Symmetry Reduction in Polar Integrals

Why Symmetry Helps

A polar double integral can become much simpler when the region and the integrand share a symmetry. If a region consists of several congruent sectors and the integrand has the same value in each, it may be enough to integrate over one sector and multiply. If corresponding points have opposite integrand values, their contributions can cancel.

The key is to check both parts of the integral. A symmetric region by itself does not guarantee that the integral is zero, and a symmetric-looking integrand does not guarantee that pieces contribute equally. The transformation must preserve the region, and we must know how it changes the integrand.

Definition: A region $D$ is invariant under a transformation $T$ if $T(D)=D$. An integrand $f$ is invariant under $T$ if $f(T(x,y))=f(x,y)$ at every point in its domain. It changes sign under $T$ if $f(T(x,y))=-f(x,y)$.

Symmetry in Polar Coordinates

Write a point in the plane as $(x,y)=(r\cos\theta,r\sin\theta)$, with $r\geq 0$. The polar area element is $dA=r\,dr\,d\theta$, so a double integral over a polar region has integrand $f(r\cos\theta,r\sin\theta)r$. A reflection or rotation about the origin leaves $r$ unchanged and transforms only the angle.

Transformation	Polar angle mapping	Effect to check
Reflection across the $x$-axis	$\theta\mapsto-\theta$	Compare $f(r,\theta)$ with $f(r,-\theta)$.
Reflection across the $y$-axis	$\theta\mapsto\pi-\theta$	Compare $f(r,\theta)$ with $f(r,\pi-\theta)$.
Rotation by angle $\alpha$	$\theta\mapsto\theta+\alpha$	Compare $f(r,\theta)$ with $f(r,\theta+\alpha)$.

Here $f(r,\theta)$ abbreviates the Cartesian function evaluated at $(r\cos\theta,r\sin\theta)$. The polar area factor $r$ is unchanged by these angle transformations. Thus the symmetry test for a polar integral is the same as the test for the function values, with the area factor retained in the integral.

When Symmetric Contributions Cancel

Suppose $D$ is unchanged by a reflection or rotation $T$, and $f(T(x,y))=-f(x,y)$ throughout $D$. The transformation preserves area. Substituting $T(x,y)$ for $(x,y)$ in the integral therefore gives the same integral over $D$, but with the integrand negated. The integral must be its own negative.

Theorem: If $D$ is an integrable region invariant under an area-preserving transformation $T$, and $f(T(x,y))=-f(x,y)$ on $D$, then $$ \iint_D f(x,y)\,dA=0. $$ The integral must exist for this conclusion to apply.

For the reflections and rotations considered here, area is preserved because they are rigid motions. In polar coordinates, the same argument is a change of angular variable; the factor $r$ is unchanged. The proof can be written in full by setting $I=\iint_D f\,dA$. The substitution by $T$ gives

I=\iint_D f(T(x,y))\,dA =-\iint_D f(x,y)\,dA =-I.

Hence $2I=0$, so $I=0$. This argument pairs contributions without needing to calculate either side separately.

Worked Example: An Odd Integrand on a Disk

Evaluate $\iint_D x\,dA$, where $D$ is the disk $x^2+y^2\leq 4$. The disk is invariant under reflection across the $y$-axis, which sends $(x,y)$ to $(-x,y)$. The integrand changes sign: $f(-x,y)=-x=-f(x,y)$. By the theorem, the integral is zero.

The cancellation can also be checked by computing the two halves. In polar coordinates, the right half has $-\pi/2\leq\theta\leq\pi/2$, and the integrand including the area factor is $r^2\cos\theta$. Thus

\iint_{D_{\mathrm{right}}}x\,dA =\int_{-\pi/2}^{\pi/2}\int_0^2 r^2\cos\theta\,dr\,d\theta =\left(\frac{8}{3}\right)(2) =\frac{16}{3}.

Reflection across the $y$-axis maps this half to the left half and changes $x$ to $-x$, so the left-half integral is $-16/3$. The total is $16/3-16/3=0$, as predicted. Substituting the reflected point into the integrand confirms the sign change directly.

When Symmetric Contributions Repeat

If a region is made up of congruent sectors related by rotations, an integrand invariant under those rotations contributes equally on each sector. Integrating over one sector and multiplying by the number of sectors is then valid, provided the sectors cover the region without overlap except along boundaries. Such boundaries have area zero and do not affect an ordinary double integral.

Theorem: Suppose $D$ is partitioned, up to boundaries of area zero, into $N$ sectors $S,T(S),\ldots,T^{N-1}(S)$, where $T$ is a rotation. If $f$ is integrable and satisfies $f(T(x,y))=f(x,y)$, then $$ \iint_D f(x,y)\,dA=N\iint_S f(x,y)\,dA. $$

To justify the formula, apply the rotation substitution to each sector integral. Rotation preserves area, and invariance gives the same integrand values at corresponding points. Therefore each sector integral equals $\iint_S f\,dA$. Adding the $N$ equal contributions gives the stated result. If the integrand changes sign under a rotation instead, repeated pieces need not all be equal; use the sign pattern rather than multiplying by $N$ without checking.

Worked Example: A Radial Integrand on a Disk

Evaluate $\iint_D (x^2+y^2)\,dA$ over the disk $x^2+y^2\leq4$, using rotational symmetry. The disk splits into four quarter-disks related by rotations through $\pi/2$. The integrand $x^2+y^2$ is unchanged by every rotation, so the full integral is four times the integral over the first quadrant.

In polar coordinates, the first-quadrant bounds are $0\leq r\leq2$ and $0\leq\theta\leq\pi/2$. Since $x^2+y^2=r^2$, including the area factor gives $r^3$. Hence

\begin{aligned} \iint_{D_{\mathrm{first\ quadrant}}}(x^2+y^2)\,dA &=\int_0^{\pi/2}\int_0^2 r^3\,dr\,d\theta\\ &=\int_0^{\pi/2}\left[\frac{r^4}{4}\right]_0^2\,d\theta\\ &=\int_0^{\pi/2}4\,d\theta =2\pi. \end{aligned}

Multiplying by the four equal sectors gives $4(2\pi)=8\pi$. As a check, integrating over the whole disk directly gives $\int_0^{2\pi}\int_0^2r^3\,dr\,d\theta=4(2\pi)=8\pi$, the same value.

Reducing an Even Integral by Reflection

When the integrand is unchanged under a reflection, the two reflected pieces add rather than cancel. For example, if a region is symmetric about the $x$-axis and $f(x,-y)=f(x,y)$, then the upper and lower halves make equal contributions. The full integral is twice the integral over either half. In polar form, reflection across the $x$-axis pairs angles $\theta$ and $-\theta$.

Worked Example: Squared Height on a Disk

Evaluate $\iint_D y^2\,dA$ over the unit disk $x^2+y^2\leq1$. Since $y^2$ is unchanged when $y$ is replaced by $-y$, the lower and upper semicircles contribute equally. Calculate on the upper semicircle and double the result.

For $0\leq r\leq1$ and $0\leq\theta\leq\pi$, $y^2=r^2\sin^2\theta$, so the polar integrand including area is $r^3\sin^2\theta$. Therefore

\begin{aligned} \iint_{D_{\mathrm{upper}}}y^2\,dA &=\int_0^\pi\int_0^1r^3\sin^2\theta\,dr\,d\theta\\ &=\left[\frac{r^4}{4}\right]_0^1 \int_0^\pi\sin^2\theta\,d\theta\\ &=\frac14\cdot\frac{\pi}{2} =\frac{\pi}{8}. \end{aligned}

Doubling gives $\iint_D y^2\,dA=\pi/4$. As a check, the full-disk polar bounds give $\frac14\int_0^{2\pi}\sin^2\theta\,d\theta=\frac14\cdot\pi=\pi/4$, matching the reduced calculation.

A Common Symmetry Error

A common mistake is to infer cancellation from a symmetric region without checking the function. On the disk in the first example, reflection across the $y$-axis sends $x$ to $-x$, so the contributions cancel. But for the integrand $x^2$, the same reflection gives $(-x)^2=x^2$: the contributions are equal, not opposite.

For instance, on the unit disk the integral of $x^2$ is positive. In polar coordinates its integrand is $r^3\cos^2\theta$, which is nonnegative, and direct integration gives

\int_0^{2\pi}\int_0^1 r^3\cos^2\theta\,dr\,d\theta =\frac14\int_0^{2\pi}\cos^2\theta\,d\theta =\frac{\pi}{4},

not zero. The correction is to test the function under the actual transformation before deciding whether paired pieces cancel or repeat. Also keep the polar factor $r$: the function $f(r,\theta)$ alone is not the full integrand with respect to $dr\,d\theta$.

A Reliable Symmetry Procedure

Use the following sequence before simplifying a polar integral:

Identify a candidate symmetry.
Check whether a reflection or rotation maps the whole region onto itself.

Transform the integrand.
Substitute the reflected or rotated coordinates into the function, and retain the polar area factor $r$.

Decide whether contributions cancel or repeat.
An integrand that changes sign can yield cancellation; an invariant integrand gives equal contributions on corresponding pieces.

Choose bounds for the reduced piece.
Write the radial and angular bounds for one half or sector, integrate there, and apply the justified factor or cancellation.

Symmetry matters because it can shorten both the bounds and the calculation, but it is also a way to reason about an integral before evaluating it. A zero result can follow from cancellation alone, while a repeated-sector calculation can reduce a large angular interval to a smaller one. In either case, the transformation supplies the justification: it preserves the region and area, and its effect on the integrand determines the relationship between contributions.

Check Your Understanding

For each question, state what transformation you would use and how it acts on the region and integrand.

A disk centered at the origin is integrated against $f(x,y)=y$. Which reflection proves the integral is zero?
A region consists of four rotated copies of one sector, and $f$ is unchanged by rotations through $\pi/2$. How is the full integral related to the integral over one sector?
Under reflection across the $x$-axis, what happens to $f(x,y)=x^2y$? What does that imply for its integral over a disk centered at the origin?
Why does an integrand that is unchanged under a reflection lead to equal contributions rather than cancellation?
In a polar double integral, why must the area factor $r$ be retained when comparing reflected or rotated integrands?

Transformation	Polar angle mapping	Effect to check
Reflection across the \(x\)-axis	\(\theta\mapsto-\theta\)	Compare \(f(r,\theta)\) with \(f(r,-\theta)\).
Reflection across the \(y\)-axis	\(\theta\mapsto\pi-\theta\)	Compare \(f(r,\theta)\) with \(f(r,\pi-\theta)\).
Rotation by angle \(\alpha\)	\(\theta\mapsto\theta+\alpha\)	Compare \(f(r,\theta)\) with \(f(r,\theta+\alpha)\).