Tue, 31 May 2022

16:00 - 18:00
L5

'My avid fellow feeling' and 'Fleas': Playing with words on the computer

Troy Astarte
(Swansea University)
Abstract

Computers have been used to process natural language for many years. This talk considers two historical examples of computers used rather to play with human language, one well-known and the other a new archival discovery: Strachey’s 1952 love letters program, and a poetry programming competition held at Newcastle University in 1968. Strachey’s program used random number generation to pick words to fit into a template, resulting in letters of varying quality, and apparently much amusement for Strachey. The poetry competition required the entrants, mostly PhD students, to write programs whose output or source code was in some way poetic: the entries displayed remarkable ingenuity. Various analyses of Strachey’s work depict it as a parody of attitudes to love, an artistic endeavour, or as a technical exploration. In this talk I will consider how these apply to the Newcastle competition and add my own interpretations.

Fri, 06 Mar 2020

15:00 - 16:00
N3.12

Estimating the reach of a submanifold

John Harvey
(Swansea University)
Abstract

The reach is an important geometric invariant of submanifolds of Euclidean space. It is a real-valued global invariant incorporating information about the second fundamental form of the embedding and the location of the first critical point of the distance from the submanifold. In the subject of geometric inference, the reach plays a crucial role. I will give a new method of estimating the reach of a submanifold, developed jointly with Clément Berenfeld, Marc Hoffmann and Krishnan Shankar.

Fri, 14 Jun 2019

15:00 - 16:00
N3.12

Multiparameter persistence vs parametrised persistence

Jeffrey Giansiracusa
(Swansea University)
Abstract

One of the key properties of 1-parameter persistent homology is that its output can entirely encoded in a purely combinatorial way via persistence diagrams or barcodes.  However, many applications of topological data analysis naturally present themselves with more than 1 parameter. Multiparameter persistence suggests itself as the natural invariant to use, but the problem here is that the moduli space of multiparameter persistence diagrams has a much more complicated structure and we lack a combinatorial diagrammatic description.  An alternative approach was suggested by work of Giansiracusa-Moon-Lazar, where they investigated calculating a series of 1-parameter persistence diagrams as the other parameter is varied. In this talk I will discuss work in progress to produce a refinement of their perspective, making use the Algebraic Stability Theorem for persistent homology and work of Bauer-Lesnick on induced matchings.

Mon, 22 Oct 2018

14:15 - 15:15
L3

Bismut Formula for Lions Derivative of Distribution Dependent SDEs and Applications

PANPAN REN
(Swansea University)
Abstract

By using Malliavin calculus, Bismut type formulas are established for the Lions derivative of , where  0,  is a bounded measurable function,  and  solves a distribution dependent SDE with initial distribution . As applications, explicit estimates are derived for the Lions derivative and the total variational distance between distributions of   solutions with different initial data. Both degenerate and non-degenerate situations are considered. Due to the lack of the semi-group property  and the invalidity of the formula =  , essential difficulties are overcome in the study.

Joint work with Professor Feng-Yu Wang

Mon, 08 May 2017

14:15 - 15:15
L3

Characterising path-independence of Girsanov transform for stochastic differential equations

JIANG-LUN WU
(Swansea University)
Abstract

This talk will address a new link from stochastic differential equations (SDEs) to nonlinear parabolic PDEs. Starting from the necessary and sufficient condition of the path-independence of the density of Girsanov transform for SDEs, we derive characterisation by nonlinear parabolic equations of Burgers-KPZ type. Extensions to the case of SDEs on differential manifolds and the case od SDEs with jumps as well as to that of (infinite dimensional) SDEs on separable Hilbert spaces will be discussed. A perspective to stochastically deformed dynamical systems will be briefly considered.

Fri, 30 Jan 2015

14:15 - 15:15
C1

Semi-Bayesian methods under ice

Martin O'Leary
(Swansea University)
Abstract

One of the main obstacles to forecasting sea level rise over the coming centuries is the problem of predicting changes in the flow of ice sheets, and in particular their fast-flowing outlet glaciers. While numerical models of ice sheet flow exist, they are often hampered by a lack of input data, particularly concerning the bedrock topography beneath the ice. Measurements of this topography are relatively scarce, expensive to obtain, and often error-prone. In contrast, observations of surface elevations and velocities are widespread and accurate.

In an ideal world, we could combine surface observations with our understanding of ice flow to invert for the bed topography. However, this problem is ill-posed, and solutions are both unstable and non-unique. Conventionally, this problem is circumvented by the use of regularization terms in the inversion, but these are often arbitrary and the numerical methods are still somewhat unstable.

One philosophically appealing option is to apply a fully Bayesian framework to the problem. Although some success has been had in this area, the resulting distributions are extremely difficult to work with, both from an interpretive standpoint and a numerical one. In particular, certain forms of prior information, such as constraints on the bedrock slope and roughness, are extremely difficult to represent in this framework.

A more profitable avenue for exploration is a semi-Bayesian approach, whereby a classical inverse method is regularized using terms derived from a Bayesian model of the problem. This allows for the inclusion of quite sophisticated forms of prior information, while retaining the tractability of the classical inverse problem. In particular, we can account for the severely non-Gaussian error distribution of many of our measurements, which was previously impossible.

Fri, 07 Nov 2014

14:15 - 15:15
C1

Semi-Bayesian methods under ice CANCELLED (will be rescheduled)

Martin O'Leary
(Swansea University)
Abstract

One of the main obstacles to forecasting sea level rise over the coming centuries is the problem of predicting changes in the flow of ice sheets, and in particular their fast-flowing outlet glaciers. While numerical models of ice sheet flow exist, they are often hampered by a lack of input data, particularly concerning the bedrock topography beneath the ice. Measurements of this topography are relatively scarce, expensive to obtain, and often error-prone. In contrast, observations of surface elevations and velocities are widespread and accurate.

In an ideal world, we could combine surface observations with our understanding of ice flow to invert for the bed topography. However, this problem is ill-posed, and solutions are both unstable and non-unique. Conventionally, this problem is circumvented by the use of regularization terms in the inversion, but these are often arbitrary and the numerical methods are still somewhat unstable.

One philosophically appealing option is to apply a fully Bayesian framework to the problem. Although some success has been had in this area, the resulting distributions are extremely difficult to work with, both from an interpretive standpoint and a numerical one. In particular, certain forms of prior information, such as constraints on the bedrock slope and roughness, are extremely difficult to represent in this framework.

A more profitable avenue for exploration is a semi-Bayesian approach, whereby a classical inverse method is regularized using terms derived from a Bayesian model of the problem. This allows for the inclusion of quite sophisticated forms of prior information, while retaining the tractability of the classical inverse problem. In particular, we can account for the severely non-Gaussian error distribution of many of our measurements, which was previously impossible.

Mon, 23 Apr 2012

12:00 - 13:00
L3

Gauge-Strings Duality and applications

Carlos Nunez
(Swansea University)
Abstract

I will discuss some recent progress connecting different quiver gauge theories and some applications of these results.

Mon, 30 Jan 2012

17:00 - 18:00
Gibson 1st Floor SR

Universal coercivity in linear elasticity and some image processing results

Kewei Zhang
(Swansea University)
Abstract

In the first part, a variational model for composition of finitely many strongly elliptic

homogenous elastic materials in linear elasticity is considered. The notion of`universal coercivity' for the variational integrals is introduced which is independent of particular compositions of materials involved. Examples and counterexamples for universal coercivity are presented.

In the second part, some results of recent work with colleagues on image processing and feature extraction will be displayed.

Subscribe to Swansea University