Friday, December 2, 2016 - 14:00 for 1 hour (actually 50 minutes)
Location
Skiles 006
Speaker
S. Suzuki – RIMS, Kyoto University
The Drinfeld double of a finite dimensional Hopf algebra is a
quasi-triangular Hopf algebra with the canonical element as the universal R
matrix, and we obtain a ribbon Hopf algebra by adding the ribbon element.
The universal quantum invariant is an invariant of framed links, and is
constructed diagrammatically using a ribbon Hopf algebra. In that
construction, a copy of the universal R matrix is attached to each positive
crossing, and invariance under the Reidemeister III move is shown by the
quantum Yang-Baxter equation of the universal R matrix.
On the other hand, R. Kashaev showed that the Heisenberg double has the
canonical element (the universal S matrix) satisfying the pentagon
relation. In this talk we reconstruct the universal quantum invariant using
Heisenberg double, and extend it to an invariant of colored ideal
triangulations of the complement. In this construction, a copy of the
universal S matrix is attached to each tetrahedron and the invariance under
the colored Pachner (2,3) move is shown by the pentagon equation of the
universal S matrix
Friday, December 2, 2016 - 13:05 for 1 hour (actually 50 minutes)
Location
Skiles 006
Speaker
Daniel Zink – Georgia Tech
Conditional gradient algorithms (also often called Frank-Wolfe algorithms) are popular due to their simplicity of only requiring a linear optimization oracle and more recently they also gained significant traction for online learning. While simple in principle, in many cases the actual implementation of the linear optimization oracle is costly. We show a general method to lazify various conditional gradient algorithms, which in actual computations leads to several orders of magnitude of speedup in wall-clock time. This is achieved by using a faster separation oracle instead of a linear optimization oracle, relying only on few linear optimization oracle calls.
Asymptotic equivalence between two statistical models means that they have the same asymptotic (large sample) properties with respect to all decision problems with bounded loss. In nonparametric (infinite-dimensional) statistical models, asymptotic equivalence has been found to be useful since it can allow one to derive certain results by studying simpler models. One of the key results in this area is Nussbaum’s theorem, which states that nonparametric density estimation is asymptotically equivalent to a Gaussian shift model, provided that the densities are smooth enough and uniformly bounded away from zero.We will review the notion of asymptotic equivalence and existing results, before presenting recent work on the extent to which one can relax the assumption of being bounded away from zero. We further derive the optimal (Le Cam) distance between these models, which quantifies how close they are for finite-samples. As an application, we also consider Poisson intensity estimation with low count data. This is joint work with Johannes Schmidt-Hieber.
The probabilistic method, pioneered by P. Erdös, has been key in many proofs from asymptotic geometric analysis. This method allows one to take advantage of numerous tools and concepts from probability theory to prove theorems which are not necessarily a-priori related to probability. The objective of this talk is to demonstrate several recent results which take advantage of stochastic calculus to prove results of a geometric nature. We will mainly focus on a specific construction of a moment-generating process, which can be thought of as a stochastic version of the logarithmic Laplace transform. The method we introduce allows us to attain a different viewpoint on the method of semigroup proofs, namely a path-wise point of view. We will first discuss an application of this method to concentration inequalities on high dimensional convex sets. Then, we will briefly discuss an application to two new functional inequalities on Gaussian space; an L1 version of hypercontractivity of the convolution operator related to a conjecture of Talagrand (joint with J. Lee) and a robustness estimate for the Gaussian noise-stability inequality of C.Borell (improving a result of Mossel and Neeman).
The Homfly skein algebra of a surface is defined using links in
thickened surfaces modulo local "skein" relations. It was shown by
Turaev that this quantizes the Goldman symplectic structure on the
character varieties of the surface. In this talk we give a complete
description of this algebra for the torus. We also show it is
isomorphic to the elliptic Hall algebra of Burban and Schiffmann,
which is an algebra whose elements are (formal sums of) sheaves on an
elliptic curve, with multiplication defined by counting extensions of
such sheaves. (Joint work with H. Morton.)
Wednesday, November 30, 2016 - 12:00 for 1 hour (actually 50 minutes)
Location
Skiles 006
Speaker
Christian Houdré – Georgia Institute of Technology
I will start with a brief presentation of the Probability activities in SOM. I will continue by presenting results obtained in SOM, over the past ten years, answering long standing questions insequences comparison.
Wednesday, November 30, 2016 - 11:00 for 1 hour (actually 50 minutes)
Location
Skiles 005
Speaker
Prof. Eugene Wayne – Boston University
The nonlinear Schroedinger
equation (NLS) can be derived as a formal approximating equation for the
evolution of wave packets in a wide array of nonlinear dispersive PDE’s
including the propagation of waves on the surface of an inviscid
fluid. In
this talk I will describe recent work that justifies this approximation
by exploiting analogies with the theory of normal forms for ordinary
differential equations.