A singular perturbation method

Fowkes, N. D. (Neville D.) (1966). A singular perturbation method PhD Thesis, School of Physical Sciences, The University of Queensland.

Attached Files (Some files may be inaccessible until you login with your UQ eSpace credentials)
Name Description MIMEType Size Downloads
THE617.pdf Full text application/pdf 3.95MB 4
Author Fowkes, N. D. (Neville D.)
Thesis Title A singular perturbation method
School, Centre or Institute School of Physical Sciences
Institution The University of Queensland
Publication date 1966
Thesis type PhD Thesis
Supervisor J. J, Mahony
Total pages 110
Language eng
Subjects 010110 Partial Differential Equations
Formatted abstract
In a paper entitled "An Expansion Method for Singular Perturbation Problems" [1 ], Mahony proposed a method for producing uniformly valid, asymptotic expansions of solutions of singular perturbation problems. This method was applied to a particular problem, and in this case worked well. In order to obtain an assessment of tho method, its application to a range of problems has been considered in this thesis. In addition, certain clarifications of mathematical difficulties noted by Mahony, but not resolved by him, have been achieved.

In order to deal with a singular point* (or line, or surface) in a solution domain, one is usually forced into introducing some type of stretched co-ordinate description of the problem close to this point. This description usually ceases to be of use away from the singular point because the stretched co-ordinate system is inappropriate away from this point. In the approach used by Mahony, however, a stretched variable "the boundary layer function" is introduced which is appropriate to the situation close to, and away from, the singular point. This 'boundary layer function" is introduced firstly as an unknown, and a solution is sought as a function of the original independent variables and this boundary layer function. The freedom of mathematical form thus introduced can be utilized to extend the range of validity of the solution away from the singular point. Further boundary layer functions can be introduced if there is more than one singular point until the complete solution domain is covered.

In Part I of this thesis, the method has been applied to the second order, linear, ordinary, differential equation

                                              E3y" - g(x,E)y = 0 ,

where E is a small positive parameter. In recent years this equation has been investigated by many authors and several of the simpler cases have been solved. The methods employed by these authors, however, are not readily extensible to apply to other than ordinary, linear, differential equations. It was found that standard results can be obtained in the cases considered, with far less labour, using the Mahony approach. Some problems associated with the above equation, which have not previously been tackled, have also been considered. The difficulties associated with nonuniqueness, which troubled Mahony, are clarified by these examples.

In Part II, the method has been applied to the partial differential equation

                                             E3V2ψ - g(x) ψ= 0 ,

with g(x) negative only in bounded domains in the finite solution domain, and with the associated boundary condition ψ → 0 as |x| → 00. This i s , of course, the potential well problem of Quantum Mechanics. The two-dimensional case is considered in detail. The application of Mahony's technique leads naturally to a particular class of eigensolutions. The boundary layer function introduced to cope with the situation in this case, determines a co-ordinate network in terms of which the problem can be solved quite generally. Thus the geometry of a particular problem is absorbed into the geometry of this co-ordinate net, no mathematical techniques special to a particular geometry being required.

Finally, in Part III , the method has been applied to a non-linear, ordinary, differential equation case of importance in the theory of shocks. Certain limitations of the method are exhibited in this case, and a combination of the Lagerstrom Kaplun Matching technique [15] and the Mahony approach leads t o suitable results.

The Mahony approach to singular perturbation problems leads to uniformly valid solution representations and thus escapes certain difficulties associated with the Lagerstrom Kaplun Matching technique. Thus, if the Mahony approach can be employed in a particular case it will produce more satisfactory results, sometimes at the expense of more analytic apparatus, but sometimes with rather less labour. It will be shown that the method extends to cases which cannot be satisfactorily attacked using the matching technique. At times, however, (as was found in the nonlinear differential equation case and in isolated portions of the solution domain in the partial differential equation case) it becomes impracticable to cover the complete solution domain with a single Mahony type expansion. Thus, if the solution domain contains more than one singular point it does not seem practicable, in general, to incorporate new "boundary layer co-ordinates" in the solution expansion to cope with the situation close to all the singular points. It is easier to employ separate matched expansions about each of the singular points in these cases.
Keyword Perturbation (Mathematics)

Document type: Thesis
Collection: UQ Theses (RHD) - UQ staff and students only
Citation counts: Google Scholar Search Google Scholar
Created: Fri, 03 Sep 2010, 13:11:24 EST by Ms Natalie Hull on behalf of Social Sciences and Humanities Library Service