HIGHER DERIVATIVES - Successive Approximations and Implicit Functions - Advanced Calculus of Several Variables

Advanced Calculus of Several Variables (1973)

Part III. Successive Approximations and Implicit Functions

Chapter 5. HIGHER DERIVATIVES

Thus far in this chapter our attention has been confined to Image mappings, or to first derivatives. This is a brief exercise section dealing with higher derivatives.

Recall (from Section II.7) that the function f : UImage, where U is an open subset of Imagen, is of class Image if all partial derivatives of f, of order at most k, exist and are continuous on U. Equivalently, f is of class Image if and only if fand its first-order partial derivatives D1 f, . . . , Dnf are all of class Image on U. This inductive form of the definition is useful for inductive proofs.

We say that a mapping F : UImagem is of class Image if each of its component functions is of class Image.

Exercise 5.1 If f and g are functions of class Image on Imagen, and Image, show that the functions f + g, fg, and af are also of class Image.

The following exercise gives the class Image chain rule.

Exercise 5.2 If f : ImagenImagem and g : ImagemImagel are class Image mappings, prove that the composition g Image f : ImagenImagel is of class Image. Use the previous exercise and the matrix form of the ordinary (class Image) chain rule.

Recall that we introduced in Section 2 the space Imagemn of m × n matrices, and showed that the differentiable mapping f : ImagenImagem is Image if and only if its derivative f′ : ImagenImagemn is continuous.

Exercise 5.3 Show that the differentiable mapping f : ImagenImagem is of class Image on the open set U if and only if f′ : ImagenImagemn is of class Image on U.

Exercise 5.4 Denote by Imagenn the open subset of Imagenn consisting of all non-singular n × n matrices, and by Image : ImagennImagenn the inversion mapping, Image(A) = A−1. Show that Image is of class Image for every positive integer k. This is simply a matter of seeing that the elements of A−1 are Image functions of those of A.

We can now establish Image versions of the inverse and implicit mapping theorems.

Theorem 5.1If, in the inverse mapping theorem, the mapping f is of class Image in a neighborhood of a, then the local inverse g is of class Image in a neighborhood of b = f(a).

Theorem 5.2If, in the implicit mapping theorem, the mapping G is of class Image in a neighborhood of (a, b), then the implicitly defined mapping h is of class Image in a neighborhood of a.

It suffices to prove Theorem 5.1, since 5.2 follows from it, just as in the Image case. Recall the formula

Image

from the last paragraph of the proof of Theorem 3.3 (the Image inverse mapping theorem). We already know that g is of class Image. If we assume inductively that g is of class Image, then Exercises 5.2 and 5.4 imply that g′ : ImagenImagenn is of class Image, so g is of class Image by Exercise 5.3.

Image

With the class Image inverse and implicit mapping theorems now available, the interested reader can rework Section 4 to develop the elementary theory of class Image manifolds—ones for which the coordinate patches are regular class Image mappings. Everything goes through with Image replaced by Image throughout.