A theorem which effectively describes how lengths, areas, volumes, and generalized n-dimensional volumes (contents) are distorted by differentiable functions. In particular, the change of variables theorem reduces the whole problem of figuring out the distortion of the content to understanding the infinitesimal distortion, i.e., the distortion of the derivative (a linear map), which is given by the linear map's determinant. So f:R^n->R^n is an area-preserving linear transformation iff left bracketing bar det(f') right bracketing bar = 1, and in more generality, if S is any subset of R^n, the content of its image is given by left bracketing bar det(f') right bracketing bar times the content of the original. The change of variables theorem takes this infinitesimal knowledge, and applies calculus by breaking up the domain into small pieces and adds up the change in area, bit by bit. The change of variable formula persists to the generality of differential k-forms on manifolds, giving the formula integral_M(f^* ω) = integral_W(ω) under the conditions that M and W are compact connected oriented manifolds with nonempty boundaries, f:M->W is a smooth map which is an orientation-preserving diffeomorphism of the boundaries. In one dimension, the explicit statement of the theorem for f a continuous function of y is integral_s f(ϕ(x))(d ϕ)/(d x) d x = integral_T f(y) d y, where y = ϕ(x) is a differential mapping on the interval [c, d] and T is the interval [a, b] with ϕ(c) = a and ϕ(d) = b. In two dimensions, the explicit statement of the theorem is integral_R f(x, y) d x d y = integral_R^* f[x(u, v), y(u, v)] left bracketing bar (d(x, y))/(d(u, v)) right bracketing bar d u d v and in three dimensions, it is integral_R f(x, y, z) d x d y d z = integral_R^* f[x(u, v, w), y(u, v, w), z(u, v, w)] left bracketing bar (d(x, y, z))/(d(u, v, w)) right bracketing bar d u d v d w, where R = f(R^*) is the image of the original region R^*, left bracketing bar (d(x, y, z))/(d(u, v, w)) right bracketing bar is the Jacobian, and f is a global orientation-preserving diffeomorphism of R and R^* (which are open subsets of R^n). The change of variables theorem is a simple consequence of the curl theorem and a little de Rham cohomology. The generalization to n dimensions requires no additional assumptions other than the regularity conditions on the boundary.

implicit function theorem | Jacobian