Two Routes to the Same Circulation
A circulation integral measures how a vector field acts along an oriented closed curve. If the curve has a convenient parametrization, it may be natural to evaluate the line integral directly. But if the curve bounds a simple surface, Stokes’ Theorem may replace that boundary calculation with a surface integral of the curl.
The useful question is not whether either route is always better. It is which route has the simpler ingredients for the particular field and boundary: a manageable parametrization and line-integral integrand, or a manageable curl and spanning surface.
Direct integration uses a parametrization \(\mathbf{r}(t)\) of \(C\), with its direction matching the stated orientation: $$ \oint_C \mathbf{F}\cdot d\mathbf{r} = \int_a^b \mathbf{F}(\mathbf{r}(t))\cdot\mathbf{r}'(t)\,dt. $$ The Stokes route instead computes the curl, chooses a spanning surface with the compatible boundary orientation, and integrates the normal component of that curl. Both routes require attention to the same oriented boundary; the surface route does not permit choosing a surface on which the field or its derivatives fail to meet the theorem’s hypotheses.
Why the Surface Integral Gives the Boundary Integral
A short calculation explains why curl flux is the relevant surface quantity. Consider a smooth parametrized surface \(\mathbf{X}(u,v)\) over a region \(D\) in the \(uv\)-plane, with orientation \(\mathbf{X}_u\times\mathbf{X}_v\). Define $$ A(u,v)=\mathbf{F}(\mathbf{X}(u,v))\cdot\mathbf{X}_u, \qquad B(u,v)=\mathbf{F}(\mathbf{X}(u,v))\cdot\mathbf{X}_v. $$ Along the boundary of the parameter region, the circulation on the surface is the line integral of \(A\,du+B\,dv\). Green’s Theorem converts this boundary integral to an integral over \(D\) of \(\partial B/\partial u-\partial A/\partial v\).
By the chain rule, the terms involving the second derivatives of \(\mathbf{X}\) cancel: $$ \frac{\partial B}{\partial u}-\frac{\partial A}{\partial v} = (D\mathbf{F}(\mathbf{X})\mathbf{X}_u)\cdot\mathbf{X}_v - (D\mathbf{F}(\mathbf{X})\mathbf{X}_v)\cdot\mathbf{X}_u = (\nabla\times\mathbf{F})(\mathbf{X})\cdot(\mathbf{X}_u\times\mathbf{X}_v). $$ The last equality follows by expanding the components of the curl and the cross product. Since \(\mathbf{X}_u\times\mathbf{X}_v\) is the oriented surface-area vector, integrating this identity over \(D\) gives the curl-flux integral over \(S\). Green’s Theorem therefore explains, for such a parametrized patch, why the boundary circulation equals the flux of curl.
Worked Example: A Circle and a Polynomial Field
Worked Example: A Circle and a Polynomial Field
Let \(C\) be the counterclockwise circle \(x^2+y^2=4\) in the \(xy\)-plane, viewed from above, and let $$ \mathbf{F}(x,y,z)=\bigl(-y(x^2+y^2),\ x(x^2+y^2),\ 0\bigr). $$ Compute the circulation directly and by using the disk bounded by \(C\).
Direct boundary calculation. Parametrize the circle by $$ \mathbf{r}(t)=(2\cos t,2\sin t,0),\qquad 0\leq t\leq 2\pi. $$ Then \(\mathbf{r}'(t)=(-2\sin t,2\cos t,0)\). On this curve, \(x^2+y^2=4\), so $$ \mathbf{F}(\mathbf{r}(t))=(-8\sin t,8\cos t,0). $$ The dot product is $$ \mathbf{F}(\mathbf{r}(t))\cdot\mathbf{r}'(t) =16\sin^2 t+16\cos^2 t=16. $$ Thus the circulation is \(\int_0^{2\pi}16\,dt=32\pi\).
Curl-flux calculation. Writing \(\mathbf{F}=(P,Q,R)\), the \(z\)-component of the curl is \(Q_x-P_y\). Here $$ Q_x=3x^2+y^2,\qquad P_y=-x^2-3y^2, $$ so \((\nabla\times\mathbf{F})_z=4(x^2+y^2)\). The disk has upward unit normal \(\mathbf{k}\), so polar coordinates give $$ \iint_S(\nabla\times\mathbf{F})\cdot\mathbf{k}\,dA =\int_0^{2\pi}\int_0^2 4r^2\,r\,dr\,d\theta =2\pi\left[r^4\right]_0^2 =32\pi. $$ Both methods give \(32\pi\). The disk route avoids carrying the full vector field through the boundary parametrization; in this example the radial curl also makes the surface integral immediate.
Worked Example: A Square Boundary
Worked Example: A Square Boundary
Use the same field $$ \mathbf{F}(x,y,z)=\bigl(-y(x^2+y^2),\ x(x^2+y^2),\ 0\bigr), $$ but let \(C\) be the counterclockwise boundary of the square \([-1,1]\times[-1,1]\) in the \(xy\)-plane. Compare the four boundary integrals with the curl flux across the square.
On the bottom edge, \(y=-1\), \(x\) runs from \(-1\) to \(1\), and \(dy=0\). Thus \(P=x^2+1\), and this edge contributes $$ \int_{-1}^{1}(x^2+1)\,dx=\frac{2}{3}+2=\frac{8}{3}. $$ On the right edge, \(x=1\), \(y\) runs from \(-1\) to \(1\), and \(dx=0\). Here \(Q=1+y^2\), giving the contribution $$ \int_{-1}^{1}(1+y^2)\,dy=2+\frac{2}{3}=\frac{8}{3}. $$
On the top edge, \(y=1\), \(x\) runs from \(1\) to \(-1\). Since \(P=-(x^2+1)\), its contribution is $$ \int_1^{-1}-(x^2+1)\,dx=\frac{8}{3}. $$ On the left edge, \(x=-1\), \(y\) runs from \(1\) to \(-1\). Since \(Q=-(1+y^2)\), this edge contributes $$ \int_1^{-1}-(1+y^2)\,dy=\frac{8}{3}. $$ Adding the four edge contributions gives \(32/3\).
For the surface route, the upward curl component is again \(4(x^2+y^2)\). Therefore $$ \iint_{[-1,1]^2}4(x^2+y^2)\,dA =4\left( \int_{-1}^1\int_{-1}^1 x^2\,dy\,dx + \int_{-1}^1\int_{-1}^1 y^2\,dy\,dx \right). $$ Each double integral equals \((2/3)(2)=4/3\), so the total is \(4(4/3+4/3)=32/3\). The results agree. Here the surface calculation is shorter than evaluating four separate edges, though the direct route remains quite manageable.
Worked Example: A Unit Circle and an Exponential Field
Worked Example: A Unit Circle and an Exponential Field
Let \(C\) be the counterclockwise unit circle in the \(xy\)-plane, and let $$ \mathbf{F}(x,y,z)=\bigl(-y e^{x^2+y^2},\ x e^{x^2+y^2},\ 0\bigr). $$ Compute its circulation using the circle and then using the flat disk it bounds.
On the parametrization \(\mathbf{r}(t)=(\cos t,\sin t,0)\), \(0\leq t\leq2\pi\), the field is \((-e\sin t,e\cos t,0)\) and \(\mathbf{r}'(t)=(-\sin t,\cos t,0)\). Their dot product is \(e(\sin^2t+\cos^2t)=e\). Hence $$ \oint_C\mathbf{F}\cdot d\mathbf{r}=\int_0^{2\pi}e\,dt=2\pi e. $$
For the disk, compute the upward component of the curl: $$ (\nabla\times\mathbf{F})_z =\frac{\partial}{\partial x}(x e^{x^2+y^2}) -\frac{\partial}{\partial y}(-y e^{x^2+y^2}) =2e^{x^2+y^2}(1+x^2+y^2). $$ Using polar coordinates and then \(t=r^2\), with \(dt=2r\,dr\), the flux is $$ \int_0^{2\pi}\int_0^1 2e^{r^2}(1+r^2)r\,dr\,d\theta =2\pi\int_0^1 e^t(1+t)\,dt =2\pi\left[t e^t\right]_0^1 =2\pi e. $$ The direct calculation is especially short because the field has constant tangential strength along the circle. The disk calculation is also efficient because the curl is radial in its \(z\)-component. A simple spanning surface helps only when its curl flux is actually easy to integrate.
How to Decide Which Route Is Shorter
Before calculating, inspect both possible integrands. A direct line integral is often attractive when the curve has a simple parametrization, when the field restricted to the curve simplifies, or when \(\mathbf{F}(\mathbf{r}(t))\cdot\mathbf{r}'(t)\) quickly reduces to a constant or a familiar expression. Curl flux is often attractive when the boundary encloses a disk or planar patch, the curl has a simple normal component, or the surface integral has symmetry that removes terms.
A simple-looking boundary does not guarantee a simpler curl calculation, and a simple-looking curl does not guarantee that the surface integral is straightforward. Stokes’ Theorem gives equivalent routes, not a promise that one route will always be shorter. It also permits replacing a curved spanning surface by another suitable surface with the same oriented boundary, provided the field is defined and has continuous first partial derivatives throughout an open region containing the surface used.
Common Pitfall: Integrating the Field Instead of Its Curl
In a Stokes calculation, the surface integrand is \((\nabla\times\mathbf{F})\cdot\mathbf{n}\), not \(\mathbf{F}\cdot\mathbf{n}\). For the polynomial field in the first example, the \(z\)-component of \(\mathbf{F}\) is zero. A mistaken calculation might therefore assert $$ \iint_S\mathbf{F}\cdot\mathbf{k}\,dA=0 \quad\Longrightarrow\quad \oint_C\mathbf{F}\cdot d\mathbf{r}=0. $$ That conclusion is wrong: the surface integral on the left is not the Stokes integrand. The correct upward curl component is \(4(x^2+y^2)\), whose disk integral is \(32\pi\), matching the direct boundary calculation. The relevant surface quantity is the flux of the curl.
Also check the theorem’s requirements before using it: the curve must be the oriented boundary of the surface, and the field must be sufficiently smooth on a region containing that surface. A convenient-looking surface is not valid if it passes through a point where the field is undefined.
Check Your Understanding
For each question, decide what information determines whether direct integration or curl flux is the more efficient route.
- What smoothness and boundary conditions are needed to apply Stokes’ Theorem?
- For a counterclockwise circle in the \(xy\)-plane, what is the natural orientation of a flat disk spanning it?
- In the square example, why does each edge contribute a nonzero amount even though the field has no \(z\)-component?
- What quantity must be integrated over a surface in the Stokes route: the field or its curl?
- Give one feature of a boundary calculation and one feature of a surface calculation that can make that route shorter.