Page 12 - Basic Structured Grid Generation
P. 12

1







                                               Mathematical

                           preliminaries – vector and tensor

                                                    analysis








                           1.1 Introduction

                        In this chapter we review the fundamental results of vector and tensor calculus which
                        form the basis of the mathematics of structured grid generation. We do not feel it
                        necessary to give derivations of these results from the perspective of modern dif-
                        ferential geometry; the derivations provided here are intended to be appropriate to
                        the background of most engineers working in the area of grid generation. Helpful
                        introductions to tensor calculus may be found in Kay (1988), Kreyzig (1968), and
                        Spain (1953), as well as many books on continuum mechanics, such as Aris (1962).
                        Nevertheless, we have tried to make this chapter reasonably self-contained. Some of
                        the essential results were presented by the authors in Farrashkhalvat and Miles (1990);
                        this book started at an elementary level, and had the restricted aim, compared with
                        many of the more wide-ranging books on tensor calculus, of showing how to use
                        tensor methods to transform partial differential equations of physics and engineer-
                        ing from one co-ordinate system to another (an aim which remains relevant in the
                        present context). There are some minor differences in notation between the present
                        book and Farrashkhalvat and Miles (1990).


                           1.2 Curvilinear co-ordinate systems and base
                                 vectors in E  3

                                                                       i
                        We consider a general set of curvilinear co-ordinates x , i = 1, 2, 3, by which points
                                                             3                          1   2  3
                        in a three-dimensional Euclidean space E may be specified. The set {x ,x ,x }
                        could stand for cylindrical polar co-ordinates {r, θ, z}, spherical polars {r, θ, ϕ},etc.
                        A special case would be a set of rectangular cartesian co-ordinates, which we shall
                        generally denote by {y 1 ,y 2 ,y 3 } (where our convention of writing the integer indices
                        as subscripts instead of superscripts will distinguish cartesian from other systems),
                                                                                 1
                                                                                       3
                                                                                    2
                        or sometimes by {x, y, z} if this would aid clarity. Instead of {x ,x ,x },it may
                        occasionally be clearer to use notation such as {ξ, η, ς} without indices.
   7   8   9   10   11   12   13   14   15   16   17