MA 105 Calculus

Autumn 2011 SRG/AA/VDS/KSK

Solutions and Marking Scheme for the End-Semester Examination

ln x
1. (i) Consider the function f : (0, ∞) → R defined by f (x) = . Find the intervals in (0, ∞)
x
where f is convex and the intervals in (0, ∞) where f is concave. Justify your answers.
[2 marks]
(ii) For each n ∈ N, let fn : [0, 1] → R be the function defined by

k
1 if x = for some integer k with 0 ≤ k ≤ n!,

fn (x) = n!
0 otherwise.

(a) Show that for each n ∈ N, the function fn is Riemann-integrable on [0, 1] and determine
Z 1
fn (x)dx.
0
(b) Determine the values of x ∈ [0, 1] for which the limit lim fn (x) exists. Suppose we
n→∞
denote lim fn (x) by f (x). Then does f define a Riemann-integrable function on [0, 1]?
n→∞
Justify your answers. [3 marks]
Solution: (i) Clearly, f is twice differentiable on (0, ∞) and
1 − ln x 2 ln x − 3
f 0 (x) = and f 00 (x) = for x ∈ (0, ∞). [1]
x2 x3
Hence
3 3
f 00 (x) < 0 when ln x < , i.e., x < e3/2 and f 00 (x) > 0 when ln x > , i.e., x > e3/2 .
2 2
and thus
f is concave on (0, e3/2 ) and convex on (e3/2 , ∞). [1]
(ii) For a fixed n ∈ N, there are only finitely many numbers of the form k/n! where k is
an integer with 0 ≤ k ≤ n!. Thus fn (x) = 0 for all except finitely many values of x and in
particular,
R1 fn has at most finitely many discontinuities. Hence fn is Riemann-integrable
and 0 fn (x)dx = 0 for all n ∈ N. [1]
Now let x ∈ [0, 1]. To determine if lim fn (x) exists, we consider two cases:
n→∞
Case 1: x is rational, say x = p/q where p, q are integers with p ≥ 0, q > 0, and q ≥ p.
In this case, for each n ≥ q, we can write x = k/n! for some integer k with 0 ≤ k ≤ n!.
Hence fn (x) = 1 for all n ≥ q and so in this case lim fn (x) = 1. [1]
n→∞
Case 2: x is irrational.
In this case, fn (x) = 0 for all n ∈ N and so lim fn (x) = 0.
n→∞
It follows that f (x) := lim fn (x) exists for all x ∈ [0, 1] and is given by
n→∞
(
1 if x is rational,
f (x) =
0 if x is irrational.
Thus f is the Dirichlet function, which is not Riemann integrable on [0, 1]. [1]

1
2. (i) Suppose f : R2 → R is such that the first order partial derivatives fx and fy exist and are
nonnegative at every point of R2 . Prove that

f (a, b) ≤ f (a + h, b + k) for all (a, b) ∈ R2 and h, k ∈ (0, ∞). [2 marks]

p
(ii) Suppose f : R2 → R is defined by f (x, y) = |xy|. Is f continuous at (0, 0)? Is f differen-
tiable at (0, 0)? Justify your answers. [3 marks]

Solution: (i) Let (a, b) ∈ R2 and h, k ∈ (0, ∞). We can write

f (a + h, b + k) − f (a, b) = f (a + h, b + k) − f (a, b + k) + f (a, b + k) − f (a, b) [1]

Now, using the Mean Value Theorem [applied to the real-valued functions x 7→ f (x, b + k)
and y 7→ f (a, y) on the intervals [a, a + h] and [b, b + k], respectively), the RHS above can
be written as

hfx (α, b + k) + kfy (a, β) for some α ∈ (a, a + h) and β ∈ (b, b + k). [1]

Since fx ≥ 0 and fy ≥ 0 on R2 and also since h > 0 and k > 0, it follows that

f (a + h, b + k) − f (a, b) ≥ 0, that is, f (a, b) ≤ f (a + h, b + k).

p p
(ii) For any x, y ∈ R, we have |x| ≤ x2 + y 2 and |y| ≤ x2 + y 2 . Hence
p
|f (x, y)| ≤ x2 + y 2 for all (x, y) ∈ R2

and so by Sandwich Theorem, it follows that lim f (x, y) exists and is equal to
(x,y)→(0,0)
f (0, 0) = 0. Thus f is continuous at (0, 0). [1]
Next, for any h, k ∈ R with h 6= 0 and k 6= 0, we have

f (h, 0) − f (0, 0) f (0, k) − f (0, 0)

=0 and =0
h k
and therefore, both fx and fy exist at (0, 0) and fx (0, 0) = fy (0, 0) = 0. [1]
Finally, for (h, k) ∈ R2 with (h, k) 6= (0, 0),
r
f (h, k) − f (0, 0) − hfx (0, 0) − kf( 0, 0) |hk|
√ = 2 + k2
2
h +k 2 h
p
and if k = mh, then the RHS is |m|/(1 + m2 ), which is nonzero if m 6= 0; it follows
that f is not differentiable at (0, 0). [1]
√
[Aliter: Consider the unit vector u = (u1 , u2 ), where u1 = u2 = 1/ 2. Now, for t 6= 0,

f (tu1 , tu2 ) − f (0, 0) 1 |t|

=
t 2 t
and since limt→0 (|t|/t) does not exist, it follows that the directional derivative Du f (0, 0)
does not exist. Hence f can not be differentiable at (0, 0).]

2
3. Let φ : R3 → R be a differentiable scalar field and let P, Q, R, S, and T denote the points
(1, 2, 1), (1, 2, 2), (1, 1, 2), (2, 2, 0), and (2, 2, 3), respectively. Suppose the directional derivatives
−−→ −→ −→
of φ at P in the directions of P Q, P R, and P S are 1, √12 , and 0, respectively. Then find a unit
vector u such that the directional derivative of φ at P in the direction of u has the maximum
−→
magnitude. Also, compute the directional derivative of φ at P in the direction of P T . [5 marks]

Solution: Since φ is differentiable, for any unit vector u in R3 , the directional derivative
Du φ(P ) exists and is equal to ∇φ(P ) · u.
−−→ −→ −→
Now let ∇φ(P ) = ai + bj + ck. The unit vectors in the direction of P Q, P R, and P S are

−j + k i−k
u−
−→ = k,
PQ
u−→ =
PR
√ , and u−→ = √
PS
[1]
2 2
Using the given values of directional derivatives of φ at P and the relation Du φ(P ) =
∇φ(P ) · u, we obtain

c = 1, −b + c = 1, and a − c = 0, which gives, (a, b, c) = (1, 0, 1).

Thus ∇φ(P ) = ai + bj + ck = i + k. [1]

Now the magnitude of the directional derivative of φ at P in the direction of u is

|Du φ(P )| = |∇φ(P )| cos θ where θ is the angle between ∇φ(P ) and u,

and this is maximum when

√ θ = 0, i.e., when u = ∇φ(P )/ |∇φ(P )|. Thus the desired unit
vector is u = (i + k) / 2. [1]
−→
Finally, the unit vectors in the direction of P T is u−→ = i+2k
√ and therefore the directional
PT 5
−→
derivative of φ at P in the direction of P T is
3
Du−→ φ(P ) = ∇φ(P ) · u−→ = √ . [2]
PT
PT
5

3
4. Let p, q be real numbers such that p > 1 and q > 1.

(i) Use the Lagrange Multiplier Method to show that

1 1 xp yq
+ ≤ + for all x, y ∈ (0, ∞) with xy = 1. [3 marks]
p q p q
1 1
(ii) Using part (i), show that if p + q = 1, then

ap bq
ab ≤ + for all a, b ∈ (0, ∞). [2 marks]
p q

Solution: Let D = {(x, y) ∈ R2 : x > 0 and y > 0} and let f, g : D → R be defined by

xp y q
f (x, y) = + and g(x, y) = xy − 1.
p q
We wish to minimize f (x, y) on C = {(x, y) ∈ D : xy = 1}, i.e., subject to the constraint
g(x, y) = 0. [1]
As per the Lagrange Multiplier Method, we try to solve the equations
∇f (x, y) = λ∇g(x, y) and g(x, y) = 0
for x, y (with x > 0 and y > 0) and λ. Thus, we consider
xp−1 = λy, y q−1 = λx, xy − 1 = 0. [ 12 ]
This gives xp = λxy = y q and using xy = 1, we find xp = (1/x)q , i.e., xp+q = 1, which
implies x = 1 and consequently, y = 1. [ 12 ]
Thus (x, y) = (1, 1) is the only common solution of ∇f (x, y) = λ∇g(x, y) and g(x, y) = 0.
Moreover, ∇g(1, 1) = (1, 1) 6= (0, 0). Consequently, if f has an absolute extremum on C,
then it must be at (1, 1). Now if (x, y) ∈ C is such that x → ∞, then f (x, y) → ∞ and
also if x → 0, then y → ∞ and so f (x, y) → ∞. Hence we can find (x1 , y1 ), (x2 , y2 ) ∈ C
with x1 < x2 such that
1 1
f (x, y) > + whenever (x, y) ∈ C with x < x1 or x > x2 .
p q
Since E = {(x, y) ∈ C : x1 ≤ x ≤ x2 } is a closed and bounded set and f is continuous, f
must attain an absolute minimum on E. Since f (1, 1) = p1 + 1q , we see that (1, 1) ∈ E and
the minimum of f on E must be the minimum of f on C. But (1, 1) is the only point of C
where f can have an absolute extremum. So it follows that f has the absolute minimum
on C at (1, 1) and therefore, for all x, y ∈ (0, ∞) with xy = 1, we have
1 1 xp y q
f (1, 1) ≤ f (x, y), i.e., + ≤ + . [1]
p q p q

1 1 a b
(ii) Suppose + = 1. Given any a, b ∈ (0, ∞), consider x = 1/p
and y = .
p q (ab) (ab)1/q
Note that x, y ∈ (0, ∞) and xy = 1. Hence from part (i) above, we obtain
1 1 xp y q ap bq ap bq
1= + ≤ + = + and so ab ≤ + . [2]
p q p q abp abq p q

4
5. (i) Consider the vector field F = (2xy +z 3 )i+x2 j+3xz 2 k on R3 . Determine a potential function
φ on R3 such that F = ∇φ. Also, find the work done by the vector field F in moving an
object from (1, −2, 1) to (3, 1, 4). [3 marks]
(ii) A vector field v on R3 is the product of a differentiable scalar field ψ and the gradient of a
scalar field φ, where φ has continuous first and second order partial derivatives. Compute
the real number v · curl v. [2 marks]

Solution: (i) The scalar field φ on R3 defined by φ(x, y, z) = x2 y + xz 3 satisfies F = ∇φ

and is hence a potential function for F. [2]
H
Now since F is conservative, the line integral F · ds is path-independent. Thus the work
done by the vector field F in moving an object from (1, −2, 1) to (3, 1, 4) is given by
I (3,1,4) I (3,1,4)
W = F · ds = ∇φ · ds
(1,−2,1) (1,−2,1)
= φ(3, 1, 4) − φ(1, −2, 1)
= (9 + 192) − (−2 + 1)
= 202 units. [1]

(ii) Since v = ψ∇φ,

curl(v) = ∇ × (ψ∇φ) = ∇ψ × ∇φ + ψ (∇ × ∇φ) = ∇ψ × ∇φ, [1]

since ∇ × ∇φ = curl(grad φ) = 0. Thus

v · (curl v) = (ψ∇φ) × (∇ψ × ∇φ) = 0, [1]

since ψ∇φ and ∇φ are parallel.

5
6. Let C be the astroid given by the parametric equations

x = 2 sin3 θ, y = 2 cos3 θ, 0 ≤ θ ≤ 2π.

(i) Using Green’s theorem, find the area of the planar region bounded by the closed curve C.
[3 marks]
(ii) Evaluate the line integral I
(2x + y 2 )dy − (2x2 + y)dx. [2 marks]
C

Solution: (i) Let R be the region bounded by C. Observe that C is a piecewise smooth
closed simple curve, which is oriented clockwise. Hence using Green’s theorem applied to
the vector field F(x, y) = xi, we get
Z
Area(R) = xdy , [1]
C−

where C − denotes the given curve C, but with opposite orientation. Hence
Z 0
Area(R) = 2 sin3 θ (2)(3) cos2 θ(− sin θ) dθ [1]
2π
Z 2π
= 12 sin4 θ cos2 θ dθ
0
Z 2π
= 12 [sin4 θ − sin6 θ] dθ
0
 
3 53 3π
= 12 π − π = . [1]
4 64 2

(ii) Consider the vector field

F(x, y) = (2x + y 2 )i + (2x2 + y)j .

Changing the orientation of C and then applying Green’s theorem (normal form), we find
I I
2 2
(2x + y )dy − (2x + y)dx = − (2x + y 2 )dy − (2x2 + y)dx
C C−
ZZ
= − (div F)d(x, y) [1]
R
ZZ
= − (2 + 1)d(x, y)
R

9π
= −3 Area(R) = − . [1]
2

6
7. Let S be the surface given by x2 + y 2 + z 2 = 9, 1 ≤ x ≤ 2, and let n beZthe
Z outward unit normal
to S. If F = (xi + yj + zk) × (i + j), then compute the surface integral (∇ × F) · n dS, using
S
Stokes’ theorem. [5 marks]

Solution: Note that

F(x, y, z) = −zi + zj + (x − y)k .
Using Stokes’ theorem, we have
ZZ Z
(∇ × F) · n dS = F · ds ,
S ∂S
where ∂S denotes the boundary of S with the induced orientation. Note that ∂S consists
of the circle y 2 + z 2 = 8, x = 1, oriented counterclockwise, and the circle y 2 + z 2 = 5,
x = 2, oriented clockwise. [1]
Thus ∂S = C1 ∪ C2 , where C1 is parametrized by
√ √
r1 (θ) = i + 2 2 cos θ j + 2 2 sin θ k, 0 ≤ θ ≤ 2π,

and C2 is parametrized by
√ √
r2 (θ) = 2i + 5 cos(2π − θ)j + 5 sin(2π − θ)k, 0 ≤ θ ≤ 2π. [1]

Now
Z Z 2π
F · ds = F(r1 (θ)) · r01 (θ) dθ
C1 0
Z 2π √ √ √ √ √
= [−2 2 sin θi + 2 2 sin θj + (1 − 2 2 cos θ)k] · [−2 2 sin θj + 2 2 cos θk]dθ
0
= −8 × 2π = −16π . [1]

and
Z Z 2π
F · ds = F(r2 (θ)) · r02 (θ) dθ
C2 0
Z 2π √ √ √ √ √
= [ 5 sin θi − 5 sin θj + (2 − 5 cos θ)k] · [− 5 sin θj − 5 cos θk]dθ
0
= 5 × 2π = 10π . [1]

Hence ZZ Z Z
∇ × F · n dS = F · ds + F · ds = −16π + 10π = −6π . [1]
S C1 C2

7
8. The sphere x2 + y 2 + z 2 = 25 is intersected by the plane z = 3. Consider the solid R bounded by
the closed surface S0 = S1 ∪ S2 , where S1 is the part of the sphere corresponding to z ≥ 3, and
S2 is the part of the plane z = 3 lying on or inside the sphere. If the outward unit normal to R is
cos αi + cos βj + cos γk, then compute the value of the surface integral
ZZ
(xz cos α + yz cos β + cos γ) dS
S
when (a) S is the spherical cap S1 , and (b) S is the planar base S2 . Furthermore, compute the
value of this surface integral when S is the complete boundary S0 by using (a) and (b), and also
by using the Gauss’ divergence theorem. [5 marks]

Solution: Clearly, both S1 and S2 project onto the disc E := {(x, y) ∈ R2 : x2 + y 2 ≤ 16}
in the xy-plane.
p
(a) The spherical cap S1 is given by z = f (x, y), (x, y) ∈ E, where f (x, y) = 25 − x2 − y 2 ,
and so the unit outward normal n = cos α i + cos βj + cos γk to S1 is
x y
−fx i − fy j + k z i+ z j+k
x y 
n= q =q and n dS = i + j + k d(x, y).
1 + fx2 + fy2 1 + fx2 + fy2 z z

Consequently,
ZZ ZZ   
x y  
(xz cos α + yz cos β + cos γ) dS = xz + yz + 1 d(x, y)
S1 z z
Z ZE
x2 + y 2 + 1 d(x, y)


=
E
Z 2π Z 4 
2 1
= (r + 1)r dr dθ = 144π. [1 ]
0 0 2

(b) The planar base S2 is given by z = 3, (x, y) ∈ E, and the unit outward normal n =
cos α i + cos βj + cos γk to S2 is simply n = −k, while n dS = −k d(x, y). Consequently,
ZZ ZZ
(xz cos α + yz cos β + cos γ) dS = −d(x, y) = −Area(E) = −16π. [1]
S2 E

Finally, combining (a) and (b), we obtain

ZZ
(xz cos α + yz cos β + cos γ) dS = 144π − 16π = 128π. [ 21 ]
S0

On the other hand, applying Gauss’ divergence Theorem to F := xz i + yz j + k, we obtain

ZZ ZZ ZZZ ZZZ
I := (xz cos α + yz cos β + cos γ) dS = F·n dS = div F dV = 2z d(x, y, z).
S0 ∂R R R
Thus, using Fubini’s theorem and then switching to polar coordinates, we find
ZZ Z √25−x2 −y2 !
I = 2zdz [1]
E 3
ZZ
= (16 − x2 − y 2 )d(x, y)
E
2π 4 r=4
16r2 r4
Z Z  
2
= (16 − r )r dr dθ = 2π − = 128π. [1]
0 0 2 4 r=0