de Rham Cohomology and Global Flux

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.

De Rham cohomology measures the obstruction to replacing a closed differential form by one global potential. A closed form of positive degree is locally exact, but its integrals over closed cycles can remain nonzero. Those periods depend only on the cohomology class of the form and provide coordinate-independent global flux data; the angular form on $S^1$ gives the simplest complete example.

Required background. Tensor Fields and Natural Operations — use differential forms, pullbacks, smooth maps, and tangent-space calculus.

Helpful background. Calculus and Linear Algebra for Relativity — use line and surface integrals, quotient vector spaces, and the fundamental theorem of calculus.

The de Rham complex

Because $d^2=0$ , differential forms form a cochain complex

0\longrightarrow\Omega^0(M) \xrightarrow{d}\Omega^1(M) \xrightarrow{d}\cdots \xrightarrow{d}\Omega^n(M) \longrightarrow0.

Let $Z^p(M)=\ker(d:\Omega^p\to\Omega^{p+1})$ be the closed forms and $B^p(M)=\operatorname{im}(d:\Omega^{p-1}\to\Omega^p)$ the exact forms. Since $B^p\subset Z^p$ , the quotient

H^p_{\mathrm{dR}}(M)=Z^p(M)/B^p(M)

is defined. Two closed forms represent the same class when they differ by an exact form. The construction uses only smooth structure; it requires no metric. A systematic treatment, including homotopy invariance and calculations, appears in Guillemin and Haine 2019, §§5.1–5.3, pp. 171–196, PDF.

For a connected manifold, $H^0_{\mathrm{dR}}(M)\cong\mathbb R$ because closed zero-forms are constant. Higher groups can detect holes, but “counts holes” is only intuition: cohomology records graded algebraic information and its products, not a literal inventory of missing points.

Periods and global flux

Here a chain is a finite linear combination of oriented parametrized pieces, with integration and boundary taken term by term. A cycle is a chain whose oriented boundary is zero; it need not be the boundary of a higher-dimensional chain in $M$ . A circle enclosing the missing origin is a cycle in the punctured plane, but the disk that would fill it is not contained in that manifold. We use compact smooth cycles, so no convergence at infinity is being assumed.

Let $C$ be such a $p$ -cycle and $\omega$ a closed $p$ -form. Its period is $\int_C\omega$ . If $\omega$ is changed by an exact form $d\eta$ , Stokes’ theorem gives

\int_C(\omega+d\eta) =\int_C\omega+\int_{\partial C}\eta =\int_C\omega.

Thus the period depends only on $[\omega]$ . Similarly, moving $C$ through the boundary of a $(p+1)$ -chain leaves the period unchanged when $d\omega=0$ .

For a smooth manifold, the de Rham theorem identifies $H^p_{\mathrm{dR}}(M)$ with singular cohomology $H^p(M;\mathbb R)$ through integration over smooth cycles. Stokes makes integration respect the two differentials; the Poincaré lemma and a local-to-global argument show that the induced map is an isomorphism. A proof using a smooth triangulation appears in Calegari 2022, §§3.2–3.5, pp. 25–30, PDF. The full comparison theorem is not needed to use a nonzero period as an obstruction to exactness.

A nonzero period proves that a closed form is not exact. The converse—that all periods vanishing implies exactness—requires an appropriate class of cycles and the de Rham theorem or a direct argument for the manifold at hand.

The circle and the angular class

Regard $S^1$ as $x^2+y^2=1$ in $\mathbb R^2$ . Restrict the one-form

\alpha=x\,dy-y\,dx

to the circle. It is closed because every two-form on a one-dimensional manifold vanishes. Under the positive parametrization $\gamma(\theta)=(\cos\theta,\sin\theta)$ ,

\gamma^*\alpha=d\theta, \qquad \int_{S^1}\alpha=2\pi.

Therefore $\alpha$ is not exact. Locally it is $d\theta$ , but no globally single-valued real angle exists. In fact the period map

H^1_{\mathrm{dR}}(S^1)\longrightarrow\mathbb R, \qquad [\omega]\longmapsto\frac{1}{2\pi}\int_{S^1}\omega

is an isomorphism. To see this constructively, write the pullback of any one-form to the real angle line as $f(\theta)\,d\theta$ , where $f$ is smooth and $2\pi$ -periodic. Set

c=\frac1{2\pi}\int_0^{2\pi}f(s)\,ds, \qquad F(\theta)=\int_0^\theta\bigl(f(s)-c\bigr)ds.

The subtraction makes $F(\theta+2\pi)=F(\theta)$ , so $F$ descends to a smooth function on $S^1$ . Thus every one-form is $c\alpha+dF$ . The number $c$ is its only cohomological information: zero period means $c=0$ and hence exactness; every real $c$ occurs.

On $\mathbb R^2\setminus\{0\}$ , the corresponding form $(-y\,dx+x\,dy)/(x^2+y^2)$ has the same period around every positively oriented circle enclosing the origin. This is the global flux that remains after every local patch admits a potential.

Interpretation and limits

Cohomology distinguishes local differential information from global topology. It does not by itself determine a physical charge. A theory must identify the relevant closed form, normalization, allowed cycles, boundary conditions, and whether sources or singularities have been excluded. If $d\omega$ is nonzero somewhere, periods can change as a cycle crosses that region.

The de Rham class also forgets exact details: forms with the same class can have very different pointwise profiles. A metric can choose preferred representatives through Hodge theory under suitable global and boundary hypotheses, but no metric is present here. Noncompact manifolds and manifolds with boundary may require compactly supported or relative cohomology for the physical question being asked.

Homotopy invariance supplies a robust check on the circle result. Radially deforming the punctured plane onto a unit circle does not change its de Rham cohomology, so the angular class detected on the circle must also generate the first cohomology of the punctured plane. Filling in the missing origin destroys that class: on the full plane every closed one-form is exact by the Poincaré lemma. The contrast isolates topology rather than local differential behavior as the source of the nonzero period.

The wedge product descends to cohomology because changing either closed representative by an exact form changes their product only by an exact form. De Rham cohomology is therefore a graded ring, not merely a list of vector spaces. This product can distinguish spaces whose individual group dimensions agree. A full computation lies beyond the present page, but the structural point matters later when characteristic forms are multiplied to produce global obstruction classes.

Periods are insensitive to exact deformations but sensitive to orientation. Reversing a cycle changes the sign of its period, while deforming it through homologous cycles does not. These two tests separate an orientation mistake from a genuine change in the enclosed topology or sources.

Exercises

The sphere $S^2$ is simply connected. Does this force every closed two-form on it to be exact? Consider the restriction to the unit sphere of

\beta=X\,dY\wedge dZ+Y\,dZ\wedge dX+Z\,dX\wedge dY.

Use the outward orientation and spherical coordinates $(X,Y,Z)=(\sin\theta\cos\phi,\sin\theta\sin\phi,\cos\theta)$ to evaluate its period. Distinguish the exterior derivative in the ambient three-dimensional space from the derivative after restriction to $S^2$ .

Solution

The ambient derivative is $d\beta=3\,dX\wedge dY\wedge dZ$ . Its pullback to $S^2$ vanishes, since a two-dimensional tangent space cannot support a three-form. Naturality of $d$ therefore makes the restricted two-form closed. The coordinate pullback is

\beta|_{S^2}=\sin\theta\,d\theta\wedge d\phi.

The coordinates fail at the poles, but the original ambient formula shows the restricted form is smooth there. Integrating by the usual limiting patches gives

\int_{S^2}\beta =\int_0^{2\pi}\int_0^\pi\sin\theta\,d\theta\,d\phi=4\pi.

If the restricted form were $dA$ for a global one-form $A$ on $S^2$ , Stokes would make this integral zero because $\partial S^2=\varnothing$ . Hence it is not exact. Simple connectedness eliminates the path-dependence obstruction for closed one-forms; it does not eliminate higher-degree cohomology.

References

Calegari, Danny. Introduction to Differential Geometry. Preliminary version, May 26, 2022, University of Chicago, §§3.2–3.5, pp. 25–30. Course PDF.
Guillemin, Victor, and Peter J. Haine. Differential Forms. World Scientific, 2019, §§5.1–5.3, pp. 171–196, PDF.