L3

Networks form the backbones of a wide variety of complex systems,
ranging from food webs, gene regulation and social networks to
transportation networks and the internet. Due to the sheer size and
complexity of many of theses systems, it remains an open challenge to
formulate general descriptions of their large-scale structures.
Although many methods have been proposed to achieve this, many of them
yield diverging descriptions of the same network, making both the
comparison and understanding of their results very
difficult. Furthermore, very few methods attempt to gauge the
statistical significance of the uncovered structures, and hence the
majority cannot reliably separate actual structure from stochastic
fluctuations.  In this talk, I will show how these issues can be tackled
in a principled fashion by formulating appropriate generative models of
network structure that can have their parameters inferred from data. I
will also consider the comparison between a variety of generative
models, including different structural features such as degree
correction, where nodes with arbitrary degrees can belong to the same
group, and community overlap, where nodes are allowed to belong to more
than one group. Because such model variants possess an increased number
of parameters, they become prone to overfitting. We demonstrate how
model selection based on the minimum description length criterion and
posterior odds ratios can fully account for the increased degrees of
freedom of the larger models, and selects the most appropriate trade-off
between model complexity and quality of fit based on the statistical
evidence present in the data.

Throughout the talk I will illustrate the application of the methods
with many empirical networks such as the internet at the autonomous
systems level, the global airport network, the network of actors and
films, social networks, citations among websites, co-occurrence of
disease-causing genes and many others.
 

Sudden cardiac arrhythmic death remains a major health challenge in Western Society. Recent advances in computational methods and technologies have made clinically-based cardiac modelling a reality. An important current focus is the use of modelling to understand the nature of arrhythmias in the setting of different forms of structural heart disease. However, many challenges remain regarding the best use of these models to inform clinical decision making and guide therapies. In this talk, I will introduce a diverse sample of applications of modelling in this context, ranging from basic science studies which aim to leverage a fundamental mechanistic understanding of different aspects of pathological cardiac function, to direct clinical-application projects which aim to use modelling to immediately inform a clinical therapy. I will also discuss the challenges involved in clinically-driven modelling, and how to both engage, and manage, the expectations of clinicians at the same time, particularly with respect to the potential uses of 'patient-specific' modelling.

I this talk I will try to give an overview of recent progress in
spatial stochastic modeling within the reaction-diffusion
framework. While much of the initial motivation for this work came
from problems in cell biology, I will also highlight some examples
from epidemics and neuroscience.

As a motivating introduction, some newly discovered properties of
optimal controls in stochastic enzymatic reaction networks will be
presented. I will next detail how diffusive and subdiffusive reactive
processes in realistic geometries at the cellular scale may be modeled
mesoscopically. Along the way, some different means by which these
models may be analyzed with respect to consistency and convergence
will also be discussed. These analytical techniques hint at how
effective (i.e. parallel) numerical implementations can be
designed. Large-scale simulations will serve as illustrations.

I present a new method for optimizing quadratic functions subject to simple bound constraints.  If the problem happens to be strictly convex, the algorithm reduces to a highly efficient method by Dostal and Schoberl.  Our algorithm, however, is also able to efficiently solve nonconcex problems. During this talk I will present the algorithm, a sketch of the convergence theory, and numerical results for convex and nonconvex problems.

We will begin by discussing the risk parity portfolio selection problem, which aims to find  portfolios for which the contributions of risk from all assets are equally weighted. The risk parity may be satisfied over either individual assets or groups of assets. We show how convex optimization techniques can find a risk parity solution in the nonnegative  orthant, however, in general cases the number of such solutions can be anywhere between zero and  exponential in the dimension. We then propose a nonconvex least-squares formulation which allows us to consider and possibly solve the general case. 

Motivated by this problem we present several alternating direction schemes for specially structured nonlinear nonconvex problems. The problem structure allows convenient 2-block variable splitting.  Our methods rely on solving convex subproblems at each iteration and converge to a local stationary point. Specifically, discuss approach  alternating directions method of multipliers and the alternating linearization method and we provide convergence rate results for both classes of methods. Moreover, global optimization techniques from polynomial optimization literature are applied to complement our local methods and to provide lower bounds.

A meromorphic connection on a complex curve can be interpreted as a representation of a simple Lie algebroid.  By integrating this Lie algebroid to a Lie groupoid, one obtains a complex surface on which the parallel transport of the connection is globally well-defined and holomorphic, despite the apparent singularities of the corresponding differential equations.  I will describe these groupoids and explain how they can be used to illuminate various aspects of the classical theory of singular ODEs, such as the resummation of divergent series solutions.  (This talk is based on joint work with Marco Gualtieri and Songhao Li.)

The first FFT-based algorithm for numerical homogenization from high-resolution images was proposed by Moulinec and Suquet in 1994 as an alternative to finite elements and twenty years later, it is still widely used in computational micromechanics of materials. The method is based on an iterative solution to an integral equation of the Lippmann-Schwinger type, whose kernel can be explicitly expressed in the Fourier domain. Only recently, it has been recognized that the algorithm has a variational structure arising from a Fourier-Galerkin method. In this talk, I will show how this insight can be used to significantly improve the performance of the original Moulinec-Suquet solver. In particular, I will focus on (i) influence of an iterative solver used to solve the system of linear algebraic equations, (ii) effects of numerical integration of the Galerkin weak form, and (iii) convergence of an a-posteriori bound on the solution during iterations.