math.wiki
Calculus › Multiple integrals › Cylindrical and spherical coordinates
Sign in

Cylindrical and spherical coordinates

A solid is shaped by spheres, cones and cylinders about an axis. Its description in x, y and z is a tangle of square roots, but in distances and angles it is a box. What does a triple integral become in those coordinates, and what factor replaces dV?

Where you are 6 in Multiple integrals

Why this exists

Layla is a glassblower. She spins a bowl on the end of her pipe while the glass is hot, and the effort it takes to start and stop the spin is set by the bowl’s moment of inertia about the pipe, an integral of x^2 + y^2 over a shell between two hemispheres. In x, y and z its upper surface needs two formulas, one over the hollow and one over the rim, and every limit is a square root. Described by the distance \rho from the center and two angles, it is a box: 10 \le \rho \le 11, all directions in the lower half.

Double integrals in polar coordinates did this in the plane, where dA = r\,dr\,d\theta. This lesson does it in space, twice: once for solids built around an axis, once for solids built around a point.

Definition

Cylindrical coordinates. A point (x, y, z) has cylindrical coordinates (r, \theta, z) when (r, \theta) are Polar coordinates of (x, y):

\begin{aligned} &x = r\cos\theta, \quad y = r\sin\theta, \\ &z = z, \qquad r \ge 0. \end{aligned}

So \displaystyle r = \sqrt{x^2 + y^2} is the distance from the z-axis.

Spherical coordinates. A point (x, y, z) has spherical coordinates (\rho, \theta, \varphi) when \rho \ge 0 is its distance from the origin, \theta is as in cylindrical coordinates, and \varphi, with 0 \le \varphi \le \pi, is the angle between the positive z-axis and \langle x, y, z \rangle (Figure 1). Then z = \rho\cos\varphi and r = \rho\sin\varphi, so

\begin{aligned} x &= \rho\sin\varphi\cos\theta, \\ y &= \rho\sin\varphi\sin\theta, \\ z &= \rho\cos\varphi, \end{aligned}

and \rho^2 = x^2 + y^2 + z^2. Since \sin\varphi \ge 0 on [0, \pi], r = \rho\sin\varphi is never negative. Physics books swap the names of the two angles, writing \theta for the angle from the z-axis and \varphi for the angle around it (and often r for \rho); this lesson keeps the order above throughout.

x y z
\theta
\varphi
\rho
r z
(x, y, z)
Figure 1One point, two systems. Cylindrical coordinates are (r, \theta, z) and spherical coordinates are (\rho, \theta, \varphi), with r = \rho\sin\varphi and z = \rho\cos\varphi.

With \theta in [0, 2\pi) and \rho > 0, a point off the z-axis has exactly one set of each coordinates. On the z-axis \theta is arbitrary, and at the origin \varphi is arbitrary too.

EquationSurface
r = c > 0the cylinder x^2 + y^2 = c^2
\theta = ca half-plane with edge the z-axis
\rho = c > 0the sphere x^2 + y^2 + z^2 = c^2
\varphi = c, \displaystyle 0 < c < \frac{\pi}{2}the half-cone z = r\cot c opening upward
\displaystyle \varphi = \frac{\pi}{2}the xy-plane

Solids. A solid is cylindrically simple if it is z-simple (Triple integrals), u_1(x, y) \le z \le u_2(x, y), over a radially simple region D: \alpha \le \theta \le \beta, h_1(\theta) \le r \le h_2(\theta) with 0 \le \beta - \alpha \le 2\pi and 0 \le h_1 \le h_2 continuous (Double integrals in polar coordinates). Write U_i(r, \theta) = u_i(r\cos\theta, r\sin\theta).

Let S(\rho, \varphi, \theta) = (\rho\sin\varphi\cos\theta, \rho\sin\varphi\sin\theta, \rho\cos\varphi). A solid E is spherically simple if E = S(P) for a parameter set

\begin{aligned} P = \{&(\rho, \varphi, \theta) : \alpha \le \theta \le \beta, \\ &\varphi_1 \le \varphi \le \varphi_2, \\ &g_1(\varphi, \theta) \le \rho \le g_2(\varphi, \theta)\}, \end{aligned}

with 0 \le \beta - \alpha \le 2\pi, 0 \le \varphi_1 \le \varphi_2 \le \pi, and 0 \le g_1 \le g_2 continuous. When g_1 = a and g_2 = b are constants, E is a spherical wedge.

Intuition

A cylindrical cell \Delta r by \Delta\theta by \Delta z is a polar cell of area about r\,\Delta r\,\Delta\theta stacked to height \Delta z, so its volume is about r\,\Delta z\,\Delta r\,\Delta\theta.

A spherical cell (Figure 2) has one straight edge, \Delta\rho along a ray. Its edge in the direction of \varphi is an arc of a circle of radius \rho, length about \rho\,\Delta\varphi. Its edge in the direction of \theta is an arc of a circle about the z-axis of radius r = \rho\sin\varphi, length about \rho\sin\varphi\,\Delta\theta. The edges meet at right angles, so the volume is about \rho^2\sin\varphi\,\Delta\rho\,\Delta\varphi\,\Delta\theta. Cells near the z-axis are thin wedges, which is why \sin\varphi appears; cells far out are large, which is why \rho^2 does.

z O
\Delta\rho
\rho\,\Delta\varphi
\rho\sin\varphi\,\Delta\theta
Figure 2A spherical cell is nearly a box with edges \Delta\rho, \rho\,\Delta\varphi and \rho\sin\varphi\,\Delta\theta. Lemma 2 makes the product exact at suitable \rho^* and \varphi^*.

Results

Theorem 1 (triple integrals in cylindrical coordinates). Let E be cylindrically simple and f continuous on E. Then

\begin{aligned} &\iiint_E f\,dV \\ &\quad = \int_{\alpha}^{\beta} \int_{h_1(\theta)}^{h_2(\theta)} \int_{U_1}^{U_2} f_C\,r\,dz\,dr\,d\theta, \end{aligned}

where f_C(r, \theta, z) = f(r\cos\theta, r\sin\theta, z).

Lemma 2 (volume of a spherical wedge). Let W = S(Q) with Q = [a, b] \times [\varphi_1, \varphi_2] \times [\theta_1, \theta_2], 0 \le a \le b, 0 \le \varphi_1 \le \varphi_2 \le \pi, 0 \le \theta_2 - \theta_1 \le 2\pi. Then

\begin{aligned} V(W) = \tfrac13&\left(b^3 - a^3\right)\left(\cos\varphi_1 - \cos\varphi_2\right) \\ &\cdot \left(\theta_2 - \theta_1\right), \end{aligned}

and there are \rho^* in [a, b] and \varphi^* in [\varphi_1, \varphi_2] with V(W) = \rho^{*2}\sin\varphi^*\,\Delta\rho\,\Delta\varphi\,\Delta\theta, where \Delta\rho = b - a, \Delta\varphi = \varphi_2 - \varphi_1, \Delta\theta = \theta_2 - \theta_1.

Theorem 3 (triple integrals in spherical coordinates). Let E = S(P) be spherically simple and f continuous on E. Then

\begin{aligned} &\iiint_E f\,dV \\ &\quad = \int_{\alpha}^{\beta} \int_{\varphi_1}^{\varphi_2} \int_{g_1}^{g_2} f_S\,\rho^2\sin\varphi \\ &\qquad\qquad d\rho\,d\varphi\,d\theta, \end{aligned}

where f_S(\rho, \varphi, \theta) = f(S(\rho, \varphi, \theta)).

On a spherical wedge the integrand is continuous on a box, so the six orders agree by Fubini’s theorem (Theorem 3 of Triple integrals).

Why it is true

Theorem 1. By Theorem 4 of Triple integrals, \iiint_E f\,dV = \iint_D F\,dA with F(x, y) = \int_{u_1}^{u_2} f(x, y, z)\,dz, and the proof there shows F is continuous on D. Theorem 2 of Double integrals in polar coordinates applied to F gives \int_{\alpha}^{\beta} \int_{h_1}^{h_2} F(r\cos\theta, r\sin\theta)\,r\,dr\,d\theta, and F(r\cos\theta, r\sin\theta) = \int_{U_1}^{U_2} f(r\cos\theta, r\sin\theta, z)\,dz. The factor r does not depend on z, so it moves inside. \blacksquare

Lemma 2. First let C_\gamma(b) = S([0, b] \times [0, \gamma] \times [\theta_1, \theta_2]), a solid cone capped by the sphere. For \displaystyle 0 < \gamma \le \frac{\pi}{2} it is z-simple over the sector 0 \le r \le b\sin\gamma, \theta_1 \le \theta \le \theta_2, between the cone z = r\cot\gamma and the sphere \displaystyle z = \sqrt{b^2 - r^2}, because \varphi \le \gamma means z \ge r\cot\gamma. By Theorem 1 with f = 1, using \cos\gamma \ge 0,

\begin{aligned} V &= \Delta\theta \int_0^{b\sin\gamma} \left(\sqrt{b^2 - r^2} - r\cot\gamma\right) r\,dr \\ &= \Delta\theta\,\tfrac{b^3}{3}\left(1 - \cos^3\gamma - \cos\gamma\sin^2\gamma\right) \\ &= \Delta\theta\,\tfrac{b^3}{3}\left(1 - \cos\gamma\right). \end{aligned}

At \gamma = 0 both sides are 0. For \displaystyle \frac{\pi}{2} < \gamma \le \pi, C_\gamma(b) is C_{\pi/2}(b) together with the reflection, by z \mapsto -z, of the part of C_{\pi/2}(b) outside C_{\pi - \gamma}(b). Reflection keeps the volume of a z-simple solid (substitute z = -s in the inner integral), so

V = \Delta\theta\,\tfrac{b^3}{3}\left(1 + \cos(\pi - \gamma)\right) = \Delta\theta\,\tfrac{b^3}{3}\left(1 - \cos\gamma\right).

The pieces in each step share only points of a cone or a sphere, sets of zero volume, so volumes add (Result 2 of Triple integrals). Removing C_{\varphi_1}(b) from C_{\varphi_2}(b) leaves the part of W with \rho \le b, of volume \displaystyle \Delta\theta\,\frac{b^3}{3}(\cos\varphi_1 - \cos\varphi_2); removing the part with \rho \le a leaves W, which gives the formula. By the mean value theorem, b^3 - a^3 = 3\rho^{*2}(b - a) and \cos\varphi_1 - \cos\varphi_2 = \sin\varphi^*\,(\varphi_2 - \varphi_1) for some \rho^* and \varphi^* between the endpoints (any point if an interval is a single point). \blacksquare

Theorem 3. Let b = \max g_2, M = \max_E \lvert f \rvert, Q = [0, b] \times [\varphi_1, \varphi_2] \times [\alpha, \beta] and T = S(Q) \supseteq E. Let G = f(S)\,\rho^2\sin\varphi on P and 0 on the rest of Q, and f^E = f on E and 0 on the rest of T. Both are bounded and continuous off a set of zero volume (graphs and their images under S), so both are integrable (the Real Analysis fact behind Result 1 of Triple integrals), and \iiint_Q G\,dV is the iterated integral of the theorem by Theorem 4 there, since P is \rho-simple over a rectangle. It remains to show \iiint_T f^E\,dV = \iiint_Q G\,dV.

For 0 \le \rho, \rho' \le b,

\begin{aligned} &\lVert S(\rho, \varphi, \theta) - S(\rho', \varphi', \theta') \rVert \\ &\quad \le \lvert \rho - \rho' \rvert + b\,\lvert \varphi - \varphi' \rvert + b\,\lvert \theta - \theta' \rvert: \end{aligned}

move along a ray, then along a circle of radius \rho' through the z-axis, then along a circle about the z-axis of radius \rho'\sin\varphi' \le b; each chord is no longer than its arc.

Cut the three intervals of Q into n equal pieces, giving cells Q_{ijk} and spherical wedges W_{ijk} = S(Q_{ijk}). Lemma 2 gives V(W_{ijk}) = \rho_i^{*2}\sin\varphi_j^*\,\Delta\rho\,\Delta\varphi\,\Delta\theta; take any \theta_k^* and let p_{ijk} = S(\rho_i^*, \varphi_j^*, \theta_k^*). The wedges cover T and overlap only on spheres, cones and half-planes, so \iiint_T f^E\,dV = \sum \iiint_{W_{ijk}} f^E\,dV. Compare term by term with \Sigma_n = \sum G(\rho_i^*, \varphi_j^*, \theta_k^*)\,\Delta\rho\,\Delta\varphi\,\Delta\theta, which tends to \iiint_Q G\,dV.

  • Outside (Q_{ijk} misses P). Both terms are 0, up to points with two preimages under S (the z-axis, and the half-plane \theta = \alpha when \beta - \alpha = 2\pi), a set of zero volume.
  • Inside (Q_{ijk} \subseteq P). Given \varepsilon > 0, uniform continuity of f on E gives \delta; once \Delta\rho + b\,\Delta\varphi + b\,\Delta\theta < \delta, \lvert f - f(p_{ijk}) \rvert < \varepsilon on W_{ijk}, so the term differs from f(p_{ijk})\,V(W_{ijk}) = G(\rho_i^*, \varphi_j^*, \theta_k^*)\,\Delta\rho\,\Delta\varphi\,\Delta\theta by at most \varepsilon\,V(W_{ijk}).
  • Boundary (the rest). Each term is at most Mb^2\,\Delta\rho\,\Delta\varphi\,\Delta\theta. The cell holds a point of P and a point not in P, so on the segment joining them \rho - g_1 or g_2 - \rho changes sign, and the cell meets a graph \rho = g_i. Let \omega_n be the largest change of g_1 or g_2 over one cell of the (\varphi, \theta) grid; \omega_n \to 0 by uniform continuity. Each graph meets at most \displaystyle \frac{\omega_n}{\Delta\rho} + 2 cells of each of the n^2 columns.

Adding up, with n\,\Delta\varphi = \varphi_2 - \varphi_1 and n\,\Delta\theta = \beta - \alpha,

\begin{aligned} \left\lvert \iiint_T f^E\,dV - \Sigma_n \right\rvert &\le \varepsilon\,V(T) \\ &\quad + 4Mb^2(\varphi_2 - \varphi_1) \\ &\quad \cdot (\beta - \alpha)(\omega_n + 2\Delta\rho). \end{aligned}

Let n \to \infty; then \varepsilon \to 0. \blacksquare

Worked examples

Example 1

Converting a point and a surface

Find cylindrical and spherical coordinates of \displaystyle (-1, 1, \sqrt{6}), and identify the surface \rho = 4\cos\varphi.

Solution

\displaystyle r = \sqrt{1 + 1} = \sqrt{2}. The angle has \displaystyle \cos\theta = -\frac{1}{\sqrt2} and \displaystyle \sin\theta = \frac{1}{\sqrt2}, so \displaystyle \theta = \frac{3\pi}{4}; the value \displaystyle \arctan\frac{y}{x} = -\frac{\pi}{4} is in the wrong quadrant. Cylindrical: \displaystyle \left(\sqrt{2}, \frac{3\pi}{4}, \sqrt{6}\right). Next \displaystyle \rho = \sqrt{2 + 6} = 2\sqrt{2} and \displaystyle \cos\varphi = \frac{\sqrt6}{2\sqrt2} = \frac{\sqrt3}{2}; since \varphi lies in [0, \pi], where cosine is one-to-one, \displaystyle \varphi = \frac{\pi}{6}. Spherical: \displaystyle \left(2\sqrt{2}, \frac{3\pi}{4}, \frac{\pi}{6}\right). Check: \displaystyle \rho\sin\varphi = 2\sqrt2 \cdot \frac12 = \sqrt2 = r.

Multiply \rho = 4\cos\varphi by \rho: \rho^2 = 4\rho\cos\varphi, that is x^2 + y^2 + z^2 = 4z, or x^2 + y^2 + (z - 2)^2 = 4. Multiplying by \rho adds only the origin, which is already on the surface (at \displaystyle \varphi = \frac{\pi}{2}). It is the sphere of radius 2 centered at (0, 0, 2), traced for \displaystyle 0 \le \varphi \le \frac{\pi}{2}, where \cos\varphi \ge 0.

Example 2

A solid off the axis

Find the volume of the solid between the paraboloid z = x^2 + y^2 and the plane z = 2y.

Solution

They meet where x^2 + y^2 = 2y, the circle x^2 + (y - 1)^2 = 1, which in polar form is r = 2\sin\theta, 0 \le \theta \le \pi. Inside it x^2 + y^2 \le 2y, so the plane is on top. The solid is cylindrically simple with h_1 = 0, h_2 = 2\sin\theta, U_1 = r^2 and U_2 = 2r\sin\theta, and by Theorem 1

\begin{aligned} V &= \int_0^{\pi} \int_0^{2\sin\theta} \int_{r^2}^{2r\sin\theta} r\,dz\,dr\,d\theta \\ &= \int_0^{\pi} \int_0^{2\sin\theta} \left(2r^2\sin\theta - r^3\right) dr\,d\theta \\ &= \int_0^{\pi} \left(\tfrac{16}{3} - 4\right)\sin^4\theta\,d\theta \\ &= \frac43 \cdot \frac{3\pi}{8} = \frac{\pi}{2}. \end{aligned}

Here \displaystyle \sin^4\theta = \frac38 - \frac12\cos 2\theta + \frac18\cos 4\theta. The axis of the coordinates need not be an axis of symmetry: the shadow is a disk through the origin, and h_2 carries its shape.

Example 3

Layla’s bowl

Layla’s bowl is the shell 10 \le \rho \le 11 (centimeters), \displaystyle \frac{\pi}{2} \le \varphi \le \pi, of glass with density 2.5 grams per cubic centimeter, spun about the z-axis. Find its mass M and moment of inertia I_z = \iiint_E (x^2 + y^2)\,\delta\,dV, and compare I_z with the thin-shell value \displaystyle \frac23 MR^2.

Solution

The bowl is a spherical wedge with \theta in [0, 2\pi]. By Lemma 2,

M = 2.5 \cdot \tfrac13(1331 - 1000)(0 + 1)(2\pi) = \frac{1655\pi}{3},

about 1733 grams. In spherical coordinates x^2 + y^2 = r^2 = \rho^2\sin^2\varphi, and the limits are constants, so by Theorem 3 the integral splits into three:

\begin{aligned} I_z &= 2.5\int_0^{2\pi} d\theta \int_{\pi/2}^{\pi} \sin^3\varphi\,d\varphi \\ &\quad \cdot \int_{10}^{11} \rho^4\,d\rho \\ &= 2.5 \cdot 2\pi \cdot \frac23 \cdot \frac{61{,}051}{5} \\ &= \frac{122{,}102\pi}{3}, \end{aligned}

about 127{,}865 gram square centimeters. The middle factor is \displaystyle \int_{-1}^{0} (1 - u^2)\,du = \frac23 with u = \cos\varphi. So \displaystyle \frac{I_z}{M} = \frac{122{,}102}{1655} \approx 73.78 square centimeters, against \displaystyle \frac23(10.5)^2 = 73.5 for a thin shell at the middle radius. In general \displaystyle \frac{I_z}{M} = \frac25 \cdot \frac{b^5 - a^5}{b^3 - a^3}, which tends to \displaystyle \frac25 \cdot \frac{5b^4}{3b^2} = \frac23 b^2 as a \to b, the thin-shell value.

Example 4

A ball through the origin

Find the volume of the part of the ball x^2 + y^2 + (z - 2)^2 \le 4 inside the cone \displaystyle \varphi \le \frac{\pi}{3}, and of the rest of the ball (Figure 3).

Solution

By Example 1 the ball is 0 \le \rho \le 4\cos\varphi, \displaystyle 0 \le \varphi \le \frac{\pi}{2}. The part in the cone is spherically simple with g_1 = 0, g_2 = 4\cos\varphi, \displaystyle 0 \le \varphi \le \frac{\pi}{3}, so by Theorem 3

\begin{aligned} V &= \int_0^{2\pi} \int_0^{\pi/3} \int_0^{4\cos\varphi} \rho^2\sin\varphi\,d\rho\,d\varphi\,d\theta \\ &= 2\pi \int_0^{\pi/3} \tfrac{64}{3}\cos^3\varphi\sin\varphi\,d\varphi \\ &= \frac{128\pi}{3}\Big[-\tfrac14\cos^4\varphi\Big]_0^{\pi/3} \\ &= \frac{128\pi}{3} \cdot \frac{15}{64} = 10\pi. \end{aligned}
x z
\varphi
\displaystyle (\sqrt{3}, 1)
\varphi = \tfrac{\pi}{3}
\rho = 4\cos\varphi
Figure 3Example 4. The ball \rho \le 4\cos\varphi and the cone \displaystyle \varphi = \frac{\pi}{3}. Along each ray inside the cone, \rho runs from 0 to 4\cos\varphi.

With \displaystyle \frac{\pi}{2} as the upper limit the bracket is \displaystyle \frac14 and the volume \displaystyle \frac{32\pi}{3} = \frac43\pi \cdot 2^3, the whole ball, so the rest has volume \displaystyle \frac{32\pi}{3} - 10\pi = \frac{2\pi}{3}. In cylindrical coordinates the part in the cone takes two pieces: for \displaystyle r \le \sqrt{3}, z runs from the cone to the upper hemisphere; for \displaystyle \sqrt3 \le r \le 2, from the lower hemisphere to the upper. Spherical coordinates centered at a point of the sphere need one.

Example 5

The volume of a spherical cap

Let 0 < h \le a. Prove that the cap x^2 + y^2 + z^2 \le a^2, z \ge a - h, has volume \displaystyle \frac{\pi h^2(3a - h)}{3}.

Solution

Let d = a - h, so 0 \le d < a. If d = 0 the cap is the wedge 0 \le \rho \le a, \displaystyle 0 \le \varphi \le \frac{\pi}{2}, of volume \displaystyle \frac{2\pi a^3}{3} by Lemma 2, which is the formula at h = a. Let d > 0 and \displaystyle \cos\gamma = \frac{d}{a} with \displaystyle 0 < \gamma < \frac{\pi}{2}. A point of the cap has z \ge d > 0, so \displaystyle \varphi < \frac{\pi}{2}, and d \le \rho\cos\varphi \le a\cos\varphi forces \varphi \le \gamma. Conversely \varphi \le \gamma and \displaystyle \frac{d}{\cos\varphi} \le \rho \le a give z = \rho\cos\varphi \ge d and \rho \le a. So the cap is spherically simple with g_1 = d\sec\varphi, g_2 = a over 0 \le \varphi \le \gamma, where g_1 \le a. By Theorem 3,

\begin{aligned} V &= 2\pi\int_0^{\gamma} \tfrac13\left(a^3 - d^3\sec^3\varphi\right)\sin\varphi\,d\varphi \\ &= \tfrac{2\pi}{3}\Big(a^3(1 - \cos\gamma) \\ &\qquad - \tfrac{d^3}{2}\left(\sec^2\gamma - 1\right)\Big), \end{aligned}

since \displaystyle \frac{1}{2\cos^2\varphi} has derivative \sec^3\varphi\sin\varphi. Now a^3(1 - \cos\gamma) = a^2(a - d) = a^2 h and \displaystyle \frac{d^3}{2}\left(\frac{a^2}{d^2} - 1\right) equals \displaystyle \frac{d}{2}(a - d)(a + d) = \frac12(a - h)h(2a - h). So

\begin{aligned} V &= \tfrac{\pi h}{3}\left(2a^2 - (a - h)(2a - h)\right) \\ &= \tfrac{\pi h}{3}\left(3ah - h^2\right) = \frac{\pi h^2(3a - h)}{3}. \quad \blacksquare \end{aligned}
Example 6

A ball acts like a point

Let B be the ball \rho \le a and \mathbf{q} a point at distance d > a from the origin. Prove that

\iiint_B \frac{dV}{\lVert \mathbf{p} - \mathbf{q} \rVert} = \frac{4\pi a^3}{3d} = \frac{V(B)}{d}.
Solution

A rotation about the origin keeps distances, the ball and volume (a case of Change of variables and the Jacobian), so take \mathbf{q} = (0, 0, d). For \mathbf{p} = S(\rho, \varphi, \theta),

\begin{aligned} \lVert \mathbf{p} - \mathbf{q} \rVert^2 &= x^2 + y^2 + (z - d)^2 \\ &= \rho^2 + d^2 - 2\rho d\cos\varphi, \end{aligned}

which is at least (d - \rho)^2 > 0, so the integrand is continuous on B. The ball is a spherical wedge; integrate in \varphi first. Write \displaystyle w(\varphi) = \sqrt{\rho^2 + d^2 - 2\rho d\cos\varphi}, so that \displaystyle w' = \frac{\rho d\sin\varphi}{w}. For 0 < \rho \le a,

\begin{aligned} \int_0^{\pi} \frac{\rho^2\sin\varphi}{w}\,d\varphi &= \frac{\rho}{d}\Big[w\Big]_0^{\pi} \\ &= \frac{\rho}{d}\left((d + \rho) - (d - \rho)\right) = \frac{2\rho^2}{d}, \end{aligned}

using w(0) = \lvert d - \rho \rvert = d - \rho. By Theorem 3 the integral is \displaystyle 2\pi\int_0^a \frac{2\rho^2}{d}\,d\rho = \frac{4\pi a^3}{3d}. \blacksquare With a uniform density this is Newton’s result that a ball’s gravitational potential outside it is that of its mass placed at the center.

Boundaries

The factors are not optional. \iiint d\rho\,d\varphi\,d\theta over [0, 1] \times [0, \pi] \times [0, 2\pi] is 2\pi^2, the volume of the parameter box; the unit ball has volume \displaystyle \frac{4\pi}{3}. The map S stretches volume by \rho^2\sin\varphi, which Change of variables and the Jacobian identifies as a Jacobian determinant (Exercise 17).

The map is not one-to-one on the z-axis and at the origin, and \theta wraps around. Those sets have zero volume, so integrals do not see them, but \theta of a point must be read from the signs of x and y (Example 1).

Coordinates should fit the solid. The cube [0, 1]^3 in spherical coordinates needs three pieces, since \rho stops on x = 1, y = 1 or z = 1 depending on the direction. A solid that is a box in x, y, z is best left there.

Singular integrands. Near the origin \rho^{-p} is integrable over the ball \rho \le 1 exactly when p < 3: over \varepsilon \le \rho \le 1 the integral is 4\pi\int_{\varepsilon}^1 \rho^{2 - p}\,d\rho, which has a finite limit as \varepsilon \to 0^+ only then. In the plane the threshold for r^{-p} is p < 2; the factor \rho^2 buys one more power.

Where this goes next

Change of variables and the Jacobian puts r and \rho^2\sin\varphi in one theorem: for a one-to-one smooth map, the volume element is the absolute value of the Jacobian determinant (What the determinant means). Parametric surfaces and surface integrals parametrizes the sphere \rho = a by \varphi and \theta, with area element a^2\sin\varphi\,d\varphi\,d\theta, the face of the cell in Figure 2. The divergence theorem turns flux through spheres and cylinders into integrals of exactly the kind computed here.

Exercises

Coordinates and surfaces

  1. Convert the cylindrical point \displaystyle \left(4, \frac{5\pi}{6}, -3\right) to rectangular coordinates.
  2. Find cylindrical and spherical coordinates of \displaystyle (1, -\sqrt{3}, 2\sqrt{3}), with \theta in [0, 2\pi).
  3. Convert the spherical point \displaystyle (\rho, \theta, \varphi) = \left(6, \frac{\pi}{4}, \frac{2\pi}{3}\right) to rectangular coordinates.
  4. Identify each surface by its rectangular equation: (a) \rho = 6\sin\varphi\sin\theta; (b) \rho\sin\varphi = 3; (c) \displaystyle \varphi = \frac{3\pi}{4}; (d) \rho^2\cos 2\varphi = 1.

Cylindrical coordinates

  1. Evaluate \iiint_E (x^2 + y^2)\,dV over the cylinder x^2 + y^2 \le 9, 0 \le z \le 2.
  2. Find the volume of the solid inside both the cylinder x^2 + y^2 = 1 and the sphere x^2 + y^2 + z^2 = 4.
  3. Evaluate \iiint_E z\,dV, where E lies above the cone \displaystyle z = \sqrt{x^2 + y^2} and below the paraboloid z = 6 - x^2 - y^2.
  4. Evaluate \displaystyle \int_0^2 \int_0^{\sqrt{4 - x^2}} \int_0^{4 - x^2 - y^2} x\,dz\,dy\,dx.
  5. Omar makes the wall of a glass vase, \displaystyle 4 \le r \le 5 - \frac{z}{40}, 0 \le z \le 20 (centimeters), 1 centimeter thick at the base and half that at the rim, with density 2.5 grams per cubic centimeter. Find its mass (ignore the base).

Spherical coordinates

  1. Evaluate \iiint_B (x^2 + y^2 + z^2)\,dV over the ball of radius 2 centered at the origin.
  2. Find the volume of the solid \rho \le 3, \displaystyle 0 \le \varphi \le \frac{\pi}{6}.
  3. Evaluate \displaystyle \int_{-1}^{1} \int_{-\sqrt{1 - x^2}}^{\sqrt{1 - x^2}} \int_{\sqrt{x^2 + y^2}}^{\sqrt{2 - x^2 - y^2}} w\,dz\,dy\,dx, where \displaystyle w = \sqrt{x^2 + y^2 + z^2}.
  4. Find \bar{z}, the height of the centroid of the solid in Exercise 11.
  5. Find the average distance from the center of a ball of radius a to its points.
  6. Find the average distance from a point on the surface of a ball of radius a to the points of the ball. (Put the point at the origin and the center at (0, 0, a).)
  7. Find \iiint_{\mathbb{R}^3} e^{-(x^2 + y^2 + z^2)^{3/2}}\,dV, as the limit of the integrals over the balls \rho \le R.
  8. Compute the determinant of the matrix of partial derivatives of S with columns in the order \rho, \varphi, \theta.

Proofs

  1. Let g be continuous on [0, a]. Prove that \displaystyle \iiint_B g\left(\sqrt{x^2 + y^2 + z^2}\right) dV = 4\pi\int_0^a g(\rho)\,\rho^2\,d\rho, where B is the ball \rho \le a.
  2. Prove that a solid ball of radius a and constant density, with mass M, has \displaystyle I_z = \frac25 Ma^2.
  3. Let \Omega be the region c \le r \le d, k_1(r) \le z \le k_2(r) of the half-plane r \ge 0, with k_1 \le k_2 continuous, and let E be the solid swept out by turning \Omega about the z-axis. Prove V(E) = 2\pi\iint_\Omega r\,dA = 2\pi\bar{r}\,A(\Omega), where \bar{r} is the r-coordinate of the centroid of \Omega (Pappus’s theorem).
  4. Prove that the cap formula of Example 5 also holds for a < h \le 2a.
Answers
  1. \displaystyle (-2\sqrt{3}, 2, -3). \displaystyle x = 4\cos\frac{5\pi}{6} and \displaystyle y = 4\sin\frac{5\pi}{6}.
  2. Cylindrical \displaystyle \left(2, \frac{5\pi}{3}, 2\sqrt{3}\right), spherical \displaystyle \left(4, \frac{5\pi}{3}, \frac{\pi}{6}\right). r = 2, the point \displaystyle (1, -\sqrt3) is in the fourth quadrant, \displaystyle \rho = \sqrt{16} = 4, \displaystyle \cos\varphi = \frac{2\sqrt3}{4}.
  3. \displaystyle \left(\frac{3\sqrt{6}}{2}, \frac{3\sqrt{6}}{2}, -3\right). \displaystyle z = 6\cos\frac{2\pi}{3} = -3 and \displaystyle r = 6\sin\frac{2\pi}{3} = 3\sqrt3, then \displaystyle x = y = 3\sqrt3 \cdot \frac{\sqrt2}{2}.
  4. (a) The sphere x^2 + (y - 3)^2 + z^2 = 9: multiply by \rho to get x^2 + y^2 + z^2 = 6y. (b) The cylinder x^2 + y^2 = 9, since \rho\sin\varphi = r. (c) The lower half-cone \displaystyle z = -\sqrt{x^2 + y^2}: \displaystyle z = \rho\cos\frac{3\pi}{4} = -r. (d) The hyperboloid of two sheets z^2 - x^2 - y^2 = 1, since \rho^2\cos 2\varphi = \rho^2\cos^2\varphi - \rho^2\sin^2\varphi = z^2 - r^2.
  5. 81\pi. \displaystyle \int_0^{2\pi} \int_0^3 \int_0^2 r^3\,dz\,dr\,d\theta = 2\pi \cdot \frac{81}{4} \cdot 2.
  6. \displaystyle \frac{4\pi}{3}\left(8 - 3\sqrt{3}\right) \approx 11.74. Over r \le 1, z runs between \displaystyle \mp\sqrt{4 - r^2}, so \displaystyle V = 2\pi\int_0^1 2\sqrt{4 - r^2}\,r\,dr = 2\pi \cdot \frac23\left(8 - 3^{3/2}\right).
  7. \displaystyle \frac{92\pi}{3}. The surfaces meet where r = 6 - r^2, at r = 2. The z-integral is \displaystyle \frac12\left((6 - r^2)^2 - r^2\right), and \displaystyle \pi\int_0^2 \left(36r - 13r^3 + r^5\right) dr = \pi\left(72 - 52 + \frac{32}{3}\right).
  8. \displaystyle \frac{64}{15}. The shadow is the quarter disk r \le 2, \displaystyle 0 \le \theta \le \frac{\pi}{2}, so the integral is \int_0^{\pi/2} \cos\theta\,d\theta = 1 times \displaystyle \int_0^2 (4 - r^2)\,r^2\,dr = \frac{32}{3} - \frac{32}{5}.
  9. \displaystyle \frac{1975\pi}{6} \approx 1034 grams. Over 4 \le r \le 4.5, z runs from 0 to 20; over 4.5 \le r \le 5, from 0 to 40(5 - r). So V = 2\pi\int_4^{4.5} 20r\,dr + 2\pi\int_{4.5}^{5} 40(5 - r)\,r\,dr, which is \displaystyle 2\pi\left(\frac{85}{2} + \frac{70}{3}\right) = \frac{395\pi}{3}, and M = 2.5\,V.
  10. \displaystyle \frac{128\pi}{5}. By Exercise 18, 4\pi\int_0^2 \rho^4\,d\rho.
  11. \displaystyle 9\pi\left(2 - \sqrt{3}\right) \approx 7.58. By Lemma 2, \displaystyle \frac13 \cdot 27 \cdot \left(1 - \frac{\sqrt3}{2}\right) \cdot 2\pi.
  12. \displaystyle \pi\left(2 - \sqrt{2}\right). The solid is above the cone \displaystyle \varphi = \frac{\pi}{4} and inside \displaystyle \rho = \sqrt2, which meet over r = 1. The integral is \displaystyle 2\pi\int_0^{\pi/4} \sin\varphi\,d\varphi \int_0^{\sqrt2} \rho^3\,d\rho = 2\pi\left(1 - \frac{\sqrt2}{2}\right) \cdot 1.
  13. \displaystyle \bar{z} = \frac{9(2 + \sqrt{3})}{16} \approx 2.10. M_{xy} = 2\pi\int_0^{\pi/6} \cos\varphi\sin\varphi\,d\varphi \int_0^3 \rho^3\,d\rho, which is \displaystyle 2\pi \cdot \frac18 \cdot \frac{81}{4} = \frac{81\pi}{16}; divide by \displaystyle 9\pi(2 - \sqrt3) and rationalize.
  14. \displaystyle \frac{3a}{4}. By Exercise 18, \displaystyle \iiint_B \rho\,dV = 4\pi \cdot \frac{a^4}{4}; divide by \displaystyle \frac43\pi a^3.
  15. \displaystyle \frac{6a}{5}. The ball is \rho \le 2a\cos\varphi, \displaystyle 0 \le \varphi \le \frac{\pi}{2} (as in Example 1). \displaystyle \iiint \rho\,dV = 2\pi\int_0^{\pi/2} 4a^4\cos^4\varphi\sin\varphi\,d\varphi = \frac{8\pi a^4}{5}; divide by \displaystyle \frac43\pi a^3.
  16. \displaystyle \frac{4\pi}{3}. Over \rho \le R the integral is \displaystyle 4\pi\int_0^R \rho^2 e^{-\rho^3}\,d\rho = \frac{4\pi}{3}\left(1 - e^{-R^3}\right), which tends to \displaystyle \frac{4\pi}{3}.
  17. \rho^2\sin\varphi. The bottom row is \cos\varphi, -\rho\sin\varphi, 0. Expanding along it, the two minors are \rho^2\sin\varphi\cos\varphi and \rho\sin^2\varphi, so the determinant is \cos\varphi \cdot \rho^2\sin\varphi\cos\varphi plus \rho\sin\varphi \cdot \rho\sin^2\varphi, which is \rho^2\sin\varphi\left(\cos^2\varphi + \sin^2\varphi\right) = \rho^2\sin\varphi. For cylindrical coordinates the same computation gives r.
  18. \displaystyle f(x, y, z) = g\left(\sqrt{x^2 + y^2 + z^2}\right) is a composition of continuous functions, so continuous on B, and f(S(\rho, \varphi, \theta)) = g(\rho). B is the spherical wedge [0, a] \times [0, \pi] \times [0, 2\pi], so by Theorem 3 with \rho outermost (any order, by Fubini), the integrand g(\rho)\rho^2\sin\varphi is integrated first in \theta and \varphi. Since \int_0^{2\pi} d\theta = 2\pi and \int_0^{\pi} \sin\varphi\,d\varphi = 2, \iiint_B f\,dV = 4\pi\int_0^a g(\rho)\rho^2\,d\rho. \blacksquare
  19. Let \delta be the density, so \displaystyle M = \frac43\pi a^3\delta. By Theorem 3, with x^2 + y^2 = \rho^2\sin^2\varphi and constant limits, I_z is \delta times the product of \int_0^{2\pi} d\theta = 2\pi, \displaystyle \int_0^{\pi} \sin^3\varphi\,d\varphi = \frac43 (with u = \cos\varphi, it is \int_{-1}^1 (1 - u^2)\,du) and \displaystyle \int_0^a \rho^4\,d\rho = \frac{a^5}{5}. So \displaystyle I_z = \frac{8\pi\delta a^5}{15}. Dividing, \displaystyle \frac{I_z}{M} = \frac{8\pi a^5/15}{4\pi a^3/3} = \frac25 a^2. \blacksquare
  20. E is z-simple over the ring D: c \le r \le d, 0 \le \theta \le 2\pi, with \displaystyle u_i(x, y) = k_i\left(\sqrt{x^2 + y^2}\right), continuous, so E is cylindrically simple with U_i = k_i(r). By Theorem 1 with f = 1, V(E) = \int_0^{2\pi} \int_c^d \left(k_2(r) - k_1(r)\right) r\,dr\,d\theta, which is 2\pi\int_c^d \left(k_2 - k_1\right) r\,dr. In the rz-plane \Omega is type I, so by Theorem 4 of Iterated integrals and Fubini's theorem the last integral is \iint_\Omega r\,dA, which by the definition of the centroid is \bar{r}\,A(\Omega). \blacksquare
  21. Let a < h \le 2a. The ball is the cap z \ge a - h together with the slab z \le a - h, sharing a disk of zero volume. The reflection z \mapsto -z, which keeps volume (Lemma 2’s proof), carries the slab to the cap z \ge h - a of height k = 2a - h, with 0 \le k < a. By Example 5 (and volume 0 when k = 0), \displaystyle V = \frac43\pi a^3 - \frac{\pi k^2(3a - k)}{3}. Now k^2(3a - k) = (2a - h)^2(a + h), which expands to 4a^3 - 3ah^2 + h^3, so \displaystyle V = \frac{\pi}{3}\left(3ah^2 - h^3\right) = \frac{\pi h^2(3a - h)}{3}. \blacksquare
math.wiki · updated 2026-11-17 · CC-BY-SA-4.0