Nanyang Technological University

School of Physical and Mathematical Sciences

Division of Mathematical Sciences

MH1811 Mathematics 2 Tutorial 4

Topics: Global and Local Extrema. Finding Stationary points of f (x, y) and classifying each as local
maximum/minimum or saddle point via discriminant D = fxx fyy − fxy fyx . Lagrange Multiplier
Method. Finding Global Extreme Values.

1. Let f (x, y) = ln(x + y) + x2 − y.

(a) What is the domain of f ?

[Solution] The domain is {(x, y)|x + y > 0}.

(b) Find all stationary point(s) of f and classify each of them.

 
1 1
[Solution] ∇f (x, y) = + 2x, −1
x+y x+y
1 1
Stationary points when ∇f (x, y) = (0, 0), i.e., + 2x = 0 and − 1 = 0. The
x+y x+y
second equation gives x + y = 1. Thus from the first equation, we have x = −1/2. Hence
y = 3/2.
We use second derivatives test :

−1 −1 −1
fxx = 2
+ 2, fxy = fyx = 2
, fyy =
(x + y) (x + y) (x + y)2
At (−1/2, 3/2), we have

fxx (−1/2, 3/2) = −1 + 2 = 1, fxy = fyx = −1, fyy = −1,

and
D = fxx fyy − fxy fyx = 1(−1) − 1 = −2 < 0.
The point (−1/2, 3/2) is a saddle point.

.............................................................................................

1
2. Locate all critical points of f and classify whether they are local (relative) maximum, local
(relative) minimum, and saddle points.

(a) f (x, y) = x4 + y 4 − 4xy + 7

[Solution] Note that fx = 4x3 − 4y = 4(x3 − y), fy = 4y 3 − 4x = 4(y 3 − x).
Thus there is no singular point. We proceed to find stationary points, i.e., both fx = 0
and fy = 0.
That means we solve x3 − y = 0 and y 3 − x = 0.
Substituting y = x3 into x = y 3 gives y = y 9 , i.e., y 9 − y = 0. Note that

y 9 − y = y(y 8 − 1) = y(y 4 − 1)(y 4 + 1) = y(y 2 − 1)(y 2 + 1)(y 4 + 1)

= y(y − 1)(y + 1)(y 2 + 1)(y 4 + 1).

Thus, we have y = 0, y = 1 or y = −1, and the corresponding values for x are x = 0,

x = 1 or x = −1.
Stationary point are: (0, 0), (1, 1) and (−1, −1).
To classify them, we compute second derivatives:
fxx = 12x2 , fxy = −4, fyx = −4, fyy = 12y 2
to evaluate
D = fxx fyy − fxy fyx = 144x2 y 2 − 16.

At (0, 0), D(0, 0) = −16 < 0. This says that (0, 0) is a saddle point.

At (1, 1) and (−1, −1), both D(1, 1) = D(−1, −1) > 0, and fxx (1, 1) = f (−1, −1) = 12 >
0. Therefore, both points (1, 1) and (−1, −1) are local minimum.

(b) f (x, y) = ex cos y

[Solution] Note that fx = ex cos y, fy = −ex sin y.
Thus there is no singular point. We proceed to find stationary points, i.e., both fx = 0
and fy = 0.
That means we solve ex cos y = 0 and ex sin y = 0. Since ex > 0, we have cos y = 0 and
sin y = 0.
Since cos2 y + sin2 y = 1, there is no real number y at which both cos y = 0 and sin y = 0.
There is no critical point.

.............................................................................................

2
3. Let f (x, y) = (2x − x2 )(y 2 − 4y) and S = (x, y) : x2 + y 2 > 4 . Classify all stationary points


of f in the region S as local maximum, local minimum or saddle points.

[Solution]
(2 − 2x)(y 2 − 4y), (2x − x2 )(2y − 4)


∇f (x, y) =

fxx (x, y) = −2(y 2 − 4y) = −2y(y − 4),

fxy (x, y) = fyx (x, y) = (2 − 2x)(2y − 4) = 4(1 − x)(y − 2),
fyy (x, y) = 2x(2 − x)

D(x, y) = −4 xy(y − 4)(2 − x) + 4(1 − x)2 (y − 2)2

Stationary points: ∇f (x, y) = 0, which is equivalent to

(2 − 2x)(y 2 − 4y) = 0 and (2x − x2 )(2y − 4) = 0

(x = 1 or y = 0 or y = 4) and (x = 0 or x = 2 or y = 2)

Stationary points are

(1, 2) ∈ S: D(1, 2) = −4 (2(−2)(1)) > 0 and fxx (1, 2) = −2(2)(2 − 4) > 0. Local Minimum.

(0, 0) 6∈ S

(2, 0) 6∈ S

(0, 4) ∈ S: D(0, 4) = −4 4(2)2 < 0. Saddle Point.

(2, 4) ∈ S: D(2, 4) = −4 4(22 ) < 0. Saddle Point.

.............................................................................................

4. Find the global (absolute) maximum and minimum of f (x, y) = xy − x − 3y on the triangular
region R with vertices (0, 0), (0, 4), and (5, 0).
(Hint: Locate all critical points of f in the interior of the region, and locate points on boundary
where f has local extrema. Lastly compare values of f .)
[Solution] Note that the triangular region R is closed and bounded, and f is continuous. Thus,
the function f has global (absolute) maximum and minimum on this region.
To find the global (absolute) extrema of a function f on a region D, we proceed as follows:

(a) Find all critical points (singular and stationary points) of f in the interior of D.
(b) Find all possible candidate for extreme values of f on the boundary of D.
(c) Compute the value of f at each point obtained in Steps 1 and 2.
The largest value is the global (absolute) maximum value while the smallest value is the
global (absolute) minimum value.

3
Step 1. To find critical points (singular and stationary points) of f in the interior of the
triangular region R.
fx = y − 1 and fy = x − 3.
Therefore, there is no singular point and (3, 1) is the only stationary point and it is in the
interior of R the critical point and f (3, 1) = 3 − 3 − 3 = −3.

Step 2. The boundary of the triangular region may be divided into three segments L1 ,L2 and
L3 .

On L1 : along y = 0 where 0 ≤ x ≤ 5, we have f (x, y) = f (x, 0) = −x. Note that f (0, 0) = 0

is a maximum value and f (5, 0) = −5 is a minimum value.

On L2 : along x = 0 we have 0 ≤ y ≤ 4 and f (x, y) = f (0, y) = −3y. Note that, f (0, 0) = 0 is

a maximum value and f (0, 4) = −12 is a minimum value.

On L3 : note that y = 4 − 54 x where 0 ≤ x ≤ 5 and

   
4 4 4 27
f (x, y) = xy − x − 3y = x 4 − x − x − 3 4 − x = − x2 + x − 12.
5 5 5 5

To find extrema of g(x) = − 45 x2 + 27

5 x − 12 on [0, 5], we use the closed interval method in
Mathematics 1 as follows:
27
g 0 (x) = − 58 x + 27
5 , which has a stationary point at x = .
8
27
Comparing the value g( ) = − 231
80 with values of g at end-points, g(0) = −12 and g(5) = −20+
8
27 − 12 = −5, we conclude that f 27 13 231


8 , 10 = − 80 is the maximum value and f (0, 4) = −12 is
the minimum value on L3 .

Step 3: From Steps 1 and 2, the absolute (global) maximum is f (0, 0) = 0 and the absolute
(global) minimum is f (0, 4) = −12 in the triangular region.

.............................................................................................

4
5. Use Lagrange Multipliers to find the maximum volume of a rectangular box with three faces
on the positive coordinate planes (i.e., xy-, xz-, yz-planes and x, y, z ≥ 0) and a vertex on the
plane x + y + z = 1.
[Solution] The problem is equivalent to maximizing V (x, y, z) = xyz subject to the constraints:
x + y + z = 1, x ≥ 0, y ≥ 0, and z ≥ 0.
Note that the constrained set is a closed and bounded triangle. Thus global maximum of
V (x, y, z) = xyz exists.

On the boundary of the triangle, we have x = 0 or y = 0 or z = 0 so that V = 0.

In the interior of the triangle, we shall use the method of Lagrange Multipliers with g (x, y, z) =
x + y + z.
We shall solve for x, y, z and λ such that ∇V = λ∇g and g (x, y, z) = 1. This gives
 

 Vx = λgx 
 yz =λ − − − − (1)
Vy = λgy xz =λ − − − − (2)
 
which become
 Vz
 = λgz 
 xy =λ − − − − (3)
x+y+z =1 x+y+z =1 − − − − (4)
 

Multiply (1) by x, (2) by y, and (3) by z, we have

xyz = λx (5)
xyz = λy (6)
xyz = λz (7)

Thus
λx = λy = λz.

As the points are in the interior of the triangle, we have xyz > 0 so that λ 6= 0 (from (1), (2),
(3)).
Since λ 6= 0, using (1) = (2) = (3), we have x = y = z.
Therefore by Equation (4),
1
x=y=z=
3
1 1 1 1


Hence we conclude that Vmax = V 3, 3, 3 = 27 .
.............................................................................................

5
6. A cardboard box without a lid is to have a volume of 32,000 cm3 . Use the method of Lagrange
multipliers to find the dimensions which minimizes the amount of cardboard used. Here, we
assume that the global minimum exists.
[Solution] Let x, y, and z denote the length, the width and the height respectively of the box.
The problem is equivalent to minimizing A (x, y, z) = xy + 2yz + 2xz subject to the constraint
g (x, y, z) = xyz = 32, 000.
Using the method of Lagrange Multipliers, we solve for x, y, z and λ such that ∇A = λ∇g and
g (x, y, z) = 32, 000. This gives


 y + 2z = λyz − − − − (1)
x + 2z = λxz − − − − (2)


 2x + 2y = λxy − − − − (3)
xyz = 32, 000 − − − − (4)


Multiply (1) by x, (2) by y, and (3) by z we have

xy + 2xz = λxyz (5)

xy + 2yz = λxyz (6)
2xz + 2yz = λxyz (7)

Since λ 6= 0 (otherwise x = y = z = 0, which will contradict (4)), we have from (5) and (6),

xy + 2xz = xy + 2yz,

which gives xz = yz. Since z 6= 0 (otherwise V = 0), so x = y. From (6) and (7) we have

xy + 2yz = 2xz + 2yz

which gives xy = 2xz and so y = 2z (since x 6= 0). Thus we have x = y = 2z. Substituting this
into Equation (4) we have
4z 3 = 32, 000
Hence z = 20 and x = y = 40.

.............................................................................................

6
7. Use Lagrange multipliers to discuss how to construct a closed cylindrical metal can with a
given volume V in a way that minimizes the surface area (the amount of metal used). (Assume
that the minimum value exists.)
[Solution] Let r be the radius and h the height and A the surface area of the cylindrical can.
Then
V = πr2 h, and A = 2πr2 + 2πrh.

Our aim is to minimize A = 2πr2 + 2πrh subject to the constraint V = πr2 h, where V is a
given volume. Let g(r, h) = πr2 h.
Using the method of Lagrange Multipliers, we need to look for r > 0, h > 0 and λ such that
∇A = λ∇g and πr2 h = V . This gives

 Ar = λgr ; i.e., 4πr + 2πh = λ(2πrh)
Ah = λgh ; i.e., 2πr = λπr2
πr2 h = V



 2r + h = λrh − − − − (1)

2
which become r = − − − − (2)
 λ
 2
πr h =V − − − − (3)

Substituting (2) into (1), we have

2(2/λ) 4
h= = − − − (4).
2−1 λ
 2  
2 4 16π
Using (3), we have V = π = 3 .
λ λ λ
r
3 16π
Thus, we have λ = .
V
r r
3 V V
From (2) and (4), we have r = and h = 2 3 .
2π 2π
And, the minimum area
r !2
3 V
A = 6π .
2π
.............................................................................................

7
8. A space probe in the shape of the ellipsoid

4x2 + y 2 + 4z 2 = 16

enters Earth’s atmosphere and its surface begins to heat. After one hour, the temperature at
the point (x, y, z) on the probe’s surface is

T (x, y, z) = 8x2 + 4yz − 16z + 600.

Find the hottest point on the probe’s surface.

 
4 4 4
(Answer: ± , − , − )
3 3 3
[Solution] The problem is to find the global (absolute) maximum value of T (x, y, z) = 8x2 +
4yz − 16z + 600 subject to the constraint 4x2 + y 2 + 4z 2 = 16.
The region defined by the constraint is closed and bounded. This problem has a global maxi-
mum.
We use Lagrange Multipliers to solve this problem. Let g(x, y, z) = 4x2 + y 2 + 4z 2
We shall solve x, y, z and λ satisfying ∇T = λ∇g and g(x, y, z) = 16, which are

16x = 8xλ, 4z = 2yλ, 4y − 16 = 8zλ, 4x2 + y 2 + 4z 2 = 16.

Simplifying gives
0 = x(λ − 2) − − − (1)
2z = yλ − − − (2)
y = 4 + 2zλ − − − (3)
4x2 + y 2 + 4z 2 = 16 − − − (4)

From (2) and (3), we have

2λ
2z = λ (4 + 2zλ) , which gives z = . − − − (5).
1 − λ2
Hence,
4
y= . − − − (6).
1 − λ2
In view of equation 1, we shall proceed to discuss the problem on two cases:

• Case 1 λ = 2
−4 −4
From (5) and (6) we have z = and y = .
3 3
4
Substituting into (4) gives x = ± .
3

8
 • Case 2 λ 6= 2.
Then x = 0, and, from (4), we have y 2 + 4z 2 = 16.
Substituting (5)and (6) into the above gives
 2  2
4 2λ
+4 = 16,
1 − λ2 1 − λ2

which gives
1 + λ2 = (1 − λ2 )2 = 1 − 2λ2 + λ4 .
√
From which, we have λ2 (λ2 − 3) = 0. Hence we have, λ = 0 or λ = ± 3.
λ = 0: x = 0, y = 4, z = 0.
√ √
λ = 3: x = 0, y = −2, z = − 3
√ √
λ = − 3: x = 0, y = −2, z = 3

Comparing values of T :

√ √ √ √
T (0, 4, 0) = 600, T (0, −2, 3) = 600 − 24 3, T (0, −2, − 3) = 600 + 24 3,
4 4 4 2
T (± , − , − ) = 642 .
3 3 3 3
4 4 4
Therefore, the hottest point on the probe’s surface is at (± , − , − ).
3 3 3
.............................................................................................

9. (Optional.) Find the point closest to the origin on the line of intersection of the planes y + 2z =
12 and x + y = 6. (Answer: (2, 4, 4))
p
[Solution 1] The distance from a point (x, y, z) to the origin is given by d(x, y, z) = x2 + y 2 + z 2 .
However, the minimum value of d(x, y) occurs at the same point as the function f (x, y, z) =
x2 + y 2 + z 2 . Therefore, we shall find the minimum value of f (x, y).
Our aim is to minimize f (x, y, z) = x2 + y 2 + z 2 subject to two constraints y + 2z = 12 and
x + y = 6. Let g(x, y, z) = y + 2z and h(x, y, z) = x + y.
Using the method of Lagrange Multipliers, we need to look for x, y, z, λ and µ such that
∇f = λ∇g + µ∇h , g(x, y, z) = 12, and h(x, y, z) = 6, which are
 µ
 2x =µ i.e., x = − − − − (1)

 2

 λ+µ
 2y = λ + µ i.e., y = − − − − (2)

2

 2z = 2λ i.e., z = λ − − − − (3)
− − − − (4)



 y + 2z = 12

x+y =6 − − − − (5)

Substituting (2) & (3) into (4):

λ+µ
+ 2λ = 12, i.e., 5λ + µ = 24. − − − −(6)
2

9
Substituting (1) & (2) into (5):

µ λ+µ
+ = 6, i.e., λ + 2µ = 12. − − − −(6)
2 2

Solving (6) & (7), we obtain

λ=4 & µ = 4.
Hence we have,
x = 2, y = 4, z = 4.
√
The value of f (2, 4, 4) = 22 + 42 + 42 = 36. The closet distance is 36 = 6.

[Solution 2]
Note that the two planes intersect on a line. From y + 2z = 12 and x + y = 6, we have
1
z = (12 − y) and x = 6 − z. Thus, the equation of the line is
2
1
x = 6 − t, y = t, z = (12 − t) .
2

Our aim to to minimize f (x, y, z) = x2 + y 2 + z 2 , where

1
x = 6 − t, y = t, z = (12 − t) .
2

Note that the composite function f (x(t), y(t), z(t)) is a single variable function.
 2
2 2 1
f (x(t), y(t), z(t)) = (6 − t) + t + (12 − t) = 72 − 18t + 9t2 /4.
2

df d2 f
We have = −18 + 9t/2, and 2 = 9/2, which is concave upward.
dt dt
df
We find the values of t at which = 0, which gives t = 4. Hence we have x = 2, y = 4, andz =
dt
4.
df
Alternatively, we may use chain rule to find t at which = 0, as follows,
dt
df ∂f dx ∂f dy ∂f dz
= + +
dt ∂x dt ∂y dt ∂z dt
df 1
i.e., = 2x(−1) + 2y(1) + 2z(−1/2) = −2(6 − t) + 2(t) − (12 − t) = −18 + 9t/2.
dt 2
df
Thus, = 0, gives t = 4. Hence x = 2, y = 4 and z = 4.
dt
.............................................................................................

10