MATH 2023 Tutorial 12 (T1A)

Green’s theorem, curl and divergence

4/5/2023

Green’s theorem
#»
• Works only for ℝ2 vector fields 𝐹(𝑥, 𝑦) = ⟨𝑃(𝑥, 𝑦) , 𝑄(𝑥, 𝑦)⟩.
• Green’s theorem:
– Let 𝐶 be positively oriented, piecewise-smooth, simply closed.
– 𝐷 be the region bounded by 𝐶.
– If 𝑃, 𝑄 ∈ 𝐶 1 (𝐷).
– Then
#» #» 𝜕𝑄 𝜕𝑃
∮ 𝐹 ⋅ d 𝑟 = ∮ [𝑃 d𝑥 + 𝑄 d𝑦] = ∬ 𝜕𝑥 − 𝜕𝑦 d𝐴 .
𝐶 𝐶 𝐷

• For domains with “holes”:

– Suppose the boundaries of holes are in negative orientation.
– Cut the domain in the bounded areas, giving pieces without holes.
– Apply Green’s theorem on each domain.
– Add the integral at each pieces back, the line integral along “cuts” are cancelled out as
they are in opposite orientation.
#»
• Can be used to avoid discontinuities of 𝐹 .

Curl
#»
• Works only for ℝ3 vector fields 𝐹 = ⟨𝑃, 𝑄, 𝑅⟩.
#»
• Assume all partial derivative of the components of 𝐹 exists.
#»
• The curl of 𝐹 :
⎡ ı⃗ ȷ⃗ k⃗ ⎤
#» 𝜕 𝜕 𝜕⎥
curl 𝐹 = ∇ × 𝐹 = det ⎢
⎢ 𝜕𝑥 𝜕𝑦 𝜕𝑗 ⎥ .
⎣ 𝑃 𝑄 𝑅⎦
– View ∇ as a operator
𝜕 𝜕 𝜕
∇=⟨ , ⟩,
𝜕𝑥 𝜕𝑦 𝜕𝑧
and take cross product.
• Interpretation: local spin of a vector field (positive if counterclockwise).
– Not necessary the rotation shown in the field.
• Theorem:

1
 #»
– If 𝑓 ∶ ℝ3 → ℝ, 𝑓 ∈ 𝐶 2 , then ∇ × (∇𝑓) = 0 .
#» #» #» #»
– Consequently, if 𝐹 is conservative (i.e. there is 𝑓 such that 𝐹 = ∇𝑓), then ∇ × 𝐹 = 0 .
• Theorem:
#»
– Let 𝐹 = ⟨𝑃, 𝑄, 𝑅⟩ be defined on the whole ℝ3 .
– 𝑃, 𝑄, 𝑅 ∈ 𝐶 1 (ℝ3 ).
#» #»
– ∇× 𝐹 = 0.
#»
– Then 𝐹 is conservative.

Divergence
• Works for ℝ𝑛 vector fields.
#» #»
• Suppose 𝐹 ∶ ℝ3 → ℝ3 , 𝐹 = ⟨𝑃, 𝑄, 𝑅⟩, all partial derivative of the components of 𝐹 exists.
#»
• The divergence of 𝐹 :
#» 𝜕𝑃 𝜕𝑄 𝜕𝑅
div 𝐹 = ∇ ⋅ 𝐹 = + + .
𝜕𝑥 𝜕𝑦 𝜕𝑧
– View ∇ as a operator and take dot product.
• Interpretation: net quantity gathering or flowing out of a point (positive if flowing out).
#»
• Theorem: If 𝑃, 𝑄, 𝑅 ∈ 𝐶 2 , then ∇ ⋅ (∇ × 𝐹 ) = 0.

Vector forms of Green’s theorem

#» #»
• Let 𝐹 = ⟨𝑃, 𝑄⟩ and 𝐺 = ⟨𝑃, 𝑄, 0⟩, then
#» 𝜕𝑄 𝜕𝑃 ⃗
∇× 𝐺 = ( − ) k.
𝜕𝑥 𝜕𝑦

• Green’s theorem (first form) can be rewritten as

#» #» #»
∮ 𝐹 ⋅ d 𝑟 = ∮ [𝑃 d𝑥 + 𝑄 d𝑦] = ∬ (∇ × 𝐺) ⋅ k⃗ d𝐴 .
𝐶 𝐶 𝐷

• Second form:
#» #» # »
∮ 𝐹 ⋅ 𝑛 d𝑠 = ∬ ∇ ⋅ 𝐹(𝑥, 𝑦) d𝐴 ,
𝐶 𝐷
#» is the unit outer normal of 𝐷.
where 𝑛

2
1 (Fitzpatrick, Ex 20.3, Q 1). Verify Green’s theorem for the case where

𝐹(𝑥, 𝑦) = ⟨𝑥2 − 𝑦2 , 2𝑥𝑦⟩,

and 𝐷 is the triangle with vertices (0, 0), (2, 0) and (1, 1).

2 (Ex 16.4, Q 16). Use Green’s theorem to evaluate

#» #»
∫ 𝐹⋅d𝑟 ,
𝐶

where
#»
𝐹(𝑥, 𝑦) = ⟨𝑒 −𝑥 + 𝑦2 , 𝑒 −𝑦 + 𝑥2 ⟩,
and 𝐶 consists of the arc of the curve 𝑦 = cos 𝑥 from (− 𝜋2 , 0) to ( 𝜋2 , 0) and the line segment
from ( 𝜋2 , 0) to (− 𝜋2 , 0).

3 (Ex 16.4, Q 25).

(a). If 𝐶 is the line segment joining (𝑥1 , 𝑦1 ) to (𝑥2 , 𝑦2 ), show that

∫ 𝑥 d𝑦 − 𝑦 d𝑥 = 𝑥1 𝑦1 − 𝑥2 𝑦2 .
𝐶

(b). If the vertices of a polygon, in counterclockwise order, are (𝑥1 , 𝑦1 ), (𝑥2 , 𝑦2 ), … , (𝑥𝑛 , 𝑦𝑛 ),
show that the area of the polygon is
1
𝐴 = [(𝑥1 𝑦2 − 𝑥2 𝑦1 ) + (𝑥2 𝑦3 − 𝑥3 𝑦2 ) + ⋯ + (𝑥𝑛−1 𝑦𝑛 − 𝑥𝑛 𝑦𝑛−1 ) + (𝑥𝑛 𝑦1 − 𝑥1 𝑦𝑛 )].
2

4 (Ex 16.4 Q 35). Suppose 𝑅, 𝑆 both satisfies the conditions in Green’s theorem.
𝑥 = 𝑥(𝑢, 𝑣) and 𝑦 = 𝑦(𝑢, 𝑣). 𝑂 is an open set containing 𝑆 and 𝜕𝑆 which 𝑥, 𝑦 ∈ 𝐶 2 (𝑂). Show
that
𝜕𝑦 𝜕𝑦
∬ d𝑥 d𝑦 = ∫ 𝑥 𝜕𝑢 d𝑥 + 𝑥 𝜕𝑣 d𝑦 = ∬ det(𝐽) d𝑢 d𝑣 ,
𝑅 𝜕𝑆 𝑆
where 𝐽 is the Jacobian matrix

𝜕(𝑥, 𝑦) 𝜕𝑥/𝜕𝑢 𝜕𝑥/𝜕𝑣

𝐽= =[ ].
𝜕(𝑢, 𝑣) 𝜕𝑦/𝜕𝑢 𝜕𝑦/𝜕𝑣

5 (Ex 16.5, Q 2). Find the curl and divergence of

#»
𝐹(𝑥, 𝑦, 𝑧) = 𝑥3 𝑦𝑧2 ȷ ⃗ + 𝑦4 𝑧3 k.⃗

6 (Ex 16.5, Q 18). Show that the vector field

#»
𝐹(𝑥, 𝑦, 𝑧) = 𝑦𝑧 sin 𝑥𝑦 ı ⃗ + 𝑥𝑧 sin 𝑥𝑦 ȷ ⃗ − cos 𝑥𝑦 k⃗
#»
is conservative, and find 𝑓 such that ∇𝑓 = 𝐹

3
 #»
7 (Ex 16.5, Q 27,28). Let 𝑓 ∶ ℝ3 → ℝ3 and 𝐹 ∶ ℝ3 → ℝ3 . For
#» #»
𝐹 = ⟨𝐹1 , 𝐹2 , 𝐹3 ⟩, and 𝑓 𝐹 = ⟨𝑓𝐹1 , 𝑓𝐹2 , 𝑓𝐹3 ⟩,

show that
#» #» #»
(a). ∇ ⋅ (𝑓 𝐹 ) = 𝑓 ∇ ⋅ 𝐹 + 𝐹 ⋅ ∇𝑓.
#» #» #»
(b). ∇ × (𝑓 𝐹 ) = 𝑓 ∇ × 𝐹 + (∇𝑓) × 𝐹 .

8 (Ex 16.5 Q 35). Suppose all derivatives are continuous and 𝐷 satisfies the require-
ments of Green’s theorem. Using Green’s theorem, prove the integration by parts formula
#» #» #» #»
∬ 𝑓 (∇ ⋅ 𝐹 ) d𝐴 = ∮ (𝑓 𝐹 ) ⋅ 𝑛 d𝑠 − ∬ 𝐹 ⋅ ∇𝑓 d𝐴 ,
𝐷 𝜕𝐷 𝐷
#» is the unit outer normal of 𝐷 and 𝜕𝐷 is the boundary of 𝐷.
where 𝑛
Also, deduce Green’s first identity

#»
∬ 𝑓Δ𝑔 d𝐴 = ∮ 𝑓(∇𝑔) ⋅ 𝑛 d𝑠 − ∬ ∇𝑓 ⋅ ∇𝑔 d𝐴 ,
𝐷 𝜕𝐷 𝐷

where Δ is the Laplacian operator, defined, for 𝐺 = ⟨𝐺1 , 𝐺2 ⟩, as

𝜕2 𝑔1 𝜕2 𝑔2
Δ𝑔 = + .
𝜕𝑥2 𝜕𝑦2

4
1 Solutions
1. Green’s theorem asserts that for 𝐶 positively-oriented, piecewise smooth, simple, closed
curve in ℝ2 and 𝐷 be the region bounded by 𝐶. If 𝑃, 𝑄 ∈ 𝐶 1 in an open region containing 𝐷,
then
𝜕𝑄 𝜕𝑃
∮ 𝑃 d𝑥 + 𝑄 d𝑦 = ∬ 𝜕𝑥 − 𝜕𝑦 d𝐴 .
𝐶 𝐷
Thus, we are require to show that the two sides above are equal.
Let us evaluate the line integral first. As the given region is a triangle, it has to be split to its
three edges. Suppose we start at (0, 0), and go counterclockwise along the triangle, then
𝐶1 ∶ #»
𝑟 (𝑡) = (𝑡, 0), 0 ≤ 𝑡 ≤ 2,
#»
𝑟 ′ (𝑡) = ⟨1, 0⟩.
𝐶2 ∶ #»
𝑟 (𝑡) = (1 − 𝑡)⟨2, 0⟩ + 𝑡⟨1, 1⟩ = ⟨2 − 𝑡, 𝑡⟩, 0 ≤ 𝑡 ≤ 1,
#»
𝑟 ′ (𝑡) = ⟨−1, 1⟩.
𝐶3 ∶ #»
𝑟 (𝑡) = (1 − 𝑡)(1, 1) = ⟨1 − 𝑡, 1 − 𝑡⟩, 0 ≤ 𝑡 ≤ 1,
#»
𝑟 ′ (𝑡) = ⟨−1, −1⟩.
For
#»
𝐹(𝑥, 𝑦) = ⟨𝑥2 − 𝑦2 , 2𝑥𝑦⟩,
we have
2
#» #» 1 |2 8
∫ 𝐹 ⋅ d 𝑟 = ∫ ⟨𝑡 2 , 0⟩ ⋅ ⟨1, 0⟩ d𝑡 = 3 𝑡 3 || = 3 .
𝐶1 0 0
1
#» #»
∫ 𝐹 ⋅ d 𝑟 = ∫ ⟨(2 − 𝑡)2 − 𝑡 2 , 2𝑡(2 − 𝑡)⟩ ⋅ ⟨−1, 1⟩ d𝑡
𝐶2 0
1
= ∫ 4𝑡 − 4 + 4𝑡 − 2𝑡 2 d𝑡
0
1
2
= [− 𝑡 3 + 4𝑡 2 − 4𝑡]
3 0
2
=− .
3
1
#» #»
∫ 𝐹 ⋅ d 𝑟 = ∫ ⟨0, 2(1 − 𝑡)2 ⟩ ⋅ ⟨−1, −1⟩ d𝑡
𝐶3 0
1
= −2 ∫ 1 − 2𝑡 + 𝑡 2 d𝑡
0
1 3 1
= −2[ 𝑡 − 𝑡2 + 𝑡
3 ]0
2
=− .
3
Therefore
#» #» #» #» #» #» #» #» 8 2 2 4
∮ 𝐹 ⋅ d 𝑟 = ∫ 𝐹 ⋅ d 𝑟 + ∫ 𝐹 ⋅ d 𝑟 + ∫ 𝐹 ⋅ d 𝑟 = 3 − 3 − 3 = 3.
𝐶 𝐶1 𝐶2 𝐶3

Next we evaluate the double integral. By sketching out the domain, we can easily identify it
as
𝐷 = {(𝑥, 𝑦) ∶ 𝑦 ≤ 𝑥 ≤ 2 − 𝑦, 0 ≤ 𝑥 ≤ 1}.

5
Because
𝜕𝑄 𝜕 𝜕𝑃 𝜕
= (2𝑥𝑦) = −2𝑦, = (𝑥2 − 𝑦2 ) = 2𝑦,
𝜕𝑥 𝜕𝑥 𝜕𝑦 𝜕𝑦
the required integral is
1 2−𝑦
𝜕𝑄 𝜕𝑃
∬ 𝜕𝑥 − 𝜕𝑦 d𝐴 = ∫ ∫ 2𝑦 − (−2𝑦) d𝑥 d𝑦
𝐷 0 𝑦
1
= 4 ∫ 𝑦(2 − 𝑦 − 𝑦) d𝑦
0
1
= 8 ∫ 𝑦 − 𝑦2 d𝑦
0
1 1 1
= 8 [ 𝑦2 − 𝑡 3 ]
2 3 0
4
= .
3
Now we see that the two sides give the same result.

2. By sketching out the integration path, we can see that 𝐶 is piecewise smooth, simply
closed, but negatively oriented. So if we are to apply Green’s theorem, a minus sign should
#»
be added to the double integral. Obviously 𝑃, 𝑄 ∈ 𝐶 1 for 𝐹 = ⟨𝑃, 𝑄⟩, so Green’s theorem
indeed applies, yielding
#» #» 𝜕𝑄 𝜕𝑃
∮ 𝐹 ⋅ d 𝑟 = − ∬ 𝜕𝑥 − 𝜕𝑦 d𝐴 .
𝐶 𝐷

In this problem,
𝑃 = 𝑒 −𝑥 + 𝑦2 and 𝑄 = 𝑒 −𝑦 + 𝑥2
𝜕𝑃 𝜕𝑄
⟹ = 2𝑦 and = 2𝑥.
𝜕𝑦 𝜕𝑥
The integration domain is
𝜋 𝜋
− ≤𝑥≤ , 0 ≤ 𝑦 ≤ cos 𝑥.
2 2
Therefore
#» #» 𝜕𝑄 𝜕𝑃
∮ 𝐹 ⋅ d 𝑟 = − ∬ 𝜕𝑥 − 𝜕𝑦 d𝐴
𝐶 𝐷
𝜋/2 cos 𝑥
= −∫ ∫ 2𝑥 − 2𝑦 d𝑦 d𝑥
−𝜋/2 0
𝜋/2
cos 𝑥
= ∫ [𝑦2 − 2𝑥𝑦]0 d𝑥
−𝜋/2
𝜋/2
= ∫ cos2 𝑥 − 2𝑥 cos 𝑥 d𝑥
−𝜋/2
𝜋/2
1 + cos 2𝑥
=∫ d𝑥 (2𝑥 cos 𝑥 is odd)
−𝜋/2 2
𝑥 sin 2𝑥 𝜋/2
=[ +
2 4 ]−𝜋/2
𝜋
= .
2

6
 𝑦
𝑦 = cos 𝑥

𝑥
− 𝜋2 0
𝜋
2

Figure 1: Integration paths

Remark 1. Here we attempt to evaluate the line integral directly. Along the first path,
𝑦 = cos 𝑥, even though not given explicitly, we can use 𝑥 as its parameter, i.e.
𝜋 𝜋
𝐶1 ∶ #»
𝑟 (𝑡) = ⟨𝑡, cos 𝑡⟩, − ≤ 𝑡 ≤ ,
2 2
#»
𝑟 ′ (𝑡) = ⟨1, − sin 𝑡⟩.

Then the line integral along 𝐶1 is

𝜋/2
#» #»
∫ 𝐹 ⋅ d 𝑟 = ∫ ⟨𝑒 −𝑡 + cos2 𝑡, 𝑒 − cos 𝑡 + 𝑡 2 ⟩ ⋅ ⟨1, − sin 𝑡⟩ d𝑡
𝐶1 −𝜋/2
𝜋/2
=∫ 𝑒 −𝑡 + cos2 𝑡 − 𝑒 − cos 𝑡 sin 𝑡 − 𝑡 2 sin 𝑡 d𝑡
−𝜋/2
𝜋/2
1 + cos 2𝑡
=∫ 𝑒 −𝑡 + d𝑡 (𝑒 − cos 𝑡 sin 𝑡, 𝑡 2 sin 𝑡 are odd)
−𝜋/2 2
−𝑡 𝑥 sin 2𝑥 𝜋/2
= [−𝑒 + +
2 4 ]−𝜋/2
𝜋
= + 𝑒 𝜋/2 − 𝑒 −𝜋/2 .
2

For 𝐶2 , the path from ( 𝜋2 , 0) to (− 𝜋2 , 0), one simple parametrization would be

𝜋 𝜋
𝐶2 ∶ #»
𝑟 (𝑡) = ⟨−𝑡, 0⟩, − ≤𝑡≤ .
2 2
So
𝜋/2
#» #»
∫ 𝐹 ⋅ d 𝑟 = ∫ ⟨𝑒 𝑡 , 1 + 𝑡 2 ⟩ ⋅ −1, 0 d𝑡
𝐶2 −𝜋/2
𝜋/2
= −∫ 𝑒 𝑡 d𝑡
−𝜋/2
𝜋/2
= −[𝑒 𝑡 ]−𝜋/2
= 𝑒 −𝜋/2 − 𝑒 𝜋/2 .

7
Combining the two paths,
#» #» #» #» #» #» 𝜋 𝜋
∫ 𝐹 ⋅ d 𝑟 = ∫ 𝐹 ⋅ d 𝑟 + ∫ 𝐹 ⋅ d 𝑟 = 2 + 𝑒 𝜋/2 − 𝑒 −𝜋/2 + 𝑒 −𝜋/2 − 𝑒 𝜋/2 = 2 .
𝐶 𝐶1 𝐶2

This result is consistent with the one obtained using Green’s theorem, and you appreciate
how much shorter the solution is using Green’s theorem.

Remark 2. In this problem, Green’s theorem is used to translate a closed-loop integral into
an area integral. Note that in the reverse scenarios, i.e. translating from a double integral
into a line integral, could also happen.

3. (a). The integration path is a line segment of two points, so we can parametrize it as

𝐶 ∶ #»
𝑟 (𝑡) = (1 − 𝑡)⟨𝑥1 , 𝑦1 ⟩ − 𝑡⟨𝑥2 , 𝑦2 ⟩
= ⟨𝑥1 + 𝑡(𝑥2 − 𝑥1 ), 𝑦1 + 𝑡(𝑦2 − 𝑦1 )⟩, 0 ≤ 𝑡 ≤ 1,
#» ′
𝑟 (𝑡) = ⟨𝑥2 − 𝑥1 , 𝑦2 − 𝑦1 ⟩.

Then
1
∫ 𝑥 d𝑦 − 𝑦 d𝑥 = ∫ 𝑥 ⋅ 𝑦′ (𝑡) − 𝑦 ⋅ 𝑥′ (𝑡) d𝑡
𝐶 0
1
= ∫ [𝑥1 + 𝑡(𝑥2 − 𝑥1 )](𝑦2 − 𝑦1 ) − [𝑦1 + 𝑡(𝑦2 − 𝑦1 )](𝑥2 − 𝑥1 ) d𝑡
0
𝑥2 − 𝑥1 𝑦 −𝑦
= [𝑥1 + ] (𝑦2 − 𝑦1 ) − [𝑦1 + 2 1 ] (𝑥2 − 𝑥1 )
2 2
1
= [(𝑥1 + 𝑥2 )(𝑦2 − 𝑦1 ) − (𝑦1 + 𝑦2 )(𝑥2 − 𝑥1 )]
2
= 𝑥1 𝑦2 − 𝑥2 𝑦1 .

(b). The key idea is that as the vertices are ordered in counterclockwise order, the edges
forms a piecewise smooth, simply closed, positively-oriented path. So we can use Green’s
theorem to claim that
𝜕𝑄 𝜕𝑃 #» #»
∬ 𝜕𝑥 − 𝜕𝑦 d𝐴 = ∮ 𝐹 ⋅ d 𝑟 ,
𝐷 𝐶
#»
for some appropriate vector field 𝐹 = ⟨𝑃, 𝑄⟩. This is not the desired form yet. To find the
area we want to calculate
∬ d𝐴 ,
𝐷
so we want
𝜕𝑄 𝜕𝑃
− ≡ 1.
𝜕𝑥 𝜕𝑦
One possible choice of such a vector field is
#» 𝑦 𝑥
𝐹 = ⟨− , ⟩.
2 2
#»
With this 𝐹 ,
#» #» 1 1
∫ 𝐹 ⋅ d 𝑟 = 2 ∫ (−𝑦) d𝑥 + 𝑥 d𝑦 = 2 ∫ 𝑥 d𝑦 − 𝑦 d𝑥 ,
𝐶 𝐶 𝐶

8
which is the result from (a) divided by 2.
By denoting 𝐶𝑖 to be the edge joining (𝑥𝑖 , 𝑦𝑖 ) and (𝑥𝑖+1 , 𝑦𝑖+1 ) for 𝑖 = 1, 2, … , 𝑛 − 1 and 𝐶𝑛 be
the edge joining (𝑥𝑛 , 𝑦𝑛 ) and (𝑥1 , 𝑦1 ), using the results from the above discussion,
#» 𝑛 #» 1 𝑛−1
𝐴 = ∬ d𝐴 = ∮ 𝐹 ⋅ d #»
𝑟 = ∑ ∫ 𝐹 ⋅ d #»
𝑟 = [ ∑ (𝑥𝑖 𝑦𝑖+1 − 𝑥𝑖+1 𝑦𝑖 ) + 𝑥𝑛 𝑦1 − 𝑥1 𝑦𝑛 ].
𝐷 𝐶 𝑖=1 𝐶𝑖 2 𝑖=1

This formula could also be written in the following form

1 𝑛−1 𝑥 𝑦𝑖 𝑥 𝑦
𝐴 = ( ∑ det [ 𝑖 ] + det [ 𝑛 𝑛 ]).
2 𝑖=1 𝑥𝑖+1 𝑦𝑖+1 𝑥1 𝑦1

4. We first show the left equality. To use Green’s theorem, we want to find a vector field
⟨𝑃, 𝑄⟩ such that
𝜕𝑄 𝜕𝑃
− = 1.
𝜕𝑥 𝜕𝑦
In the previous problem, we have chosen ⟨𝑃, 𝑄⟩ = ⟨−𝑦/2, 𝑥/2⟩. This time we take ⟨𝑃, 𝑄⟩ =
⟨0, 𝑥⟩. Then, by Green’s theorem

∬ d𝑥 d𝑦 = ∮ ⟨0, 𝑥⟩ ⋅ ⟨𝑥′ (𝑡) 𝑦′ (𝑡)⟩ d𝑡

𝑅 𝜕𝑅

= ∮ 𝑥𝑦′ (𝑡) d𝑡
𝜕𝑅
𝜕𝑦 d𝑢 𝜕𝑦 d𝑣
= ∮ 𝑥( + ) d𝑡 (1)
𝜕𝑆 𝜕𝑢 d𝑡 𝜕𝑣 d𝑡
𝜕𝑦 𝜕𝑦
=∮ 𝑥 d𝑢 + 𝑥 d𝑣 .
𝜕𝑆 𝜕𝑢 𝜕𝑣
Note that in Eq. (1), although we changed the integration path from boundary 𝜕𝑅 to 𝜕𝑆, the
range of the underlying variable 𝑡 is not changed.
The right equality can be obtained by invoking Green’s theorem again. If we consider
𝜕𝑦 𝜕𝑦
𝑃 =𝑥 , 𝑄 =𝑥
𝜕𝑢 𝜕𝑣
as in the usual Green’s theorem, then
𝜕𝑦 𝜕𝑦 𝜕 𝜕𝑦 𝜕 𝜕𝑦
∮ 𝑥 𝜕𝑢 d𝑢 + 𝑥 𝜕𝑣 d𝑣 = ∬ 𝜕𝑢 (𝑥 𝜕𝑣 ) − 𝜕𝑣 (𝑥 𝜕𝑢 ) d𝑢 d𝑣
𝜕𝑆 𝑆
𝜕𝑥 𝜕𝑦 𝜕2 𝑦 𝜕𝑥 𝜕𝑦 𝜕2 𝑦
=∬ + − − d𝑢 d𝑣
𝑆 𝜕𝑢 𝜕𝑣 𝜕𝑢𝜕𝑣 𝜕𝑣 𝜕𝑢 𝜕𝑣𝜕𝑢
𝜕𝑥 𝜕𝑦 𝜕𝑥 𝜕𝑦
=∬ − d𝑢 d𝑣 (2)
𝑆 𝜕𝑢 𝜕𝑣 𝜕𝑣 𝜕𝑢
= ∬ det(𝐽) d𝑢 d𝑣 ,
𝑆

where in Eq. (1), the assumption 𝑥, 𝑦 ∈ 𝐶 2 and the mixed partials theorem is used to conclude
that
𝜕2 𝑦 𝜕2 𝑦
= .
𝜕𝑢𝜕𝑣 𝜕𝑣𝜕𝑢

9
Remark 3. From this problem, one can see that the sign of an “area” is determined by the
orientation of its boundary. To compute the unsigned area of a domain as we did in double
and tripel integrals, it is necessary to take absolute value on det 𝐽.

5. We compute the curl first.

⎡ ı⃗ ȷ⃗ k⃗ ⎤
#» 𝜕 𝜕 𝜕
∇ × 𝐹 = det ⎢ 𝜕𝑥 𝜕𝑦 𝜕𝑧 ⎥
⎣ 0 𝑥3 𝑦𝑧2 𝑦4 𝑧3 ⎦
= ⟨3𝑦3 𝑧3 − 2𝑥3 𝑦𝑧, 0, 3𝑥2 𝑦𝑧3 ⟩.

Next, the divergence is

#» 𝜕 𝜕 3 2 𝜕
∇⋅ 𝐹 = (0) + (𝑥 𝑦𝑧 ) + (𝑦4 𝑧3 ) = 𝑥3 𝑧2 + 3𝑦4 𝑧2 = (𝑥3 + 3𝑦4 )𝑧2 .
𝜕𝑥 𝜕𝑦 𝜕𝑧

#»
(a) Vector field 𝐹 (b) With curl
#»
Figure 2: Vector field 𝐹 = 𝑥3 𝑦𝑧 k⃗ + 𝑦4 𝑧3 k⃗ at 𝑧 = 1.

#» #» #»
6. To check if a ℝ3 vector field 𝐹 is conservative, we check test if ∇ × 𝐹 = 0 . By direct
computation,

⎡ ı⃗ ȷ⃗ k⃗ ⎤
#» 𝜕 𝜕 𝜕
∇ × 𝐹 = det ⎢ 𝜕𝑥 𝜕𝑦 𝜕𝑧 ⎥
⎣𝑦𝑧 sin 𝑥𝑦 𝑥𝑧 sin 𝑥𝑦 − cos 𝑥𝑦⎦
= ⟨𝑥 sin 𝑥𝑦 − 𝑥 sin 𝑥𝑦, 𝑦 sin 𝑥𝑦 − 𝑦 sin 𝑥𝑦, 𝑧 sin 𝑥𝑦 − 𝑧 sin 𝑥𝑦⟩
#»
= 0.
#» #»
Moreover, 𝐹 is defined on the whole ℝ3 and has continuous partial derivatives, thus 𝐹 is
conservative.
The technique for finding the potential function is the same as the ℝ2 case, meaning we inte-
grate some of the components, then differentiate it to compare with the other components.
#»
Suppose 𝑓 satisfies ∇𝑓 = 𝐹 , then

𝑓𝑥 = 𝑦𝑧 sin 𝑥𝑦, 𝑓𝑦 = 𝑥𝑧 sin 𝑥𝑦, 𝑓𝑧 = − cos 𝑥𝑦.

10
Start by integrating 𝑓𝑧 w.r.t. 𝑧, giving us

𝑓 = −𝑧 cos 𝑥𝑦 + 𝑔(𝑥, 𝑦).

Again, integrating w.r.t. 𝑧 introduces a function of 𝑥 and 𝑦, since it vanishes when 𝑓 is

differentiated w.r.t. 𝑧.
Then differentiating 𝑓 w.r.t. 𝑥 and comparing with 𝑓𝑥 yields

𝜕𝑔
𝑦𝑧 sin 𝑥𝑦 + = 𝑦𝑧 sin 𝑥𝑦
𝜕𝑥
𝜕𝑔
⟹ =0
𝜕𝑥
⟹ 𝑔(𝑥, 𝑦) = ℎ(𝑦).

Finally, differentiate 𝑓 w.r.t. 𝑦 to obtain

𝑥𝑧 sin 𝑥𝑦 + ℎ′ (𝑦) = 𝑥𝑧 sin 𝑥𝑦

⟹ ℎ′ (𝑦) = 0
⟹ ℎ(𝑦) = 𝐶 , 𝐶 constant.

Hence any 𝑓 of the form

𝑓(𝑥, 𝑦, 𝑧) = −𝑧 cos 𝑥𝑦 + 𝐶
#»
satisfies ∇𝑓 = 𝐹 . In particular, we can choose 𝐶 = 0, i.e.

𝑓(𝑥, 𝑦, 𝑧) = −𝑧 cos 𝑥𝑦.

(a) Projection to 𝑥𝑦 plane (b) Viewed from first octant

#»
Figure 3: Vector field 𝐹 = ⟨𝑦𝑧 sin 𝑥𝑦, 𝑥𝑧 sin 𝑥𝑦, − cos 𝑥𝑦⟩. Blue: 𝑧 = 1; green: 𝑧 = 0.

7. (a). When the formula consists of only divergence, which is a sum of the partial deriva-
tives, it is much easier to consider the individual partial derivatives then adding them
together.
Relabel the coordinates to 𝑥1 , 𝑥2 , 𝑥3 . In details, for 𝑖 = 1, 2, 3,

𝜕(𝑓𝐹𝑖 ) 𝜕𝐹 𝜕𝑓
=𝑓 𝑖 + 𝐹.
𝜕𝑥𝑖 𝜕𝑥𝑖 𝜕𝑥𝑖 𝑖

11
Hence
#» 3 𝜕(𝑓𝐹 ) 3 𝜕𝐹 3 𝜕𝑓 #» #»
𝑖
∇ ⋅ (𝑓 𝐹 ) = ∑ = 𝑓 ∑ 𝑖 + ∑ 𝐹𝑖 = 𝑓∇ ⋅ 𝐹 + 𝐹 ⋅ ∇𝑓.
𝑖=1 𝜕𝑥𝑖 𝑖=1 𝜕𝑥𝑖 𝑖=1 𝜕𝑥𝑖

(b). The formula involving curl is not as easy, since each coordinate depends also on the
other coordinates. The most effective yet tedius way is to really write out the formula of curl.
#»
For this problem, it is easier to first compute the general curl then substitute it with 𝑓 𝐹 . Let
𝐺 = ⟨𝐺1 , 𝐺2 , 𝐺3 ⟩ , then

⎡ ı⃗ ȷ ⃗ k⃗ ⎤
𝜕𝐺 𝜕𝐺 𝜕𝐺 𝜕𝐺 𝜕𝐺 𝜕𝐺
𝜕
∇ × 𝐺 = det ⎢ 𝜕𝑥 𝜕 𝜕
𝜕𝑧 ⎥
= ⟨ 3 − 2 , 1 − 3 , 2 − 1 ⟩.
𝜕𝑦 𝜕𝑦 𝜕𝑧 𝜕𝑧 𝜕𝑥 𝜕𝑥 𝜕𝑦
⎣𝐺1 𝐺2 𝐺3 ⎦

#»
Next, we substitute 𝐺 = 𝑓 𝐹 . By chain rule, we separate it into one vector with partial
#»
derivatives of 𝐹 , and one with partial derivatives of 𝑓, i.e.

#» 𝜕(𝑓𝐹3 ) 𝜕(𝑓𝐹2 ) 𝜕(𝑓𝐹1 ) 𝜕(𝑓𝐹3 ) 𝜕(𝑓𝐹2 ) 𝜕(𝑓𝐹1 )

∇ × (𝑓 𝐹 ) = ⟨ − , − , − ⟩
𝜕𝑦 𝜕𝑧 𝜕𝑧 𝜕𝑥 𝜕𝑥 𝜕𝑦
𝜕𝐹 𝜕𝐹 𝜕𝐹 𝜕𝐹 𝜕𝐹 𝜕𝐹
= 𝑓⟨ 3 − 2 , 1 − 3 , 2 − 1 ⟩
𝜕𝑦 𝜕𝑧 𝜕𝑧 𝜕𝑥 𝜕𝑥 𝜕𝑦
𝜕𝑓 𝜕𝑓 𝜕𝑓 𝜕𝑓 𝜕𝑓 𝜕𝑓
+ ⟨𝐹3 − 𝐹2 , 𝐹1 − 𝐹3 , 𝐹2 − 𝐹1 ⟩.
𝜕𝑦 𝜕𝑧 𝜕𝑧 𝜕𝑥 𝜕𝑥 𝜕𝑦
#»
For the first term, the vector after 𝑓 is precisely ∇× 𝐹 . The second term is less obvious, but it
#» #»
is actually ∇𝑓 × 𝐹 (try replacing the second row in curl by ∇𝑓, and the last row by 𝐹 ). Hence
#» #» #»
∇ × (𝑓 𝐹 ) = 𝑓(∇ × 𝐹 ) + ∇𝑓 × 𝐹.

#»
8. For 𝐹 = ⟨𝑃, 𝑄⟩, one of the forms of Green’s theorem is
#» #» #»
∮ 𝐹 ⋅ 𝑛 d𝑠 = ∬ ∇ ⋅ 𝐹 d𝐴 .
𝐶 𝐷

#» #»
Replacing 𝐹 by 𝑓 𝐹 gives
#» #» #»
∮ (𝑓 𝐹 ) ⋅ 𝑛 d𝑠 = ∬ ∇ ⋅ (𝑓 𝐹 ) d𝐴 .
𝐶 𝐷

We can then use the (a) in the last problem to conclude that
#» #» #»
∮ (𝑓 𝐹 ) ⋅ 𝑛 d𝑠 = ∬ 𝑓(∇ ⋅ 𝑓) d𝐴 + ∬ 𝐹 ⋅ ∇𝑓 d𝐴 ,
𝐶 𝐷 𝐷
#» #» #»
⟹ ∬ 𝑓(∇ ⋅ 𝑓) d𝐴 = ∮ (𝑓 𝐹 ) ⋅ 𝑛 d𝑠 ∬ 𝐹 ⋅ ∇𝑓 d𝐴 .
𝐷 𝐶 𝐷

#»
In addition, because Δ𝑔 = ∇ ⋅ (∇𝑔), replace 𝐹 by ∇𝑔 in the integration by parts formula to
obtain
#»
∬ 𝑓Δ𝑔 d𝐴 = ∮ (𝑓∇𝑔) ⋅ 𝑛 d𝑠 ∬ ∇𝑓 ⋅ ∇𝑔 d𝐴 .
𝐷 𝐶 𝐷

12
2 Demos
The following are interactive plots to some of the problems:
5 . https://www.math3d.org/M5u7lLLIJ
6 . https://www.math3d.org/dFxynlvE1

13