Singularities and the Smoothness Hypotheses

Why Smoothness Must Be Checked on the Surface

Stokes’ theorem can turn a circulation around a boundary curve into a surface integral of curl. When several surfaces share the same boundary, it can be tempting to choose whichever one makes the calculation easiest. That choice is legitimate only when the vector field satisfies the theorem’s hypotheses on the chosen surface.

A field may be perfectly well defined and smooth along the boundary curve but fail to be defined at a point inside a proposed spanning surface. In that case, the line integral can still exist, but Stokes’ theorem cannot be applied to that surface. The distinction matters: a calculation using the curl away from the singularity does not repair the missing hypothesis at the singularity.

The essential check: Before applying Stokes’ theorem, verify that the vector field is defined and continuously differentiable on an open set containing the entire surface—not merely on its boundary.

Stokes’ Theorem and Its Hypotheses

Let \(S\) be an oriented, piecewise smooth surface with positively oriented boundary curve \(C=\partial S\). Let \(\mathbf{F}\) be a vector field whose component functions have continuous first partial derivatives on an open set containing \(S\). Then Stokes’ theorem states that

\oint_C \mathbf{F}\cdot d\mathbf{r} = \iint_S (\nabla\times\mathbf{F})\cdot\mathbf{n}\,dS.

Here \(\mathbf{n}\) is the unit normal determined by the chosen orientation of \(S\), and the orientation of \(C\) is induced from that surface orientation by the right-hand rule. The theorem equates the circulation of the field along the boundary with the flux of its curl through the surface.

The smoothness condition is stronger than requiring \(\mathbf{F}\) to have a value at each point of \(S\). The first partial derivatives used to form \(\nabla\times\mathbf{F}\) must exist and be continuous on an open neighborhood of the surface. In particular, a point where the formula for \(\mathbf{F}\) has a zero denominator cannot be ignored just because it is a single point or lies away from the boundary.

Surface replacement rule: Two oriented surfaces with the same oriented boundary do not automatically give two valid applications of Stokes’ theorem. Check the hypotheses separately for each surface.

A Surface That Stays Away from a Singularity

First consider a field with a genuine singular set, but choose a surface that avoids it. This example shows what a valid smoothness check looks like.

Worked Example: A Rectangle in a Smooth Region

Let \(\mathbf{F}(x,y,z)=(0,0,y/x)\), defined where \(x\ne 0\). Let \(S\) be the rectangle parametrized by \(\mathbf{r}(y,z)=(2,y,z)\), with \(0\le y\le 1\) and \(0\le z\le 3\), oriented by the normal \(\mathbf{n}=(1,0,0)\). Check the hypotheses and verify Stokes’ theorem.

The singular set of the formula is the plane \(x=0\). The whole rectangle lies in \(x=2\). For example, the open slab \(1<x<3\) contains the rectangle and contains no singular points. In that slab, the components of \(\mathbf{F}\) have continuous first partial derivatives, so the smoothness requirement is satisfied.

Compute the curl component by component:

\nabla\times\mathbf{F} = \left( \frac{\partial}{\partial y}\frac{y}{x}, -\frac{\partial}{\partial x}\frac{y}{x}, 0 \right) = \left(\frac{1}{x},\frac{y}{x^2},0\right).

The parametrization has \(\mathbf{r}_y=(0,1,0)\) and \(\mathbf{r}_z=(0,0,1)\), so \(\mathbf{r}_y\times\mathbf{r}_z=(1,0,0)\), consistent with the chosen orientation. On \(S\), \(x=2\), and hence

(\nabla\times\mathbf{F})\cdot(\mathbf{r}_y\times\mathbf{r}_z) = \left(\frac12,\frac{y}{4},0\right)\cdot(1,0,0) = \frac12.

The curl flux is therefore

\iint_S(\nabla\times\mathbf{F})\cdot\mathbf{n}\,dS = \int_0^1\int_0^3 \frac12\,dz\,dy = \int_0^1 \frac32\,dy = \frac32.

Now traverse the boundary in the orientation induced by the normal \(+\mathbf{i}\). In the \(yz\)-coordinates, the four sides run from \((0,0)\) to \((1,0)\), then to \((1,3)\), then to \((0,3)\), and then back to \((0,0)\). Since \(\mathbf{F}\cdot d\mathbf{r}=(y/2)\,dz\) along this rectangle, the two sides with \(dz=0\) contribute zero. The upward side at \(y=1\) contributes \(\int_0^3 \tfrac12\,dz=\tfrac32\), and the downward side at \(y=0\) contributes zero. Thus the circulation is \(\tfrac32\), matching the curl flux.

The check at the end is \(\oint_C\mathbf{F}\cdot d\mathbf{r}=\tfrac32=\iint_S(\nabla\times\mathbf{F})\cdot\mathbf{n}\,dS\). Both the integral calculation and the theorem’s smoothness requirement hold on this surface.

The Punctured-Plane Field

Consider the field

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

This field is undefined on the \(z\)-axis, where \(x=y=0\). Away from that axis, its components have continuous first partial derivatives. Its curl there is zero. Indeed, writing the components as \(P=-y/(x^2+y^2)\), \(Q=x/(x^2+y^2)\), and \(R=0\), the only potentially nonzero curl component is

\frac{\partial Q}{\partial x}-\frac{\partial P}{\partial y} = \frac{y^2-x^2}{(x^2+y^2)^2} - \frac{y^2-x^2}{(x^2+y^2)^2} = 0.

This calculation establishes that the curl vanishes wherever the field is defined. It does not establish that the field is defined or smooth on a surface that meets the \(z\)-axis.

Worked Example: Why the Flat Disk Is Not a Valid Replacement

Let \(C\) be the unit circle in the \(xy\)-plane, traversed counterclockwise as viewed from above. Compute its circulation for \(\mathbf{F}\), and test whether Stokes’ theorem can be applied using the flat disk it bounds.

Parametrize the curve by \(\mathbf{r}(t)=(\cos t,\sin t,0)\) for \(0\le t\le 2\pi\). Then \(\mathbf{r}'(t)=(-\sin t,\cos t,0)\), and on the curve \(x^2+y^2=1\), so \(\mathbf{F}(\mathbf{r}(t))=(-\sin t,\cos t,0)\). Therefore

\mathbf{F}(\mathbf{r}(t))\cdot\mathbf{r}'(t) = (-\sin t)(-\sin t)+(\cos t)(\cos t) = 1.

Consequently, the circulation is

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

The flat disk \(D\) bounded by \(C\) contains the origin. At that point, \(x^2+y^2=0\), so \(\mathbf{F}\) is undefined. No open set containing the whole disk can also be a region where \(\mathbf{F}\) is continuously differentiable. Thus Stokes’ theorem does not apply to \(D\).

A tempting but incorrect calculation would integrate the curl, which is zero away from the origin, and conclude that the disk’s curl flux is zero. That calculation treats the curl as if it were defined on the whole disk. It is not: the field, and therefore its curl, is undefined at the point where the disk meets the singular axis. The correct conclusion is not that the circulation equals zero; it is that this surface is not eligible for Stokes’ theorem.

Why the Hypothesis Matters

The theorem relates two integrals because the field behaves smoothly throughout the surface. In a smooth region, local circulation and curl fit together consistently as one moves across the surface. A singularity breaks that argument: there may be circulation around the boundary even when the curl vanishes at every point where it can be computed.

The punctured-plane example makes the danger precise. The circle’s circulation is \(2\pi\), while the curl is zero wherever the field is defined. There is no contradiction with Stokes’ theorem, because the disk needed for that proposed application contains a point outside the field’s domain. A theorem’s hypotheses are part of its conclusion’s guarantee; they cannot be replaced by a calculation that happens to look plausible.

When replacing one spanning surface by another, compare the surfaces with the singular set. The same boundary curve and compatible orientation are not enough. The field must be continuously differentiable on an open neighborhood of the particular surface used. A replacement that avoids a singularity might be admissible if it still has the required boundary and orientation; one that intersects a singularity is not.

Identify the domain.
Find where each component of the field is defined; inspect denominators, logarithms, square roots, and other restrictions.

Locate the singular set.
Determine where the field or its first partial derivatives fail to exist or be continuous.

Check the whole surface.
Verify that the chosen surface lies in an open region on which the field is continuously differentiable.

Then check orientation and compute.
Only after the hypotheses hold should the curl-flux integral be equated with the boundary circulation.

Common Mistake: Checking Only the Boundary

A common error is to observe that the field is smooth along \(C\), calculate its circulation, and then choose a convenient surface without checking the interior. For the punctured-plane field, this reasoning might run: “The field is smooth on the unit circle, the disk is easy, and the curl is zero, so the circulation is zero.” The first observation is true, but the conclusion does not follow. The disk contains the origin, where the field is undefined, so the theorem’s hypotheses fail.

The correction is to separate the questions. The line integral asks whether the field is defined along the curve. Stokes’ theorem asks whether the field is continuously differentiable on an open neighborhood of the entire spanning surface. The first condition may hold while the second does not. Always make the surface-wide check before substituting a curl integral for a circulation.

Check Your Understanding

For each question, identify the relevant domain or hypothesis before deciding whether Stokes’ theorem applies.

State the smoothness condition on \(\mathbf{F}\) required for the usual Stokes’ theorem.
Why is smoothness along the boundary curve alone insufficient?
For the punctured-plane field, what is its singular set, and why does the unit disk meet it?
The field \(\mathbf{F}(x,y,z)=(0,0,y/x)\) is considered on a surface contained in the plane \(x=2\). Give an open slab containing that surface on which the field is smooth.
What is the specific invalid step in concluding that the unit circle’s circulation is zero from the vanishing curl of the punctured-plane field?