Seminar

Title: The spectrum of the random-to-below Markov chain

Abstract:

The random-to-below shuffle of a deck of cards consists of removing any card randomly (with uniform probability), and inserting it anywhere below (with uniform probability). When looking at the eigenvalues of its transition matrix, they all seem to be rational and positive.

Title: Two conjectures on the spread of graphs

Abstract:

Given a graph $G$ let $\lambda_1$ and $\lambda_n$ be the maximum and minimum eigenvalues of its adjacency matrix and define the spread of $G$ to be $\lambda_1 - \lambda_n$. In this talk we discuss solutions to a pair of 20 year old conjectures of Gregory, Hershkowitz, and Kirkland regarding the spread of graphs.

Title: A primal-dual interior-point algorithm fo rnonsymmetric conic optimization

Abstract:

It is well known that primal-dual interior-point algorithms for linear optimization can easily be extended to the case of symmetric conic optimization, as shown by Nesterov and Todd (NT) in their 1997 paer about self-scaled barriers. Although many convex optimization problems can be expressed using symmetric cones then models involving for instance exponential functions do not belong to the class of symmetric conic optimization problems.

Title: In Memoriam: Tom Coleman’s Contributions to Applied Mathematics and Optimization

Description:

Thomas F. Coleman—a leader in optimization and scientific computing, professor at the University of Waterloo, and a SIAM Fellow—passed away on April 20, 2021. Tom served as the Director of the Theory Center at Cornell and then as Dean of the Faculty of Mathematics at the University of Waterloo. His research spanned continuous optimization, combinatorial scientific computing, automatic differentiation, financial optimization, mathematical software, etc. In this session, his wife and collaborator, Yuying Li, and five of his students and colleagues will describe the pioneering contributions that Tom made to these fields in his research.

Title: Asymptotic quantum state transfer using two loops with large weights

Abstract:

We study the question of asymptotic transfer strength between two nodes of a graph when large weight loop edges are located at these nodes.  It turns out that the limiting strength can be exactly computed and depends only on the extended neighborhoods of the nodes.

Title: The $k$-Independence Number

Abstract:

An independent set, also known as a stable set or coclique, in a graph is a set of vertices, no two of which are adjacent. The size of a largest independent set is called the independence number. Two classical eigenvalue bounds on the independence number, proved using eigenvalue interlacing are the Hoffman's ratio bound and the Cvetkovi\'{c}'s inertia bound.

Title: State transfer for paths with weighted loops at the end vertices

Abstract:

We consider a continuous time quantum walk on an unweighted path on n vertices, to which a loop of weight w has been added at each end vertex. Let f(t) denote the fidelity of state transfer from one end vertex to the other at time t. In this talk we give upper and lower bounds on f(t) in terms of w, n and t; further, given a > 0 we discuss the values of t for which f(t) > 1-a.

Title: Average Mixing Matrices of Trees and Stars

Abstract:

We define the average mixing matrix (AMM) of a continuous-time quantum walk on a graph using the orthogonal projections onto the eigenspaces of the adjacency matrix A. From there, one of the properties that has been studied is the rank of the AMM. This is easiest to do if the eigenvalues of A are simple, and we’ll review some of the results on this from Coutinho et. al. (2018).

Title: Arctic curves for groves

Abstract:

The limit shape phenomenon is a "law of large numbers" for random surfaces: the random surface looks macroscopically like the "average surface". The first result of this kind was the celebrated arctic circle theorem for domino tilings of the aztec diamond. The limit shape has macroscopic regions with different qualitative behavior, and the arctic curve is the boundary separating these regions.

Title: From low probability to high confidence in stochastic convex optimization

Abstract:

Standard results in stochastic convex optimization bound the number of samples that an algorithm needs to generate a point with small function value in expectation. More nuanced high probability guarantees are rare, and typically either rely on “light-tail” noise assumptions or exhibit worse sample complexity. In this work, we show that a wide class of stochastic optimization algorithms can be augmented with high confidence bounds at an overhead cost that is only logarithmic in the confidence level and polylogarithmic in the condition number.