Karin Baur (University of Leeds)
Cluster algebras and cluster categories via surfaces abstract
Abstract:
xTwenty years of research in cluster theory have established deep links between cluster algebras, surface geometry and representation theory. Cluster structures can be defined using surface geometry, with curves corresponding to cluster variables and to rigid objects in associated categories. The classical Ptolemy relations give rise to exchange phenomena or mutation in the associated cluster structures.In these lectures I will focus on the geometric approach to cluster algebras and cluster categories. Topics discussed will include cluster structures on the Grassmannian, friezes, laminations, Postnikov diagrams, dimer models, root combinatorics.
10:15 • online
Michael Krivelevich (Tel Aviv University)
Random Graphs abstract
Abstract:
Random graphs is a generic name for discrete probability spaces, whose ground sets are composed of graphs, or more generally of discrete structures. Since its inception by Erdős and Rényi some sixty years ago, random graphs have grown to be one of the key disciplines in modern Combinatorics, at the same time serving as an indispensable tool for other branches of Combinatorics and Computer Science.The course will serve as an introduction to random graphs, covering classical topics as well as addressing recent developments and tools.Course syllabus (tentative): Models of random graphs and of random graph processes; illustrative examples; random regular graphs, configuration model; small subgraphs; long paths and Hamiltonicity; hitting time results; coloring problems; extremal problems in random graphs; pseudo-random graphs.Desirable background: Working knowledge of Graph Theory, familiarity with basic notions of Probability and Linear Algebra.
14:15 • Zoom
Gunter Malle (Universität Kaiserslautern)
Overgroups of regular unipotent elements in simple algebraic groups abstract
Abstract:
What are the subgroups of simple algebraic groups that contain a regular unipotent element? This question has a long history and different aspects of it have been considered by various authors. In our talk we intend to first give an introduction to this problem and then to present our recent extension of a result of Testerman-Zalesski to non-connected positive-dimensional subgroups. It differs from the connected case in that interesting new examples do arise. This is joint work with Donna Testerman.
15:00 • EPF Lausanne, Zoom Seminar
Dr. Woonam Lim (ETH Zürich)
Virtual invariants of Quot schemes of surfaces III abstract
Abstract:
Let S be a smooth projective surface. Grothendieck’s Quot scheme on S admits a 2-term perfect obstruction theory when it parametrizes torsion quotients. This allows us to define various virtual invariants of Quot schemes, including homological (K-theoretic) descendent invariants and virtual Segre/Verlinde numbers. The study of these invariants is partially motivated by the parallel theory of moduli of sheaves. A special feature of the Quot scheme theory is the conjectural rationality of the homological (K-theoretic) descendent series. We explain how this can be proven for all surfaces with p_g>0 using the multiplicative structural formula involving Seiberg-Witten invariants. The same method applies to the study of the virtual Segre/Verlinde series. We also explain the virtual Segre/Verlinde correspondence and a special symmetry for punctual Quot schemes which is reminiscent of the numerical strange duality.
15:00 • Zoom
Onlline Seminar
Dr. Stefan Tohaneanu (University of Idaho)
Linear codes with prescribed projective codewords of minimum weight abstract
Abstract:<button>Video<i class="fa fa-play-circle"></button> (**This eSeminar will take place on Zoom, using the same meeting details as previous seminars. If you do not have meeting details, please contact simran.tinani@math.uzh.ch **)
Consider $C$, an $[n,k,d]$-linear code. Every projective codeword of minimum weight $d$ corresponds to a point in $\\mathbb P^{k-1}$, and there are strong connections between the algebraic and geometric properties of these points and the parameters of $C$, especially with the minimum distance $d$. The most non-trivial connection is the fact that the Castelnuovo-Mumford regularity of the coordinate ring of these points is a lower bound for $d$. Conversely, given a finite set of points $X$ in $\\mathbb P^{k-1}$, it is possible to construct linear codes with projective codewords of minimum weight corresponding to $X$. We will discuss about these constructions, and we will also look at the particular case when the constructed linear code has minimum distance equal to the regularity.
16:00 • Onlline Seminar
Dr. Anirbit Mukherjee (University of Pennsylvania, USA)
Some Recent Progresses in the Mathematics of Neural Training abstract
Abstract:
One of the most intriguing mathematical mysteries of our times is to be able to explain the phenomenon of deep-learning. Neural nets can be made to paint while imitating classical art styles or play chess better than any machine or human ever and they seem to be the closest we have ever come to achieving "artificial intelligence". But trying to reason about these successes quickly lands us into a plethora of extremely challenging mathematical questions - typically about discrete stochastic processes. Some of these questions remain unsolved for even the smallest neural nets! In this talk we will describe two of the most recent themes of our work in this direction. Firstly, we will explain how under mild distributional conditions we can construct iterative algorithms which can train a ReLU gate in the realizable setting in linear time while also keeping track of mini-batching. We will show how this algorithm does approximate training when there is a data-poisoning attack on the training labels. Such convergence proofs remain unknown for S.G.D, but we will show via experiments that our algorithm very closely mimics the behaviour of S.G.D. Lastly, we will review this very new concept of "local elasticity" of a learning process and demonstrate how it appears to reveal certain universal phase changes during neural training. Then we will introduce a mathematical model which reproduces some of these key properties in a semi-analytic way. We will end by delineating various open questions in this theme of macroscopic phenomenology with neural nets. This is joint work with Weijie Su (Wharton, Statistics), Sayar Karmakar (U Florida, Statistics) and Phani Deep (Amazon, India)