Following a Boundary Through Its Corners
A boundary need not be a single smooth curve. It may consist of straight segments joined to one another, smooth arcs joined at corners, or a mixture of both. Green’s theorem still applies to many such boundaries, provided the boundary pieces fit together into a suitably oriented, simple closed curve and the field satisfies the required smoothness conditions.
The practical task is to follow the boundary without losing track of its direction. At each corner, one piece ends and the next begins. The direction changes abruptly, but the line integral is assembled in the same way: compute the contribution from each piece in its assigned direction, then add the contributions.
For a vector field \(\mathbf{F}=\langle P,Q\rangle\), the circulation integral along a parametrized piece \(\mathbf{r}(t)=\langle x(t),y(t)\rangle\), \(a\le t\le b\), is
For a boundary made of \(m\) consecutive pieces \(C_1,\ldots,C_m\), the integral is the sum of the integrals over those pieces. Each piece must be traversed in the direction in which the full boundary is being followed.
Orientation and the Green’s Theorem Conditions
For a simple closed boundary, the positive orientation means that the region enclosed by the boundary stays on the left as you travel around it. For a region without holes, this is counterclockwise travel. The direction must continue through every corner: finishing one piece does not restart the traversal or reverse the next piece.
Here, “simple” means that the boundary does not cross itself. The hypotheses on \(P\) and \(Q\) ensure that the curl density in the double integral is continuous and that Green’s theorem applies. If the boundary is traversed in the negative direction instead, the line integral changes sign. The region and the integrand on the right do not change.
Why Corners Add No Extra Term
Green’s theorem relates an integral around the entire boundary to an integral over the enclosed area. A corner changes the direction of the tangent, but it does not add a separate corner contribution to the line integral. The reason is that the line integral is accumulated along the pieces of the curve; a single joining point has no length over which an additional contribution could accumulate.
One way to see this is to round each corner within a small neighborhood. The original and rounded boundaries differ only on short pieces whose lengths tend to zero as the neighborhoods shrink. Since \(P\) and \(Q\) are continuous near the boundary, they are bounded there. If their absolute values are at most \(M\), then a changed boundary piece of length \(L\) contributes an amount whose absolute value is at most \(\sqrt{2}ML\): this follows from \(|P\,dx+Q\,dy|\le \sqrt{P^2+Q^2}\,ds\). Thus the difference in boundary integrals tends to zero with \(L\).
The small area added or removed when rounding also tends to zero. The curl density is bounded near the boundary, so its integral over that small area tends to zero as well. Consequently, the limiting boundary and area integrals agree just as they do for the rounded boundaries. This explains why ordinary corners require no correction term. It also explains why the smoothness conditions on the pieces and the field matter: the argument relies on bounded field values and well-behaved, finite-length pieces near the joins.
Worked Example: A Triangle with a Slanted Edge
Worked Example: A Triangle with a Slanted Edge
Let \(C\) be the triangle with vertices \((0,0)\), \((2,0)\), and \((0,1)\), traversed in that order and then back to \((0,0)\). Compute \(\oint_C x^2\,dx+xy\,dy\), and check the result using Green’s theorem.
The listed order is counterclockwise: the triangle remains on the left as each edge is traversed. Write \(P=x^2\) and \(Q=xy\). Split the boundary into three directed edges.
On the bottom edge, use \(x=t,\ y=0\), with \(0\le t\le2\). Then \(dx=dt\), \(dy=0\), so
On the slanted edge from \((2,0)\) to \((0,1)\), use \(x=2-2t,\ y=t\), with \(0\le t\le1\). Thus \(dx=-2\,dt\) and \(dy=dt\). Substitution gives
On the vertical edge from \((0,1)\) to \((0,0)\), \(x=0\) and \(dx=0\). Both terms in the integrand vanish, so the contribution is \(0\). Adding the three directed contributions gives
For the check, \(\frac{\partial Q}{\partial x}=y\) and \(\frac{\partial P}{\partial y}=0\). The region is described by \(0\le x\le2\) and \(0\le y\le1-\frac{x}{2}\). Therefore
The area integral agrees with the sum over the three edges.
Working with an Arc Joined to Line Segments
An arc is handled just like a line segment once its parametrization and direction are specified. At the point where the arc meets a segment, note the endpoint of the arc and the starting point of the segment. They must be the same point for the pieces to form a continuous traversal. The tangent directions need not match: that mismatch is precisely what makes the join a corner.
Record the starting and ending point of every segment and arc so that consecutive endpoints match.
Trace the whole boundary once, keeping the enclosed region on your left for positive orientation.
Use a parametrization that runs from its listed start to its listed end, then add the resulting contributions.
Integrate the curl density over the region and compare it with the sum along the boundary.
Worked Example: A Quarter-Circle Arc and Two Radii
Let \(C\) bound the quarter-disk \(x^2+y^2\le4\), \(x\ge0\), \(y\ge0\). Traverse from \((0,0)\) to \((2,0)\), then along the quarter-circle to \((0,2)\), then down the vertical axis to \((0,0)\). Compute \(\oint_C -y\,dx+x\,dy\).
This is positive orientation. On the horizontal radius \(y=0\) and \(dy=0\), so the contribution is \(0\). Parametrize the arc by \(x=2\cos t,\ y=2\sin t\), \(0\le t\le\frac{\pi}{2}\). Then \(dx=-2\sin t\,dt\) and \(dy=2\cos t\,dt\), giving
On the vertical radius \(x=0\) and \(dx=0\), so that contribution is also \(0\). The full boundary integral is \(2\pi\). For the check, \(\frac{\partial Q}{\partial x}=1\) and \(\frac{\partial P}{\partial y}=-1\), so the curl density is \(2\). The quarter-disk has area \(\frac14\pi(2)^2=\pi\), and hence
which matches the sum over the radii and arc.
Worked Example: A Semicircle Joined to Its Diameter
Let \(C\) consist of the upper unit semicircle, traversed from \((1,0)\) to \((-1,0)\), followed by the diameter from \((-1,0)\) back to \((1,0)\). Compute \(\oint_C -y\,dx\).
The arc is traversed counterclockwise, so the enclosed upper half-disk stays on the left. Parametrize it by \(x=\cos t,\ y=\sin t\), \(0\le t\le\pi\). Since \(dx=-\sin t\,dt\), its contribution is
Along the diameter, \(y=0\), so the contribution is \(0\). Thus the boundary integral is \(\frac{\pi}{2}\). Here \(P=-y\) and \(Q=0\), so \(\frac{\partial Q}{\partial x}-\frac{\partial P}{\partial y}=1\). The enclosed half-disk has area \(\frac{\pi}{2}\), and Green’s theorem gives the same value. At each endpoint the arc and diameter meet at a corner; neither corner changes the calculation.
A Common Direction Error
A frequent mistake is to compute one piece in the reverse direction while leaving the other pieces unchanged. The result is not the integral around the stated boundary: the pieces no longer join in the order of a single traversal. In the triangle example, the slanted edge correctly runs from \((2,0)\) to \((0,1)\) and contributes \(-\frac{7}{3}\). Reversing only that edge would give \(+\frac{7}{3}\), so the sum would incorrectly become \(\frac{8}{3}+\frac{7}{3}=5\). That mixed list of directions is not a closed traversal of the triangle, and Green’s theorem cannot be used to justify it.
The correction is to keep each piece’s start and end points consistent with the same traversal. If the entire boundary is reversed, every piece must be reversed, and the complete integral changes sign. If the boundary is positively oriented, check that the region stays on the left as you proceed through every segment and arc.
Why Piecewise Boundaries Matter
Many useful planar regions have boundaries assembled from simple pieces. A polygon has corners; a sector may have circular arcs joined to radii; a region with a curved cap may have an arc meeting a straight segment. Piecewise-smooth boundaries let Green’s theorem handle these shapes without requiring a single smooth tangent direction all the way around.
The essential bookkeeping is local but the conclusion is global: choose one direction for the whole boundary, make consecutive pieces meet at their endpoints, compute every contribution in that direction, and add. Corners affect the parametrization and require care with orientation, but they contribute no extra term. A direct calculation can then be checked against the double integral of the curl density, provided the theorem’s hypotheses hold.
Check Your Understanding
Use the definitions and procedures in this tutorial to answer the following questions.
- What does piecewise-smooth mean, and where are the corner points in such a boundary?
- For positive orientation around a simple region, which side of the traversed boundary contains the region?
- Why does a corner contribute no separate term to the Green’s theorem line integral?
- What must be true of the endpoints and directions of consecutive parametrized pieces?
- If every piece of a positively oriented boundary is reversed, how does the line integral change?