University of Bristol

Jakobshavn Isbrae and many other fast flowing outlet glaciers of present

and past ice sheets lie in deep troughs which often have several

overdeepened sections. To make their fast flow possible their bed needs

to be slippery which in turn means high basal water pressures. I will

present a model of subglacial water flow and its application to

Jakobshavn. I find that, somewhat surprisingly, the reason for

Jakobshavn's fast flow might be the pressure dependence of the melting

point of ice. The model itself describes the unusual fluid dynamics occurring underneath the ice; it has an interesting mathematical structure that presents computational challenges.

Turbidity currents - submarine flows of sediment - are capable of transporting particulate material over large distance. However direct observations of them are extremely rare and much is inferred from the deposits they leave behind, even though the characteristics of their source are often not known. The submarine flows of volcanic ash from the Soufriere Hills Volcano, Monsterrat provide a unique opportunity to study a particle-driven flow and the deposit it forms, because the details of the source are relatively well constrained and through ocean drilling, the deposit is well sampled.

We have formed simple mathematical models of this motion that capture ash transport and deposit. Our description brings out two dynamical features that strongly influence the motion and which have previously often been neglected, namely mixing between the particulate flow and the oceanic water and the distribution of sizes suspended by the flow. We show how, in even simple situations, these processes alter our views of how these currents propagate.

We present a variational model for the quasi-static evolution of brutal brittle damage for geometrically-linear elastic materials. We

allow for multiple damaged states. Moreover, unlike current formulations, the materials are allowed to be anisotropic and the

deformations are not restricted to anti-plane shear. The model can be formulated either energetically or through a strain threshold. We

explore the relationship between these formulations. This is joint work with Christopher Larsen, Worcester Polytechnic Institute.

We present some recent results concerning domain wall motion in one-dimensional nanowires, including the existence, velocity and stability of travelling-wave and precessing solutions.  We consider the case of unixial anisotropy, characteristic of wires with symmetrical (e.g., circular) cross-section, as opposed to strongly anisotropic geometries (films and strips) that have received greater attention.  This is joint work with Arseni Goussev and Valeriy Slastikov.

A central challenge in socio-physics is understanding how groups of self-interested agents make collective decisions. For humans many insights in the underlying opinion formation process have been gained from network models, which represent agents as nodes and social contacts as links. Over the past decade these models have been expanded

to include the feedback of the opinions held by agents on the structure of the network. While a verification of these adaptive models in humans is still difficult, evidence is now starting to appear in opinion formation experiments with animals, where the choice that is being made concerns the direction of movement. In this talk I show how analytical insights can be gained from adaptive networks models and how predictions from these models can be verified in experiments with swarming animals. The results of this work point to a similarity between swarming and human opinion formation and reveal insights in the dynamics of the opinion formation process. In particular I show that in a population that is under control of a strongly opinionated minority a democratic consensus can be restored by the addition of

uninformed individuals.

Support Vector Machines are a new and very promising approach to

machine learning. They can be applied to a wide range of tasks such as

classification, regression, novelty detection, density estimation,

etc. The approach is motivated by statistical learning theory and the

algorithms have performed well in practice on important applications

such as handwritten character recognition (where they currently give

state-of-the-art performance), bioinformatics and machine vision. The

learning task typically involves optimisation theory (linear, quadratic

and general nonlinear programming, depending on the algorithm used).

In fact, the approach has stimulated new questions in optimisation

theory, principally concerned with the issue of how to handle problems

with a large numbers of variables. In the first part of the talk I will

overview this subject, in the second part I will describe some of the

speaker's contributions to this subject (principally, novelty

detection, query learning and new algorithms) and in the third part I

will outline future directions and new questions stimulated by this

research.