Math 2011-Introduction to Multivariable Calculus (Edited by Dr.

Hon-Ming HO)
Practice Exercises 9: Chain Rule for Partial Differentiation

Exercise 1:

𝑑𝑤
Given that 𝑤 = ln|𝑥 2 + 𝑦 2 + 𝑧 2 | , 𝑥 = cos 𝑡 , 𝑦 = sin 𝑡 , 𝑧 = 4√𝑡 . Express as a function of 𝑡 both
𝑑𝑡

1) By using Chain Rule for partial differentiation,

2) By expressing 𝑤 in terms of 𝑡 before differentiating with respect to 𝑡 .

Compare the resulting expressions obtained in part 1) and part 2). Make a conclusion.

Solution to Exercise 1:

Using Chain Rule for partial differentiation: Things to do which using

Dependent variable 𝑤 = ln|𝑥 2 + 𝑦 2 + 𝑧 2 |
chain rule: Identify all independent variables, all intermediate variables
and dependent variables stated in the question. Identify which variables 𝝏𝒘 𝝏𝒘
𝝏𝒘
𝝏𝒙 𝝏𝒛
control and determine which other variables. Draw a tree diagram 𝝏𝒚
𝑥 = cos 𝑡 𝑦 = sin 𝑡 𝑧 = 4√𝑡
showing the relation and dependence among all stated variables. In this
example, the right figure shows the tree diagram for the appropriate 𝒅𝒙 𝒅𝒚
𝒅𝒛
chain rule. Since there are three paths starting from the independent 𝒅𝒕 𝒅𝒕
𝒅𝒕
𝑑𝑤
variable 𝑡 and ending at 𝑤 , the derivative is a sum of three Independent variable 𝑡
𝑑𝑡
products. We have A tree diagram showing the dependence among all stated
𝑑𝑤 𝝏𝒘 𝒅𝒙 𝝏𝒘 𝒅𝒚 𝝏𝒘 𝒅𝒛 variables.
= ∙ + ∙ + ∙ ,
𝑑𝑡 𝝏𝒙 𝒅𝒕 𝝏𝒚 𝒅𝒕 𝝏𝒛 𝒅𝒕
𝝏𝒘 𝝏𝒘 𝝏𝒘 𝟐
= ∙ (− 𝒔𝒊𝒏 𝒕) + ∙ (𝒄𝒐𝒔 𝒕) +
). ∙(
𝝏𝒙 𝝏𝒚 √𝒕 𝝏𝒛
The next operation is to perform is the partial differentiation of 𝑤 with respect to variables 𝑥 , 𝑦 , 𝑧 . Performing partial
differentiation is a kind of pattern recognition, recognizing which partial differentiation rules are to be applied.

𝑤 = ln|𝑥 2 + 𝑦 2 + 𝑧 2 |

(Constant Multiple Rule)

(General power rule)
𝑘♦ ∆
ῧ
(General logarithmic rule)
(General exponential rule)
?
𝒍𝒏 | ῧ | ? ? 𝒆ῧ
▭1
?
(÷ Rule) ▭1 + ▭2 (+ Rule) ?
▭2 ?
? 𝒔𝒊𝒏−𝟏 ῧ 𝐬𝐢𝐧 ῧ (General sine rule)

? ? ?
𝒄𝒐𝒔−𝟏 ῧ ? ? 𝐜𝐨𝐬 ῧ (General cosine rule)
(▭1 ) ∙ (▭2 ) ?
𝐭𝐚𝐧−𝟏 ῧ 𝐭𝐚𝐧 ῧ
(× Rule)

▭1 and ▭2 are expressions in terms of 𝑥 and 𝑦 . ῧ , ῧ , ῧ , ῧ , ῧ , ῧ , ῧ , ῧ , ῧ are expressions in terms of 𝑥 and 𝑦 .

We observe that 𝑤 = ln|𝑥 2 + 𝑦 2 + 𝑧 2 | = 𝒍𝒏 | ῧ | where ῧ = 𝑥 2 + 𝑦 2 + 𝑧 2 . We apply the general logarithmic rule for partial
differentiation. We have

𝝏𝒘 𝝏 𝜕 1 𝜕 2𝑥
= [ ln|𝑥 2 + 𝑦 2 + 𝑧 2 | ] = [ 𝒍𝒏 | ῧ | ] = ∙ [ ῧ ]= 2 ,
𝝏𝒙 𝝏𝒙 𝜕𝑥 ῧ 𝜕𝑥 𝑥 + 𝑦2 + 𝑧2

𝝏𝒘 𝝏 𝜕 1 𝜕 2𝑦
= [ ln|𝑥 2 + 𝑦 2 + 𝑧 2 | ] = [ 𝒍𝒏 | ῧ | ] = ∙ [ ῧ ]= 2 ,
𝝏𝒚 𝝏𝒚 𝜕𝑦 ῧ 𝜕𝑦 𝑥 + 𝑦2 + 𝑧2

Practice Exercises 9 (edited by Dr. Hon-Ming HO)-page 1

 𝝏𝒘 𝝏 𝜕 1 𝜕 2𝑧
= [ ln|𝑥 2 + 𝑦 2 + 𝑧 2 | ] = [ 𝒍𝒏 | ῧ | ] = ∙ [ ῧ ]= 2 .
𝝏𝒛 𝝏𝒛 𝜕𝑧 ῧ 𝜕𝑧 𝑥 + 𝑦2 + 𝑧2

Thus chain rule implies that

𝟐
2𝑧 ( )
𝑑𝑤 𝝏𝒘 𝝏𝒘 𝝏𝒘 𝟐 2𝑥(− sin 𝑡 ) 2𝑦(cos 𝑡) √𝒕
= ∙ (− 𝒔𝒊𝒏 𝒕) + ∙ (𝒄𝒐𝒔 𝒕) + ∙( )= 2 + + ,
𝑑𝑡 𝝏𝒙 𝝏𝒚 𝝏𝒛 √𝒕 𝑥 + 𝑦2 + 𝑧2 𝑥 2 + 𝑦2 + 𝑧2 𝑥 2 + 𝑦2 + 𝑧2
4
−2(cos 𝑡)(sin 𝑡) + 2(sin 𝑡)(cos 𝑡) + (4√𝑡) ( )
√ 𝑡 16
= 2 = .
(cos 𝑡) + (sin 𝑡) + (4√𝑡)
2 2 1 + 16𝑡

Expressing 𝒘 in terms of 𝒕 before performing partial differentiation:

2 𝑑𝑤 16
𝑤 = ln|𝑥 2 + 𝑦 2 + 𝑧 2 | = ln |(cos 𝑡)2 + (sin 𝑡)2 + (4√𝑡) | = ln|1 + 16𝑡| ⟹ = .
𝑑𝑡 1 + 16𝑡

Conclusion: the 2nd way verifies the proposed chain rule for partial differentiation stated in the 1st way. In this example, both ways
can work out the same result.

Exercise 2:

𝜕𝑤 𝜕𝑤
Write down an appropriate chain rule for each of the following partial derivatives: and where 𝑤 = 𝑓(𝑟 , 𝑠 , 𝑡) , 𝑟 =
𝜕𝑥 𝜕𝑦
𝑔(𝑥, 𝑦) , 𝑠 = ℎ(𝑦) , 𝑡 = 𝑘(𝑥) .

Solution to Exercise 2:

Things to do which using chain rule: Identify all independent variables,

Dependent variable 𝑤 = 𝑓(𝑟, 𝑠, 𝑡)
all intermediate variables and dependent variables stated in the
question. Identify which variables control and determine which other 𝝏𝒘 𝝏𝒘
𝝏𝒘
𝝏𝒓 𝝏𝒕
variables. Variables appearing inside pairs of brackets are supposed to be 𝝏𝒔
𝑟 𝝏𝒓 𝑠 𝑡
variables controlling and determining other variables. Draw a tree
diagram showing the relation and dependence among all stated 𝝏𝒚
𝒅𝒕
variables. In this example, the right figure shows the tree diagram for the 𝝏𝒓 𝒅𝒙
𝝏𝒙
appropriate chain rule. Since there are two paths starting from the 𝑥 𝑦
𝜕𝑤
independent variable 𝑥 and ending at 𝑤 , the partial derivative is
𝜕𝑥 A tree diagram showing the dependence among all stated
a sum of two products. variables.

𝜕𝑤 𝝏𝒘 𝝏𝒓 𝝏𝒘 𝒅𝒕 𝜕𝑤 𝝏𝒘 𝝏𝒓 𝝏𝒘 𝒅𝒔
= ∙ + ∙ 𝑎𝑛𝑑 = ∙ + ∙ .
𝜕𝑥 𝝏𝒓 𝝏𝒙 𝝏𝒕 𝒅𝒙 𝜕𝑦 𝝏𝒓 𝝏𝒚 𝝏𝒔 𝒅𝒚

Exercise 3:

𝜕𝑧 𝜕𝑧
Given an equation 𝑧 3 − 𝑥𝑦 + 𝑦𝑧 + 𝑦 3 − 2 = 0 . Find and at point (1 ,1 ,1) .
𝜕𝑥 𝜕𝑦

Solution to Exercise 3:

1) Things to do: Look at the desired partial derivatives. Identify which variables are variables of partial
differentiation. Variables of partial differentiation are supposed to control and determine other variable(s) in the given
𝜕𝑧 𝜕𝑧
equation. In this example, the desired partial derivatives are and . Thus variables 𝒙 and 𝒚 are variables of partial
𝜕𝒙 𝜕𝒚
differentiation.
2) Things to do: Look at the given equation carefully. Ask yourself whether or not it is easy to solve the equation for 𝑧 in terms
of the variables of partial differentiation found at step 1. In this example, is it easy for us to express 𝑧 in terms
of the variables of partial differentiation 𝑥 and 𝑦 ? Answer: it is not easy. Then move to the next step.
3
𝒛𝟑 − 𝑥𝑦 + 𝑦𝒛 + 𝑦 3 − 2 = 0 ⟹ 𝒛 = ⋯ ? ⋯ ⋯ √? ? ? ? ? ? ⋯ ⋯ ⋯ ? ⋯ ⋯ (𝑛𝑜 𝑧 𝑎𝑝𝑝𝑒𝑎𝑟𝑠 𝑜𝑛 𝑡ℎ𝑒 𝑅. 𝐻. 𝑆. . )

Practice Exercises 9 (edited by Dr. Hon-Ming HO)-page 2

3) Things to do: Use implicit partial differentiation to find the desired derivatives. The idea behind the calculations is to treat 𝑧
as a locally (near the stated point) differentiable function of the variables of partial differentiation. In this
example, we treat 𝑧 as a locally differentiable function of 𝑥 and 𝑦 , i.e. 𝑧 = 𝑓(𝑥, 𝑦) near the point (1,1,1) . Although we
do not know the explicit formula for function 𝑓 , it does not really matter. When performing partial differentiation below,
we bear in the mind that 𝑧 = 𝑓(𝑥, 𝑦) holds near the point (1,1,1) .
∵ 𝑧 3 − 𝑥𝑦 + 𝑦𝑧 + 𝑦 3 − 2 = 0 ,
𝜕 3 𝜕
∴ [𝑧 − 𝑥𝑦 + 𝑦𝑧 + 𝑦 3 − 2] = [0] ,
𝜕𝑥 𝜕𝑥
𝜕 3 𝜕 𝜕 𝜕 3 𝜕
∴ [𝑧 ] − [𝑥𝑦] + [𝑦𝑧] + [𝑦 ] − [2] = 0 ,
𝜕𝑥 𝜕𝑥 𝜕𝑥 𝜕𝑥 𝜕𝑥
𝜕𝑧 𝜕𝑧
∴ (3𝑧 2 ) −𝑦+𝑦 +0−0=0 ,
𝜕𝑥 𝜕𝑥
𝜕𝑧
∴ (3𝑧 2 + 𝑦) =𝑦 ,
𝜕𝑥
𝜕𝑧 𝑦
∴ = 2 .
𝜕𝑥 3𝑧 + 𝑦
Similar calculation shows that
𝜕𝑧 𝑥 − 𝑧 − 3𝑦 2
= .
𝜕𝑦 3𝑧 2 + 𝑦

𝜕𝑧 1 𝜕𝑧 −3
Thus we have (1,1,1) = and (1,1,1) = .
𝜕𝑥 4 𝜕𝑦 4

Exercise 4:

The voltage 𝑉 in a circuit that satisfies the law 𝑉 = 𝐼 ∙ 𝑅 is slowly dropping as

the battery wears out. At the same time, the resistance 𝑅 is increasing as the
resistor heats up. Find how the current 𝐼 is changing at the instant when 𝑅 =
𝑑𝑅 𝑑𝑉 current 𝐼
600 𝑜ℎ𝑚𝑠 , 𝐼 = 0.04 𝑎𝑚𝑝 , = 0.5 𝑜ℎ𝑚/𝑠𝑒𝑐 and = −0.01 𝑣𝑜𝑙𝑡/𝑠𝑒𝑐 .
𝑑𝑡 𝑑𝑡

Solution to Exercise 4:

1) Things to do: Identify what changing quantities with time there are. In this example, three quantities, namely voltage, current
and resistance, are changing with time. They can be regarded as three functions of time 𝑡 .
2) Things to do: Identify which variables control and determine which other variables. Draw a tree diagram showing the relation
and dependence among all stated variables. In this example, there are two paths starting from time 𝑡 to voltage 𝑉 . Thus
chain rule implies that
𝑑𝑉 𝜕𝑉 𝑑𝐼 𝜕𝑉 𝑑𝑅
= ∙ + ∙
𝑑𝑡 𝜕𝐼 𝑑𝑡 𝜕𝑅 𝑑𝑡
3) Things to do: Plug all given derivatives into the chain rule formula above. Compute all required partial derivatives, if any.
𝑑𝑉 𝜕𝑉 𝑑𝐼 𝜕𝑉 𝑑𝑅 𝑑𝐼 𝑑𝑅 𝑑𝐼
−0.01 = = ∙ + ∙ =𝑅∙ +𝐼∙ = (600) ∙ + (0.04)(0.5) .
𝑑𝑡 𝜕𝐼 𝑑𝑡 𝜕𝑅 𝑑𝑡 𝑑𝑡 𝑑𝑡 𝑑𝑡
𝑑𝐼 𝑑𝐼
Solve the equation for the unknown above, we have = −0.00005 𝑎𝑚𝑝𝑠/𝑠𝑒𝑐 .
𝑑𝑡 𝑑𝑡

Exercise 5:

Let 𝑇 = 𝑇(𝑥, 𝑦) be the temperature at the point (𝑥, 𝑦) on the unit circle centered at the origin of the 𝑥𝑦-plane and suppose
𝜕𝑇 𝜕𝑇
that
𝜕𝑥
= 8𝑥 − 4𝑦 and
𝜕𝑦
= 8𝑦 − 4𝑥 . Find where the (local) maximum and (local) minimum temperatures on the
circle occur.
Solution to Exercise 5:

Past experience in doing optimization (優化) in the course Calculus I and II (Math 1013 and Math 1014):

We learnt that if we are given a twice differentiable function 𝑦 = 𝑓(𝑥) , then to locate the point at which the function 𝑓 attains
its local maximum and local minimum, we may use the second derivative test for local extrema as follow:

Practice Exercises 9 (edited by Dr. Hon-Ming HO)-page 3

 Second derivative test for local extrema:

Suppose that 𝑓 ′′ is continuous on an open interval that contains 𝑥 = 𝑐 .

1) If 𝑓 ′ (𝑐) = 0 and 𝑓 ′′ (𝑐) < 0 , then function 𝑓 has a local maximum value at point 𝑥 = 𝑐 .
2) If 𝑓 ′ (𝑐) = 0 and 𝑓 ′′ (𝑐) > 0 , then function 𝑓 has a local minimum value at point 𝑥 = 𝑐 .
3) If 𝑓 ′ (𝑐) = 0 and 𝑓 ′′ (𝑐) = 0 , then the test fails. The function 𝑓 may have a local maximum, a local minimum, or neither.

Present situation:

At first glance, the given temperature distribution function 𝑇 = 𝑇(𝑥, 𝑦) has two variables, instead of a single variable. However,
to use the second derivative test above, the testing function 𝑦 = 𝑓(𝑥) must be a function of a single variable. It seems to us that
the second derivative test is not application at first glance. So to apply the second derivative test, we need to transform the
temperature distribution function into a function of a single variable. How? We enter into vectorial perspective.

𝑇(𝑥2 , 𝑦2 ) 𝑇(𝑥1 , 𝑦1 ) 𝑇(𝑟(𝑡2 )) 𝑇(𝑟(𝑡1 ))

(𝑥2 , 𝑦2 ) (𝑥1 , 𝑦1 ) (𝑥(𝑡2 ) , 𝑦(𝑡2 )) (𝑥(𝑡1 ) , 𝑦(𝑡1 ))

Enter into vectorial perspective
𝑟(𝑡2 ) 𝑟(𝑡1 )

𝑟(𝑡3 ) 𝑟(𝑡4 )
(𝑥3 , 𝑦3 ) (𝑥4 , 𝑦4 ) (𝑥(𝑡3 ) , 𝑦(𝑡3 )) (𝑥(𝑡4 ) , 𝑦(𝑡4 ))

𝑇(𝑥3 , 𝑦3 ) 𝑇(𝑥4 , 𝑦4 ) 𝑇(𝑟(𝑡3 )) 𝑇(𝑟(𝑡4 ))

Enter into vectorial perspective

The underlying sense of the red object: The vectorial sense of the red object:
The red object above is the unit circle To transform temperature function 𝑇 into a
centered at the origin with radius 1, where function of a single variable, we treat the red
we do optimization, where we maximize and object above as a curve traced out by a vector
minimize the temperature function 𝑇 . ⃑ (𝒕) .
function 𝒓

There are infinitely many possible formulations for the vector function 𝒓⃑ (𝒕) above. One natural choice is to make use of the
definitions of sine and cosine. We propose the following 𝒓 ⃑ (𝒕) =< cos 𝑡 , sin 𝑡 > where 0 ≤ 𝑡 ≤ 2𝜋 . Then the temperature
distribution function along the unit circle is given by 𝑇 = 𝑇(𝑟 (𝑡)) where 0 ≤ 𝑡 ≤ 2𝜋 . Now we succeed to transform 𝑇 into
a function of a single variable 𝑡 . Then we apply the second derivative test to 𝑇 = 𝑇(𝑟(𝑡)) .

Things to do which using chain rule: Identify all independent variables, all intermediate variables and dependent variables stated
in the question. Identify which variables control and determine which other variables. Variables appearing inside pairs of brackets
are supposed to be variables controlling and determining other variables. Draw a tree diagram showing the relation and
dependence among all stated variables. In this example, we have

𝑑𝑇 𝑑 𝜕𝑇 𝑑𝑥 𝜕𝑇 𝑑𝑦
∵ = [𝑇(𝑟 (𝑡))] = ∙ + ∙ = (8𝑥 − 4𝑦)(− sin 𝑡) + (8𝑦 − 4𝑥)(cos 𝑡) ,
𝑑𝑡 𝑑𝑡 𝜕𝑥 𝑑𝑡 𝜕𝑦 𝑑𝑡
𝑑𝑇
∴ = (8 cos 𝑡 − 4 sin 𝑡)(− sin 𝑡) + (8 sin 𝑡 − 4 cos 𝑡)(cos 𝑡) = 4 𝑠𝑖𝑛2 𝑡 − 4 𝑐𝑜𝑠 2 𝑡 .
𝑑𝑡

Practice Exercises 9 (edited by Dr. Hon-Ming HO)-page 4

 𝑑2𝑇 𝑑𝑇
It implies that = 16 sin 𝑡 cos 𝑡 . Solve equation = 0 for the unknown 𝑡 . We obtain four points within the range [0 ,2𝜋]
𝑑𝑡 2 𝑑𝑡
𝜋 3𝜋 5𝜋 7𝜋 𝑑2 𝑇
which are listed as follows 𝑡 = , , , . Plugging them into the second derivative = 16 sin 𝑡 cos 𝑡 , calculation
4 4 4 4 𝑑𝑡 2
shows that

𝜋 𝜋 √2 √2
1) The distribution function 𝑇 attains a local minimum value at point (𝑥, 𝑦) = (cos , sin ) = ( , ) .
4 4 2 2
3𝜋 3𝜋 √2 √2
2) The distribution function 𝑇 attains a local maximum value at point (𝑥, 𝑦) = (cos , sin ) = (− , ) .
4 4 2 2
5𝜋 5𝜋 √2 √2
3) The distribution function 𝑇 attains a local minimum value at point (𝑥, 𝑦) = (cos , sin ) = (− ,− ) .
4 4 2 2
7𝜋 7𝜋 √2 √2
4) The distribution function 𝑇 attains a local maximum value at point (𝑥, 𝑦) = (cos , sin )=( ,− ) .
4 4 2 2

Practice Exercises 9 (edited by Dr. Hon-Ming HO)-page 5