Surface integrals of 2-forms; flux of a vector field through an oriented surface

Intuition Beginner

Imagine a fishing net held open in a moving river. The amount of water passing through the net each second depends on three things: how fast the water moves, how the net is tilted relative to the flow, and how big the net is. Hold the net face-on to the current and a lot of water goes through. Turn it edge-on and almost nothing passes. This quantity, the rate at which something flows through a surface, is called the flux.

To make the count honest you must pick a side of the surface as "out". Water crossing from the in-side to the out-side counts as positive; water crossing the other way counts as negative. Choosing a consistent side everywhere is what mathematicians mean by orienting the surface.

Flux is everywhere once you look. It is the electric field streaming through a closed shell, heat leaving a warm body, or fluid pumped through a membrane. In each case you add up a flow, weighted by how squarely the surface faces it, over the whole surface.

Visual Beginner

Picture a curved surface peppered with little arrows showing the flow at each point. At every patch, only the part of the arrow poking straight through the patch contributes; the part sliding along the surface does not.

The chosen outward side fixes the sign: arrows leaving through the out-side add up positively, arrows entering subtract. Add the contributions over every patch and you get one number, the total flux.

Worked example Beginner

Take a flat square panel of side 2, lying in a horizontal plane, with area 4. Let the flow point straight up at a steady speed of 3 units, and choose "up" as the out-side.

Because the flow pierces the panel face-on, every bit of the 3 units of speed counts. The flux is speed times area: 3 times 4, which is 12.

Now tilt the panel so it stands vertically while the flow still points straight up. The flow now slides along the panel and pierces none of it. The flux drops to 0.

What this tells us: flux rewards the part of the flow that goes through the surface, scaled by area, and it collapses to nothing when the surface turns edge-on to the flow.

Check your understanding Beginner

Formal definition Intermediate+

Let $S$ be an oriented surface in ${\mathbb{R}}^3$ and let $\Phi :D\to {\mathbb{R}}^3$ be a smooth parametrisation of $S$ on a region $D\subset {\mathbb{R}}^2$ with coordinates $(u,v)$, chosen so that $\Phi$ agrees with the orientation of $S$. For a 2-form $\omega$ defined near $S$, the surface integral of $\omega$ over $S$ is the integral of the pullback over the parameter domain:

$${\int }_S\omega ={\int }_D{\Phi }^{\ast }\omega .$$

Here ${\Phi }^{\ast }$ denotes the pullback, which precomposes $\omega$ with $\Phi$ and substitutes $d{x}^i={\sum }_j\frac{\partial {\Phi }^i}{\partial u^j}d{u}^j$; the right-hand integral is an ordinary double integral over $D$. Writing $\omega =ady\wedge dz+bdz\wedge dx+cdx\wedge dy$, the pullback evaluates to

$${\Phi }^{\ast }\omega =(a\frac{\partial (y,z)}{\partial (u,v)}+b\frac{\partial (z,x)}{\partial (u,v)}+c\frac{\partial (x,y)}{\partial (u,v)})du\wedge dv,$$

where each $\frac{\partial (\cdot ,\cdot )}{\partial (u,v)}$ is the $2\times 2$ Jacobian minor.

To a vector field $F=(F^1,F^2,F^3)$ associate the 2-form obtained by feeding $F$ into the standard volume form $dV=dx\wedge dy\wedge dz$ through the interior product ${\iota }_F$ (contraction in the first slot, $({\iota }_{F}dV)(X,Y)=dV(F,X,Y)$):

$${\omega }_F={\iota }_F(dx\wedge dy\wedge dz)=F^{1}dy\wedge dz+F^{2}dz\wedge dx+F^{3}dx\wedge dy.$$

The flux of $F$ through $S$ is then ${\int }_S{\omega }_F$. The cross product ${\Phi }_u\times {\Phi }_v$ controls everything: the scalar surface element is $dA=|{\Phi }_u\times {\Phi }_v|dudv$, and the vector surface element is $\mathbf{n}dA=({\Phi }_u\times {\Phi }_v)dudv$, where $\mathbf{n}$ is the unit normal compatible with the orientation. With this notation the flux is the classical surface integral

$${\int }_S{\omega }_F={\iint }_{S}F\cdot \mathbf{n}dA={\iint }_{D}F(\Phi )\cdot ({\Phi }_u\times {\Phi }_v)dudv,$$

since ${\Phi }^{\ast }{\omega }_F=F(\Phi )\cdot ({\Phi }_u\times {\Phi }_v)du\wedge dv$ by direct expansion of the three Jacobian minors ^{[Shifrin Ch. 8]}. A common slip is to forget that ${\Phi }_u\times {\Phi }_v$ already encodes both the magnitude $dA$ and the orientation $\mathbf{n}$; using $|{\Phi }_u\times {\Phi }_v|$ for a flux integral drops the sign information that the 2-form is built to carry.

Key theorem with proof Intermediate+

Theorem (invariance under orientation-preserving reparametrisation). Let $\Phi :D\to S$ and $\Psi :E\to S$ be two smooth parametrisations of the same oriented surface $S$, related by an orientation-preserving change of coordinates $g:E\to D$ with $\Phi \circ g=\Psi$ and $\det Dg>0$. Then for every 2-form $\omega$ near $S$,

$${\int }_D{\Phi }^{\ast }\omega ={\int }_E{\Psi }^{\ast }\omega .$$

Proof. Pullback is functorial: ${\Psi }^{\ast }\omega =(\Phi \circ g{)}^{\ast }\omega =g^{\ast }({\Phi }^{\ast }\omega )$. Write ${\Phi }^{\ast }\omega =hdu\wedge dv$ for a scalar function $h$ on $D$. Pulling a top-degree 2-form back by $g$ multiplies its coefficient by the Jacobian determinant of $g$:

$$g^{\ast }(hdu\wedge dv)=(h\circ g)\det (Dg)ds\wedge dt,$$

with $(s,t)$ the coordinates on $E$. Integrating over $E$,

$${\int }_E{\Psi }^{\ast }\omega ={\int }_E(h\circ g)\det (Dg)dsdt.$$

Since $g$ is orientation-preserving, $\det (Dg)=|\det (Dg)|>0$, so this is exactly the right-hand side of the ordinary change-of-variables theorem for the double integral of $h$ over $D=g(E)$:

$${\int }_E(h\circ g)|\det (Dg)|dsdt={\int }_{D}hdudv={\int }_D{\Phi }^{\ast }\omega .$$

The two surface integrals therefore agree. If instead $g$ reverses orientation, $\det (Dg)<0$; the algebraic Jacobian factor $\det (Dg)$ from the pullback and the absolute value $|\det (Dg)|$ from change of variables differ by a sign, and the integral changes sign. This is why the integral of a 2-form depends on $S$ only through its orientation, not its parametrisation. $\square$

Bridge. This invariance builds toward 03.04.03 (integration on manifolds), where the same pullback-and-Jacobian mechanism is the engine that glues chart-local integrals into a global integral, and it appears again in 03.04.05 (Stokes' theorem), where surface flux is paired against the exterior derivative of a 1-form. The orientation-sensitivity isolated here is the same datum that 03.04.06 (de Rham cohomology) records as a cohomology class, and it carries through to 03.07.05 (Yang-Mills action), where curvature 2-forms are integrated over surfaces to extract topological charge.

Exercises Intermediate+

Lean formalization Intermediate+

lean_status: none — Mathlib provides the Lebesgue change-of-variables theorem and a growing differential-forms layer, but it does not package the surface integral of a 2-form as the integral of a pullback over a parameter domain, nor the vector-field-to-2-form correspondence ${\omega }_F={\iota }_F(dV)$, nor the solid-angle flux computation.

A formalization route would need: oriented parametrised surfaces, the pullback of a 2-form, the interior product with the volume form, the orientation-preserving-reparametrisation invariance theorem proved above, and the $4\pi$ flux of the inverse-square field via the closedness of ${\omega }_F$ on ${\mathbb{R}}^3\setminus \{0\}$.

Advanced results Master

The flux integral is the codimension-one face of the calculus of forms, and three structural facts organise its behaviour.

First, the divergence theorem is the Stokes identity for ${\omega }_F$. For a compact region $U\subset {\mathbb{R}}^3$ with boundary $\partial U$ oriented outward,

$${\int }_{\partial U}{\omega }_F={\int }_{U}d{\omega }_F={\int }_U(\mathrm{div}F)dx\wedge dy\wedge dz,$$

because $d{\omega }_F=d\left({\iota }_{F}dV\right)=(\mathrm{div}F)dV$ by the Cartan formula ${\mathcal{L}}_{F}dV=d{\iota }_{F}dV+{\iota }_{F}ddV=d{\iota }_{F}dV$ together with ${\mathcal{L}}_{F}dV=(\mathrm{div}F)dV$ ^{[Spivak Ch. 5]}. The classical Stokes theorem for the curl is the same statement one degree down: for a vector field $G$ with associated 1-form ${\alpha }_G=G^{1}dx+G^{2}dy+G^{3}dz$, one has $d{\alpha }_G={\omega }_{\mathrm{curl}G}$, so ${\int }_{\partial S}{\alpha }_G={\int }_S{\omega }_{\mathrm{curl}G}={\iint }_S(\mathrm{curl}G)\cdot \mathbf{n}dA$.

Second, the value $4\pi$ for the inverse-square field is cohomological. The 2-form ${\omega }_F$ with $F=\mathbf{r}/|\mathbf{r}{|}^3$ restricts on the unit sphere $S^2$ to $4\pi$ times the generator of $H_{\mathrm{dR}}^2(S^2)\cong \mathbb{R}$. Concretely ${\omega }_F{|}_{S^2}$ is the area form $\sigma$ of $S^2$ (total mass $4\pi$), which is closed but not exact: there is no globally defined 1-form $\eta$ on $S^2$ with $d\eta =\sigma$, since otherwise Stokes on the closed surface $S^2$ would force ${\int }_{S^2}\sigma =0$. The non-vanishing flux is exactly the statement that the area class is a nonzero element of $H^2(S^2)$ ^{[Bott-Tu §I.4]}. The reason the flux is the same for every enclosing surface is that any two such surfaces are homologous in ${\mathbb{R}}^3\setminus \{0\}$, and the integral of a closed form depends only on the homology class.

Third, orientability is a genuine obstruction, not a bookkeeping nuisance. The vector surface element $\mathbf{n}dA=({\Phi }_u\times {\Phi }_v)dudv$ requires a globally consistent choice of normal field. On the Mobius band no such field exists: transporting $\mathbf{n}$ once around the core curve returns it reversed. A 2-form cannot be integrated over a non-orientable surface to give an orientation-respecting flux, because the local sign data ${\Phi }_u\times {\Phi }_v$ cannot be patched coherently. The orientation double cover restores integrability at the cost of counting each point twice.

Synthesis. Flux is the pairing of a 2-form with an oriented surface, and the pullback formula ${\int }_S\omega ={\int }_D{\Phi }^{\ast }\omega$ makes this pairing coordinate-free: the Jacobian minors that appear are precisely the components of ${\Phi }_u\times {\Phi }_v$, so the analytic surface integral and the algebraic pullback are one object viewed two ways. The interior product ${\omega }_F={\iota }_F(dV)$ is the dictionary between vector fields and 2-forms, and under it the divergence theorem becomes ${\int }_{\partial U}{\omega }_F={\int }_{U}d{\omega }_F$, a single instance of Stokes that simultaneously contains the curl theorem one degree below. The inverse-square flux $4\pi$ is not an accident of the sphere but the area class generating $H^2(S^2)$, closed yet not exact, which is why every enclosing surface returns the same number and why the field detects the puncture at the origin. Non-orientability, finally, is the precise condition under which the whole construction fails, locating flux squarely inside the homological theory of the de Rham complex rather than alongside it.

Full proof set Master

Proposition (the flux 2-form computes the classical surface integral). Let $\Phi :D\to S\subset {\mathbb{R}}^3$ orient $S$, let $F$ be a vector field near $S$, and let ${\omega }_F=F^{1}dy\wedge dz+F^{2}dz\wedge dx+F^{3}dx\wedge dy$. Then

$${\int }_S{\omega }_F={\iint }_{D}F(\Phi )\cdot ({\Phi }_u\times {\Phi }_v)dudv.$$

Proof. Write $\Phi =(x,y,z)$ with each coordinate a function of $(u,v)$. The pullback acts on a wedge of two coordinate differentials by replacing them with the corresponding Jacobian minor times $du\wedge dv$:

$${\Phi }^{\ast }(dy\wedge dz)=\frac{\partial (y,z)}{\partial (u,v)}du\wedge dv=(y_u{z}_v-y_v{z}_u)du\wedge dv,$$

and cyclically for $dz\wedge dx$ and $dx\wedge dy$. Hence

$${\Phi }^{\ast }{\omega }_F=(F^1(y_u{z}_v-y_v{z}_u)+F^2(z_u{x}_v-z_v{x}_u)+F^3(x_u{y}_v-x_v{y}_u))du\wedge dv,$$

with each $F^i$ evaluated at $\Phi (u,v)$. The three Jacobian minors are exactly the three components of the cross product ${\Phi }_u\times {\Phi }_v$, where ${\Phi }_u=(x_u,y_u,z_u)$ and ${\Phi }_v=(x_v,y_v,z_v)$:

$${\Phi }_u\times {\Phi }_v=(y_u{z}_v-y_v{z}_u,z_u{x}_v-z_v{x}_u,x_u{y}_v-x_v{y}_u).$$

Therefore the bracketed coefficient is the dot product $F(\Phi )\cdot ({\Phi }_u\times {\Phi }_v)$. Integrating ${\Phi }^{\ast }{\omega }_F$ over $D$ gives the stated double integral. $\square$

Proposition (orientation-reversal sign). With $-S$ the surface $S$ given the opposite orientation, ${\int }_{-S}{\omega }_F=-{\int }_S{\omega }_F$. Proof. An orientation-reversing reparametrisation has $\det (Dg)<0$. As in the Key theorem, the pullback contributes the signed factor $\det (Dg)$ while change of variables contributes $|\det (Dg)|=-\det (Dg)$; the two differ by exactly $-1$, so the integral negates. Equivalently, reversing orientation replaces $\mathbf{n}$ by $-\mathbf{n}$, and $F\cdot (-\mathbf{n})=-F\cdot \mathbf{n}$ pointwise. $\square$

Proposition (the unit-sphere area form is closed but not exact). Let $\sigma$ be the area 2-form on $S^2$. Then $d\sigma =0$ and there is no $\eta \in {\Omega }^1(S^2)$ with $d\eta =\sigma$. Proof. Closedness holds because $\sigma$ is a top-degree form on the 2-manifold $S^2$, so $d\sigma \in {\Omega }^3(S^2)=0$. If $\sigma =d\eta$ for some global 1-form $\eta$, then by Stokes' theorem on the closed (boundaryless) surface $S^2$, ${\int }_{S^2}\sigma ={\int }_{S^2}d\eta ={\int }_{\partial S^2}\eta =0$, contradicting ${\int }_{S^2}\sigma =4\pi >0$. Hence $\sigma$ represents a nonzero class in $H_{\mathrm{dR}}^2(S^2)$. $\square$

Connections Master

Integration on manifolds 03.04.03 supplies the pullback-and-partition-of-unity machinery that the surface integral specialises; the orientation-reversal sign proved there is the same sign that governs flux under orientation-reversing reparametrisation.

Stokes' theorem 03.04.05 contains the divergence theorem as the case ${\int }_{\partial U}{\omega }_F={\int }_{U}d{\omega }_F$ with $d{\omega }_F=(\mathrm{div}F)dV$, and contains the classical curl theorem as the analogous statement one degree lower.

De Rham cohomology 03.04.06 is where the $4\pi$ lives: the area form of $S^2$ is closed but not exact, generating $H^2(S^2)$, and the flux of the inverse-square field reads off this class.

The exterior derivative 03.04.04 produces $d{\omega }_F=(\mathrm{div}F)dV$, so the divergence operator of vector calculus is the exterior derivative seen through the 2-form dictionary.

The Yang-Mills action 03.07.05 integrates curvature 2-forms over surfaces to extract topological charges, a direct generalisation of integrating ${\omega }_F$ over $S$.

Historical & philosophical context Master

The flux of an inverse-square field through a closed surface was computed by Gauss in his 1813 memoir on the attraction of ellipsoids, where the solid-angle argument that yields $4\pi$ for an enclosing surface and $0$ otherwise first appears in print ^{[Gauss 1813]}. The line-integral counterpart now called the classical Stokes theorem was circulated by Stokes as the 1854 Smith's Prize examination question, having been communicated to him by Kelvin; the modern unification of these classical integral theorems into the single identity ${\int }_{\partial M}\omega ={\int }_{M}d\omega$ is due to the twentieth-century reformulation of integration in terms of differential forms, presented in its now-standard pedagogical shape by Spivak in 1965 ^{[Spivak Ch. 5]}. Shifrin's account threads the classical surface-integral computations to the form-theoretic statement at the level of a rigorous undergraduate course ^{[Shifrin Ch. 8]}, and the cohomological reading of the $4\pi$ as the generator of $H^2(S^2)$ is the entry point Bott and Tu use to compute the de Rham cohomology of spheres.

Bibliography Master

@book{shifrin2005multivariable,
  author    = {Shifrin, Theodore},
  title     = {Multivariable Mathematics: Linear Algebra, Multivariable Calculus, and Manifolds},
  publisher = {John Wiley \& Sons},
  year      = {2005}
}

@book{spivak1965calculus,
  author    = {Spivak, Michael},
  title     = {Calculus on Manifolds},
  publisher = {W. A. Benjamin},
  year      = {1965}
}

@incollection{gauss1813theoria,
  author    = {Gauss, Carl Friedrich},
  title     = {Theoria attractionis corporum sphaeroidicorum ellipticorum homogeneorum methodo nova tractata},
  booktitle = {Commentationes Societatis Regiae Scientiarum Gottingensis Recentiores},
  volume    = {2},
  year      = {1813}
}

@article{stokes1854smith,
  author    = {Stokes, George Gabriel},
  title     = {Smith's Prize Examination Paper (statement of the curl theorem)},
  journal   = {Cambridge University},
  year      = {1854}
}

@book{botttu1982differential,
  author    = {Bott, Raoul and Tu, Loring W.},
  title     = {Differential Forms in Algebraic Topology},
  publisher = {Springer},
  year      = {1982}
}