Past Colloquia

8 May 2009
16:30
Professor Steven N. Evans
Abstract
<span lang="EN-GB"> <p> A common question in evolutionary biology is whether evolutionary processes leave some sort of signature in the shape of the phylogenetic tree of a collection of present day species. </p> <p> Similarly, computer scientists wonder if the current structure of a network that has grown over time reveals something about the dynamics of that growth. </p> <p> Motivated by such questions, it is natural to seek to construct``statistics'' that somehow summarise the shape of trees and more general graphs, and to determine the behaviour of these quantities when the graphs are generated by specific mechanisms. </p> <p> The eigenvalues of the adjacency and Laplacian matrices of a graph are obvious candidates for such descriptors. </p> <p> I will discuss how relatively simple techniques from linear algebra and probability may be used to understand the eigenvalues of a very broad class of large random trees. These methods differ from those that have been used thusfar to study other classes of large random matrices such as those appearing in compact Lie groups, operator algebras, physics, number theory, and communications engineering. </p> <p> This is joint work with Shankar Bhamidi (U. of British Columbia) and Arnab Sen (U.C. Berkeley). </p> <p> &nbsp; </p> </span>
6 March 2009
16:30
Professor Bao Chau Ngo
Abstract
Coefficients of the characteristic polynomial are generators of the ring of polynomial functions on the space of matrices which are invariant under the conjugation. This was generalized by Chevalley to general reductive groups. By looking closely on the centralisers, one is lead to a very natural 2-category attached to Chevalley characteristic morphism. This abstract, but yet elementary, construction helps one to understand the symmetries of the fibres of the Hitchin fibration, as well as those of affine Springer fibers.<br /> <br /> We will also explain how these groups of symmetries are related to the notion of endoscopic groups, which was introduced by Langlands in his stabilisation of the trace formula. We will also briefly explain how the symmetry groups help one to acquire a rather good understanding of the cohomology of the Hitchin fibration and eventually the proof of the fundamental lemma in Langlands' program.
6 February 2009
16:30
Professor Ivar Ekeland
Abstract
In classical economic theory, one discounts future gains or losses at a constant rate: one pound in t years is worth exp(-rt) pounds today. There are now very good reasons to consider non-constant discount rates. This gives rise to a problem of time-inconsistency: a policy which is optimal today will no longer be optimal tomorrow. The concept of optimality then no longer is useful. We introduce instead a concept of equilibrium solution, and characterize it by a non-local variant of the Hamilton-Jacobi equation. We then solve the classical Ramsey model of endogenous growth in this framework, using the central manifold theorem<br /> <br />
7 November 2008
16:30
Abstract
A random environment (in Z^d) is a collection of (random) transition probabilities, indexed by sites. Perform now a random walk using these transitions. This model is easy to describe, yet presents significant challenges to analysis. In particular, even elementary questions concerning long term behavior, such as the existence of a law of large numbers, are open. I will review in this talk the model, its history, and recent advance, focusing on examples of unexpected behavior.
6 June 2008
16:30
Prof. Michael Harris
Abstract
Let E be an elliptic curve defined by a cubic equation with rational coefficients. <br />The Sato-Tate Conjecture is a statistical assertion about the variation of the number of points of E over finite fields. I review some of the main steps in my proof of this conjecture with Clozel, Shepherd-Barron, and Taylor, in the case when E has non-integral j-invariant. Emphasis will be placed on the steps involving moduli spaces of certain Calabi-Yau hypersurfaces with level structure.<br /><br />If one admits a version of the stable trace formula that should soon be available, the same techniques imply that, when E and E' are two elliptic curves that are not isogenous, then the numbers of their points over finite fields are statistically independent. For reasons that have everything to do with the current limits to our understanding of the Langlands program, the analogous conjectures for three or more non-isogenous elliptic curves are entirely out of reach.<br /><br />
9 May 2008
16:30
Hans G. Othmer
Abstract
New techniques in cell and molecular biology have produced huge advances in our understanding of signal transduction and cellular response in many systems, and this has led to better cell-level models for problems ranging from biofilm formation to embryonic development. However, many problems involve very large numbers of cells, and detailed cell-based descriptions are computationally prohibitive at present. Thus rational techniques for incorporating cell-level knowledge into macroscopic equations are needed for these problems. In this talk we discuss several examples that arise in the context of cell motility and pattern formation. We will discuss systems in which the micro-to-macro transition can be made more or less completely, and also describe other systems that will require new insights and techniques.
29 February 2008
15:30
Professor Etienne Ghys
Abstract
A lattice in the plane is a discrete subgroup in R^2 isomorphic to Z^2 ; it is unimodular if the area of the quotient is 1. The space of unimodular lattices is a venerable object in mathematics related to topology, dynamics and number theory. In this talk, I'd like to present a guided tour of this space, focusing on its topological aspect. I will describe in particular the periodic orbits of the modular flow, giving rise to beautiful "modular knots". I will show some animations
2 November 2007
15:30
Prof Ari Laptev
Abstract
We shall begin with simple Weyl type asymptotic formulae for the spectrum of Dirichlet Laplacians and eventually prove a new result which I have recently obtained, jointly with J. Dolbeault and M. Loss. Following Eden and Foias, we derive a matrix version of a generalised Sobolev inequality in one dimension. This allows us to improve on the known estimates of best constants in Lieb-Thirring inequalities for the sum of the negative eigenvalues for multi-dimensional Schrödinger operators. Bio: Ari Laptev received his PhD in Mathematics from Leningrad University (LU) in 1978, under the supervision of Michael Solomyak. He is well known for his contributions to the Spectral Theory of Differential Operators. Between 1972 - 77 and 1977- 82 he was employed as a junior researcher and as Assistant Professor at the Mathematics & Mechanics Department of LU. In 1981- 82 he held a post-doc position at the University of Stockholm and in 1982 he lost his position at LU due to his marriage to a British subject. Up until his emigration to England in 1987 he was working as a builder, constructing houses in small villages in the Novgorod district of Russia. In 1987 he was employed in Sweden, first as a lecturer at Linköping University and then from 1992 at the Royal Institute of Technology (KTH). In 1999 he became a professor at KTH and also Vice Chairman of its Mathematics Department. In 1992 he was granted Swedish citizenship. Ari Laptev was the President of the Swedish Mathematical Society from 2001 to 2003 and the President of the Organizing Committee of the Fourth European Congress of Mathematics in Stockholm in 2004. From January 2007 he has been employed by Imperial College London. Ari Laptev has supervised twelve PhD students. From January 2007 until the end of 2010 he is President of the European Mathematical Society.
19 October 2007
16:30
Professor John Cardy
Abstract
Random planar curves arise in a natural way in statistical mechanics, for example as the boundaries of clusters in critical percolation or the Ising model. There has been a great deal of mathematical activity in recent years in understanding the measure on these curves in the scaling limit, under the name of Schramm-Loewner Evolution (SLE) and its extensions. On the other hand, the scaling limit of these lattice models is also believed to be described, in a certain sense, by conformal field theory (CFT). In this talk, after an introduction to these two sets of ideas, I will give a theoretical physicist's viewpoint on possible direct connections between them. John Cardy studied Mathematics at Cambridge. After some time at CERN, Geneva he joined the physics faculty at Santa Barbara. He moved to Oxford in 1993 where he is a Senior Research Fellow at All Souls College and a Professor of Physics. From 2002-2003 and 2004-2005 he was a member of the IAS, Princeton. Among other work on the applications of quantum field theory, in the 1980s he helped develop the methods of conformal field theory. Professor Cardy is a Fellow of the Royal Society, a recipient of the 2000 Paul Dirac Medal and Prize of the Institute of Physics, and of the 2004 Lars Onsager Prize of the American Physical Society "for his profound and original applications of conformal invariance to the bulk and boundary properties of two-dimensional statistical systems."

Pages