Differential Forms and Stokes' Theorem

Vector fields tell you which way to move; differential forms are the things you integrate. A form is the natural integrand on a manifold: it eats tangent vectors and spits out numbers, in just the antisymmetric way that oriented area and volume demand. Build the theory up and it delivers one of the most beautiful results in mathematics — the generalized Stokes theorem, a single equation that swallows the fundamental theorem of calculus, Green's theorem, the classical curl theorem, and the divergence theorem all at once.

What a differential form is

A differential k-form is a smooth field of alternating multilinear gadgets: at each point it is a function that takes k tangent vectors and returns a real number, is linear in each slot, and is totally antisymmetric — swap any two inputs and the sign flips. In the language of tensors, a k-form is precisely an alternating (0,k)-tensor field.

That antisymmetry is the whole personality of the subject. It is what makes forms measure oriented volume: swapping two edges of a parallelogram reverses its orientation and negates its signed area. By degree, on an n-manifold:

a 0-form is just a smooth function;
a 1-form eats one vector — e.g. df, or dx^i;
an n-form is a top form — a volume element, the thing you integrate over the whole manifold;
a k-form with k > n is identically 0 — you cannot fit more independent antisymmetric slots than the dimension allows.

The wedge product

Forms multiply with the wedge product \wedge, which builds a (k+l)-form out of a k-form and an l-form. Degrees add:

\deg(\alpha \wedge \beta) = \deg\alpha + \deg\beta.

The wedge is associative and bilinear, but graded-anticommutative: for a k-form \alpha and an l-form \beta,

\alpha \wedge \beta = (-1)^{kl}\, \beta \wedge \alpha.

The most-used consequence: for a 1-form the sign is -1, so dx \wedge dy = -\,dy \wedge dx, and in particular

dx \wedge dx = 0.

Repeating a 1-form annihilates the product — the antisymmetric echo of "a parallelogram with two identical edges has zero area". This is why an n-form is the top of the ladder: on \mathbb{R}^3 the only non-repeating triple is dx \wedge dy \wedge dz, and any fourth factor would repeat one and vanish.

The exterior derivative

The exterior derivative d turns a k-form into a (k+1)-form — it raises degree by one. On a 0-form (a function f) it is just the differential you already know:

df = \sum_{i=1}^{n} \frac{\partial f}{\partial x^i}\, dx^i,

a 1-form packaging all the partial derivatives. Applied again, d extends by the graded Leibniz rule — and its single most important property is that doing it twice always gives zero:

d(d\omega) = 0, \qquad\text{i.e.}\qquad d^2 = 0.

This one identity, unwound in coordinates, is exactly the statement that \operatorname{curl}(\operatorname{grad} f) = 0 and \operatorname{div}(\operatorname{curl}\mathbf{F}) = 0 — the two "second derivatives vanish" facts of vector calculus, revealed as one line about forms. It holds because mixed partial derivatives commute while the wedge anticommutes, so every term cancels its twin.

Stokes' theorem — one theorem to rule them all

Everything above exists to make one equation possible. If \omega is a (k-1)-form and M is an oriented k-dimensional manifold with boundary \partial M, then the integral of d\omega over the inside equals the integral of \omega over the edge.

For an oriented compact k-manifold M with boundary \partial M and a smooth (k-1)-form \omega:

\int_M d\omega = \int_{\partial M} \omega;
k = 1 is the fundamental theorem of calculus, \int_a^b f'\,dx = f(b) - f(a) (the boundary is the two endpoints);
k = 2 in the plane is Green's theorem; on a surface in space it is the classical Stokes / curl theorem;
k = 3 is the divergence (Gauss) theorem;
"the derivative of a thing, summed over a region, is the thing summed over the boundary" — a single antiderivative-versus-boundary law, valid in every dimension.

Worked: degrees, a differential, and a wedge

Degree bookkeeping. Wedging a p-form with a q-form gives a (p+q)-form; applying d to a k-form gives a (k+1)-form. So on \mathbb{R}^3, if \alpha is a 1-form and \beta a 2-form, then \alpha \wedge \beta is a 3-form (a top form), while \beta \wedge \beta is a 4-form on a 3-manifold — hence 0.

A differential. For f = x^2 y on \mathbb{R}^2,

df = \frac{\partial f}{\partial x}\,dx + \frac{\partial f}{\partial y}\,dy = 2xy\,dx + x^2\,dy,

a 1-form, as promised.

A wedge. With \alpha = dx and \beta = x\,dy,

\alpha \wedge \beta = dx \wedge (x\,dy) = x\, dx \wedge dy,

and swapping the order costs a sign: \beta \wedge \alpha = x\,dy \wedge dx = -x\,dx \wedge dy. Order and sign matter.

Generations of students memorise the fundamental theorem of calculus, Green's theorem, the curl theorem and the divergence theorem as four separate incantations, each with its own diagram and its own way of orienting the boundary. Stokes' theorem reveals they are a single sentence read at four different dimensions: \int_M d\omega = \int_{\partial M} \omega.

The pattern is always the same. Integrate a derivative over the interior; the answer only depends on values along the boundary. In 1D the "boundary" is two endpoints and the derivative is f'; in 3D the boundary is a closed surface and the derivative is the divergence. Once you translate grad, curl and div into the exterior derivative d — which they secretly always were — the four theorems collapse into one. Unifications like this are why differential forms feel less like a new topic and more like the language the old topics were always trying to speak.

d^2 = 0 says every exact form (one that equals d\omega for some \omega) is automatically closed (has d(d\omega) = 0). It is tempting to read the converse — "closed \Rightarrow exact" — but that is false in general. A closed form need not be exact; whether it is depends on the shape of the domain.

The classic counterexample is the angle form \omega = \dfrac{-y\,dx + x\,dy}{x^2 + y^2} on the punctured plane: it is closed everywhere it is defined, yet it is not exact, because the plane has a hole at the origin. Exactly how badly "closed" fails to be "exact" is measured by de Rham cohomology — the holes in a space are detected by closed-but-not-exact forms. And keep the algebra straight while you are at it: the wedge is anticommutative, so order and sign matter, and \alpha \wedge \alpha = 0 for any 1-form \alpha. Do not treat \wedge like ordinary multiplication.