The Exterior Derivative and the Poincare Lemma

Scrollable formula. Use Left and Right arrow keys to reveal hidden content.

Scrollable table. Use Left and Right arrow keys to reveal additional columns.

Scrollable figure. Use Left and Right arrow keys to inspect hidden content.

The exterior derivative is the unique natural first-order operator $d:\Omega^p(M)\to\Omega^{p+1}(M)$ that extends the differential of functions, obeys a graded product rule, and satisfies $d^2=0$ . Consequently every exact form is closed. The Poincaré lemma supplies the local converse on contractible coordinate neighborhoods, but the angular form on the punctured plane shows why closed need not mean globally exact.

Required background. Tensor Fields and Natural Operations — use differential forms, pullbacks, vector fields, and antisymmetrization.

Helpful background. Calculus and Linear Algebra for Relativity — use partial derivatives, the fundamental theorem of calculus, and homotopies of Euclidean domains.

The exterior derivative

For

\omega=\frac{1}{p!}\omega_{i_1\cdots i_p} dx^{i_1}\wedge\cdots\wedge dx^{i_p},

define

d\omega=\frac{1}{p!}\partial_j\omega_{i_1\cdots i_p} dx^j\wedge dx^{i_1}\wedge\cdots\wedge dx^{i_p}.

Although written with partial derivatives, the result is chart independent. It is characterized by $df(X)=Xf$ , linearity, locality, $d^2=0$ , and

d(\alpha\wedge\beta) =d\alpha\wedge\beta+(-1)^p\alpha\wedge d\beta

for a $p$ -form $\alpha$ . The coordinate formula gives $d^2=0$ because mixed partial derivatives are symmetric while the corresponding wedge factors are antisymmetric. These axioms also determine $d$ uniquely: write each form locally as a sum of functions times wedges of the $dx^i$ , use the product rule, and use $d(dx^i)=d^2x^i=0$ . Pullback commutes with $d$ : $F^*(d\omega)=d(F^*\omega)$ . See Guillemin and Haine 2019, §2.4, pp. 53–57; §2.6, pp. 61–64, PDF.

For a one-form $\omega=P\,dx+Q\,dy$ on the plane, the abstract rule becomes

\begin{aligned} d\omega &=(P_x\,dx+P_y\,dy)\wedge dx +(Q_x\,dx+Q_y\,dy)\wedge dy\\ &=(Q_x-P_y)\,dx\wedge dy. \end{aligned}

The minus sign comes from $dy\wedge dx=-dx\wedge dy$ . Thus closedness is the familiar cross-derivative condition $Q_x=P_y$ , stated without choosing a metric.

A form is closed if $d\omega=0$ and exact if $\omega=d\eta$ . Exactness implies closedness immediately. The converse is a question about both the domain and the degree.

The Poincaré lemma and its proof mechanism

On a star-shaped open set $U\subset\mathbb R^n$ , every closed $p$ -form with $p\gt0$ is exact. Translate the center to the origin, so that $H_t(x)=tx$ stays in $U$ for $0\le t\le1$ . With $R=x^i\partial_i$ , define the operator $K:\Omega^p(U)\to\Omega^{p-1}(U)$ by

K\omega=\int_0^1 \frac{1}{t}\,H_t^*(\iota_R\omega)\,dt.

This formula is exact, with its integrand smoothly extended to $t=0$ . To see why it is regular and how to use it, write it in components:

K\omega =\frac{1}{(p-1)!}\left[\int_0^1 t^{p-1} x^j\omega_{j i_2\cdots i_p}(tx)\,dt\right] dx^{i_2}\wedge\cdots\wedge dx^{i_p}.

There is no singular factor here, including when $p=1$ . For $t\gt0$ , differentiation of $H_t^*\omega$ gives

\frac{d}{dt}H_t^*\omega =\frac1t H_t^*\bigl(d\iota_R\omega+\iota_Rd\omega\bigr).

This identity follows either by differentiating the components or by the Cartan formula for a radial flow. Integration in $t$ yields

dK+Kd=H_1^*-H_0^*.

The map $H_1$ is the identity; $H_0$ is constant, so $H_0^*\omega=0$ in positive degree. If $d\omega=0$ , therefore, $\omega=d(K\omega)$ . The same argument with a smooth contraction proves the result on a contractible smooth manifold; it does not require the manifold itself to be a Euclidean star-shaped set. The homotopy identity and its Poincaré-lemma consequence are developed in Guillemin and Haine 2019, §2.6, Theorem 2.6.15 and Exercises 2.6.vi–vii, pp. 64–67, PDF.

The proof is local because a coordinate ball can be chosen star-shaped. Therefore every closed positive-degree form is locally exact. It does not follow that one potential patches together globally.

The angular form on the punctured plane

On $M=\mathbb R^2\setminus\{0\}$ , define

\eta=\frac{-y\,dx+x\,dy}{x^2+y^2}.

Direct differentiation gives $d\eta=0$ . On any simply connected sector where an angle $\theta$ is single valued, $\eta=d\theta$ , so the Poincaré lemma is visible locally. Yet around the positively oriented circle $C_R(\theta)=(R\cos\theta,R\sin\theta)$ ,

\int_{C_R}\eta =\int_0^{2\pi}d\theta =2\pi.

If a global function $f$ satisfied $\eta=df$ , the integral around every closed curve would vanish by the fundamental theorem of calculus. Therefore $\eta$ is closed but not exact. The missing origin creates a global obstruction invisible in a sufficiently small neighborhood.

This example also exposes a frequent notation trap. Writing $\eta=d\theta$ is valid only on a chart with a chosen branch of the angle. There is no globally single-valued smooth $\theta:M\to\mathbb R$ whose differential is $\eta$ .

Scope and later use

The Poincaré lemma is not the claim that every closed form on every manifold is exact. Contractibility is a sufficient domain hypothesis, not a necessary one; on a general manifold exactness must be established separately. Nor does $d^2=0$ depend on a metric, connection, or field equation. In gauge theory it underlies local potentials; in conservation laws it turns a local differential identity into a candidate flux statement. Whether a potential exists globally is measured by de Rham cohomology; interpreting its periods as physical charges additionally requires a theory and normalization.

For degree zero, “closed” means locally constant because $df=0$ ; there is no degree $-1$ potential in the ordinary de Rham complex. Boundary conditions and support restrictions can also matter when one asks for compactly supported potentials, questions not covered by the elementary lemma here.

There is also a direct coordinate check on nilpotence. Applying the exterior derivative twice produces second partial derivatives symmetric in their derivative indices, multiplied by basis one-forms antisymmetric in the same indices. Their contraction vanishes. This explanation requires the usual smoothness that allows mixed partial derivatives to commute. It complements the homotopy proof: nilpotence is local algebra, while exactness of a closed form depends on the domain’s global shape.

Naturality gives another stringent test. For every smooth map, pulling back after differentiation must equal differentiating after pullback. This identity lets one prove closedness on a parametrized submanifold without recalculating in its coordinates. It also means a potential pulls back to a potential. The converse need not hold: a pullback can erase information when the map has insufficient rank, so exactness after pullback does not prove exactness on the target.

For one-forms on a star-shaped plane domain, the lemma reduces to the equality of cross derivatives familiar from elementary calculus. The homotopy operator upgrades that fact to every positive degree and dimension while displaying the topology needed for a global primitive.

Exercises

On $\mathbb R^2$ , show that $\omega=2xy\,dx+x^2\,dy$ is closed and use the homotopy idea—or direct inspection—to find a potential. Why does this example not test the global limitation of the Poincaré lemma?

Solution

Writing $\omega=Pdx+Qdy$ gives

d\omega=(\partial_xQ-\partial_yP)dx\wedge dy =(2x-2x)dx\wedge dy=0.

For a one-form, the component formula for $K$ gives a function. Here

K\omega(x,y) =\int_0^1\bigl[2(tx)(ty)x+(tx)^2y\bigr]dt =3x^2y\int_0^1t^2dt=x^2y.

Differentiating verifies $d(K\omega)=2xy\,dx+x^2\,dy$ . The domain $\mathbb R^2$ is star shaped and contractible; it has no hole on which a nonzero period could obstruct a global potential. The angular-form example is the discriminating global test.

References

Guillemin, Victor, and Peter J. Haine. Differential Forms. World Scientific, 2019, §2.4, pp. 53–57; §2.6, pp. 61–67, Open PDF.