MATH2023 Multivariable Calculus Chapter 6 Vector calculus

L2/L3 (Fall 2019)

Chapter 6 Vector calculus

1. Vector fields

Definition 6.1 A vector field (in 𝑛 dimensions) is a vector-valued function 𝐅: 𝐴 → ℝ𝑛 , where 𝐴

is a subset of ℝ𝑛 . It assigns an 𝑛-dimensional vector to each point (𝑥1 , 𝑥2 , … , 𝑥𝑛 ) in 𝐴.

From now on, in order to distinguish from a “vector field” which is a vector-valued function, a
“function” refers only to a scalar-valued (or real-valued) function unless otherwise specified.

Example 6.2 Let 𝐷 be a region in ℝ3 and let 𝑓: 𝐷 → ℝ be a function of three real variables
having first partial derivatives. Then its gradient ∇𝑓: 𝐷 → ℝ3 is a vector field given by
𝜕𝑓 𝜕𝑓 𝜕𝑓
∇𝑓(𝑥, 𝑦, 𝑧) = 〈 (𝑥, 𝑦, 𝑧), (𝑥, 𝑦, 𝑧), (𝑥, 𝑦, 𝑧)〉.
𝜕𝑥 𝜕𝑦 𝜕𝑧

Example 6.3 (For those who are taking or have taken MATH2350/2351/2352) Let 𝑓: 𝐴 → ℝ be a
function of two real variables. Then the vector field 𝐅̂: 𝐴 → ℝ2 given by
1 𝑦
𝐅̂(𝑥, 𝑦) = 〈1, 𝑓(𝑥, 𝑦)〉
√1 + [𝑓(𝑥, 𝑦)]2
is called the direction field (or the slope field) for the first-order
differential equation
𝑥
𝑦 ′ = 𝑓(𝑥, 𝑦).
Given each point (𝑎, 𝑏) ∈ 𝐴, there is a solution 𝑦 = 𝑦(𝑥) whose
graph passes through (𝑎, 𝑏) (i.e. 𝑦(𝑎) = 𝑏). 𝐅̂(𝑎, 𝑏) is then a unit
tangent vector of such a graph at (𝑎, 𝑏). Direction field for the
differential equation 𝑦 ′ = 𝑥 − 𝑦

As in Example 6.3, we may visually present a vector field 𝐅 in two or three dimensions as follows.
We first choose some particular points in the domain of 𝐅. Then at each of these chosen points
(𝑎, 𝑏), we draw the vector 𝐅(𝑎, 𝑏) which is assigned to the point (𝑎, 𝑏) by 𝐅.

Example 6.4 Sketch the vector fields 𝐅(𝑥, 𝑦) = 𝑥𝐢 + 𝑦𝐣 and 𝐆(𝑥, 𝑦) = (−𝑦)𝐢 + 𝑥𝐣.
Solution:
𝑦 𝑦
(a) (b)

𝑥 𝑥

Page 1 of 40
MATH2023 Multivariable Calculus Chapter 6 Vector calculus
L2/L3 (Fall 2019)

Definition 6.5 Let 𝐷 be a region in ℝ𝑛 and 𝐅: 𝐷 → ℝ𝑛 be a vector field whose component

functions are continuous. A field line (or trajectory, or streamline) of 𝐅 is a curve in 𝐷 such that
𝐅 is tangent to it at every point. In other words, if 𝐫: 𝐼 → 𝐷 is a parametrization of a field line of
𝐅, then there exists a continuous function 𝜆: 𝐼 → (0, +∞) such that
𝐫 ′ (𝑡) = 𝜆(𝑡)𝐅(𝐫(𝑡))
for every 𝑡.

Remark 6.6 In the simple case when 𝑛 = 2, we may write 𝐅(𝑥, 𝑦) = 〈𝑓(𝑥, 𝑦), 𝑔(𝑥, 𝑦)〉. Then if
𝐫(𝑡) = 〈𝑥(𝑡), 𝑦(𝑡)〉 is a field line, its components 𝑥 and 𝑦 must satisfy the differential equation
𝑑𝑥 𝑑𝑦
= .
𝑓(𝑥, 𝑦) 𝑔(𝑥, 𝑦)
We can therefore find the equations (or parametric equations) of the field lines of 𝐅 if we are able
to solve the above differential equation. In Example 6.3, the field lines of the direction field for the
differential equation 𝑦 ′ = 𝑓(𝑥, 𝑦) are simply the graphs of solutions to this differential equation.

Example 6.7 Sketch the field lines of the vector field 𝐅(𝑥, 𝑦) = 〈−𝑦, 𝑥〉.

Solution:
We solve the differential equation 𝑦

𝑑𝑥 𝑑𝑦
= ,
−𝑦 𝑥
i.e. 𝑥 𝑑𝑥 = −𝑦 𝑑𝑦. Taking antiderivatives on both sides we get 𝑥
2 2
𝑥 + 𝑦 = 𝐶,
where 𝐶 is an arbitrary non-negative constant. So the field lines
of 𝐅 are circles centered at the origin.

Example 6.8 Find the parametric equations of the field lines of the vector field
𝐅(𝑥, 𝑦, 𝑧) = 𝑥𝑧𝐢 + 𝑦𝑧𝐣 + 𝑥𝐤.
Solution:
On solving the differential equations
𝑑𝑥 𝑑𝑦 𝑑𝑧
= = ,
𝑥𝑧 𝑦𝑧 𝑥
1
𝑑𝑥 =
1
𝑑𝑦 ln|𝑥| = ln|𝑦| + 𝐶 𝑦 = 𝐶1 𝑥
i.e. {𝑥 𝑦 , we obtain { 1 2 , i.e. { 1 where 𝐶1 , 𝐶2 are arbitrary
𝑑𝑥 = 𝑧 𝑑𝑧 𝑥 = 2 𝑧 + 𝐶2 𝑥 = 2 𝑧 2 + 𝐶2
constants, 𝐶1 ≠ 0 and 𝐶2 > 0. Therefore the field lines of 𝐅 have parametric equations
𝑥 = 𝑡 2 /2 + 𝐶2
{𝑦 = 𝐶1 (𝑡 2 /2 + 𝐶2 ) , where 𝑡 ∈ (−∞, +∞).
𝑧=𝑡

Page 2 of 40
MATH2023 Multivariable Calculus Chapter 6 Vector calculus
L2/L3 (Fall 2019)

Definition 6.9 Let 𝐷 be a region in ℝ𝑛 . We say that a vector field 𝐅: 𝐷 → ℝ𝑛 is conservative

in 𝑫 if there exists a function 𝑓: 𝐷 → ℝ of 𝑛 real variables such that
𝐅 = ∇𝑓,
i.e. if 𝐅 is the gradient of some function 𝑓. In this case we also say that 𝑓 is a potential function
of the conservative vector field 𝐅 on 𝐷.

Remark 6.10 There are “conservative” vector fields, but there are no “liberal” vector fields.

Example 6.11 The vector field 𝐅: ℝ3 ∖ {(0, 0, 0)} → ℝ3 defined by

1 Denoting 𝐫 ≔ 〈𝑥, 𝑦, 𝑧〉 and
𝐅(𝑥, 𝑦, 𝑧) = − 3
(𝑥𝐢 + 𝑦𝐣 + 𝑧𝐤) 𝑟 ≔ ‖𝐫‖, we may also write
(𝑥 2 + 𝑦 2 + 𝑧 2 )2 1
𝐅(𝐫) = − 𝐫̂
𝑟2
is called a gravitational field. 𝐅 is conservative in ℝ3 ∖ {(0, 0, 0)}
and its potential function is
because it is the gradient of the function
1
1 𝜙(𝐫) = .
𝑟
𝜙(𝑥, 𝑦, 𝑧) = .
√𝑥 2 + 𝑦 2 + 𝑧 2

Example 6.12 Determine whether the vector field 𝐅: ℝ2 → ℝ2 defined by

𝐅(𝑥, 𝑦) = 〈𝑥 2 + 𝑦 2 , 2𝑥𝑦〉
is conservative in ℝ2 .

Solution:
To determine whether 𝐅 is conservative in ℝ2 , we ask whether there exists a function 𝑓: ℝ2 → ℝ
such that 𝐅 = ∇𝑓, i.e.
𝑓 (𝑥, 𝑦) = 𝑥 2 + 𝑦 2
{𝑥 .
𝑓𝑦 (𝑥, 𝑦) = 2𝑥𝑦
The second equation of the system requires that

𝑓(𝑥, 𝑦) = ∫ 2𝑥𝑦 𝑑𝑦 = 𝑥𝑦 2 + 𝑔(𝑥)

for some function 𝑔 depending on 𝑥 only. Taking partial derivatives with respect to 𝑥 we have
𝑓𝑥 (𝑥, 𝑦) = 𝑦 2 + 𝑔′ (𝑥),
so the first equation of the system requires that 𝑔′ (𝑥) = 𝑥 2 , i.e.
1 3
𝑔(𝑥) = ∫ 𝑥 2 𝑑𝑥 = 𝑥 +𝑘
3
for some constant 𝑘. Thus if we simply take 𝑘 = 0 and let
1 3
𝑥 + 𝑥𝑦 2 ,
𝑓(𝑥, 𝑦) =
3
then 𝐅 = ∇𝑓. This shows that 𝐅 is indeed conservative in ℝ2 .

Page 3 of 40
MATH2023 Multivariable Calculus Chapter 6 Vector calculus
L2/L3 (Fall 2019)

Example 6.13 Show that the vector field 𝐅: ℝ3 → ℝ3 defined by

𝐅(𝑥, 𝑦, 𝑧) = 𝑥𝐣
is not conservative in ℝ3 .

Proof:
Suppose on the contrary that 𝐅 is conservative in ℝ3 . Then there exists a function 𝑓: ℝ3 → ℝ
such that 𝐅 = ∇𝑓, i.e.
𝑓𝑥 (𝑥, 𝑦, 𝑧) = 0
{𝑓𝑦 (𝑥, 𝑦, 𝑧) = 𝑥
𝑓𝑧 (𝑥, 𝑦, 𝑧) = 0
for every (𝑥, 𝑦, 𝑧) ∈ ℝ3 . But then this system implies that
𝜕 𝜕
𝑓𝑥𝑦 (𝑥, 𝑦, 𝑧) = 0=0 but 𝑓𝑦𝑥 (𝑥, 𝑦, 𝑧) = 𝑥=1
𝜕𝑦 𝜕𝑥
which contradicts the mixed derivative theorem. Therefore 𝐅 cannot be conservative. ∎

Theorem 6.14 Let 𝑅 be a region in ℝ2 and 𝐅: 𝑅 → ℝ2 be a vector field

𝐅(𝑥, 𝑦) = 〈𝑓(𝑥, 𝑦), 𝑔(𝑥, 𝑦)〉
whose components 𝑓 and 𝑔 have continuous partial derivatives. If 𝐅 is conservative in 𝑅, then
𝑔𝑥 − 𝑓𝑦 = 0.

Theorem 6.15 Let 𝐷 be a region in ℝ3 and 𝐅: 𝐷 → ℝ3 be a vector field

𝐅(𝑥, 𝑦, 𝑧) = 〈𝑓(𝑥, 𝑦, 𝑧), 𝑔(𝑥, 𝑦, 𝑧), ℎ(𝑥, 𝑦, 𝑧)〉
whose components 𝑓, 𝑔 and ℎ have continuous partial derivatives. If 𝐅 is conservative in 𝐷,
then
〈ℎ𝑦 − 𝑔𝑧 , 𝑓𝑧 − ℎ𝑥 , 𝑔𝑥 − 𝑓𝑦 〉 = 〈0, 0, 0〉.

Proof. If 𝐅 is conservative in 𝐷, then there exists a function 𝜙: 𝐷 → ℝ such that 𝐅 = ∇𝜙, i.e.
〈𝑓(𝑥, 𝑦, 𝑧), 𝑔(𝑥, 𝑦, 𝑧), ℎ(𝑥, 𝑦, 𝑧)〉 = 〈𝜙𝑥 (𝑥, 𝑦, 𝑧), 𝜙𝑦 (𝑥, 𝑦, 𝑧), 𝜙𝑧 (𝑥, 𝑦, 𝑧)〉.
Therefore
ℎ𝑦 − 𝑔𝑧 = 𝜙𝑧𝑦 − 𝜙𝑦𝑧 = 0
{𝑓𝑧 − ℎ𝑥 = 𝜙𝑥𝑧 − 𝜙𝑧𝑥 = 0
𝑔𝑥 − 𝑓𝑦 = 𝜙𝑦𝑥 − 𝜙𝑥𝑦 = 0
by the mixed derivative theorem. ∎

Definition 6.16 Let 𝐅 = 〈𝑓, 𝑔, ℎ〉 be a vector field in three dimensions whose component
functions 𝑓, 𝑔 and ℎ have continuous partial derivatives. Then the curl of 𝐅 is the vector field
in three dimensions defined by
∇ × 𝐅 ≔ 〈ℎ𝑦 − 𝑔𝑧 , 𝑓𝑧 − ℎ𝑥 , 𝑔𝑥 − 𝑓𝑦 〉.

Page 4 of 40
MATH2023 Multivariable Calculus Chapter 6 Vector calculus
L2/L3 (Fall 2019)

Remark 6.17 We use the notation ∇ × 𝐅 to denote the curl of 𝐅 = 〈𝑓, 𝑔, ℎ〉 because the formula
can be easily remembered using a formal determinant
𝐢 𝐣 𝐤
𝜕 𝜕 𝜕 𝜕ℎ 𝜕𝑔 𝜕ℎ 𝜕𝑓 𝜕𝑔 𝜕𝑓
| | = ( − ) 𝐢 − ( − ) 𝐣 + ( − ) 𝐤 = 〈ℎ𝑦 − 𝑔𝑧 , 𝑓𝑧 − ℎ𝑥 , 𝑔𝑥 − 𝑓𝑦 〉.
𝜕𝑥 𝜕𝑦 𝜕𝑧 𝜕𝑦 𝜕𝑧 𝜕𝑥 𝜕𝑧 𝜕𝑥 𝜕𝑦
𝑓 𝑔 ℎ
𝜕 𝜕 𝜕
If we think of ∇ = 〈𝜕𝑥 , 𝜕𝑦 , 𝜕𝑧〉, then the above formal determinant is the same as what we obtain

when computing the “cross product” ∇ × 𝐅. This understanding of the symbol ∇ agrees with the
previous notation ∇𝑓 for the gradient of a function 𝑓, because we may also think of
𝜕 𝜕 𝜕 𝜕𝑓 𝜕𝑓 𝜕𝑓
∇𝑓 = 〈 , , 〉 𝑓 = 〈 , , 〉.
𝜕𝑥 𝜕𝑦 𝜕𝑧 𝜕𝑥 𝜕𝑦 𝜕𝑧

Remark 6.18 The curl ∇ × 𝐅 of a vector field 𝐅 is also a vector field.

(∇ × 𝐅)(𝑎, 𝑏, 𝑐)
Suppose that 𝐅 represents the velocity field of a fluid flowing in ℝ3 .
Then at each point (𝑎, 𝑏, 𝑐), the fluid tends to rotate about the axis
indicated by the direction of the vector (∇ × 𝐅)(𝑎, 𝑏, 𝑐), following the
(𝑎, 𝑏, 𝑐)
right-handed convention; while the length of the vector (∇ × 𝐅)(𝑎, 𝑏, 𝑐)
measures how quickly the fluid rotates around the axis. So we
Direction of fluid flow
sometimes say that 𝐅 is irrotational if ∇ × 𝐅 ≡ 𝟎. The field lines of an
irrotational vector field cannot form closed loops. In other words, if the
field lines of 𝐅 form closed loops, then ∇ × 𝐅 ≢ 𝟎.

Example 6.19 Let 𝑓 be a differentiable function of one real variable. Find the curl of
𝐅(𝑥, 𝑦, 𝑧) = 𝑓(𝑥 2 + 𝑦 2 + 𝑧 2 )(2𝑥𝐢 + 2𝑦𝐣 + 2𝑧𝐤).
𝐅(𝐫) = 2𝑟𝑓(𝑟 2 )𝐫̂

Solution:
The curl of 𝐅 is given by
𝐢 𝐣 𝐤
𝜕 𝜕 𝜕
∇ × 𝐅 = || ||
𝜕𝑥 𝜕𝑦 𝜕𝑧
2𝑥𝑓(𝑥 2 + 𝑦 2 + 𝑧 2 ) 2𝑦𝑓(𝑥 2 + 𝑦 2 + 𝑧 2 ) 2𝑧𝑓(𝑥 2 + 𝑦 2 + 𝑧 2 )
= [2𝑧𝑓 ′ (𝑥 2 + 𝑦 2 + 𝑧 2 )(2𝑦) − 2𝑦𝑓 ′ (𝑥 2 + 𝑦 2 + 𝑧 2 )(2𝑧)]𝐢
−[2𝑧𝑓 ′ (𝑥 2 + 𝑦 2 + 𝑧 2 )(2𝑥) − 2𝑥𝑓 ′ (𝑥 2 + 𝑦 2 + 𝑧 2 )(2𝑧)]𝐣
+[2𝑦𝑓 ′ (𝑥 2 + 𝑦 2 + 𝑧 2 )(2𝑥) − 2𝑥𝑓 ′ (𝑥 2 + 𝑦 2 + 𝑧 2 )(2𝑦)]𝐤
= 𝟎.
[It turns out that 𝐅 is also conservative in ℝ3 . In fact, if 𝑔 is an antiderivative of 𝑓, then
𝜙(𝑥, 𝑦, 𝑧) = 𝑔(𝑥 2 + 𝑦 2 + 𝑧 2 ) is a potential function of 𝐅 on ℝ3 .] 𝜙(𝐫) = 𝑔(𝑟 2 )

Page 5 of 40
MATH2023 Multivariable Calculus Chapter 6 Vector calculus
L2/L3 (Fall 2019)

Theorem 6.15 can now be rephrased as follows.

Corollary 6.20 Let 𝐷 be a region in ℝ3 . If 𝑓: 𝐷 → ℝ is a function of three real variables that

has continuous second partial derivatives, then its gradient vector field has zero curl on 𝐷, i.e.
∇ × (∇𝑓) = 𝟎.
In particular, every conservative vector field 𝐅 in 𝐷 has zero curl on 𝐷, i.e.
∇ × 𝐅 = 𝟎.

Example 6.21 Show that the vector field

𝐅(𝑥, 𝑦, 𝑧) = 〈𝑥𝑧, 𝑥𝑦𝑧, −𝑦 2 〉
is not conservative in ℝ3 .

Proof:
The curl of 𝐅 is given by
𝐢 𝐣 𝐤
𝜕 𝜕 𝜕
∇ × 𝐅 = || ||
𝜕𝑥 𝜕𝑦 𝜕𝑧
𝑥𝑧 𝑥𝑦𝑧 −𝑦 2
𝜕 𝜕 𝜕 𝜕 𝜕 𝜕
=〈 (−𝑦 2 ) − (𝑥𝑦𝑧), (𝑥𝑧) − (−𝑦 2 ), (𝑥𝑦𝑧) − (𝑥𝑧)〉
𝜕𝑦 𝜕𝑧 𝜕𝑧 𝜕𝑥 𝜕𝑥 𝜕𝑦
= 〈−2𝑦 − 𝑥𝑦, 𝑥, 𝑦𝑧〉
≠ 𝟎.
Since the curl of 𝐅 is not the zero vector field, 𝐅 is not conservative in ℝ3 . ∎

Definition 6.22 Let 𝐅 = 〈𝑓1 , 𝑓2 , … , 𝑓𝑛 〉 be a vector field in 𝑛 dimensions, whose components

have continuous partial derivatives. Then the divergence of 𝐅 is the function of 𝑛 real variables
defined by
𝜕𝑓1 𝜕𝑓2 𝜕𝑓𝑛
∇⋅𝐅 = + +⋯+ .
𝜕𝑥1 𝜕𝑥2 𝜕𝑥𝑛
In particular, the divergence of a vector field 𝐅 = 〈𝑓, 𝑔, ℎ〉 (in three dimensions) is the function
∇ ⋅ 𝐅 = 𝑓𝑥 + 𝑔𝑦 + ℎ𝑧
The divergence of 𝐅 is also known as
of three real variables. the flux density of 𝐅. We will explain
this alternative name in Section 5.

Remark 6.23 We use the notation ∇ ⋅ 𝐅 for the divergence of 𝐅 because the formula can also be
𝜕 𝜕 𝜕
recovered if we think of ∇ = 〈𝜕𝑥 , 𝜕𝑦 , 𝜕𝑧〉 as before, whose “dot product” with 𝐅 = 〈𝑓, 𝑔, ℎ〉 is

𝜕 𝜕 𝜕 𝜕𝑓 𝜕𝑔 𝜕ℎ
〈 , , 〉 ⋅ 〈𝑓, 𝑔, ℎ〉 = + + .
𝜕𝑥 𝜕𝑦 𝜕𝑧 𝜕𝑥 𝜕𝑦 𝜕𝑧

Page 6 of 40
MATH2023 Multivariable Calculus Chapter 6 Vector calculus
L2/L3 (Fall 2019)

Remark 6.24 The divergence ∇ ⋅ 𝐅 of a vector field 𝐅 is a function. Suppose that 𝐅 represents
the velocity field of a fluid flowing in ℝ3 . Then at each point (𝑎, 𝑏, 𝑐), the value (∇ ⋅ 𝐅)(𝑎, 𝑏, 𝑐)
measures how quickly the fluid is “spreading away” from the point (𝑎, 𝑏, 𝑐) (the fluid is “expanding”
if the value is positive, and is “compressing” if the value is negative). So we sometimes say that 𝐅
is incompressible if ∇ ⋅ 𝐅 ≡ 0. The field lines of an incompressible vector field do not begin or end;
they either form closed loops or go away to infinity in both ends.

Example 6.25 Find the divergence of the gravitational field

1
𝐅(𝑥, 𝑦, 𝑧) = − 3
(𝑥𝐢 + 𝑦𝐣 + 𝑧𝐤).
(𝑥 2 + 𝑦2 + 𝑧 2 )2

Solution:
𝜕 −𝑥 𝜕 −𝑦 𝜕 −𝑧
∇⋅𝐅= 3 + 3 + 3
𝜕𝑥 𝜕𝑦 𝜕𝑧
(𝑥 2 + 𝑦 2 + 𝑧 2 )2 (𝑥 2 + 𝑦 2 + 𝑧 2 )2 (𝑥 2 + 𝑦 2 + 𝑧 2 )2

2𝑥 2 − 𝑦 2 − 𝑧 2 2𝑦 2 − 𝑥 2 − 𝑧 2 2𝑧 2 − 𝑥 2 − 𝑦 2
= 5+ 5+ 5 = 0.
(𝑥 2 + 𝑦2 + 𝑧 2 )2 (𝑥 2 + 𝑦2 + 𝑧 2 )2 (𝑥 2 + 𝑦2 + 𝑧 2 )2

Theorem 6.26 If 𝐅 is a vector field in three dimensions whose components have continuous
second partial derivatives, then the curl of 𝐅 has zero divergence, i.e.
∇ ⋅ (∇ × 𝐅) = 0.

Proof. Let 𝐅 = 〈𝑓, 𝑔, ℎ〉. Then

𝜕 𝜕 𝜕
∇ ⋅ (∇ × 𝐅) = (ℎ𝑦 − 𝑔𝑧 ) + (𝑓𝑧 − ℎ𝑥 ) + (𝑔𝑥 − 𝑓𝑦 )
𝜕𝑥 𝜕𝑦 𝜕𝑧
= (ℎ𝑦𝑥 − 𝑔𝑧𝑥 ) + (𝑓𝑧𝑦 − ℎ𝑥𝑦 ) + (𝑔𝑥𝑧 − 𝑓𝑦𝑧 ) = 0
by the mixed derivatives theorem. ∎

Example 6.27 Show that

𝐅(𝑥, 𝑦, 𝑧) = 〈𝑥𝑧, 𝑥𝑦𝑧, −𝑦 2 〉
is not the curl of any vector field.

Proof:
The divergence of 𝐅 is the function
𝜕 𝜕 𝜕
∇⋅𝐅= (𝑥𝑧) + (𝑥𝑦𝑧) + (−𝑦 2 ) = 𝑧 + 𝑥𝑧.
𝜕𝑥 𝜕𝑦 𝜕𝑧
Since the divergence of 𝐅 is not the zero function, 𝐅 is not the curl of any vector field. ∎

Page 7 of 40
MATH2023 Multivariable Calculus Chapter 6 Vector calculus
L2/L3 (Fall 2019)

2. Line integrals

Definition 6.28 Let 𝑓 be a continuous function of 𝑛 real variables and let 𝐶 be a smooth curve
in ℝ𝑛 parametrized by a vector-valued function 𝐫: [𝑎, 𝑏] → ℝ𝑛 . Then the line integral of the
function 𝑓 over 𝐶 is
𝑏
∫𝑓 𝑑𝑠 ≔ ∫ 𝑓(𝐫(𝑡))‖𝐫 ′ (𝑡)‖𝑑𝑡.
𝐶 𝑎 𝑑𝑠 = ‖𝐫 ′ (𝑡)‖𝑑𝑡

Conceptually, a “line integral” should better be called a “curve integral”. Let’s stick to the tradition
and use the name “line integral” though.
𝑧
𝑧 = 𝑓(𝑥, 𝑦)
𝑦
Remark 6.29 Geometrically, if 𝐶 is a smooth curve
2
in ℝ and 𝑓 is a function of two real variables with
𝑓(𝑥, 𝑦) ≥ 0 for every point (𝑥, 𝑦) on the curve 𝐶,
then ∫𝐶 𝑓 𝑑𝑠 represents the area of the “curtain” 𝐶

under the graph of 𝑓 over 𝐶.

𝑥

Remark 6.30 The arc-length of a smooth curve 𝐶 is the line integral of 1 over the curve 𝐶, i.e.
𝑏
∫𝑑𝑠 = ∫ ‖𝐫 ′ (𝑡)‖𝑑𝑡.
𝐶 𝑎

Remark 6.31 If a wire is described by a smooth curve 𝐶 in ℝ3 and 𝑓(𝑥, 𝑦, 𝑧) is the linear
density (or charge density) of the wire at the point (𝑥, 𝑦, 𝑧), then ∫𝐶 𝑓(𝑥, 𝑦, 𝑧)𝑑𝑠 represents the
total mass of (or total charge carried by) the wire.

Lemma 6.32 The line integral of a function over a smooth curve is independent of
reparametrization of the curve.

Example 6.33 Let 𝐶 be the circular helix parametrized by

𝐫(𝑡) = 〈cos 𝑡 , sin 𝑡 , 𝑡〉
for 𝑡 ∈ [0, 2𝜋]. Evaluate the line integral ∫𝐶 𝑒 −𝑧 𝑑𝑠.

Solution:
Since 𝐫 ′ (𝑡) = 〈− sin 𝑡 , cos 𝑡 , 1〉, we have
‖𝐫 ′ (𝑡)‖ = √(− sin 𝑡)2 + (cos 𝑡)2 + 12 = √2
for every 𝑡 ∈ (0, 2𝜋). Therefore
2𝜋 2𝜋
∫𝑒 −𝑧 𝑑𝑠 = ∫ 𝑒 −𝑡 ‖𝐫 ′ (𝑡)‖𝑑𝑡 = ∫ 𝑒 −𝑡 √2 𝑑𝑡 = √2(1 − 𝑒 −2𝜋 ).
𝐶 0 0

Page 8 of 40
MATH2023 Multivariable Calculus Chapter 6 Vector calculus
L2/L3 (Fall 2019)

Example 6.34 Let 𝑎 > 0 and 𝑏 > 0. Compute the line integral of the function
𝑓(𝑥, 𝑦) = √(𝑏𝑥/𝑎)2 + (𝑎𝑦/𝑏)2
over the ellipse 𝐶 with equation 𝑥 2 /𝑎2 + 𝑦 2 /𝑏 2 = 1.

Solution:
The given ellipse 𝐶 can be parametrized by 𝐫(𝑡) = 〈𝑎 cos 𝑡 , 𝑏 sin 𝑡〉 for 𝑡 ∈ [0, 2𝜋] . Now
𝐫 ′ (𝑡) = 〈−𝑎 sin 𝑡 , 𝑏 cos 𝑡〉 and ‖𝐫 ′ (𝑡)‖ = √(−𝑎 sin 𝑡)2 + (𝑏 cos 𝑡)2 for every 𝑡 ∈ (0, 2𝜋), so

2𝜋 2 2
𝑏 𝑎
∫𝑓(𝑥, 𝑦)𝑑𝑠 = ∫ √( 𝑎 cos 𝑡) + ( 𝑏 sin 𝑡) √(−𝑎 sin 𝑡)2 + (𝑏 cos 𝑡)2 𝑑𝑡
𝐶 0 𝑎 𝑏
2𝜋
= ∫ (𝑎2 sin2 𝑡 + 𝑏 2 cos 2 𝑡)𝑑𝑡
0
= 𝜋(𝑎2 + 𝑏 2 ).

𝐶1
Lemma 6.35 Let 𝐶 be a curve obtained by joining finitely many 𝐶5

smooth curves 𝐶1 , 𝐶2 , … , 𝐶𝑛 end to end, and let 𝑓 be a function. Then

𝐶4
∫𝑓 𝑑𝑠 = ∫ 𝑓 𝑑𝑠 + ∫ 𝑓 𝑑𝑠 + ⋯ + ∫ 𝑓 𝑑𝑠.
𝐶 𝐶1 𝐶2 𝐶𝑛 𝐶2
𝐶3

From now on, we will focus only on piecewise-smooth curves, which are those as in Lemma 6.35.

Definition 6.36 Let 𝐅 be a continuous vector field in 𝑛 dimensions and let 𝐶 be a curve in ℝ𝑛
parametrized by a vector-valued function 𝐫: [𝑎, 𝑏] → ℝ𝑛 . Then the line integral of the vector field
𝐅 along 𝐶 is
𝑏
∫𝐅 ⋅ 𝑑𝐫 ≔ ∫ 𝐅(𝐫(𝑡)) ⋅ 𝐫 ′ (𝑡)𝑑𝑡. 𝑑𝐫 = 𝐫 ′ (𝑡)𝑑𝑡
𝐶 𝑎
3
If 𝐶 is a curve in ℝ and 𝐅 = 〈𝑓, 𝑔, ℎ〉 is a vector field, then the line integral of 𝐅 along 𝐶 can
also be denoted as

∫𝑓(𝑥, 𝑦, 𝑧)𝑑𝑥 + 𝑔(𝑥, 𝑦, 𝑧)𝑑𝑦 + ℎ(𝑥, 𝑦, 𝑧)𝑑𝑧.

𝐶

Remark 6.37 Let 𝐫(𝑡) be the position at time 𝑡 of a particle moving in ℝ3 , and let 𝐅(𝑥, 𝑦, 𝑧)
be the force acting on the particle when it is located at the point (𝑥, 𝑦, 𝑧). Then ∫𝐶 𝐅 ⋅ 𝑑𝐫
represents the work done by the force 𝐅 on the particle as it moves along a curve 𝐶.

Remark 6.38 Let 𝐶 be a curve inside a pipe and let 𝐅(𝑥, 𝑦, 𝑧) be the flow velocity of water at
the point (𝑥, 𝑦, 𝑧) in the pipe. Then ∫𝐶 𝐅 ⋅ 𝑑𝐫 measures the tendency of water to flow along 𝐶,
and is called the flow of 𝐅 along 𝐶.

Page 9 of 40
MATH2023 Multivariable Calculus Chapter 6 Vector calculus
L2/L3 (Fall 2019)

Remark 6.39 Let 𝐶 be a curve parametrized by 𝐫 and let 𝐓 ̂ be the unit tangent vector of 𝐶, i.e.
1
̂ (𝑡) ≔
𝐓 𝐫 ′ (𝑡).
‖𝐫 ′ (𝑡)‖
Then the line integral of a vector field 𝐅 along 𝐶 is just the same as the line integral of the
̂ over 𝐶, since
function 𝐅 ⋅ 𝐓
𝑏 𝑏
1
∫𝐅 ⋅ 𝑑𝐫 = ∫ 𝐅(𝐫(𝑡)) ⋅ 𝐫 ′ (𝑡)𝑑𝑡 = ∫ 𝐅(𝐫(𝑡)) ⋅ 𝐫 ′ (𝑡) ‖𝐫 ̂ 𝑑𝑠 ;
⏟ ′ (𝑡)‖𝑑𝑡 = ∫𝐅 ⋅ 𝐓
𝐶 𝑎 𝑎 ‖𝐫 ′ (𝑡)‖
⏟ 𝐶
𝑑𝑠
̂ (𝑡)
𝐓

and conversely, the line integral of a function 𝑓 over 𝐶 is just the same as the line integral of the
̂ along 𝐶. (Can you verify this on your own?)
vector field 𝑓𝐓

Remark 6.40 Given a curve 𝐶, if we denote by −𝐶 the curve having the same image as that of 𝐶
but is in the opposite orientation of 𝐶, then we have

∫ 𝐅 ⋅ 𝑑𝐫 = − ∫𝐅 ⋅ 𝑑𝐫
−𝐶 𝐶
because the direction of the unit tangent vector 𝐓 ̂ along the curve is reversed. Note however
that the line integral of a function 𝑓 over 𝐶 is independent of the orientation of 𝐶, i.e.

∫ 𝑓 𝑑𝑠 = ∫𝑓 𝑑𝑠.
−𝐶 𝐶

Example 6.41 Let 𝐅(𝑥, 𝑦, 𝑧) = 〈𝑦, 𝑧, 𝑥〉 be a vector field. Evaluate

∫𝐅 ⋅ 𝑑𝐫 𝑧
𝐶 𝑥 2 + 𝑦2 = 1
2 2
where 𝐶 is the curve of intersection of the cylinder 𝑥 + 𝑦 = 1 and the
plane 𝑧 = 2𝑦 + 1, oriented counterclockwise as viewed from above. 𝑦

Solution:
𝑥
The curve 𝐶 can be parametrized by the vector-valued function 𝐶
𝑧 = 2𝑦 + 1
𝐫(𝑡) = 〈cos 𝑡 , sin 𝑡 , 2 sin 𝑡 + 1〉
where 𝑡 ∈ [0, 2𝜋]. So
2𝜋
∫𝐅 ⋅ 𝑑𝐫 = ∫ 𝐅(𝐫(𝑡)) ⋅ 𝐫 ′ (𝑡)𝑑𝑡
𝐶 0
2𝜋
= ∫ 〈sin 𝑡 , 2 sin 𝑡 + 1, cos 𝑡〉 ⋅ 〈− sin 𝑡 , cos 𝑡 , 2 cos 𝑡〉𝑑𝑡
0
2𝜋
= ∫ (− sin2 𝑡 + 2 sin 𝑡 cos 𝑡 + cos 𝑡 + 2 cos2 𝑡)𝑑𝑡
0
= −𝜋 + 0 + 0 + 2𝜋
= 𝜋.

Page 10 of 40
MATH2023 Multivariable Calculus Chapter 6 Vector calculus
L2/L3 (Fall 2019)

Example 6.42 Let 𝐅(𝑥, 𝑦) = 〈𝑥 2 − 𝑦 2 , 2𝑥𝑦〉 be a vector field. Evaluate

∫𝐅 ⋅ 𝑑𝐫
𝐶
𝑦
2
(a) if 𝐶 is the upper unit semicircle in ℝ centered at the origin, going
clockwise from (−1, 0) to (1, 0).
(b) if 𝐶 is the curve in ℝ2 consisting of the line segment from (−1, 0)
to (0, 1) and then the line segment from (0, 1) to (1, 0). 𝑥

Solution:
(a) The curve 𝐶 can be parametrized by the vector-valued function 𝐫: [0, 𝜋] → ℝ2 defined by
𝐫(𝑡) = 〈− cos 𝑡 , sin 𝑡〉,
so
𝜋
∫𝐅 ⋅ 𝑑𝐫 = ∫ 𝐅(𝐫(𝑡)) ⋅ 𝐫 ′ (𝑡)𝑑𝑡
𝐶 0
𝜋
= ∫ 〈(− cos 𝑡)2 − (sin 𝑡)2 , 2(− cos 𝑡)(sin 𝑡)〉 ⋅ 〈sin 𝑡 , cos 𝑡〉𝑑𝑡
0
𝜋
= ∫ (− sin 𝑡)𝑑𝑡 = [cos 𝑡]𝜋0 = −2.
0

(b) The curve 𝐶 can be decomposed into two smooth curves 𝐶1 and 𝐶2 , so that 𝐶1 is the line
segment from (−1, 0) to (0, 1), which can be parametrized by
𝐫1 (𝑡) = 〈−1 + 𝑡, 𝑡〉
for 𝑡 ∈ [0, 1], and 𝐶2 is the line segment from (0, 1) to (1, 0), with a parametrization
𝐫2 (𝑡) = 〈𝑡, 1 − 𝑡〉
for 𝑡 ∈ [0, 1]. So

∫𝐅 ⋅ 𝑑𝐫 = ∫ 𝐅 ⋅ 𝑑𝐫1 + ∫ 𝐅 ⋅ 𝑑𝐫2
𝐶 𝐶1 𝐶2
1 1
′ (𝑡)𝑑𝑡
= ∫ 𝐅(𝐫1 (𝑡)) ⋅ 𝐫1 + ∫ 𝐅(𝐫2 (𝑡)) ⋅ 𝐫2 ′ (𝑡)𝑑𝑡
0 0
1
= ∫ 〈(−1 + 𝑡)2 − 𝑡 2 , 2(−1 + 𝑡)(𝑡)〉 ⋅ 〈1, 1〉𝑑𝑡
0
1
+ ∫ 〈𝑡 2 − (1 − 𝑡)2 , 2(𝑡)(1 − 𝑡)〉 ⋅ 〈1, −1〉𝑑𝑡
0
1 1
= ∫ [(−1 + 𝑡) − 𝑡 + 2(−1 + 𝑡)(𝑡)]𝑑𝑡 + ∫ [𝑡 2 − (1 − 𝑡)2 − 2(𝑡)(1 − 𝑡)]𝑑𝑡
2 2
0 0
1
2 2
= 2 [ 𝑡3 − 𝑡2] = − .
3 0 3

Page 11 of 40
MATH2023 Multivariable Calculus Chapter 6 Vector calculus
L2/L3 (Fall 2019)

Let 𝑃 and 𝑄 be two given points. Example 6.42 shows that the line integrals of a vector field
along different curves joining 𝑃 and 𝑄 are different in general. But in fact for conservative
vector fields, the line integral is always the same for every curve joining 𝑃 and 𝑄.

Theorem 6.43 Let 𝐷 be a region in ℝ𝑛 , let 𝑓: 𝐷 → ℝ be a function having continuous partial

derivatives, and let 𝐶 be a curve in 𝐷 parametrized by 𝐫: [𝑎, 𝑏] → 𝐷. Then

∫∇𝑓 ⋅ 𝑑𝐫 = 𝑓(𝐫(𝑏)) − 𝑓(𝐫(𝑎)).

𝐶
In this case, we say that the line integrals of ∇𝑓 are path-independent on 𝑫, i.e. they depend only
on the initial point and the terminal point of the curve, but do not depend on the shape of the curve.
So if 𝐶 ′ is another curve in 𝐷 also joining the points 𝐫(𝑎) and 𝐫(𝑏), then we have

∫∇𝑓 ⋅ 𝑑𝐫 = ∫ ∇𝑓 ⋅ 𝑑𝐫 .
𝐶 𝐶′
In particular, the line integrals of a conservative vector field in 𝐷 are path-independent on 𝐷.

Proof. To make calculations simple we just consider the case 𝑛 = 2, and the general case is similar.
Let 𝐫(𝑡) = 〈𝑢(𝑡), 𝑣(𝑡)〉. Then chain rule gives
𝑑 𝑑 𝜕𝑓 𝜕𝑓
𝑓(𝐫(𝑡)) = 𝑓(𝑢(𝑡), 𝑣(𝑡)) = (𝑢(𝑡), 𝑣(𝑡))𝑢′ (𝑡) + (𝑢(𝑡), 𝑣(𝑡))𝑣 ′ (𝑡) = ∇𝑓(𝐫(𝑡)) ⋅ 𝐫 ′ (𝑡),
𝑑𝑡 𝑑𝑡 𝜕𝑥 𝜕𝑦
and so
𝑏 𝑏
𝑑
∫∇𝑓 ⋅ 𝑑𝐫 = ∫ ∇𝑓(𝐫(𝑡)) ⋅ 𝐫 ′ (𝑡)𝑑𝑡 = ∫ 𝑓(𝐫(𝑡))𝑑𝑡 = 𝑓(𝐫(𝑏)) − 𝑓(𝐫(𝑎))
𝐶 𝑎 𝑎 𝑑𝑡
by the Fundamental Theorem of Calculus for functions of one real variable. ∎

Example 6.44 Let 𝐅: ℝ3 ∖ {(0, 0, 0)} → ℝ3 be the gravitational field acting on a particle, given by
1
𝐅(𝑥, 𝑦, 𝑧) = − 3
(𝑥𝐢 + 𝑦𝐣 + 𝑧𝐤) for every (𝑥, 𝑦, 𝑧) ≠ (0, 0, 0).
(𝑥 2 + 𝑦 2 + 𝑧 2 )2

Find the work done by 𝐅 in moving the particle along any curve 𝐶 from the point (4, 3, 12) to
the point (2, 2, 1).

Solution:
We have seen in Example 6.11 that the gravitational field 𝐅 is conservative on ℝ3 ∖ {(0, 0, 0)} and
1
𝜙(𝑥, 𝑦, 𝑧) =
√𝑥 2 + 𝑦 2 + 𝑧 2
is a potential function of 𝐅 on ℝ3 ∖ {(0, 0, 0)}, i.e. 𝐅 = ∇𝜙. Therefore the work is given by
1 1 10
∫𝐅 ⋅ 𝑑𝐫 = ∫∇𝜙 ⋅ 𝑑𝐫 = 𝜙(2, 2, 1) − 𝜙(4, 3, 12) = − = .
𝐶 𝐶 √22 + 22 + 12 √42 + 32 + 122 39

Page 12 of 40
MATH2023 Multivariable Calculus Chapter 6 Vector calculus
L2/L3 (Fall 2019)

Definition 6.45 A curve 𝐶 parametrized by 𝐫: [𝑎, 𝑏] → ℝ𝑛 is called a closed curve if its initial
point and terminal point are the same point, i.e.
𝐫(𝑎) = 𝐫(𝑏).

A closed curve Not a closed curve

Remark 6.38 If 𝐶 is a closed curve and 𝐅 is a vector field, then the line integral ∫𝐶 𝐅 ⋅ 𝑑𝐫 is
sometimes also denoted as

∮𝐅 ⋅ 𝑑𝐫 .
𝐶
The flow of a velocity field 𝐅 along a closed curve 𝐶 is also called the circulation of 𝐅 around 𝐶.

Corollary 6.46 Let 𝐷 be a region in ℝ𝑛 . If 𝐅 is a conservative vector field in 𝐷 and 𝐶 is a

closed curve in 𝐷, then

∮𝐅 ⋅ 𝑑𝐫 = 0.
𝐶

Example 6.47 Let 𝐅: ℝ3 → ℝ3 be the vector field defined by

𝐅(𝑥, 𝑦, 𝑧) = 〈𝑦 2 , 2𝑥𝑦 + 𝑒 3𝑧 , 3𝑦𝑒 3𝑧 〉.
Evaluate

∫𝐅 ⋅ 𝑑𝐫 ,
𝐶
3
where 𝐶 is the pentagon in ℝ as shown in the diagram below.
𝑧

𝑦
(4, 1, 3)
(−1, 0, 2)

(4, 1, 0)
0
𝑥
(2, −1, 0)

Solution:
[The line integral is complicated because the curve 𝐶 consists of several portions. But since 𝐶 is
a closed curve, the line integral ∮𝐶 𝐅 ⋅ 𝑑𝐫 is just zero if we can show that 𝐅 is conservative in ℝ3 .]

Page 13 of 40
MATH2023 Multivariable Calculus Chapter 6 Vector calculus
L2/L3 (Fall 2019)

Let’s check whether 𝐅 is conservative in ℝ3 , so we check whether there exists 𝑓: ℝ3 → ℝ such

that 𝐅 = ∇𝑓, i.e.
𝑓𝑥 (𝑥, 𝑦, 𝑧) = 𝑦 2
{𝑓𝑦 (𝑥, 𝑦, 𝑧) = 2𝑥𝑦 + 𝑒 3𝑧 .
𝑓𝑧 (𝑥, 𝑦, 𝑧) = 3𝑦𝑒 3𝑧
The first equation of the system requires that

𝑓(𝑥, 𝑦, 𝑧) = ∫ 𝑦 2 𝑑𝑥 = 𝑥𝑦 2 + 𝑔(𝑦, 𝑧)

for some function 𝑔 depending on 𝑦 and 𝑧 only. Taking partial derivatives with respect to 𝑦,
we have
𝑓𝑦 (𝑥, 𝑦, 𝑧) = 2𝑥𝑦 + 𝑔𝑦 (𝑦, 𝑧),
so comparing this with the second equation of the system we must have 𝑔𝑦 (𝑦, 𝑧) = 𝑒 3𝑧 , i.e.

𝑔(𝑦, 𝑧) = ∫ 𝑒 3𝑧 𝑑𝑦 = 𝑦𝑒 3𝑧 + ℎ(𝑧)

and so
𝑓(𝑥, 𝑦, 𝑧) = 𝑥𝑦 2 + 𝑦𝑒 3𝑧 + ℎ(𝑧)
for some function ℎ depending on 𝑧 only. Taking partial derivatives with respect to 𝑧 we have
𝑓𝑧 (𝑥, 𝑦, 𝑧) = 3𝑦𝑒 3𝑧 + ℎ′ (𝑧),
so comparing this with the third equation of the system we must have ℎ′ (𝑧) = 0, i.e.

ℎ(𝑧) = ∫ 0 𝑑𝑧 = 𝑘

for some constant 𝑘. Thus if we let

𝑓(𝑥, 𝑦, 𝑧) = 𝑥𝑦 2 + 𝑦𝑒 3𝑧 + 𝑘,
then 𝐅 = ∇𝑓. This shows that 𝐅 is conservative in ℝ3 . Finally 𝐶 is a closed curve in ℝ3 , so

∮𝐅 ⋅ 𝑑𝐫 = 0.
𝐶

Example 6.48 Determine whether the vector field

𝑦 𝑥
𝐅(𝑥, 𝑦) = 〈− 2 , 〉
𝑥 + 𝑦2 𝑥2 + 𝑦2
is conservative in its domain ℝ2 ∖ {(0, 0)}.

Solution:
[If 𝑔𝑥 − 𝑓𝑦 ≠ 0, then we can conclude right away that 𝐅 = 〈𝑓, 𝑔〉 is not conservative in its domain
according to Theorem 6.14. But for the given vector field 𝐅 we unfortunately have
𝜕 𝑥 𝜕 −𝑦
− = 0,
𝜕𝑥 𝑥 2 + 𝑦 2 𝜕𝑦 𝑥 2 + 𝑦 2
so no conclusion about 𝐅 can be drawn this way. Thus we try to use Corollary 6.46 instead.]

Page 14 of 40
MATH2023 Multivariable Calculus Chapter 6 Vector calculus
L2/L3 (Fall 2019)

Let 𝐶 be the unit circle in ℝ2 centered at the origin, which is parametrized by

𝐫(𝑡) = 〈cos 𝑡 , sin 𝑡〉
for 𝑡 ∈ [0, 2𝜋]. Then 𝐶 is a closed curve in ℝ2 ∖ {(0, 0)}. Now 𝐫 ′ (𝑡) = 〈− sin 𝑡 , cos 𝑡〉 for
𝑡 ∈ (0, 2𝜋), so
2𝜋
sin 𝑡 cos 𝑡
∮𝐅 ⋅ 𝑑𝐫 = ∫ 〈− , 〉 ⋅ 〈− sin 𝑡 , cos 𝑡〉𝑑𝑡
𝐶 0 cos2 𝑡 + sin 𝑡 cos 𝑡 + sin2 𝑡
2 2

2𝜋 2𝜋
sin2 𝑡 + cos2 𝑡
=∫ 𝑑𝑡 = ∫ 𝑑𝑡 = 2𝜋
0 cos2 𝑡 + sin2 𝑡 0
≠ 0,
and so the vector field 𝐅 is not conservative in ℝ2 ∖ {(0, 0)}.

The converse of Theorem 6.43 is also true.

Theorem 6.49 Let 𝐷 be a region in ℝ𝑛 and 𝐅: 𝐷 → ℝ𝑛 be a vector field whose component

functions are continuous. If the line integrals of 𝐅 are path-independent on 𝐷, then 𝐅 is
conservative in 𝐷.

Proof. To make calculations simple we just consider the case 𝑛 = 2, and the general case is similar.
Let’s write 𝐅 = 〈𝑓, 𝑔〉, let the point (𝑎, 𝑏) ∈ 𝐷 be fixed and define a function 𝜙: 𝐷 → ℝ by

𝜙(𝑥, 𝑦) = ∫ 𝐅 ⋅ 𝑑𝐫 .
a curve in 𝐷 from (𝑎,𝑏) to (𝑥,𝑦)

This function 𝜙 is well-defined because the line integral is path-independent on 𝐷.

Now it remains to verify that 𝐅 = ∇𝜙 on 𝐷. For every (𝑥, 𝑦) ∈ 𝐷, we have

𝜕𝜙 𝜙(𝑥 + ℎ, 𝑦) − 𝜙(𝑥, 𝑦) 1
(𝑥, 𝑦) = lim = lim ∫ 𝐅 ⋅ 𝑑𝐫
𝜕𝑥 ℎ→0 ℎ ℎ→0 ℎ a curve in 𝐷 from (𝑥,𝑦) to (𝑥+ℎ,𝑦)

1
= lim ∫ 𝐅 ⋅ 𝑑𝐫
ℎ→0 ℎ the line segment from (𝑥,𝑦) to (𝑥+ℎ,𝑦)

1 ℎ
= lim ∫ 〈𝑓(𝑥 + 𝑡, 𝑦), 𝑔(𝑥 + 𝑡, 𝑦)〉 ⋅ 〈1, 0〉 𝑑𝑡
ℎ→0 ℎ 0

1 ℎ
= lim ∫ 𝑓(𝑥 + 𝑡, 𝑦)𝑑𝑡
ℎ→0 ℎ 0
𝑓 is continuous
= 𝑓(𝑥, 𝑦),
and similarly
𝜕𝜙
(𝑥, 𝑦) = 𝑔(𝑥, 𝑦).
𝜕𝑦
This shows that 𝐅 = ∇𝜙 on 𝐷, so 𝐅 is conservative in 𝐷. ∎

Page 15 of 40
MATH2023 Multivariable Calculus Chapter 6 Vector calculus
L2/L3 (Fall 2019)

3. Green’s Theorem

Green’s Theorem provides another way of evaluating the line integral of a vector field along a
“simple” closed curve in ℝ2 , which is sometimes quicker than evaluating the line integral directly.

Definition 6.50 A closed curve 𝐶 parametrized by 𝐫: [𝑎, 𝑏] → ℝ𝑛 is called a simple closed curve
if it does not have any “self-intersection”, i.e. if
𝐫(𝑡1 ) ≠ 𝐫(𝑡2 )
for any distinct 𝑡1 , 𝑡2 ∈ [𝑎, 𝑏], except when 𝑡1 = 𝑎 and 𝑡2 = 𝑏.

A simple closed curve A closed curve which is not simple

Theorem 6.51 (Jordan curve theorem) The image of a simple closed curve in ℝ2 divides the
whole plane ℝ2 into two regions. One of these regions is bounded, and the other one is
unbounded.

Definition 6.52 Let 𝐶 be a simple closed curve in ℝ2 .

(i) The bounded region as in the above Jordan curve theorem is called the interior of 𝐶. The
other unbounded region is called the exterior of 𝐶.
(ii) 𝐶 is positively oriented if the interior of 𝐶 is always on the left when moving along 𝐶.
(iii) The outward unit normal vector to 𝐶 is the unit normal vector to 𝐶 which points toward the
exterior of 𝐶.

Exterior of 𝐶
̂(𝑡) to 𝐶
Outward unit normal vectors 𝐧

Interior of 𝐶

A positively oriented simple closed curve 𝐶

Remark 6.53 If 𝐶 is a positively oriented smooth simple closed curve in ℝ2 ,

parametrized by 𝐫: [𝑎, 𝑏] → ℝ2 with 𝐫(𝑡) = 〈𝑢(𝑡), 𝑣(𝑡)〉 so that
𝐶
𝐫 ′ (𝑡) = 〈𝑢′ (𝑡), 𝑣 ′ (𝑡)〉,
𝐫 ′ (𝑡)
then the outward unit normal vector to 𝐶 is given by
̂(𝑡)
𝐧
1
̂(𝑡) = ′
𝐧 〈𝑣 ′ (𝑡), −𝑢′ (𝑡)〉
‖𝐫 (𝑡)‖
for every 𝑡 ∈ (𝑎, 𝑏).

Page 16 of 40
MATH2023 Multivariable Calculus Chapter 6 Vector calculus
L2/L3 (Fall 2019)

Theorem 6.54 (Green) Let 𝐅 = 〈𝑓, 𝑔〉 be a vector field in two dimensions whose components 𝑓
and 𝑔 have continuous partial derivatives, and let 𝐶 be a positively oriented simple closed curve
in ℝ2 , whose interior is the region 𝑅. Then
𝜕𝑔 𝜕𝑓
∮𝐅 ⋅ 𝑑𝐫 = ∬ ( − ) 𝑑𝐴.
𝐶 𝑅 𝜕𝑥 𝜕𝑦

𝑅
𝐶

Proof. Let’s take for granted the fact that the region 𝑅 can always be decomposed into
subregions which are both “𝑦-simple” and “𝑥-simple”, so without loss of generality we may suppose
that 𝑅 itself is both “𝑦-simple” and “𝑥-simple”, i.e.
𝑅 = {(𝑥, 𝑦) ∈ ℝ2 : 𝑎 ≤ 𝑥 ≤ 𝑏 and 𝑝1 (𝑥) ≤ 𝑦 ≤ 𝑞1 (𝑥)}
= {(𝑥, 𝑦) ∈ ℝ2 : 𝑐 ≤ 𝑦 ≤ 𝑑 and 𝑝2 (𝑦) ≤ 𝑥 ≤ 𝑞2 (𝑦)}
for some real numbers 𝑎, 𝑏, 𝑐, 𝑑 and single-variable continuous functions 𝑝1 , 𝑞1 : [𝑎, 𝑏] → ℝ and
𝑝2 , 𝑞2 : [𝑐, 𝑑] → ℝ. Then we have
𝑏
∮𝑓 𝑑𝑥 = ∫ 𝑓(𝑥, 𝑝1 (𝑥))𝑑𝑥 + ∫vertical line segment joining 𝑓 𝑑𝑥 𝑦
𝐶 𝑎
⏟(𝑏,𝑝1(𝑏)) and (𝑏,𝑞1(𝑏))
𝑦 = 𝑞1 (𝑥)
=0
𝑎
+ ∫ 𝑓(𝑥, 𝑞1 (𝑥))𝑑𝑥 + ∫vertical line segment joining 𝑓 𝑑𝑥 𝑅
𝑏
⏟(𝑎,𝑞1(𝑎)) and (𝑎,𝑝1 (𝑎))
=0 𝑦 = 𝑝1 (𝑥)
𝑏
𝑥
= − ∫ [𝑓(𝑥, 𝑞1 (𝑥)) − 𝑓(𝑥, 𝑝1 (𝑥))]𝑑𝑥 𝑎 𝑏
𝑎
𝑏 𝑞1 (𝑥)
𝜕𝑓 𝜕𝑓
= −∫ ∫ (𝑥, 𝑦)𝑑𝑦 𝑑𝑥 = − ∬ 𝑑𝐴 , 𝑦
𝑎 𝑝1 (𝑥) 𝜕𝑦 𝑅 𝜕𝑦
𝑑
and similarly
𝑥 = 𝑞2 (𝑦)
𝑑
∮𝑔 𝑑𝑦 = ∫ [𝑔(𝑞2 (𝑦), 𝑦) − 𝑔(𝑝2 (𝑦), 𝑦)]𝑑𝑥 𝑥 = 𝑝2 (𝑦) 𝑅
𝐶 𝑐
𝑑 𝑞2 (𝑦)
𝜕𝑔 𝜕𝑔 𝑐
=∫ ∫ (𝑥, 𝑦)𝑑𝑥 𝑑𝑦 = ∬ 𝑑𝐴.
𝑐 𝑝2 (𝑦) 𝜕𝑥 𝑅 𝜕𝑥 𝑥

Therefore
𝜕𝑔 𝜕𝑓
∮𝐅 ⋅ 𝑑𝐫 = ∮𝑓 𝑑𝑥 + 𝑔 𝑑𝑦 = ∬ ( − ) 𝑑𝐴
𝐶 𝐶 𝑅 𝜕𝑥 𝜕𝑦
as desired. ∎

Page 17 of 40
MATH2023 Multivariable Calculus Chapter 6 Vector calculus
L2/L3 (Fall 2019)

Example 6.55 Let 𝐶 be the unit circle in ℝ2 centered at the origin, oriented counterclockwise.
Evaluate the line integral

∮[𝑥 2 𝑒 𝑥 + 𝑦 − ln(1 + 𝑥 2 )]𝑑𝑥 + (8𝑥 + sin 𝑦)𝑑𝑦.

𝐶

Solution:
Let 𝑅 be the unit disk in ℝ2 centered at the origin. Then 𝐶 is a positively oriented simple
closed curve in ℝ2 whose interior is 𝑅. Now if we let 𝐅(𝑥, 𝑦) = 〈𝑓(𝑥, 𝑦), 𝑔(𝑥, 𝑦)〉, where
𝑓(𝑥, 𝑦) = 𝑥 2 𝑒 𝑥 + 𝑦 − ln(1 + 𝑥 2 ) and 𝑔(𝑥, 𝑦) = 8𝑥 + sin 𝑦 ,
then by Green’s Theorem we have

∮[𝑥 2 𝑒 𝑥 + 𝑦 − ln(1 + 𝑥 2 )]𝑑𝑥 + (8𝑥 + sin 𝑦)𝑑𝑦

𝐶
𝜕𝑔 𝜕𝑓
= ∮𝐅 ⋅ 𝑑𝐫 = ∬ ( − ) 𝑑𝐴 = ∬ (8 − 1)𝑑𝐴
𝐶 𝑅 𝜕𝑥 𝜕𝑦 𝑅

= 7 ∬ 𝑑𝐴 = 7(Area of 𝑅) = 7𝜋.
𝑅

Example 6.56 Let 𝐅: ℝ2 → ℝ2 be the vector field defined by

𝐅(𝑥, 𝑦) = 〈𝑦 2 + 1, 3𝑥𝑦〉
and let 𝐶 be the curve in ℝ2 consisting of
(i) the upper semicircle centered at the origin joining (2, 0) to (−2, 0), followed by
(ii) the line segment joining (−2, 0) to (−1, 0), followed by
(iii) the upper semicircle centered at the origin joining (−1, 0) to (1, 0).
Evaluate the line integral ∫𝐶 𝐅 ⋅ 𝑑𝐫.

Solution: 𝑦
The given curve 𝐶 is not a closed curve, but if we let 𝐶 ′ be the
𝐶
line segment joining (1, 0) and (2, 0), then the curve consisting of 𝑅
𝐶 followed by 𝐶 ′ becomes a positively oriented simple closed
𝑥
curve in ℝ2 , whose interior is the region 𝐶′
𝑅 = {(𝑟, 𝜃): 1 ≤ 𝑟 ≤ 2 and 0 ≤ 𝜃 ≤ 𝜋}
Technique: “Complete the closed curve”
defined in polar coordinates. So by Green’s Theorem, we have
𝜕 𝜕
∫𝐅 ⋅ 𝑑𝐫 + ∫ 𝐅 ⋅ 𝑑𝐫 = ∬ [ (3𝑥𝑦) − (𝑦 2 + 1)] 𝑑𝐴
𝐶 𝐶 ′ 𝑅 𝜕𝑥 𝜕𝑦
𝜋 2
= ∬ 𝑦 𝑑𝐴 = ∫ ∫ (𝑟 sin 𝜃)𝑟 𝑑𝑟 𝑑𝜃
𝑅 0 1
𝜋 2
7 14
= (∫ sin 𝜃 𝑑𝜃) (∫ 𝑟 2 𝑑𝑟) = (2) ( ) = .
0 1 3 3

Page 18 of 40
MATH2023 Multivariable Calculus Chapter 6 Vector calculus
L2/L3 (Fall 2019)

On the other hand, the line segment 𝐶 ′ can be parametrized by 𝐫(𝑡) = 〈𝑡, 0〉 for 𝑡 ∈ [1, 2], so
the line integral of 𝐅 along 𝐶 ′ is simply
2 2
∫ 𝐅 ⋅ 𝑑𝐫 = ∫ 〈02 + 1, 3(𝑡)(0)〉 ⋅ 〈1, 0〉 𝑑𝑡 = ∫ 𝑑𝑡 = 1.
𝐶′ 1 1
Therefore
14 11
∫𝐅 ⋅ 𝑑𝐫 = −1= .
𝐶 3 3

Corollary 6.57 Let 𝐶 be a positively oriented simple closed curve in ℝ2 . Then the area of the
interior of 𝐶 is given by
1
∮−𝑦 𝑑𝑥 + 𝑥 𝑑𝑦 ,
2 𝐶
i.e. the area equals half of the line integral of the vector field 𝐅(𝑥, 𝑦) = 〈−𝑦, 𝑥〉 along 𝐶.

Proof. Let 𝑅 be the interior of 𝐶. Then by Green’s Theorem we have

𝜕 𝜕
∮𝐅 ⋅ 𝑑𝐫 = ∬ [ 𝑥 − (−𝑦)] 𝑑𝐴 = 2 ∬ 𝑑𝐴 ,
𝐶 𝑅 𝜕𝑥 𝜕𝑦 𝑅
so the area of 𝑅 is given by
1 1
∬ 𝑑𝐴 = ∮𝐅 ⋅ 𝑑𝐫 = ∮−𝑦 𝑑𝑥 + 𝑥 𝑑𝑦 .
𝑅 2 𝐶 2 𝐶
∎

Example 6.58 Let 𝑎 > 0 and 𝑏 > 0, and let 𝐶 be the ellipse in ℝ2 with equation
𝑥2 𝑦2
+ = 1.
𝑎2 𝑏 2
Find the area of the region bounded by 𝐶.

Solution:
The ellipse 𝐶 can be parametrized by the function 𝐫: [0, 2𝜋] → ℝ2 defined by
𝐫(𝑡) = 〈𝑎 cos 𝑡 , 𝑏 sin 𝑡〉.
Let 𝑅 be the interior of the ellipse 𝐶. Then the area of 𝑅 is
1
∬ 𝑑𝐴 = ∮−𝑦 𝑑𝑥 + 𝑥 𝑑𝑦
𝑅 2 𝐶
1 2𝜋
= ∫ [(−𝑏 sin 𝑡)(−𝑎 sin 𝑡) + (𝑎 cos 𝑡)(𝑏 cos 𝑡)]𝑑𝑡
2 0
1 2𝜋
= ∫ 𝑎𝑏 𝑑𝑡
2 0
= 𝜋𝑎𝑏.

Page 19 of 40
MATH2023 Multivariable Calculus Chapter 6 Vector calculus
L2/L3 (Fall 2019)

Remark 6.59 Green’s Theorem can still be applied to a region in ℝ2 whose boundary consists of
more than one connected components. The region 𝑅 in the following diagram has a boundary
that consists of three simple closed curves 𝐶, 𝐶1 and 𝐶2 which do not intersect each other.
Suppose that these boundary curves are oriented positively with respect to 𝑹, i.e. the outermost
boundary curve 𝐶 is oriented counterclockwise while the inner curves are all oriented clockwise.

If we introduce “cross cuts” joining 𝐶 with each of 𝐶1 and 𝐶2 , then we

obtain one single boundary curve of 𝑅. Since the two line integrals 𝑅
𝐶 𝐶2
along each of the cross cuts cancel, we are left with the conclusion that
𝜕𝑔 𝜕𝑓 𝐶1
∮𝐅 ⋅ 𝑑𝐫 + ∮ 𝐅 ⋅ 𝑑𝐫 + ∮ 𝐅 ⋅ 𝑑𝐫 = ∬ ( − ) 𝑑𝐴.
𝐶 𝐶1 𝐶2 𝑅 𝜕𝑥 𝜕𝑦
This is sometimes called the “general version” of Green’s Theorem.

𝑦 𝑥
Example 6.60 Let 𝐅(𝑥, 𝑦) = 〈− 𝑥 2 +𝑦 2 , 𝑥 2 +𝑦 2 〉. Show that

∮𝐅 ⋅ 𝑑𝐫 = 2𝜋
𝐶
for every positively oriented simple close curve 𝐶 in ℝ2 whose interior contains the origin.

Technique: “Drill a hole around the bad point”

Solution:
Given any positively oriented simple close curve 𝐶 in ℝ2 whose interior contains the origin, we
let 𝐶 ′ be a counterclockwise oriented circle in ℝ2 centered at the origin, with radius 𝑎 > 0
which is small enough so that 𝐶 ′ lies completely in the interior of 𝐶. Also let 𝑅 be the region
bounded between 𝐶 and 𝐶 ′ . Then the general version of Green’s Theorem gives
𝜕 𝑥 𝜕 −𝑦 𝑦
∮𝐅 ⋅ 𝑑𝐫 + ∮ 𝐅 ⋅ 𝑑𝐫 = ∬ ( 2 2
− 2 + 𝑦2
) 𝑑𝐴
𝐶 −𝐶 ′ 𝑅 𝜕𝑥 𝑥 + 𝑦 𝜕𝑦 𝑥 𝐶
𝑦2 − 𝑥2 𝑦2 − 𝑥2
=∬[ 2 2 2
− ] 𝑑𝐴 = 0,
𝑅 (𝑥 + 𝑦 ) (𝑥 2 + 𝑦 2 )2 𝑅 𝐶′
𝑥
so 𝑎

∮𝐅 ⋅ 𝑑𝐫 = − ∮ 𝐅 ⋅ 𝑑𝐫 = ∮ 𝐅 ⋅ 𝑑𝐫.
𝐶 −𝐶 ′ 𝐶′
′
But when we parametrize the circle 𝐶 as usual by 𝐫(𝑡) = 〈𝑎 cos 𝑡 , 𝑎 sin 𝑡〉 for 𝑡 ∈ [0, 2𝜋], the
last line integral along the circle 𝐶 ′ is simply
2𝜋 2𝜋 (−𝑎
′ (𝑡)𝑑𝑡
sin 𝑡)(−𝑎 sin 𝑡) + (𝑎 cos 𝑡)(𝑎 cos 𝑡)
∮ 𝐅 ⋅ 𝑑𝐫 = ∫ 𝐅(𝐫(𝑡)) ⋅ 𝐫 =∫ 𝑑𝑡 = 2𝜋.
𝐶′ 0 0 (𝑎 cos 𝑡)2 + (𝑎 sin 𝑡)2
Therefore the line integral along the curve 𝐶 is also

∮𝐅 ⋅ 𝑑𝐫 = 2𝜋.
𝐶

Page 20 of 40
MATH2023 Multivariable Calculus Chapter 6 Vector calculus
L2/L3 (Fall 2019)

Definition 6.61 We say that a (connected) region 𝑅 in the plane ℝ2 is simply connected if for
every simple closed curve 𝐶 in 𝑅, the interior of 𝐶 also lies completely in 𝑅.

A simply connected region Not a simply connected region

Remark 6.62 Intuitively, we may regard a simply connected region in ℝ2 as a connected region
which has no “holes”.

We have seen that a vector field 𝐅: 𝐷 → ℝ𝑛 is conservative if and only if the line integrals of 𝐅 are
path-independent on 𝐷. On a simply connected region in ℝ2 , we have another characterization
for conservative vector fields.

Theorem 6.63 Let 𝐷 be a simply connected region in ℝ2 and let 𝐅 = 〈𝑓, 𝑔〉: 𝐷 → ℝ2 be a
vector field, whose components 𝑓 and 𝑔 have continuous partial derivatives. If
𝑔𝑥 − 𝑓𝑦 = 0
on 𝐷, then 𝐅 is conservative in 𝐷.

Proof. For each simple closed curve 𝐶 in 𝐷, its interior 𝑅 also lies in 𝐷, so we have
𝜕𝑔 𝜕𝑓
∮𝐅 ⋅ 𝑑𝐫 = ∬ ( − ) 𝑑𝐴 = ∬ 0 𝑑𝐴 = 0
𝐶 𝑅 𝜕𝑥 𝜕𝑦 𝑅
by Green’s theorem, so the line integrals of 𝐅 are path-independent on 𝐷. Therefore 𝐅 is
conservative in 𝐷 by Theorem 6.49. ∎

Remark 6.64 A vector field 𝐅 = 〈𝑓, 𝑔〉 defined on a non-simply connected region may not be
conservative even if 𝑔𝑥 − 𝑓𝑦 = 0. To see this, consider the vector field 𝐅 = 〈𝑓, 𝑔〉 with
𝑦 𝑥
𝑓(𝑥, 𝑦) = − 2 2
and 𝑔(𝑥, 𝑦) = 2 .
𝑥 +𝑦 𝑥 + 𝑦2
 Although we have
𝑦2 − 𝑥2
𝑔𝑥 (𝑥, 𝑦) = = 𝑓𝑦 (𝑥, 𝑦)
(𝑥 2 + 𝑦 2 )2
for every (𝑥, 𝑦) ∈ ℝ2 ∖ {(0, 0)}, 𝐅 is still not conservative in ℝ2 ∖ {(0, 0)} as we have seen
in Example 6.48. Note that ℝ2 ∖ {(0, 0)} is not a simply connected region.
 But 𝐅 is conservative in the open right half-plane 𝑅 = {(𝑥, 𝑦) ∈ ℝ2 : 𝑥 > 0} with a potential
𝑦
function 𝜙(𝑥, 𝑦) = arctan 𝑥 . Note that 𝑅 is simply connected.

Page 21 of 40
MATH2023 Multivariable Calculus Chapter 6 Vector calculus
L2/L3 (Fall 2019)

Corollary 6.65 (2-D Divergence Theorem) Let 𝐅 = 〈𝑓, 𝑔〉 be a vector field in two dimensions
whose components 𝑓 and 𝑔 have continuous partial derivatives, and let 𝐶 be a positively
oriented simple closed curve, whose interior is the region 𝑅, and whose outward unit normal vector
̂. Then
is 𝐧
𝜕𝑓 𝜕𝑔
∮𝐅 ⋅ 𝐧̂ 𝑑𝑠 = ∬ ( + ) 𝑑𝐴.
𝐶 𝑅 𝜕𝑥 𝜕𝑦

𝑅 ̂(𝑡)
𝐧

Proof. Let 𝐫: [𝑎, 𝑏] → ℝ2 with 𝐫(𝑡) = 〈𝑢(𝑡), 𝑣(𝑡)〉 be a parametrization of 𝐶 . Since 𝐶 is

positively oriented, its outward unit normal vector is
1
̂(𝑡) =
𝐧 〈𝑣 ′ (𝑡), −𝑢′ (𝑡)〉.
‖𝐫 ′ (𝑡)‖
Thus
𝑏
1
̂ 𝑑𝑠 = ∫ 〈𝑓(𝐫(𝑡)), 𝑔(𝐫(𝑡))〉 ⋅
∮𝐅 ⋅ 𝐧 〈𝑣 ′ (𝑡), −𝑢′ (𝑡)〉 ⏟
‖𝐫 ′ (𝑡)‖𝑑𝑡
𝐶 𝑎 ‖𝐫 ′ (𝑡)‖
⏟ 𝑑𝑠
̂
𝐧
𝑏
= ∫ [𝑓(𝐫(𝑡))𝑣 ′ (𝑡) − 𝑔(𝐫(𝑡))𝑢′ (𝑡)]𝑑𝑡
𝑎
𝑏
= ∫ 〈−𝑔(𝐫(𝑡)), 𝑓(𝐫(𝑡))〉 ⋅ 〈𝑢′ (𝑡), 𝑣 ′ (𝑡)〉𝑑𝑡 = ∮〈−𝑔, 𝑓〉 ⋅ 𝑑𝐫 .
𝑎 𝐶
Applying Green’s Theorem to the vector field 〈−𝑔, 𝑓〉, we obtain
𝜕𝑓 𝜕(−𝑔) 𝜕𝑓 𝜕𝑔
∮〈−𝑔, 𝑓〉 ⋅ 𝑑𝐫 = ∬ [ − ] 𝑑𝐴 = ∬ ( + ) 𝑑𝐴 ,
𝐶 𝑅 𝜕𝑥 𝜕𝑦 𝑅 𝜕𝑥 𝜕𝑦
so the result follows. ∎

Remark 6.66 Let’s regard the vector field 𝐅 = 〈𝑓, 𝑔〉 as a vector field in three dimensions having
zero 𝐤-component, i.e. 𝐅 = 〈𝑓, 𝑔, 0〉. Then Green’s theorem (Theorem 6.54) can be written as

̂ 𝑑𝑠 = ∬ (∇ × 𝐅) ⋅ 𝐤 𝑑𝐴 ,
∮𝐅 ⋅ 𝐓
𝐶 𝑅
which says that the integral of the curl of 𝐅 across a region equals the sum of all components of 𝐅
tangential to the boundary. The 2-D Divergence Theorem (Corollary 6.65) becomes

̂ 𝑑𝑠 = ∬ ∇ ⋅ 𝐅 𝑑𝐴 ,
∮𝐅 ⋅ 𝐧
𝐶 𝑅
which says that the integral of the divergence of 𝐅 over a region equals the sum of all components
of 𝐅 normal to the boundary.

Page 22 of 40
MATH2023 Multivariable Calculus Chapter 6 Vector calculus
L2/L3 (Fall 2019)

4. Surface integrals

Definition 6.67 Let 𝑓 be a continuous function of 𝑛 real variables and let 𝑆 be a smooth
surface in ℝ𝑛 parametrized by a vector-valued function 𝐫: 𝑅 → ℝ𝑛 . Then the surface integral of
the function 𝑓 over 𝑆 is
Compare with Definition 6.28.
∬𝑓 𝑑𝑆 ≔ ∬ 𝑓(𝐫(𝑢, 𝑣))‖𝐫𝑢 × 𝐫𝑣 ‖𝑑𝑢 𝑑𝑣 .
𝑆 𝑅 𝑑𝑆 = ‖𝐫𝑢 × 𝐫𝑣 ‖𝑑𝑢 𝑑𝑣

Remark 6.68 The area of a smooth surface 𝑆 is the integral of 1 over 𝑆, i.e.

∬𝑑𝑆 = ∬ ‖𝐫𝑢 × 𝐫𝑣 ‖𝑑𝑢 𝑑𝑣 ,

𝑆 𝑅
as seen in Definition 5.40.

Remark 6.69 If a piece of material of the shape of a thin sheet occupies the smooth surface 𝑆 in
ℝ3 and 𝑓(𝑥, 𝑦, 𝑧) is the surface density of the material at the point (𝑥, 𝑦, 𝑧), then ∬𝑆 𝑓(𝑥, 𝑦, 𝑧)𝑑𝑆
represents the total mass of that piece of material.

Example 6.70 Let 𝑆 be the portion of the circular cylinder 𝑥 2 + 𝑦 2 = 1 which lies between the
𝑥𝑦-plane and the plane 𝑧 = 1 + 𝑥. Evaluate ∬𝑆(𝑥 + 𝑦 + 𝑧)𝑑𝑆.

Solution:
The whole circular cylinder has parametric equations
𝑥 = cos 𝑢
{𝑦 = sin 𝑢 where 0 ≤ 𝑢 ≤ 2𝜋 and − ∞ < 𝑣 < +∞.
𝑧=𝑣
The specified portion 𝑆 also needs to satisfy 0 ≤ 𝑧 ≤ 1 + 𝑥, i.e. 0 ≤ 𝑣 ≤ 1 + cos 𝑢. Therefore
𝑆 can be parametrized by the vector-valued function 𝐫: 𝑅 → ℝ3 defined by
𝐫(𝑢, 𝑣) = 〈cos 𝑢 , sin 𝑢 , 𝑣〉,
2
where 𝑅 = {(𝑢, 𝑣) ∈ ℝ : 0 ≤ 𝑢 ≤ 2𝜋 and 0 ≤ 𝑣 ≤ 1 + cos 𝑢}. Now we have
𝐢 𝐣 𝐤
(𝐫𝑢 × 𝐫𝑣 )(𝑢, 𝑣) = |− sin 𝑢 cos 𝑢 0 | = 〈cos 𝑢 , sin 𝑢 , 0〉
0 0 1
for every (𝑢, 𝑣) in the interior of 𝑅, so

∬(𝑥 + 𝑦 + 𝑧)𝑑𝑆 = ∬ (cos 𝑢 + sin 𝑢 + 𝑣)‖𝐫𝑢 × 𝐫𝑣 ‖𝑑𝑢 𝑑𝑣

𝑆 𝑅

= ∬ (cos 𝑢 + sin 𝑢 + 𝑣)√(cos 𝑢)2 + (sin 𝑢)2 + 02 𝑑𝑢 𝑑𝑣

𝑅
2𝜋 1+cos 𝑢
5𝜋
=∫ ∫ (cos 𝑢 + sin 𝑢 + 𝑣)𝑑𝑣 𝑑𝑢 = .
0 0 2

Page 23 of 40
MATH2023 Multivariable Calculus Chapter 6 Vector calculus
L2/L3 (Fall 2019)

Remark 6.71 If the smooth surface 𝑆 in ℝ3 is the graph of a function 𝑓: 𝑅 → ℝ of two real
variables, i.e. if 𝑆 is defined by the equation 𝑧 = 𝑓(𝑥, 𝑦), then 𝑆 can be parametrized by
𝐫(𝑢, 𝑣) = 〈𝑢, 𝑣, 𝑓(𝑢, 𝑣)〉,
Compare this with Corollary 5.42.
so that 𝐫𝑢 × 𝐫𝑣 = 〈−𝑓𝑢 , −𝑓𝑣 , 1〉 and we simply have
2
∬𝑔(𝑥, 𝑦, 𝑧)𝑑𝑆 = ∬ 𝑔(𝑥, 𝑦, 𝑓(𝑥, 𝑦))√1 + [𝑓𝑥 (𝑥, 𝑦)]2 + [𝑓𝑦 (𝑥, 𝑦)] 𝑑𝐴.
𝑆 𝑅

Example 6.72 Let 𝑆 be the unit upper hemisphere in ℝ3 centered at the origin. Evaluate

∬(𝑥 2 + 𝑦 2 )𝑑𝑆.
𝑆

Solution:
Let 𝑅 be the unit disk in ℝ2 centered at the origin. Then the given surface 𝑆 is the graph of the
function 𝑓: 𝑅 → ℝ defined by
𝑓(𝑥, 𝑦) = √1 − 𝑥 2 − 𝑦 2 .
Now in the interior of 𝑅 we have
𝑥 𝑦
𝑓𝑥 (𝑥, 𝑦) = − and 𝑓𝑦 (𝑥, 𝑦) = − ,
√1 − 𝑥 2 − 𝑦 2 √1 − 𝑥 2 − 𝑦 2
so
2 2
2 2 )𝑑𝑆
𝑥 𝑦
∬(𝑥 + 𝑦 =∬ (𝑥 2 +𝑦 2 )√1
+ (− ) + (− ) 𝑑𝐴
𝑆 𝑅 √1 − 𝑥 2 − 𝑦 2 √1 − 𝑥 2 − 𝑦 2
2𝜋 1
𝑥2 + 𝑦2 𝑟2
=∬ 𝑑𝐴 = ∫ 𝑟𝑑𝑟 𝑑𝜃∫
𝑅 √1 − 𝑥2 − 𝑦2 √1 − 𝑟 2 0 0
𝑡 = −√1 − 𝑟 2 2𝜋 0
2 4𝜋
= (∫ 𝑑𝜃) [∫ (1 − 𝑡 2 )𝑑𝑡] = (2𝜋) ( ) = .
0 −1 3 3

Definition 6.73 Let 𝑆 be a surface in ℝ3 . We say that 𝑆 is a closed surface if it divides the
whole space ℝ3 into two regions, one of which is bounded and the other is unbounded. In this
case, the bounded region is called the interior of 𝑆, and the other unbounded region is called the
exterior of 𝑆.

Exterior of 𝑆
Outward unit normal vectors 𝐧 to 𝑆

Interior of 𝑆

A closed surface 𝑆

Page 24 of 40
MATH2023 Multivariable Calculus Chapter 6 Vector calculus
L2/L3 (Fall 2019)

Definition 6.74 Let 𝑆 be a surface in ℝ3 . We say that 𝑆 is an orientable surface if it is

possible to choose unit normal vectors at each point of 𝑆 in a continuous manner. Given such a
choice 𝐧̂, 𝑆 is called an oriented surface. The following are some conventions of choosing the
unit normal vectors 𝐧̂ to an orientable surface.
(i) If 𝑆 is an orientable parametrized surface, then regarding a parametrization 𝐫, the choice
1
̂=
𝐧 (𝐫 × 𝐫𝑣 )
‖𝐫𝑢 × 𝐫𝑣 ‖ 𝑢
is called the orientation of 𝑆 induced by the parametrization 𝐫.
(ii) If 𝑆 is the graph of a function 𝑓 of two real variables, then the choice
1
̂=
𝐧 〈−𝑓𝑥 , −𝑓𝑦 , 1〉
2 2
√1 + 𝑓𝑥 + 𝑓𝑦

is called the upward orientation of 𝑆.

̂ to be the outward unit normal vector, i.e. the unit
(iii) If 𝑆 is a closed surface, then choosing 𝐧
normal vector to 𝐶 which points toward the exterior of 𝐶, is said to give the positive
orientation of 𝑆.

The Möbius strip is a surface in ℝ3 which is not orientable. The Klein bottle is a surface in ℝ4
which is not orientable.

Möbius strip
Klein bottle

In Problem Set 7 we will prove the non-orientability of the Möbius strip using its parametrization.

Definition 6.75 Let 𝐅 be a continuous 3-D vector field and 𝑆 be an oriented surface in ℝ3 with
̂. Then the surface integral (or flux) of the vector field 𝐅 across 𝑆 is
unit normal vector 𝐧

̂ 𝑑𝑆.
∬𝐅 ⋅ 𝑑𝐒 ≔ ∬𝐅 ⋅ 𝐧
𝑆 𝑆

Remark 6.76 Let 𝑆 be an oriented surface in the sea and let 𝐅(𝑥, 𝑦, 𝑧) be flow velocity of water
at the point (𝑥, 𝑦, 𝑧) in the sea. Then ∬𝑆 𝐅 ⋅ 𝑑𝐒 represents the volume of water flowing across
𝑆 per unit time. If 𝑆 is a closed surface, then the flux ∬𝑆 𝐅 ⋅ 𝑑𝐒 is sometimes also denoted as
∯𝑆 𝐅 ⋅ 𝑑𝐒, and in this case − ∯𝑆 𝐅 ⋅ 𝑑𝐒 represents the volume of water accumulated in the interior
of 𝑆 per unit time.

Page 25 of 40
MATH2023 Multivariable Calculus Chapter 6 Vector calculus
L2/L3 (Fall 2019)

Remark 6.77 Let 𝑆 be an oriented surface with unit normal vector 𝐧 ̂. Then the surface integral
̂ over 𝑆, i.e.
of a vector field 𝐅 across 𝑆 is by definition the surface integral of the function 𝐅 ⋅ 𝐧

̂ 𝑑𝑆 ;
∬𝐅 ⋅ 𝑑𝐒 ≔ ∬𝐅 ⋅ 𝐧
𝑆 𝑆
and conversely, the surface integral of a function 𝑓 over 𝑆 is just the same as the surface integral
̂ across 𝑆. (Can you verify this on your own?)
of the vector field 𝑓𝐧

Remark 6.78 Given an oriented surface 𝑆, if we denote by −𝑆 the same surface but in the
opposite orientation of 𝑆, then we have

∬ 𝐅 ⋅ 𝑑𝐒 = − ∬𝐅 ⋅ 𝑑𝐒
−𝑆 𝑆
because the direction of the unit normal vector 𝐧 ̂ to the surface is reversed. Note however that
the surface integral of a function 𝑓 over 𝑆 is independent of the orientation of 𝑆 (if any), i.e.

∬ 𝑓 𝑑𝑆 = ∬𝑓 𝑑𝑆.
−𝑆 𝑆

Remark 6.79 The surface integral

∬𝐅 ⋅ 𝑑𝐒
𝑆
can be computed explicitly under different situations.
(i) If 𝑆 is an oriented parametrized surface whose orientation is induced
Compare with Definition
by a parametrization 𝐫: 𝑅 → ℝ𝑛 , then 6.36 and Remark 6.37.

1
∬𝐅 ⋅ 𝑑𝐒 = ∬ 𝐅 ⋅ (𝐫 × 𝐫𝑣 ) ⏟
‖𝐫𝑢 × 𝐫𝑣 ‖𝑑𝑢 𝑑𝑣 = ∬ 𝐅 ⋅ (𝐫𝑢 × 𝐫𝑣 )𝑑𝑢 𝑑𝑣.
𝑆 𝑆 ‖𝐫𝑢 × 𝐫𝑣 ‖ 𝑢
⏟ 𝑅 𝑑𝑆
̂
𝐧
𝑑𝐒 = (𝐫𝑢 × 𝐫𝑣 )𝑑𝑢 𝑑𝑣
(ii) If 𝑆 is the graph of a function 𝑓: 𝑅 → ℝ of two real variables, and is
given the upward orientation, then a natural parametrization is 𝐫(𝑢, 𝑣) = 〈𝑢, 𝑣, 𝑓(𝑢, 𝑣)〉, so
〈−𝑓𝑢 , −𝑓𝑣 , 1〉 𝑑𝑢 𝑑𝑣 = ∬ 𝐅 ⋅ 〈−𝑓𝑥 , −𝑓𝑦 , 1〉𝑑𝐴.
∬𝐅 ⋅ 𝑑𝐒 = ∬ 𝐅 ⋅ ⏟
𝑆 𝑅 𝐫𝑢 ×𝐫𝑣 𝑅

(iii) If 𝑆 is a level surface of a function 𝑓 of three real variables, then according to Theorem 3.74,
1
̂ = ‖∇𝑓‖ ∇𝑓, so given this orientation of 𝑆 we have
a unit normal vector to 𝑆 is given by 𝐧

𝐅 ⋅ ∇𝑓
∬𝐅 ⋅ 𝑑𝐒 = ∬ 𝑑𝑆 .
𝑆 𝑆 ‖∇𝑓‖
This formula is useful when the surface integral on the right is easy to compute without writing
down a parametrization of 𝑆.

Page 26 of 40
MATH2023 Multivariable Calculus Chapter 6 Vector calculus
L2/L3 (Fall 2019)

Example 6.80 Let 𝐅(𝑥, 𝑦, 𝑧) = 𝐤 be a (constant) vector field and let 𝑆 be the positively oriented
closed surface in ℝ3 consisting of
Positive orientation of 𝑆 means that
2 2 1 1 the outward unit normal is chosen.
(i) the disk 𝑥 + 𝑦 ≤ 2 on the plane 𝑧 = and
√2

1
(ii) the portion of the sphere 𝑥 2 + 𝑦 2 + 𝑧 2 = 1 above the plane 𝑧 = .
√2

Evaluate the surface integral 𝑧

𝑆2
∯𝐅 ⋅ 𝑑𝐒. 1
𝑆
1
Solution: √2 𝑆1
1
Let 𝑆1 denote the downward oriented disk 𝑥 2 + 𝑦 2 ≤ 2 on the plane
𝑦

1
𝑧= , and 𝑆2 denote the upward oriented portion of the sphere 𝑥
√2

1
𝑥 2 + 𝑦 2 + 𝑧 2 = 1 above the plane 𝑧 = . Then
√2

(i) ̂ = −𝐤, so
The unit normal vector to 𝑆1 is given by 𝐧

̂ 𝑑𝑆 = ∬ 𝐤 ⋅ (−𝐤)𝑑𝑆 = ∬ (−1)𝑑𝑆
∬ 𝐅 ⋅ 𝑑𝐒 = ∬ 𝐅 ⋅ 𝐧
𝑆1 𝑆1 𝑆1 𝑆1
2
1 𝜋
= (−1)(Area of 𝑆1 ) = (−1) [𝜋 ( ) ]=− .
√2 2
1
(ii) Let 𝑅 be the disk in ℝ2 defined by 𝑥 2 + 𝑦 2 ≤ 2. Then 𝑆2 is the upward oriented graph of

the function 𝑓: 𝑅 → ℝ defined by

𝑓(𝑥, 𝑦) = √1 − 𝑥 2 − 𝑦 2 .
For each (𝑥, 𝑦) in the interior of 𝑅 we have
𝑥 𝑦
𝑓𝑥 (𝑥, 𝑦) = − and 𝑓𝑦 (𝑥, 𝑦) = − ,
√1 − 𝑥 2 − 𝑦 2 √1 − 𝑥 2 − 𝑦 2
so

∬ 𝐅 ⋅ 𝑑𝐒 = ∬ 𝐅 ⋅ 〈−𝑓𝑥 , −𝑓𝑦 , 1〉𝑑𝐴

𝑆2 𝑅
𝑥 𝑦
= ∬ 〈0, 0, 1〉 ⋅ 〈 , , 1〉 𝑑𝐴
𝑅 √1 − 𝑥 2 − 𝑦 2 √1 − 𝑥 2 − 𝑦 2
2
1 𝜋
= ∬ 𝑑𝐴 = Area of 𝑅 = 𝜋 ( ) = .
𝑅 √2 2
Therefore combining (i) and (ii) we have
𝜋 𝜋
∯𝐅 ⋅ 𝑑𝐒 = ∬ 𝐅 ⋅ 𝑑𝐒 + ∬ 𝐅 ⋅ 𝑑𝐒 = − + = 0.
𝑆 𝑆1 𝑆2 2 2

Page 27 of 40
MATH2023 Multivariable Calculus Chapter 6 Vector calculus
L2/L3 (Fall 2019)

Example 6.81 Let 𝑆 be the positively oriented unit sphere in ℝ3 centered at the origin, and let
1
𝐅(𝑥, 𝑦, 𝑧) = 3
〈𝑥, 𝑦, 𝑧〉.
(𝑥 2 + 𝑦2 + 𝑧 2 )2 1
𝐅(𝐫) = 𝐫̂
𝑟2
Find the flux of 𝐅 across 𝑆.

Solution:
The sphere 𝑆 is defined by the equation 𝑥 2 + 𝑦 2 + 𝑧 2 = 1. It is a level surface of the function
𝑓(𝑥, 𝑦, 𝑧) = 𝑥 2 + 𝑦 2 + 𝑧 2 , whose gradient vector is
∇𝑓 = 2𝐫
∇𝑓(𝑥, 𝑦, 𝑧) = 〈2𝑥, 2𝑦, 2𝑧〉.
This gradient vector ∇𝑓 gives a normal vector to 𝑆 which points upward above the 𝑥𝑦-plane and
points downward below the 𝑥𝑦-plane. The orientation given by ∇𝑓 thus agrees with the positive
orientation of 𝑆, so the flux of 𝐅 across 𝑆 is

𝐅 ⋅ ∇𝑓 1 〈𝑥, 𝑦, 𝑧〉 ⋅ 〈2𝑥, 2𝑦, 2𝑧〉

∯𝐅 ⋅ 𝑑𝐒 = ∯ 𝑑𝑆 = ∯ 3 𝑑𝑆
𝑆 𝑆 ‖∇𝑓‖ 𝑆 (𝑥 2 + 𝑦2 + 𝑧 2 )2 √(2𝑥)
2 + (2𝑦)2 + (2𝑧)2

1
=∯ 𝑑𝑆 = ∯𝑑𝑆 = Surface area of 𝑆 = 4𝜋.
𝑆 𝑥2 + 𝑦2 + 𝑧2 𝑆

5. Gauss’ Divergence Theorem

Gauss’ Divergence Theorem provides another way of evaluating the surface integral of a vector
field across a closed surface in ℝ3 , which is sometimes quicker than evaluating the surface integral
directly.

Theorem 6.82 (Gauss’ Divergence Theorem) Let 𝐅 be a vector field in three dimensions whose
components have continuous partial derivatives, and let 𝑆 be a positively oriented closed surface in
ℝ3 whose interior is the bounded region 𝐷. Then

∯𝐅 ⋅ 𝑑𝐒 = ∭ ∇ ⋅ 𝐅 𝑑𝑉 .
Compare this with 2-D Divergence
𝑆 𝐷
Theorem, the second equation in
Remark 6.66.

Page 28 of 40
MATH2023 Multivariable Calculus Chapter 6 Vector calculus
L2/L3 (Fall 2019)

Proof. Write 𝐅 = 〈𝑓, 𝑔, ℎ〉. Similar to the proof of Green’s Theorem, we assume without loss of
generality that the region 𝐷 can be written as
𝐷 = {(𝑥, 𝑦, 𝑧) ∈ ℝ3 : (𝑥, 𝑦) ∈ 𝑅1 and 𝑝1 (𝑥, 𝑦) ≤ 𝑧 ≤ 𝑞1 (𝑥, 𝑦)}
= {(𝑥, 𝑦, 𝑧) ∈ ℝ3 : (𝑥, 𝑧) ∈ 𝑅2 and 𝑝2 (𝑥, 𝑧) ≤ 𝑦 ≤ 𝑞2 (𝑥, 𝑧)}
= {(𝑥, 𝑦, 𝑧) ∈ ℝ3 : (𝑦, 𝑧) ∈ 𝑅3 and 𝑝3 (𝑦, 𝑧) ≤ 𝑥 ≤ 𝑞3 (𝑦, 𝑧)}
for some planar regions 𝑅1 , 𝑅2 , 𝑅3 and some two-variable continuous functions 𝑝1 , 𝑞1 : 𝑅1 → ℝ,
𝑝2 , 𝑞2 : 𝑅2 → ℝ and 𝑝3 , 𝑞3 : 𝑅3 → ℝ.
𝑧 𝑦 𝑧 = 𝑞1 (𝑥, 𝑦)

𝑅1

𝑥
𝑧 = 𝑝1 (𝑥, 𝑦)
Then we have
Top: Upward oriented graph of 𝑞1
∯ℎ𝐤 ⋅ 𝑑𝐒 = (∬ + ∬ +∬ ) ℎ𝐤 ⋅ 𝑑𝐒 Bottom: Downward oriented graph of 𝑝1
𝑆 top bottom side
𝜕𝑞1 𝜕𝑞1
= ∬ 〈0, 0, ℎ(𝑥, 𝑦, 𝑞1 (𝑥, 𝑦))〉 ⋅ 〈− ,− , 1〉 𝑑𝑥𝑑𝑦
𝑅1 𝜕𝑥 𝜕𝑦
𝜕𝑝1 𝜕𝑝1
+ ∬ 〈0, 0, ℎ(𝑥, 𝑦, 𝑝1 (𝑥, 𝑦))〉 ⋅ 〈 , , −1〉 𝑑𝑥𝑑𝑦 + 0
𝑅1 𝜕𝑥 𝜕𝑦

= ∬ [ℎ(𝑥, 𝑦, 𝑞1 (𝑥, 𝑦)) − ℎ(𝑥, 𝑦, 𝑝1 (𝑥, 𝑦))]𝑑𝑥𝑑𝑦

𝑅1
𝑞1 (𝑥,𝑦)
𝜕ℎ 𝜕ℎ
=∬ ∫ (𝑥, 𝑦, 𝑧)𝑑𝑧 𝑑𝑥𝑑𝑦 = ∭ 𝑑𝑉 ,
𝑅1 𝑝1 (𝑥,𝑦) 𝜕𝑧 𝐷 𝜕𝑧
and similarly we have

∯𝑔𝐣 ⋅ 𝑑𝐒 = ∬ [𝑔(𝑥, 𝑞2 (𝑥, 𝑧), 𝑧) − 𝑔(𝑥, 𝑝2 (𝑥, 𝑧), 𝑧)]𝑑𝑥𝑑𝑧

𝑆 𝑅2
𝑞2 (𝑥,𝑧)
𝜕𝑔 𝜕𝑔
=∬ ∫ (𝑥, 𝑦, 𝑧)𝑑𝑦 𝑑𝑥𝑑𝑧 = ∭ 𝑑𝑉
𝑅2 𝑝2 (𝑥,𝑧) 𝜕𝑦 𝐷 𝜕𝑦
and

∯𝑓𝐢 ⋅ 𝑑𝐒 = ∬ [𝑓(𝑞3 (𝑦, 𝑧), 𝑦, 𝑧) − 𝑓(𝑝3 (𝑦, 𝑧), 𝑦, 𝑧)]𝑑𝑦𝑑𝑧

𝑆 𝑅3
𝑞3 (𝑦,𝑧)
𝜕𝑓 𝜕𝑓
=∬ ∫ (𝑥, 𝑦, 𝑧)𝑑𝑥 𝑑𝑦𝑑𝑧 = ∭ 𝑑𝑉 .
𝑅3 𝑝3 (𝑦,𝑧) 𝜕𝑥 𝐷 𝜕𝑥
Therefore
𝜕𝑓 𝜕𝑔 𝜕ℎ
∯𝐅 ⋅ 𝑑𝐒 = ∯(𝑓𝐢 + 𝑔𝐣 + ℎ𝐤) ⋅ 𝑑𝐒 = ∭ ( + + ) 𝑑𝑉 = ∭ ∇ ⋅ 𝐅 𝑑𝑉
𝑆 𝑆 𝐷 𝜕𝑥 𝜕𝑦 𝜕𝑧 𝐷
as desired. ∎

Page 29 of 40
MATH2023 Multivariable Calculus Chapter 6 Vector calculus
L2/L3 (Fall 2019)

Example 6.83 Let 𝑆 be the positively oriented surface of the solid in ℝ3 bounded by the
cylinder 𝑦 2 + 𝑧 2 = 1 and the planes 𝑥 = −1 and 𝑥 = 2, and let
𝐅(𝑥, 𝑦, 𝑧) = 3𝑥𝑦 2 𝐢 + 𝑥𝑒 𝑧 𝐣 + 𝑧 3 𝐤.
Find the flux of 𝐅 across 𝑆.
𝑧
Solution:
Let 𝐷 be the solid in ℝ3 bounded by the cylinder 𝑦 2 + 𝑧 2 = 1 and 𝑆 −1
the planes 𝑥 = −1 and 𝑥 = 2. Then by Gauss’ Divergence Theorem,
𝑦
the flux of 𝐅 across 𝑆 is 𝐷
2
∯𝐅 ⋅ 𝑑𝐒 = ∭ ∇ ⋅ 𝐅 𝑑𝑉
𝑆 𝐷 𝑥

= ∭ (3𝑦 2 + 0 + 3𝑧 2 )𝑑𝑉
The cylindrical coordinates 𝐷
(𝑥, 𝑟, 𝜃) here are taken in 2𝜋 1 2
the direction of the 𝑥-axis: =∫ ∫ ∫ 3𝑟 2 𝑑𝑥 𝑟𝑑𝑟 𝑑𝜃
𝑥=𝑥 0 0 −1
{𝑦 = 𝑟 cos 𝜃. 2𝜋 1 2
3
9𝜋
𝑧 = 𝑟 sin 𝜃 = (∫ 𝑑𝜃 ) (∫ 3𝑟 𝑑𝑟) (∫ 𝑑𝑥) = .
0 0 −1 2

Example 6.84 Let 𝑆 be the upward oriented portion of the circular paraboloid 𝑧 = 4 − 𝑥 2 − 𝑦 2
above the 𝑥𝑦-plane, and let
𝐅(𝑥, 𝑦, 𝑧) = 〈4𝑥 + 𝑒 𝑦𝑧 sin 𝑦 , 𝑥𝑒 𝑧 cos 𝑥 , 1 − 2𝑧〉.
Find the flux of 𝐅 across 𝑆.
Technique: “Complete the closed surface”
Solution:
𝑧
[It looks difficult to compute the surface integral directly, so we aim
to use Gauss’ Divergence Theorem.] 𝑆

𝐷
The given surface 𝑆 is not a closed surface. However, if we let
𝑆 ′ be the downward oriented closed disk on the 𝑥𝑦-plane with
𝑦
radius 2 and centered at the origin, then the surface consisting of 𝑆′
𝑆 together with 𝑆 ′ becomes a positively oriented closed surface 𝑥

in ℝ3 , whose interior is the region

𝐷 = {(𝑟, 𝜃, 𝑧): 0 ≤ 𝜃 ≤ 2𝜋 and 0 ≤ 𝑟 ≤ 2 and 0 ≤ 𝑧 ≤ 4 − 𝑟 2 }
defined in cylindrical coordinates. So by Gauss’ Divergence Theorem, we have
2𝜋 2 4−𝑟 2
∬𝐅 ⋅ 𝑑𝐒 + ∬ 𝐅 ⋅ 𝑑𝐒 = ∭ ∇ ⋅ 𝐅 𝑑𝑉 = ∫ ∫ ∫ (4 + 0 − 2)𝑑𝑧 𝑟𝑑𝑟 𝑑𝜃
𝑆 𝑆′ 𝐷 0 0 0
2
= 4𝜋 ∫ (4𝑟 − 𝑟 3 )𝑑𝑟 = 16𝜋.
0

Page 30 of 40
MATH2023 Multivariable Calculus Chapter 6 Vector calculus
L2/L3 (Fall 2019)

̂ to the disk
On the other hand, with the downward orientation chosen, the unit normal vector 𝐧
′ ′
̂ = 〈0, 0, −1〉. So the flux of 𝐅 across 𝑆 is
𝑆 is given by 𝐧

̂ 𝑑𝑆
∬ 𝐅 ⋅ 𝑑𝐒 = ∬ 𝐅 ⋅ 𝐧
𝑆′ 𝑆′

= ∬ 〈4𝑥 + 𝑒 𝑦𝑧 sin 𝑦 , 𝑥𝑒 𝑧 cos 𝑥 , 1 − 2𝑧〉 ⋅ 〈0, 0, −1〉𝑑𝑆

𝑆′

= ∬ (2𝑧 − 1)𝑑𝑆 = −1 ∬ 𝑑𝑆 = −1[𝜋(2)2 ] = −4𝜋.

𝑆′ 𝑆′
Therefore

∬𝐅 ⋅ 𝑑𝐒 = 16𝜋 − (−4𝜋) = 20𝜋.

𝑆

Example 6.85 Let 𝑆 be a positively oriented closed surface in ℝ3 whose interior contains the
origin, and let
1
𝐅(𝑥, 𝑦, 𝑧) = 3
〈𝑥, 𝑦, 𝑧〉.
(𝑥 2 + 𝑦2 + 𝑧 2 )2

Find the flux of 𝐅 across 𝑆.

Technique: “Drill a hole around the bad point”

Solution:
Note that 𝐅 is undefined at the origin, so we cannot directly apply Gauss’ Divergence Theorem to
𝐅 on the interior of 𝑆. Instead, we let 𝑆 ′ be a positively oriented sphere centered at the origin,
with radius 𝑎 > 0 small enough so that 𝑆 ′ lies completely in the interior of 𝑆, and we let 𝐷 be
the region inside 𝑆 but outside 𝑆 ′ . Then since
∇⋅𝐅= 0 𝑆′
as computed in Example 6.25, by Gauss’ Divergence Theorem we have

∯𝐅 ⋅ 𝑑𝐒 + ∯ 𝐅 ⋅ 𝑑𝐒 = ∭ ∇ ⋅ 𝐅 𝑑𝑉 = ∭ 0 𝑑𝑉 = 0,
𝐷
𝑆 −𝑆 ′ 𝐷 𝐷
and so 𝑆

∯𝐅 ⋅ 𝑑𝐒 = − ∯ 𝐅 ⋅ 𝑑𝐒 = ∯ 𝐅 ⋅ 𝑑𝐒.
𝑆 −𝑆 ′ 𝑆′
But similar to Example 6.81, the last surface integral is simply
1 1 1
∯ 𝐅 ⋅ 𝑑𝐒 = ∯ 2 2 2
𝑑𝑆 = ∯ 2 𝑑𝑆 = 2 (4𝜋𝑎2 ) = 4𝜋,
𝑆′ 𝑆′ 𝑥 + 𝑦 + 𝑧 𝑆′ 𝑎 𝑎
so

∯𝐅 ⋅ 𝑑𝐒 = 4𝜋
𝑆
as well.

Page 31 of 40
MATH2023 Multivariable Calculus Chapter 6 Vector calculus
L2/L3 (Fall 2019)

Remark 6.86 (An interpretation of the divergence) Consider a very small closed surface 𝑆 in ℝ3 ,
whose interior 𝐷 contains a point (𝑎, 𝑏, 𝑐). Let 𝐅 be a vector field on ℝ3 whose components
have continuous partial derivatives, so that ∇ ⋅ 𝐅 is almost a constant function on the small region
𝐷. According to Gauss’ Divergence Theorem we have

∯𝐅 ⋅ 𝑑𝐒 = ∭ ∇ ⋅ 𝐅 𝑑𝑉
𝑆 𝐷

≈ (∇ ⋅ 𝐅)(𝑎, 𝑏, 𝑐) ∭ 𝑑𝑉
𝐷
= (∇ ⋅ 𝐅)(𝑎, 𝑏, 𝑐)(Volume of 𝐷).
As a result, the divergence (∇ ⋅ 𝐅)(𝑎, 𝑏, 𝑐) can also be regarded as the “flux density” of 𝐅 near
the point (𝑎, 𝑏, 𝑐):
∯𝑆 𝐅 ⋅ 𝑑𝐒
(∇ ⋅ 𝐅)(𝑎, 𝑏, 𝑐) ≈ .
Volume of 𝐷

6. Stokes’ Theorem

Stokes’ Theorem provides another way to evaluate the line integral of a vector field along a simple
closed curve in ℝ3 . In Definition 6.52 we have defined the positive orientation for a simple
closed curve in ℝ2 . Now we need a similar notion for a simple closed curve in ℝ3 .

Definition 6.87 Let 𝐶 be a simple closed curve in ℝ3 and let 𝑆 be an oriented surface in ℝ3
whose boundary curve is 𝐶. Then we say that 𝐶 is positively oriented with respect to 𝑺 if the
orientation of 𝐶 follows the right-hand rule, i.e. whenever the fingers are curved in the direction
along 𝐶, the thumb will point toward the chosen normal vector 𝐧 ̂ to the surface 𝑆.

Theorem 6.88 (Stokes) Let 𝐅 be a vector field in three dimensions whose components have
continuous partial derivatives, and let 𝑆 be an oriented surface in ℝ3 , whose boundary curve 𝐶 is
a simple closed curve which is positively oriented with respect to 𝑆. Then

∮𝐅 ⋅ 𝑑𝐫 = ∬(∇ × 𝐅) ⋅ 𝑑𝐒. Compare this with Green’s

𝐶 𝑆 Theorem, the first equation
in Remark 6.66.

Page 32 of 40
MATH2023 Multivariable Calculus Chapter 6 Vector calculus
L2/L3 (Fall 2019)

Proof. Write 𝐅 = 〈𝑓, 𝑔, ℎ〉. We assume without loss of generality that the orientation of the
surface 𝑆 is induced by a smooth parametrization 𝛔: 𝑅 → ℝ3 , where 𝑅 is a region in ℝ2
containing the parameters (𝑢, 𝑣), and the boundary of 𝑅 is a positively oriented smooth simple
closed curve 𝐶 ′ in ℝ2 with a parametrization 𝐬: [𝑎, 𝑏] → ℝ2 .

𝑣
Curve 𝐶′ Surface 𝑆 𝑆
parametrized by parametrized by
𝑅 𝐶′ 𝛔: 𝑅 → ℝ3
𝐬: [𝑎, 𝑏] → ℝ2
[ ] 𝐶
𝑎 𝑡 𝑏
𝑢

Since 𝐶 is the positively oriented boundary of 𝑆, it has a parametrization 𝐫 = 𝛔 ∘ 𝐬: [𝑎, 𝑏] → ℝ3 .

So we have
𝑏 𝑏
′ (𝑡)𝑑𝑡
𝑑𝑢 𝑑𝑣
∮𝐅 ⋅ 𝑑𝐫 = ∫ 𝐅(𝐫(𝑡)) ⋅ 𝐫 = ∫ 𝐅 (𝛔(𝑢(𝑡), 𝑣(𝑡))) ⋅ (𝛔𝑢 + 𝛔𝑣 ) 𝑑𝑡
𝐶 𝑎 𝑎 𝑑𝑡 𝑑𝑡

= ∮ (𝐅 ⋅ 𝛔𝑢 )𝑑𝑢 + (𝐅 ⋅ 𝛔𝑣 )𝑑𝑣.
𝐶′
Now consider the vector field in two dimensions defined by
𝐆(𝑢, 𝑣) = 〈(𝐅 ⋅ 𝛔𝑢 )(𝑢, 𝑣), (𝐅 ⋅ 𝛔𝑣 )(𝑢, 𝑣)〉.
Then

∮ (𝐅 ⋅ 𝛔𝑢 )𝑑𝑢 + (𝐅 ⋅ 𝛔𝑣 )𝑑𝑣 = ∮ 𝐆 ⋅ 𝑑𝐬
𝐶′ 𝐶′
𝜕(𝐅 ⋅ 𝛔𝑢 ) 𝜕(𝐅 ⋅ 𝛔𝑣 )
=∬[ − ] 𝑑𝑢𝑑𝑣
𝑅 𝜕𝑢 𝜕𝑣
according to Green’s Theorem. On the other hand, we also have
ℎ𝑦 − 𝑔𝑧 𝑓𝑧 − ℎ𝑥 𝑔𝑥 − 𝑓𝑦 𝜕(𝐅 ⋅ 𝛔𝑣 ) 𝜕(𝐅 ⋅ 𝛔𝑢 )
(∇ × 𝐅) ⋅ (𝛔𝑢 × 𝛔𝑣 ) = | 𝑥𝑢 𝑦𝑢 𝑧𝑢 | = −
𝑥𝑣 𝑦𝑣 𝑧𝑣 𝜕𝑢 𝜕𝑣

(verify the last equality using chain rule!), and so

∬(∇ × 𝐅) ⋅ 𝑑𝐒 = ∬ (∇ × 𝐅) ⋅ (𝛔𝑢 × 𝛔𝑣 )𝑑𝑢𝑑𝑣

𝑆 𝑅
𝜕(𝐅 ⋅ 𝛔𝑣 ) 𝜕(𝐅 ⋅ 𝛔𝑢 )
=∬[ − ] 𝑑𝑢𝑑𝑣
𝑅 𝜕𝑢 𝜕𝑣
as well. Therefore

∮𝐅 ⋅ 𝑑𝐫 = ∬(∇ × 𝐅) ⋅ 𝑑𝐒
𝐶 𝑆
as desired. ∎

Page 33 of 40
MATH2023 Multivariable Calculus Chapter 6 Vector calculus
L2/L3 (Fall 2019)

Example 6.89 Let 𝐅: ℝ3 → ℝ3 be the vector field defined by

𝐅(𝑥, 𝑦, 𝑧) = 〈𝑧, 𝑥, 𝑦〉
and let 𝑆 be the portion of the paraboloid 𝑥 2 + 𝑦 2 + 𝑧 = 1 above the 𝑥𝑦-plane, oriented
upward. Verify that Stokes’ Theorem holds for the vector field 𝐅 and the oriented surface 𝑆.

Solution: 𝑧

Let 𝐶 be boundary curve of 𝑆 which is positively oriented with 𝑆

respect to 𝑆. The issue is to compute

∮𝐅 ⋅ 𝑑𝐫 and ∬(∇ × 𝐅) ⋅ 𝑑𝐒
𝐶 𝑆
𝑦
separately and verify that they are equal.
𝐶
𝑥
 On the one hand, 𝐶 is the circle
2 2
{𝑥 + 𝑦 = 1
𝑧=0
oriented counterclockwise when viewed from above the 𝑥𝑦-plane because of the upward
orientation of 𝑆. So we may parametrize 𝐶 by
𝐫(𝑡) = 〈cos 𝑡 , sin 𝑡 , 0〉
where 𝑡 ∈ [0, 2𝜋]. Thus 𝐫 ′ (𝑡) = 〈− sin 𝑡 , cos 𝑡 , 0〉 for every 𝑡 ∈ (0, 2𝜋), and we have
2𝜋 2𝜋
∮𝐅 ⋅ 𝑑𝐫 = ∫ 〈0, cos 𝑡 , sin 𝑡〉 ⋅ 〈− sin 𝑡 , cos 𝑡 , 0〉𝑑𝑡 = ∫ cos 2 𝑡 𝑑𝑡 = 𝜋.
𝐶 0 0
 On the other hand, 𝑆 is the graph of the function
𝑓(𝑥, 𝑦) = 1 − 𝑥 2 − 𝑦 2
defined on the unit disk 𝑅 in ℝ2 centered at the origin, with the natural upward orientation.
Since
𝐢 𝐣 𝐤
𝜕 𝜕 𝜕
∇×𝐅= | | = 〈1, 1, 1〉,
𝜕𝑥 𝜕𝑦 𝜕𝑧
𝑧 𝑥 𝑦
we have

∬(∇ × 𝐅) ⋅ 𝑑𝐒 = ∬ (∇ × 𝐅) ⋅ 〈−𝑓𝑥 , −𝑓𝑦 , 1〉𝑑𝐴 = ∬ 〈1, 1, 1〉 ⋅ 〈2𝑥, 2𝑦, 1〉𝑑𝐴

𝑆 𝑅 𝑅

= ∬ (2𝑥 + 2𝑦 + 1)𝑑𝐴 = ∬ 𝑑𝐴 = 𝜋.
𝑅 𝑅
Since the disk 𝑅 is symmetric about the
line 𝑥 = 0 and is also symmetric about
The above two paragraphs have shown that the line 𝑦 = 0, we have

∮𝐅 ⋅ 𝑑𝐫 = ∬(∇ × 𝐅) ⋅ 𝑑𝐒. ∬ 𝑥 𝑑𝐴 = 0 and ∬ 𝑦 𝑑𝐴 = 0.

𝐶 𝑆 𝑅 𝑅

Page 34 of 40
MATH2023 Multivariable Calculus Chapter 6 Vector calculus
L2/L3 (Fall 2019)

Example 6.90 Let 𝑎 > 0 and let 𝐶 be the circle 𝑥 2 + 𝑦 2 = 𝑎2 lying on the 𝑥𝑦-plane in ℝ3 ,
oriented counterclockwise when viewed from above the 𝑥𝑦-plane. Evaluate the line integral
2
∮𝑥 2 𝑦 3 𝑑𝑥 + 𝑒 𝑦 cos 𝑧 2 𝑑𝑦 + 𝑥𝑦 sin 𝑒 𝑧 𝑑𝑧.
𝐶

Solution:
[The given line integral is along a curve in ℝ3 and looks difficult to be computed directly, so we aim
to use the Stokes’ Theorem.]

Let 𝑆 be the upward oriented disk in ℝ3 lying on the 𝑥𝑦-plane, 𝑧

which is defined by the expressions
2 2 2
{𝑥 + 𝑦 ≤ 𝑎 , 𝑆
𝑧=0
so that 𝐶 is the boundary curve of 𝑆 which is positively oriented
𝑦
with respect to 𝑆. Then 𝑆 is simply the graph of the zero function
𝐶
𝑓(𝑥, 𝑦) = 0, 𝑥
defined on the disk 𝑅 in ℝ2 with radius 𝑎 centered at the origin.

Also let 𝐅: ℝ3 → ℝ3 be the vector field defined by

2
𝐅(𝑥, 𝑦, 𝑧) = 〈𝑥 2 𝑦 3 , 𝑒 𝑦 cos 𝑧 2 , 𝑥𝑦 sin 𝑒 𝑧 〉,
so that the curl of 𝐅 is
𝐢 𝐣 𝐤
𝜕 𝜕 𝜕
∇ × 𝐅 = || || = 〈𝑥 sin 𝑒 𝑧 − 2𝑧𝑒 𝑦 2 sin 𝑧 2 , −𝑦 sin 𝑒 𝑧 , −3𝑥 2 𝑦 2 〉.
𝜕𝑥 𝜕𝑦 𝜕𝑧
2
𝑥2𝑦3 𝑒 𝑦 cos 𝑧 2 𝑥𝑦 sin 𝑒 𝑧

Then by Stokes’ Theorem we have

2
∮𝑥 2 𝑦 3 𝑑𝑥 + 𝑒 𝑦 cos 𝑧 2 𝑑𝑦 + 𝑥𝑦 sin 𝑒 𝑧 𝑑𝑧
𝐶

= ∮𝐅 ⋅ 𝑑𝐫 = ∬(∇ × 𝐅) ⋅ 𝑑𝐒 = ∬ (∇ × 𝐅) ⋅ 〈−𝑓𝑥 , −𝑓𝑦 , 1〉𝑑𝐴

𝐶 𝑆 𝑅

= ∬ 〈𝑥 sin 𝑒 0 − 0, −𝑦 sin 𝑒 0 , −3𝑥 2 𝑦 2 〉 ⋅ 〈0, 0, 1〉𝑑𝐴 = −3 ∬ 𝑥 2 𝑦 2 𝑑𝐴

𝑅 𝑅
2𝜋 𝑎
= −3 ∫ ∫ (𝑟 cos 𝜃)2 (𝑟 sin 𝜃)2 𝑟𝑑𝑟 𝑑𝜃
0 0
𝑎 2𝜋
5
1 2 𝜋
= −3 (∫ 𝑟 𝑑𝑟) (∫ sin 2𝜃 𝑑𝜃) = − 𝑎6 .
0 0 4 8

Page 35 of 40
MATH2023 Multivariable Calculus Chapter 6 Vector calculus
L2/L3 (Fall 2019)

Example 6.91 Let 𝐅: ℝ3 → ℝ3 be the vector field defined by

𝐅(𝑥, 𝑦, 𝑧) = 〈𝑦, −𝑥, 𝑥 2 𝑦 2 𝑧〉
and let 𝑆 be the portion of the ellipsoid 𝑥 2 + 𝑦 2 + 3𝑧 2 = 1 below the 𝑥𝑦-plane, oriented
downward. Compute the surface integral ∬𝑆(∇ × 𝐅) ⋅ 𝑑𝐒 using Stokes’ Theorem.

Solution: 𝑧
Let 𝐶 be the circle in ℝ3 lying on the 𝑥𝑦-plane defined by
𝐶
2 2
{𝑥 + 𝑦 = 1 ,
𝑧=0
which is oriented clockwise when viewed from above the 𝑥𝑦-plane, 𝑦

so that 𝐶 is the boundary curve of 𝑆 which is positively oriented 𝑥

with respect to 𝑆. 𝐶 can be parametrized by
𝑆
𝐫(𝑡) = 〈cos 𝑡 , − sin 𝑡 , 0〉
for 𝑡 ∈ [0, 2𝜋]. By Stokes’ Theorem we have
2𝜋
∬(∇ × 𝐅) ⋅ 𝑑𝐒 = ∮𝐅 ⋅ 𝑑𝐫 = ∫ 𝐅(𝐫(𝑡)) ⋅ 𝐫 ′ (𝑡)𝑑𝑡
𝑆 𝐶 0
2𝜋 2𝜋
= ∫ 〈− sin 𝑡 , − cos 𝑡 , 0〉 ⋅ 〈− sin 𝑡 , − cos 𝑡 , 0〉𝑑𝑡 = ∫ 𝑑𝑡 = 2𝜋.
0 0

Example 6.92 Let 𝐅 be a vector field in three dimensions whose components have continuous
partial derivatives, and let 𝑆 be a smooth closed surface in ℝ3 . Show that

∯(∇ × 𝐅) ⋅ 𝑑𝐒 = 0.
𝑆
Proof:
By Stokes’ Theorem we have

∯(∇ × 𝐅) ⋅ 𝑑𝐒 = ∮𝐅 ⋅ 𝑑𝐫
𝑆 𝐶
where 𝐶 is the boundary curve of 𝑆, oriented positively with respect to 𝑆. But since 𝑆 is a
closed surface, it does not have any boundary curve at all. Consequently

∯(∇ × 𝐅) ⋅ 𝑑𝐒 = 0.
𝑆
∎
Alternative proof:
Let 𝐷 be the interior of the closed surface 𝑆, and equip 𝑆 with the positive orientation. By
Gauss’ Divergence Theorem we have

∯(∇ × 𝐅) ⋅ 𝑑𝐒 = ∭ ∇ ⋅ (∇ × 𝐅) 𝑑𝑉 = ∭ 0 𝑑𝑉 = 0,
𝑆 𝐷 𝐷
since ∇ ⋅ (∇ × 𝐅) = 0 by Theorem 6.26. ∎

Page 36 of 40
MATH2023 Multivariable Calculus Chapter 6 Vector calculus
L2/L3 (Fall 2019)

Remark 6.93 (An interpretation of the curl) Consider a very small simple closed curve 𝐶 in ℝ3 ,
which is the boundary of a very small oriented surface 𝑆, and let (𝑎, 𝑏, 𝑐) be a point on 𝑆. Let 𝐅
be a vector field on ℝ3 whose components have continuous partial derivatives, so that ∇ × 𝐅 is
almost a constant vector on the small surface 𝑆. According to Stokes’ Theorem we have

∮𝐅 ⋅ 𝑑𝐫 = ∬(∇ × 𝐅) ⋅ 𝑑𝐒
𝐶 𝑆

≈ [(∇ × 𝐅)(𝑎, 𝑏, 𝑐) ⋅ 𝐧
̂] ∬𝑑𝑆
𝑆
= [(∇ × 𝐅)(𝑎, 𝑏, 𝑐) ⋅ 𝐧
̂](Surface area of 𝑆).
3
̂ in ℝ , the quantity
As a result, given any unit vector 𝐧
(∇ × 𝐅)(𝑎, 𝑏, 𝑐) ⋅ 𝐧
̂
(i.e. the scalar projection of the curl onto the direction of 𝐧 ̂) can be regarded as the “circulation
density” of 𝐅 near the point (𝑎, 𝑏, 𝑐) around the direction of 𝐧 ̂ following the right-hand rule:
∮𝐶 𝐅 ⋅ 𝑑𝐫
(∇ × 𝐅)(𝑎, 𝑏, 𝑐) ⋅ 𝐧
̂≈ .
Surface area of 𝑆

Definition 6.94 We say that a (connected) region 𝐷 in ℝ3 is simply connected if for every
simple closed curve 𝐶 in 𝐷, there exists a surface 𝑆 lying completely in 𝐷 whose boundary
curve is 𝐶.

Example 6.95 Consider the following regions in ℝ3 :

𝐷1 = {(𝑥, 𝑦, 𝑧): 1 ≤ 𝑥 2 + 𝑦 2 + 𝑧 2 ≤ 4}
and
𝐷2 = {(𝑥, 𝑦, 𝑧): 𝑥 2 + 𝑦 2 + 𝑧 2 ≤ 4 and 𝑥 2 + 𝑦 2 ≥ 1}.
In other words, as in the following diagram, 𝐷1 consists of all points inside the ball of radius 2 but
outside the ball of radius 1; while 𝐷2 consists of all points inside the ball of radius 2 but outside
the solid cylinder. Then 𝐷1 is simply connected, but 𝐷2 is not simply connected.

𝐷1
𝐷2

A simple closed curve in 𝐷2 which is not the boundary

curve of any surface lying completely in 𝐷2

Page 37 of 40
MATH2023 Multivariable Calculus Chapter 6 Vector calculus
L2/L3 (Fall 2019)

On a simply connected region in ℝ3 , conservative vector fields can be characterized by zero curl.

Theorem 6.96 (Curl test) Let 𝐷 be a simply connected region in ℝ3 and 𝐅: 𝐷 → ℝ3 be a vector
field. If
∇ × 𝐅 = 𝟎,
on 𝐷, then 𝐅 is conservative in 𝐷.

Sketch of proof. For each simple closed curve 𝐶 in 𝐷, there exists an orientable surface 𝑆 lying
completely in 𝐷 whose boundary curve is 𝐶. By Stokes’ theorem we always have

∮𝐅 ⋅ 𝑑𝐫 = ∬(∇ × 𝐅) ⋅ 𝑑𝐒 = ∬𝟎 ⋅ 𝑑𝐒 = 0.
𝐶 𝑆 𝑆
So line integrals of 𝐅 are path-independent in 𝐷, and 𝐅 is conservative in 𝐷 by Theorem 6.49. ∎

Remark 6.97 Let 𝐷 be a simply connected region in ℝ3 and let 𝐅: 𝐷 → ℝ3 be a vector field.
Then the following statements are logically equivalent:
(i) 𝐅 is conservative in 𝐷.
(ii) There exists a function 𝑓: 𝐷 → ℝ (a potential function of 𝐅) such that 𝐅 = ∇𝑓 on 𝐷.
(iii) Line integrals of 𝐅 are path-independent on 𝐷.
(iv) For every closed curve in 𝐷, we have

∮𝐅 ⋅ 𝑑𝐫 = 0.
𝐶
(v) ∇ × 𝐅 = 𝟎 on 𝐷.

Remark 6.98 Here is a summary on the results about vector calculus that we have learnt so far.

(i) Green’s Theorem:

 Consider a region 𝑅 in ℝ2 , whose boundary is a simple closed curve 𝐶.
 Consider a vector field 𝐅 = 〈𝑓, 𝑔〉, with 𝑅
We regard 𝐅 = 〈𝑓, 𝑔, 0〉. 𝐶
(∇ × 𝐅) ⋅ 𝐤 = 𝑔𝑥 − 𝑓𝑦 .
 ∬𝑅(𝑔𝑥 − 𝑓𝑦 )𝑑𝐴 = ∮𝐶 𝐅 ⋅ 𝑑𝐫 is saying that the integral of ∇ × 𝐅 across the region 𝑅
equals the sum of the components of 𝐅 tangent to the boundary 𝐶.

(ii) 2-D Divergence Theorem:

 Consider a region 𝑅 in ℝ2 , whose boundary is a simple closed curve 𝐶.
 Consider a vector field 𝐅 = 〈𝑓, 𝑔〉, with 𝑅
𝐶
∇ ⋅ 𝐅 = 𝑓𝑥 + 𝑔𝑦 .
 ̂ 𝑑𝑠 is saying that the integral of ∇ ⋅ 𝐅 over the region 𝑅
∬𝑅(𝑓𝑥 + 𝑔𝑦 )𝑑𝐴 = ∮𝐶 𝐅 ⋅ 𝐧
equals the sum of the components of 𝐅 normal to the boundary 𝐶.

Page 38 of 40
MATH2023 Multivariable Calculus Chapter 6 Vector calculus
L2/L3 (Fall 2019)

(iii) Gauss’ Divergence Theorem:

 Consider a region 𝐷 in ℝ3 , whose boundary is a closed surface 𝑆. 𝐷
𝑆
 Consider a vector field 𝐅 and its divergence ∇ ⋅ 𝐅.
 ∭𝐷 ∇ ⋅ 𝐅 𝑑𝑉 = ∯𝑆 𝐅 ⋅ 𝐧 ̂ 𝑑𝑆 is saying that the integral of ∇ ⋅ 𝐅 over the region 𝐷 equals
the sum of the components of 𝐅 normal to the boundary 𝑆.

𝑆
(iv) Stokes’ Theorem:
 Consider a surface 𝑆 in ℝ3 , whose boundary is a simple closed curve 𝐶.
 Consider a vector field 𝐅 and its curl ∇ × 𝐅. 𝐶
 ∬𝑆(∇ × 𝐅) ⋅ 𝑑𝐒 = ∮𝐶 𝐅 ⋅ 𝑑𝐫 is saying that the integral of ∇ × 𝐅 across the surface 𝑆
equals the sum of the components of 𝐅 tangent to the boundary 𝐶.

(v) Fundamental Theorem of Calculus:

 Consider an interval [𝑎, 𝑏], whose boundary just consists of the two numbers 𝑎 and 𝑏.
̂(𝑎) = −𝐢 and 𝐧
The “outward unit normal vector” to the boundary is 𝐧 ̂(𝑏) = 𝐢.
 Consider the “vector field” 𝐅(𝑥) = 𝑓(𝑥)𝐢 defined on [𝑎, 𝑏], with
[ ]
(∇ ⋅ 𝐅)(𝑥) = 𝑓 ′ (𝑥). 𝑎 𝑏
𝑏
 ∫𝑎 𝑓 ′ (𝑥)𝑑𝑥 = 𝑓(𝑏) − 𝑓(𝑎) is saying that the integral of ∇ ⋅ 𝐅 over the interval [𝑎, 𝑏]

equals the sum of the components of 𝐅 normal to the boundary of [𝑎, 𝑏].

(vi) Path-independence of conservative fields:

 Consider a curve 𝐶 of finite length with arc-length parametrization 𝐫: [𝑎, 𝑏] → ℝ𝑛 ,
whose boundary just consists of the two end-points 𝐫(𝑎) and 𝐫(𝑏). The “outward unit
̂(𝐫(𝑎)) = −𝐫 ′ (𝑎) and 𝐧
normal vector” to the boundary is 𝐧 ̂(𝐫(𝑏)) = 𝐫 ′ (𝑏).
 Consider the “vector field” 𝐅(𝐫(𝑠)) = 𝑓(𝐫(𝑠))𝐫 ′ (𝑠) defined on 𝐶, with 𝐶
𝐫(𝑎)
𝑑
(∇ ⋅ 𝐅)(𝐫(𝑠)) = 𝑓(𝐫(𝑠)) = ∇𝑓(𝐫(𝑠)) ⋅ 𝐫 ′ (𝑠).
𝑑𝑠 𝐫(𝑏)
 ∫𝐶 ∇𝑓 ⋅ 𝑑𝐫 = 𝑓(𝐫(𝑏)) − 𝑓(𝐫(𝑎)) is saying that the integral of ∇ ⋅ 𝐅 over the curve 𝐶
equals the sum of the components of 𝐅 normal to the boundary of 𝐶.

All these results are reflected in the following unifying principle:

The integral of a differential operator (∇ ⋅ or ∇ ×) acting on a vector field over a geometric object
equals the sum of the appropriate field components (normal or tangential) over the boundary of
the geometric object.

Page 39 of 40
MATH2023 Multivariable Calculus Chapter 6 Vector calculus
L2/L3 (Fall 2019)

Summary of Chapter 6

The following are what you need to know in this chapter in order to get a pass (a distinction) in this
course:

 Vector fields and their visual presentation, field lines of a vector field
 Curl of a vector field in three dimensions
 Divergence of a vector field
 Conservative vector fields in a region, to find a potential function of a conservative vector field

 Line integral of a function over a curve, 𝑑𝑠 = ‖𝐫 ′ (𝑡)‖𝑑𝑡

 ̂ 𝑑𝑠 = 𝐫 ′ (𝑡)𝑑𝑡
Line integral of a vector field along a curve, orientation of a curve, 𝑑𝐫 = 𝐓
 Physical interpretations of the line integral of a vector field
 Work done by a force field on an object moving along a curve
 Flow or circulation of a velocity field
 Path-dependence of line integrals
 Path-independence of line integrals of conservative vector fields
 Green’s Theorem: Another way of evaluating line integrals of a vector field along a simple
closed curve in ℝ𝟐

 Surface integral of a function over a surface

 𝑑𝑆 = ‖𝐫𝑢 × 𝐫𝑣 ‖𝑑𝑢 𝑑𝑣 if 𝑆 has a parametrization 𝐫

 𝑑𝑆 = √1 + 𝑓𝑥 2 + 𝑓𝑦 2 𝑑𝐴 if 𝑆 is the graph of 𝑓, i.e. 𝑧 = 𝑓(𝑥, 𝑦)

 Surface integral (or flux) of a vector field across an oriented surface, orientation of an
̂ 𝑑𝑆
orientable surface, 𝑑𝐒 = 𝐧
 𝑑𝐒 = (𝐫𝑢 × 𝐫𝑣 )𝑑𝑢 𝑑𝑣 if 𝑆 has orientation induced by a parametrization 𝐫
 𝑑𝐒 = 〈−𝑓𝑥 , −𝑓𝑦 , 1〉𝑑𝐴 if 𝑆 is the upward oriented graph of 𝑓, i.e. 𝑧 = 𝑓(𝑥, 𝑦)
 Gauss’ Divergence Theorem: Another way of evaluating surface integrals of a vector field
𝟑
across a closed surface in ℝ
 Stokes’ Theorem: Another way of evaluating line integrals of a vector field along a simple
closed curve in ℝ𝟑

 Characterization of conservative vector fields in a region 𝐷:

 Has potential functions on 𝐷
 Line integrals are path-independent on 𝐷
 Zero line integral around any closed curve in 𝐷
 Zero curl on 𝐷 (if 𝐷 is a simply connected region)

Page 40 of 40