Zero Curl, Circulation, and Punctured Domains

Zero Curl Does Not Always Mean Zero Circulation

A vector field’s curl describes a local property: it measures rotation in the field near a point. Circulation around a closed curve is a global quantity, obtained by accumulating the field’s tangential component all the way around that curve. On a domain with a hole, those two facts can differ in a striking way: the curl can vanish at every point of the domain, while a loop around the hole still has nonzero circulation.

The key is that a field must be defined on the region where we apply theorems about it. If an axis is excluded, a curve may lie entirely in the field’s domain even though a disk bounded by that curve crosses the excluded axis. That distinction is enough to block a tempting but invalid use of Stokes’ Theorem.

Central idea: Zero curl is a local condition. It does not by itself guarantee zero circulation around every closed curve when the domain has a hole.

Definitions: Curl-Free Fields and Circulation

Let \(D\) be an open subset of three-dimensional space and let \(\mathbf F=(P,Q,R)\) be a vector field whose component functions have continuous first partial derivatives on \(D\). The curl of \(\mathbf F\) is

\nabla\times\mathbf F = \left( \frac{\partial R}{\partial y}-\frac{\partial Q}{\partial z}, \frac{\partial P}{\partial z}-\frac{\partial R}{\partial x}, \frac{\partial Q}{\partial x}-\frac{\partial P}{\partial y} \right).

The field is curl-free on \(D\) if \(\nabla\times\mathbf F=\mathbf 0\) at every point of \(D\). If \(C\) is a piecewise smooth, oriented curve contained in \(D\), the circulation of \(\mathbf F\) along \(C\) is the line integral \(\int_C\mathbf F\cdot d\mathbf r\). For a parametrization \(\mathbf r(t)\), \(a\le t\le b\), this is \(\int_a^b\mathbf F(\mathbf r(t))\cdot\mathbf r'(t)\,dt\).

A field is conservative on \(D\) if there is a scalar function \(\phi\) defined on \(D\) such that \(\mathbf F=\nabla\phi\). For a conservative field, the chain rule gives \(\mathbf F(\mathbf r(t))\cdot\mathbf r'(t)=\frac{d}{dt}\phi(\mathbf r(t))\). Consequently, its integral around any closed piecewise smooth curve is zero. The reverse implication, from curl-free to conservative, needs an appropriate condition on the domain; it does not hold for every domain with holes.

A Field Around an Excluded Axis

Consider the domain obtained by removing the entire \(z\)-axis:

D=\{(x,y,z)\in\mathbb R^3:x^2+y^2>0\}.

On this domain define

\mathbf F(x,y,z) = \left( -\frac{y}{x^2+y^2}, \frac{x}{x^2+y^2}, 0 \right).

The denominator is nonzero at every point of \(D\), so the field is smooth there. It is not defined on the excluded axis. In particular, the formula does not describe a field on all of three-dimensional space, and statements about its curl or line integrals apply only where it is defined.

Worked Example: Verify That the Field Is Curl-Free

Write \(P=-y/(x^2+y^2)\), \(Q=x/(x^2+y^2)\), and \(R=0\). Since the components do not depend on \(z\), the first two components of the curl are

\frac{\partial R}{\partial y}-\frac{\partial Q}{\partial z}=0-0=0, \qquad \frac{\partial P}{\partial z}-\frac{\partial R}{\partial x}=0-0=0.

For the third component, use the quotient rule:

\frac{\partial Q}{\partial x} = \frac{(x^2+y^2)(1)-x(2x)}{(x^2+y^2)^2} = \frac{y^2-x^2}{(x^2+y^2)^2},

\frac{\partial P}{\partial y} = \frac{-(x^2+y^2)(1)-(-y)(2y)}{(x^2+y^2)^2} = \frac{y^2-x^2}{(x^2+y^2)^2}.

Subtracting gives \(\frac{\partial Q}{\partial x}-\frac{\partial P}{\partial y}=0\). Thus every component of the curl vanishes, and \(\nabla\times\mathbf F=\mathbf 0\) throughout \(D\). The calculation makes no claim at points on the axis, which do not belong to \(D\).

Circulation Around a Circle Enclosing the Axis

Take a circle of radius \(a>0\) in the plane \(z=h\), centered on the excluded axis. Orient it counterclockwise when viewed from the positive \(z\)-direction. A parametrization is \(\mathbf r(t)=(a\cos t,a\sin t,h)\), for \(0\le t\le2\pi\). Every point of this curve lies in \(D\).

Worked Example: Circulation Around a Circle

Along the parametrized circle, substitute \(x=a\cos t\) and \(y=a\sin t\) into the field:

\mathbf F(\mathbf r(t)) = \left( -\frac{a\sin t}{a^2}, \frac{a\cos t}{a^2}, 0 \right) = \left( -\frac{\sin t}{a}, \frac{\cos t}{a}, 0 \right).

The tangent vector is \(\mathbf r'(t)=(-a\sin t,a\cos t,0)\). Their dot product is

\mathbf F(\mathbf r(t))\cdot\mathbf r'(t) = \left(-\frac{\sin t}{a}\right)(-a\sin t) + \left(\frac{\cos t}{a}\right)(a\cos t) = \sin^2 t+\cos^2 t = 1.

Therefore the circulation is

\int_C\mathbf F\cdot d\mathbf r = \int_0^{2\pi}1\,dt = 2\pi.

As a check, the integrand is \(1\) at every parameter value, so its integral over an interval of length \(2\pi\) is indeed \(2\pi\). The result does not depend on the radius \(a\) or height \(h\). In particular, this is a closed curve in a region where the field is curl-free, but its circulation is not zero.

A Second Loop Around the Hole

The same nonzero circulation can be verified on a square, without relying on a circular parametrization. This calculation also illustrates how contributions from different parts of a closed path add together.

Worked Example: Circulation Around a Square

Take the square with vertices \((1,-1,0)\), \((1,1,0)\), \((-1,1,0)\), and \((-1,-1,0)\), traversed in that order. This is counterclockwise when viewed from above, and its boundary avoids the excluded axis. On the right edge, \(x=1\), \(y\) runs from \(-1\) to \(1\), and \(dx=0\). Thus \(\mathbf F\cdot d\mathbf r=Q\,dy\), and

\int_{-1}^{1}\frac{1}{1+y^2}\,dy = \arctan(1)-\arctan(-1) = \frac{\pi}{4}-\left(-\frac{\pi}{4}\right) = \frac{\pi}{2}.

On the top edge, \(y=1\), \(x\) runs from \(1\) to \(-1\), and \(\mathbf F\cdot d\mathbf r=P\,dx\). Its contribution is

\int_{1}^{-1}-\frac{1}{x^2+1}\,dx = -\bigl(\arctan(-1)-\arctan(1)\bigr) = \frac{\pi}{2}.

On the left edge, \(x=-1\), \(y\) runs from \(1\) to \(-1\). Here \(Q=-1/(1+y^2)\), so the contribution is

\int_{1}^{-1}-\frac{1}{1+y^2}\,dy = -\bigl(\arctan(-1)-\arctan(1)\bigr) = \frac{\pi}{2}.

On the bottom edge, \(y=-1\), \(x\) runs from \(-1\) to \(1\). Here \(P=1/(x^2+1)\), giving

\int_{-1}^{1}\frac{1}{x^2+1}\,dx = \arctan(1)-\arctan(-1) = \frac{\pi}{2}.

Adding the four edge contributions gives \(\frac{\pi}{2}+\frac{\pi}{2}+\frac{\pi}{2}+\frac{\pi}{2}=2\pi\). This checks the total circulation and shows that the nonzero result is not specific to a circular path.

Why a Nearby Closed Curve Can Have Zero Circulation

Not every closed curve in \(D\) has nonzero circulation. For instance, a circle that stays entirely in the half-space \(x>0\) can be enclosed in a region that does not touch the removed axis. On that half-space, an ordinary single-valued potential is available.

Worked Example: A Circle That Does Not Enclose the Axis

Let \(c>b>0\), and consider the circle \(x=c+b\cos t\), \(y=b\sin t\), \(z=0\), for \(0\le t\le2\pi\). Since \(x\ge c-b>0\), the entire circle lies in \(x>0\). Define \(\phi(x,y,z)=\arctan(y/x)\) on that half-space. Differentiating gives

\frac{\partial\phi}{\partial x} = -\frac{y}{x^2+y^2}, \qquad \frac{\partial\phi}{\partial y} = \frac{x}{x^2+y^2}, \qquad \frac{\partial\phi}{\partial z}=0.

Thus \(\nabla\phi=\mathbf F\) there. Along the circle, the chain rule gives \(\mathbf F(\mathbf r(t))\cdot\mathbf r'(t)=\frac{d}{dt}\phi(\mathbf r(t))\). At \(t=0\) and \(t=2\pi\), the circle is at \((c+b,0,0)\), where \(\phi=0\). Therefore

\int_C\mathbf F\cdot d\mathbf r = \int_0^{2\pi}\frac{d}{dt}\phi(\mathbf r(t))\,dt = \phi(\mathbf r(2\pi))-\phi(\mathbf r(0)) = 0-0 = 0.

The check is the equality of the potential’s endpoint values on the closed path. This example contrasts with the circles around the axis: on \(x>0\), the angle \(\arctan(y/x)\) is single-valued and provides a potential, whereas following a loop around the axis would require the angle to return after changing by \(2\pi\).

Why the Curl Calculation Does Not Force Zero Circulation

For a field that is continuously differentiable on an open, simply connected domain, the standard curl criterion says that zero curl implies the field is conservative. The domain matters: \(D\), with the \(z\)-axis removed, is not simply connected. A loop around the axis cannot be continuously contracted to a point while staying in \(D\). Thus the usual sufficient condition for a global potential does not apply here.

In fact, the circle calculation proves that \(\mathbf F\) cannot have a single-valued potential on all of \(D\). If \(\mathbf F=\nabla\phi\) on \(D\), the chain rule would make the integral around every closed curve equal to zero. But the circle has integral \(2\pi\), a contradiction. The field does have potentials on suitable smaller regions, as the half-space example showed; the obstruction is to one potential defined consistently throughout the punctured domain.

Common Mistake: Applying Stokes’ Theorem Across the Hole

A tempting argument is: the circle has a disk as its spanning surface, the curl is zero, so Stokes’ Theorem gives zero circulation. The problem is that Stokes’ Theorem requires the field to be continuously differentiable on an open region containing the surface. The disk bounded by the circle centered on the axis intersects the excluded axis, where \(\mathbf F\) is undefined. The hypotheses fail, even though the field is smooth along the boundary circle.

Correction: Before applying a surface theorem, check that the field is defined and sufficiently smooth on the entire surface, not merely on its boundary. For the disk crossing the excluded axis, Stokes’ Theorem cannot be used to conclude that the circulation is zero.

There is no conflict between the zero curl on \(D\) and the nonzero circulation around the axis. The curl calculation is valid at each point of \(D\); it does not fill in the missing axis or make a spanning disk that crosses it a valid surface for Stokes’ Theorem.

Why This Distinction Matters

In flow language, the example models a field circulating around an excluded axis. Away from the axis there is no local curl, yet a particle moving around the axis accumulates a nonzero tangential line integral. This illustrates why local measurements and global circulation answer different questions. A curl test can identify local rotation where the field is defined, but it cannot by itself determine circulation around every possible loop if the domain has holes.

When analyzing a curl-free field, keep both the formula and its domain in view. First check where the field is defined. Then calculate the curl on that domain. For a closed curve of interest, evaluate circulation directly or establish that a potential exists on a region containing the curve. Do not infer zero circulation solely from zero curl unless the domain conditions that support that conclusion are satisfied.

Check Your Understanding

Use the definitions and calculations in this tutorial to answer the following questions.

Why is the field \(\mathbf F=(-y/(x^2+y^2),x/(x^2+y^2),0)\) not defined on the \(z\)-axis?
What is the circulation around a counterclockwise circle centered on the excluded axis, and why does it not depend on the circle’s radius?
Why does a nonzero integral around a closed curve rule out a single-valued potential on the entire domain?
What assumption needed for Stokes’ Theorem fails when the spanning disk crosses the excluded axis?
Why can a closed curve contained in \(x>0\) have zero circulation for this same field?