The Generalized Stokes' Theorem

The generalised Stokes' Theorem is one of the most elegant equations in all of mathematics. In a single line, it unifies the Fundamental Theorem of Calculus, Green's Theorem, the classical Stokes' Theorem, and the Divergence Theorem — four results that fill chapters of a calculus textbook. Once you see this unification, the integral theorems are no longer four disconnected facts to memorise; they are four faces of one geometric truth: the integral of a derivative is a boundary value. This section is the payoff of everything in chapters 5 and 6.

Statement of the theorem

Let $M$ be a compact oriented $n$-dimensional smooth manifold with boundary $\partial M$ (given the induced orientation). Let $\omega$ be a smooth $(n-1)$-form on $M$. Then

That is it. The integral of $d\omega$ (an $n$-form) over the $n$-manifold $M$ equals the integral of $\omega$ over its boundary $\partial M$, an $(n-1)$-manifold.

How each classical theorem drops out

• FTC ($n=1$): $M = [a,b]$; $\omega = f$ is a 0-form; $d\omega = f',dx$; $\partial M = {b} - {a}$. The theorem reads $\int_a^b f',dx = f(b) - f(a)$.
• Green ($n=2$): $M = D \subset\mathbb{R}^2$; $\omega = P,dx + Q,dy$ is a 1-form; $d\omega = (Q_x - P_y),dx\wedge dy$. The theorem reads $\iint_D (Q_x - P_y),dA = \oint_{\partial D}(P,dx + Q,dy)$.
• Classical Stokes (surface in $\mathbb{R}^3$, $n=2$): $M = S$; $\omega$ is a 1-form (corresponding to $\mathbf{F}$); $d\omega$ is a 2-form (corresponding to $\nabla\times\mathbf{F}$). The theorem reads $\iint_S (\nabla\times\mathbf{F})\cdot d\mathbf{S} = \oint_{\partial S}\mathbf{F}\cdot d\mathbf{r}$.
• Divergence ($n=3$): $M = V\subset\mathbb{R}^3$; $\omega$ is a 2-form (corresponding to $\mathbf{F}$); $d\omega$ is a 3-form (corresponding to $\nabla\cdot\mathbf{F}$). The theorem reads $\iiint_V \nabla\cdot\mathbf{F},dV = \iint_{\partial V}\mathbf{F}\cdot d\mathbf{S}$.

Same equation. Different dimension. Different name in the textbook. One geometric statement underneath.

Proof strategy

The proof reduces, via partition-of-unity arguments, to the case where $M$ is a coordinate patch in $\mathbb{R}^n$. On a box, the theorem follows from the FTC applied along each axis in turn, exactly as in §5.5. Globally, the contributions from interior chart overlaps cancel (the same edge is integrated once with each orientation), leaving only the boundary contributions. The proof is one of the cleanest in geometry: most of the work is set-up; once $d$ and the volume of a parallelepiped are defined correctly, the theorem is essentially the multivariable FTC.

Why $d^2 = 0$ matches $\partial^2 = \emptyset$

If $\omega = d\alpha$ is exact, then on a closed manifold (one with $\partial M = \emptyset$) we have $\int_M \omega = \int_M d\alpha = \int_{\partial M}\alpha = \int_\emptyset \alpha = 0$. So integrals of exact forms on closed manifolds vanish. Conversely, the algebraic identity $d^2 = 0$ mirrors the topological fact that “the boundary of a boundary is empty.” This duality between algebra (forms) and topology (manifolds) is the seed of de Rham cohomology: the quotient $\ker d / \text{im},d$ at each degree measures topological holes.

(The full beauty of this theorem is best appreciated by writing out each special case on paper and noticing how they collapse into one formula. The references include detailed worked-out examples.)

Pause and think: If $M$ is a closed orientable surface (e.g., a torus) and $\omega = d\alpha$ is exact, what does the generalised Stokes' Theorem force $\int_M \omega$ to be? What does this imply about whether non-trivial top-degree forms on $M$ can ever be exact?

Try it

• Verify by direct calculation: take $M = [0,1]$ and $\omega = x^2$. Compute both sides of $\int_M d\omega = \int_{\partial M}\omega$.
• Take $M = D$ (unit disc in $\mathbb{R}^2$) and $\omega = -y,dx + x,dy$. Compute $d\omega$, then check both sides of Stokes. (Both should equal $2\pi$.)
• Use the generalised theorem with $M = V$ (unit ball) and $\omega = x,dy\wedge dz + y,dz\wedge dx + z,dx\wedge dy$. Compute $d\omega$; both sides should give $4\pi$.
• True or false: the integral of an exact $n$-form $d\alpha$ over a closed $n$-manifold is always zero. Use Stokes to justify.

A trap to watch for

Orientation, again. The induced orientation on $\partial M$ is the one for which Stokes' Theorem holds as stated. Reverse it and you flip the sign of the boundary integral. The rule: at each boundary point, the outward-pointing normal (in $M$) followed by the boundary orientation should agree with the orientation of $M$. Get this wrong on a multiple-component boundary (e.g., a spherical shell) and the inner-vs-outer signs cancel incorrectly.

What you now know

You can state the generalised Stokes' Theorem, recover the four classical integral theorems as special cases, and recognise the algebra-topology duality between $d^2 = 0$ and $\partial^2 = \emptyset$. This completes the Garrity chapters on vector calculus and differential forms. You now have the geometric vocabulary used across general relativity, gauge theory, symplectic mechanics, and modern topology.

References

• Garrity, T. (2002). All the Mathematics You Missed. Cambridge University Press, ch. 6.
• Spivak, M. (1965). Calculus on Manifolds. W. A. Benjamin, ch. 5 (the generalised theorem and its applications).
• Bachman, D. (2012). A Geometric Approach to Differential Forms (2nd ed.). Birkhäuser.
• Lee, J. M. (2012). Introduction to Smooth Manifolds (2nd ed.). Springer, ch. 16-17.
• Hubbard, J. H., Hubbard, B. B. (2015). Vector Calculus, Linear Algebra, and Differential Forms (5th ed.). Matrix Editions.