Chapter 5

Multiple Integrals; Applications of Integration

5.1Single integral

Video Lecture 15, Part 1: From Area to Volume (0:00)

Illustration of the definite integral as the signed area under a curve.
Figure 5.1. Illustration of the definite integral as the signed area under a curve.
DefinitionDefinite Integral as Signed Area

If f ( x ) is continuous on the closed interval [ a , b ] , the definite integral

∫ a b f ( x ) d x (5.1)

represents the accumulated signed area between the curve y = f ( x ) and the x -axis, from x = a to x = b .

  • Area above the x -axis contributes positive value.

  • Area below the x -axis contributes negative value.

Example

Over one full period of the sine,

∫ 0 2 π sin ⁡ ( x ) d x = [ − cos ⁡ ( x ) ] 0 2 π = − 1 − ( − 1 ) = 0 . (5.2)

The shaded area is not zero: the arch above the axis, from 0 to π , has area ∫ 0 π sin ⁡ ( x ) d x = 2 , and the arch below, from π to 2 π , counts as − 2 . The two cancel (Fig. 5.2).

The integral keeps the sign: over one period of (x) the arch above the axis counts +2 and the arch below counts -2, so 0^2(x)\,dx = 0.
Figure 5.2. The integral keeps the sign: over one period of sin ⁡ ( x ) the arch above the axis counts + 2 and the arch below counts − 2 , so ∫ 0 2 π sin ⁡ ( x ) d x = 0 .

5.2Double and Triple Integrals

In many physical problems, quantities depend on two or three spatial variables, e.g. temperature T ( x , y , z ) , mass density ρ ( x , y , z ) , or electric potential Φ ( x , y , z ) . To accumulate such quantities over a region in the plane or in space, we use multiple integrals.

DefinitionDouble Integral

Let f ( x , y ) be a continuous function on a region A ⊂ ℝ 2 . The double integral

∬ A f ( x , y ) d A (5.3)

represents the accumulated value of f over the region A .

  • If f ( x , y ) ≥ 0 , it represents the volume under the surface z = f ( x , y ) above the region A .

  • If f ( x , y ) takes negative values, the integral subtracts the volume between the surface and the x y -plane where f ( x , y ) < 0 : it is the volume above the plane minus the volume below it. This keeps the sign information of the function.

In particular, if f ( x , y ) ≡ 1 ,

∬ A 1 d A = Area ( A ) . (5.4)
The double integral as a sum of thin columns f(x,y)\,dA, here under z = 0.9 - 0.75x + 0.25y over [0,2][0,2]. Where the surface dips below the xy-plane the columns hang below it (red) and count as negative, just as area below the axis does for a single integral.
Figure 5.3. The double integral as a sum of thin columns f ( x , y ) d A , here under z = 0.9 − 0.75 x + 0.25 y over [ 0 , 2 ] × [ 0 , 2 ] . Where the surface dips below the x y -plane the columns hang below it (red) and count as negative, just as area below the axis does for a single integral.

With f ≡ 1 every column has height 1 , so the solid is a slab of thickness one and its volume is the same number as the area of its floor. That is why ∬ A 1 d A is an area, not a volume: the volume under a surface needs the height f ( x , y ) inside the integral.

Example

Find the volume of the solid below the plane z = 1 + y , bounded by the coordinate planes and the vertical plane 2 x + y = 2 .

Geometry of the solid bounded by the coordinate planes, the vertical plane 2x + y = 2, and the plane z = 1 + y.
Figure 5.4. Geometry of the solid bounded by the coordinate planes, the vertical plane 2 x + y = 2 , and the plane z = 1 + y .
Solution

Setup. The projection A of the solid onto the x y -plane is the triangle

A = { ( x , y ) : x ≥ 0 ,   y ≥ 0 ,   2 x + y ≤ 2 } . (5.5)

Since z = 1 + y , the volume is

V = ∬ A z d A = ∬ A ( 1 + y ) d x d y . (5.6)

Video Lecture 15, Part 2: The Volume Under a Plane (5:58)

Method 1: integrate w.r.t. y first. For a fixed x , y runs from 0 to 2 − 2 x and 0 ≤ x ≤ 1 . Thus

V = ∫ x = 0 1 ( ∫ y = 0 2 − 2 x ( 1 + y ) d y ) d x = ∫ 0 1 [ y + y 2 2 ] 0 2 − 2 x d x = ∫ 0 1 ( ( 2 − 2 x ) + 1 2 ( 2 − 2 x ) 2 ) d x = ∫ 0 1 ( 4 − 6 x + 2 x 2 ) d x = 5 3 . (5.7)
Integrating y first. For a fixed x the strip runs from y=0 up to the line, and the element of area is A= y\, x.
Figure 5.5. Integrating y first. For a fixed x the strip runs from y = 0 up to the line, and the element of area is Δ A = Δ y Δ x .

Video Lecture 15, Part 3: Horizontal Strips (11:07)

Method 2: integrate w.r.t. x first. For a fixed y , x runs from 0 to 1 − y 2 and 0 ≤ y ≤ 2 . Hence

V = ∫ y = 0 2 ( ∫ x = 0 1 − y / 2 ( 1 + y ) d x ) d y = ∫ 0 2 ( 1 + y ) [ x ] 0 1 − y / 2 d y = ∫ 0 2 ( 1 + y ) ( 1 − y 2 ) d y = ∫ 0 2 ( 1 + y 2 − y 2 2 ) d y = 5 3 . (5.8)
Integrating x first. For a fixed y the strip runs from x=0 to x=1-y/2, and the element of area is A= x\, y.
Figure 5.6. Integrating x first. For a fixed y the strip runs from x = 0 to x = 1 − y / 2 , and the element of area is Δ A = Δ x Δ y .

5.3When to Integrate y -First vs. x -First

Video Lecture 15, Part 4: Type I and Type II (14:24)

In evaluating a double integral, the choice of integration order depends on which limits are simpler or which inner integral is easier to evaluate. We typically choose the order that makes the region description and computation more convenient.

Type I Regions (Vertical Strips)

When the region A is bounded above and below by curves

y = y 2 ( x ) , y = y 1 ( x ) , (5.9)

and x varies between fixed limits a and b , we integrate with respect to y first:

∬ A f ( x , y ) d x d y = ∫ x = a b ( ∫ y = y 1 ( x ) y 2 ( x ) f ( x , y ) d y ) d x . (5.10)

The test for a Type I region is a vertical strip swept from x = a to x = b : every strip must enter the region on the same curve y 1 ( x ) and leave it on the same curve y 2 ( x ) . The inner limits are then those two curves, and the outer limits are plain numbers.

Type I region A: vertical slices, integrating y first.
Figure 5.7. Type I region A : vertical slices, integrating y first.

Type II Regions (Horizontal Strips)

When the region A is bounded on the left and right by curves

x = x 1 ( y ) , x = x 2 ( y ) , (5.11)

and y varies between fixed limits c and d , we integrate with respect to x first:

∬ A f ( x , y ) d x d y = ∫ y = c d ( ∫ x = x 1 ( y ) x 2 ( y ) f ( x , y ) d x ) d y . (5.12)

The test is the same, turned sideways: every horizontal strip, swept from y = c to y = d , must enter the region on x 1 ( y ) and leave it on x 2 ( y ) .

Type II region A: horizontal slices, integrating x first.
Figure 5.8. Type II region A : horizontal slices, integrating x first.

Either Description Works

If the region A satisfies both descriptions, either order of integration yields the same result:

∬ A f ( x , y ) d x d y = ∫ x = a b ( ∫ y = y 1 ( x ) y 2 ( x ) f ( x , y ) d y ) d x = ∫ y = c d ( ∫ x = x 1 ( y ) x 2 ( y ) f ( x , y ) d x ) d y . (5.13)
Regions where either order is valid.
Figure 5.9. Regions where either order is valid.

How to Choose the Order

Draw the region and try both strips. If an end of the strip switches from one boundary curve to another on the way across, that order needs the region split into pieces, one double integral for each piece. If the inner integral has no elementary antiderivative, try the other order. The two examples below show each case.

Example

Let A be the triangle under the lines y = x and y = 2 − x , above y = 0 (corners ( 0 , 0 ) , ( 2 , 0 ) , ( 1 , 1 ) ). Find ∬ A y d A .

Solution.

Vertical strips. For 0 ≤ x ≤ 1 the top of the strip lies on y = x ; for 1 ≤ x ≤ 2 it lies on y = 2 − x . The top switches lines at x = 1 , so the region must be split:

∫ 0 1 ∫ 0 x y d y d x + ∫ 1 2 ∫ 0 2 − x y d y d x = ∫ 0 1 x 2 2 d x + ∫ 1 2 ( 2 − x ) 2 2 d x = 1 6 + 1 6 = 1 3 . (5.14)

Horizontal strips. For every y in [ 0 , 1 ] the strip runs from the line x = y to the line x = 2 − y , the same two lines all the way up, so one double integral is enough:

∫ 0 1 ∫ y 2 − y y d x d y = ∫ 0 1 y ( 2 − 2 y ) d y = 1 − 2 3 = 1 3 . (5.15)

The same result, with half the work (Fig. 5.10).

The same triangle read two ways. A vertical strip ends on y=x for x&lt;1 and on y=2-x for x&gt;1, so that order needs two integrals; a horizontal strip always runs from x=y to x=2-y.
Figure 5.10. The same triangle read two ways. A vertical strip ends on y = x for x < 1 and on y = 2 − x for x > 1 , so that order needs two integrals; a horizontal strip always runs from x = y to x = 2 − y .
Example

Evaluate ∫ 0 1 ∫ x 1 e y 2 d y d x .

Solution.

The inner integral cannot be done as it stands: e y 2 has no elementary antiderivative. The limits describe the region 0 ≤ x ≤ 1 , x ≤ y ≤ 1 , the triangle above the diagonal y = x with corners ( 0 , 0 ) , ( 0 , 1 ) , ( 1 , 1 ) . Read with horizontal strips, the same triangle is 0 ≤ y ≤ 1 , 0 ≤ x ≤ y (Fig. 5.11). In the new inner integral e y 2 does not depend on x , so

∫ 0 1 ∫ 0 y e y 2 d x d y = ∫ 0 1 y e y 2 d y = [ 1 2 e y 2 ] 0 1 = e − 1 2 . (5.16)
The region of 0^1x^1 e^y^2\,dy\,dx. With y inside (left) the inner integral of e^y^2 cannot be done; read the same triangle with x inside (right) and it becomes y\,e^y^2.
Figure 5.11. The region of ∫ 0 1 ∫ x 1 e y 2 d y d x . With y inside (left) the inner integral of e y 2 cannot be done; read the same triangle with x inside (right) and it becomes y e y 2 .

A common mistake when changing the order is to keep the old limits. In the solid of the first example, the limit 2 − 2 x belongs to the order with y inside; with x inside, x runs from 0 to 1 − y 2 . Whenever the order changes, read the limits again from the picture.

RemarkProduct Integrands over Rectangular Regions

If the region A = [ a , b ] × [ c , d ] is a rectangle and the integrand separates as f ( x , y ) = g ( x ) h ( y ) , then

∬ A f ( x , y ) d x d y = ( ∫ a b g ( x ) d x ) ( ∫ c d h ( y ) d y ) . (5.17)

This property can greatly simplify double-integral evaluations.

Video Lecture 15, Part 5: Volume and Mass (21:02)

DefinitionTriple integral

For a continuous f on a solid R ⊂ ℝ 3 ,

∭ R f ( x , y , z ) d V (5.18)

is the accumulated signed total of f throughout the solid R . (If f ≡ 1 , it equals the volume of R ; if f = ρ is mass density, it equals the mass.)

Example

Find the volume of the solid bounded by the coordinate planes, the vertical plane 2 x + y = 2 , and the plane z = 1 + y via a triple integral

Solution.

Region and bounds. Projecting onto the x y -plane gives the triangle

A = { ( x , y ) :   0 ≤ x ≤ 1 ,     0 ≤ y ≤ 2 − 2 x } . (5.19)

For each ( x , y ) ∈ A , z ranges from the bottom z = 0 to the top z = 1 + y .

Triple–integral setup.

d V = d x d y d z (5.20)

as a result:

V = ∭ V d V = ∫ x = 0 1 ∫ y = 0 2 − 2 x ∫ z = 0 1 + y d z d y d x . (5.21)

Evaluate (inner integral first).

∫ z = 0 1 + y 1 d z = 1 + y , ⇒ V = ∫ 0 1 ∫ 0 2 − 2 x ( 1 + y ) d y d x . (5.22)

The remaining double integral is elementary:

∫ 0 1 [ y + y 2 2 ] 0 2 − 2 x d x = ∫ 0 1 ( 4 − 6 x + 2 x 2 ) d x =   5 3   . (5.23)
Example

On the same solid V as above, suppose the density (mass per unit volume) is ρ ( x , y , z ) = x + z . Find the total mass M .

Solution.

Triple–integral setup.

d M = ρ ( x , y , z ) d x d y d z (5.24)

as a result:

M = ∭ V ρ ( x , y , z ) d V = ∫ x = 0 1 ∫ y = 0 2 − 2 x ∫ z = 0 1 + y ( x + z ) d z d y d x . (5.25)

Step 1. Integrate in z .

∫ 0 1 + y ( x + z ) d z = [ x z + z 2 2 ] 0 1 + y = x ( 1 + y ) + ( 1 + y ) 2 2 . (5.26)

Step 2. Integrate in y , then x .

M = ∫ 0 1 ∫ 0 2 − 2 x ( x ( 1 + y ) + ( 1 + y ) 2 2 ) d y d x = ∫ 0 1 ( 2 x 3 3 − 5 x + 13 3 ) d x =   2   . (5.27)

5.4Applications of Integration

Video Lecture 15, Part 6: The Centre of Mass (25:26)

Center of mass

DefinitionCenter of mass

For any mass distribution with mass element d m and position vector r → = x 𝐢 ˆ + y 𝐣 ˆ + z 𝐤 ˆ , the center of mass is

r → cm = 1 M ∫ r → d m , M = ∫ d m . (5.28)

Specializations (same formula, different d m ):

Surface (lamina): d m = σ ( x , y ) d A   ⇒   M = ∬ σ d A , r → cm = 1 M ∬ r → σ d A . Volume (solid): d m = ρ ( x , y , z ) d V   ⇒   M = ∭ ρ d V , r → cm = 1 M ∭ r → ρ d V . (5.29)

Components: x cm = 1 M ∫ x d m ,   y cm = 1 M ∫ y d m ,   z cm = 1 M ∫ z d m .

Right triangular lamina with vertices (0,0), (L,0), (0,H).
Figure 5.12. Right triangular lamina with vertices ( 0 , 0 ) , ( L , 0 ) , ( 0 , H ) .
Example

Centroid of a Right Triangle (Uniform Density σ 0 )

Solution.

Step 1. Describe the region. The hypotenuse runs from ( L , 0 ) to ( 0 , H ) , so its equation is x L + y H = 1 , that is y = H ( 1 − x L ) . A convenient Type I description is therefore

R = { ( x , y ) :   0 ≤ x ≤ L , 0 ≤ y ≤ H ( 1 − x L ) } . (5.30)

Step 2. Mass.

M = ∬ R σ 0 d A = σ 0 ∫ 0 L ∫ 0 H ( 1 − x / L ) d y d x = σ 0 ∫ 0 L ( H − H L x ) d x = σ 0 H L 2 . (5.31)

Step 3. First moments.

∬ R x σ 0 d A = σ 0 ∫ 0 L ∫ 0 H ( 1 − x / L ) x d y d x = σ 0 ∫ 0 L x ( H − H L x ) d x = σ 0 H L 2 6 , ∬ R y σ 0 d A = σ 0 ∫ 0 L ∫ 0 H ( 1 − x / L ) y d y d x = σ 0 ∫ 0 L 1 2 ( H − H L x ) 2 d x = σ 0 H 2 L 6 . (5.32)

Step 4. Coordinates of the center of mass.

x c m = 1 M ∬ R x σ 0 d A = σ 0 ( H L 2 / 6 ) σ 0 ( H L / 2 ) = L 3 , y c m = 1 M ∬ R y σ 0 d A = σ 0 ( H 2 L / 6 ) σ 0 ( H L / 2 ) = H 3 . (5.33)

Result. The centroid is

( x c m , y c m ) = ( L 3 , H 3 ) . (5.34)

5.5Change of Variables in Integrals

Video Lecture 16, Part 1: Why New Coordinates (0:00)

Cartesian coordinates are rarely the natural ones for a physical region. A cylinder, a sphere or a cone has boundaries that are awkward to describe with x , y and z but trivial in the right coordinates: a sphere of radius a is simply r = a . The price is that the volume element is no longer d x d y d z , and the next two subsections derive what it becomes.

Round bodies come in two kinds. A body with spherical symmetry is the same in every direction from a centre (a planet, an atom, a charged ball); spherical coordinates fit it. A body with cylindrical symmetry is the same all around an axis (a wire, a pipe, a spinning top); cylindrical coordinates fit it. Any other change of variables is handled by the Jacobian (Section 5.6).

A spinning top shows both sides of this. It is nearly a solid cone standing on its point, and its mass, its balance point and its moment of inertia are all triple integrals over the cone (the example at the end of this section). With the apex at the origin, the axis along z , and height and base radius both h , the cone is x 2 + y 2 ≤ z , 0 ≤ z ≤ h . In Cartesian coordinates every horizontal slice is a disk, and its round edge needs square roots in the limits:

V = ∫ 0 h ∫ − z z ∫ − z 2 − x 2 z 2 − x 2 d y d x d z = 1 3 π h 3 . (5.35)

Measured instead by the distance ρ = x 2 + y 2 from the axis and the angle ϕ around it, the side of the cone is just ρ = z , and every limit is simple: 0 ≤ ρ ≤ z , 0 ≤ ϕ ≤ 2 π , 0 ≤ z ≤ h . Figure 5.13 shows why: small squares never fit a round edge, while pieces of rings fill it exactly.

The cone's slice at height z, a disk of radius z. Cut into squares dx\,dy (left), only the blue squares fit inside; the red ones stick out over the round edge however small they are. Cut into pieces of rings (right), the pieces fill the disk exactly.
Figure 5.13. The cone's slice at height z , a disk of radius z . Cut into squares d x d y (left), only the blue squares fit inside; the red ones stick out over the round edge however small they are. Cut into pieces of rings (right), the pieces fill the disk exactly.
RemarkThe price

The small volume is not d ρ d ϕ d z . With it the cone would have ∫ 0 h ∫ 0 2 π ∫ 0 z d ρ d ϕ d z = π h 2 , a length squared: an area, not a volume. The angle d ϕ has no length; the length around the axis is ρ d ϕ .

Change of Variables: Cartesian → Cylindrical Coordinates

Video Lecture 16, Part 2: Cylindrical Coordinates (6:18)

1. Definition of Cylindrical Coordinates

A point P ( x , y , z ) in space can be described using cylindrical coordinates ( ρ , ϕ , z ) defined as:

{ x = ρ cos ⁡ ( ϕ ) , y = ρ sin ⁡ ( ϕ ) , z = z . (5.36)

Here:

The position vector of the point is

r → = x i ˆ + y j ˆ + z k ˆ . (5.37)

Substituting the cylindrical coordinate relations:

r → = ρ ( cos ⁡ ( ϕ ) i ˆ + sin ⁡ ( ϕ ) j ˆ ) + z k ˆ . (5.38)

Thus, we can write

r → = ρ ρ ˆ + z k ˆ , (5.39)

where ρ ˆ is the radial unit vector defined by

ρ ˆ = cos ⁡ ( ϕ ) i ˆ + sin ⁡ ( ϕ ) j ˆ . (5.40)

2. The Local Unit Vectors

At each point, three mutually perpendicular unit vectors are defined:

ρ ˆ = cos ⁡ ( ϕ ) i ˆ + sin ⁡ ( ϕ ) j ˆ , ϕ ˆ = − sin ⁡ ( ϕ ) i ˆ + cos ⁡ ( ϕ ) j ˆ , k ˆ = k ˆ . (5.41)

These three vectors form a right-handed orthonormal basis:

ρ ˆ × ϕ ˆ = k ˆ , ϕ ˆ × k ˆ = ρ ˆ , k ˆ × ρ ˆ = ϕ ˆ . (5.42)

3. Differential Displacement d r →

Starting from

r → = ρ ρ ˆ + z k ˆ , (5.43)

we differentiate:

d r → = d ρ ρ ˆ + ρ d ρ ˆ + d z k ˆ . (5.44)

Since ρ ˆ depends only on ϕ ,

d ρ ˆ = ∂ ρ ˆ ∂ ϕ d ϕ . (5.45)

Compute this derivative:

∂ ρ ˆ ∂ ϕ = − sin ⁡ ( ϕ ) i ˆ + cos ⁡ ( ϕ ) j ˆ = ϕ ˆ . (5.46)

Hence,

d ρ ˆ = ϕ ˆ d ϕ . (5.47)

Substitute into d r → :

d r → = ρ ˆ d ρ + ρ ϕ ˆ d ϕ + k ˆ d z . (5.48)
Differential volume element in cylindrical coordinates.
Figure 5.14. Differential volume element in cylindrical coordinates.

4. Small Surface Element d s →

Each coordinate surface has a small area element with magnitude and direction as follows:

d s → ρ = ( ρ d ϕ ϕ ˆ ) × ( d z k ˆ ) = ( ρ d ϕ d z ) ρ ˆ , (surface of a cylinder, constant  ρ ) , d s → ϕ = ( d z k ˆ ) × ( d ρ ρ ˆ ) = ( d ρ d z ) ϕ ˆ , (surface of a plane at constant  ϕ ) , d s → z = ( d ρ ρ ˆ ) × ( ρ d ϕ ϕ ˆ ) = ( ρ d ρ d ϕ ) k ˆ , (surface of a disk, constant  z ) . (5.49)
d s → ρ = ρ d ϕ d z ρ ˆ , d s → ϕ = d ρ d z ϕ ˆ , d s → z = ρ d ρ d ϕ k ˆ . (5.50)

5. Small Volume Element d V

The infinitesimal volume element is formed by the product of the three small lengths:

d ρ , ρ d ϕ , d z . (5.51)

Thus,

d V = ρ d ρ d ϕ d z . (5.52)
RemarkWhy the factor ρ

Two small elements with the same d ρ , d ϕ and d z are not the same size: the one farther from the axis is bigger, because the same angle d ϕ sweeps a longer arc ρ d ϕ . Twice as far out, twice the volume. A common mistake is to write d V = d ρ d ϕ d z ; it forgets that d ϕ is an angle, not a length. As a check, the surface elements d s → z added over a disk of radius a give its area, ∫ 0 2 π ∫ 0 a ρ d ρ d ϕ = 2 π a 2 2 = π a 2 .

QuantityExpressionDescription
r → ρ ρ ˆ + z k ˆ Position vector
d r → ρ ˆ d ρ + ρ ϕ ˆ d ϕ + k ˆ d z Differential displacement
d s → ρ ρ d ϕ d z ρ ˆ Surface element on cylindrical wall
d s → ϕ d ρ d z ϕ ˆ Surface element on plane at constant ϕ
d s → z ρ d ρ d ϕ k ˆ Surface element on disk (constant z )
d V ρ d ρ d ϕ d z Volume element
Table 5.1. Summary of differential elements in cylindrical coordinates

Change of Variables: Cartesian → Spherical Coordinates

Video Lecture 16, Part 3: Spherical Coordinates (13:27)

1. Definition of Spherical Coordinates

A point P ( x , y , z ) in space can be described using spherical coordinates ( r , θ , ϕ ) defined as:

{ x = r sin ⁡ ( θ ) cos ⁡ ( ϕ ) , y = r sin ⁡ ( θ ) sin ⁡ ( ϕ ) , z = r cos ⁡ ( θ ) . (5.53)

Here:

The position vector of the point is

r → = x i ˆ + y j ˆ + z k ˆ . (5.54)

Substituting the spherical coordinate relations:

r → = r ( sin ⁡ ( θ ) cos ⁡ ( ϕ ) i ˆ + sin ⁡ ( θ ) sin ⁡ ( ϕ ) j ˆ + cos ⁡ ( θ ) k ˆ ) . (5.55)

Thus, we can write

r → = r r ˆ , (5.56)

where r ˆ is the radial unit vector defined by

r ˆ = sin ⁡ ( θ ) cos ⁡ ( ϕ ) i ˆ + sin ⁡ ( θ ) sin ⁡ ( ϕ ) j ˆ + cos ⁡ ( θ ) k ˆ . (5.57)

2. The Local Unit Vectors

At each point, three mutually perpendicular unit vectors are defined:

r ˆ = sin ⁡ ( θ ) cos ⁡ ( ϕ ) i ˆ + sin ⁡ ( θ ) sin ⁡ ( ϕ ) j ˆ + cos ⁡ ( θ ) k ˆ , θ ˆ = cos ⁡ ( θ ) cos ⁡ ( ϕ ) i ˆ + cos ⁡ ( θ ) sin ⁡ ( ϕ ) j ˆ − sin ⁡ ( θ ) k ˆ , ϕ ˆ = − sin ⁡ ( ϕ ) i ˆ + cos ⁡ ( ϕ ) j ˆ . (5.58)

These three vectors form a right-handed orthonormal basis:

r ˆ × θ ˆ = ϕ ˆ , θ ˆ × ϕ ˆ = r ˆ , ϕ ˆ × r ˆ = θ ˆ . (5.59)

3. Differential Displacement d r →

Differentiating the position vector

r → = r r ˆ , (5.60)

we obtain

d r → = d r r ˆ + r d r ˆ . (5.61)

To find d r ˆ , note that r ˆ depends on both θ and ϕ :

d r ˆ = ∂ r ˆ ∂ θ d θ + ∂ r ˆ ∂ ϕ d ϕ . (5.62)

Compute these derivatives:

∂ r ˆ ∂ θ = θ ˆ , ∂ r ˆ ∂ ϕ = sin ⁡ ( θ ) ϕ ˆ . (5.63)

Hence,

d r ˆ = θ ˆ d θ + sin ⁡ ( θ ) ϕ ˆ d ϕ . (5.64)

Substitute into d r → = d r r ˆ + r d r ˆ :

d r → = r ˆ d r + r θ ˆ d θ + r sin ⁡ ( θ ) ϕ ˆ d ϕ . (5.65)
Differential displacement in spherical coordinates
Figure 5.15. Differential displacement in spherical coordinates

4. Small Surface Element d s →

Each coordinate surface has a small area element with magnitude and direction as follows:

d s → r = ( r d θ θ ˆ ) × ( r sin ⁡ ( θ ) d ϕ ϕ ˆ ) = ( r 2 sin ⁡ ( θ ) d θ d ϕ ) r ˆ , (surface of a sphere, constant  r ) , d s → θ = ( r sin ⁡ ( θ ) d ϕ ϕ ˆ ) × ( d r r ˆ ) = ( r sin ⁡ ( θ ) d r d ϕ ) θ ˆ , (surface of a cone, constant  θ ) , d s → ϕ = ( d r r ˆ ) × ( r d θ θ ˆ ) = ( r d r d θ ) ϕ ˆ , (surface of a half-plane, constant  ϕ ) . (5.66)
d s → r = r 2 sin ⁡ ( θ ) d θ d ϕ r ˆ , d s → θ = r sin ⁡ ( θ ) d r d ϕ θ ˆ , d s → ϕ = r d r d θ ϕ ˆ . (5.67)

5. Small Volume Element d V

The infinitesimal volume is formed by the product of the three differential lengths:

d r , r d θ , r sin ⁡ ( θ ) d ϕ . (5.68)

Thus,

d V = r 2 sin ⁡ ( θ ) d r d θ d ϕ . (5.69)
RemarkWhy the factor sin ⁡ ( θ )

The arc along a circle of latitude has the radius r sin ⁡ ( θ ) , the distance from the z -axis. Near the poles that circle shrinks, so elements with the same d r , d θ and d ϕ become thinner and thinner, and their volume goes to zero with sin ⁡ ( θ ) .

ExampleThe surface of a sphere

On the sphere r = a every surface element is d s r = a 2 sin ⁡ ( θ ) d θ d ϕ . Adding them over the whole sphere,

S = ∫ 0 2 π ∫ 0 π a 2 sin ⁡ ( θ ) d θ d ϕ = a 2 ⋅ 2 π ⋅ [ − cos ⁡ ( θ ) ] 0 π = 4 π a 2 . (5.70)
RemarkCommon mistake

Forgetting the sin ⁡ ( θ ) . With d V = r 2 d r d θ d ϕ a ball of radius a would have the volume a 3 3 ⋅ π ⋅ 2 π instead of 4 3 π a 3 : the sin ⁡ ( θ ) is what makes the elements near the poles small.

QuantityExpressionDescription
r → r r ˆ Position vector
d r → r ˆ d r + r θ ˆ d θ + r sin ⁡ ( θ ) ϕ ˆ d ϕ Differential displacement
d s → r r 2 sin ⁡ ( θ ) d θ d ϕ r ˆ Surface element on sphere ( r = const )
d s → θ r sin ⁡ ( θ ) d r d ϕ θ ˆ Surface element on cone ( θ = const )
d s → ϕ r d r d θ ϕ ˆ Surface element on half-plane ( ϕ = const )
d V r 2 sin ⁡ ( θ ) d r d θ d ϕ Volume element
Table 5.2. Summary of differential elements in spherical coordinates

5.6Connection with the Jacobian Matrix

Video Lecture 16, Part 4: The Jacobian (19:50)

When transforming a multiple integral from one set of variables to another, the Jacobian determinant accounts for the change in scale between coordinate systems.

TheoremChange of Variables for Triple Integrals

Suppose we have a triple integral

∭ f ( u , v , w ) d u d v d w (5.71)

in some variables u , v , w . Let r , s , t be another set of variables related by

u = u ( r , s , t ) , v = v ( r , s , t ) , w = w ( r , s , t ) . (5.72)

as a result:

d u = ∂ u ∂ r d r + ∂ u ∂ s d s + ∂ u ∂ t d t , d v = ∂ v ∂ r d r + ∂ v ∂ s d s + ∂ v ∂ t d t , d w = ∂ w ∂ r d r + ∂ w ∂ s d s + ∂ w ∂ t d t . (5.73)

In matrix (differential) form,

( d u d v d w ) = ( ∂ u ∂ r ∂ u ∂ s ∂ u ∂ t ∂ v ∂ r ∂ v ∂ s ∂ v ∂ t ∂ w ∂ r ∂ w ∂ s ∂ w ∂ t ) ⏟ J ( u , v , w r , s , t ) ( d r d s d t ) . (5.74)

where J is the Jacobian of u , v , w with respect to r , s , t , then the triple integral in the new variables becomes

∭ f ( u , v , w ) d u d v d w = ∭ f ( u ( r , s , t ) , v ( r , s , t ) , w ( r , s , t ) ) | J | d r d s d t . (5.75)

The integrand on the right is the same function f , evaluated at the point the new coordinates describe; it is not a new function of r , s , t . The factor | J | is what corrects for the fact that a box d r d s d t in the new coordinates does not have the same volume as a box d u d v d w in the old ones.

Remark

The function f and the determinant J must both be expressed in the new variables, and the limits of integration must be adjusted to correspond to the transformed region.

RemarkWhy a determinant

A matrix M stretches every area (or volume) by the same factor, | det ⁡ ( M ) | (Chapter 3). A change of variables is not a matrix, but near any one point its small steps are related by one: ( d u , d v , d w ) is the Jacobian matrix times ( d r , d s , d t ) . So a small box d r d s d t is stretched by | J | , and the factor may change from point to point because the matrix does. Figure 5.16 shows it for polar coordinates.

Equal squares d\,d (left), carried by x = (), y = () onto the xy-plane (right), become pieces of rings of unequal size: each piece is times its square, a quarter near the origin and almost twice as big farther out. That stretching factor is the Jacobian, |J| = .
Figure 5.16. Equal squares d ρ d ϕ (left), carried by x = ρ cos ⁡ ( ϕ ) , y = ρ sin ⁡ ( ϕ ) onto the x y -plane (right), become pieces of rings of unequal size: each piece is ρ times its square, a quarter near the origin and almost twice as big farther out. That stretching factor is the Jacobian, | J | = ρ .
RemarkWhich way round

The Jacobian that multiplies d u d v is that of the old variables with respect to the new ones, J ( x , y u , v ) : the derivatives of x and y . Taking the derivatives of u and v with respect to x and y instead gives the Jacobian of the inverse map, 1 | J | .

Verifying the Jacobian for Cylindrical Coordinates

For the transformation ( ρ , ϕ , z ) → ( x , y , z ) ,

{ x = ρ cos ⁡ ( ϕ ) , y = ρ sin ⁡ ( ϕ ) , z = z . (5.76)

The Jacobian matrix is

J ( x , y , z ρ , ϕ , z ) = ( ∂ x ∂ ρ ∂ x ∂ ϕ ∂ x ∂ z ∂ y ∂ ρ ∂ y ∂ ϕ ∂ y ∂ z ∂ z ∂ ρ ∂ z ∂ ϕ ∂ z ∂ z ) = ( cos ⁡ ( ϕ ) − ρ sin ⁡ ( ϕ ) 0 sin ⁡ ( ϕ )   ρ cos ⁡ ( ϕ ) 0 0 0 1 ) . (5.77)

Compute its determinant:

det ⁡ ( J ) = | cos ⁡ ( ϕ ) − ρ sin ⁡ ( ϕ ) 0 sin ⁡ ( ϕ )   ρ cos ⁡ ( ϕ ) 0 0 0 1 | = ρ . (5.78)

Expanding along the last row, only the 1 in the corner survives, times the determinant left when its row and column are struck out:

det ⁡ ( J ) = 1 ⋅ ( ρ cos 2 ⁡ ( ϕ ) + ρ sin 2 ⁡ ( ϕ ) ) = ρ . (5.79)
RemarkCylindrical Volume Element
d V = | J | d ρ d ϕ d z = ρ d ρ d ϕ d z . (5.80)

Verifying the Jacobian for Spherical Coordinates

For the transformation ( r , θ , ϕ ) → ( x , y , z ) ,

{ x = r sin ⁡ ( θ ) cos ⁡ ( ϕ ) , y = r sin ⁡ ( θ ) sin ⁡ ( ϕ ) , z = r cos ⁡ ( θ ) . (5.81)

The Jacobian matrix is

J ( x , y , z r , θ , ϕ ) = ( ∂ x ∂ r ∂ x ∂ θ ∂ x ∂ ϕ ∂ y ∂ r ∂ y ∂ θ ∂ y ∂ ϕ ∂ z ∂ r ∂ z ∂ θ ∂ z ∂ ϕ ) . (5.82)

Compute each partial derivative:

∂ x ∂ r = sin ⁡ ( θ ) cos ⁡ ( ϕ ) , ∂ x ∂ θ = r cos ⁡ ( θ ) cos ⁡ ( ϕ ) , ∂ x ∂ ϕ = − r sin ⁡ ( θ ) sin ⁡ ( ϕ ) , ∂ y ∂ r = sin ⁡ ( θ ) sin ⁡ ( ϕ ) , ∂ y ∂ θ = r cos ⁡ ( θ ) sin ⁡ ( ϕ ) , ∂ y ∂ ϕ = r sin ⁡ ( θ ) cos ⁡ ( ϕ ) , ∂ z ∂ r = cos ⁡ ( θ ) , ∂ z ∂ θ = − r sin ⁡ ( θ ) , ∂ z ∂ ϕ = 0. (5.83)

Therefore,

det ⁡ ( J ) = | sin ⁡ ( θ ) cos ⁡ ( ϕ ) r cos ⁡ ( θ ) cos ⁡ ( ϕ ) − r sin ⁡ ( θ ) sin ⁡ ( ϕ ) sin ⁡ ( θ ) sin ⁡ ( ϕ ) r cos ⁡ ( θ ) sin ⁡ ( ϕ ) r sin ⁡ ( θ ) cos ⁡ ( ϕ ) cos ⁡ ( θ ) − r sin ⁡ ( θ ) 0 | . (5.84)

Expanding along the last row, with the signs + − + , each entry multiplies the 2 × 2 determinant left when its row and column are struck out:

det ⁡ ( J ) = cos ⁡ ( θ ) [ r 2 sin ⁡ ( θ ) cos ⁡ ( θ ) cos 2 ⁡ ( ϕ ) + r 2 sin ⁡ ( θ ) cos ⁡ ( θ ) sin 2 ⁡ ( ϕ ) ] − ( − r sin ⁡ ( θ ) ) [ r sin 2 ⁡ ( θ ) cos 2 ⁡ ( ϕ ) + r sin 2 ⁡ ( θ ) sin 2 ⁡ ( ϕ ) ] + 0 = r 2 sin ⁡ ( θ ) cos 2 ⁡ ( θ ) + r 2 sin 3 ⁡ ( θ ) = r 2 sin ⁡ ( θ ) ( cos 2 ⁡ ( θ ) + sin 2 ⁡ ( θ ) ) = r 2 sin ⁡ ( θ ) . (5.85)

Hence

| J | = r 2 sin ⁡ ( θ ) . (5.86)
RemarkSpherical Volume Element
d V = | J | d r d θ d ϕ = r 2 sin ⁡ ( θ ) d r d θ d ϕ . (5.87)
Example

Video Lecture 16, Part 6: The Solid Cone (31:28)

Centroid and Moment of Inertia of a Uniform Solid Cone. Find the z -coordinate of the centroid z c m of a uniform solid right circular cone (one nappe) whose height and base radius are both equal to h , that is, one with semi-vertical angle 45 ∘ . Also find the moment of inertia I z about its symmetry axis.

The uniform solid cone of height h and base radius r=h, with its vertex at the origin and its symmetry axis along z.
Figure 5.17. The uniform solid cone of height h and base radius r = h , with its vertex at the origin and its symmetry axis along z .
Solution

Geometry and coordinates. Place the apex at the origin and the axis along + z . The semi-vertical angle is 45 ∘ , so the base radius equals the height: both are h . That is why the radius of the cross-section at height z is itself z , and the lateral surface is ρ = z (for 0 ≤ z ≤ h ). Use cylindrical coordinates with d V = ρ d ρ d ϕ d z and constant density σ . Note that σ denotes the volume density here, written this way because ρ is already in use as the cylindrical radial coordinate. By symmetry x c m = y c m = 0 : turning the cone about its axis (changing ϕ ) changes nothing, a cylindrical symmetry, f ( ϕ + ϕ ′ ) = f ( ϕ ) (a body with spherical symmetry, f ( θ + θ ′ , ϕ + ϕ ′ ) = f ( θ , ϕ ) , would call for spherical coordinates). So cylindrical coordinates fit, and the cone is a stack of thin disks, the one at height z of radius z :

0 ≤ ϕ ≤ 2 π , 0 ≤ ρ ≤ z , 0 ≤ z ≤ h . (5.88)

Step 1: Mass.

M = ∭ σ d V = σ ∫ z = 0 h ∫ ρ = 0 z ∫ ϕ = 0 2 π ρ d ϕ d ρ d z = σ ⋅ 2 π ∫ 0 h [ ρ 2 2 ] 0 z d z = σ π ∫ 0 h z 2 d z = σ π [ z 3 3 ] 0 h = σ π h 3 3 . (5.89)

Step 2: Centroid z c m . Use z c m = 1 M ∭ z σ d V .

∭ z σ d V = σ ∫ 0 h ∫ 0 z ∫ 0 2 π z ρ d ϕ d ρ d z = σ ⋅ 2 π ∫ 0 h z [ ρ 2 2 ] 0 z d z = σ π ∫ 0 h z 3 d z = σ π [ z 4 4 ] 0 h = σ π h 4 4 . (5.90)

Hence

z c m = σ π h 4 / 4 σ π h 3 / 3 = 3 4 h . (5.91)

The balance point is above the middle: the disks grow toward the top, so most of the mass is high up (Fig. 5.18).

The cone as a stack of thin disks of radius z. They grow upward, so the centre of mass is at 34h, above the middle height 12h.
Figure 5.18. The cone as a stack of thin disks of radius z . They grow upward, so the centre of mass is at 3 4 h , above the middle height 1 2 h .

Step 3: Moment of inertia about the axis. I z = ∭ ρ 2 d m = σ ∭ ρ 2 d V .

I z = σ ∫ 0 h ∫ 0 z ∫ 0 2 π ρ 2 ( ρ d ϕ d ρ d z ) = σ ⋅ 2 π ∫ 0 h [ ρ 4 4 ] 0 z d z = σ π 2 ∫ 0 h z 4 d z = σ π 2 [ z 5 5 ] 0 h = σ π h 5 10 . (5.92)

In terms of M ,

I z = σ π h 5 10 = 3 10 ( σ π h 3 3 ) h 2 = 3 10 M h 2 . (5.93)

Common mistake. The upper limit of ρ is z , not h : the radius of the cone grows with the height. With ρ running to h the integral describes a cylinder, σ ∫ 0 h ∫ 0 2 π ∫ 0 h ρ d ρ d ϕ d z = π σ h 3 , three times the cone's mass.

Example

Video Lecture 16, Part 5: Straightening a Region (26:41)

Evaluate

∫ y = 0 1 ∫ x = 0 1 − y exp ( x − y x + y ) d x d y (5.94)

by the substitution u = x − y , v = x + y .

Solution.

Step 1. Map and Jacobian. The inverse relations are

x = u + v 2 , y = v − u 2 . (5.95)

Hence

J ( x , y u , v ) = | ∂ x ∂ u ∂ x ∂ v ∂ y ∂ u ∂ y ∂ v | = | 1 2 1 2 − 1 2 1 2 | = 1 2 ⇒ d x d y = 1 2 d u d v . (5.96)

Step 2. Region in ( u , v ) . The original region is R = { ( x , y ) : x ≥ 0 ,   y ≥ 0 ,   x + y ≤ 1 } . On R ,

v = x + y ∈ [ 0 , 1 ] , u = x − y ∈ [ − v , v ] (5.97)

(because along x = 0 ⇒ u = − v and along y = 0 ⇒ u = v ).

Step 3. Transform the integrand and limits.

x − y x + y = u v , ∬ R e ( x − y ) / ( x + y ) d x d y = ∫ v = 0 1 ∫ u = − v v e u / v 1 2 d u d v . (5.98)

Step 4. Integrate. For fixed v > 0 ,

∫ − v v e u / v d u = v [ e u / v ] − v v = v ( e − e − 1 ) = 2 v sinh ⁡ ( 1 ) . (5.99)

Therefore

∫ 0 1 ∫ − v v 1 2 e u / v d u d v = 1 2 ∫ 0 1 2 v sinh ⁡ ( 1 ) d v = sinh ⁡ ( 1 ) ∫ 0 1 v d v = 1 2 sinh ⁡ ( 1 ) . (5.100)
∫ 0 1 ∫ 0 1 − y exp ( x − y x + y ) d x d y = 1 2 sinh ⁡ ( 1 ) (5.101)

As a check, the original integral in x and y , computed numerically, gives 0.5876 , the value of 1 2 sinh ⁡ ( 1 ) .

The substitution u = x - y, v = x + y carries the triangle (left) onto the triangle -v u v, 0 v 1 (right): it turns by 45^ and grows, each side landing on one of the lines u = -v, u = v, v = 1, and every small square doubles its area, which is why dx\,dy = 12\,du\,dv.
Figure 5.19. The substitution u = x − y , v = x + y carries the triangle (left) onto the triangle − v ≤ u ≤ v , 0 ≤ v ≤ 1 (right): it turns by 45 ∘ and grows, each side landing on one of the lines u = − v , u = v , v = 1 , and every small square doubles its area, which is why d x d y = 1 2 d u d v .
Remark

Why these variables? The integrand depends only on x − y x + y , which is constant along every ray from the origin (it depends only on y x ); in the new variables it is e u / v , whose integral in u is easy. A common mistake is to change the integrand but not the limits: the new limits must describe the same region, found by mapping each of its sides (Fig. 5.19).

5.7Where Multiple Integrals Are Used

Every integral in this chapter is an accumulation: chop a region into pieces, evaluate a quantity on each piece, and add. What changes from one application to the next is only what sits inside the integral.

Remark

Two habits are worth taking from this chapter. First, draw the region and decide the order of integration before writing anything down; most of the difficulty in a multiple integral is in the limits, not the antiderivatives. Second, choose coordinates that match the symmetry of the region. A sphere described in Cartesian coordinates produces limits containing a 2 − x 2 − y 2 ; in spherical coordinates the same region is just r ≤ a , and the price is the single factor r 2 sin ⁡ ( θ ) .

Summary Table

ItemExpressionNotes
Double integral ∬ A f d A Volume under z = f ( x , y ) ; area if f ≡ 1 .
Triple integral ∭ R f d V Volume if f ≡ 1 ; mass if f = ρ .
Type I region ∫ a b ∫ y 1 ( x ) y 2 ( x ) f d y d x Vertical strips; y first.
Type II region ∫ c d ∫ x 1 ( y ) x 2 ( y ) f d x d y Horizontal strips; x first.
Separable integrand ( ∫ a b g ) ( ∫ c d h ) Only on a rectangle, with f = g ( x ) h ( y ) .
Centre of mass r → cm = 1 M ∫ r → d m d m = σ d A or ρ d V .
Cylindrical x = ρ cos ⁡ ( ϕ ) , y = ρ sin ⁡ ( ϕ ) , z = z d V = ρ d ρ d ϕ d z .
Spherical x = r sin ⁡ ( θ ) cos ⁡ ( ϕ ) , , z = r cos ⁡ ( θ ) d V = r 2 sin ⁡ ( θ ) d r d θ d ϕ .
Jacobian J = det ( ∂ ( u , v , w ) ∂ ( r , s , t ) ) d u d v d w = | J | d r d s d t .
Cylindrical Jacobian | J | = ρ Recovers d V = ρ d ρ d ϕ d z .
Spherical Jacobian | J | = r 2 sin ⁡ ( θ ) Recovers d V = r 2 sin ⁡ ( θ ) d r d θ d ϕ .

For the Interested Reader

The hardest part of a multiple integral is almost never the antiderivative — it is setting up the limits and choosing the coordinates. The sources below have far more practice at exactly that than we have room for. Everything listed is free.

Setting up double and triple integrals

Cylindrical, spherical and the Jacobian

Remark

We derived the volume elements twice on purpose: once geometrically, by multiplying the three small lengths along the coordinate directions, and once algebraically, as the Jacobian determinant. The geometric route shows why the factors ρ and r 2 sin ⁡ ( θ ) appear; the Jacobian route is the one that generalises to coordinates with no easy picture. It is worth being comfortable with both, because physics arguments usually quote the first and calculations usually use the second.

Video lectures

Each section of this chapter links to the part of a lecture that teaches it. The full lectures are below, and all of them are in the course playlist.

Video Lecture 15: Double and Triple Integrals: Volume, Mass and Centre of Mass (full lecture)

Video Lecture 16: Change of Variables and the Jacobian (full lecture)