10

• prove the sandwich rule

• identify null sequences and solve problems involving null sequences

• prove the power rule for null sequences

• give the definition of a sequence that tends to infinity

• use the rules in connection with infinity( 3.3.3 H)

• prove and use the reciprocal rule (3.3.4H)

• give the definition of a subsequence of a sequence, and use subsequences to see if a sequence converges or
diverges

• give the definition of a monotone sequence and prove that a bounded monotone sequence is convergent

• do problems involving monotone sequences

• state and prove the Bolzano-Weierstrass Theorem

The following is a list of

PROOFS YOU NEED NOT BE ABLE TO REPRODUCE.

N.B. You still need to know the statements of all of these theorems and lemmas and must be able to apply them
where necessary.

p. 73 3.1.2:
proofs of the product rule, quotient rule and scalar product rule (but note the result in the first bit of the proof of the
product rule: every convergent sequence is bounded)

p. 77 3.1.4:
composite rule (proof on p. 152)
p. 80 3.2.2:
basic null sequences (a) to (e)
p. 96 3.4.4:
convergent sequences are Cauchy and conversely.

3.1 CONVERGENT SEQUENCES

a1 , a2 , a3 , ..., an , ...

and as such can be pictured by plotting the terms an on the real line:

Open Rubric
 11 MAT2613/1

a2 a1 a3 a4

Alternatively, (an ) can be regarded as a function from N to R which maps the natural number n onto the real
number an . As such, the sequence can be pictured by plotting the real number an against the natural number n in
the coordinate plane:

a4 (4, a4)

(3, a3)
a3
1 2 3 4

a1
(1 , a1)
a2
(2 , a2)

Haggarty first introduces the concept of a limit in Definition 3.3.1 H. We suggest that you work carefully through
the discussion on pp. 70-71 H following Definition 3.1.1 H.
Also note that " an → L as n → ∞" can be written as " lim an = L" and it is read as "an tends to L as n tends to
n→∞
infinity".
When we prove a result directly from a definition, we say it is a proof from first principles. Work carefully through
Examples 2 and 3 on pp. 71-72 H.
In Example 3 we see that Haggarty obtains an inequality of the form
k
|an − L| <
n
where (an ) is the given sequence, L is the expected limit and k is some positive real number. Then N is chosen to
k
be the smallest integer greater than where is the given " tolerance" . Then we have, for every n ≥ N that

k k k
≤ < = .
n N k/
Thus
|an − L| < for all n ≥ N.

The general method of proving from first principles that lim an = L for a given sequence (an ) is as follows.
n→∞
First try to obtain an inequality of the form
k
|an − L| ≤
np
for some constant k and some positive real number p. Let > 0 be given. Then choose N to be an integer greater
1/
than k p .
Then if n ≥ N we have
k k k
p
≤ p < = .
n N k/
 12

It then follows that

If we cannot find an inequality of the desired form which holds for all n ∈ N , it is sufficient to find such an inequality
which holds from a certain term onwards. If you have obtained that
k
|an − L| ≤ for all n ≥ n o
np
1/
k p
then choose N to be any integer which is greater than both and n o , i.e.

1/
N ≥ max{(k/ ) p
, n o }.

Then it follows for all n ≥ N that |an − L| < ε.

Example:
Prove from first principles that the sequence

4n 2 − 1
(an ) = converges.
5n 2 − 3n
When we work with such a problem we follow the following steps:

1. Always (if not given in the form lim an = L, L = ∞ ) find the limit by dividing by the largest power of n.
n→∞
In this case
4n 2 − 1 4 − n12 4
lim 2
= lim 3
= .
n→∞ 5n − 3n n→∞ 5 − 5
n

2. Given ε > 0 we now try to find an N ∈ N such that

4n 2 − 1 4
if n > N then 2
− <ε
5n − 3n 5

To do this we follow the following steps:

(a) Obtain a common denominator:

4n 2 − 1 4 5 4n 2 − 1 − 4 5n 2 − 3n
2
− =
5n − 3n 5 5 5n 2 − 3n
−5 + 12n
=
5 5n 2 − 3n

(b) Estimate the numerator (if possible in a way that is valid for all n ∈ N, otherwise we have a conditional
estimate as discussed above).
Now
|−5 + 12n| ≤ |12n| for all n ∈ N

We thus have
−5 + 12n 12n 12 1
≤ =
5 5n 2 − 3n 5n (5n − 3) 5 5n − 3
 13 MAT2613/1

(c) Estimate the denominator (if possible in a way that is valid for all n ∈ N, otherwise we have a conditional
estimate as discussed on p. 12S).
Since |5n − 3| ≥ 5n − 3n = 2n > 0 for all n ≥ 1 (always try to obtain one term i.e. of the form cn p
where c is a constant and p a power of n) we have that
1 1
≤ for all n ∈ N
5n − 3 2n
N.B. It is important to use the inequality signs correctly
(d) Then
4n 2 − 1 4 12 1 6
2
− ≤ =
5n − 3n 5 5 2n 5n
(e) Now
6 6
< ε provided n > .
5n 5ε
6
(f) By 2.3.1(H) there exists an integer N exceeding 5ε
. For such an N,
6
n>N ⇒ n>
5ε
6
⇒ <ε
5n
4n 2 − 1 4
⇒ 2
− <ε
5n − 3n 5
In other words
4n 2 − 1 4
lim 2
=
n→∞ 5n − 3n 5

When you are asked to use the rules for convergent sequences to evaluate a certain limit you may apply all the rules
for convergent sequences directly, see Example 4, pp. 75-76 H.
Note that the rules 3.1.2 H can be rewritten in the following way:
3.1.1 Rules
Suppose lim an and lim bn both exist.
n→∞ n→∞
Then the following rules apply:
lim (an + bn ) = lim an + lim bn
n→∞ n→∞ n→∞
lim an bn = lim an lim bn
n→∞ n→∞ n→∞

an lim an
n→∞
lim = provided bn = 0 for each n and lim bn = 0
n→∞ bn lim bn n→∞
n→∞
lim kan = k lim an for every real number k.
n→∞ n→∞
3.2 NULL SEQUENCES
Remember that all the rules which apply to convergent sequences also apply to null sequences since null sequences
are only those convergent sequences which converge to the limit zero.
The null sequences which are given in 3.2.2 H may be used to identify null sequences in problems where limits have
to be determined. In all such problems the first step is to identify the dominant term (that is the term with the fastest

Open Rubric