Skip to main content
Library homepage
 

Text Color

Text Size

 

Margin Size

 

Font Type

Enable Dyslexic Font
Mathematics LibreTexts

2.5: Limit Superior and Limit Inferior

( \newcommand{\kernel}{\mathrm{null}\,}\)

We begin this section with a proposition which follows from Theorem 2.3.1. All sequences in this section are assumed to be of real numbers.

Proposition 2.5.1

Let {an} be a bounded sequence. Define

sn=sup{ak:kn}

and

tn=inf{ak:kn}.

Then {sn} and {tn} are convergent.

Proof

If nm, then {ak:km}{ak:kn}. Therefore, it follows from Theorem 2.5.3 that snsm and, so, the sequence {sn} is decreasing. Since {an} is bounded, then so is {sn}. In particular, {sn} is bounded below. Similarly, {tn} is increasing and bounded above. Therefore, both sequences are convergent by Theorem 2.3.1.

Definition 2.5.1: Limit Superior

Let {an} be a sequence. Then the limit superior of {an}\), denoted by lim supnan, is defined by

lim supnan=limnsup{ak:kn}.

Note that lim supnan=limnsn, where sn is defined in (2.8).

Similarly, the limit inferior of {an}, denoted by lim infnan, is defined by

lim infnan=lim infn{ak:kn}.

Note that lim infnan=limntn, where tn is defined in (2.9).

Theorem 2.5.2

If {an} is not bounded above, then

limnsn=,

where {sn} is defined in (2.8).

Similarly, if {an} is not bounded below, then

limntn=,

where {tn} is defined in (2.9).

Proof

Suppose {an} is not bounded above. Then for any kN, the set {ai:ik} is also not bounded above. Thus, sk=sup{ai:ik}= for all k. Therefore, limksk=. The proof for the second case is similar.

Remark 2.5.3

By Theorem 2.5.2, we see that if {an} is not bounded above, then

lim supnan=.

Similarly, if {an} is not bounded below, then

lim infnan=.

Theorem 2.5.4

Let {an} be a sequence and R. The following are equivalent:

  1. lim supnan=.
  2. For any ε>0, there exists NR such that

an<+ε for all nN,

and there exists a subsequence of {ank} of {an} such that

limkank=.

Proof

Suppose lim supnan=. Then limnsn=, where Sn is defined as in (2.8). For any ε>0, there exists NN such that

ε<sn<+ε for all nN.

This implies sN=sup{an:nN}<+ε. Thus,

an<+ε for all nN

Moreover, for ε=1, there exists N1N such that

1<sN1=sup{an:nN1}<+1.

Thus, there exists n1N such that

1<an1<+1.

For ε=12, there exists N2N and N2>n1 such that

12<sN2=sup{an:nN2}<+12.

Thus, there exists n2>n1 such that

12<an2<+12.

In this way, we can construct a strictly increasing sequence {nk} of positive integers such that

1k<ank<+1k.

Therefore, limkank=.

We now prove the converse. Given any ε>0, there exists NN such that

an<+ε and ε<ank<+ε

for all nM and kN. Let any mN, we have

sm=sup{ak:km}+ε.

By Lemma 2.1.8, nmm, so we also have

sm=sup{ak:km}anm>ε.

Therefore, limmsm=lim supman=.

The following result is proved in a similar way.

Theorem 2.5.5

Let {an} be a sequence and R. The following are equivalent:

  1. lim infnan=.
  2. For any ε>0, there exists NN such that

an>ε for all nN,

and there exists a subsequence of {ank} of {an} such that

limkank=.

Proof

Add proof here and it will automatically be hidden

The following corollary follows directly from Theorems 2.5.4 and 2.5.5.

Corollary 2.5.6

Let {an} be a sequence. Then

limnan= if and only if lim supnan=lim infnan=.

Proof

Add proof here and it will automatically be hidden

Corollary 2.5.7

Let {an} be a sequence.

  1. Suppose lim supnan= and {ank} is a subsequence of {an} with

limkank=.

Then .

  1. Suppose lim infnan= and {ank} is a subsequence of {an} with

limkank=.

Then .

Proof

We prove only (a) because the proof of (b) is similar. By Theorem 2.5.4 and the definition of limits, for any ε>0, there exists NN such that

an<+ε and ε<ank<+ε

for all nN and kN. Since nNN, this implies

ε<anN<+ε.

Thus, <+2ε and, hence, because E is arbitrary.

Remark 2.5.8: Subsequential Limit

Let {an} be a bounded sequence. Define

A={xR: there exists a subsequence {ank} with limank=x}.

Each element of the set A called a subsequential limit of the sequence {an}. It follows from Theorem 2.5.4, Theorem 2.5.5, and Corollary 2.5.7 that A and

lim supnan=maxA and lim infnan=minA.

Theorem 2.5.9

Suppose {an} is a sequence such that an>0 for every nN and

lim supnan+1an=<1.

Then limnan=0.

Proof

Choose ε>0 such that +ε<1. Then there exists NN such that

an+1an<+ε for all nN.

Let q=+ε. Then 0<q<1. By induction,

0<anqnNaN for all nN.

Since limnqnNaN=0, one has limnan=0.

By a similar method, we obtain the theorem below.

Theorem 2.5.10

Suppose {an} is a sequence such that an>0 for every nN and

lim infnan+1an=>1.

Then limnan=.

Proof

Add proof here and it will automatically be hidden

Example 2.5.1

Given a real number α, define

an=αnn!,nN.

Solution

When α=0, it is obvious that limnan=0. Suppose α>0. Then

lim supnan+1an=limnαn+1=0<1.

Thus, limnan=0. In the general case, we can also show that limnan=0 by considering limn|an| and using Exercise 2.1.3.

All sequences in this set of exercises are assumed to be in R.

Exercise 2.5.1

Find lim supnan and lim infnan for each sequence.

  1. an=(1)n.
  2. an=sin(nπ2).
  3. an=1+(1)nn.
  4. an=nsin(nπ2).

Exercise 2.5.2

For a sequence {an}, prove that:

  1. lim infnan= if and only if limnan=.
  2. lim supnan= if and only if limnan=.

Exercise 2.5.3

Let {an} and {bn} be bounded sequences. Prove that:

  1. supkn(an+bn)supknak+supknbk.
  2. infkn(an+bn)infknak+infknbk.

Exercise 2.5.4

Let {an} and {bn} be bounded sequences.

  1. Prove that lim supn(an+bn)lim supnan+lim supnbn.
  2. Prove that lim infn(an+bn)lim infnan+lim infnbn.
  3. Find the two counter examples to show that the equalities may not hold in part (a) and part (b).

Exercise 2.5.5

Let {an} be a convergent sequence and let {bn} be an arbitrary sequence. Prove that

  1. lim supn(an+bn)=lim supnan+lim supnbn=limnan+lim supnbn.
  2. lim infn(an+bn)=lim infnan+lim infnbn=limnan+lim infnbn.

This page titled 2.5: Limit Superior and Limit Inferior is shared under a CC BY-NC-SA license and was authored, remixed, and/or curated by Lafferriere, Lafferriere, and Nguyen (PDXOpen: Open Educational Resources) .

Support Center

How can we help?

2.4: The Bolazno-Weierstrass Theorem
2.6: Open Sets, Closed Sets, Compact Sets, and Limit Points