MA 105 : Calculus (Autumn 2014)

Solutions to Tutorial Sheets

August 20, 2017

 Solutions to Tutorial Sheet 1

(1) For a given  > 0, we have to find n0 ∈ N such that |an | <  for all n ≥ n0 . Select

n0 ∈ N (This is possible by the archimedean property of R - but you should not

probably not mention this to your students for whom this fact is surely self evident.

If someone brings it up in class, you can acknowledge the comment and leave it at

that.) such that

10
(i) n0 > ,

5−
(ii) n0 > ,
3
2
1 1 n3
(iii) n0 > 3 as 1 > > |an |,
 n3 n+1
 
2 2 1 1
(iv) n0 > as > 2− = |an |.
 n n n+1
n2 n2
(2) (i) 2 ≤ an ≤ 2 ⇒ lim an = 1.
n +n  n +1 n→∞
   
n! n−1 n−2 2 1 1 n!
(ii) 0 < n = ··· ≤ ⇒ lim n = 0.
n n n n n n n→∞ n
3 2 2 4
n + 3n + 1 (1/n) + (3/n ) + (1/n ) 4
(iii) 0 < 4 2
= 2 4
< ⇒ lim an = 0.
n + 8n + 2 1 + (8/n ) + (2/n ) n n→∞
1
(iv) Let n n = 1 + hn . Then, for n ≥ 2, one has
   
n n 2 n 2
n = (1 + hn ) ≥ 1 + nhn + hn > h .
2 2 n
2 1
Thus 0 < h2n < n−1
(n ≥ 2) giving limn→∞ hn = 0. So limn→∞ n n = 1.

(v) Since
 √ 
 cos(π n) 
0<
 ≤ 1,
n 2  n2

it follows that limn→∞ an = 0.

√
√ √ √ n 1 1
(vi) n( n + 1 − n) = √ √ =q → as n → ∞.
n+1+ n 1 + n1 + 1 2

n2 1 1
(3) (i) { = (n − 1) + }n≥1 is not convergent since → 0 as n → ∞.
n+1 n+1 n+1

1
 (−1)n (−1)n (−1)n
 
n 1 1
(ii) {(−1) − = − }n≥1 is not convergent since → 0 as
2 n 2 n n
n → ∞.

1 1
(4) (i) Decreasing as an = 1/ n + and {n + }n≥1 is increasing.
n n
an+1 6
(ii) Increasing as = > 1.
an 5
n(n − 1) − 1
(iii) Increasing as an+1 − an = 2 > 0 for n ≥ 2.
n (1 + n)2
√
(5) (i) By the AM-GM inequality, we see that an ≥ 2 for all n ≥ 2. Consequently,
2−a2n
an+1 − an = 2an
≤ 0 for n ≥ 2. Thus {an }n≥2 is monotonically decreasing
√ √
and bounded below by 2. So limn→∞ an = a (say) exists, and a ≥ 2. Also
√
a = 12 a + a2 , i.e., a2 = 2. It follows that a = 2.

√
(ii) By induction, 2 ≤ an < 2 ∀n. Hence an+1 − an = (2−a √ n
)(1+an )
2+an +an
> 0 ∀n. Thus

limn→∞ an = a (say) exists and arguing as in (i), we find a = 2.

6−an
(iii) By induction, 2 ≤ an < 6 ∀n. Hence an+1 − an = 2
> 0 ∀n. Thus limn→∞ an =

a (say) exists and arguing as in (i), we find a = 6.

(6) It is clear that limn→∞ an+1 = L. The inequality ||an | − |L|| ≤ |an − L| implies that

limn→∞ |an | = |L|.

(7) Take  = |L|/2. Then  > 0 and since an → L, there exists n0 ∈ N such that |an −L| < 

∀n ≥ n0 . Now ||an | − |L|| ≤ |an − L| and hence |an | > |L| −  = |L|/2 ∀n ≥ n0 .
√ 
(8) Given  > 0, there exists n0 ∈ N such that |an | < 2 ∀n ≥ n0 . Hence  an  < 

∀n ≥ n0 . [Note: For a corresponding result when an → L, see, e.g., part (ii) of

Propositions 1.9 and 2.4 of [GL-1].]

(9) Both the statements are false. Consider, for example, an = 1 and bn = (−1)n .

(10) The implication “⇒” is obvious. For the converse, suppose both {a2n }n≥1 and {a2n+1 }n≥1

converge to `. Let  > 0 be given. Choose n1 , n2 ∈ N such that |a2n − `| <  for all

2
 n ≥ n1 and |a2n+1 − `| <  for all n ≥ n2 . Let n0 = max{n1 , n2 }. Then

|an − `| <  for all n ≥ 2n0 + 1.

(11) (i) The statement is false. For example, consider a = −1, b = 1, c = 0 and define

f, g : (−1, 1) → R by

 1

if x = 0
f (x) = x and g(x) =
 1/x2 if x 6= 0.


(ii) The statement is true since |g(x)| ≤ M for all x ∈ (a, b) implies that 0 ≤

|f (x)g(x)| ≤ M |f (x)| for all x ∈ (a, b).

(iii) The statement is true since lim f (x)g(x) = lim f (x) lim g(x).
x→c x→c x→c

(12) Suppose lim f (x) = L. Then lim f (α + h) = L. and since

x→α h→0

|f (α + h) − f (α − h)| ≤ |f (α + h) − L| + |f (α − h) − L|

it follows that

lim |f (α + h) − f (α − h)| = 0.
h→0

The converse is false; e.g. consider α = 0 and


 1 if x = 0

f (x) =
 1 if x =

6 0.
|x|

(13) (i) Continuous everywhere except at x = 0. To see that f is not continuous at 0,

consider the sequences {xn }n≥1 , {yn }n≥1 where

1 1
xn := and yn := .
nπ 2nπ + π2

Note that both xn , yn → 0, but f (xn ) → 0 and f (yn ) → 1.

3
 (ii) Continuous everywhere. For ascertaining the continuity of f at x = 0, note that

|f (x)| ≤ |x| and f (0) = 0.

(iii) Continuous everywhere on [1, 3] except at x = 2.

(14) Taking x = 0 = y, we get f (0 + 0) = 2f (0) so that f (0) = 0. By the assumption of

the continuity of f at 0, limx→0 f (x) = 0. Thus,

lim f (c + h) = lim [f (c) + f (h)] = f (c)

h→0 h→0

showing that f is continuous at x = c.

Optional: First verify the equality for all k ∈ Q and then use the continuity of f to

establish it for all k ∈ R.

(15) Clearly, f is differentiable for all x 6= 0 and the derivative is

   
0 1 1
f (x) = 2x sin − cos , x 6= 0.
x x

Also,
h2 sin( h1 ) − 0
f 0 (0) = lim = 0.
h→0 h

Clearly, f 0 is continuous at any x 6= 0. However, lim f 0 (x) does not exist. Indeed, for
x→0

any δ > 0, we can choose n ∈ N such that x := 1/nπ, y := 1/(n + 1)π are in (−δ, δ),

but |f 0 (x) − f 0 (y)| = 2.

(16) The inequality

 
 f (x + h) − f (x) 
0≤
  ≤ c|h|α−1
h 

implies, by the Sandwich Theorem, that

 
 f (x + h) − f (x) 
lim   = 0 ∀x ∈ (a, b).
h→0  h 

4
(17) For the first part, observe that
 
f (c + h) − f (c − h) 1 f (c + h) − f (c) f (c − h) − f (c)
lim = lim +
h→0+ 2h h→0+ 2 h −h
1 0
= [f (c) + f 0 (c)] = f 0 (c).
2

The converse is false; consider, for example, f (x) = |x| and c = 0.

(18) Since f (x + y) = f (x)f (y), we obtain, in particular, f (0) = f (0)2 and therefore f (0) =

0 or 1. If f (0) = 0, then

f (x + 0) = f (x)f (0) ⇒ f (x) = 0 ∀x.

Thus, trivially, f is differentiable. If f (0) = 1, then

 
0 f (c + h) − f (c) f (h) − f (0)
f (c) = lim = f (c) lim = f 0 (0)f (c).
h→0 h h→0 h

(19) (i) Let f (x) = cos(x). Then f 0 (x) = − sin(x) 6= 0 for x ∈ (0, π).

Thus g(y) = f −1 (y) = cos−1 (y), −1 < y < 1 is differentiable

and
1
g 0 (y) = , where x is such that f (x) = y.
f 0 (x)

Therefore,
−1 −1 −1
g 0 (y) = =p =p .
sin(x) 1 − cos2 (x) 1 − y2
(ii) Note that
1
cosec−1 (x) = sin−1 for |x| > 1.
x

Since
d 1
sin−1 (x) = √ for |x| < 1,
dx 1 − x2

5
 one has, by the Chain rule,
 
d 1 −1
cosec−1 (x) = q , |x| > 1.
dx (1 − 1
) x2
x2

(20) By the Chain rule,

   
dy 2x − 1 d 2x − 1
= f0
dx x + 1 dx x + 1
 2    2
2x − 1 3 3 2x − 1
= sin = sin .
x+1 (x + 1)2 (x + 1)2 x+1

Optional exercises

(1) Note that f (n) = nf (1) for all n ∈ N. Show that f (r) = rf (1), for every rational

number r. By continuity f (λ) = λf (1) for all real λ. If we set f (1) = k, we are done.

(2) This can be easily done inductively, since every derivative of f satisfies the same

property as f .

(3) Consider f (x) := |x| + |1 − x| for x ∈ R.

(4) For c ∈ R, select a sequence {an }n≥1 of rational numbers and a sequence {bn }n≥1

of irrational numbers, both converging to c. Then {f (an )}n≥1 converges to 1 while

{f (bn )}n≥1 converges to 0, showing that limit of f at c does not exist.

(5) Suppose c 6= 1/2. If {an }n≥1 is a sequence of rational numbers and {bn }n≥1 a sequence

of irrational numbers, both converging to c, then g(an ) = an → c, while g(bn ) =

1 − bn → 1 − c, and c 6= 1 − c. Thus g is not continuous at any c 6= 1/2. Further, if

{an }n≥1 is any sequence converging to c = 1/2, then g(an ) → 1/2 = g(1/2). Hence, g

is continuous at c = 1/2.

(6) Let L = limx→c f (x). Take  = L − α. Then  > 0 and so there exists a δ > 0 such

that

|f (c + h) − L| <  for 0 < |h| < δ.

6
 Consequently, f (c + h) > L −  = α for 0 < |h| < δ.

(7) (i) ⇒ (ii): Choose δ > 0 such that (c − δ, c + δ) ⊆ (a, b). Take α = f 0 (c) and

 f (c + h) − f (c) − αh , if h 6= 0

1 (h) = h
 0,

if h = 0.

|f (c + h) − f (c) − αh|
(ii) ⇒ (iii): lim = lim |1 (h)| = 0
h→0 |h| h→0

 
 f (c + h) − f (c)  f (c + h) − f (c)
(iii) ⇒ (i): lim 
 − α = 0 ⇒ lim exists
h→0 h h→0 h
and is equal to α.

0
(8) f (0) = 0, f (0) = 1 and f 0 (x) = 1 + x2

(9) Follows easily from the definitions.

(10) Apply the Intermediate Value Theorem to the function f (x) − x.

7
 Solutions to Tutorial Sheet 2

√
(1) f (x) = x3 − 6x + 3 has stationary points at x = ± 2.
√ √ √ √
Note that f (− 2) = 4 2 + 3 > 0, f (+ 2) = −4 2 + 3 < 0. Therefore f has a
√ √
root in (− 2, 2). Also, f → −∞ as x → −∞ implying that f has a root in
√
(−∞, − 2). Similarly, f → +∞ as x → +∞ implying that f has a root in
√
( 2, ∞). Since f has at most three roots, all its root are real.

(2) For f (x) = x3 + px + q, p > 0, f 0 (x) = 3x2 + p > 0. Therefore f is strictly in-

creasing and can have at most one real root. Since

p q
lim + 3 = 0,
x→±∞ x2 x
f (x) p q
3
=1+ 2 + 3 >0
x x x
for |x| very large. Thus f (x) > 0 if x is large positive and f (x) < 0 if x is large

negative. By the Intermediate Value Property (IVP) f must have at least

one real root.

(3) By the IVP, there exists at least one x0 ∈ (a, b) such that f (x0 ) = 0. If there

were another y0 ∈ (a, b) such that f (y0 ) = 0, then by Rolle’s theorem there

would exist some c between x0 and y0 (and hence between a and b) with

f 0 (c) = 0, leading to a contradiction.

(4) Since f has 3 distinct roots, say, r1 < r2 < r3 , by Rolle’s theorem f 0 (x) has at

least two real roots, say, x1 and x2 such that r1 < x1 < r2 and r2 < x2 < r3 .
p
Since f 0 (x) = 3x2 + p, this implies that p < 0, and x1 = − −p/3, x2 =
p
−p/3. Now, f 00 (x1 ) = 6x1 < 0 =⇒ f has a local maximum at x = x1 .

Similarly, f has a local minimum at x = x2 . Since the quadratic f 0 (x) is

8
 negative between its roots x1 and x2 (so that f is decreasing over [x1 , x2 ]) and

f has a root r2 in (x1 , x2 ), we must have f (x1 ) > 0 and f (x2 ) < 0. Further,

r r
−4p3 −4p3
f (x1 ) = q + , f (x2 ) = q −
27 27
so that
4p3 + 27q 2
= f (x1 )f (x2 ) < 0.
27
(5) For some c between a and b, one has
 
 sin(a) − sin(b) 
  = | cos(c)| ≤ 1.
 a−b 
 
(a+b)
(6) By Lagrange’s Mean Value Theorem (MVT) there exists c1 ∈ a, 2
such that

f a+b


− f (a)
2
b−a

= f 0 (c1 )
2
a+b


and there exists c2 ∈ 2
,b such that
a+b


f (b) − f
b−a

2 = f 0 (c2 ).
2

Clearly one has c1 < c2 , and adding the above equations one obtains
f (b) − f (a)
f 0 (c1 ) + f 0 (c2 ) = b−a

= 2 (as f (b) = b, f (a) = a).
2

(7) By Lagrange’s MVT, there exists c1 ∈ (−a, 0) and there exists c2 ∈ (0, a) such

that

f (0) − f (−a) = f 0 (c1 )a and f (a) − f (0) = f 0 (c2 )a.

Using the given conditions, we obtain

f (0) + a ≤ a and a − f (0) ≤ a

which implies f (0) = 0.

Optional: Consider g(x) = f (x) − x, x ∈ [−a, a]. Since g 0 (x) = f 0 (x) − 1 ≤ 0, g

is decreasing over [−a, a]. As g(−a) = g(a) = 0, we have g ≡ 0.

9
(8) (i) No such function exists in view of Rolle’s theorem.
x2
(ii) f (x) = 2
+x

(iii) f 00 ≥ 0 ⇒ f 0 increasing. As f 0 (0) = 1, by Lagrange’s MVT we have

f (x) − f (0) ≥ x for x > 0. Hence f with the required properties cannot

exist.

(iv) 
1
if x ≤ 0


1−x
f (x) =
 1 + x + x2 if x > 0.


(9) The points to check are the end points x = −2 and x = 5, the point of non-

differentiability x = 0, and the stationary point x = 2. The values of f at

these points are given by

f (−2) = f (2) = 13, f (0) = 1, f (5) = −14.

Thus, global max = 13 at x = ±2, and global min = −14 at x = 5.

(10) To be done in the tutorial. Asymptotes have not been discussed in class, but this

curve doesn’t have any, so you can skip that part.

(i) f (x) = 2x3 + 2x2 − 2x − 1 ⇒ f 0 (x) = 6x2 + 4x − 2 = 2(x + 1)(3x − 1).

Thus, f 0 (x) > 0 in (−∞, −1)∪(1/3, ∞) so that f (x) is strictly increasing

in those intervals, and f 0 (x) < 0 in (−1, 1/3) so that f (x) is strictly decreasing in that in

Thus, f (x) has a local maximum at x = −1, and a local minimum atx =
1
. As f 00 (x) = 12x + 4 we have that f (x) is convex in − 31 , ∞ and concave in −∞, − 13



3
1
with a point of inflection at x = − .
3
(ii) f (x) = 1 + 12|x| − 3x2 ; f is not differentiable at x = 0; f (0) = 1. Further,

f 0 (x) = 0 at x = ±2, f 0 (x) < 0 in (−2, 0) ∪ (2, 5], f 0 (x) > 0 in (0, 2),

10
 and

f 00 (x) = −6 in (−2, 0) ∪ (0, 5). Thus f is concave in(−2, 0)∪(0, 5), decreasing in

(−2, 0) ∪ (2, 5), and increasing in (0, 2); further, f has an absolute

maximum at x = ±2.

(11) To be done in the tutorial.

√
(12) (i) x2 , (ii) x, (iii) x1 , (iv) sin x

(13) (i) True (easy).

(ii) False. Take f (x) = g(x) = x3 at the point x = 0.

x2
(14) This exercise involves asymptotes so you will have to introduce this idea. y = 2 ⇒ lim y = 1 ⇒
x +1 x→±∞
2x
y0 = 2 ⇒ y is increasing in (0, ∞) and decreasing in (−∞, 0).
(x + 1)2
2 −1)
Further, y 00 = − 2(3x
(x2 +1)3
implies that y 00 > 0 if |x| < √13 , and y 00 < 0 if |x| > √13 .

Therefore,
   
1 1 1 1
y is convex in − √ , √ and concave in R \ − √ , √
3 3 3 3
1
with the points x = ± √ being the points of inflection.
3

11
 Solutions to Tutorial Sheet 3

(1) The remainder terms are easy in these cases. The series are given by

(i)
∞
x 2 x4 X (−1)k
cos x = 1 − + + ... = · x2k .
2! 4! k=0
2k!
(ii)
∞
x3 X (−1)n
arctan x = x − + ... = x2n+1 for |x| < 1.
3! n=0
2n + 1
For the series for arctan x we can proceed in two ways. The easier way is to observe
1
that f (1) (x) = 1+x2
and integrate this term by term. If you are using this solution, it

is best to read the paragraph before 3.5 first and then do this problem.

I recall seeing a (somewhat complicated) direct solution where one can evaluate the

derivatives directly at 0, but I am not able to reproduce it now. Its probably floating

around on the internet somewhere.

(2) The Taylor series is just (x − 1)3

(3) The Taylor series is simply

1729x1729 + 1728x1728 + 28x28 + 6x6 + 1729.

Indeed, for any polynomial, the Taylor series about the point 0 just gives you the

polynomial back.

(4) Let us denote the partial sums of of the given series by sm (x). We would like to show

that |sm (x) − sn (x)| can be made arbitrarily small whenever m and n are sufficiently

large. Let us assume that m > n. We see that

   
m
xk   xn  1 1 |xn |

 X 1
|sm (x) − sn (x)| =  ≤  + + . . . m−n ≤ 2 · .


k=n+1
k!  n! 2 4 2 n!
If N is made sufficiently large and n > N , the last expression can be made as small

as we please.

12
 (5) We simply integrate term by term to get
∞
x2 X xk
log x + x + + . . . = log x + .
2 · 2! k=1
k · k!

(6) The solution will be provided later.

(7) Follow the hints given in the question.

(8)
∞
x3 2x5 X B2n (−4)n (1 − 4n ) π
tan x = x + + + ··· = x2n−1 for |x| < .
3 15 n=1
(2n)! 2
Here B2n are the Bernoulli numbers defined by
∞
t X tm
= Bm ,
et − 1 m=0 m!
that is, the Bernoulli numbers Bm are the numbers that appear in the Taylor series
t
expansion for et −1
.

(9) The key to the solution is the function


e−1/x if x > 0


h(x) =

0 if x ≤ 0.


Recall that this is a smooth function for which h(n) (0) = 0 for all n. Let us construct

a smooth function that is 0 outside of [c, d] and 1 on [a, b] ⊂ [c, d], for any such pairs

of intervals. Let f1 (x) = h(x − c) and f2 (x) = h(a − x). Then

f1
f (x) =
f1 + f2
has the property that f is identically 0 to the left of c and identically 1 to the right

of a. Similarly, let g1 (x) = h(d − x) and g2 (x) = h(x − b). Then g = g1 /g1 + g2 is

identically 0 to the right of d and identically 1 to the left of b. Then k = f g is the

desired function.

(10) Follow the hints given in the question.

13
 Solutions to Tutorial Sheet 4

(1) The given fumction is integrable as it is monotone. Let Pn be the partition of [0, 2]

into 2 × 2n equal parts. Then U (Pn , f ) = 3 and

1 (2n − 1)
L(Pn , f ) = 1 + 1 × + 2 × →3
2n 2n
R2
as n → ∞. Thus, 0
f (x)dx = 3.
Rb
(2) (a) f (x) ≥ 0 ⇒ U (P, f ) ≥ 0, L(P, f ) ≥ 0 ⇒ a f (x)dx ≥ 0.
Rb
Suppose, moreover, f is continuous and a f (x)dx = 0. Assume f (c) > 0 for some
f (c)
c in [a, b]. Then f (x) > 2
in a δ-nbhd of c for some δ > 0. This implies that

f (c)
U (P, f ) > δ ×
2
Rb
for any partition P, and hence, a
f (x)dx ≥ δf (c)/2 > 0, a contradiction.

(b) on [0, 1] take f (x) = 0 for all x 6= 0 and f (0) = 0.

n  3 Z 1
1X i 2 2
(3) (i) Sn = −→ (x)3/2 dx =
n i=1 n 0 5
n Z 1
1X 1 dx π
(ii) Sn = 2 −→ 2
=
n i=1 i + 1 0 x +1 4


n
n
1X 1
Z 1
dx √
(iii) Sn = q −→ √ = 2( 2 − 1)
n i=1 i
+1 0 x+1
n
n Z 1
1X iπ
(iv) Sn = cos −→ cos πxdx = 0
n i=1 n 0

1 2 √
Z 1 Z 2 Z 3
3/2 19
(v) Sn −→ xdx + x dx + x2 dx = + (4 2 − 1) +
0 1 2 2 5 3
Rx
(4) Let F (x) = a f (t)dt. Then F 0 (x) = f (x). Note that
Z v(x) Z v(x) Z u(x)
f (t)dt = f (t)dt − f (t)dt = F (v(x)) − F (u(x)).
u(x) a a

14
 By the Chain Rule one has
Z v(x)
d
f (t)dt = F 0 (v(x))v 0 (x) − F 0 (u(x))u0 (x)
dx u(x)

= f (v(x))v 0 (x) − f (u(x))u0 (x).

d2 y
dy
p
(a) dx
= 1
dx/dy
= 1 + y2, dx2
= √y dy
= y.
1+y 2 dx

(b) (i) F 0 (x) = cos((2x)2 )2 = 2 cos(4x2 ).

(ii) F 0 (x) = cos(x2 )2x = 2x cos(x2 )

(5) Define
Z x+p
F (x) = f (t)dt, x ∈ R.
x

Then F 0 (x) = 0 for every x.

(6) Write sin λ(x − t) as sin(λx) cos(λt) − cos(λx) sin(λt) in the integrand, take trems in x

outside the integral, evaluate g 0 (x), g 00 (x), and simplify to show LHS=RHS; from the

expressions for g(x) and g 0 (x) it should be clear that g(0) = g 0 (0) = 0.

The problem could also be solved by appealing to the following theorem:

∂h
Theorem A. Let h(t, x) and ∂x
(t, x) be continuous functions of t and x on the rect-

angle [a, b] × [c, d]. Let u(x) and v(x) be differentiable functions of x on [c, d] such

that, for each x in [c, d], the points (u(x), x) and (v(x), x) belong to [a, b] × [c, d]. Then
Z v(x) Z v(x)
d ∂h
h(t, x)dt = (t, x)dt − u0 (x)h(u(x), x) + v 0 (x)h(v(x), x).
dx u(x) u(x) ∂x

Consider now
Z x
1
g(x) = f (t) sin λ(x − t)dt.
λ 0

15
 Let h(t, x) = λ1 f (t) sin λ(x − t), u(x) = 0, and v(x) = x. Then it follows from Theorem

A that
Z x
0
g (x) = f (t) cos λ(x − t).
0

Again applying Theorem A, we have

Z x
00
g (x) = −λ f (t) sin λ(x − t) + f (x).
0

Thus

g 00 (x) + λ2 g(x) = f (x).

That g(0) = g 0 (0) = 0 is obvious from the expressions for g(x) and g 0 (x).
R1 R1 √ 1
(7) (i) 0
y dx = 0
(1 + x − 2 x) dx = 6
R2 R2 128
(ii) 2 0
(2x2 − (x4 − 2x2 )) dx = 2 0
(4x2 − x4 ) dx = 15
R3 R3 4
(iii) 1
(3y − y 2 − (3 − y)) dy = 1
(4y − y 2 − 3) dy = 3

R 1−a R 1−a (1−a)3

(8) 0
(x − x2 − ax) dx = 0
((1 − a)x − x2 ) dx = 4.5 gives 6
= 4.5 so that a =

−2.
R π/3 R π/3
(9) Required area = 2 × 0 21 (r22 − r12 )dθ = 4a2 0 (8 cos2 θ − 2 cos θ − 1)dθ = 4πa2 .
Z 2π q Z 2π
2
Rπ
(10) (i) Length = 2
(1 − cos(t)) + sin (t)dt = 2| sin(t/2)|dt = 4 0 | sin(u)|du = 8.
0 0
π/4 π/4 √ R π/4
Z p Z p
(ii) Length = 1 + y 02 dx = 1 + cos(2x)dx = 2 0 | cos(x)|dx = 1.
0 0

 
dy 2 1
(11) =x + − 2 .
dx 4x
s  2 r
ds dy 1 1 1
= 1+ = 1 + x4 + 4
− = x2 + 2 .
dx dx 16x 2 4x
Therefore,
3 3
x3
Z   
1
2 1 53
Length = x + 2 dx = − = .
1 4x 3 4x 1 6

16
 The surface area is
3 3
x3
Z Z   
ds 1 2 1
S= 2π(y + 1) dx = 2π + +1 x + 2 dx
1 dx 1 3 4x 4x
3
x6 x3 x2

1 1
= 2π + + − 2
− .
18 3 6 32x 4x 1

(12) The diameter of the circle at a point x is given by

(8 − x2 ) − x2 , −2 ≤ x ≤ 2.

So the area of the cross-section at x is A(x) = π(4 − x2 )2 . Thus

Z 2 Z 2
2 2 512π
Volume = π(4 − x ) dx = 2π (4 − x2 )2 dx = .
−2 0 15

(13) In the first octant, the sections perpendicular to the y-axis are squares with
p p
0≤x≤ a2 − y 2 , 0 ≤ z ≤ a2 − y 2 , 0 ≤ y ≤ a.

p
Since the squares have sides of length a2 − y 2 , the area of the cross-section at y is

A(y) = 4(a2 − y 2 ). Thus the required volume is

a a
16a3
Z Z
A(y)dy = 8 (a2 − y 2 )dy = .
−a 0 3

(14) Let the line be along z-axis, 0 ≤ z ≤ h. For any fixed z, the section is a square of area
Rh
r2 . Hence the required volume is 0 r2 dz = r2 h.

(15) Washer Method

Area of washer = π(1 + y)2 = π(1 + (3 − x2 ))2 = π(4 − x2 )2 so that

R2
Volume = −2 π(4 − x2 )2 dx = 512π/15.

(This is the same integral as in (6) above).

17
 Shell method
√
Area of shell = 2π(y − (−1))2x = 4π(1 + y) 3 − y so that
Z 3 p
Volume = 4π(1 + y) 3 − ydy = 512π/15.
−1

(14) Washer Method

Required volume = Volume of the sphere -Volume generated by revolving the shaded
R1 √ R1
region around the y-axis = 32π/3−[ −1 πx2 dy −π( 3)2 2] = 32π/3−2π[ 0 (4−y 2 )dy −

3] = 32π/3 − 2π[11/3 − 3] = 28π/3

Shell Method

Required volume = Volume of the sphere -Volume generated by revolving the shaded
R2 R2 √
region around the y-axis = 32π/3 − √3 2πx(2y)dx = 32π/3 − 4π √3 x 4 − x2 dx =

32π/3 − 4π(1/3) = 28π/3

18
 Solutions to Tutorial Sheet 5

(1) (i) {(x, y) ∈ R2 | x 6= ±y}

(ii) R2 − {(0, 0)}

(2) (i) A level curve corresponding to any of the given values of c is the straight line

x − y = c in the xy-plane. A contour line corresponding to any of the

given values of c is the same line shifted to the plane z = c in R3 .

(ii) Level curves do not exist for c = −3, −2, −1. The level curve corre-

sponding to c = 0 is the point (0, 0). The level curves corresponding

to c = 1, 2, 3, 4 are concentric circles centered at the origin in the xy-

plane. Contour lines corresponding to c = 1, 2, 3, 4 are the cross-sections

in R3 of the paraboloid z = x2 + y 2 by the plane z = c, i.e., circles in the

plane z = c centered at (0, 0, c).

(iii) For c = −3, −2, −1, level curves are rectangular hyperbolas xy = c

in the xy-plane with branches in the second and fourth quadrant. For

c = 1, 2, 3, 4, level curves are rectangular hyperbolas xy = c in the xy-

plane with branches in the first and third quadrant. For c = 0, the

corresponding level curve (resp. the contour line) is the union of the x-

axis and the y-axis in the xy-plane (resp. in the xyz-space). A contour

line corresponding to a non-zero c is the cross-section of the hyperboloid

z = xy by the plane z = c, i.e., a rectangular hyperbola in the plane

z = c.

(3) (i) Discontinuous at (0, 0). (Check lim f (x, y) using y = mx3 ).
(x,y)→(0,0)

(ii) Continuous at (0, 0) :

19
  x2 − y 2  2 2
 
xy  ≤ |xy| x + y = |xy|.
 x2 + y 2  x2 + y 2
(iii) Continuous at (0, 0) :

p
|f (x, y)| ≤ 2(|x| + |y|) ≤ 4 x2 + y 2 .

(4) (i) Use the sequential definition of limit: (xn , yn ) → (a, b) =⇒ xn → a and yn →

b =⇒ f (xn ) → f (a) and g(yn ) → g(b) =⇒ f (xn ) ± g(yn ) → f (a) ± g(b)

by the continuity of f, g and limit theorems for sequences.

(ii) (xn , yn ) → (a, b) =⇒ xn → a and yn → b =⇒ f (xn ) → f (a) and g(yn ) →

g(b) =⇒ f (xn )g(yn ) → f (a)g(b) by the continuity of f, g and limit the-

orems for sequences.

(iii) Follows from (i) above and the following:

f (x) + g(y) |f (x) − g(y)|

min{f (x), g(y)} = − ,
2 2
f (x) + g(y) |f (x) − g(y)|
max{f (x), g(y)} = + .
2 2

(5) Note that limits are different along different paths: f (x, x) = 1 for every x and

f (x, 0) = 0.

(6) (i) fx (0, 0) = 0 = fy (0, 0).

(ii)
sin2 (h)/|h| sin2 (h)
fx (0, 0) = lim = lim
h→0 h h→0 h|h|

does not exist (Left Limit 6= Right Limit). Similarly, fy (0, 0) does not

exist.

(7) |f (x, y)| ≤ x2 + y 2 ⇒ f is continuous at(0, 0).

It is easily checked that fx (0, 0) = fy (0, 0) = 0.

20
 Now,
    
1 1 1
fx = 2x sin − 2 cos .
x2 + y 2 x + y2 x2 + y 2
 
1
The function 2x sin x2 +y2 is bounded in any disc centered at (0, 0),
 
2x 1
while 2 cos is unbounded in any such disc.
x + y2 x2 + y 2  
(To see this, consider (x, y) = √1nπ , 0 for n a large positive integer.)

Thus fx is unbounded in any disc around (0, 0).

f (h, 0) − f (0, 0) 1
(8) fx (0, 0) = lim = lim sin does not exist. Similarly fy (0, 0) does
h→0 h h→0 h
not exist. Clearly, f is continuous at (0, 0).

(9) (i) Let ~v = (a, b) be any unit vector in R2 . We have

 
a2 −b2
f (h~v ) f (ha, hb) h2 ab a2 +b2
(D~v f ) (0, 0) = lim = lim = lim = 0.
h→0 h h→0 h h→0 h

Therefore (D~v f ) (0, 0) exists and equals 0 for every unity vector ~v ∈ R2 .

For considering differentiability, note that fx (0, 0) = (Dî f ) (0, 0) = 0 =

fy (0, 0) = Dĵ f (0, 0). We have then

|f (h, k) − f (0, 0) − hfx (0, 0) − kfy (0, 0)| |hk(h2 − k 2 )|

lim √ = lim =0
(h,k)→(0,0) h2 + k 2 (h,k)→(0,0) (h2 + k 2 )3/2

since
√ √
|hk(h2 − k 2 )| |hk| h2 + k 2 h2 + k 2 h2 + k 2 √ 2
0≤ ≤√ ≤ √ = h + k2.
(h2 + k 2 )3/2 h2 + k 2 h2 + k 2 h2 + k 2

Thus f is differentiable at (0, 0).

(ii) Note that, for any unit vector ~v = (a, b) in R2 , we have

h3 a3 a3 a3
D~v f (0, 0) = lim = lim = .
h→0 h(h2 (a2 + b2 )) h→0 (a2 + b2 ) (a2 + b2 )

To consider differentiability, note that fx (0, 0) = 1, fy (0, 0) = 0 and

|f (h, k) − h × 1 − k × 0| |h3 /(h2 + k 2 ) − h|

lim √ = lim √
(h,k)→(0,0) h2 + k 2 (h,k)→(0,0) h2 + k 2

21
 |hk 2 |
= lim
(h,k)→(0,0) (h2 + k 2 )3/2

does not exist (consider, for example, k = mh). Hence f is not differen-

tiable at (0, 0).

(iii) For any unit vector ~v ∈ R2 , one has

h i
2 2 2 1
h (a + b ) sin h2 (a2 +b2 )
(D~v f ) (0, 0) = lim = 0.
h→0 h
Also,
 h i
 2 2 1
(h + k ) sin (h2 +k2 ) 

√
 
1
lim √ = lim h2 + k 2 sin = 0;
(h,k)→(0,0) h2 + k 2 (h,k)→(0,0) h2 + k 2
therefore f is differentiable at (0, 0).
p
(10) f (0, 0) = 0, |f (x, y)| ≤ x2 + y 2 =⇒ f is continuous at (0, 0).

Let ~v be a unit vector in R2 .

For ~v = (a, b), with b 6= 0, one has

√
1 hb √ 2 2 ( a2 + b2 )b
(D~v ) f (0, 0) = lim h a + h2 b2 = .
h→0 h |hb| |b|
If ~v = (a, 0), then (D~v f ) (0, 0) = 0. Hence (D~v f ) (0, 0) exists for every unit

vector ~v ∈ R2 . Further,

fx (0, 0) = 0, fy (0, 0) = 1,

and
 √ 
k 2 + k2 − k
|f (h, k) − 0 − h × 0 − k × 1|  |k|
h 
lim √ = lim √
(h,k)→(0,0) h2 + k 2 (h,k)→(0,0) h2 + k 2
 
k k 
= lim  −√ 
(h,k)→(0,0) |k| h2 + k 2 
does not exist (consider, for example, k = mh) so that f is not differentiable

at (0, 0).

(11) Continuity follows from the fact that |xy| ≤ x2 + y 2 .

22
 Solutions to Tutorial Sheet 6

.(1) Solution provided in the class slides.

(2) Solution provided in the class slides.

∂F ∂F ∂F
(3) (∇F ) (1, −1, 3) = (
(1, −1, 3), (1, −1, 3), (1, −1, 3)) = 4j + 6k.
∂x ∂y ∂z
The tangent plane to the surface F (x, y, z) = 7 at the point (1, −1, 3) is given

by

0 × (x − 1) + 4 × (y + 1) + 6 × (z − 3) = 0, i.e., 2y + 3z = 7.

The normal line to the surface F (x, y, z) = 7 at the point (1, −1, 3) is given by

x = 1, 3y − 2z + 9 = 0.
 
(2, 2, 1) 2 2 1 2 1
(4) u = √ = , , = (i + j) + k
22 + 22 + 12 3 3 3 3 3
and

(∇F ) ((2, 2, 1)) = 3i − 5j + 2k.

Therefore,
6 10 2 2
(Du F ) (2, 2, 1) = (∇F )(2, 2, 1) · u = − + =− .
3 3 3 3
(5) Given that sin (x + y) + sin (y + z) = 1 (with cos(y + z) 6= 0).

(The student may assume that z is a sufficiently smooth function of x and y).

Differentiating w.r.t. x while keeping y fixed, we get

∂z
cos (x + y) + cos (y + z) ∂x = 0. (∗)

Similarly, differentiating w.r.t.y while keping x fixed, we get

 
∂z
cos (x + y) + cos (y + z) 1 + ∂y = 0. (∗∗)

Differentiating (*) w.r.t y we have

 
∂z ∂z ∂2z
− sin (x + y) − sin (y + z) 1 + ∂y ∂x
+ cos (y + z) ∂x∂y = 0.

23
 Thus, using (*) and (**), we have

∂ 2z
∂x∂y
   
1 ∂z ∂z
= sin (x + y) + sin (y + z). 1 +
cos (y + z) ∂y ∂x

   
1 cos (x + y) cos (x + y)
= sin (x + y) + sin (y + z) − −
cos (y + z) cos (y + z) cos (y + z)

sin (x + y) cos2 (x + y)
= + tan (y + z) 2 .
cos (y + z) cos (y + z)

(6) We have
fx (0, k) − fx (0, 0)
fxy (0, 0) = lim ,
k→0 k
where (noting that k 6= 0)

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

fx (0, k) = lim = −k and fx (0, 0) = lim = 0.
h→0 h h→0 h

Therefore,

−k − 0
fxy (0, 0) = lim = −1 ; similarly fyx (0, 0) = 1.
k→0 k

Thus

fxy (0, 0) 6= fyx (0, 0).

By directly computing fxy , fyx for (x, y) 6= (0, 0), one observes that these are

not continuous at (0, 0).

24
 (In the following Hf (a, b) denotes the Hessian matrix of a sufficiently smooth

function f at the point (a, b)).

(5) (i) We have

 
 12 0 
 
fx (−1, 2) = 0 = fy (−1, 2); Hf (−1, 2) = 

.

 
0 48
D = 12 × 48 > 0, fxx (−1, 2) = 12 > 0 ⇒ (−1, 2) is a point of local minimum of f .

(ii) We have
 
 6 −2 
 
fx (0, 0) = 0 = fy (0, 0); Hf (0, 0) = 

.

 
−2 10
D = 60 − 4 > 0, fxx (0, 0) = 6 > 0 ⇒ (0, 0) is a point of local minimum of f .

(x2 +y 2 ) (x2 +y 2 )
(8) (i) fx = e− 2(2x − x3 + xy 2 ) , fy = e− 2 (−2y + y 3 − x2 y) .
√ √
Critical points are (0, 0), (± 2, 0), (0, ± 2).
 
 2 0 
Hf (0, 0) =   ⇒ (0, 0) is a saddle point of f .
0 −2
 
4
√  −e 0  √
Hf (± 2, 0) =   ⇒ (± 2, 0) is a point of local maximum of f .
0 − 4e
 
4
√  e 0  √
Hf (0, ± 2) =   ⇒ (0, ± 2) is a point of local minimum of f .
0 4e

(ii) fx = 3x2 −3y 2 and fy = −6xy imply that (0, 0) is the only critical point of f.

Now,  
 0 0 
Hf (0, 0) =  .
0 0

25
 Thus, the standard derivative test fails.

However, f (±, 0) = ±3 for any  so that (0, 0) is a saddle point of f.

(9) From f (x, y) = (x2 − 4x) cos y (1 ≤ x ≤ 3, −π/4 ≤ y ≤ π/4), we have

fx = (2x − 4) cos y and fy = −(x2 − 4x) sin y.

Thus the only critical point of f is P = (2, 0); note that f (P ) = −4.
(x2 −4x)
Next, g± (x) ≡ f (x, ± π4 ) = √
2
(1 ≤ x ≤ 3) has x = 2 as the only critical point
−4
so that we consider P± = (2, ± π4 ); note that f (P± ) = √
2
.

We also need to check g± (1) = f (1, ± π4 ) (≡ f (Q± )) and g± (3) = f (3, ± π4 ) (≡

−3 −3
f (S± )); note that f (Q± ) = √
2
, f (S± ) = − √ 2
.

Next, consider h(y) ≡ f (1, y) = −3 cos y (−π/4 ≤ y ≤ π/4). The only critical

point of h is y = 0; note that h(0) = f (1, 0) (≡ f (M )) = −3. (h(±π/4) is just

f (Q± )).

Finally, consider k(y) ≡ f (3, y) = −3 cos y (−π/4 ≤ y ≤ π/4). The only critical

point of k is y = 0; note that k(0) = f (3, 0) (≡ f (T )) = −3. (k(±π/4) is just

f (S± )).

Summarizing, we have the following table:

Points P+ P− Q+ Q− S+ S− T P M

Values − √42 − √42 − √32 − √32 − √32 − √32 -3 -4 -3

By inspection one finds that

fmin = −4 is attained at P = (2, 0) and

fmax = − √32 at Q± = (1, ±π/4) and at S± = (3, ±π/4).

26
(10) Consider 5(T + λg) = 0, g = 0, where

T (x, y, z) = 400xyz and g(x, y, z) = x2 + y 2 + z 2 − 1.

Thus one has

400yz + 2λx = 0, 400xz + 2λy = 0, 400xy + 2λz = 0, so that

400xyz = −2λx2 = −2λy 2 = −2λz 2 .

If λ 6= 0, then x = ±y, y = ±z and z = ±x; combining this with g = 0 one obtains

  √ √
(x, y, z) = ± √13 , ± √13 , ± √13 ( with the corresponding λ being either −200/ 3 or 200/ 3).

If λ = 0, then yz = xz = xy = 0; combining this with g = 0 one obtains

(x, y, z) = (±1, 0, 0) or (0, ±1, 0) or (0, 0, ±1).

Thus we need to check T at 8 + 6 = 14 points. The first set of eight points gives
400 400
T = √
3 3
or − √ ,
3 3
and the second set of six points gives T = 0. Hence the
400
highest temperature on the unit sphere is √ .
3 3

(11) Consider

5f + λ 5 g1 + µ 5 g2 = 0, g1 = 0 = g2 , where

f (x, y, z) = xyz, g1 (x, y, z) = x + y + z − 40, g2 (x, y, z) = x + y − z.

Thus one has

yz + λ + µ = 0, zx + λ + µ = 0, xy + λ − µ = 0. (∗)

Now, g1 = 0 = g2 gives z = 20 and the first two equalities in (∗) then give x = y;

since one has x + y = 20 in view of g2 = 0, one is led to x = y = 10, z = 20.

Then f (10, 10, 20) = 2000 is the maximum value of f subject to the given

constraints. (That 2000 is the maximum value (and not the minimum value)

of f under the given constraints can be deduced by checking the value of f at

some other point satisfying the constraints such as (x, y, z) = (5, 15, 20)).

27
(12) Consider

5f + λ 5 g1 + µ 5 g2 = 0, g1 = 0 = g2 , where

f (x, y, z) = x2 + y 2 + z 2 , g1 (x, y, z) = x + 2y + 3z − 6, g2 (x, y, z) = x + 3y + 4z − 9.

Thus one has

2x + λ + µ = 0 (A)

2y + 2λ + 3µ = 0 (B)

2z + 3λ + 4µ = 0 (C)

x + 2y + 3z − 6 = 0 (D)

x + 3y + 4z − 9 = 0 (E)

Considering (E)−(D), one obtains y+z = 3; using this and considering (B)+(C)

one next obtains

5λ + 7µ = −6 (F )

Considering 2 × (E) − 3 × (D), one obtains x + z = 0; using this and considering

(A) + (C) one next obtains

4λ + 5µ = 0 (G)

Solving (F ) and (G) for λ and µ, one has

λ = 10 and µ = −8.

It follows now from (A), (B) and (C) that

x = −1, y = 2 and z = 1.

Then f (−1, 2, 1) = 6 is the minimum value of f subject to the given constraints.

(That 6 is the minimum value (and not the maximum value) of f under the

given constraints can be deduced by checking the value of f at some other

point satisfying the constraints such as (x, y, z) = (−3, 0, 3)).

28
 Solutions to Tutorial Sheet 7

Z e Z 1 
(1) (i) dx dy
1 ln y
Z 1 Z 1 
(ii) f (x, y)dy dx.
−1 x2
Z π Z y  Z π
sin(y)
(2) (i) dx dy = sin(y)dy = 2.
0 0 y 0
Z 1 Z x  Z 1
2 xy 2
(ii) x e dy dx = x(ex − 1)dx
0 0 0
1 1 1
= (e − 1) − = (e − 2).
2 2 2
Z 2 Z 2 Z πx 
−1 −1 dy
(iii) (tan (πx) − tan (x))dx = 2
dx
0Z Z Z 2 Z y 0  1 + Zy 2π Z 2
x 
d(x, y) dx dx
= 2
= 2
dy + 2
dy
R1 +R2 1 + y 0 y/π 1 + y 2 y/π 1 + y
Z 2 Z 2π
1 ydy y dy
= (1 − ) + (2 − )
0 π 1 + y2 2 π 1 + y2
π−1 1
= ln(1 + y 2 )|20 + 2 tan−1 y|2π 2 − ln(1 + y 2 )|2π
2
2π 2π 
2

π−1 −1 −1 1 (4π + 1)
= ln 5 + 2(tan 2π − tan 2) − ln .
2π 2π 5
Z Z Z 1 Z 2x  Z 1
x2 x2 2
(3) e d(x, y) = e dy dx = 2xex dx = e − 1.
D 0 0 0

(4) Put
u−v u+v
x= ,y = .
2 2
Then the rectangle R = {π ≤ u ≤ 3π, −π ≤ v ≤ π}

in the uv-plane gets mapped to D, a parallelogram in the xy-plane.

Further,  
 
 1/2 −1/2  1
J = =
 
 2
 1/2 1/2 


and then
Z Z Z Z
2 2 1
(x − y) sin (x + y)dxdy = v 2 sin2 (u) dudv
D R 2

29
 π 3π
π3 π4
Z  Z   
1 2 2 1
= v dv sin (u)du = 2× (π) = .
2 −π π 2 3 3
(5) Put

u
x= , y = uv.
v
Then the rectangle R = {1 ≤ u ≤ 3, 1 ≤ v ≤ 2} in the uv-plane gets mapped to D

in the xy-plane.

Further,  
 
 1/v −u/v 2  2u
J = =
 
v
 v u
 

and then
Z Z Z Z Z 3  Z 2 
2u dv
d(x, y) = Area(D) = dudv = 2udu = 8 ln 2.
D R v 1 1 v

(6) (i) Setting

x = ρ cos(θ), y = ρ sin(θ), 0 ≤ ρ ≤ r, 0 ≤ θ ≤ 2π,

and using J = ρ, we get

Z Z Z 2π Z r
−(x2 +y 2 ) 2 2
e d(x, y) = e−ρ ρdρdθ = π(1 − e−r ).
D(r) 0 0

Therefore, letting r → ∞, we obtain the limit to be π.

π 2
 π
(ii) By symmetry, the required limit is lim 1 − e−r = .
r→∞ 4 4
(iii) Let

I(r) = {|x| ≤ r, |y| ≤ r} .

Then
Z Z Z Z
−(x2 +y 2 ) 2 +y 2 )
e d(x, y) ≤ e−(x d(x, y)
D(r) I(r)
Z Z
2 2
≤ √
e−(x +y )d(x,y) .
D(r 2)

Therefore, letting r → ∞, we obtain the limit to be π using the Sandwich theorem.

30
 π
(iv) The required integral, being one-fourth of the integral in (iii), is .
4
(7) By symmetry, the given volume is 8 times the volume in the positive octant.

In that octant the volume lies above the region Q = {x ≥ 0, y ≥ 0, x2 + y 2 ≤ a2 }

and underneath the cylinder x2 + z 2 = a2 .

Therefore, √ √ ! !
a a2 −x2 a2 −x2
16a3
Z Z Z
V =8 1dz dy dx = .
0 0 0 3

(8)

√ √ p
D = {(x, y, z)| − 1 ≤ x ≤ 1, − 1 − x2 ≤ y ≤ 1 − x2 , x2 + y 2 ≤ z ≤ 1}.

(9) √ √ !
Z 2 Z 2−x2 Z 2 
I= xdz dy dx.
0 0 x2 +y 2

We can also write the region of integration D as

√ p
D = {(x, y, z)|0 ≤ z ≤ 2, 0 ≤ y ≤ z, 0 ≤ x ≤ z − y 2 }.

Thus
Z 2 Z √
z Z √z−y2 ! √
!
8 2
I= xdx dy dz = .
0 0 0 15

(10) (i) Using cylindrical coordinates, one has

Z 1 Z 2π Z 1
I= (z 2 r2 )rdrdθdz = π/3.
−1 0 0

(ii) Using spherical coordinates, one has

2π π 1
4π(e − 1)
Z Z Z
I= (exp(r3 )r2 sin φdrdφdθ = .
0 0 0 3

31
 Solutions to Tutorial Sheet 8

i ∂(f g)
i ∂f ∂g
P P P
(1) (i) ∇(f g) = ∂x
= ∂x
g+ if ∂x = g∇f + f ∇g.
∂f n
(ii) Since ∂x
= nf n−1 ∂f
∂x
, hence ∇f n = nf n−1 ∇f .

(iii) Since,
   
∂ f −2 ∂f ∂g
=g g −f ,
∂x g ∂x ∂x
  X    X X ∂g 
f ∂ f −2 ∂f
∇ = i =g g i −f i ,
g ∂x g ∂x ∂x
which is the desired result.

(2) (i) Note that

∂r ∂(x2 + y 2 + z 2 )1/2
= = x/r.
∂x ∂x
∂r ∂r
Similarly, ∂y
= y/r and ∂z
= z/r. Now,
∂rn ∂r
= nrn−1 = nrn−2 x, etc.
∂x ∂x
Hence, ∇(rn ) = nrn−2 r.

(ii) Letting n = −1 in (i) we have ∇(r−1 ) = −r−3 r. Hence, a · ∇(r−1 ) =

−r−3 (a · r).

(iii) First we compute ∇(a · ∇(r−1 )):

∂ a · r ∂  a1 x + a2 y + a3 z  a1 ∂r−3 a1
3
= 3
= 3
+ a · r = 3 − 3xr−5 (a · r).
∂x r ∂x r r ∂x r
Hence,

b · ∇(a · ∇(r−1 )) = −r−3 (a · b) + 3r−5 (a · r)(b · r).

P ∂
P ∂f P ∂v1
(3) (i) ∇ · (f v) = ∂x
(f v1 ) = v
∂x 1
+f ∂x
= ∇f · v + f ∇ · v.
P 
∂
i × ( ∂f i × (f ∂v ∂f ∂
P P P P
(ii) ∇ × (f v) = i× ∂x
(f v) = ∂x
v) + ∂x
)= ∂x
i ×v+f i× ∂x
v

= ∇f × v + f (∇ × v).

32
 P  PP 
∂ ∂ ∂2v
P
(iii) ∇ × (∇ × v) = i × j× = ∂x
i × (j ×
∂y
v ∂x∂y
)
   
∂2v
P 2
− ∂∂xv2 = j j ∂y
P P P ∂ P ∂
= i j j i· ∂x∂y i i· ∂x v −∆v = ∇(∇·v)−∆v.

(iv) Since
P  P  
P ∂ ∂g ∂ ∂g
∇ · (f ∇g) = i· ∂x
f ∂y j = ∂x
f ∂x
= ∇f · ∇g + f ∆g,

we have

∇ · (f ∇g) − ∇ · (g∇f ) = f ∆g − g∆f .

P  P P 
∂ ∂ ∂2v
P
(v) ∇ · (∇ × v) = i · ∂x j × ∂y v = i· j× ∂x∂y
PP ∂2v
P  ∂2v ∂2v

= (i × j) · ∂x∂y = k · ∂x∂y − ∂y∂x = 0,

by the equality of mixed partials.

P  PP P  ∂2f 
∂2f ∂2f
∂
j ∂f
P
(vi) ∇ × (∇f ) = i × ∂x ∂y
= (i × j) ∂x∂y
= k ∂x∂y − ∂y∂x
= 0.

(vii) Note that

X ∂f X ∂g X X ∂f ∂g
g∇f × f ∇g = g i ×f j = f g(i × j)
∂x ∂y ∂x ∂y
 
f g ∂f ∂g
− ∂f ∂g
P
= ∂x ∂y ∂y ∂x
k
 
 
 i j k 
 
  P  ∂f ∂g ∂g ∂f 
= fg 
 ∂f ∂f ∂f  = fg
∂y ∂z
− ∂y ∂z i.
∂x ∂y ∂z 
 
 ∂g ∂g ∂g 
 ∂x ∂y ∂z 
Hence,
 
∂
if g ∂f ∂g ∂g ∂f
P P
∇ · (g∇f × f ∇g) = i · ∂x ∂y ∂z
− ∂y ∂z
P ∂   ∂f ∂g ∂g ∂f  P ∂g  ∂f ∂g ∂g ∂f  P ∂f  ∂f ∂g ∂g ∂f

= ∂x
f g ∂y ∂z − ∂y ∂z = f ∂x ∂y ∂z − ∂y ∂z + g ∂x ∂y ∂z − ∂y ∂z
  P  2 
P ∂ 2 f ∂g ∂ 2 f ∂g ∂ g ∂f ∂ 2 g ∂f
+ f g ∂x∂y ∂z
− ∂x∂z ∂y
+ f g ∂x∂z ∂y
− ∂y∂x ∂z
.

Each of the four sums vanishes individually.

∂f (r) 0 ∂r 0
(4) (i) Since ∂x
= f (r) ∂x = f (r)x/r, we have
X ∂  0 x
div(∇f (r)) = f (r)
∂x r

33
 0
X 00 x2 X f (r) X x2 0 00 2 0
= f (r) 2 + − 3
f (r) = f (r) + f (r).
r r r r
∂ 2
(ii) div(rn r) = (rn x) = (rn + nrn−1 xr ) = 3rn + nrn = (3 + n)rn .
P P
∂x
 n+2 
(iii) Note that ∇ rn+2 = rn r, by exercise 1(i). So,
 rn+2 
curl(rn r) = curl(grad ) = 0,
n+2

by exercise 3(vi). If n = −2, then

∇(ln r) = r−2 r

and hence,

curl(r−2 r) = curl(∇ ln r) = 0.

(iv) Using part (i) it follows that

d2  1  2  d −1 
div(∇r−1 ) = + (r ) = 0.
dr2 r r dr
X ∂
X ∂u X ∂v
(5) (i) ∇ · (u × v) = i· (u × v) = i·( × v) + i · (u × )
∂x ∂x ∂x
X ∂u X ∂v
= (i × )·v− (i × ) · u = v · curl u − u · curl v.
∂x ∂x
(Def: A vector-field u is said to be irrotational if ∇ × u = 0. A vector-

field u is said to be solenoidal if ∇ · u = 0. We have now proved that if

u and v are irrotational then u × v is solenoidal. )

X ∂ X ∂u
X ∂v

(ii) ∇ × (u × v) = i× (u × v) = i× ×v + i× u×
∂x ∂x ∂x
X ∂u X ∂u
X ∂v
X ∂v
= (i · v) − i· v+ i· u− (i · u)
∂x ∂x ∂x ∂x
= (v · ∇)u − (∇ · u)v + (∇ · v)u − (u · ∇)v.
X ∂ X ∂u
X ∂v

(iii) ∇(u · v) = i (u · v) = i ·v + i ·u
∂x ∂xh ∂x
X h ∂u
∂u i X ∂v
∂v i
= v· i − (v · i) + u· i − (u · i) +(v · ∇)u + (u · ∇)v
∂x  X ∂x ∂  ∂x ∂x
X ∂ 
=v× i× u +u× i× v + (v · ∇)u + (u · ∇)v
∂x ∂x
= v × (∇ × u) + u × (∇ × v) + (v · ∇)u + (u · ∇)v.

34
(6) (i) Let w = f u, where f is a scalar field and u is a constant vector. Then,

w · (∇ × w) = f u · (∇ × f u) = f u · (f ∇ × u + ∇f × u) (using 3(ii) ) =

f 2 (u · (∇ × u)) + f u · ∇f × u = 0 (using 3(v)).

(ii) Here r = xi + yj + zk. Thus, ∇ × v = ∇ × (w × r) = (r · ∇)w − (w · ∇)r +

(∇ · r)w − (∇ · w)r (using 5(ii) ) = (∇ · r)w − (w · ∇)r = 3w − w = 2w.

(iii) Let f = ρ−1 . Then, using problem 3(ii) and 3(vi),

∇ × v = ∇ × (f ∇p) = f (∇ × ∇p) + ∇f × ∇p = −f 2 (∇ρ × ∇p).

Hence

v · (∇ × v) = −f 3 ∇p · (∇ρ × ∇p) = 0.

(7) Parameterize the curve C as r(t) = ti + t2 j. Then, r0 (t) = i + 2tj. Thus

Z Z 1
 2
(t − 2t3 )i + (t4 − 2t3 )j · (i + 2tj)dt

f · dr =
C −1
Z 1
14
= (t2 − 2t3 ) + 2t(t4 − 2t3 )dt = − .
−1 15

(8) A parametrization of the ellipse is given by

r(θ) = a cos θi + b sin θj, 0 ≤ θ ≤ 2π,

and

r0 (θ) = −a sin θi + b cos θj.

Thus,

f (a cos θ, b sin θ)·r0 (θ) = [(a2 cos2 θ+b2 sin2 θ)i+(a cos θ−b sin θ)j]·[−a sin θi+b cos θj]

and
Z Z 2π
(−a3 cos2 θ sin θ − ab2 sin3 θ) + (ab cos2 θ − b2 sin θ cos θ) dθ
 
f (x, y) · dr =
C 0

= πab.

35
(9) A parametrization of the curve is given by

r(θ) = a cos θi + a sin θj, 0 ≤ θ ≤ 2π.

Thus, x(θ) = a cos θ, y = a sin θ, and the required integral is

2π 2π
a2 (cos θ + sin θ)(− sin θ) + a2 (sin θ − cos θ) cos θ −a2
Z Z
dθ = dθ = −2π.
0 a2 0 a2

(10) For the curve z = xy, x2 + y 2 = 1, we can use the parametrization

x = cos θ, y = sin θ, z = sin θ cos θ, 0 ≤ θ ≤ 2π.

Thus,
Z Z
ydx + zdy + xdz = ydx + (xy)dy + x(xdy + ydx)
Z 2π
= [sin θ(− sin θ) + sin θ cos θ cos θ + cos2 θ cos θ + sin θ cos θ(− sin θ)]dθ
0
Z 2π
= [− sin2 θ + sin θ cos2 θ + cos3 θ − sin2 θ cos θ]dθ
0
Z 2π Z 2π
2
=− sin θdθ + sin θ cos2 θdθ
0 0
Z 2π Z 2π
+ cos3 θdθ − sin2 θ cos θdθ
0 0

= −π.

36
 Solutions to Tutorial Sheet 9

(1) Let

r(t) = a cos ti + a sin tj + ctk.

Then r0 (t) = −a sin ti + a cos tj + ck and

0 √
kr (t)k = a2 + c 2 .

Hence
Z t√ √
s= a2 + c2 du = ( a2 + c2 )t.
0

Thus the arc length parametrization is

   
s s cs
r(s) = a cos √ i + a sin √ j+ √ k.
2
a +c 2 2
a +c 2 a + c2
2

(2)
+1 Z +π/4
x2 dx
Z Z
1
=− 2 2
=− sin2 θ = −π/4 + ,
C1 −1 (1 + x ) −π/4 2
Z Z +1 Z +π/4
dy
=− 2 )2
=− cos2 θdθ = −π/4 − 1/2,
C2 −1 (1 + y −π/4
Z +1 2
−x2 dx
Z
1
= 2
= −π/4 + ,
C3 −1 (1 + x ) 2
Z +1
−dy
Z
= 2 2
= −π/4 − 1/2.
C4 −1 (1 + y )

Hence
Z Z Z Z Z
= + + + = −π.
C C1 C2 C3 C4

(3) A parametrization of C is

r(t) = cos ti + sin tj, 0 ≤ t ≤ 2π.

37
 Note that the outward unit normal to the circle at r(t) is the radial vector

n = r(t). Also,

∇(x2 − y 2 ) = 2x i − 2y j.

Thus
I Z 2π
2 2
∇(x − y ) · dn = (2 cos t i − 2 sin tj) · (− sin ti + cos tj)dt
C 0
Z 2π
= (−2 sin 2t)dt = 0.
0

(4) Parameterize C as

r(t) = t i + t3 j, 0 ≤ t ≤ 2.

Then r0 (t) = i + 3t2 j. Since ∇(x2 − y 2 ) = 2ti − 2t3 j, we have

Z Z 2
2 2
∇(x − y ) · dr = (2t − 6t5 )dt = 4 − 64 = −60.
C 0

(5) The required integral is

Z Z Z Z
dx + dy dx + dy dx + dy dx + dy
= + + + .
C1 |x| + |y| C2 |x| + |y| C3 |x| + |y| C4 |x| + |y|

Along C1 : x + y = 1 and |x| + |y| = x + y = 1. Thus

Z Z 0 Z 0
dx + dy
= dx − dx = 0.
C1 |x| + |y| 1 1

Along C2 : −x + y = 1 and |x| + |y| = −x + y = 1. Thus

Z Z −1 Z −1
dx + dy
= dx + dx = −2.
C2 |x| + |y| 0 0

Along C3 : x + y = −1 and |x| + |y| = −x − y = 1. Thus

Z Z 0 Z 0
dx + dy
= dx − dx = 0.
C3 |x| + |y| −1 −1

Along C4 : x − y = 1 and |x| + |y| = x − y = 1. Thus

Z Z 1 Z 1
dx + dy
= dx + dx = 2.
C4 |x| + |y| 0 0

38
 Hence
Z
dx + dy
= 2 − 2 = 0.
C |x| + |y|

(6)
Z Z
Work W = (xyi + x6 y 2 j) · (idx + jdy)
F · dr =
ZC1 C
Z 1
b+1
= ax dx + (a2 x2b+6 )(abxb−1 )dx
0 0
a a3 b
= +
b + 2  3b + 6 
a2 b 3 + a2 b
 
a
= 1+ =a .
b+2 3 3(b + 2)

dW (b+2)a2 −(3+a2 b)
This will be independent of b iff db
= 0 iff 0 = (b+2)2
iff a =
q
3
2
(as a > 0).

(7) First we observe that the cylinder is given by

 a 2 a2
x− + y2 = .
2 4

From the equations of the sphere and the cylinder we have that, on the inter-

section C,

z 2 = a2 − ax.

Noting the requirement z ≥ 0, a parametrization of C is given by

a a a θ
x= + cos θ, y= sin θ, z = a sin ; 0 ≤ θ ≤ 2π.
2 2 2 2

39
 Then
2π
a2
Z Z  
a θ a
F · dr = [ sin θ (− sin θ) + (a2 sin2 )( cos θ)
2
C 0 4 2 2 2
 2 2 2

a a 2 a a θ
+ + cos θ + cos θ ( cos )]dθ
4 4 2 2 2
Z 2π 3 3 3 3
a a θ a θ a θ
= [− sin3 θ + sin2 cos θ + cos + cos2 θ cos
0 8 2 2 8 2 8 2
3
a θ
+ cos θ cos ]dθ
4 2
πa3
= − .
4
∂f1 ∂f2
(8) = 3x, = 3x2 y where (f1 , f2 ) = f. Now
∂y ∂x
∂f1 ∂f2
= iff 3x = 3x2 y
∂y ∂x
iff either x = 0 or xy = 1.

Since the sets {(x, y)|x = 0}, {(x, y)|xy = 1} are not open, F(x, y) is not the

gradient of a scalar field on any open subset of R2 .

(9)

∂f1 y 2 − x2 ∂f2
= 2 2 2
= on R2 \ {(0, 0)}.
∂y (x + y ) ∂x

However, F 6= ∇f for any f . Indeed, let C to be the unit circle x2 + y 2 = 1,

oriented anticlockwise. Then one has

I Z 2π
F · dr = (− sin ti + cos tj) · (− sin ti + cos tj)dt
C 0

= 2π 6= 0

(10) Suppose v = ∇φ for some φ.

Then
∂φ
= 2xy + z 3 ⇒ φ(x, y, z) = x2 y + z 3 x + f (y, z)
∂x

40
 for some f (y, z). Assuming f has partial derivatives, we get

∂φ ∂f
= x2 + = x2
∂y ∂y
∂f
so that = 0
∂y

and f (y, z) depends only on z

∂φ
Let f (y, z) = g(z). Then φ(x, y, z) = x2 y + z 3 x + g(z) ⇒ = 3z 2 x + g 0 (z) =
∂z
3z 2 x ⇒ g 0 (z) = 0. Let us select g(z) = 0. It can be checked that φ(x, y, z) =

x2 y + z 3 x satisfies ∇φ = v.

Hence
I
v · dr = 0
C

for every smooth closed curve C.

(11) F(x, y, z) = f (r)r = f (r)x i + f (r)y j + f (r)z k.

Since

r = (x2 + y 2 + z 2 )1/2 ,

∂r x ∂r y ∂r z
= , = , = .
∂x r ∂y r ∂z r

If F is to be ∇φ for some φ, then we must have φx = f (r)x, φy = f (r)y, φz =

f (r)z; that is,

x ∂r
φx = xf (r) = rf (r) = rf (r),
r ∂x
y ∂r
φy = yf (r) = rf (r) = rf (r),
r ∂y
z ∂r
φz = zf (r) = rf (r) = rf (r).
r ∂z
Z r
Now it can be seen that φ(x, y, z) = tf (t)dt, with some t0 fixed, satisfies
t0
all the desired equations.

41
 Solutions to Tutorial Sheet 10

(1) We have to show that

ZZ I
(gx − fy ) dxdy = (f dx + gdy).
R ∂R

(i) !
ZZ Z 1 Z 1−x2
1
LHS = 4xy dxdy = 4xy dy dx = .
R 0 0 3
Observe that the boundary of R consists of three smooth curves: a seg-

ment of the x-axis, a part of the parabola y = 1 − x2 and a segment of the

y-axis. The integral on the RHS vanishes on both the axes. We choose

the parametrization t 7→ (t, 1−t2 ), (t ∈ [0, 1]) for the part of the parabola

traced in the opposite direction. This gives

I
RHS = (−xy 2 dx + x2 y dy)
∂R
Z 1
1
=− [−t(1 − t2 )2 + t2 (1 − t2 )(−2t)] dt = .
0 3
(ii)
ZZ Z 1 Z x 
x x
LHS = (e + 2x − 2x) dxdy = e dy dx = 1.
R 0 0

and
I
RHS = [2xy dx + (ex + x2 ) dy]
∂R
Z 1 Z 0
= (e + 1)dy + (3t2 + et ) dt = e + 1 − e = 1.
0 1

(Observe that f and dy are zero on the horizontal segment of the curve,

whereas on the vertical segment dx = 0.)

(2) (i) Here

f (x, y) = y 2 ; g(x, y) = x.

42
 Therefore, the given path integral is equal to
ZZ Z 2Z 2 Z 2
(1 − 2y) dxdy = (1 − 2y)dy dx = 4 − 4 dx = 4 − 8 = −4.
R 0 0 0

(ii) Here
ZZ ZZ Z 1 Z 1
(1 − 2y) dxdy = dxdy + (−2y)dy dx = 4 + 0 = 4.
R R −1 −1

(iii) Here
"Z √ #
ZZ ZZ Z 2 2−x2
(1 − 2y) dxdy = dxdy + √
(−2y)dy dx = 4π + 0.
R R −2 − 2−x2

3πa2
(3) (i) A = 2
. (ii) A = a2 /2.

(4) (i) The required area is bounded by the curves

C1 : r = a(1 − cos θ), 0 ≤ θ ≤ π/2

and C2 which is a portion of the y-axis. In any case, the required area is

equal to
I
1
r(θ)2 dθ.
2 C

Since θ is a constant along the y-axis, this integral is equal to

1 π/2 2 a2
Z
r dθ = (3π − 8).
2 0 8

(ii) The required area is

I
1
xdy − ydx.
2 C

Here the boundary curve consists of the interval [0, 2π] and the cycloid

above traced in the opposite direction. But the integrand is zero on the

x-axis, since both y and dy vanish there. Hence the required area is

a2 2π
Z
− (t − sin t)d(1 − cos t) − (1 − cos t)d(t − sin t) = 2πa2 .
2 0

43
 (iii) Here we use the polar coordinate form as in the previous exercise:
I
1
A= r2 dθ.
2 C

(This formula follows from Green’s Theorem.)

Since θ is a constant on the two axes, this integral is equal to

1 π/2
 
1 3π − 8
Z
2
(1 − 2 cos θ) dθ = .
2 0 2 2

(5) Observe that

2
−y 2 2 −y 2 x2 e−y
xe dx + (−x ye )dy = d( ).
2
Hence the integral of this term along a closed path vanishes. So the given

integral is equal to
I
dy
C x2 + y2.
We compute this directly. Observe that dy = 0 along the two horizontal

parts. But then the integral along one vertical segment cancels with that on

the other since the integrands are the same and the segments are traced in

the opposite direction. So the value of the required integral is equal to 0.

(6) Take f = −y 3 and g = x3 and apply Green’s theorem. We get

ZZ
RHS = (3x2 + 3y 2 ) dxdy = 3I0 .
R

(7)

u · v = (ax − ay )a + (bx − by )b.

Therefore taking f = (a2 + b2 )/2 = g, we see that

RR RR
D
u · v dxdy = D
(gx − fy ) dxdy
H
= ∂D
(f dx + g dy)
1
H
= 2 ∂D
(1 + y 2 )(dx + dy) = 0.

44
 (The last equality follows by considering the parametrizaton (cos θ, sin θ), 0 ≤

θ ≤ 2π). Likewise we see that

ZZ I Z 2π
u · w dxdy = (ab)(dx + dy) = − sin2 θ dθ = −π.
D ∂D 0

(8) Since ∇2 (x2 − y 2 ) = 0, using one of Green’s identities (refer to (9),(i)) one has

∂(x2 − y 2 )
I I ZZ
2 2
∇(x − y ).nds = ds = ∇2 (x2 − y 2 ) dxdy = 0.
C C ∂n R

(9) (a)
I
2 ∂w
∇ w = 0 hence, dS = 0
C ∂n

(b) Put H = F − G. Then curl H = 0. Since D is simply connected, there

exists u such that grad u = H. Now div H = 0 implies that ∇2 u = 0, i.e.,

∂u
u is harmonic. Finally H · n = 0, i.e., ∇u · n = ∂n
= 0 on the boundary

implies, using (9),(ii), that u is a constant. But then H = grad u = 0

and hence F = G.

(10) (i) There are two distinct cases to be considered:

Case (a): Suppose the curve does not enclose the origin. Take

y x
f (x, y) = , g(x, y) = 2
x2 +y 2 x + y2

and apply Green’s theorem in the region R boundedby C. So the integral

is equal to
ZZ
(gx − fy ) dxdy.
R

A simple computation show that gx = fy and hence the integral vanishes.

Case (b): Suppose the curve encloses the origin, i.e, (0, 0) ∈ R. (Now the

above argument does not work!) We choose a small disc D around the

origin contained in R and apply Green’s theorem in the closure of R0 =

45
 R \ D. As before, the double integral vanishes. But since the boundary

of R0 consists of C and −∂D it follows that

y dx − x dy y dx − x dy
I I
= .
C x2 + y 2 ∂D x2 + y 2

We can compute this now by using polar coordinates and see that this is

equal to −2π.

(ii) Here again, we take D to be a small disc of radius  around the origin

and contained in the square R and apply Green’s theorem in the closure

of R \ D. Taking

x2 y x3
f (x, y) = , g(x, y) = ,
(x2 + y 2 )2 (x2 + y 2 )2

we once again observe that the double integral vanishes since gx = fy =

(x2 +y 2 )(x4 −3x2 y 2 )
(x2 +y 2 )4
. Hence the given line integral is equal to the correspond-

ing line integral taken over the boundary of D. This can be computed

by using the parametrizaton ( cos θ,  sin θ), 0 ≤ θ ≤ 2π. The answer is

−π/4.

(iii) We have
∂(ln r) y ∂(ln r) x
= 2 and = 2 .
∂y x + y2 ∂x x + y2
By part (i), the required line integral is −2π.

46
 Solutions to Tutorial Sheet 11

(1) On S,

1
z = (4 + y − x) = h(x, y) so that
2
1
r(x, y) = xi + yj + (4 + y − x)k, (x, y) ∈ R2
2

can be chosen as one parametrization.

The normal vector is

1 1
rx × ry = i − j + k.
2 2

(ii) For S : y 2 + z 2 = a2 , a parametrization is

r(u, v) = ui + a sin vj + a cos vk, u ∈ R, 0 ≤ v ≤ 2π.

The normal vector is

ru × rv = a sin vj + a cos vk.

i+j+k
(iii) If e = √ , then e is a unit vector along the axis of the cylinder.
3
Consider the planar cross-section of the cylinder through the origin O. This is a
−→
circle C of radius 1. Fix a point P on C. Then OP is a unit vector, say u.

Let v = e × u. Then a point on the cylinder is representable as

r(θ, t) = cos θu + sin θv + te, 0 ≤ θ ≤ 2π, t ∈ R.

The normal vector is

rθ × rt = cos θu + sin θv.

47
(2) (a) The area SA of the surface S with projection R on the xy-plane is given by
ZZ
SA = sec γdxdy
R

where γ is the acute angle between n and k at a generic point on the surface.

Thus, if this angle is the same at every pointon S, we have

ZZ
SA = sec γ dxdy = sec γSAxy ,
R

where SAxy is the area of R. Hence,

SAxy = SA cos γ.

(b) By (a) above, one has (for appropriate α, β and γ)

S1 = S cos α,

S2 = S cos β,

S3 = S cos γ.

Thus S12 + S22 + S32 = S 2 (cos2 α + cos2 β + cos2 γ) = S 2 in view of the fact

that cos α, cos β, cos γ are the direction cosines of n.

(3) There are two pieces of the surface - one below and one above the xy-plane, both

having the same area. Let S be the upper piece. Then one has
ZZ q
Area(S) = 1 + zx2 + zy2 dxdy,
T

where T is the disc

n a a o
{x2 + y 2 ≤ ay} = x2 + (y − )2 ≤ ( )2 ,
2 2
p
and z = a2 − x2 − y 2 . Since

x y
zx = − and zy = − ,
z z

48
 it follows that
ZZ ZZ
adxdy adxdy
Area(S) = = p .
T z T a2 − x 2 − y 2
Now T is described in polar coordinates by

x = r cos θ, y = r sin θ; 0 ≤ θ ≤ π, 0 ≤ r ≤ a sin θ.

Therefore,
Z π Z a sin θ 
ardr
Area(S) = √ dθ
0 0 a2 − r 2
Z π √
=a [− a2 − r2 ]|a0 sin θ dθ
0
Z π
=a (−a| cos θ| + a)dθ = (π − 2)a2 .
0

Thus the required area is 2(π − 2)a2 .

(4) (i) A point (x, y, z) on the surface satisfies z = x2 + y 2 . (The surface is thus

a portion of a paraboloid of revolution). The given portion lies between the

planes z = 0 and z = 16. u = c gives a horizontal circular section, while v = c

gives a profile curve which is the portion of a half parabola.

(ii) ru × rv = −2u2 (cos vi + sin vj) + uk

R 2π R 4 R4 √ √
(iii) S = v=0 u=0 |ru × rv |dudv = 2π 0 u 4u2 + 1 du = π6 (65 65 − 1).

Therefore, n = 6.

(5) The area of the paraboloid x2 + z 2 = 2ay between y = 0 and y = a is given by

s  2  2
ZZ
∂y ∂y
S= 1+ + dxdz
T ∂x ∂z

where T is the region {z 2 + x2 ≤ 2a2 } in the zx-plane. Hence,

ZZ r
x2 z 2
S= 1 + 2 + 2 dxdz
T a a
Z 2π Z a√2 r
r2
= 1 + 2 rdrdθ
0 0 a

49
 2π √
= (3 3 − 1)a2 .
3
(6) We choose the coordinate system in such a way that the center of the sphere is

located at the origin and the central axis of the circular cylinder coincides

with the z-axis. We consider the case when one plane is cutting the sphere

at height h above the xy-plane and the other plane is cutting the sphere at

depth k below the xy-plane. (Other cases can be treated similarly). We

compute the surface areas S1 and S2 of the ‘upper’ and ‘lower’ caps of the

sphere and subtract their sum from 4πa2 . We are expected to get the result

to be 2πa(h + k).

Note that the plane cutting the sphere at height h above the xy-plane intersects

the sphere in the circle x2 + y 2 = a2 − h2 . A parametrization for the upper

p
cap of the sphere is thus given by r(x, y) = (x, y, a2 − x2 − y 2 ) with (x, y) ∈

D = {(x, y) : x2 + y 2 ≤ a2 − h2 }. We have then

Z Z s
−x −y
S1 = 1 + (p )2 + ( p )2 dxdy
2
a −x −y2 2 2 2
a −x −y 2
D
√
Z Z Z 2π Z a2 −h2
a a
p dxdy = √ rdrdθ
D a2 − x 2 − y 2 0 0 a2 − r 2
= 2πa(−h + a).

Similarly, S2 = 2πa(−k + a); and then 4πa2 − (S1 + S2 ) = 2πa(h + k), as desired.

(7) (i) Note that ru × rv = −2(i + j + k) has negative z-component. Thus, using

ndS = −ru × rv dudv = 2(i + j + k)dudv

and

F · ndS = 2(x + y + z)dudv = 2dudv,

one has
ZZ
F · ndS = 2Area(S ∗ ),
S

50
 where S ∗ is the parametrizing region in the uv-plane. As the components of r

are affine-linear in u and v, S ∗ is also a triangle whose vertices are pre-images

1 1 1 1
of the vertices of S. Now the vertices of S ∗ are (0, 0), ( , ), and ( , − ) so
2 2 2 2
∗ 1 RR 1
that the area of S is , and hence S F · ndS = .
4 2
(ii) The surface S satisfies z = 1 − x − y ≥ 0, x ≥ 0, y ≥ 0. Thus, using

ndS = (−zx , −zy , −1)dxdy

and

F · ndS = (x, y, z) · (−zx , −zy , 1)dxdy = (x + y + z)dxdy = dxdy

and S1∗ = {x + y ≤ 1, x ≥ 0, y ≥ 0} as the parametrizing region, one has

ZZ ZZ
1
F · n dS = dxdy = Area(S1∗ ) = .
S S1∗ 2

(8) A parametrization of S is

r(u, v) = a sin v cos u i + a sin v sin u j + a cos v k, 0 ≤ u ≤ 2π, 0 ≤ v ≤ π

and

ru × rv = a sin v r(u,v)

is the outward normal. The integrand is

F · (ru × rv ) = a4 sin3 v cos v(1 + cos2 u).

Thus the required integral is

Z 2π Z π
a4 sin3 v cos v(1 + cos2 u)dudv
u=0 v=0
Z π  Z 2π 
4 3 2
=a sin v cos v dv (1 + cos u)du = 0.
0 0

51
(9) The hemisphere satisfies
p
z= 1 − x2 − y 2 , x2 + y 2 ≤ 1.

Using
x y
ndS = (−zx , −zy , 1)dxdy = ( , , 1)dxdy
z z
and
x y (1 − 2xy − 2y 2 )
F · ndS = (x, −2x − y, z) · ( , , 1) = dxdy
z z z
and T = {(x, y) : x2 + y 2 ≤ 1} as the parametrizing region, one has

(1 − 2xy − 2y 2 )
ZZ ZZ
F · ndS = p dxdy
S T 1 − x2 − y 2
2π 1
(1 − r2 sin 2θ − 2r2 sin2 θ)rdrdθ
Z Z
2π
= √ = .
0 0 1−r 2 3

(10) The flux through the base T is

ZZ
F · (−k)dxdy = 0,
T

as F · (−k) = −z = 0 along T . The total flux is therefore the same as in the

2π
previous problem, namely, .
3

52
 Solutions to Tutorial Sheet 12

p p
(1) The cone z = x2 + y 2 is parametrized as r(x, y) = xi + yj + x2 + y 2 k. One

has then ndS = (− √ x

, − √ 2y 2 , 1)dxdy.
Further, curlF = (0, 0, 2).
x2 +y 2 x +y
p
(a) If S is the surface lying on the cone z = x2 + y 2 and bounded by the

intersection C of the hemisphere x2 + (y − a)2 + z 2 = a2 , z ≥ 0 with the cone,

then the projection R of S onto the xy-plane is given by x2 +(y−a/2)2 ≤ a2 /4.

Thus one has

ZZ ZZ ZZ
curlF · ndS = 2 dS = 2 dS = πa2 /2.
S R

With the choice of the normal n to S as indicated above, the induced ori-

entation on C is counterclockwise (when viewed from high above). The pro-

jection of C onto the xy-plane can then be described by ( a2 cos θ, a2 + a2 sin θ)

(0 ≤ θ ≤ 2π).

Thus,
I I
F · dr = (x − y)dx + (x + z)dy + (y + z)dz
C IC
1
= (xdy − ydx) + d(yz) + d(x2 + z 2 )
2
IC 2
a
= (xdy − ydx) = 2π = πa2 /2.
C 4

Stokes’ Theorem now stands verified.

p
(b) If S is the surface lying on the cone z = x2 + y 2 and bounded by the

intersection C of the cylinder x2 + (y − a)2 = a2 , z ≥ 0 with the cone, then

the projection R of S onto the xy-plane is given by x2 + (y − a)2 ≤ a2 .

Thus one has

ZZ ZZ ZZ
curlF · ndS = 2 dS = 2 dS = 2πa2 .
S R

53
 With the choice of the normal n to S as indicated above, the induced ori-

entation on C is counterclockwise (when viewed from high above). The pro-

jection of C onto the xy-plane can then be described by (a cos θ, a + sin θ)

(0 ≤ θ ≤ 2π).

Thus,
I I
F · dr = (x − y)dx + (x + z)dy + (y + z)dz
C IC
1
= (xdy − ydx) + d(yz) + d(x2 + y 2 )
2
IC
= (xdy − ydx) = 2πa2 .
C

Stokes’ Theorem now stands verified.

(2) For F = yzi + xzj + xyk,

 
 
 i j k 

 
curl (F) =  ∂ ∂ ∂  = (x − x)i + (y − y)j + (z − z)k = 0.
 ∂x ∂y ∂z 

 yz zx xy 
 

Thus the required line integral is

ZZ
curl (F) · ndS = 0.
S

(3) By Stokes’ theorem, we have

ZZ I I
curl (v) · ndS = v · ds + v · ds
S C1 C2

where C1 is the circle x2 +y 2 = 4, z = −3 with the counterclockwise orientation

when viewed from high above, and C2 is the circle x2 + y 2 = 4, z = 0 with the

opposite orientation. Now,

curl v = (−3zy 2 − 3xz 2 )i + (z 3 − 1)k and

54
 v · ds = ydx + xz 3 dy − zy 3 dz.

I Z 2π
Along C2 , z = 0 ; v · ds = ydx and ydx = − (−4 sin2 θ| dθ = 4π.
C2 0

Along C1 , z = −3 ; dz = 0; v · ds = ydx − 27xdy = d(xy) − 28xdy and

I I Z 2π
d(xy) − 28xdy = −28 4 cos2 θdθ = −112π.
C1 C1 0

Hence
ZZ
curl (v) · ndS = −108π.
S

(4) Note that, to apply Stokes’ Theorem, one would have to work inside U = R3 \

z−axis, as v would not make sense at a point on the z-axis. But there is

no surface in U = R3 \ z−axis whose boundary C is. Hence Stokes’ theorem

cannot be applied.

Using the parametrization (cos θ, − sin θ), (0 ≤ θ ≤ 2π) one has

Z 2π
−ydx + xdy
I I
v · dr = =− dθ = −2π.
C C x2 + y 2 0

(5) Note that

F = (y 2 − z 2 )i + (z 2 − x2 )j + (x2 − y 2 )k,

curl F = (−2y − 2z)i + (−2z − 2x)j + (−2x − 2y)k,

i+j+k
and n = √ .
3
3a
Thus, along the surface S which is part of the plane x + y + z = 2
and which

is bounded by C, one has

2
curlF · n = − √ (y + z + z + x + x + y)
3
4 4 3a
= − √ (x + y + z) = − √ .
3 3 2

55
 Hence
ZZ √ ZZ √
curl F · ndS = −2 3a dS = (−2 3a)(Area of S).
S S

The surface S is a regular hexagon with vertices (a/2, 0, a), (a, 0, a/2), (a, a/2, 0), (a/2, a, 0),
√ √
3 2 3 3 a2
(0, a, a/2), (0, a/2, a). Hence its area is 3 (length of side) = .
2 2 2
9a3
I
Stokes’ theorem then yields that (y 2 −z 2 )dx+(z 2 −x2 )dy +(x2 −y 2 )dz = − .
C 2
(6) We have

F = yi + zj + xk,

curl F = −(i + j + k).

Parametrize the surface lying on the hyperbolic paraboloid z = xy/b and

xy
bounded by the curve C as xi + yj + b
(x2 + y 2 ≤ a2 ) so that ndS =

(− yb , − xb , 1)dxdy and
ZZ ZZ
1
curl F · ndS = (y + x − b)dxdy
S b x2 +y 2 ≤a2
Z 2π Z a
1
= (r sin θ + r cos θ − b)rdrdθ = −πa2
b 0 0

(7) Letting r = xi + yj + zk and using S to denote the planar area enclosed by C,

one has
I
1
I := a(ydz − zdy) + b(zdx − xdz) + c(xdy − ydx)
2 C
I
1
= n × r · ds
2 C
ZZ
1
= ∇ × (n × r) · ndS
2 S
ZZ ZZ
1
= 2n · ndS = dS = Area(S).
2 S S

56
If C is parametrized as u cos t + v sin t (0 ≤ t ≤ 2π), then
I
1
Area(S) = n × r · ds
2 C
1 2π
Z
= n × (u cos t + v sin t) · (−u sin t + v cos t)dt
2 0
1 2π
Z
= n · (u cos t + v sin t) × (−u sin t + v cos t)dt
2 0
1 2π
Z
= n · u × v dt
2 0
u×v
so that, letting n = ||u×v||
, we get
Z 2π
1
Area(S) = ||u × v|| dt = π||u × v||.
2 0

57
 Solutions to Tutorial Sheet 13

(1) We have

R = {(x, y, z) : y 2 + z 2 ≤ x2 , 0 ≤ x ≤ 4} and
[
∂R = S1 S2 , where

S1 : r(x, θ) = xi + x cos θj + x sin θk, 0 ≤ θ ≤ 2π,

S2 : x = 4, y 2 + z 2 ≤ 16.

Along S1 , rx × rθ = xi − x cos θj − x sin θk = xi − yj − zk so that the outward

normal is −rx × rθ = −xi + yj + zk. Thus

ZZ Z 4 Z 2π
F · ndS = (−x2 y 2 + y 2 z 2 + z 2 x2 )dxdθ
S1 0 0
Z 4 Z 2π
= x4 (− cos2 θ + cos2 θ sin2 θ + sin2 θ)dxdθ
0Z 40  Z 2π 
4 2 2 2 2
= x dx (− cos θ + cos θ sin θ + sin θ)
0 0
45 π π
= = 44 .
5 4 5

Also, along S2 , the outward normal (to x = 4 ≡ f (y, z)) is (1, 0, 0) = i. Thus
ZZ ZZ ZZ
2
F · ndS = 4y dS = 4y 2 dydz
S2 S2 y 2 +z 2 ≤16
Z 2π Z 4
= 4 r3 cos2 θdrdθ = 44 π.
0 0

Now,
ZZZ ZZZ Z 4 Z x Z 2π  
2 2 2 2 2
div F dV = (x + y + z )dV = (x + r )rdrdθ dx
y 2 +z 2 ≤x2
R 0≤x≤4
0 0 0
Z 4 Z x 
= 2π (x2 r + r3 )dr dx
0 0
Z 4 4
3x 6π
= 2π dx = 44 .
0 4 5

58
 Since 44 6π
5
= 44 π5 + 44 π, the Divergence Theorem is verified.

(2) We have divF = (y + z + x) and

ZZZ
I = (x + y + z)d(x, y, z) =
Z Z ZR ZZZ ZZZ
= xd(x, y, z) + yd(x, y, z) + zd(x, y, z)
R R R
Z c Z b(1− z ) Z a(1− y − z )
c b c
= xdxdydz + (· · · ) + (· · · )
0 0 0
a bc ab c abc2
2 2
= + +
24 24 24
abc
= (a + b + c).
24

Now,
ZZ ZZ ZZ ZZ ZZ
F · ndS = F · ndS + F · ndS + F · ndS + F · ndS
S S1 S2 S3 S4

where
x y
S1 : z = 0; + ≤ 1, x, y ≥ 0 and
a b
x z
S2 : y = 0; + ≤ 1, x, z ≥ 0 and
a c
z y
S3 : x = 0; + ≤ 1, y, z ≥ 0 and
c b
x y z
S4 : + + = 1, x, y, z ≥ 0.
a b c
Also,
along S1 , n = −k ⇒ F · n = −xz = 0 (as z = 0 on S1 );

along S2 , n = −j ⇒ F · n = −yz = 0 (as y = 0 on S2 );

along S3 , n = −i ⇒ F · n = −xy = 0 (as x = 0 on S3 ).

59
 Along S4 , the outward normal (to z = c(1 − x
a
− yb ) ≡ f (x, y)) is ( ac , cb , 1) so that
ZZ ZZ  cxy cyz 
F · ndS = + + zx d(x, y)
S4 x
a
+ yb ≤1;x,y≥0 a b
Z a b(1− x
Z )
cxy a
= dxdy + (· · · ) + (· · · )
0 0 a
ab2 c abc2 a2 bc
= + +
24 24 24
abc
= (a + b + c).
24

(3) Consider F = uvi. By the Divergence Theorem, one has

ZZ ZZZ
∂
uvnx dS = (uv)d(x, y, z)
∂R ∂x
Z Z ZR ZZZ
∂v ∂u
= u d(x, y, z) + v d(x, y, z)
R ∂x R ∂x

Hence
ZZZ ZZ ZZZ
∂v ∂u
u dV = (uvnx )dS − v dV.
R ∂x ∂R ∂x

∇ · (φ∇φ) = k∇φk2 + φ∇2 φ

(4) ⇒ φ∇2 φ = 10φ − 4φ

⇒ ∇2 φ = 6.
Thus
ZZ ZZ
∂φ
dS = gradφ · ndS
S ∂n
Z ZSZ ZZZ
= div(gradφ)dV = 6 dV
V V
4π
= 6 (Volume of the sphere) = 6 × = 8π.
3

(5) Let F = xi. Using the Divergence Theorem, we get

ZZZ ZZ
V = d(x, y, z) = xnx dS
R S

60
 Similarly, letting F = yj, we get
ZZ
V = yny dS
S

and letting F = zk, we get

ZZ
V = znz dS.
S

(6) Consider
ZZ
I = x2 dy ∧ dz + y 2 dz ∧ dx + z 2 dx ∧ dy
Z ZS
= (F · n)dS,
S

where F = x2 i + y 2 j + z 2 k. We have then, using the Divergence Theorem,

ZZZ Z 1Z 1Z 1
I= (div F)dV = 2(x + y + z)dxdydz = 3.
0 0 0

ZZ
(7) The required integral is I = (F · n)dS, where F = yzi + zxj + xyk.
∂R
Since div(F) = 0, one has
ZZZ
I= (div F)dV = 0.
R

(8) By the Divergence Theorem, one has

ZZ ZZZ
S
(∇ × u) · ndS = ∇ · (∇ × u)dV = 0,
S S1 R

where S1 is the disc x2 + y 2 + z 2 = 1, z = 1/2. Thus

ZZ ZZ
(∇ × u) · ndS = (∇ × u) · ndS,
S S1
RR
where n in the R.H.S. integral is the vector k; so R.H.S.= S1
(∇ × u) · kdS.

But the coefficient of k in curl(u) is 0. Hence, curl(u) · k = 0. Thus

ZZ
(∇ × u) · kdS = 0.
S1

61
 This gives
ZZ
(∇ × u) · ndS = 0.
S

(9) Note that

( x2 , y2 , z2 ) 1
p=n·r= q a b c · (x, y, z) = q .
x2 2 2 x2 2 z2
a4
+ yb4 + zc4 a4
+ yb4 + c4

(a) Let F = xi + yj + zk (= r). Then

ZZ ZZ ZZZ ZZZ  
4π
F · n dS = p dS = divF dV = 3 dV = 3 abc = 4πabc.
S S V V 3

1
(b) Let F = p2
(xi + yj + zk). Then
ZZ ZZ ZZZ
1
dS = F · n dS = divF dV
S p S V
x2 y 2 z 2
ZZZ  
= 5 + 4 + 4 d(x, y, z).
V a4 b c

Let x = ar sin φ cos θ, y = br sin φ sin θ, z = cr cos φ. Then

Z π Z 2π Z 1  3
sin φ cos2 θ sin3 φ sin2 θ sin φ cos2 φ
ZZ 
1 4
dS = 5abc r + + drdθdφ
S p 0 0 0 a2 b2 c2
Z π Z 2π  3
sin φ cos2 θ sin3 φ sin2 θ sin φ cos2 φ

= abc + + dθdφ
0 0 a2 b2 c2
Z π 3
sin φ sin3 φ 2 sin φ cos2 φ

= πabc + + dφ
0 a2 b2 c2
 
4 1 1 1 4π 2 2
= πabc 2
+ 2+ 2 = (b c + c2 a2 + a2 b2 ).
3 a b c 3abc

(We used the fact that the Jacobian of (x, y, z) with respect to (r, φ, θ) is

abcr2 sin φ).

62
 x2 y2 z2
Aliter: If ψ = a2
+ b2
+ c2
, then

∇ψ
ZZ ZZ  ZZ
1 n 
dS = · n dS = · n dS
S p S r·n S r · ∇ψ
∇ψ
ZZ
= · n dS
S 2
∇2 ψ
ZZZ  Z Z Z  
1 1 1 4 1 1 1
= dV = + + dV = πabc + + .
V 2 a2 b 2 c 2 V 3 a2 b 2 c 2

(10) For a simple closed (and sufficiently smooth) plane curve C : r(s) = x(s)i + y(s)j,

parametrized by the arc length s, the outward unit normal at x(s) is n =

y 0 (s)i − x0 (s)j. Let R be the region enclosed by the curve C.

Let F = Qi − P j be a continuously differentiable vector field in a region including

S
C R. Then

∂Q ∂P
div(F) = − and F · n = Qy 0 + P x0 .
∂x ∂y

Thus one has

I I ZZ   ZZ
∂Q ∂P
(F · n) ds = P dx + Qdy = − d(x, y) = div(F) d(x, y).
C C R ∂x ∂y R

63