1/73

6 Sequences

7 Sequences II

8 Sequences III

9 Sequences IV

10 Sequences V

11 Sequences VI

12 Sequences VII

Instructor: David Earn Mathematics 3A03 Real Analysis I

 Sequences 2/73

Mathematics
and Statistics
Z Z
dω = ω
M ∂M

Mathematics 3A03
Real Analysis I

Instructor: David Earn

Lecture 6
Sequences
Friday 18 January 2019

Instructor: David Earn Mathematics 3A03 Real Analysis I

 Sequences Announcements 3/73

Announcements

Solutions to Assignment 1 will be posted later today.

Study them!
Assignment 2: check the course web page over the weekend.
Remember that solutions to assignments and tests from the 2016
and 2017 versions of the course are available on the course web site.
Take advantage of these problems and solutions. They provide
many useful examples that should help you prepare for tests and the
final exam. (However, note that while most of the content of the
course is the same this year, there are some differences.)
No late submission of assignments. No exceptions.
However, best 5 of 6 assignments will be counted.
Always due 5 minutes before class on the due date.
Note as stated on course info sheet: Only a selection of
problems on each assignment will be marked; your grade on each
assignment will be based only on the problems selected for marking.
Problems to be marked will be selected after the due date.
Instructor: David Earn Mathematics 3A03 Real Analysis I
 Sequences Announcements 4/73

Announcements for week of 21–25 January 2019

Office hour on Monday 21 Jan 2019 will be at 3:30pm

(rather than the usual 1:30pm).

Wednesday’s lecture will be given by Niky Hristov.

Instructor: David Earn Mathematics 3A03 Real Analysis I

 Sequences Definition 5/73

Sequences

A sequence is a list that goes on forever.

There is a beginning (a “first term”) but no end, e.g.,

1 1 1 1 1
, , , , ..., , ...
1 2 3 4 n

We use the natural numbers N to label the terms of a

sequence:
a1 , a2 , a3 , . . . , an , . . .

Instructor: David Earn Mathematics 3A03 Real Analysis I

 Sequences Definition 6/73

Formal definition of a sequence

Definition (Sequence of Real Numbers)

A sequence of real numbers is a function

f : N → R.

A lot of different notation is common for sequences:

f (1), f (2), f (3), . . . {f (n)}∞

n=1
f1 , f2 , f3 , . . . {f (n)}
{f (n) : n = 1, 2, 3, . . .} {fn }∞
n=1
{f (n) : n ∈ N} {fn }

Instructor: David Earn Mathematics 3A03 Real Analysis I

 Sequences Definition 7/73

Specifying sequences

There are two main ways to specify a sequence:

1. Direct formula.
Specify f (n) for each n ∈ N.

Example (arithmetic progression with common difference d)

Sequence is:
c, c + d, c + 2d, c + 3d, . . .

∴ f (n) = c + (n − 1)d, n∈N

i.e., xn = c + (n − 1)d, n = 1, 2, 3, . . .

Instructor: David Earn Mathematics 3A03 Real Analysis I

 Sequences Specifying sequences 8/73

Specifying sequences
2. Recursive formula.
Specify first term and function f (x) to iterate.

i.e., Given x1 and f (x), we have xn = f (xn−1 ) for all n > 1.

x2 = f (x1 ), x3 = f (f (x1 )), x4 = f (f (f (x1 ))), ...

Example (arithmetic progression with common difference d)

x1 = c, f (x) = x + d

∴ xn = xn−1 + d, n = 2, 3, 4, . . .

Note: f is the most typical function name for both the direct and
recursive specifications. The correct interpretation of f should be
clear from context.
Instructor: David Earn Mathematics 3A03 Real Analysis I
 Sequences Specifying sequences 9/73

Specifying sequences
Example (geometric progression with common ratio r )
Sequence is: c, cr , cr 2 , cr 3 , . . .
Direct formula: xn = f (n) = cr n−1 , n = 1, 2, 3, . . .
Recursive formula: x1 = c, f (x) = rx, xn = f (xn−1 )
Number line representation of {xn } with c = 1 and r = 43 :

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

Graph of f (n):
1.0
0.8
0.6
xn 0.4
0.2
0.0 n
5
Instructor: David Earn 10 Mathematics 3A03 15Real Analysis I 20
 Sequences Specifying sequences 10/73

Specifying sequences
1
Example (f (n) = 1 + n2 )
5 10 17
Sequence is: 2, 4 , 9 16 , ...
1
Direct formula: xn = f (n) = 1 + n2 , n = 1, 2, 3, . . .
Recursive formula: x1 = 2, f (x) = 1 + [1 + (x − 1)−1/2 ]−2
Get this formula by solving for n in terms of x in
x = 1 + 1/(n − 1)2 (= xn−1 ).
Such an inversion will NOT always be possible.
Number line representation of {xn }:

1 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 1.9 2

Graph of f (n):
2.0
1.8
1.6
xn 1.4
1.2
1.0 n
5 10 15 20
Instructor: David Earn Mathematics 3A03 Real Analysis I
 Sequences Convergence 11/73

Convergence of sequences

We know from previous experience that:

cr n−1 → 0 as n → ∞ (if |r | < 1).
1
1+ n2
→ 1 as n → ∞.
How do we make our intuitive notion of convergence
mathematically rigorous?

Informal definition: “xn → L as n → ∞” means “we can make the

difference between xn and L as small as we like by choosing n big
enough”.

More careful informal definition: “xn → L as n → ∞” means

“given any error tolerance, say ε, we can make the distance
between xn and L smaller than ε by choosing n big enough”.

Instructor: David Earn Mathematics 3A03 Real Analysis I

 Sequences Convergence 12/73

Convergence of sequences
Definition (Limit of a sequence)
A sequence {sn } converges to L if, given any ε > 0 there is some
integer N such that

if n ≥ N then |sn − L| < ε .

In this case, we write lim sn = L or sn → L as n → ∞

n→∞
and we say that L is the limit of the sequence {sn }.

Note: To use this definition to prove that the limit of a sequence is

L, we start by imagining that we are given some error tolerance
ε > 0. Then we have to find a suitable N, which will depend on ε.
This means that the N that we find will be a function of ε.
Shorthand:
def
lim sn = L = ∀ε > 0 ∃N ∈ N –) n ≥ N =⇒ |sn − L| < ε.
n→∞
Instructor: David Earn Mathematics 3A03 Real Analysis I
 Sequences Convergence 13/73

Convergence of sequences

Convergence terminology:
A sequence that converges is said to be convergent.

A sequence that is not convergent is said to be divergent.

Remark (Sequences in spaces other than R)

The formal definition of a limit of a sequence works in any space
where we have a notion of distance if we replace |sn − L| with
d(sn , L).

Instructor: David Earn Mathematics 3A03 Real Analysis I

 Sequences Convergence 14/73

Convergence of sequences

Example
Use the formal definition of a limit of a sequence to prove that

n2 + 1
→1 as n → ∞ .
n2

(solution on board)

Note: Our strategy here was to solve for n in the inequality

|sn − L| < ε. From this we were able to infer how big N has to be
in order to ensure that |sn − L| < ε for all n ≥ N. That much was
“rough work”. Only after this rough work did we have enough
information to be able to write down a rigorous proof.

Instructor: David Earn Mathematics 3A03 Real Analysis I

 Sequences II 15/73

Mathematics
and Statistics
Z Z
dω = ω
M ∂M

Mathematics 3A03
Real Analysis I

Instructor: David Earn

Lecture 7
Sequences II
Monday 21 January 2019

Instructor: David Earn Mathematics 3A03 Real Analysis I

 Sequences II Announcements 16/73

Announcements

Solutions to Assignment 1 have been posted. Study them!

Assignment 2 has been posted. Due Friday 1 Feb 2019 at

1:25pm.

Instructor: David Earn Mathematics 3A03 Real Analysis I

 Sequences II Convergence 17/73

Convergence of sequences

Example
Use the formal definition of a limit of a sequence to prove that

n5 − n3 + 1
→0 as n → ∞ .
n8 − n5 + n + 1

(solution on board)

Note: In this example, it was not possible to solve for n in the

inequality |sn − L| < ε. Instead, we first needed to bound |sn − L|
by a much simpler expression that is always greater than |sn − L|.
If that bound is less than ε then so is |sn − L|.

Instructor: David Earn Mathematics 3A03 Real Analysis I

 Sequences II Convergence 18/73

Uniqueness of limits

Theorem (Uniqueness of limits)

If lim sn = L1 and lim sn = L2 then L1 = L2 .
n→∞ n→∞

(solution on board)

So, we are justified in referring to “the” limit of a convergent

sequence.

Instructor: David Earn Mathematics 3A03 Real Analysis I

 Sequences II Divergence 19/73

Divergence of sequences

Divergence is the logical opposite (negation) of convergence.

We can infer the formal meaning of divergence by taking the
logical negation of the formal definition of convergence.
Doing so, we find that the sequence {sn } diverges (i.e., does not
converge to any L ∈ R) iff

∀L ∈ R, ∃ε > 0 such that: ∀N ∈ N ∃n ≥ N –) |sn − L| ≥ ε.

Notes:
The n that exists will, in general, depend on L, ε and N.
This is the meaning of not converging to any limit, but it does
not tell us anything about what happens to the sequence {sn }
as n → ∞.

Instructor: David Earn Mathematics 3A03 Real Analysis I

 Sequences II Divergence 20/73

Divergence to ±∞
Definition (Divergence to ∞)
The sequence {sn } of real numbers diverges to ∞ if, for every
real number M there is an integer N such that

n≥N =⇒ sn ≥ M ,

in which case we write sn → ∞ as n → ∞ or lim sn = ∞.

n→∞

Definition (Divergence to −∞)

The sequence {sn } of real numbers diverges to −∞ if, for every
real number M there is an integer N such that

n≥N =⇒ sn ≤ M .

Instructor: David Earn Mathematics 3A03 Real Analysis I

 Sequences II Divergence 21/73

Divergence to ∞

Example
Use the formal definition to prove that
 3 
n −1
diverges to ∞ .
n+1

(solution on board)

Approach: Find a lower bound for the sequence that is a simple

function of n and show that that can be made bigger than any
given M.

Instructor: David Earn Mathematics 3A03 Real Analysis I

 Sequences III 22/73

Mathematics
and Statistics
Z Z
dω = ω
M ∂M

Mathematics 3A03
Real Analysis I

Instructor: David Earn

Lecture 8
Sequences III
Wednesday 23 January 2019

Instructor: David Earn Mathematics 3A03 Real Analysis I

 Sequences III 23/73

What we’ve done so far on sequences

Definition of convergence.

Definition of divergence.

Definition of divergence to ±∞.

Examples.

Instructor: David Earn Mathematics 3A03 Real Analysis I

 Sequences III 24/73

Divergence to ∞

Example (Example from last time)

n3 − 1
 
Use the formal definition to prove that diverges to ∞.
n+1

Clean proof.
Given M ∈ R>0 , let N = dMe + 1. Then N − 1 = dMe ≥ M.
∴ ∀n ≥ N, n − 1 ≥ M. Now observe that

(n − 1)(n + 1) n2 − 1 n3 − 1
∀n ∈ N, n−1= = ≤ .
n+1 n+1 n+1
∴ ∀n ≥ N we have
n3 − 1
≥M,
n+1
as required.
Instructor: David Earn Mathematics 3A03 Real Analysis I
 Sequences III Series 25/73

Sequences of partial sums (a.k.a. Series)

Given a sequence {xn }, we define the sequence of partial sums

of {xn } to be {sn }, where
n
X
sn = xk = x1 + x2 + · · · + xn .
k=1

Note: We can start from any integer, not necessarily k = 1.

Instructor: David Earn Mathematics 3A03 Real Analysis I

 Sequences III Boundedness 26/73

Boundedness of sequences

A sequence is said to be bounded if its range is a bounded set.

Definition (Bounded sequence)
A sequence {sn } is bounded if there is a real number M such that
every term in the sequence satisfies |sn | ≤ M.

Theorem (Every convergent sequence is bounded.)

L ∈ R ∧ lim sn = L =⇒ ∃M > 0 –) |sn | ≤ M ∀n ∈ N.
n→∞

(solution on board)

Note: The converse is FALSE.

Proof? Find a counterexample, e.g., {(−1)n }.

Instructor: David Earn Mathematics 3A03 Real Analysis I

 Sequences III Boundedness 27/73

Boundedness of sequences
Corollary (Unbounded sequences diverge)
If {sn } is unbounded then {sn } diverges.

Example (The harmonic series diverges)

1 1
Consider the harmonic series sn = 1 + 2 + 3 + · · · + n1 .

sn 3

1 n
0 20 40 60 80 100

Prove that sn diverges to ∞.

(solution on board)
Instructor: David Earn Mathematics 3A03 Real Analysis I
 Sequences III Harmonic series 28/73

Harmonic series – idea for proof of divergence

Approach: Group terms and use the corollary above.

 1 1 1 1 1 1 1
1+ + + + + + + + ···
| {z 2 } | 3 {z 4 } | 5 6 {z 7 8 }
> 1× 1 > 2× 14 > 4× 18
| {z 2 }
s2 > 1× 12
| {z }
s4 > 2× 12
| {z }
s8 > 3× 21

1
=⇒ s2n > n ×
2

Note: These sorts calculations are just “rough work”, not a formal proof.
A proof must be a clearly presented coherent argument from beginning to end.
Instructor: David Earn Mathematics 3A03 Real Analysis I
 Sequences III Harmonic series 29/73

Harmonic series – clean proof of divergence

Proof.
Part (i). Prove (e.g., by induction) that s2n > n/2 ∀n ∈ N.

Part (ii). Suppose we are given M ∈ R.

If M ≤ 0 then note that sn > 0 ∀n ∈ N.
If M > 0, let Ñ = 2 dMe and N = 2Ñ . Then, ∀n ≥ N, we
have sn ≥ sN = s2Ñ > Ñ/2 = dMe ≥ M, as required.

Instructor: David Earn Mathematics 3A03 Real Analysis I

 Sequences III Algebra of limits 30/73

Algebra of limits
Theorem (Algebraic operations on limits)
Suppose {sn } and {tn } are convergent sequences and C ∈ R.


1 lim C sn = C lim sn ;
n→∞ n→∞



2 lim (sn + tn ) = lim sn + lim tn ;
n→∞ n→∞ n→∞



3 lim (sn − tn ) = lim sn − lim tn ;
n→∞ n→∞ n→∞



4 lim (sn tn ) = lim sn lim tn ;
n→∞ n→∞ n→∞

5 if tn 6= 0 for all n and limn→∞ tn 6= 0 then

 s  lim
n n→∞ sn
lim = .
n→∞ tn limn→∞ tn

(solution on board)
Instructor: David Earn Mathematics 3A03 Real Analysis I
 Sequences III Algebra of limits 31/73

Revisit example

Example (previously proved directly from definition)

Use the algebraic properties of limits to prove that

n5 − n3 + 1
→0 as n → ∞ .
n8 − n5 + n + 1

(solution on board)

Instructor: David Earn Mathematics 3A03 Real Analysis I

 Sequences IV 32/73

Mathematics
and Statistics
Z Z
dω = ω
M ∂M

Mathematics 3A03
Real Analysis I

Instructor: David Earn

Lecture 9
Sequences IV
Friday 25 January 2019

Instructor: David Earn Mathematics 3A03 Real Analysis I

 Sequences IV 33/73

Announcements

Assignment 2 is posted.
Due next Friday, 1 Feb 2019, at 1:25pm.

Instructor: David Earn Mathematics 3A03 Real Analysis I

 Sequences IV 34/73

What we’ve done so far on sequences

Definition of convergence.

Definition of divergence.

Definition of divergence to ±∞.

Examples.

Every convergent sequence is bounded.

Harmonic series diverges.

Algebra of limits (more today).

Instructor: David Earn Mathematics 3A03 Real Analysis I

 Sequences IV 35/73

Product Rule for Limits

The 4th item in the algebra of limits theorem was:
Theorem (Product Rule for Limits)
If sn → S and tn → T as n → ∞ then sn tn → ST as n → ∞.

Proof.
For any n ∈ N, |sn tn − ST | = |sn tn − ST + sn T − sn T |
= |sn (tn − T ) + T (sn − S)|
≤ |sn ||tn − T | + |T ||sn − S|
Now, {sn } converges, so it is bounded by some M > 0, i.e.,
|sn | ≤ M ∀n ∈ N. Therefore, given ε > 0, choose N ∈ N such that
ε ε
|tn − T | < and |sn − S| < .
2M 2(1 + |T |)
Then |sn tn − ST | < ε/2 + ε/2 = ε, as required.
Instructor: David Earn Mathematics 3A03 Real Analysis I
 Sequences IV 36/73

Quotient Rule for Limits

Quotient Rule was the 5th item in the algebra of limits theorem.
Lemma (Reciprocal Rule for Limits)
If tn 6= 0 ∀n and tn → T 6= 0 then 1/tn → 1/T .

Proof.
1 1 tn − T 1 1
For any n ∈ N, − = = |tn − T | · · .
tn T tn T |tn | |T |
Since {tn } converges, ∃N1 ∈ N such that ∀n ≥ N1 , |tn | > |T |/2
(details on next slide) and hence 1/|tn | < 2/|T |.
Now choose N ≥ N1 such that |tn − T | < ε|T |2 /2. Then

1 1 1 1 ε|T |2 2 1
− = |tn − T | · · < · · = ε,
tn T |tn | |T | 2 |T | |T |

as required.
Instructor: David Earn Mathematics 3A03 Real Analysis I
 Sequences IV 37/73

Quotient Rule for Limits

|T |
Details missing on previous slide: (consider ε = 2 )
|T |
Since tn → T , ∃N1 ∈ N such that ∀n ≥ N1 , |tn − T | < ,
2
|T | |T | |T | |T |
i.e., − < tn −T < , i.e., T − < tn < T + .
2 2 2 2
If T > 0 this says
T 3T
0< < tn < ,
2 2
whereas if T < 0 it says
3T T
< tn < < 0.
2 2
|T |
In either case, ∀n ≥ N1 , we have 0< < |tn | .
2
Instructor: David Earn Mathematics 3A03 Real Analysis I
 Sequences IV Order properties 38/73

Order properties of limits (§2.8)

Theorem (Limits retain order)

If {sn } and {tn } are convergent sequences then

sn ≤ t n ∀n ∈ N =⇒ lim sn ≤ lim tn .
n→∞ n→∞

(solution on board)

Note: If sn < tn for all n ∈ N, can we conclude that

lim sn < lim tn

n→∞ n→∞
?

No! No! No! No! No! No!! NO!!!!!!!!!!

Instructor: David Earn Mathematics 3A03 Real Analysis I
 Sequences IV Order properties 39/73

Order properties of limits (§2.8)

Theorem (Limits retain bounds)

If {sn } is a convergent sequence then

α ≤ sn ≤ β ∀n ∈ N =⇒ α ≤ lim sn ≤ β .
n→∞

(solution on board)

Instructor: David Earn Mathematics 3A03 Real Analysis I

 Sequences IV Order properties 40/73

Order properties of limits (§2.8)

Theorem (Squeeze Theorem)
If {sn } and {tn } are convergent sequences such that
(i) sn ≤ xn ≤ tn ∀n ∈ N, (xn is always between them)

(ii) lim sn = lim tn = L. (both approach the same limit)

n→∞ n→∞

Then {xn } is convergent and lim xn = L.

n→∞

Proof? (What’s WRONG?).

{sn } and {tn } are both bounded since they both converge. {xn } is
therefore bounded (by the lower bound of {sn } and the upper
bound of {tn }). {xn } therefore converges, say xn → X . Hence, by
order retension, L ≤ X ≤ L =⇒ X = L.
(solution on board) •
Instructor: David Earn Mathematics 3A03 Real Analysis I
 Sequences V 41/73

Mathematics
and Statistics
Z Z
dω = ω
M ∂M

Mathematics 3A03
Real Analysis I

Instructor: David Earn

Lecture 10
Sequences V
Monday 28 January 2019

Instructor: David Earn Mathematics 3A03 Real Analysis I

 Sequences V 42/73

Announcements

Assignment 2.
Due next Friday, 1 Feb 2019, at 1:25pm.

Office hour today at 4:30pm (not the usual 1:30pm)

Niky Hristov’s notes for his lecture (Lecture 8, last

Wednesday) are now posted on the Lectures page course web
site.

Instructor: David Earn Mathematics 3A03 Real Analysis I

 Sequences V 43/73

What we’ve done so far on sequences

Definition of convergence.
Definition of divergence.
Definition of divergence to ±∞.
Every convergent sequence is bounded.
Harmonic series diverges.
Algebra of limits (sums, products, quotients).
Order properties of limits; squeeze theorem
Today:
Proof of Squeeze Theorem
Absolute value and max/min of limits.
Monotone convergence.

Instructor: David Earn Mathematics 3A03 Real Analysis I

 Sequences V 44/73

Order properties of limits (§2.8)

Theorem (Limits of Absolute Values)
If {sn } converges then so does {|sn |}, and

lim |sn | = lim sn .

n→∞ n→∞

(solution on board)

Proof.
Suppose sn → L. Given ε > 0, choose N ∈ N such that
∀n ≥ N,|sn − L| < ε. But for any x, y ∈ R, we know? from the
triangle inequality that ||x| − |y || ≤ |x − y |. Therefore, ∀n ≥ N,
||sn | − |L|| ≤ |sn − L| < ε. Thus, |sn | → |L|, as required.
?
To see this, observe that |x| = |x − y + y | ≤ |x − y | + |y |, which implies |x| − |y | ≤ |x − y | .
Similarly, |y | = |y − x + x| ≤ |y − x| + |x|, which implies |y | − |x| ≤ |y − x|, which can in turn be
rewritten −(|x| − |y |) ≤ |x − y |. Combining these inequalities, we have ||x| − |y || ≤ |x − y |.

Instructor: David Earn Mathematics 3A03 Real Analysis I

 Sequences V 45/73

Order properties of limits (§2.8)

Corollary (Max/Min of Limits)
If {sn } and {tn } converge then {max{sn , tn }} and {min{sn , tn }}
both converge and
n o
lim max{sn , tn } = max lim sn , lim tn ,
n→∞ n→∞ n→∞
n o
lim min{sn , tn } = min lim sn , lim tn .
n→∞ n→∞ n→∞

Idea for proof:

x +y |x − y |
∀x, y ∈ R max{x, y } = +
2 2
x +y |x − y |
∀x, y ∈ R min{x, y } = −
2 2
Prove these facts, then use theorems on sums and absolute values
of limits.
Instructor: David Earn Mathematics 3A03 Real Analysis I
 Sequences V Monotone convergence 46/73

Monotone convergence (§2.9)

Definition (Monotonic sequence)

The sequence {sn } is monotonic iff it satisfies any of the following
conditions:
(i) Increasing: s1 < s2 < s3 < · · · < sn < sn+1 < · · · ;

(ii) Decreasing: s1 > s2 > s3 > · · · > sn > sn+1 > · · · ;

(iii) Non-decreasing: s1 ≤ s2 ≤ s3 ≤ · · · ≤ sn ≤ sn+1 ≤ · · · ;

(iv) Non-increasing: s1 ≥ s2 ≥ s3 ≥ · · · ≥ sn ≥ sn+1 ≥ · · · .

Instructor: David Earn Mathematics 3A03 Real Analysis I

 Sequences V Monotone convergence 47/73

Monotone convergence (§2.9)

Theorem (Monotone Convergence Theorem)

A monotonic sequence {sn } is convergent iff it is bounded.
In particular,
(i) {sn } non-decreasing and unbounded =⇒ sn → ∞ ;
(ii) {sn } non-decreasing and bounded =⇒ sn → sup{sn } ;
(iii) {sn } non-increasing and unbounded =⇒ sn → −∞ ;
(iv) {sn } non-increasing and bounded =⇒ sn → inf{sn } .

(solution on board)

Instructor: David Earn Mathematics 3A03 Real Analysis I

 Sequences V Subsequences 48/73

Subsequences
Definition (Subsequence)
Let {s1 , s2 , s3 , . . .} be a sequence. If {n1 , n2 , n3 , . . .} is an
increasing sequence of natural numbers then {sn1 , sn2 , sn3 , . . .} is a
subsequence of {s1 , s2 , s3 , . . .}.

Example (Subsequences)
Consider the sequence {sn } defined by sn = n2 for all n ∈ N.
What are the first few terms of these subsequences?
{sn : n even} {22 , 42 , 62 , . . .}
{sn : n = 2k + 1, ∃k ∈ N} {32 , 52 , 72 , . . .}
{s2n+1 } Same as line above
{s2n } {22 , 42 , 82 , . . .}
{sn2 } {12 , 42 , 92 , . . .}

Instructor: David Earn Mathematics 3A03 Real Analysis I

 Sequences V Subsequences 49/73

Subsequences

Given any sequence {sn }, can you always find a subsequence that
is monotonic?

Theorem
Every sequence contains a monotonic subsequence.

Let’s draw some pictures to help us visualize how we might

construct a proof. . .

Instructor: David Earn Mathematics 3A03 Real Analysis I

 Sequences V Monotonic subsequences 50/73

Idea for proof that every sequence contains a monotonic

subsequence (“point of no return”)
Given a sequence {s1 , s2 , s3 , . . .}, try to build a subsequence
{sn1 , sn2 , sn3 , . . .} that is non-decreasing (sn1 ≤ sn2 ≤ sn3 ≤ · · · ) by
discarding any terms that are less than the running maximum:
12
10
8
sn 6
4
2
n1 n2 n3 n4 n5 n6 n7 n8
0 n
5 10 15 20

If this works indefinitely then we have a non-decreasing

subsequence. But if we can find only finitely many such terms then
we’re stuck because our subsequence is defined using earlier terms.
Instructor: David Earn Mathematics 3A03 Real Analysis I
 Sequences VI 51/73

Mathematics
and Statistics
Z Z
dω = ω
M ∂M

Mathematics 3A03
Real Analysis I

Instructor: David Earn

Lecture 11
Sequences VI
Wednesday 30 January 2019

Instructor: David Earn Mathematics 3A03 Real Analysis I

 Sequences VI 52/73

Announcements

Niky’s Monday office hour has been been switched to

Tuesdays at 2:30pm. Note that Math 3D03 students have
priority during that hour.

Please send me an e-mail ASAP if you have a conflict with

either of the midterm tests.

Proof of Limits of Absolute Values theorem is now in the

slides for Lecture 10.

Plan for today:

Prove Monotone Convergence Theorem
Prove that Every sequence contains a monotonic subsequence.
Maybe state and prove Bolzano-Weierstrass Theorem.

Instructor: David Earn Mathematics 3A03 Real Analysis I

 Sequences VI Monotonic convergence 53/73

Monotone convergence (§2.9)

Theorem (Monotone Convergence Theorem)

A monotonic sequence {sn } is convergent iff it is bounded.
In particular,
(i) {sn } non-decreasing and unbounded =⇒ sn → ∞ ;
(ii) {sn } non-decreasing and bounded =⇒ sn → sup{sn } ;
(iii) {sn } non-increasing and unbounded =⇒ sn → −∞ ;
(iv) {sn } non-increasing and bounded =⇒ sn → inf{sn } .

(solution on board)

Instructor: David Earn Mathematics 3A03 Real Analysis I

 Sequences VI Monotonic convergence 54/73

Proof of Monotone Convergence Theorem

Given a monotonic sequence {sn } we must show that

{sn } converges ⇐⇒ {sn } is bounded

Proof of “=⇒” and part (ii).

=⇒ For any sequence (monotonic or not) convergent implies bounded.
⇐= [part (ii)] Suppose {sn } is non-decreasing, i.e., sn ≤ sn+1 for all
n ∈ N. Since {sn } is bounded, it has a least upper bound, say
L = sup{sn }. We will now show that sn → L, i.e., ∀ε > 0 ∃N ∈ N –)
∀n ≥ N, |sn − L| < ε.

Before proceeding, note that since L = sup{sn }, it follows that

|sn − L| < ε ⇐⇒ L − sn < ε ⇐⇒ L − ε < sn .

Given ε > 0, choose N ∈ N such that sN > L − ε (which is possible ∵ L

is the least upper bound of {sn }). But {sn } is non-decreasing, so ∀n ≥ N
we have sN ≤ sn =⇒ −sn ≤ −sN =⇒ L − sN ≤ L − sn < ε.
Instructor: David Earn Mathematics 3A03 Real Analysis I
 Sequences VI Monotonic convergence 55/73

Proof of Monotone Convergence Theorem

h i
Monotonic =⇒ {sn } converges ⇐⇒ {sn } is bounded

Proof of parts (i), (iii), (iv).

[part (i)] Suppose {sn } is non-decreasing and unbounded. It follows that
{sn } diverges, since convergent sequences are bounded. Since {sn } is
non-decreasing, it is bounded below (by s1 , for example). Hence {sn }
(which is unbounded) must not be bounded above. Consequently, given
any M ∈ R, ∃N ∈ N such that sN > M. But {sn } is non-decreasing, so
sn > M for all n ≥ N, as required.

Proof of [part (iii)] is similar to [part (i)].

Proof of [part (iv)] is similar to [part (ii)].

Instructor: David Earn Mathematics 3A03 Real Analysis I

 Sequences VI Monotonic subsequences 56/73

Monotonic subsequences

Given any sequence {sn }, can you always find a subsequence that
is monotonic?

Theorem
Every sequence contains a monotonic subsequence.

(Textbook (TBB) §2.11, Theorem 2.39, p. 79)

There are no pictures accompanying the proof in the textbook.

So let’s draw some pictures to help us visualize how we might
construct a proof. . .

Instructor: David Earn Mathematics 3A03 Real Analysis I

 Sequences VI Monotonic subsequences 57/73

Idea for proof that every sequence contains a monotonic

subsequence (“point of no return”)
Given a sequence {s1 , s2 , s3 , . . .}, try to build a subsequence
{sn1 , sn2 , sn3 , . . .} that is non-decreasing (sn1 ≤ sn2 ≤ sn3 ≤ · · · ) by
discarding any terms that are less than the running maximum:
12
10
8
sn 6
4
2
n1 n2 n3 n4 n5 n6 n7 n8
0 n
5 10 15 20

If this works indefinitely then we have a non-decreasing

subsequence. But if we can find only finitely many such terms then
we’re stuck because our subsequence is defined using earlier terms.
Instructor: David Earn Mathematics 3A03 Real Analysis I
 Sequences VI Monotonic subsequences 58/73

Better idea for proof that every sequence contains a

monotonic subsequence (“turn-back point”)
Given a sequence {s1 , s2 , s3 , . . .}, try to build a subsequence
{sn1 , sn2 , sn3 , . . .} that is non-decreasing (sn1 ≤ sn2 ≤ sn3 ≤ · · · ) by
identifying terms that are less than or equal to all later terms.
12
10
8
sn 6
4
2
n1 n2 n3 n4 n5 n6 n7 n8
0 n
5 10 15 20

If this works indefinitely then we have a non-decreasing

subsequence. What if there are only finitely many such terms?
(There might not be any at all!)
Instructor: David Earn Mathematics 3A03 Real Analysis I
 Sequences VI Monotonic subsequences 59/73

Better idea for proof that every sequence contains a

monotonic subsequence (“turn-back point”)

If there are only finitely many sni such that sni ≤ sn ∀n > ni . . .
12
10
8
sn 6
4
2
n1 n2 n3 n4 m1 m2 m3 m4 m5
0 n
5 10 15 20

. . . then after the last “turn-back point” (sn4 above) there must be
some m1 > n4 such that sm1 is not ≤ all later terms, i.e.,
∃m2 > m1 with sm2 < sm1 , and similarly for m2 , so there must be a
decreasing subsequence sm1 > sm2 > sm3 > · · ·

Instructor: David Earn Mathematics 3A03 Real Analysis I

 Sequences VI Convergent subsequences 60/73

Convergent subsequences

Theorem (Bolzano-Weierstrass theorem)

Every bounded sequence of real numbers contains a convergent
subsequence.

Proof.
Suppose {xn } is a bounded sequence. It follows from the previous
theorem that {xn } contains a subsequence {xmk } that is
monotonic. Since {xn } is bounded, the subsequence {xmk } is
bounded as well (by the same bound). Thus, {xmk } is a
subsequence of {xn } that is both bounded and monotonic. Hence,
it converges by the Monotone Convergence Theorem.

Instructor: David Earn Mathematics 3A03 Real Analysis I

 Sequences VII 61/73

Mathematics
and Statistics
Z Z
dω = ω
M ∂M

Mathematics 3A03
Real Analysis I

Instructor: David Earn

Lecture 12
Sequences VII
Friday 1 February 2019

Instructor: David Earn Mathematics 3A03 Real Analysis I

 Sequences VII 62/73

Announcements

Assignment 2 was due before class today.

Assignment 3 will be posted this weekend.
Due Friday 15 Feb 2019 at 1:25pm.
IMPORTANT CHANGE:
For the remainder of the term, assignments must be submitted
electronically, not as a hardcopy.
By the end of this weekend, you should have received a link for
Assignment 3 via e-mail from crowdmark. If you have not
received such an e-mail by the end of the weekend, please
e-mail earn@math.mcmaster.ca.
If you write your solutions by hand, you will need to scan or
photograph them to submit them via the online system.
If you use LATEX to create a pdf file, you will need to separate
your solutions for each question.
Marked assignments will be available online, rather than being
returned in tutorial.
Instructor: David Earn Mathematics 3A03 Real Analysis I
 Sequences VII Sequences Finale 63/73

Sequences Finale!

Instructor: David Earn Mathematics 3A03 Real Analysis I

 Sequences VII Cauchy sequences 64/73

Cauchy sequences

Definition (Cauchy sequence)

A sequence {sn } is said to be a Cauchy sequence iff for all ε > 0
there exists N ∈ N such that if m ≥ N and n ≥ N then
|sn − sm | < ε.

Theorem (Cauchy criterion)

A sequence of real numbers {sn } is convergent iff it is a Cauchy
sequence.

Remark: The proof of the “only if” direction is easy. The proof of
the “if” direction contains only one tricky feature: showing that
every Cauchy sequence {sn } is bounded.

Instructor: David Earn Mathematics 3A03 Real Analysis I

 Sequences VII Cauchy sequences 65/73

Cauchy sequences

Proof of Cauchy criterion

“only if”: If {sn } converges then, given ε > 0, choose N ∈ N such that
for all n ≥ N, |sn − L| < ε/2. Then, for any m, n > N we have
|sm − sn | = |sm − L + L − sn | ≤ |sm − L| + |sn − L| < (ε/2) + (ε/2) = ε,
as required.
“if”: If we take ε = 1 in the definition of a Cauchy sequence, we find
that there is some N such that |sm − sn | < 1 for all m, n > N. In
particular, this means that |sm − sN+1 | < 1 for all m > N, i.e.,

sN+1 − 1 < sm < sN+1 + 1 ∀m > N.

Thus {sm : m > N} is bounded; moreover, since there are only finitely
many other si ’s, the whole sequence {sn } is bounded. Hence, by the
Bolzano-Weierstrass theorem, some subsequence of sn converges; let’s
write this subsequence as {smk }, and its limit as L.
. . . continued on next slide. . .
Instructor: David Earn Mathematics 3A03 Real Analysis I
 Sequences VII Cauchy sequences 66/73

Cauchy sequences
Proof of Cauchy criterion (cont’d).
We will show that {sn } converges to L. To prove this, consider any
ε > 0. Since the sequence {sn } is Cauchy, there is some N ∈ N such that
ε
|sn − sm | < for all n, m ≥ N.
2
Since the subsequence {smk } converges to L, there is some N 0 so that
ε
|smk − L| < for all k ≥ N 0 .
2
Now fix an integer k satisfying k ≥ N 0 and mk ≥ N. Then ∀n ≥ N,

|sn − L| ≤ |sn − smk + smk − L|

ε ε
≤ |sn − smk | + |smk − L| < + = ε,
2 2
as required.
Instructor: David Earn Mathematics 3A03 Real Analysis I
 Sequences VII Cauchy sequences 67/73

Cauchy sequences

Notes:
The Cauchy criterion is sometimes easier to use in proofs than
the original definition of convergence.

Its significance is more evident in spaces other than R, where

Cauchy sequences do not necessarily converge.

Instructor: David Earn Mathematics 3A03 Real Analysis I

 Sequences VII Countability 68/73

Bijectivity

Definition (Bijection)
Let f : A → B be a function. Then
(i) f is injective (or one-to-one) if ∀a1 , a2 ∈ A,
f (a1 ) = f (a2 ) =⇒ a1 = a2 .

(ii) f is surjective (or onto) if ∀b ∈ B, ∃a ∈ A so that f (a) = b.

(iii) Then f is bijective if it is both injective and surjective.

A bijective function is said to be a bijection.

Note: A bijection is sometimes called a one-to-one

correspondence. This termniology can be confusing because it
means one-to-one and onto.

Instructor: David Earn Mathematics 3A03 Real Analysis I

 Sequences VII Countability 69/73

Countability

Definition (Countable set)

A set S is countable if S is finite, or if there is a bijective function
f : N → S.
A set is uncountable if it is not countable.

Sequences {sn }∞
n=1 are generalizations of the intuitive notion of
sets whose elements can be counted.

Example
Suppose S is a subset of R. Then S is countable if and only if S is
the range of a sequence.

Instructor: David Earn Mathematics 3A03 Real Analysis I

 Sequences VII Countability 70/73

Countability

Theorem
The natural numbers N are countable.
(solution on board)

Theorem
The integers Z are countable.

(solution on board)

Theorem
The rational numbers Q are countable.

(solution on board) What about R ?. . .

Instructor: David Earn Mathematics 3A03 Real Analysis I

 Sequences VII Countability 71/73

Countability

Theorem (Cantor)
The real numbers R are uncountable.
(solution on board)

Notes:
The main argument in the proof is known as “Cantor’s
diagonal argument”.

We can infer that not only are some real numbers not
rational, but there are “many more” real numbers than
rational numbers.

Cantor’s proof depends on there being a binary expansion for

any real number number. . .
Instructor: David Earn Mathematics 3A03 Real Analysis I
 Sequences VII Countability 72/73

Countability
Theorem (Existence and uniqueness of binary expansions)
If x ∈ [0, 1) then there is a sequence {an } such that an ∈ {0, 1} ∀n and
∞
X an
x= n
.
n=1
2

n−1
$ ! %
X ai
Specifically, an = x− i
2n .
=
2
i 1

Moreover, this binary representation is unique unless x = m/2k for

some k ∈ N and m ∈ N, in which case there are exactly two binary
representations, the second being given by {bn } where

an n < k,

bn = 0 n = k,

1 n > k.


Instructor: David Earn Mathematics 3A03 Real Analysis I

 Sequences VII Countability 73/73

Countability

Theorem (Properties of countable sets)

(i) Any subset of a countable set is countable.

(ii) The union of a sequence of countable sets is countable.

(iii) No (non-degenerate∗ ) interval is countable.

∗
We normally assume implicitly that the endpoints of intervals are
distinct. If the endpoints are the same then the interval is degenerate,
e.g., (a, a) = ∅ and [a, a] = {a}.

Instructor: David Earn Mathematics 3A03 Real Analysis I