Anma 2930

21-4-2005 17:22
c M. K. Warby, J. E. Furter MA2930 ANALYSIS, Exercises Page 1
Exercises on Sequences and Series of Real Numbers

1. This was about half of question 1 of the June 2004 MA2930 paper.
Let (xn ) denote a sequence of real numbers.
(a) (i) Define what it means for the sequence (xn ) to converge, using the usual and
N notation.
[1 mark]
(ii) Define what it means for the sequence (xn ) to be strictly increasing.
[1 mark]
(iii) If the sequence is bounded above then define the least upper bound (i.e. the
supremum) of (xn ).
[1 mark]
(iv) If a sequence (xn ) is both increasing and bounded above then state what you
can deduce about the convergence or divergence of the sequence?
[1 mark]
(b) Explain why each of the following sequences converges and in the case of (i) and
(ii) determine the limits.
103n2 − 8
(i) xn := . [2 marks]
4n2 + 99n − 3
√
(ii) xn := −n + n2 + 3n. [2 marks]
n
X 3k 2 + 2k
(iii) xn := . [2 marks]
k=1
2k
ANSWER
(a) (i) (xn ) converges to x ∈ R if for every > 0 there exists a N such that
|xn − x| < for all n ≥ N .
1 mark
(ii) (xn ) is strictly increasing if xn+1 > xn for n = 1, 2, · · · .
1 mark
(iii) u ∈ R is a least upper bound or supremum of (xn ) if it is an upper bound of

(xn ) and no number smaller than u is an upper bound.
1 mark
21-4-2005 17:22
(iv) A sequence (xn ) which is increasing and bounded above converges. It converges
to the least upper bound of the set {xn }.
1 mark
(b) (i) The sequence converges because it is a combination of standard convergent

sequences. We have
103n2 − 8 103 − 8/n2 103

xn = 2 = 2
→ as n → ∞.
4n + 99n − 3 4 + 99/n − 3/n 4
2 marks
(ii)
√
xn = −n + n2 + 3n
(n2 + 3n) − n2
= p
(n2 + 3n) + n
3n
= p
n (1 + 3/n) + n
3 3
= p → as n → ∞.
(1 + 3/n) + 1 2
2 marks
(iii) (xn ) is a sequence of partial sums (i.e. a series). With ak = (3k 2 + 2k)/2k the
ratio test gives
1 3(k + 1)2 + 2(k + 1)

ak+1
=
ak 2 3k 2 + 2k
1 3(1 + 1/k)2 + 2(1/k + 1/k 2 )

1
= → as k → ∞.
2 3 + 2/k 2
As this limit is less than 1 in magnitude the series converges.
2 marks
2. This was question 1 of the January 2003 MA2034A paper. Cauchy sequences will be
covered later in the module this year and thus you will not be able to answer (a)(ii) and
(b) yet.
(a) (i) Define what it means for the sequence (xn ) to converge using the usual and
N notation.
[1 mark]
(ii) Define what it means for the sequence (xn ) to be a Cauchy sequence.
[2 marks]
21-4-2005 17:22
(iii) Define what it means for the sequence (xn ) to be strictly increasing.
[1 mark]
(b) Prove that if a sequence (xn ) converges then it is a Cauchy sequence.
[4 marks]
(c) Determine the limits of the following sequences (xn ) whose nth term xn is given
below.
(i)
5n3 + 3n + 1
xn := . [2 marks]
15n3 + n2 + 2
(ii)
sin(n2 + 1)
xn := . [2 marks]
n2 + 1
(iii) √ √
n+2− n+1
xn := √ √ . [2 marks]
n+1− n
(d) Let
n n
1 k X X
ak = k , bk = k , sn = ak and tn = bk .
k2 2 k=1 k=1
(i) Find the limits of the sequences (ak+1 /ak ) and (bk+1 /bk ).
[2 marks]
(ii) Given that

k−2
1 3
ak ≤ k ≤ b k and bk ≤ , k ≥ 3,
2 4
explain why (sn ) and (tn ) both converge with
lim sn ≤ 1 ≤ lim tn ≤ 4.
n→∞ n→∞
[4 marks]
21-4-2005 17:22
ANSWER
(a) (i) (xn ) converges to x ∈ R if for every > 0 there exists a N such that
|xn − x| < for all n ≥ N .
1 mark
(ii) (xn ) is a Cauchy sequence if for every > 0 there exists a N such that
|xn − xm | < for all n, m satisfying n ≥ N and m ≥ N .
2 marks
(iii) (xn ) is strictly increasing if xn+1 > xn for n = 1, 2, · · · .
1 mark
(b) From the definition of convergence of (xn ) to x there exists a N = N (/2) such
that
|xn − x| < /2 for all n ≥ N .
If both n ≥ N and m ≥ N then
|xn − xm | = |(xn − x) − (xm − x)| ≤ |(xn − x)| + |(xm − x)| < /2 + /2 =
by the triangle inequality and the above bound. Thus (xn ) is a Cauchy sequence.
4 marks
(c) (i)
5n3 + 3n + 1 5 + 3/n2 + 1/n3 5 1
xn := 3 2 = 3 → = as n → ∞.
15n + n + 2 15 + 1/n + 2/n 15 3
In the above we have used the result that 1/n → 0 as n → ∞ and a result about
combining convergent sequences and noting that the denominator converges to
a non-zero value.
2 marks
(ii) | sin(n2 + 1)| ≤ 1 and hence
1 1/n2
|xn | ≤ = → 0 as n → ∞.
n2 + 1 1 + 1/n2
Thus xn → 0 as n → ∞.
2 marks
(iii) Applying the identity a − b = (a2 − b2 )/(a + b) to the numerator and the
denominator gives
√ √ √ √
n+2− n+1 n+1+ n
xn := √ √ =√ √
n+1− n n+2+ n+1
p
1 + 1/n + 1 1+1
= p p → = 1 as n → ∞.
1 + 2/n + 1 + 1/n 1+1
2 marks
21-4-2005 17:22
(d) (i)
ak+1 k2k k
= (k+1)
=
ak (k + 1)2 2(k + 1)
1 1
= → as k → ∞.
2(1 + 1/k) 2
bk+1 (k + 1)2k k+1
= (k+1)
=
bk k2 2k
1 + 1/k 1
= → as k → ∞.
2 2
2 marks
(ii) sn − sn−1 = an > 0 and tn − tn−1 = bn > 0 and thus both (sn ) and (tn ) are
strictly increasing. They both converge if they are bounded.
n n n−1
!
1 1 − (1/2)n

X X 1 1 X 1 1
sn = ak ≤ k
= k
= = 1 − n < 1.
1 1
2 2 0
2 2 1 − (1/2) 2
Thus lim sn ≤ 1.
n n k−2 n−3 k
X
X X 3 1 1 3 3
tn = bk ≤ b1 + b2 + = + +
1 3
4 2 2 4 k=0 4

3 1
< 1+ = 4.
4 1 − (3/4)
Thus lim tn ≤ 4.
Finally as
1
tn ≥ 1 −
2n
for all n it follows that tn ≥ 1.
4 marks
3. These are parts of the question 1 of the January 1999–2002 examination papers. The
marks shown in bold and in square brackets next to the question are the part marks that
were shown on the January 2001 and 2002 examination papers. The mark breakdown
was not indicated on MA2034A examination papers before January 2001.
(i) Define what it means for the sequence (xn ) to be bounded.

[1 mark]
ANSWER
(xn ) is bounded if there exists a M such that |xn | ≤ M for all n.
1 mark
21-4-2005 17:22
(ii) Prove that if a sequence (xn ) is strictly increasing and bounded above then it
converges to u where u is the least upper bound of (xn ).
[4 marks]
ANSWER
As u is the least upper bound then no smaller number u − is an upper bound for
any > 0. Thus there must be an xN with
u − < xN ≤ u .
The increasing property of the sequence then implies that
u − < xN ≤ xN +1 ≤ · · · ≤ u
and we satisfy the convergence definition for all n ≥ N .
4 marks
(iii) Determine the limits of the following sequences (xn ) whose nth term xn is given
below.
(a)
7n4 + n2 − 2
xn := . [2 marks]
14n4 + 5n − 4
ANSWER
7n4 + n2 − 2 7 + (1/n2 ) − (2/n4 ) 7 1

xn = 4 = 3 4 → = as n → ∞.
14n + 5n − 4 14 + (5/n ) − (4/n ) 14 2
a non-zero value.
2 marks
(b)
n3 + 2n2 + 1
xn := . [2 marks]
6n3 + n + 4
ANSWER
n3 + 2n2 + 1 1 + (2/n) + (1/n3 ) 1

xn = 3 = 2 3 → as n → ∞.
6n + n + 4 6 + (1/n ) + (4/n ) 6
a non-zero value.
2 marks
21-4-2005 17:22
(c)
n2 + n + 1
xn := .
3n2 + 4
ANSWER
n2 + n + 1 1 + (1/n) + (1/n2 ) 1
xn = 2 = 2 → as n → ∞.
3n + 4 3 + (4/n ) 3
In the above we have used the result that 1/n → 0 as n → ∞ and a result
about combining convergent sequences. The denominator converges and it is
at least 3 for all n.
2 marks
(d) √
xn := n4 + n2 − n2 . [2 marks]
ANSWER
√ n2
xn = n4 + n2 − n2 = √
n4 + n2 + n2
1 1 1
= p → = as n → ∞.
1 + 1/n2 + 1 1+1 2
2 marks
It is acceptable here to use the general binomial expansion to give
xn = n2 (1 + 1/n2 )1/2 − 1 = n2 ((1/2)(1/n2 ) + O(1/n4 ))

= 1/2 + O(1/n2 ) → 1/2 as n → ∞.
(e) √
xn := −n + n2 + n . [2 marks]
ANSWER
√ n
xn = n2 + n − n = √
n2 + n + n
1 1 1
= p → = as n → ∞.
1 + 1/n + 1 1+1 2
2 marks
(f)
sin n √ √
xn := + ( n + 1 − n) .
n
21-4-2005 17:22
ANSWER
Since | sin n| ≤ 1 we have

sin n 1
n ≤ n → 0 as n → ∞ .

Also,
√ √ 1
n+1− n= √ √ → 0 as n → ∞.
n+1+ n
Thus xn → 0 as n → ∞.
2 marks
(iv) (a) Show that the sequence (xn ) whose nth term is
n3 + 3n2
xn := − n2
n+1
is unbounded.
[2 marks]
ANSWER
n3 + 3n2 n3 + 3n2 − (n3 + n2 )

xn = − n2 =
n+1 n+1
2 2
2n 2n
= ≥ =n
n+1 2n
(xn ) is unbounded by comparison with n.
(b) Show that the sequence (xn ) whose nth term is
xn := (n + 1/n)3 − n3
is unbounded.
ANSWER
xn = (n3 + 3n + 3/n + 1/n3 ) − n3 = 3n + 3/n + 1/n3 > 3n.

(xn ) is unbounded by comparison with 3n.
(c) Show that the sequence (xn ) whose nth term is
4
xn := n + 1/n2 − n4 [3 marks]
is unbounded.
21-4-2005 17:22
ANSWER
4
xn = n + 1/n2 − n4
= (n4 + 4n + 6/n2 + 4/n5 + 1/n8 ) − n4
= 4n + 6/n2 + 4/n5 + 1/n8 > 4n .
The sequence (n) is unbounded and hence by comparison (xn ) is also un-
bounded.
3 marks
(v) (a) Given that k! ≥ 2k−1 for all k ≥ 1, show that the sequence (xn ) whose nth
term is n
X 1
xn :=
k=0
k!
is bounded above by 3. Explain why you can deduce that it converges.
ANSWER
k−1
We are given that k! ≥ 2 for k ≥ 1 and thus
1 1
≤ k−1 for k ≥ 1.
k! 2
Thus
n
X 1
xn = 1+
k=1
k!
n
X 1
≤ 1+ k−1
k=1
2
1
1− n
= 1+ 2 (by summing the geometric series)
1
1−
2
1
= 1+2 1− n <3 .
2
Since xn+1 − xn = 1/(n + 1)! > 0 the sequence is strictly increasing and as it
is also bounded above it converges by the monotone convergence theorem.
(b) Given that k k ≥ 2k for all k ≥ 2, show that the sequence (xn ) whose nth term
is n
X 1
xn :=
k=1
kk
is bounded above by 3/2. Explain why you can deduce that it converges.
21-4-2005 17:22
ANSWER
For k ≥ 2,
1 1
k k ≥ 2k implies ≤ .
kk 2k
Thus
n
X 1
xn − 1 =
k=2
kk
n n−2
X 1 1X 1
≤ =
k=2
2k 4 j=0 2j
1 1 − (1/2)n−1

=
4 1 − (1/2)

1 1 1
< =
4 1 − (1/2) 2
by summing the geometric series. Hence xn ≤ 3/2 for all n.
4 marks
Since xn − xn−1 = 1/nn > 0 the sequence is strictly increasing and as it is also
bounded above it converges by the monotone convergence theorem.
(c) Let
n n
1 k X X
ak = k , bk = k , sn = ak and tn = bk .
k2 2 k=1 k=1
Given that k−2

1 3
ak ≤ k ≤ b k and bk ≤ , k ≥ 3,
2 4
explain why (sn ) and (tn ) both converge with
lim sn ≤ 1 ≤ lim tn ≤ 4.
n→∞ n→∞
[4 marks]
ANSWER
This repeats the last part of question 1. See the answer to the last part of
question 1.
4. If the sequence (xn ) converges then show that its limit is unique.
21-4-2005 17:22
ANSWER
This is a proof by contradiction. Hence you start by assuming that the converse is true
which in this case means that you assume that the sequence converges to 2 distinct
limits x 6= y. We need now to show that this always leads to a contradiction.
All we know about the sequence is that it converges and thus we start by using this.
We have from the definition of convergence that for every > 0 there exists Nx and Ny
such that
|xn − x| < for all n ≥ Nx and |xn − y| < for all n ≥ Ny .
As we are aiming to show that it is impossible for x and y to be different the next step
is to consider the difference x − y. If n > max{Nx , Ny } then by the triangle inequality
we have
0 6= x − y = (x − xn ) + (xn − y)
0 < |x − y| = |(x − xn ) + (xn − y)| ≤ |xn − x| + |xn − y| < + = 2 .
This is true for all and by taking sufficiently small we get our contradiction, specifically
we get a contradiction if we take
|x − y|
= >0.
4
5. Show that if (xn ) is a sequence of real numbers which converges to x then the sequence
(sn ) where
x1 + x2 + · · · + xn
sn :=
n
also converges to x.
ANSWER
(xn ) converges to x implies that for every > 0 there exists an Ñ such that
|xn − x| < for all n ≥ Ñ .
We now divide the sum into those terms before Ñ and those after Ñ . We have
   
n Ñ −1 n
1 X 1 X 1 X
|sn − x| = (xi − x) ≤ |xi − x| +  |xi − x|

n 1 n 1
n
Ñ
Ñ −1
C X
≤ + , where C = |xi − x| .
n 1
Now
C C
+ < 2 provided < , which requires that n > C/.
n n
Thus if we let N = max{Ñ , C/} then for all n ≥ N we have |sn − x| < 2 which proves
that the sequence (sn ) converges to x.
21-4-2005 17:22
6. Given that (1 + 1/n)n → e = 2.718 · · · as n → ∞ and for c > 0, c1/n → 1 as n → ∞,

show the following.
(i) n
1
1+ 2 → 1 as n → ∞.
n
(ii) The sequence (xn ) defined by
n
1
xn := 1+ √
n
is unbounded.
(iii) If r = p/q ∈ Q is a rational number (i.e. p, q ∈ N with q 6= 0) and assuming that
the sequence (tn ) defined by r n
tn := 1 +
n
converges then show that the subsequence (tnp ) = (tp , t2p , · · · ) converges to er .
ANSWER
The key here is to introduce the convergent sequence (yn ) where

n
1
yn := 1 + → e = 2.718 · · · as n → ∞
n
and to express the given sequences in terms of (yn ).
(i) Let n
1
xn := 1+ 2 = (yn2 )1/n .
n
Now yn → e implies that yn is near e for sufficiently large n. In particular, yn < 3
for all sufficiently large n. (A closer inspection would reveal that this is true for all
n but this amount of detail is not required to establish that the given sequence is
convergent.) Thus
1 ≤ xn = (yn2 )1/n < 31/n .
The right hand side converges to 1 as n → ∞ and thus the squeeze theorem implies
that xn → 1 as n → ∞.
(ii) Now let n
1
xn := 1+ √ .
n
Observe that n2
1
xn2 = 1+ = ynn .
n
yn → e implies that yn is near e for sufficiently large n. In particular, yn ≥ 2 for
sufficiently large n. (A closer inspection would reveal that this is true for all n but,
21-4-2005 17:22
as in part (i), this amount of detail is not required to establish that the sequence
is unbounded.) Thus
xn2 ≥ 2n
and hence the subsequence (xn2 ) is unbounded. This in turn implies that (xn ) is
unbounded.
(iii) We are told that the sequences (tn ) converges and we know that the any subsequence
of a convergent sequence also converges with the same limit. Now
np np
p/q 1
tnp := 1 + = 1+ .
np nq
Observe that nq
1
ynq := 1+ .
nq
Thus
r
tnp = ynq .
As yn → e as n → ∞ we have for the subsequence that ynq → e as n → ∞. From
r
this it follows that ynq → er as n → ∞ by a result about this type of function of a
convergent sequence. Hence we have shown that
r
lim tn = lim tnp = lim ynq = er .
n→∞ n→∞ n→∞
7. Let (sn ) be the sequence given by

1 1 1
sn := + + ··· + .
n+1 n+2 2n
Show that the sequence is increasing. Does it converge?
By noting that Z k+1
1 1 1 1
0< dx − < −
k x k+1 k k+1
show that
1
0 < ln 2 − sn < .
2n
ANSWER

1 1 1 1
sn+1 − sn = + ··· + − + ··· +
n+2 2n + 2 n+1 2n
1 1 1
= + −
2n + 1 2n + 2 n + 1
(4n + 3) 2
= −
(2n + 1)(2n + 2) 2n + 2
(4n + 3) − 2(2n + 1) 1
= = >0.
(2n + 1)(2n + 2) (2n + 1)(2n + 2)
21-4-2005 17:22
1 . Thus
Thus (sn ) is increasing. (sn ) is the sum of n terms the largest of which is n + 1
n
sn ≤ <1.
n+1
(sn ) is hence increasing and bounded and it converges to a limit s ≤ 1.
1 1 1
≤ ≤ for x ∈ (k, k + 1).
k+1 x k
Now Z k+1 Z k+1
1 dx 1 1
= , and hence − dx > 0
k+1 k k+1 k x k+1
from which it follows that
2n−1
X Z k+1 1 Z 2n
1 1
0< dx − = dx − sn = [ln x]2n
n − sn = ln 2 − sn .
k=n k x k + 1 n x
Thus ln 2 bounds the sequence. Also note that

2n−1
X Z k+1 2n−1
X 1
1 1 1
0 < ln 2 − sn = dx − ≤ −
k=n k x k + 1 k=n
k k+1

1 1 1 1 1 1
= − + − + ··· + −
n n+1 n+1 n+2 2n − 1 2n

1 1 1
= − = → 0 as n → ∞.
n 2n 2n
Thus we have shown that sn → ln 2 as n → ∞.
8. In each of the following cases determine whether or not the series converges.
(a)
∞
X 1
.
n=1
2n +1
ANSWER
We could show convergence here by using the ratio or root test or more simply by
using the comparison test by noting that
1 1
0≤ ≤ .
2n + 1 2n
The upper bound is a term from a convergent geometric series.
(b)
∞
X 4n2 − n + 3
.
n=1
n3 + 2n
21-4-2005 17:22
ANSWER
This is divergent.
4n2 − n + 3 1 4 − 1/n + 3/n2

an = 3
= cn , cn = → 4 as n → ∞.
n + 2n n 1 + 2/n2
cn → 4 implies that there exists

PN such that cn > 3 for n ≥ N . Hence for n P
≥N
we have an ≥ 3/n and since 1/n diverges we have by comparison that an
diverges.
(c)
∞ √
X n+ n
.
n=1
2n3 − 1
ANSWER
This converges.
√ √
n+ n 1 1 + 1/ n 1
an = 3 = 2 cn , cn = → as n → ∞.
2n − 1 n 2 − 1/n3 2
cn → 1/2 implies that there exists

P N2 such that cn < 1 for n ≥ N . Hence for nP
≥N
we have an ≤ 1/n2 and since 1/n converges we have by comparison that an
diverges.
(d)
∞
2
X
n4 e−n .
n=1
ANSWER
By the root test
1/n
4 −n2 −n2
an = n e , a1/n
n = (n 1/n 4
) e = (n1/n )4 e−n → 0 as n → ∞.
Here the results is as a consequence of n1/n → 1 and e−n → 0. By the root test the
series converges.
9. You will see questions like this in the section on series of functions.
For each of the following series determine the values of x ∈ R such that the given series
converges.
(a)
∞
X xk
k=0
k!
21-4-2005 17:22
ANSWER
Let ak = xk /k! and use the ratio test. We have
ak+1 xk+1 /(k + 1)! x

= k
= → 0 as k → ∞.
ak x /k! k+1
By the ratio test the series converges (absolutely) for all x ∈ R.
(b) In the following α ∈ R is not an integer.

∞
X α(α − 1) · · · (α − k + 1) α(α − 1) 2
xk = 1 + αx + x + ···
k=0
k! 2
ANSWER
Let ak = α(α − 1) · · · (α − k + 1)xk /k!. Using the ratio test
ak+1 α−k α/k − 1

= x= x → x as k → ∞.
ak k+1 1 + 1/k
P
Thus the series ak converges absolutely if |x| < 1 which in turn implies that the
series converges for |x| < 1.
If |x| > 1 then the terms of the series are unbounded and thus the series diverges.
What happens when x = −1 or x = 1 needs more refined tests to determine if
the series converges or diverges and the outcome depends on α. This will not be
considered further here.
(c)
∞
X k 3 xk
.
k=0
3k
ANSWER
The root test is the easiest test to use here. With ak = k 3 xk /3k we have
(k 1/k )3 |x| |x|

|ak |1/k = → as k → ∞.
3 3
By the root test the series converges (absolutely) if |x| < 3, it diverges if |x| > 3.
If |x| = 3 then |ak | = k 3 and since these terms become unbounded it follows that
the series diverges when |x| = 3.
(d)
∞
X
k k xk .
k=0
21-4-2005 17:22
ANSWER
The root test is the easiest test to use here. With ak = k k xk we have
1/k
|ak | = |kx|.
This only converges if x = 0 and is unbounded for x 6= 0. Hence the series only
converges when x = 0.
(e)
∞
X
ak xk = 1 + 2x + x2 + 2x3 + x4 + · · · ,
k=0
i.e. with a2k = 1 and a2k+1 = 2 for k = 0, 1, 2, · · · .
ANSWER
Let bk = ak xk . The ratio test does not give any information here as ak+1 /ak does
not have a limit as k → ∞. However we can still use the root test. Since
1/k
1 ≤ ak ≤ 2, 1 ≤ ak ≤ 21/k → 1 as k → ∞.
Thus
1/k
|bk |1/k = ak |x| → |x| as k → ∞.
The series converges (absolutely) if |x| < 1 and diverges if |x| > 1. By inspection
the series diverges if x = 1 as the terms of the series do not tend to 0 as k → ∞.
It can be shown that the series also diverges when x = −1.
(f) √
∞
X x2 + k − |x|
.
k=1
k2
ANSWER
Let
√
x2 + k − |x| (x2 + k) − x2 1 1
ak = 2 = √ 2
= √ ≤ 3/2 for all x
k 2
( x + k + |x|)k 2
( x + k + |x|)k k
since x2 ≥ 0. Since 0 P
≤ ak ≤ 1/k 3/2 the series
P
ak converges by comparison with
the convergent series 1/k 3/2 .
(g)
∞
X cos kx sin kx
+3 2 .
k=1
k3 k
21-4-2005 17:22
ANSWER
We can test for absolute convergence. If ak denotes the kth term then by the
triangle inequality and that | sin kx| ≤ 1 and | cos kx| ≤ 1 we have
1 1
|ak | ≤ 3 +3 2
k k
1/k 3 and 1/k 2 are standard convergent series it follows
P P
for allPx ∈ R. Since
that |ak | converges by the comparison test. Hence the original series converges
for all x.
21-4-2005 17:22
Exercises on continuity and the contraction mapping

theorem
1. This was the first part of question 2 of the June 2004 MA2930 exam paper.
(a) Let f : [a, b] → R denote a function. Define what it means for f to be Lipschitz on
[a, b] and explain when such a function is also a contraction on [a, b].
[2 marks]
(b) Show that the following function is Lipschitz on [−1, 1] and give the smallest Lip-
schitz constant.
(
x2 , 0 ≤ x ≤ 1,
f : [−1, 1] → R, f (x) :=
−3x, −1 ≤ x < 0.
[3 marks]
(c) If f : [a, b] → R is continuously differentiable on [a, b] then it can be shown that f
is Lipschitz on [a, b] with the smallest Lipschitz constant L given by
L = max |f 0 (x)| .
x∈[a,b]
Use this to obtain the smallest Lipschitz constant for the following functions on
their domains of definition.
(i) f : [−1, 1] → R, f (x) := 2x − 2x3 . [3 marks]
(ii) f : [−2, 2] → R, f (x) := (x2 − 4)2 . [4 marks]
ANSWER
(a) f is Lipschitz on [a, b] if there exists a constant L ≥ 0 such that
|f (x) − f (y)| ≤ L|x − y|
for all x, y ∈ [a, b]. f is a contraction if 0 ≤ L < 1.
2 marks
(b) If x, y ∈ [−1, 0) then
|f (x) − f (y)| = 3|x − y|.
For x > 0, f 0 (x) = 2x and thus if x, y ∈ [0, 1] then
|f (x) − f (y)| ≤ f 0 (1)|x − y| = 2|x − y|.
If x > 0, y < 0 then x − y = |x| + |y| and
|f (x) − f (0)| ≤ 2|x|,
|f (0) − f (y)| ≤ 3|y|,
|f (x) − f (y)| ≤ 2|x| + 3|y| < 3(|x| + |y|) = 3|x − y|.
3 is the smallest Lipschitz constant.
3 marks
21-4-2005 17:22
(c) (i)
f 0 (x) = 2 − 6x2 .
f 00 (x) = −12x = 0 at x = 0. Possible values for the maximum of |f 0 (x)| are
x = 0 and the end points x = ±1. Thus the smallest constant is
L = |f 0 (1)| = 4.
3 marks
(ii)
f 0 (x) = 2(x2 − 4)(2x) = 4x(x2 − 4),

f 00 (x) = 4(3x2 − 4).
f 00 (x) = 0 when x2 = 4/3 which gives two points in [−2, 2].

√

0 0 2 4 64
f (±2) = 0 and |f (±2/ 3)| = 4 √ 4− = √ .
3 3 3 3
Thus the smallest constant is
64
L= √ .
3 3
4 marks
2. This was question 2 of the Jan 2003 MA2034A exam paper.
(a) Let I ⊂ R be an interval and let f : I → R be a function.

(i) Give the definition of the continuity of the function f at the point c ∈ I using
the usual , δ notation.
[2 marks]
(ii) Define what it means for f to be Lipschitz on I and show that such a function
is continuous on I.
[3 marks]
√
(b) Let f : [0, 1] → R, f (x) := x. Show that f is continuous at x = 0 but that f is
not Lipschitz on [0, 1].
[4 marks]

L = max |f 0 (x)| .
x∈[a,b]
(i) f : [5, 13] → R, f (x) := 2x2 − 17.
[3 marks]
21-4-2005 17:22
(ii) f : [−1, 1] → R, f (x) := (1 − x2 )2 .
[4 marks]
(iii) f : [0, 4] → R, f (x) := (x + 1)e−x .
[4 marks]
ANSWER
(a) (i) f is continuous at c ∈ I if for every > 0 there exists a δ such that
|f (x) − f (c)| < whenever |x − c| < δ .
2 marks
(ii) f is Lipschitz on I if there exists a constant L ≥ 0 such that
|f (x) − f (y)| ≤ L|x − y|
for all x, y ∈ I.
|f (x) − f (y)| ≤ L|x − y| < provided |x − y| < /L .

Hence we satisfy the continuity requirement by taking δ = /L.
3 marks
(b) For x ≥ 0 √
|f (x) − f (0)| = |f (x)| = x < provided x < 2 .
Thus we satisfy the –δ definition with δ = 2 .
2 marks
For x > 0, y > 0 and x 6= y we have

√ √
f (x) − f (y) x− y 1
= =√ √ .
x−y x−y x+ y
As x → 0 and y → 0 this becomes unbounded and hence no constant L ≥ 0 can
exist as the Lipschitz constant of f on [0, 1].
2 marks
(c) (i)
f 0 (x) = 4x and f 00 (x) = 4.
There are no turning points of f 0 and
L = max{|f 0 (5)|, |f 0 (13)|} = f 0 (13) = 52.
3 marks
21-4-2005 17:22
(ii)
f 0 (x) = 2(1 − x2 )(−2x) = 4x(x2 − 1) = 4(x3 − x) and f 00 (x) = 4(3x2 − 1) .

√
There are turning points of f 0 at ±1/ 3.
√ √ 8
f 0 (−1) = f 0 (1) = 0 and |f 0 (±1/ 3)| = (4/ 3)(1 − 1/3) = √ .
3 3
Thus
8
L= √ .
3 3
4 marks
(iii)
f 0 (x) = e−x − (x + 1)e−x = −xe−x

f 00 (x) = xe−x − e−x = (x − 1)e−x .
f 0 has a local turning point at x = 1 ∈ [0, 4]. Thus
L = max{|f 0 (0)|, |f 0 (1)|, |f 0 (4)|} = max{0, e−1 , 4e−4 } = e−1 .
4 marks
3. The following were parts of question 2 from the Jan 2000, 2001 and 2002 MA2034A exam
papers.
(a) Let I ⊂ R be an interval and let f : I → R be a function.

Give the definition of the continuity of the function f at the point c ∈ I which
involves convergent sequences.
[1 mark]
ANSWER
f is continuous at c ∈ I if for every sequence (xn ) in I which converges to c then
the corresponding sequence (f (xn )) converges to f (c).
1 mark
(b) Show from the definition that the function f : [−1, 1] → R defined by
(
x if 0 ≤ x ≤ 1
f (x) :=
−2x if −1 ≤ x < 0
is Lipschitz on [−1, 1].

[3 marks]
21-4-2005 17:22
ANSWER
Let x, y ∈ [−1, 1]. In the case x, y ∈ [−1, 0] we have
|f (x) − f (y)| = |x − y|.
In the case x, y ∈ [0, 2] we have
|f (x) − f (y)| = 2|x − y|.
If x ∈ [−1, 0] and y ∈ [0, 1] then y − x = y + |x| and we have
|f (x) − f (y)| = |x + 2y| ≤ |x| + 2y ≤ 2(|x| + y) = 2|x − y|.
f is Lipschitz on [−1, 1] with a Lipschitz constant of 2.
3 marks

L = max |f 0 (x)| .
x∈[a,b]
(i) f : [−5, 5] → R, f (x) := x3 .
[3 marks]
ANSWER
f : [−5, 5] → R, f (x) := x3 , f 0 (x) := 3x2 , f 00 (x) := 6x .
f 00 (x) = 0 when x = 0. This is the only turning value of f 0 . Thus
max |f 0 (x)| = max{|f 0 (−5)|, |f 0 (0)|, |f 0 (5)|} = 3(5)2 = 75 .

[−5,5]
The smallest constant is 75.

3 marks
(ii) f : [−2, 0] → R, f (x) := x4 + 6x3 + 12x2 − x.
[4 marks]
21-4-2005 17:22
ANSWER
f : [−2, 0] → R, f (x) := x4 + 6x3 + 12x2 − x,

f 0 (x) := 4x3 + 18x2 + 24x − 1,
f 00 (x) := 12x2 + 36x + 24 = 12(x2 + 3x + 2)
= 12(x + 1)(x + 2) .
f 00 (x) = 0 when x = −1 and when x = −2.
max |f 0 (x)| = max{|f 0 (−2)|, |f 0 (−1)|, |f 0 (0)|} .

[−2,0]
|f 0 (0)| = 1, |f 0 (−1)| = | − 4 + 18 − 24 − 1| = 11 and |f 0 (−2)| = | − 32 + 72 −

48 − 1| = 9. The smallest constant is 11.
4 marks
√
(iii) f : [10−6 , 1] → R, f (x) := x.
[3 marks]
ANSWER
1 √
f : [10−6 , 1] → R, f 0 (x) := √ .
f (x) := x,
2 x
f 0 is decreasing and positive on the interval and hence the smallest constant is
f 0 (10−6 ) = 103 /2 = 500.
3 marks
(iv) f : [0, 1] → R, f (x) := x2 /2 − x3 /3.
[4 marks]
ANSWER
f : [0, 1] → R, f (x) := x2 /2 − x3 /3,

f 0 (x) := x − x2 = x(1 − x),
f 00 (x) := 1 − 2x .
f 00 (x) = 0 when x = 1/2.
max |f 0 (x)| = max{|f 0 (0)|, |f 0 (1)|, |f 0 (1/2)|} .

[0,1]
f 0 (0) = f 0 (1) = 0 and f 0 (1/2) = 1/4. The smallest constant is 1/4.
4 marks
21-4-2005 17:22
(v) f : [0, 2] → R, f (x) := x4 − 6x3 + 12x2 + 4x.

[4 marks]
ANSWER
f : [0, 2] → R, f (x) := x4 − 6x3 + 12x2 + 4x,

f 0 (x) := 4x3 − 18x2 + 24x + 4,
f 00 (x) := 12x2 − 36x + 24,
= 12(x2 − 3x + 2) = 12(x − 1)(x − 2) .
f 0 has two turning values at x = 1 and at x = 2 (which is also an end point).
Now f 0 (0) = 4, f 0 (1) = 14 and f 0 (2) = 32 − 72 + 48 + 4 = 12. The smallest
Lipschitz constant is
L = max{4, 14, 12} = 14 .
4 marks
(vi) f : [0, 1/2] → R, f (x) := x3 − x2 .
ANSWER
f 0 (x) = 3x2 − 2x = x(3x − 2) and f 00 (x) = 6x − 2 .

f 0 has a turning point at x = 1/3. To determine the maximum of |f 0 | on [0, 1/2]
we need to consider the end points and any intermediate turning points.
f 0 (0) = 0, f 0 (1/3) = −1/3 and f 0 (1/2) = −1/4.
Thus
L = max{0, 1/3, 1/4} = 1/3.
3 marks
4. Let f : R → R, f (x) := ln(x2 + k 2 ) with k > 0. Show that the smallest Lipschitz
constant is 1/k.
ANSWER
0 2x (x2 + k 2 )2 − (2x)(2x)
00 2(k 2 − x2 )
f (x) = 2 and f (x) = = 2 .
x + k2 (x2 + k 2 )2 (x + k 2 )2
f 0 has turning points when f 00 (x) = 0, i.e. when x = ±k. As f 0 (0) = 0 and f 0 (x) → 0
as |x| → ∞ these local turning points are where f 0 attains its maximum in magnitude.
Thus the smallest Lipschitz constant is
L := f 0 (k) = 1/k .
21-4-2005 17:22
5. The following all involve making use of the (Bolzano) Intermediate Value Theorem for
continuous functions.
(a) Show that every polynomial of odd degree with real coefficients has at least one
real root.
ANSWER
Let n ≥ 1 be an odd number and let
p(x) := an xn +an−1 xn−1 +· · ·+a1 x+a0 = xn an + an−1 /x + · · · + a1 /xn−1 + a0 /xn

where an 6= 0. Observe that
an + an−1 /x + · · · + a1 /xn−1 + a0 /xn → an as |x| → ∞.
Thus for sufficiently large |x|, p(x)/xn has the same sign as an . As n is odd this
implies that if A > 0 then p(−A) and p(A) have opposite sign when A is sufficiently
large. This in turn implies by the intermediate value theorem that p has a root in
[−A, A] as the polynomial p is continuous on R.
(b) Let p : R → R be the polynomial
p(x) := a0 + a1 x + a2 x2 + · · · + an xn , ai ∈ R.
Show that if n ≥ 2, n is even, an = 1 and a0 < 0 then p has at least two real roots.
ANSWER
p(x)/xn → 1 as x → ∞. Thus if M > 0 is sufficiently large then p(−M ) > 0 and
p(M ) > 0. Also p(0) = a0 < 0. Thus by the intermediate value theorem p has a
root in [−M, 0] and in [0, M ] as the polynomial p is continuous on R.
(c) Show that if φ : [a, b] → [a, b] is continuous then there exists at least one point
x∗ ∈ [a, b] with φ(x∗ ) = x∗ .
ANSWER
This follows by applying the intermediate value theorem to the continuous function
g : [a, b] → R, g(x) := φ(x) − x .
Since φ : [a, b] → [a, b] we have φ(a), φ(b) ∈ [a, b] and in particular φ(a) ≥ a and
φ(b) ≤ b giving g(a) ≥ 0 and g(b) ≤ 0. Thus either g has a root at x = a or at
x = b or by the intermediate value theorem at some point between a and b.
21-4-2005 17:22
(d) This was part of question 2 of the January 1999 paper.

Let [a, b] denote a closed bounded interval and let f : [a, b] → R be a function
which is continuous on [a, b]. Also let c, d ∈ [a, b] be such that
f (c) = min f (x) and f (d) = max f (x) .

[a,b] [a,b]
The intermediate value theorem implies that f maps [a, b] onto [f (c), f (d)]. In the
case c < d and for a function f for which
f (c) < f (a) = f (b) < f (d)
show that for each y ∈ (f (c), f (d)) there is at least 2 distinct values of x in [a, b]
for which f (x) = y.
ANSWER
Consider g : [a, b] → R, g(x) := f (x) − y which is continuous on [a, b].

If y ∈ (f (c), f (a)) then g(a) = f (a) − y > 0 and g(c) = f (c) − y < 0. Hence g
changes sign on (a, c) and has a root in (a, c) by the intermediate value theorem.
Similarly by considering the interval (c, d), g(d) = f (d) − y > 0 and g changes sign
on (c, d) and has a root in (c, d) by the intermediate value theorem. Thus we have
at least 2 points at which f (x) = y.
If y = f (a) = f (b) then x = a and x = b are 2 points at which f (x) = y.
If y ∈ (f (a), f (d)) then g(c) = f (c) − y < 0 and g(d) = f (d) − y > 0. Hence g
changes sign on (c, d) and has a root in (c, d) by the intermediate value theorem.
Similarly by considering the interval (d, b), g(b) = f (b) − y = f (a) − y < 0 and g
changes sign on (d, b) and has a root in (d, b) by the intermediate value theorem.
Thus we have at least 2 points at which f (x) = y.

Let φ : [0, 1] → [0, 1] be a function.
(a) (i) Explain what it means for φ to be a contraction on [0, 1].
[2 marks]
(ii) Explain the term onto in the statement φ maps [0, 1] onto [m, M ] where 0 ≤
m ≤ M ≤ 1.
[1 mark]
(iii) Given that a contraction mapping φ : [0, 1] → [0, 1] has a fixed point, use a
proof by contradiction to show that the fixed point is unique.
[3 marks]
(iv) If L ≥ 0 is the smallest Lipschitz constant of φ : [0, 1] → [m, M ] given in (ii)

then explain why M − m ≤ L.
[2 marks]
21-4-2005 17:22
(b) (i) Let φ1 : [0, 1] → R, φ1 (x) = (x3 + 3)/5. Determine m and M such that
φ1 : [0, 1] → [m, M ] is onto and show that φ1 satisfies the conditions of the
contraction mapping theorem.
[3 marks]
(ii) Let φ2 : [0, 1] → R, φ2 (x) = 4x(1 − x). Determine m and M such that
φ2 : [0, 1] → [m, M ] is onto and show that φ2 is not a contraction on [0, 1].
Determine the fixed points of φ2 and classify them as stable or unstable.
[5 marks]
√ √
(iii) Let a > 1 and √ a] → [ a, (1 + a)/2], φ3 (x) = 12 (x + a/x). Explain
let φ3 : [ a, √
why φ3 maps [ a, a] onto [ a, (1 + a)/2]. √
For what values of a > 1 is φ3 a contraction on [ a, a]?
Determine the fixed point or points of φ3 and classify any fixed point found as
stable or unstable.
[4 marks]
ANSWER
(a) (i) φ is a contraction on [0, 1] if there exists a constant L, 0 ≤ L < 1 such that
|φ(x) − φ(y)| ≤ L|x − y|
for all x, y ∈ [0, 1].

2 marks
(ii) The statement φ maps [0, 1] onto [m, M ] means that for all y ∈ [m, M ] there
exists an x ∈ [0, 1] such that y = φ(x).
1 mark
(iii) Suppose x∗1 and x∗2 are two different fixed points, i.e.
x∗1 6= x∗2 with φ(x∗1 ) = x∗1 and φ(x∗2 ) = x∗2 .
Thus
0 < |x∗1 − x∗2 | = |φ(x∗1 ) − φ(x∗2 )| ≤ L|x∗1 − x∗2 | < |x∗1 − x∗2 |
by the contraction property with 0 ≤ L < 1. This is a contradiction and hence
there is only one fixed point.
3 marks
(iv) If α, β ∈ [0, 1] are such that φ(α) = m and φ(β) = M then the Lipschitz
condition is
φ(β) − φ(α) M − m
L ≥ = ≥M −m
β − α |β − α|
because |β − α| ≤ 1.
2 marks
21-4-2005 17:22
(b) (i)
3x2 6x
φ01 (x) =
≥ 0 and φ001 (x) = .
5 5
φ01 (x) ≥ 0 implies that φ1 (x) is increasing and thus m = φ1 (0) = 3/5 and
M = φ1 (1) = 4/5.
φ001 (x) = 0 when x = 0, an edge point, φ01 (0) = 0 and thus the smallest Lipschitz
constant of φ1 on [0, 1] is
3
L = φ01 (1) = < 1.
5
As [m, M ] ⊂ [0, 1] and L < 1 the function φ1 satsifies the condition of the
contraction mapping theorem.
3 marks
(ii)
φ2 (x) = 4x(1 − x) = 4x − 4x2 and φ02 (x) = 4(1 − 2x) .
φ02 (x) > 0 for 0 ≤ x < 1/2 and φ02 (x) < 0 for 1/2 < x ≤ 1. Thus φ2 is increasing
in [0, 1/2) and decreasing in (1/2, 1]. φ2 (0) = φ2 (1) = 0 and φ2 (1/2) = 1. Thus
φ2 : [0, 1] → [0, 1]
is onto.
φ02 has no turning points on [0, 1] and hence the maximum of |φ02 (x)| is attained
at the end points. The smallest Lipschitz constant for the region is hence
L = max {|φ02 (x)|} = max{|φ02 (0)|, |φ02 (1)|} = 4 .
x∈[0,1]
The function is hence not contractive on [0, 1].

A fixed point of φ2 satisfies
x = 4x(1 − x) thus x = 0 or 1 = 4(1 − x), i.e. x = 3/4.
φ02 (0) = 4 and φ02 (3/4) = −2. In both cases |φ02 (x∗ )| > 1 and thus x∗ = 0 and
x∗ = 3/4 are both unstable fixed points.
5 marks
(iii)
1 a
= φ03 (x)
1 − 2 ≥ 0 for x2 ≥ a.
2 x
√ √
φ3 increases in [ a, a] and φ3 hence maps the interval onto [ a, (1 + a)/2].

00 1 2a
φ3 (x) = > 0.
2 x3
√
φ003 has no local turning points, φ03 ( a) = 0 and thus the smallest Lipschitz
constant is
1 − 1/a 1
L = |φ03 (a)| = ≤ .
2 2
Hence the mapping is a√contraction for all√a > 1.
The fixed point at x = a is stable as φ03 ( a) = 0.
4 marks
21-4-2005 17:22

(i) Explain what it means for φ to be a contraction on [0, 1].
[2 marks]
(ii) If φ is a contraction on [0, 1] and x∗ is the unique fixed point of φ then show that
if x∗ ∈ [0, 1] and xn+1 = φ(xn ), n = 0, 1, 2, · · · then
|xn − x∗ | ≤ Ln |x0 − x∗ |, n = 1, 2, · · ·
where L is the smallest Lipschitz constant of φ on [0, 1].
[3 marks]
(iii) In the case of the function

3 + e−x
φ(x) =
5
explain why
φ : [0, 1] → [φ(1), φ(0)] ⊂ [0, 1]
and determine the smallest Lipschitz constant of φ on [0, 1].
[5 marks]
State any conclusion you can make about fixed points of φ in [0, 1].
[2 marks]
(iv) Let φ1 : [0, 1] → R, φ1 (x) := x((2 − 4a)x + (4a − 1)) where a ∈ R, a 6= 1/2.
(a) Determine the fixed points of φ1 and classify them as stable or unstable de-
pending on the value of a.
[4 marks]
(b) Explain why for 0 ≤ x ≤ 1 we have
min{4a − 1, 3 − 4a} ≤ φ0 (x) ≤ max{4a − 1, 3 − 4a}.
[2 marks]
(c) By using part (b), or otherwise, explain why φ1 : [0, 1] → [0, 1] is one-to-one
and onto when 1/4 ≤ a ≤ 3/4.
[2 marks]
21-4-2005 17:22
ANSWER
(i) φ is a contraction on [0, 1] if there exists a constant L, 0 ≤ L < 1 such that
|φ(x) − φ(y)| ≤ L|x − y|
for all x, y ∈ [0, 1].
2 marks
(ii)
|xn − φ(x∗ )| = |φ(xn−1 ) − φ(x∗ )| ≤ L|xn−1 − x∗ | ≤ · · · ≤ Ln |x0 − x∗ |
by repeated use of the contraction property.
3 marks
(iii)
−e−x
φ0 (x) = <0
5
and hence φ is decreasing. From this it follows that
φ : [0, 1] → [φ(1), φ(0)] .
Now φ(0) = 4/5 < 1 and φ(1) = (3+1/e)/5 > 3/5 > 0 and thus [φ(1), φ(0)] ⊂ [0, 1].
+e−x
φ00 (x) = > 0.
5
As the second derivative is positive, the maximum of |φ0 | on [0, 1] is at an end point.
The smallest Lipschitz constant is hence L = |φ0 (0)| = 1/5.
5 marks
All the conditions of the contraction mapping theorem are satisfied and hence φ
has a unique fixed point in [0, 1].
2 marks
(iv) (a)
φ1 (x) = x((2 − 4a)x + (4a − 1)) = x
when
x = 0 and (2 − 4a)x + (4a − 1) = 1.
Thus we have fixed points at
2 − 4a
x = 0 and at x = = 1.
2 − 4a
To classify the fixed points we need φ0 (x).
φ0 (x) = 2(2 − 4a)x + (4a − 1), φ0 (0) = 4a − 1, φ0 (1) = 3 − 4a.
The fixed point at x = 0 is stable if −1 < 4a − 1 < 1, 0 < a < 1/2. It is
unstable if a < 0 or a > 1/2.
The fixed point at x = 1 is stable if −1 < 3 − 4a < 1, −1 < 4a − 3 < 1, i.e.
1/2 < a < 1. It is unstable if a < 1/2 or a > 1.
4 marks
21-4-2005 17:22
(b) As φ0 is linear the minimum and maximum on [0, 1] occur at the end points.
Hence for all x ∈ [0, 1].
min{φ0 (0), φ0 (1)} ≤ φ0 (x) ≤ max{φ0 (0), φ0 (1)}.
2 marks
(c) If a ≥ 1/4 then 4a − 1 ≥ 0 and 3 − 4a ≤ 2. If a ≤ 3/4 then 4a − 1 ≤ 2 and
3 − 4a ≥ 0. In all cases the result in part (b) shows that φ0 (x) ≥ 0. Thus φ0 (x)
is increasing and as φ(0) = 0 and φ(1) = 1 it follows that φ : [0, 1] → [0, 1] is
one-to-one and onto.
2 marks
8. The following were parts of question 3 of the Jan 2001 and Jan 2000 papers.
(a) In the the case of the functions φ1 and φ2 defined below, show that both functions
map [0, 1] onto [0, 1] and show that both functions are not contractive on [0, 1].
φ1 : [0, 1] → [0, 1], φ1 (x) := 4x(1 − x).

1
φ2 : [0, 1] → [0, 1], φ2 (x) := (1 − cos 2πx).
2
[9 marks]
In the case of φ1 determine the fixed points and classify each fixed point as stable
or unstable.
[2 marks]
ANSWER
φ1 : [0, 1] → [0, 1], φ1 (x) := 4x(1 − x) = 4x − 4x2 ,

φ01 (x) = 4(1 − 2x),
φ001 (x) = −8 .
φ1 increases in [0, 1/2) and decreases in (1/2, 1] Since φ1 (0) = φ1 (1) = 0 and
φ1 (1/2) = 1, φ1 maps [0, 1] onto [0, 1].
3 marks
φ1 is continuously differentiable on [0, 1]. As φ001 does not change sign, the smallest
Lipschitz constant on [0, 1] is L = max{|φ01 (0)|, |φ01 (1)|} = 4. As this constant is
greater than 1 the function is not contractive on [0, 1].
2 marks
21-4-2005 17:22
Because of properties of the cosine function, cos 2πx takes all values between −1
and +1 as x varies between 0 and 1. Thus 1 − cos 2πx takes all values between 0
and 2 and φ2 takes all values between 0 and 1.
2 marks
1
φ2 : [0, 1] → [0, 1], φ2 (x) := (1 − cos 2πx),
2
φ02 (x) = π sin 2πx .
As φ02 (1/2) = π sin π/2 = π > 1 the smallest Lipschitz constant is greater than 1
and the function is not contractive on [0, 1].
2 marks
A fixed point of φ1 satisfies
x = 4x(1 − x) thus x = 0 or 1 = 4(1 − x), i.e. x = 3/4.
φ01 (0) = 4 and φ01 (3/4) = −2. In both cases |φ0 (x∗ )| > 1 and thus x∗ = 0 and
x∗ = 3/4 are both unstable fixed points.
2 marks
(b) Let φ denote the function

x2 − 2x + 5
φ : [0, 2] → R, φ(x) =
.
4
Show that the conditions of the contraction mapping theorem are satisfied and give
the fixed point.
[5 marks]
ANSWER
x2 − 2x + 5
φ : [0, 2] → R, φ(x) = ,
4
2x − 2 1
φ0 (x) = = (x − 1),
4 2
00 1
φ (x) = .
2
As φ0 (x) < 0 in (0, 1) and φ0 (x) > 0 in (1, 2) the function φ decreases in (0, 1)
and increases in (1, 2). φ(0) = 5/4, φ(1) = 1 and φ(2) = 5/4. Thus φ : [0, 2] →
[1, 5/4] ⊂ [0, 2].
As φ00 (x) does not change sign on [0, 2] the smallest Lipschitz constant is
1
L = max{|φ0 (0)|, |φ0 (2)|} = < 1 .
2
Thus φ is contractive on [0, 2] and satisfies all the conditions of the contraction
mapping theorem.
5 marks
21-4-2005 17:22
(c) Determine which of the following is a contraction on their domain of definition.

(i)
π
φ1 : [−π/8, π/8] → [−π/8, π/8], φ1 (x) := sin(2x) .
8
ANSWER
π π
φ01 (x) =
cos(2x) and hence |φ01 (x)| ≤ < 1 .
4 4
φ1 is a contraction on [−π/8, π/8] with the constant
π
L= max |φ01 (x)| = <1.
[−π/8,π/8] 4
3 marks
(ii)
π
φ2 : [−π/4, π/4] → [−π/4, π/4], φ2 (x) :=
sin(2x) .
4
Classify, as stable or unstable, the three fixed points x = −π/4, x = 0 and
x = π/4 in this case.
ANSWER
π
φ02 (x) =
cos(2x) .
2
The smallest Lipschitz constant for the domain is
π
L= max |φ02 (x)| = >1.
[−π/4,π/4] 2
Thus φ2 is not contractive on [−π/4, π/4].
3 marks
(d) In the case of the function

1 + ex
φ(x) :=
4
determine the smallest Lipschitz constant L on [0, 1] and explain why
φ : [0, 1] → [φ(0), φ(1)] ⊂ [0, 1] .

21-4-2005 17:22
ANSWER
φ0 (x) = ex /4 > 0 and φ00 (x) = ex /4 > 0 .

Thus φ0 is increasing on [0, 1] and the maximum of |φ0 (x)| is attained at x = 1 and the
smallest Lipschitz constant L is given by
L = φ0 (1) = e/4 < 1 . 3 marks
Since φ is increasing on [0, 1] we have
φ : [0, 1] → [φ(0), φ(1)] .
φ(0) = 1/2 > 0 and φ(1) = (1 + e)/4 < 1 and hence [φ(0), φ(1)] ⊂ [0, 1].
2 marks
21-4-2005 17:22
Exercises 8A: Cauchy sequences and subsequences in R

1. Prove that if a sequence (xn ) converges then it is a Cauchy sequence. (This has been
asked in previous MA2034A exam papers.)
ANSWER
From the definition of convergence of (xn ) to x there exists a N = N (/2) such that
|xn − x| < /2 for all n ≥ N .
If both n ≥ N and m ≥ N then
|xn − xm | = |(xn − x) − (xm − x)| ≤ |(xn − x)| + |(xm − x)| < /2 + /2 =
by the triangle inequality and the above bound. Thus (xn ) is a Cauchy sequence.
2. Show that if (xn ) satisfies the property

1
|xn+1 − xn | ≤ , n = 1, 2, · · ·
2n
then it is a Cauchy sequence.
Given that a Cauchy sequence converges show that if x denotes the limit of the sequence
then
1
|xn − x| ≤ n−1 .
2
With S denoting the set
S := {xn : n = 1, 2, · · · }
also show that
3
sup S − inf S ≤ .
2
This was the last part of the Jan 1999 question 1. Most people found this difficult.
ANSWER
We are given |xn+1 − xn | < 1n . Thus

2
xn+j − xn = (xn+j − xn+j−1 ) + (xn+j−1 − xn+j−2 )
+ · · · + (xn+1 − xn )
21-4-2005 17:22
and by the triangle inequality
|xn+j − xn | ≤ |xn+j − xn+j−1 | + |xn+j−1 − xn+j−2 |

+ · · · + |xn+1 − xn |
n+j−1 n
1 1
≤ + ··· +
2 2
n j−1 !
1 1 1
= 1 + + ··· +
2 2 2
n n−1
1 1
≤ 2 = .
2 2
Since the right hand side does not involve j and tends to 0 as n → ∞ the sequence is a
Cauchy sequence and hence converges.
With
x = lim xm ,
m→∞
and letting j → ∞ in the above we get

n−1
1
|x − xn | ≤ .
2
As a consequence of this inequality the set {x2 , x3 , · · · } is contained in the interval

[x − 1/2, x + 1/2]. Only x1 of the set s = {x1 , x2 , x3 , · · · } may lie outside of this interval.
The complete set is thus contained in [x − 1, x + 1/2] or [x − 1/2, x + 1] from which it
follows that
3
sup S − inf S ≤ .
2
3. (This is similar to question 2.)

Let (xn ) be a sequence of vectors in Rp which are such that x0 = 0 and
kxn+1 − xn k2 ≤ rr , n = 0, 1, 2, · · · for some r ∈ [0, 1).
Show that the sequence converges to some x ∈ Rp and that x satisfies

1
kxk2 ≤ .
1−r
21-4-2005 17:22
ANSWER
To show that the sequence converges we show that the sequence is a Cauchy sequence.
xn+q − xn = (xn+q − xn+q−1 ) + · · · + (xn+1 − xn ).

kxn+q − xn k2 ≤ k(xn+q − xn+q−1 k2 + · · · + kxn+1 − xn k2
∞
n+q−1 n
X rn
≤ r + ··· + r ≤ rk = .
n
1 − r
Since |r| < 1 the right hand side tends to 0 as n → ∞ for all q > 0 which is sufficient to
show that the sequence is a Cauchy sequence.
Since (xn ) is a Cauchy sequence it converges to some x ∈ Rp .
Since x0 = 0 we have
n
X n
X
xn = x0 + (xk − xk−1 ) = (xk − xk−1 ).
k=1 k=1
By the triangle inequality we have

n
X
kxn k2 ≤ kxk − xk−1 k2
k=1
n ∞
X
k−1
X 1
≤ r ≤ rk = .
k=1 k=0
1−r
4. The following question relates to the Bolzano-Weierstrass theorem concerning sequences

in R.
Let (xn ) denote the bounded sequence given by
√
xn := sin n.
Observe that since the sine function sin x is increasing in [−π/2, π/2] then for all y ∈
[−1, 1] there is unique x ∈ [−π/2, π/2] with sin x = y and from the periodicity of the
sine function
sin(x + 2kπ) = y, for k = ±1, ±2, · · ·
By defining nk ∈ N to be the integer part of (x + 2kπ)2 and by considering the subse-
quence (xnk ) of (xn ), show that xnk → y as k → ∞.
(In your answer you can assume all the usual properties and identities of the sine function
such as
sin(x ± y) = sin x cos y ± cos x sin y etc.
and
| sin x| ≤ |x| for all x ∈ R.)
21-4-2005 17:22
ANSWER
The key point here is to note that y is given by
y = sin(x) = sin(x + 2kπ) for all k.
Thus
p
xnk − y = sin b(x + 2kπ)2 c − sin(x + 2kπ),
where bzc = integer part of z ∈ R.
To rewrite this in a form where we can establish that this difference is small we note the
trigonometric identities
sin(a + b) = sin a cos b + cos a sin b

sin(a − b) = sin a cos b − cos a sin b
sin(a + b) − sin(a − b) = 2 cos a sin b
which with c = a + b and d = a − b gives

c+d c−d
sin c − sin d = 2 cos sin .
2 2
Using the properties of the cosine and sine functions (which can be established from a
geometric definition of these functions) that
| cos u| ≤ 1 and | sin u| ≤ |u| for all u ∈ R
we have
| sin c − sin d| ≤ |c − d| .
In our case with p
c= b(x + 2kπ)2 c and d = x + 2kπ
we have
c2 − d2 b(x + 2kπ)2 c − (x + 2kπ)2
c−d= =p
c+d b(x + 2kπ)2 c + (x + 2kπ)
and by noting that the integer part of a number can differ by at most 1 from the number
we get
1
|xnk − y| ≤ p → 0 as k → ∞.

2
b(x + 2kπ) c + (x + 2kπ)
21-4-2005 17:22
Exercises 8B: Sequences and series of functions

1. The following was the last part of question 2 of the 2004 MA2930 exam paper.
Let (fn ), fn : [a, b] → R denote a sequence of functions. Define what it means for (fn )
to converge pointwise on [a, b].
[2 marks]
(i) Determine the pointwise limit function in the case

fn : (−1, 1] → R, fn (x) := xn , n = 1, 2, · · · .
Explain why the convergence is not uniform.
[2 marks]
(ii) Show that the pointwise limit function of the sequence
nx
fn : [0, 1] → R, fn (x) := , n = 1, 2, · · ·
1 + n2 x2
is the zero function but that
max{|fn (x)| : 0 ≤ x ≤ 1} 6→ 0 as n → ∞. [4 marks]
ANSWER
(fn ) converges pointwise on [a, b] if the sequence of numbers (fn (x)) converges for all
x ∈ [a, b].
2 marks
(i) If |x| < 1 then xn → 0 as n → ∞. If x = 1 then xn = 1 for all n. The pointwise

limit function is f : (−1, 1] → R,
(
0, if −1 < x < 1,
f (x) :=
1, if x = 1.
Each function fn is continuous on (−1, 1] and the limit function is f is discontinuous
at x = 1 which implies that the convergence is not uniform.
2 marks
(ii) For x = 0 we have fn (0) = 0.
For x > 0 we have
x/n 0
fn (x) = 2 2
→ = 0 as n → ∞.
1/(n ) + x 0 + x2
Thus for all x ∈ [0, 1] we have fn (x) → 0 as n → ∞.
By inspection when x = 1/n we have fn (1/n) = 1/(1 + 1) = 1/2 for all n. Thus
1
max{|fn (x)| : 0 ≤ x ≤ 1} ≥ fn (1/n) =
2
and we hence do not have convergence to 0.
4 marks
21-4-2005 17:22
2. The following was the last part of question 1 of the 2004 MA2930 exam paper.
Let (fn ), fn : [a, b] → R denote a sequence of continuous functions.
Define the uniform norm of (fn ) and explain what it means for (fn ) to converge to
f : [a, b] → R uniformly on [a, b].
[2 marks]
P
The Weierstrass M-test gives a sufficient condition for the series fn to converge uni-
formly on [a, b]. State the conditions of this test.
[2 marks]
Use the test to show that the following series converge uniformly on the given domain.
∞
X xn
(i) on |x| ≤ r < ∞. [2 marks]
0
nn
∞
X (n!)2 n
(ii) x on |x| ≤ r < 4. [2 marks]
0
(2n)!
∞
X 1
(iii) on R. [2 marks]
0
x2 + 2n
ANSWER
The uniform norm of the continuous function fn is

kfn k∞ = sup{|fn (x)| : a ≤ x ≤ b} = max{|fn (x)| : a ≤ x ≤ b}.
(fn ) converges to f uniformly means that kfn − f k∞ → 0 as n → ∞.
2 marks
The conditions of the Weierstrass M-test are satisfied if kfn k∞ ≤ Mn with the series
P
Mn converging.
2 marks
(i) In this case fn (x) = xn /nn . On the domain {x : |x| ≤ r} we have

rn
kfn k∞ =
=: Mn .
nn
P
We test for the convergence of the series Mn by using the root test.
r
Mn1/n = → 0 as n → ∞.
n
As the limit is less than 1 the series converges and by the M-test the power series
converges uniformly.
2 marks
21-4-2005 17:22
(ii) In this case fn (x) = (n!)2 xn /(2n)!. On the domain {x : |x| ≤ r} we have
(n!)2 n
kfn k∞ = r =: Mn .
(2n)!
P
We test for the convergence of the series Mn by using the ratio test.
Mn+1 (n + 1)2
= r
Mn (2n + 2)(2n + 1)
(1 + 1/n)2 r
= r→ as n → ∞.
(2 + 2/n)(2 + 1/n) 4
As the limit is less than 1 when r < 4 the series converges and by the M-test the
power series converges uniformly.
2 marks
(iii) In this case fn (x) = 1/(x2 + 2n ) and on the domain
P∞ R we have by inspection that
n n
kfn k∞ = fn (0) = 1/2 . The geometric series n=0 1/2 is convergent and by the
M-test the series converges uniformly on R.
2 marks
(a) Let I ⊂ R and let f be a bounded function on I. Define the uniform norm kf k of
f on I.
[1 mark]
(b) Let I ⊂ R and let fn : I → R, n = 1, 2, · · · and f : I → R be functions.

(i) Define what it means for (fn ) to converge pointwise on I.
[2 marks]
(ii) Define what it means for (fn ) to converge to f uniformly on I.
[2 marks]
(c) In each of the following cases of sequences of functions, determine the pointwise
limit function and determine whether or not the convergence is uniform.
(i)
xn
fn : [0, ∞) → R, fn (x) := .
1 + xn
[4 marks]
(ii)
cos(nx)
fn : R → R, fn (x) := .
n
[3 marks]
21-4-2005 17:22
(iii)
fn : [0, 1] → R, fn (x) := xn (1 − x).
[4 marks]
(iv)
fn : [0, ∞) → R, fn (x) := xe−nx .
[4 marks]
ANSWER
(a)
kf k := sup{|f (x)| : x ∈ I} .
1 mark
(b) (i) (fn ) converges pointwise on I if the sequence of numbers (fn (x)) converges for
every x ∈ I.
2 marks
(ii) (fn ) converges uniformly on I to f , f : I → R, if
kfn − f k → 0 as n → ∞.
2 marks
(c) (i) If 0 ≤ x < 1 then xn → 0 as n → ∞ and we get fn (x) → 0 as n → ∞.
If x = 1 then fn (1) = 1/2 → 1/2 as n → ∞.
If x > 1 then
1 1
fn (x) = n → = 1 as n → ∞.
(1/x ) + 1 0+1
Hence the pointwise limit function f is the discontinuous function

0,
 0 ≤ x < 1,
f : [0, ∞) → R, f (x) := 1/2, x = 1,

1, x > 1.

As each fn is continuous and the pointwise limit is discontinuous the conver-

gence is not uniform.
4 marks
(ii) For all x ∈ R
cos(nx) 1
n ≤ n → 0 as n → ∞.

Thus (fn ) converges uniformly to the zero function on R which is thus the
pointwise limit.
3 marks
21-4-2005 17:22
(iii) If x = 1 then fn (1) = 0 and (fn (1)) is a constant sequence. If 0 ≤ x < 1 then
xn → 0 as n → ∞. Thus (fn ) converges pointwise to the zero function on
[0, 1]. To test for uniform convergence we determine the uniform norm of each
fn .
As fn (x) ≥ 0 with fn (0) = fn (1) = 0 we need to find the maximum of fn (x),
0 < x < 1.
fn0 (x) = nxn−1 − (n + 1)xn = xn−1 (n − (n + 1)x).
The maximum occurs at x = n/(n + 1). Since for this x, xn < 1 we have
n 1
|fn (n/(n + 1))| ≤ 1 − = → 0 as n → ∞.
n+1 n+1
Hence fn → 0 uniformly on [0, 1].
4 marks
(iv) If x = 0 then fn (0) = 0 and (fn (0)) is a constant sequence. For x > 0 the ratio
test gives
fn+1 (x) e−(n+1)x
= = e−x < 1
fn (x) e−nx
and hence fn (x) → 0 as n → ∞. Thus (fn ) converges pointwise to the zero
function on [0, 1].
To test for uniform convergence we determine the uniform norm of each fn .
1
fn0 (x) = e−nx (1 − nx) = 0 when x = .
n
As fn0 (x) > 0 in [0, 1/n) and fn0 (x) < 0 in (1/n, ∞) this is a local maximum.
We have
e−1
kfn k = fn (1/n) = → 0 as n → ∞.
n
The sequence converges uniformly to the zero function.
4 marks
(i) Let I ⊂ R and let f be a bounded function on I. Define the uniform norm kf k of
f on I.
[1 mark]
(ii) Let I ⊂ R and let fn : I → R, n = 1, 2, · · · and f : I → R be functions.

(a) Define what is means by saying that fn → f pointwise on I.
[2 marks]
(b) Define what is means by saying that fn → f uniformly on I.
[2 marks]
21-4-2005 17:22
(iii) In the following I ⊂ R and (fn ) is a sequence of functions such that fn : I → R.

(a) Explain why when I = (0, 1) and fn (x) := xn , kfn k = 1 for all n. Hence show
that the sequence (fn ) converges pointwise on I but not uniformly on I.
[3 marks]
(b) Show that when I = [0, 1] and
nx
fn (x) := ,
1 + n2 x2
kfn k = 1/2 for all n. Hence show that the sequence (fn ) converges pointwise
on I but not uniformly on I.
[5 marks]
(c) Suppose that (fn ) is a sequence of continuous functions which converges point-
wise on I to f . What can you conclude about the uniformity of the convergence
in the following cases: (1) when f is continuous on I, (2) when f is discontin-
uous on I.
[2 marks]
(iv) Let fn : I → R, n = 1, 2, · · · be a sequence of bounded functions. Define what
it means for (fn ) to be a Cauchy sequence in the uniform norm and show that if
fn → f uniformly then (fn ) is a Cauchy sequence. (In your answer you can assume
that the uniform norm satisfies all the norm axioms, e.g. the triangle inequality.)
[5 marks]
ANSWER
(i)
kf k := sup{|f (x)| : x ∈ I} .
1 mark
(ii) (a) (fn ) converges pointwise on I if the sequence of numbers (fn (x)) converges for
every x ∈ I.
2 marks
(b) (fn ) converges uniformly on I to f , f : I → R, if
kfn − f k → 0 as n → ∞.
2 marks
(iii) (a) For all x ∈ I, xn < 1 and hence 1 is an upper bound. Also, limx→1 xn = 1.
Given any > 0, the definition of the limit implies that 1 − is exceeded by
some x ∈ I and thus no number less than 1 can be an upper bound. Hence
the least upper bound is 1 and kfn k = 1.
For all x ∈ I we have |x| < 1 and consequently xn → 0 as n → ∞. The
sequence converges pointwise to the zero function but as kfn k 6→ 0 the sequence
does not converge uniformly.
3 marks
21-4-2005 17:22
(b) As fn is continuous on [0, 1] it attains its maximum on [0, 1]. fn (0) = 0 and
fn (1) = n/(1 + n2 ) = 1/((1/n) + n). f1 (1) = 1/2 and for n ≥ 2 fn (1) < 1/n ≤
1/2. To find the maximum we consider turning points of fn .
(1 + n2 x2 )n − (nx)(2n2 x) n(1 − n2 x2 )
fn0 (x) = = =0
(1 + n2 x2 )2 (1 + n2 x2 )2
when x = 1/n. fn increases in [0, 1/n) and decreases in (1/n, ∞). Thus
1
kfn k = fn (1/n) = .
2
For the pointwise convergence observe that for x = 0, fn (x) = 0 for all n. For
x > 0,
1/(nx) 0
fn (x) = 2 → = 0 as n → ∞.
(1/nx) + 1 0+1
Thus the sequence converges pointwise to the zero function. As kfn k 6→ 0 the
sequence does not converge uniformly.
5 marks
(c) (1) If the limit function is discontinuous then this is sufficient to prove that
the convergence is not uniform. (2) If the limit function is continuous then
nothing can be concluded about whether or not the convergence is uniform.
2 marks
(iv) (fn ) is a Cauchy sequence if for every > 0 there exists an N such that
kfn − fm k < , for all n ≥ N and m ≥ N .
(fn ) converges uniformly to f means that for every > 0 there exists a N such
that
kfn − f k < /2 for all n ≥ N .
Then for all m, n ≥ N and for all x ∈ I we compare both fm and fn with the limit
f to give
|fm (x) − fn (x)| = |(fm (x) − f (x)) + (f (x) − fn (x))|

≤ |fm (x) − f (x)| + |f (x) − fn (x)|
≤ kfm − f k + kf − fn k < /2 + /2 = .
As this is true for for all x ∈ I we have kfm − fn k < as required.
5 marks
21-4-2005 17:22
(a) State the Weierstrass M -test.

[3 marks]
(b) In the following you may assume in your answer that the series
∞
X 1
k=1
kp
converges for all p > 1 and diverges for all p ≤ 1.
Let n
X cos(kx)
fn : R → R, fn (x) := .
k=1
k3
Use the the Weierstrass M-test to explain why (fn ) and (fn0 ) converge uniformly on
R.
[4 marks]
Does the sequence (fn00 ) converge pointwise on R?
[1 mark]
(c) Determine the radius of convergence of the following power series and state regions
in which the series converge uniformly.
(i)
∞
X xk
k=0
k!
[3 marks]
(ii) In the following α ∈ R is not an integer.
∞
X α(α − 1) · · · (α − k + 1) α(α − 1) 2
xk = 1 + αx + x + ···
k=0
k! 2
[4 marks]
(iii)
∞
X k 3 xk
.
k=0
3k
[3 marks]
(iv)
∞
X
k k xk .
k=0
[2 marks]
21-4-2005 17:22
ANSWER
(a) The Weierstrass M -test relates to series of functions (fk ). Let fk : I → R and
suppose that
kfk k ≤ Mk ,
P
where kfk k is the uniform norm
P of fk on I. If the series of numbers Mk converges
then the series of functions fk converges uniformly on I.
3 marks
(b) To apply the M-test to the series for fn (x) we note that | cos(kx)| ≤ 1 so that the
kth component function is bounded by
1
Mk = .
k3
1/k 3 converges and thus the series for fn converges uniformly on R.
P
The series
0
fn is given by
n
X sin(kx)
fn0 (x) = − 2 .
k=1
k
The kth component function is bounded by
1
Mk = .
k2
1/k 2 converges and thus the series for fn0 converges uniformly on R.
P
The series
4 marks
fn00 is given by
n
X cos(kx)
fn00 (x) = − .
k=1
k
When x = 0, cos(0) = 1 and −fn00 (0)
is the nth partial sum of the divergent
harmonic series. Hence the series does not converge pointwise on R.
1 mark
(c) (i) Let fk (x) = xk /k!. On |x| ≤ r
rk
|fk (x)| ≤ =: Mk .
k!
By the ratio test
Mk+1 rk+1 /(k + 1)! r
= k
= → 0 as k → ∞.
Mk r /k! k+1
P
The series Mk converges for all r and hence the radius of convergence is ∞.
The series converges uniformly in any region of the form |x| ≤ r.
3 marks
21-4-2005 17:22
(ii) Let fk (x) = ak xk where

α(α − 1) · · · (α − k + 1)
ak := .
k!
On |x| ≤ r
|fk (x)| ≤ |ak |rk =: Mk .
Using the ratio test

Mk+1 ak+1 α − k α/k − 1
= r=
r=
r → r as k → ∞.
Mk ak k +1 1 + 1/k
Thus the series converges absolutely if r < 1 and diverges for r > 1. The radius
of convergence is R = 1 and the series converges uniformly in [−r, r] for all r
satisfying 0 ≤ r < 1.
4 marks
(iii) Let fk (x) = ak xk where
k3
ak := .
3k
On |x| ≤ r
|fk (x)| ≤ |ak |rk =: Mk .
Mk+1 ak+1 (k + 1)3 (1 + 1/k)3 r

= r= 3 r = r→ .
Mk ak 3k 3 3
The series converges if r < 3 and diverges for r > 3. The radius of convergence
is R = 3. The series converges uniformly in |x| ≤ r for all r < 3.
3 marks
(iv) Let fk (x) = k k xk . On |x| ≤ r
|fk (x)| ≤ (kr)k =: Mk .
Using the root test

1/k
Mk = kr.
1/k
The sequence (Mk ) only converges when r = 0 The radius of convergence is
R = 0.
2 marks
21-4-2005 17:22
(i) State the Weierstrass M -test.

[3 marks]
(ii) In the following you may assume in your answer that the series
∞
X 1
k=1
kp
converges for all p > 1.

(a) Let
1
fk (x) := .
x + k2
2
Show that
C 9
|fk0 (x)| ≤ , C= √ , for all x ∈ R.
k3 8 3
[4 marks]
Hence use the Weierstrass M -test to show that
X∞ X∞
fk (x) and fk0 (x)
k=1 k=1
both converge uniformly on R.

[2 marks]
(b) Show that the following series converges uniformly on [0, ∞):
∞ k
X 1 1−x
2 .
k=1
k 1 + x
[Hint: make the substitution y = (1 − x)/(1 + x).]

[3 marks]
(iii) Determine the radius of convergence of the following power series and state regions
(a)
∞
X 2k xk
. [3 marks]
k=0
k!
(b)
∞
X k 3 xk
[3 marks]
k=0
3k
(c)
∞
X
k!xk . [2 marks]
k=0
21-4-2005 17:22
ANSWER
(i) The Weierstrass M -test relates to series of functions (fk ). Let fk : I → R and
suppose that
kfk k ≤ Mk ,
P
3 marks
(ii) (a)
−2x
fk0 (x) = .
(x + k 2 )2
2
To determine the maximum of |fk0 (x)| we need to consider turning points of fk0
which correspond to points at which fk00 (x) = 0.
(x2 + k 2 )2 (−2) − (−2x)2(x2 + k 2 )2x

fk00 (x) =
(x2 + k 2 )4
and hence fk00 (x) = 0 when
(x2 + k 2 )2 (−2) = −8x2 (x2 + k 2 ),

i.e. when x2 + k 2 = 4x2 , 3x2 = k 2 .
√
At x = ±k/ 3, x2 + k 2 = 4k 2 /3 and
√
√ C 2/ 3 9
|fk0 (±k/ 3)| = 3 where C = 2 =
√ .
k (4/3) 8 3
As we only have 2 turning values we obtain the bound
C
|fk0 (x)| ≤ for all x ∈ R.
k3
4 marks
As for all x ∈ R,
1 C
and |fk0 (x)| ≤ 3
|fk (x)| ≤
2
k k
2 3
P P
and 1/k and 1/k are standard convergent series the series converges
uniformly by the Weierstrass M-test.
2 marks
(b) Let
1−x yk
y= and fk (x) = 2 .
1+x k
For 0 ≤ x ≤ 1, |1 − x| ≤ 1 ≤ 1 + x. For 1 < x, |1 − x| = x − 1 ≤ 1 + x. That
is for all x ≥ 0, |1 − x| ≤ |1 + x|. Thus |y| ≤ 1 for all x ≥ 0 and
1
|fk (x)| ≤ .
k2
21-4-2005 17:22
1/k 2 is a standard convergent series it follows that the series converges

P
As
uniformly when x ≥ 0 by the Weierstrass M-test.
3 marks
(iii) (a) Let fk (x) = 2k xk /k!. As (k + 1)! = (k + 1)k! etc. we have for x 6= 0,

fk+1 (x) 2x
fk (x) k + 1 → 0 as k → ∞.
=
By the ratio test the series converges for all x ∈ R and by the M-test the series
converges uniformly in all regions of the form [−R, R], R > 0. The radius of
convergence is ∞.
3 marks
(b) Let fk (x) = k 3 xk /3k . We have for x 6= 0,
fk+1 (x) (k + 1)3 x (1 + 1/k)3 x

→ |x| as k → ∞.
fk (x) = 3k 3 =

3 3
By the ratio test the series converges for |x| < 3 and by the M-test the series
converges uniformly in all regions of the form [−r, r], r < 3. The radius of
convergence is 3.
3 marks
(c) Let fk (x) = k!xk . We have for x 6= 0,

fk+1 (x)
fk (x) = (k + 1)|x|.

This is unbounded for all x 6= 0 and by the ratio test the series diverges. Hence
the series only converges at x = 0. The radius of convergence is 0.
2 marks
7. These were part of question 4 of the Jan 2000 and Jan 2001 MA2034A exam papers.
(a) Let (fn ), fn : I → R, denote a sequence of continuous functions defined on I. If

(fn ) converges to f uniformly on I then what properties will the limit function f
have?
(b) In each of the following cases of sequences of functions, determine whether or not
the sequence converges pointwise on its given domain. If the sequence does converge
pointwise then give the pointwise limit function and determine whether or not the
convergence is uniform.
(i)
fn : (−1, 1] → R, fn (x) := xn , n = 1, 2, · · · . [2 marks]
21-4-2005 17:22
(ii)
sin nx
fn : R → R, fn (x) := n = 1, 2, · · · . [3 marks]
n
(iii)
fn : R → R, fn (x) := cos nx . [3 marks]
(iv)
xn
fn : (−1, 1] → R, fn (x) := , n = 1, 2, · · · .
(1 + xn )2
(v)
xn
fn : [0, ∞) → R, fn (x) := , n = 1, 2, · · · .
(1 + xn )n
ANSWER
(a) As each fn is continuous and (fn ) converges to f uniformly on I then f is continuous

on I.
1 mark
(b) (i)
xn → 0 as n → ∞ if |x| < 1.
Thus for x ∈ (−1, 1), fn (x) → 0 as n → ∞. Also, fn (1) = 1 for all n. Thus
(fn (x)) converges for all x in (−1, 1] and hence the sequence of functions does
converge pointwise. The pointwise limit function is
(
0, if x ∈ (−1, 1),
f (x) := lim fn (x) =
n→∞ 1, if x = 1.
As each fn is continuous and f is discontinuous this indicates that the conver-

2 marks
(ii) By the properties of the sine function we have for all x ∈ R that
1
|fn (x)| ≤
n
and thus
1
kfn k ≤
→ 0 as n → ∞.
n
Thus the sequence converges uniformly to the function f
f : R → R, f (x) := 0 .
Uniform convergence implies pointwise convergence and hence f is also the

pointwise limit.
3 marks
21-4-2005 17:22
(iii) If we let x = π then fn (π) = cos nπ = (−1)n . The sequence of numbers (fn (π))
does not converge and thus the sequence does not converge pointwise and as a
consequence it does not converge uniformly.
3 marks
(iv)
xn → 0 as n → ∞ if |x| < 1.
Thus for x ∈ (−1, 1), fn (x) → 0/(1 + 0)2 = 0 as n → ∞. Also, fn (1) = 1/4 for
all n. Thus (
0, if x ∈ (−1, 1),
f (x) := lim fn (x) =
n→∞ 1/4, if x = 1.
As each fn is continuous and f is discontinuous this indicates that the conver-
4 marks
(v) Let gn : R → R,
x (1/x)n−1
gn (x) := = .
1 + xn (1/x)n + 1
If 0 ≤ x < 1 then
n
x
fn (x) = ≤ xn → 0 as n → ∞.
1 + xn
If x = 1 then
fn (1) = (1/2)n → 0 as n → ∞.
If x > 1 then (1/x)n → 0 as n → ∞ and
n n(n−1)
(1/x)n−1

1
fn (x) = n < → 0 as n → ∞.
(1/x) + 1 x
The pointwise limit function is f : R → R, f (x) := 0.
4 marks
To establish that the convergence is uniform we need to bound gn and hence
fn . We note that gn (0) = 0 and gn (x) → 0 as |x| → ∞. For turning values we
have
(1 + xn ) − x(nxn−1 )
gn0 (x) = n 2 = 0 when 1 = (n − 1)xn .
(1 + x )
As there is only one turning value it is the point where gn has a global maxi-
mum. Thus
1/n
(1/(n − 1))1/n

1
max |gn (x)| ≤ <
R 1 + 1/(n − 1) (n − 1)
and
1
kfn k = max |fn (x)| ≤ → 0 as n → ∞.
R n−1
6 marks
21-4-2005 17:22
8. This was most of question 5 of the January 2001 MA2034A exam paper.
(i) State the Weierstrass M -test.

[3 marks]
Use the Weierstrass M -test to show that the following series converge uniformly on
R.
(a)
∞
X cos kx sin kx
+3 2 .
k=1
k3 k
[3 marks]
(b) √
∞
X x2 + k − |x|
.
k=1
k2
[4 marks]
In your answer you may assume that the series

∞
X 1
k=1
kp
converges for all p > 1.

(ii) Determine the radius of convergence of the following power series and state regions
(a)
∞ 2 2 2
X k k 1 2 2 3
x = x+ 2 x + x3
k=0
5k 5 5 53
2 2
4 4 5
+ 4 x + x5 + · · · .
5 55
[4 marks]
(b)
∞
X α(α − 1) · · · (α − k + 1) x k
k=0
k! 2
x α(α − 1) x 2
= 1+α + + ··· .
2 2 2
[4 marks]
21-4-2005 17:22
ANSWER
(i) The Weierstrass M -test relates to series of functions (fk ). Let fk : I → R and
suppose that
kfk k ≤ Mk ,
P
3 marks
(a) With
cos kx sin kx
fk (x) := 3 +3 2
k k
it follows that for all x ∈ R
1 3
|fk (x)| ≤ Mk := 3 + 2 .
k k
2
1/k 3 are standard convergent series and thus
P P P
1/kP and Mk converges.
Thus fk converges uniformly on R.
3 marks
(b) Let
√
x2 + k − |x|
fk (x) :=
k2
k 1 1
= 2
√ = √ ≤ 3/2 =: Mk
k ( x2 + k + |x|) k( x2 + k + |x|) k
where the bound was obtained by observing that the denominator takes its
smallest
P value in R when x = 0. P
Mk is a standard convergent series and thus the series fk converges uni-
formly on R.
4 marks
(ii) (a) Let
k2 k
fk (x) = x .
5k
On |x| ≤ r
k2 k
|fk (x)| ≤ k r =: Mk .
5
2
(k + 1)2 r

Mk+1 1 r r
= 2 = 1+ → as k → ∞.
Mk k 5 k 5 5
P P
Thus the series Mk converges if r < 5 and the series fk converges uni-
formly in [−r, r] for all r satisfying 0 ≤ r < 5. The radius of convergence is
R = 5.
4 marks
21-4-2005 17:22
(b) Let
α(α − 1) · · · (α − k + 1) x k
fk (x) := .
k! 2
On |x| ≤ r

α(α − 1) · · · (α − k + 1) r k
|fk (x)| ≤

2 =: Mk .

k!

Mk+1 α − k r α/k − 1 r r
= = → as k → ∞.
Mk k+1 2 1 + 1/k 2
2
P P
Thus the series Mk converges if r < 2 and the series fk converges uni-
formly in [−r, r] for all r satisfying 0 ≤ r < 2. The radius of convergence is
R = 2.
4 marks
9. This was part of question 5 of the January 2000 paper.

Use the Weierstrass M -test to show that the following series converge uniformly on R.
(a)
∞
X 1
.
k=1
x + k2
2
(b)
∞
X sin kx
.
k=1
k 3/2
(c) Determine the radius of convergence of the following power series and state the
region in which the series converge uniformly.
(i)
∞
X k k 1 2 2 3
x = x+ 2 x + x3 + · · · .
k=0
2k 2 2 23
(ii)
∞
X 1 k 1 2 1
k
x =1+x+ 2 x + 3 x3 + · · · .
k=0
k 2 3
21-4-2005 17:22
ANSWER
(a) The denominator takes its smallest value when x = 0 and hence
1
|fk (x)| ≤
.
k2
1/k 2 is a standard convergent series and hence the series of functions
P
The series
converges uniformly on R.
3 marks
(b)
sin kx 1
k 3/2 ≤ k 3/2

1/(k 3/2 ) is a standard convergent series. Hence the series of functions converges
P
and
uniformly on R.
2 marks
(c) (i) Let
k k
ak := x .
2k
ak+1 (k + 1) x x x
= = (1 + 1/k) → as k → ∞.
ak k 2 2 2
Thus the series converges absolutely if |x| < 2 and converges uniformly in [−r, r]
for all r satisfying 0 ≤ r < 2.
4 marks
(ii) Let
x k
ak := .
k
Using the root test
|x|
|ak |1/k =
→ 0 as k → ∞.
k
Thus the series converges absolutely for all x ∈ R and converges uniformly in [−r, r]
for all r satisfying 0 ≤ r < ∞.
4 marks
10. Book work question.

Let I ⊂ R and let (fn ) be a sequence of continuous functions such that fn : I → R.
Suppose that (fn ) converges uniformly to f on I and note that
|f (x) − f (c)| = |(f (x) − fn (x)) + (fn (x) − fn (c)) + (fn (c) − f (c))|
≤ |f (x) − fn (x)| + |fn (x) − fn (c)| + |fn (c) − f (c)|

≤ 2 sup |f (x) − fn (x)| + |fn (x) − fn (c)| .
x∈I
Show that f is also continuous on I.

21-4-2005 17:22
ANSWER
Let > 0. By the uniform convergence there exists N such that
sup |f (x) − fn (x)| <

x∈I
for n ≥ N . The continuity of fN gives the existence of δ such that
|fN (x) − fN (c)| < whenever |x − c| < δ .
Thus for |x − c| < δ we have
|f (x) − f (c)| < 3 whenever |x − c| < δ
which is sufficient to establish the continuity of f at the arbitrary point c.
11. The following is a miscellaneous collection of questions relating to sequences and series
of functions.
(a) Show that if fn (x) := x + 1/n and f (x) := x for all x ∈ R then fn → f uniformly
on R but that (fn2 ) does not converge uniformly on R.
(b) Let C (1) [a, b] denote the set of continuously differentiable functions on a finite in-
terval [a, b]. For each f in C (1) [a, b], define
kf kC 1 := kf k + kf 0 k .
Show that k.kC 1 satisfies the norm requirements of non-negativity, linearity and
the triangle inequality. (Remark: It can be shown that the linear space C (1) [a, b] is
complete in this norm.)
P P
(c) Use the Weierstrass M -test to deduce that if |an | and |bn | converge then the
Fourier series ∞
a0 X
+ (an cos nx + bn sin nx)
2 n=1
(d) Let
n
X cos(kx)
sn (x) := .
k=1
k5
Explain why (sn ), (s0n ), (s00n ) and (s000 0000
n ) all converge uniformly on R but that (sn (0))
does not converge.
21-4-2005 17:22
ANSWER
(a) fn (x) = x + 1/n and f (x) = x and hence fn (x) − f (x) = 1/n. Thus
1
kfn − f k∞ = → 0 as n → ∞.
n
fn2 − f 2 = (fn + f )(fn − f )
and thus in this case
2 1
2 1
fn (x) − f (x) = 2x +
n n
and
2 2
1 1
sup fn (x) − f (x) = sup 2x + = ∞ for all n.
x∈R n x∈R n
(b) Nonnegativity: Clearly kf kC 1 ≥ 0 since kf k ≥ 0 and kf 0 k ≥ 0. If kf kC 1 = 0 then
kf k = 0 and hence f (x) = 0 for all x ∈ [a, b].
Linearity: By using the linearity of the uniform norm
kαf kC 1 = kαf k + kαf 0 k = |α| kf k + |α| kf 0 k = |α| kf kC 1 .
Triangle inequality: By the triangle inequality for the uniform norm
kf + gkC 1 = kf + gk + kf 0 + g 0 k
≤ (kf k + kgk) + (kf 0 k + kg 0 k)
= kf kC 1 + kgkC 1 .
(c) With fn (x) := an cos nx + bn sin nx we have
|fn (x)| ≤ |an | + |bn |

P
P triangle inequality and that | cosP
by the nx| ≤ 1 and | sin nx| ≤ 1. As |an |
and |bn | both converge we have that (|an | + |bn |) also converges. Hence the
conditions of the Weierstrass M -test apply with Mn = |an | + |bn | and the series
(d)
n
X cos(kx)
sn (x) := ,
k=1
k5
n
X sin(kx)
s0n (x) := − ,
k=1
k4
n
X cos(kx)
s00n (x) := − ,
k=1
k3
n
X sin(kx)
s000
n (x) := ,
k=1
k2
n
X cos(kx)
s0000
n (x) := .
k=1
k
21-4-2005 17:22
The kth term in the series of sn , s0n , s00n and s000

n are bounded on R by respectively
5 4 3 2
1/k , 1/k , 1/k and 1/k and as the series of the bounds converge the sequence of
functions converge uniformly on R. However
n
X 1
s0000
n (0) = .
k=1
k
are the partial sums of the harmonic series which is divergent.

Anma 2930

Uploaded by

Copyright:

Available Formats

Anma 2930

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Anma 2930

Uploaded by

Copyright:

Available Formats

21-4-2005 17:22

c M. K. Warby, J. E. Furter MA2930 ANALYSIS, Exercises Page 1

Exercises on Sequences and Series of Real Numbers

|xn − x| <  for all n ≥ N .

(iii) u ∈ R is a least upper bound or supremum of (xn ) if it is an upper bound of

(b) (i) The sequence converges because it is a combination of standard convergent

103n2 − 8 103 − 8/n2 103

1 3(k + 1)2 + 2(k + 1)

(b) Prove that if a sequence (xn ) converges then it is a Cauchy sequence.

(ii) Given that

(i) Define what it means for the sequence (xn ) to be bounded.

The increasing property of the sequence then implies that

and we satisfy the convergence definition for all n ≥ N .

7n4 + n2 − 2 7 + (1/n2 ) − (2/n4 ) 7 1

n3 + 2n2 + 1 1 + (2/n) + (1/n3 ) 1

xn = n2 (1 + 1/n2 )1/2 − 1 = n2 ((1/2)(1/n2 ) + O(1/n4 ))

= 1/2 + O(1/n2 ) → 1/2 as n → ∞.

n3 + 3n2 n3 + 3n2 − (n3 + n2 )

xn = (n3 + 3n + 3/n + 1/n3 ) − n3 = 3n + 3/n + 1/n3 > 3n.

by summing the geometric series. Hence xn ≤ 3/2 for all n.

Given that  k−2

6. Given that (1 + 1/n)n → e = 2.718 · · · as n → ∞ and for c > 0, c1/n → 1 as n → ∞,

The key here is to introduce the convergent sequence (yn ) where

and to express the given sequences in terms of (yn ).

7. Let (sn ) be the sequence given by

Thus ln 2 bounds the sequence. Also note that

Thus we have shown that sn → ln 2 as n → ∞.

4n2 − n + 3 1 4 − 1/n + 3/n2

cn → 4 implies that there exists

cn → 1/2 implies that there exists

ak+1 xk+1 /(k + 1)! x

By the ratio test the series converges (absolutely) for all x ∈ R.

(b) In the following α ∈ R is not an integer.

ak+1 α−k α/k − 1

(k 1/k )3 |x| |x|

i.e. with a2k = 1 and a2k+1 = 2 for k = 0, 1, 2, · · · .

Exercises on continuity and the contraction mapping

(i) f : [−1, 1] → R, f (x) := 2x − 2x3 . [3 marks]

(ii) f : [−2, 2] → R, f (x) := (x2 − 4)2 . [4 marks]

f 0 (x) = 2(x2 − 4)(2x) = 4x(x2 − 4),

f 00 (x) = 0 when x2 = 4/3 which gives two points in [−2, 2].

2. This was question 2 of the Jan 2003 MA2034A exam paper.

(a) Let I ⊂ R be an interval and let f : I → R be a function.

(c) If f : [a, b] → R is continuously differentiable on [a, b] then it can be shown that f

(ii) f : [−1, 1] → R, f (x) := (1 − x2 )2 .

(iii) f : [0, 4] → R, f (x) := (x + 1)e−x .

|f (x) − f (c)| <  whenever |x − c| < δ .

|f (x) − f (y)| ≤ L|x − y|

|f (x) − f (y)| ≤ L|x − y| <  provided |x − y| < /L .

For x > 0, y > 0 and x 6= y we have

L = max{|f 0 (5)|, |f 0 (13)|} = f 0 (13) = 52.

f 0 (x) = 2(1 − x2 )(−2x) = 4x(x2 − 1) = 4(x3 − x) and f 00 (x) = 4(3x2 − 1) .

f 0 (x) = e−x − (x + 1)e−x = −xe−x

f 0 has a local turning point at x = 1 ∈ [0, 4]. Thus

L = max{|f 0 (0)|, |f 0 (1)|, |f 0 (4)|} = max{0, e−1 , 4e−4 } = e−1 .

(a) Let I ⊂ R be an interval and let f : I → R be a function.

is Lipschitz on [−1, 1].

|f (x) − f (y)| = |x − y|.

In the case x, y ∈ [0, 2] we have

|f (x) − f (y)| = 2|x − y|.

|xn − x| < for all n ≥ N .

Given that k−2

|f (x) − f (c)| < whenever |x − c| < δ .

|f (x) − f (y)| ≤ L|x − y| < provided |x − y| < /L .

|xn − x| < /2 for all n ≥ N .