I will present a gentle introduction to the theory of conformal dimension, focusing on its applications to the boundaries of hyperbolic groups, and the difficulty of classifying groups whose boundaries have conformal dimension 1.

I will introduce a McKean—Vlasov problem arising from a simple mean-field model of interacting neurons. The equation is nonlinear and captures the positive feedback effect of neurons spiking. This leads to a phase transition in the regularity of the solution: if the interaction is too strong, then the system exhibits blow-up. We will cover the mathematical challenges in defining, constructing and proving uniqueness of solutions, as well as explaining the connection to PDEs, integral equations and mathematical finance.

In this talk we introduce a novel dynamic programming (DP) approximation that exploits the inherent network structure present in revenue management problems. In particular, our approximation provides a new lower bound on the value function for the DP, which enables conservative revenue forecasts to be made. Existing state of the art approximations of the revenue management DP neglect the network structure, apportioning the prices of each product, whereas our proposed method does not: we partition the network of products into clusters by apportioning the capacities of resources. Our proposed approach allows, in principle, for better approximations of the DP to be made than the decomposition methods currently implemented in industry and we see it as an important stepping stone towards better approximate DP methods in practice.

One of the key challenges in revenue management is unconstraining demand data. Existing state of the art single-class unconstraining methods make restrictive assumptions about the form of the underlying demand and can perform poorly when applied to data which breaks these assumptions. In this talk, we propose a novel unconstraining method that uses Gaussian process (GP) regression. We develop a novel GP model by constructing and implementing a new non-stationary covariance function for the GP which enables it to learn and extrapolate the underlying demand trend. We show that this method can cope with important features of realistic demand data, including nonlinear demand trends, variations in total demand, lengthy periods of constraining, non-exponential inter-arrival times, and discontinuities/changepoints in demand data. In all such circumstances, our results indicate that GPs outperform existing single-class unconstraining methods.

I will present a new approach to study the RG flow in 3d N=4 gauge theories, based on an analysis of the Coulomb branch of vacua. The Coulomb branch is described as a complex algebraic variety and important information about the strongly coupled fixed points of the theory can be extracted from the study of its singularities. I will use this framework to study the fixed points of U(N) and Sp(N) gauge theories with fundamental matter, revealing some surprising scenarios at low amount of matter.

Lein Applied Diagnostics has a novel optical measurement technique that is used to measure various parameters in the body for medical applications.

Two particular areas of interest are non-invasive glucose measurement for diabetes care and the diagnosis of diabetes. Both measurements are based on the eye and involve collecting complex data sets and modelling their links to the desired parameter.

If we take non-invasive glucose measurement as an example, we have two data sets – that from the eye and the gold standard blood glucose reading. The goal is to take the eye data and create a model that enables the calculation of the glucose level from just that eye data (and a calibration parameter for the individual). The eye data consists of measurements of apparent corneal thickness, anterior chamber depth, optical axis orientation; all things that are altered by the change in refractive index caused by a change in glucose level. So, they all correlate with changes in glucose as required but there are also noise factors as these parameters also change with alignment to the meter etc. The goal is to get to a model that gives us the information we need but also uses the additional parameter data to discount the noise features and thereby improve the accuracy.