S. D. V. Stephens

notes / Studies in Real and Complex Analysis / chapter 6: Analysis in Several Variables

26 Change of Variables and Differential Forms

· Denis Auroux

Original Professor Auroux’s handwritten notes (PDF), behind a password for now:

Iterated and Riemann integrals in several variables

a continuous function on an -cell we can define

(why ? clearer after differential forms) either

  1. as an iterated integral: or in any order;
  2. as a Riemann integral: split into small cubes , and bound between the piecewise constant functions on and on , giving the estimate If is continuous, hence uniformly continuous, then as , so this defines the integral uniquely.

Fubini’s theorem says: for continuous , the iterated integrals for different orders of integration are all equal.

If is only piecewise continuous, integrability still holds if the regions of where is continuous are sufficiently regular—e.g. delimited by smooth hypersurfaces. Specifically: when decomposing into small cubes , we want

as one subdivides further. Over such cubes, doesn’t as the step size , but if vol we still have .

Thus we can define integrals over regions of delimited by hypersurfaces, either by

  • extending by outside of the given region, and integrating the resulting piecewise continuous function, or
  • using changes of coordinates (via the implicit function theorem) to make the region of integration an -cell. This requires change of variables…

Theorem 26.1 (Change of variables). a diffeomorphism, continuous on ; then

(Won’t prove. The geometric input is that if is a small cube , then a small parallelepiped , with .)

Path integrals

We also want to consider path integrals such as the following: given a path , , and a differential (1-form) ( ), the path integral is

  • This is independent of the parametrization of the path, by change of variable + chain rule.
  • If we reverse the path, , then .
  • Given , define ; then .

Differential forms

This generalizes to arbitrary dimensions, using the language of differential forms.

On , the symbols can be viewed as the differentials of the coordinate functions ; they form a basis of , the linear forms on the space of tangent vectors (, the th component). Differential 1-forms are therefore functions with values in .

We now consider the exterior powers = the vector space with basis , which are parts of the exterior algebra generated by , i.e. the quotient of the tensor algebra by setting . (NB: .) ( and for all 1-forms.)

Definition 26.2. A -form on an open subset is a function with values in :

We can evaluate it on vectors , :

(an alternating multilinear form).

The space of -forms on is usually denoted (). We can multiply -forms by functions, or take exterior products ():

( if , and if ).

The exterior derivative is

Example 26.3. : .

: .

Proposition 26.4. , i.e. , .

(This follows from and .)

Say is closed if , and exact if for some . The above says: exact closed.

Theorem 26.5 (Poincaré lemma). For convex open and , is exact iff is closed.

Remark. This leads to de Rham cohomology, a key invariant in differential topology!

The Poincaré lemma says for convex and , while detects that isn’t simply connected.

Pullback of differential forms

If is a smooth map (, ), then we have a map defined by

Basic properties:

  1. for functions (), ;
  2. ;
  3. (follows from the chain rule).

Concretely, denoting by coordinates on and on : (), and

where

Especially: for and ,

Integration of differential forms

Given , we can integrate over a -dimensional submanifold parametrized by a smooth map from a -cell to (or any other nice enough domain for integration), , :

splitting into small grid parallelepipeds:

A curved 3-by-3 grid drawn on a surface M; the corner cell at the point x_i is shaded as the small parallelepiped spanned by the vectors v_1 and v_k.

i.e.

Explicitly in components, writing , if :

Check: for 1-forms this agrees with the path integral formula .

What this formula means is:

  • for -forms on , ;
  • for general , , where is a -form on a usual integral.

We can similarly integrate -forms over = a finite union of parametrized pieces.

Theorem 26.6 (Pullback formula). Given a smooth map , , and a -dimensional :

This is basically equivalent to the change of variables formula for the usual , and implies that is independent of the manner in which we parametrize as the image of a map (or union of pieces), as long as all reparametrizations are orientation-preserving (i.e. we compose with a diffeomorphism , , such that everywhere).

Example 26.7. on ; = the circle of radius , oriented counterclockwise (as a path ).

Show solution

Pulling back via (polar coordinates),

So

Note: (by direct calculation, or using ), i.e. is closed; but it is not exact! If on such that , then the path integral would be . So .

But… the path integral is independent of the radius , or in fact the same for any closed curve going once around the origin. This is a consequence of Stokes’ theorem (next time).