Chapter 1.

1: Lim inf and Lim sup

Recall from Basic Analysis II the notion of convergence of a sequence (xn )∞
n=1 of real
numbers:

Definition 1.1 For l ∈ R, the sequence (xn ) converges to l – denoted

lim xn = l
n→∞

(sometimes people also write “xn → l” or “xn → l as n → ∞”) – if and only if the following
holds: For every ε > 0, there exists an integer N = Nε such that |xn −l| < ε for all n ≥ N .

This is a definition you should already be intimately familiar with. You should be able
to recite it word for word if someone were to wake you up in the middle of the night and
demand it from you. More importantly, you should be able to explain why the words in
the definition have to be exactly the way they are (for example, why it would be wrong
to begin the definition with “For some ε > 0 ...”) and have a clear intuition for the vague
idea the definition of convergence formalises. These things will not be covered again in
this course, as they were the subject of Basic Analysis. If you realise now (or at any time
in the course) that you have not mastered this or other important concepts from Basic
Analysis, you need to revise the BA notes and get extra help from the tutor or me.

In this section, we introduce tools (the lim sup and lim inf) for dealing with sequences
that do not necessarily converge to any point. As motivation, consider the following se-
quence (xn ), which (check this if you do not immediately see it) does not converge to any
l ∈ R:

Example 1.1 For any n = 1, 2, 3, . . ., define

1
xn := (−1)n + .
n
(If you have not seen this notation before, we use “:=” to emphasise that the thing on
the left-hand side is being defined by what is written on the right-hand side.) Then the
sequence (xn ) does not converge to any real number.

To deal with sequences like this, we first recall a few more important notions from
Basic Analysis:

Definition 1.2 If A ⊂ R is a non-empty set, we say A is bounded above (or that A has
an upper bound ) if and only if there exists a real number M such that a ≤ M holds for
all a ∈ A, and any such M is called an upper bound of A. If A is bounded above, we
define the supremum of A, denoted sup A, as the least upper bound of A. This means
that a ≤ sup A holds for all a ∈ A; and that if M ∈ R is any upper bound of A, then
sup A ≤ M . We know from Basic Analysis that every subset A ⊂ R that is bounded
above has a (unique) supremum sup A ∈ R. (The same is not true, for example, of the set
of rational numbers Q.) We sometimes write sup A = ∞ if A ⊂ R is a set that is NOT
bounded above.

1
Similarly, a set A ⊂ R is bounded below if and only if there exists m ∈ R such that m ≤ a
holds for all a ∈ A; and if A is bounded below we define the infimum of A, denoted inf A,
as the greatest lower bound. Therefore, inf A ≤ a holds for all a ∈ A; and if m ∈ R is
any lower bound of A, then m ≤ inf A. We will also write inf A = −∞ if the set A is not
bounded below.

Now suppose that we are given a real sequence (xn ). We will first define the lim sup
of the sequence, as follows: for each n = 1, 2, 3, . . ., define
bn := sup{xn , xn+1 , xn+2 , . . .}.
Note that if the sequence (xn ) is not bounded above, then bn = ∞ for all n.

On the other hand, if (xn ) is bounded above, and M ∈ R is an upper bound of (xn )
(so xm ≤ M holds for all m = 1, 2, 3, . . .), then M is also an upper bound for the sequence
(bn ). Furthermore, from the definition it follows that the sequence (bn ) is monotonically
decreasing: bn ≥ bn+1 holds for all n = 1, 2, 3, . . .. From results proven in Basic Analysis,
we can therefore conclude that the sequence (bn ) either converges to a finite limit, or
limn→∞ bn = −∞. Now we can define:

Definition 1.3 If (xn ) is a real sequence, we define its limit superior, or lim sup, as:
(a) If (xn ) is bounded above, then
lim sup xn := lim bn ,
n→∞ n→∞

where bn := sup{xm | m ≥ n} as above.

(b) If (xn ) is not bounded above, then
lim sup xn := ∞.
n→∞

You can think of the lim sup as being the largest number that some subsequence of
(xn ) converges to. We will do TUT problems to justify this intuition. To get a better
understanding of this definition, let’s apply it to the sequence from Example 1.1 above:

Example 1.1 cont’d: For xn := (−1)n + n1 as above, it is clear that (xn ) is bounded
above. We calculate:
 
n 1 n+1 1 n+2 1
bn = sup (−1) + , (−1) + , (−1) + ,...
n n+1 n+2
(
1 + n1 if n is even
= 1
1 + n+1 if n is odd.
Hence, limn→∞ bn = 1, so we conclude that
lim sup xn = 1.
n→∞

The definition of the limit inferior, or lim inf, of a sequence (xn ) is similar. We first
define, for each n = 1, 2, 3, . . .,
an := inf{xn , xn+1 , . . .},

2
and note again that if (xn ) is not bounded below we have an = −∞ for all n; whereas if
(xn ) is bounded below, then (an ) is also bounded below, and the sequence (an ) is mono-
tone increasing: an ≤ an+1 for all n. This means that in the case that (xn ) is bounded
below, the sequence (an ) has a limit which is either finite or +∞.

Definition 1.4 If (xn ) is a real sequence, we define its limit inferior, or lim inf, as:
(a) If (xn ) is bounded below, then

lim inf xn := lim an ,

n→∞ n→∞

where an := inf{xm | m ≥ n} as above.

(b) If (xn ) is not bounded below, then

lim inf xn := −∞.

n→∞

Example 1.1 cont’d again: Let xn := (−1)n + n1 again. It is also clear that (xn ) is
bounded below. This time, we calculate:
 
n 1 n+1 1 n+2 1
an = inf (−1) + , (−1) + , (−1) + ,...
n n+1 n+2
= −1

(So the value of an does not depend on whether n is even or odd.) Hence,

lim inf xn = lim an = −1.

n→∞ n→∞