Lecture 23: Monotone sequences,

limsup and liminf of sequences

Winfried Just
Department of Mathematics, Ohio University

Companion to Advanced Calculus

Winfried Just, Ohio University MATH4/5301, Lecture 23: Monotone sequences

Monotone sequences
A sequence (an )∞ ∗
n=m of elements of R is:

increasing if an ≤ an+1 for all n ≥ m,

strictly increasing if an < an+1 for all n ≥ m,
decreasing if an ≥ an+1 for all n ≥ m,
strictly decreasing if an > an+1 for all n ≥ m,
monotone if it is increasing or decreasing,
strictly monotone if it is either strictly increasing or strictly
decreasing.

Example L23.1: Any constant sequence is simultaneously

increasing and decreasing, but neither strictly increasing nor
strictly decreasing.

Example L23.2: The sequence (an )∞ n ∞

n=1 := ((−1) )n=1 is
not monotone. However, notice that this sequence is bounded.
Winfried Just, Ohio University MATH4/5301, Lecture 23: Monotone sequences
Monotone sequences: Example 3
Example L23.3: Let an := n1 for all n ≥ 1. Then (an )∞n=1 is
strictly decreasing, convergent, and limn→∞ an = 0 = inf(an )∞
n=1 .

+ −
For every N ≥ 1, let aN := sup(an )∞ ∞
n=N and aN := inf(an )n=N .
+ 1 −
Then in this example: aN = aN = N and aN = 0 for all N ≥ 1.
Winfried Just, Ohio University MATH4/5301, Lecture 23: Monotone sequences
Monotone sequences: Example 4
Example L23.4: Let bn := arctan n for all n ∈ N.
Then (bn )∞
n=0 is strictly increasing, convergent, and
limn→∞ bn = π2 = sup(bn )∞ n=1 ≈ 1.5708.

+ −
For every N ∈ N, let bN := sup(bn )∞ ∞
n=N and bN := inf(bn )n=N .
+ −
Question L23.1: What are bN and bN in this example?
+ π −
bN = bN = 2 and bN = bN = arctan N for all N ∈ N.
Winfried Just, Ohio University MATH4/5301, Lecture 23: Monotone sequences
Examples 3 and 4 generalize
Not all monotone sequences converge. For example, the sequence
(n)∞
n=0 is strictly increasing, but divergent. And as Example L23.2
shows, non-monotone bounded sequences may also diverge.
But bounded monotone sequences always converge:

Proposition 6.3.8: (Monotone bounded sequences converge) Let

(an )∞
n=m be a bounded sequence of real numbers and let M > 0 be
a bound in R for this sequence.
(a) If (an )∞
n=m is increasing, then it is convergent, and in fact
limn→∞ an = sup(an )∞ n=m ≤ M.
(b) If (an )∞
n=m is decreasing, then it is convergent, and in fact
limn→∞ an = inf(an )∞ n=m ≥ −M.

The above version of this result gives more information than the
textbook version, which is restricted to our part (a).
We will formally prove this proposition in Module 23.
Winfried Just, Ohio University MATH4/5301, Lecture 23: Monotone sequences
Non-monotone sequences: Example 5
Example L23.5: Let cn := sin n for all n ∈ N. Then c0 = 0,
c1 ≈ 0.8415, c2 ≈ 0.9093, c3 ≈ 0.1411, c4 ≈ −0.7568.
This sequence is not monotone. It is bounded, since |cn | ≤ 1 for
every n ∈ N, but it is not convergent.

−
For every N ∈ N, let cN+ := sup(cn )∞ ∞
n=N and cN := inf(cn )n=N .
Question L23.2: What appear cN+ and cN− to be in this example?
Here cN+ = 1 and cN− = −1 for all N ∈ N.
Winfried Just, Ohio University MATH4/5301, Lecture 23: Monotone sequences
 − ∞
The sequences (aN+ )∞
N=m and (aN )N=m

Let (an )∞
n=m be any sequence or reals. For every N ≥ m define
+ −
aN := sup(an )∞ ∞
n=N and aN := inf(an )n=N .
+ −
In general, aN and aN are in R∗ .
But when (an )∞
n=m is bounded, they are always real numbers.

Question L23.3: What can we say about the monotonicity

+ ∞
properties of the sequence (aN )N=m ?
This sequence will be decreasing, but not always strictly decreasing.
For the sequence (an )∞ ∞
n=1 = (1/n)n=1 of Example L23.3 we have
+ 1 + ∞
aN = aN = n for all N ≥ 1 so that (aN )N=1 is strictly decreasing.
+
For the sequence (bn )∞ π
n=0 of Example L23.4 we have bN = 2 for
+ ∞
all N ≥ 0. Thus (bN )N=1 is decreasing, but not strictly decreasing.
− ∞
Similarly, the sequence (aN )N=m will always be increasing,
but it may or may not be strictly increasing.
Winfried Just, Ohio University MATH4/5301, Lecture 23: Monotone sequences
Limit superior
Definition 6.4.6: (Limit superior) Suppose that (an )∞
n=m is a
+ ∞
sequence. We define a new sequence (aN )N=m by the formula
+
aN := sup(an )∞
n=N .
+
More informally, aN is the supremum of all the elements in the
sequence from aN onwards.
We then define the limit superior of the sequence (an )∞
n=m ,
denoted lim supn→∞ an , by the formula
+ ∞
lim supn→∞ an := inf(aN )N=m .

Notice that if (an )∞

n=m is bounded, then

sup(an )∞
n=N ∈ R,
+ ∞
(aN )N=m is bounded and decreasing,
+
lim supn→∞ an = limN→∞ aN .

The last item follows from part (b) of Proposition 6.3.8.

Winfried Just, Ohio University MATH4/5301, Lecture 23: Monotone sequences
Limit inferior
Definition 6.4.6: (Limit inferior) Suppose that (an )∞
n=m is a
− ∞
sequence. We define a new sequence (aN )N=m by the formula
−
aN := inf(an )∞
n=N .
−
More informally, aN is the infimum of all the elements in the
sequence from aN onwards.
We then define the limit inferior of the sequence (an )∞
n=m , denoted
lim inf n→∞ an , by the formula
− ∞
lim inf n→∞ an := sup(aN )N=m .

Notice that if (an )∞

n=m is bounded, then

inf(an )∞
n=N ∈ R,
− ∞
(aN )N=m is bounded and increasing,
−
lim inf n→∞ an = limN→∞ aN .

The last item follows from part (a) of Proposition 6.3.8.

Winfried Just, Ohio University MATH4/5301, Lecture 23: Monotone sequences
Liminf and limsup for the sequences in our examples
Example L23.6: Let (an )∞ ∞
n=1 = (−n)n=1 .
− +
Then aN = −∞ and aN = −N for all N ≥ 1.
Thus lim inf n→∞ an = lim supn→∞ an = −∞.

For the sequence (an )∞ n ∞

n=1 := ((−1) )n=1 of Example L23.2
− +
we have aN = −1 and aN = 1 for all N ≥ 1.
Thus lim inf n→∞ an = −1 and lim supn→∞ an = 1.

Similarly, for the sequence (cn )∞ ∞

n=0 := (sin n)n=0 of Example L23.5
we have lim inf n→∞ cn = −1 and lim supn→∞ cn = 1.

Question L23.4: Find lim inf n→∞ an and lim supn→∞ an

1 ∞
for the sequence (an )∞


n=1 := n n=1 of Example L23.3.
lim inf n→∞ an = lim supn→∞ an = limn→∞ an = 0.
∞
Also, for the sequence (bn )∞
n=0 := (arctan n)n=0 of Example L23.4
we get lim inf n→∞ bn = lim supn→∞ bn = limn→∞ bn = π2 .
Winfried Just, Ohio University MATH4/5301, Lecture 23: Monotone sequences
Take-home message

We defined several types of monotone sequences: increasing,

decreasing, strictly increasing, and strictly increasing ones.
Bounded monotone sequences are always convergent.
With every sequence (an )∞
n=m of real numbers we associate two
+ ∞ − ∞
sequences (aN )N=m and (aN )N=m of extended real numbers
defined by
+ −
aN := sup(an )∞
n=N and aN := inf(an )∞
n=N .
− ∞
The sequence (aN )N=m is always increasing and the sequence
+ ∞
(aN )N=m is always decreasing.
Two extended real numbers, called the limit inferior and limit
superior, respectively, of (an )∞
n=m were defined as follows:
− ∞
lim inf n→∞ an := sup(aN )N=m ,
+ ∞
lim supn→∞ an := inf(aN )N=m .

Winfried Just, Ohio University MATH4/5301, Lecture 23: Monotone sequences

Take-home message, continued

For unbounded sequences (an )∞

n=m , it is possible that
lim inf n→∞ an = −∞ or lim supn→∞ an = −∞
or lim inf n→∞ an = ∞ or lim supn→∞ an = ∞.

When the sequence (an )∞

n=m is bounded, then:
− +
aN and aN are real numbers for all N ≥ m,
lim inf n→∞ an is a real number,
lim supn→∞ an is a real number,
−
lim inf n→∞ an = limN→∞ aN ,
+
lim supn→∞ an = limN→∞ aN .

Winfried Just, Ohio University MATH4/5301, Lecture 23: Monotone sequences