid mechanics, solid mechanics and biomedicine including wake flows, shock tube problems, material characterization, brain aneurysms, etc, where traditional methods fail due to lack of boundary and initial conditions or material properties.