Introduction to Analysis - BX27

Tutorial IV

Unbounded sequences

Exercise 1. Let (an )∞ ∞ ∞

n=0 and (bn )n=0 be two real-valued sequences. Assume that (an )n=0
∞
tends to +∞ as n tends to infinity and that (bn )n=0 is bounded from below. Prove that
the sequence (an + bn )∞
n=0 tends to +∞ as n tends to infinity.

We need to prove that, given R > 0, there exists N ∈ N such that, for all n ⩾ N ,
an + bn > R.

Assume R > 0 is given. Since (bn )∞ n=0 is bounded from below, there exists M ∈ R
such that, for all n ∈ N, we have bn > M . Since (an )∞ n=0 tends to +∞ as n tends
to infinity, there exists N ∈ N such that, for all n ⩾ N , we have an > R − M .

Then, for all n ⩾ N , we have an + bn > R − M + M = R. This completes the proof

of the fact that (an + bn )∞
n=0 tends to +∞ as n tends to infinity.

Exercise 2. Find counterexamples to the following statements which have been found
in works submitted by students:

(i) “Since the sequence converges it is monotone”.

n
The sequence ( (−1) ∞
n )n=1 converges to 0 but it is not monotone.

(ii) “Since the sequence is monotone, it converges”.

The sequence (n2 )∞
n=0 is monotone but it does not converge

(iii) “Since the sequence is increasing and bounded from above by 1, it converges to 1”.

The sequence ( 12 − n1 )∞
n=1 is increasing, bounded by 1, but does not converge to 1 (it
converges to 1/2).

(iv) “Since the sequence is not bounded, it tends to infinity”.

1
 The sequence (un )∞n=0 defined by u2n := n and u2n+1 := 0, for all n ∈ N, is not
bounded but it does not tend to +∞.

(v) “Since the sequence diverges, it tends to infinity”.

The sequence ((−1)n )∞
n=0 does not converge but it does not tend to +∞ nor does it
tand to −∞.

Adjacent sequences

Exercise 3. Define the sequences (un )∞ ∞

n=1 and (vn )n=1 by

1 1 1 1
un := + + ... + 2 and vn := un + ,
12 22 n n
for all n ⩾ 1. Prove that (un )∞ ∞
n=1 and (vn )n=1 are adjacent sequences. Conclude.

Since
1
un+1 = un + > un ,
(n + 1)2
the sequence (un )∞
n=1 is increasing. Moreover,

1
vn − un = ,
n
and hence the sequence (vn − un )∞n=1 converges to 0. It remains to prove that the
sequence (vn )∞
n=1 is decreasing. To this aim, let us compute
1 1
vn+1 − vn = un+1 + − un −
n+1 n
1 1 1
= 2
+ −
(n + 1) n+1 n
1
= −
n (n + 1)2
< 0,

which proves that (vn )∞

n=1 is decreasing. This completes the proof of the fact that the
sequences (un )∞ ∞
n=1 and (vn )n=1 are adjacent. A result from the Lecture then implies
∞ ∞
that (un )n=1 and (vn )n=1 both converge, to the same limit.

Exercise 4. Definition and irrationality of e.

(i) Recall that n! = 1 × 2 × . . . × n for n ∈ N \ {0}. Let (xn )∞ ∞
n=1 and (yn )n=1 be the
sequences defined by
1 1 1 1
xn := 1 + + + ··· + and yn := xn + .
1! 2! n! n!
Prove that the two sequences converge to the same limit, which will be denoted by e.

2
 We claim that the two sequences (xn )∞ ∞
n=2 and (yn )n=2 are adjacent.

First, for all n ⩾ 1, Observe that the

1 sequence is not
un+1 − un = > 0, only increasing
(n + 1)! but strictly
increasing and
and this shows that the sequence (xn )∞
n=1 is strictly increasing. this observation
will be useful in
the next question.
Next, for n ⩾ 1, we have
1 1
yn+1 − yn = xn+1 + − xn −
(n + 1)! n!
2 1
= −
(n + 1)! n!
1−n
=
(n + 1)!
and hence, we get yn+1 ⩽ yn for n ⩾ 1 and yn+1 < yn for n ⩾ 2. Thus, (yn )∞
n=1 is
decreasing and strictly decreasing for n ⩾ 2. Observe that
the sequence is
Finally, lim (yn − xn ) = 0 and this completes the proof of the fact that the two strictly increasing
n→+∞ for n ⩾ 2 and this
sequences are adjacent. A result from the Lecture implies that two sequences both observation
be useful in the
will

converge to the same limit. next question.

(ii) Prove that e is not a rational number. Hint: Argue by contradiction and use the
previous question.

We have seen in the previous question that the sequences (xn )∞ ∞

n=1 and (yn )n=1 are
monotone and even strictly monotone for n ⩾ 2. Moreover, they both converge to e,
hence we have the inequalities
x1 < x2 < . . . xn < xn+1 ⩽ e ⩽ yn+1 < yn < . . . y2 = y1 ,
for all n ⩾ 2. In particular, we get
xn < e < yn ,
for all n ⩾ 1.
We now argue by contradiction and for the sake of contradiction, we assume that
e = pq where p, q ∈ N \ {0}. We use the fact that xq < e < yq to obtain
1 1 1 p 1 1 1 1
1+ + + ··· + < < 1 + + + ··· + + .
1! 2! q! q 1! 2! q! q!
We multiply these inequalities by q! to conclude that
k < p (q − 1)! < k + 1,
where
q! q!
k = q! + + + . . . + q + 1.
1! 2!
Observe that k is a natural number. Therefore, the natural number p (q − 1)! is found
to be (strictly) larger than k and (strictly) smaller than k + 1. This is clearly a
contradiction.

3
Exercise 5. (Difficult) We define the sequence (un )∞
n=1 by
s
…
√
q »
un := 1 + 2 + 3 + 4 + . . . + n − 1 + n,

for all n ⩾ 1 and the sequence (vn )∞

n=1 by
s
…
√
q »
vn := 1+ 2+ 3+ 4 + ... n − 1 + 2n,
for all n ⩾ 1.
(i) Prove that
n
|vn − un | ⩽ √ ,
2n n!
for all n ⩾ 1.
We use the identity
√ √ b−a
b− a= √ √ .
b+ a
to prove by induction that
√
q
√
» p q » p
k + k + 1 + . . . + 2n − k + k + 1 + . . . + n
vn − u n ⩽ p ,
2k−1 × (k − 1)!
for k = 1, . . . , n − 1. For example,
√
q
√
» p q » p
1 + 2 + 3 + . . . + 2n − 1 − 2 + 3 + . . . + n
vn − un = … …
√
q
√
» p q » p
1 + 2 + 3 + . . . + 2n + 1 + 2 + 3 + . . . + n
√
q
√
» p q » p
2 + 3 + . . . + 2n − 2 + 3 + . . . + n
⩽ √ ,
2 1
and
» p √ q» p √
1 2 + 3 + . . . + 2n − 2 − 3 + ... + n
vn − un ⩽ √ q
2 1
» p √ q » p √
2 + 3 + . . . + 2n + 2 + 3 + . . . + n
√
q
√
» p q » p
3 + 4 + . . . + 2n − 3 + 4 + . . . + n
⩽ √ ,
22 × 2!
In particular (for k = n − 1),
p √ p √
n − 1 + 2n − n − 1 + n
vn − un ⩽ p
n−2
√ 2 √× (n − 2)!
2n − n
⩽ p
2n−1 × (n − 1)!
n
⩽ √ .
n
2 × n!

4
(ii) Prove that the sequences (un )∞ ∞
n=1 and (vn )n=1 converge to the same limit.

Clearly
un ⩽ vn ,
for all n ⩾ 1 and lim (vn − un ) = 0 thanks to the previous question. Moreover,
n→+∞
since
√ » √
n< n+ n + 1,
for all n ⩾ 1, we conclude that the sequence (un )∞
n=1 is (strictly) increasing. Finally,
observe that q » √
n + 2(n + 1) < 2n,
for all n ⩾ 3 and using this, we conclude that vn+1 < vn . Therefore, the sequence
(vn )∞
n=3 is strictly decreasing. Using the previous question, we conclude that the se-
quences (un )∞ ∞
n=3 and (vn )n=3 are adjacent and hence they both converge to the same
limit.

Subsequences

Exercise 6. Assume that (an )∞ n=0 is a real-valued sequence which is not bounded from
above. Prove that it has a subsequence which tends to +∞ as n tends to infinity.
Since (an )∞
n=0 is not bounded from above, given M > 0, there exists n > N such
that
an ⩾ M.

This proof of
this exercise is
Now, given R > 0 and N ∈ N, we set interesting since
it explains how,
using some induc-
M := max (R, a0 , . . . , aN ) , tion argument,
one can extract
and thanks to the above, we get the existence of n ∈ N such that a subsequence
satisfying some
specific property.
an ⩾ M.
In particular, an ⩾ R and n > N since an > a0 , an > a1 , . . . , an > aN . Therefore,
the following statement holds:
Given R > 0 and N ∈ N, there exists n > N such that aN ⩾ R.
Taking R = 0 and N = 0, we get the existence of n0 ∈ N such that an0 ⩾ 0. Taking
R = 1 and N = n0 , we get the existence of n1 > n0 such that an1 ⩾ 1.
We now argue by induction and assume that we have constructed n0 < n1 < . . . < nk
such that ani ⩾ i, for all i = 0, . . . , k. Taking R = k + 1 and N = nk , we get the
existence of nk+1 > nk such that ank+1 ⩾ k.

We define
φ(k) := nk .
By construction φ : N → N is strictly increasing and aφ(k) ⩾ k. Therefore,
(aφ(n) )∞ ∞
n=0 is a subsequence of the sequence (an )n=0 which tends to +∞ as n tends
to infinity.

5
Exercise 7. What can be said about a monotone real-valued sequence which has a con-
vergent subsequence?
Let us prove that a monotone sequence which has a convergent subsequence, con-
verges.

A monotone sequence is either increasing or decreasing. Let us assume that the

sequence under consideration is increasing (the case of a decreasing sequence can
be treated similarly). Let (an )∞
n=0 be an increasing sequence which has a convergent
subsequence (aφ(n) )∞n=0 where φ : N → N is a strictly increasing function.

Since the sequence (aφ(n) )∞

n=0 converges, it is bounded from above. Therefore, there
exists M > 0 such that
aφ(n) ⩽ M,
for all n ∈ N. Now φ is strictly increasing and we have seen in the Lecture that this
implies that φ(n) ⩾ n, for all n ∈ N. Since the sequence (an )∞n=0 is increasing, we
conclude that
an ⩽ aφ(n) ⩽ M.
for all n ∈ N. Therefore, (an )∞
n=0 is an incrasing sequence which is bounded from
above, and hence it converges.

Exercise 8. (Difficult) The aim of this exercise is to prove that, any real-valued sequence
has a monotone subsequence. Let (an )∞ n=0 be a real-valued sequence. We define

E := {n ∈ N : for all p > n, an > ap }.

(i) Assume that E is infinite. Prove that (an )∞
n=0 has a strictly decreasing subsequence.

By assumption E is infinite. We write

E = {nj : nj < nj+1 for all j ∈ N}.
Then, by construction and definition of E,
anj > anj+1 .
It is enough to defined φ(j) := nj . The function φ : N → N is strictly increasing and
the subsequence (aφ(j) )∞
j=0 is strictly decreasing.

(ii) Assume that E is finite. Prove that (an )∞

n=0 has an increasing subsequence.

We define N := max E. For all m > N , m ∈ / E and hence there exists n > m such
that an ⩾ an . Using this and an argument by induction, one can build ψ : N → N
strictly increasing, such that
aψ(n) ⩽ aψ(n+1) ,
for all n ∈ N. Indeed, define ψ(0) := N + 1 and, assuming that ψ(k) has already
been defined, we use the fact that ψ(k) ∈
/ E to get the existence of n > ψ(k) such that
un ⩾ uψ(k) . Then we define ψ(k + 1) := n.
The sequence (aψ(n) )∞
n=0 is increasing.

6
Exercise 9. Let (an )∞
n=0 be a real-valued sequence.

(i) Assume that the subsequences (a2n )∞ ∞

n=0 and (a2n+1 )n=0 converge to the same limit.
Prove that the sequence (an )∞
n=0 converges.

Denote by ℓ, the common limit of the subsequences (a2k )∞ ∞

k=0 and (a2k+1 )k=0 . Let
∞
ε > 0 be given. Since (a2k )k=0 converges to ℓ, there exists N ∈ N such that, for all
k ∈ N, if k ⩾ N , then |a2k − ℓ| < ε. Since (a2k+1 )∞ k=0 converges to ℓ, there exists
N ′ ∈ N such that, for all k ∈ N, if k ⩾ N ′ , then |a2k+1 − ℓ| < ε.
Therefore, for all n ∈ N, if n ⩾ max(2N, 2N ′ + 1), then |an − ℓ| < ε. And this
completes the proof of the fact that (an )∞
n=0 converges (to ℓ).

(ii) Assume that the subsequences (u2n )∞ ∞ ∞

n=0 , (u2n+1 )n=0 and (u3n )n=0 converge. Prove
that the sequence (un )∞
n=0 converges.

Denote by ℓ the limit of (u2n )∞ ′ ∞

n=0 , denote by ℓ the limit of (u2n+1 )n=0 and denote by
′′ ∞
ℓ the limit of (u3n )n=0 . The idea is to use the fact that a subsequence of a sequence
which converges to some limit, also converges to the same limit. Now, observe that
(u6n )∞ ∞ ∞
n=0 is both a subsequence of (u2n )n=0 and a subsequence of (u3n )n=0 , hence
′′ ∞
we conclude that ℓ = ℓ . Moreover, the sequence (u6n+3 )n=0 is both a subsequence of
(u2n+1 )∞ ∞ ′ ′′
n=0 and a subsequence of (u3n )n=0 , hence we conclude that ℓ = ℓ . Therefore,
′ ∞
ℓ = ℓ and, thanks to the first question, we conclude that (an )n=0 converges.