Welcome to

Bob Grundy's Personal Home Page

Address: Room 121, Mathematics Institute School of Mathematical and Computational Sciences,
University of St Andrews,
St Andrews, KY16 9SS, U.K.

Telephone: Office 01334 463739

Telephone: Home 01334 472120

E-mail: reg@st-andrews.ac.uk

FAX: 01334 463748


Main Research Interests



Recent Papers

Abstract
In this paper we analyse initial-boundary value problems for partial differential equations involving nonlocal nonlinearities using two-point Hermite interpolation. The technique enables us to construct polynomial approximations which provide extensive qualitative and quantitative information regarding the structure of solutions. We proceed via a series of examples from the literature chosen for their extensive numerical and analytic provision where we can compute steady states and their stability properties, blow-up thresholds and blow-up times. The method is particularly apposite for studying feedback problems an example of which is given in the paper.
 
Abstract
In this paper we examine the general utility of multipoint Hermite type expansions in the analysis of two-point ordinary boundary value problems. We show that when the solutions are entire functions or any singularities in the complex plane are sufficiently distant from the real interval over which a solution is sought, it is possible to construct convergent Hermite expansions simply by fitting an equal number of derivatives using the end points as nodes. When singularities do influence convergence we show it is possible to change the convergence domains in the complex plane either by introducing additional nodes or by fitting different numbers of derivatives at each node. In this way convergence can often be restored and an appropriate Hermite expansion constructed. We illustrate the application of these ideas to the solution of ordinary boundary value problems via a series of linear and nonlinear examples for which a symbolic computational facility is an indispensable tool in the analysis.
 
Abstract
This paper introduces what we believe is a novel method for obtaining periodic solutions of ordinary differential equations using Hermite trigonometric polynomials. By seeking solutions in terms of such representations we show how we can derive equations relating the frequency and any parameters which are present enabling periodic solutions to be constructed. We illustrate the idea using an example which models the nonlinear pendulum. If small parameters are present we show as a useful adjunct, how we can reproduce the results of classical asymptotic analysis without recourse to the elimination of secular terms. Throughout the paper MAPLE is used as an essential aid in all the algebraic manipulations and computations that we need to do.
 
Abstract
This paper introduces what we believe is a novel method for obtaining periodic solutions of ordinary differential equations using Hermite trigonometric polynomials. By seeking solutions in terms of such representations we show how we can derive periodicity conditions relating the frequency and any parameters which are present thereby enabling the solution to be constructed. We illustrate the idea via an example which models an inverted nonlinear pendulum with a linear restoring force. If small parameters are present we show as a useful adjunct, how we can reproduce the results of classical asymptotic analysis without recourse to the elimination of secular terms. Throughout the paper MAPLE is used as an essential aid in all the algebraic manipulations and computations that we need to do.
 
Abstract
This paper extends the use of Hermite trigonometric interpolation to the computation of periodic solutions of ordinary differential equations with periodic forcing. Without loss of generality we may consider 2π-periodic solutions and construct trigonometric polynomials which fit function values and any number of derivatives at the end points of the interval [0, π] and which are thereby rendered automatically of period 2π. By seeking solutions in terms of these trigonometric polynomials we show how we can derive periodicity conditions relating the frequency, amplitude and phase of the periodic solutions to the parameters appearing in the given equation. We proceed via the example of the forced Duffing equation wherein a major aim is to demonstrate numerical convergence as the number of derivatives fitted increases. We compare our results with exact solutions when these are available or more generally by numerical confirmation. If small parameters are present we show as a useful adjunct, how we can reproduce the results of classical asymptotic analysis without recourse to the elimination of secular terms. Throughout the paper MAPLE is used as an essential aid in all the algebraic manipulations and numerical calculations that we need to do.