APPLIED MATHEMATICS AND MATHEMATICAL PHYSICS SEMINAR

AMMP 2010-2011

DEPARTMENT OF MATHEMATICS IMPERIAL COLLEGE LONDON

Unless otherwise specified, the AMMP colloquium takes place on
Tuesdays at 4:10 pm in Huxley 130 (Coffee and tea at the Mathematics common room at 3:40 pm). For further information contact Greg Pavliotis at g.pavliotis@imperial.ac.uk.

Term 1

 

05.10.2010

V. JERDJEVIC (U. TORONTO)

 

Optimal Control Problems on Lie Groups and Integrable Systems

 

 

12.10.2010

S. REICH (U. POTSDAM)

 

Taming unpredictability: combining model dynamics with observations

 

Abstract: The talk is primarily dedicated to the problem of data assimilation.

Data assimilation is a technique for combining mathematical models of

physical systems with measurements in order to estimate either the state of

the system or the parameters of the model. In this talk we will view data

assimilation as a continuous deformation process of the underlying 

probability density functions (PDFs) and their approximation by empirical measures.

After a discussion of the general mathematical methodology we will discuss our approach 

in more detail for variance minimizing filters such as ensemble

Kalman filters. 

19.10.2010

G. GOTTWALD (U. Sydney)

 

Title: "A variance constraining ensemble Kalman filter: How to improve forecast using climatic data of unobserved variables."

 

Abstract:

 

Data assimilation aims to solve one of the fundamental problems of numerical weather prediction - estimating the optimal state of the atmosphere given a numerical model of the dynamics, and sparse, noisy observations of the system. A standard tool in attacking this filtering problem is the Kalman filter.

 

We consider the problem when only partial observations are available. In particular we consider the situation where the observational space consists of variables which are directly observable with known observational error, and of variables of which only their climatic variance and mean are given. We derive the corresponding Kalman filter in a variational setting.

 

We analyze the variance constraining Kalman filter (VCKF) filter for a simple linear toy model and determine its range of optimal performance. We explore the variance constraining Kalman filter in an ensemble transform setting for the Lorenz-96 system, and show that incorporating the information on the variance on some un-observable variables can improve the skill and also increase the stability of the data assimilation procedure.

 

Using methods from dynamical systems theory we then systems where the un-observed variables evolve deterministically but chaotically on a fast time scale.

 

This is joint work with Lewis Mitchell and Sebastian Reich

 

26.10.2010

D. HENRY (DUBLIN CITY UNIVERSITY)

 

Title: "Infinite propagation speed for a two component Camassa-Holm equation."

 

Abstract: In this talk we will examine the rate of propagation of a class of solutions of an integrable two component generalisation of the celebrated Camassa-Holm equation. We examine the propagation behaviour of compactly supported solutions, namely whether solutions which are initially compactly supported will retain this property throughout their time of evolution. In the negative case, where we show that solutions have an infinite speed of propagation, we present a description of how the solutions retain weaker properties throughout their existence time, namely they decay at an exponentially fast rate for the duration of their existence.

 

 

02.11.2010

B. GODDARD (IC)

 

Non-adiabatic transitions through avoided crossings.

 

Abstract

The photo-dissociation of diatomic molecules is one of the paradigmatic chemical reactions of quantum chemistry.  The associated mathematical problem is the study of non-adiabatic transitions at avoided crossings in a two-level system, with one effective spatial degree of freedom.  Such problems are highly multi-scale; the transmitted wavefunction is typically very small.  This leads to great difficulty in performing accurate numerical calculations, and an alternative method is required.   Using superadiabatic representations (which will be defined), and an approximation of the dynamics near the crossing region, we obtain an explicit formula for the transmitted wavefunction. Our results agree extremely well with high precision ab-initio calculations.

09.11.2010

A.M. STUART (U. WARWICK)

 

Connections Bewtween (S)PDEs and MCMC in a Hilbert Space

 

I will show that a range of inverse problems, when formulated using the Bayesian

approach, lead to the problem of sampling a probability measure on a Banach space.

When the prior is a Gaussian random field, then the posterior measure is

defined via its density wrt the Gaussian. I will discuss properties of the posterior

measure and show how these properties can be used to design efficient

MCMC-based sampling algorithms which are robust under refinement of finite

dimensional approximation of the Banach space. In particular I will demonstrate

the central role of stochastic PDEs in the construction of these methods.

16.11.2010

K. MATTHIES (U. BATH)

 

Hard-sphere dynamics and Boltzmann equations

 

 

abstract:

 

The derivation of the continuum models from deterministic atomistic

descriptions is a longstanding and fundamental challenge. In particular

the emergence of irreversible  macroscopic evolution from reversible

deterministic microscopic evolution is still not fully understood. We

study a classic system: N balls that interact with each other via a

hard-core potential and show rigorously that in the case of kinetic

annihilation (particles annihilate each other upon collision) the

asymptotic behavior as N tends to infinity is correctly described by the

Boltzmann equation without gain-term for non-concentrated initial

distributions. The mean-field description fails, when there are

concentrations in the space or the velocity coordinates. Extension

towards the full collision case are outlined. Joint work with Florian

Theil.

23.11.2010

S. WHITTINGTON (U. TORONTO)

 

Title: Counting knotted curves and surfaces in lattices

 

Simple closed curves can be knotted in 3-space and these knots

can be seen experimentally in circular DNA molecules, where the entanglements affect cellular processes.  It has  been known for over twenty years that the knot probability

goes to unity as the size increases.  It is natural to ask what happens in higher dimensions.  2-spheres can be knotted in 4-space and one might ask if the knot probability goes to unity as the  area of the 2-sphere increases.  This seminar will first review the situation in three dimensions, then discuss why the higher dimensional case is more difficult and why the lower dimensional argument fails in 4-space.  Finally we shall discuss some results in this direction where the 2-sphere is confined to a tube in the 4-dimensional hypercubic lattice.

 

 

 

 

 

 

30.11.2010

R. CRASTER (IC)

 

Title: High frequency homogenization

 

Abstract: It is highly desirable to be able to create continuum equations that embed a known microstructure through effective

or averaged quantities such as wavespeeds or shear moduli. The methodology for achieving this at low frequencies and for waves long relative to the microstructure is well-known and such static or quasi-static theories are well developed. However, at high frequencies the multiple scattering by the elements of the microstructure, which is now of a similar scale to the wavelength,

 has apparently prohibited any homogenization theory. Recently we have developed an asymptotic approach that overcomes this limitation and continuum equations are developed even though the microstructure and wavelength are of the same order.

 The general theory will be described and applications to continuum, discrete and frame lattice structures will be outlined. The results and  methodology are confirmed versus various illustrative exact/ numerical calculations. 

 

07.12.2010

Ch. ORTNER (OXFORD U.)

 

Atomistic/Continuum Hybrid Methods: Construction, Ghost Forces, Consistency

 

Low energy equilibria of crystalline materials are typically characterised by localized defects that interact with their environment through long-range elastic fields. By coupling atomistic models of the defects with continuum models for the elastic far field one can, in principle, obtain models with near-atomistic accuracy at significantly reduced computational cost. However, several pitfalls need to be overcome to find a reliable coupling mechanism. Possibly the most widely discussed among these pitfalls are the so-called Òghost forcesÓ that typically arise in energy-based atomistic/continuum coupling mechanisms.

In this talk I will first describe the construction and of energy-based atomistic/continuum coupling methods with and without ghost forces. I will then explain the resulting model errors due to different types of interface treatment. To this end I will first review some results on one-dimensional model problems and then describe some recent results in higher dimensions. I will show that, in 1D, absence of ghost forces does imply Òhigh accuracyÓ of the coupling scheme but that in 2D/3D this is not clear at this point.

 

14.12.2010

V. SMYSHLYAEV (UCL and U. Bath)

 

Title: TBA