Math 341 – Lecture Notes on Chapter 6 – Sequences and Series of

Functions

Lecture 29–30

§6.2: Uniform Convergence of a Sequence of Functions

Pointwise Convergence
For each n ∈ N, let fn : A → R be a real-valued function on A. The sequence
(fn ) of functions converges pointwise on A to a function f if, for all x ∈ A,


the sequence of real number fn (x) converges to the real number f (x). We
often write
lim fn (x) = f (x) or lim fn = f.
n→∞ n→∞

Example 1

x2 + nx x2
fn (x) = = +x
n n
x2
lim fn (x) = lim + x = 0 + x = x.
n→∞ n→∞ n
If f (x) = x, then fn → f as n → ∞. In this case, the functions fn were every-
where continuous and differentiable and the limit function was also everywhere
continuous and differentiable.

1
Example 2

gn (x) = xn on the set [0, 1].

0≤x<1 ⇒ lim gn (x) = 0
n→∞
x=1 ⇒ lim gn (x) = lim 1n = 1.
n→∞ n→∞

Thus, limn→∞ gn (x) = g(x) where

(
0 if 0 ≤ x < 1,
g(x) =
1 if x = 1.

In this case, the functions gn (x) are continuous on [0, 1], but the limit function
g(x) is not continuous at every point of [0, 1].

Example 3

1
hn (x) = x1+ 2n−1 on [−1, 1],
1
lim hn (x) = x lim x 2n−1 = |x|,
n→∞ n→∞
h(x) := |x|
lim hn (x) = h(x) = |x| for x ∈ [−1, 1].
n→∞

Each of the functions hn (x) in the sequence is continuous and differentiable on

[−1, 1], but the limit function h(x) = |x| is not differentiable at all points of
[−1, 1].

2
 Important Question
Suppose f (x) = limn→∞ fn (x) for all x ∈ A. What additional hypothesis would
ensure the following?

(i) If each fn is continuous on A, then f is continuous on A.

(ii) if each fn is differentiable on A, then f is differentiable on A.

The best general answer to these questions has to do with the notion of uniform
convergence.

Definition (Uniform Convergence)

Let (fn ) be a sequence of functions defined on A ⊆ R. We say that (fn )
converges uniformly on A to the limit function f defined on A if for every
 > 0 there exists an N ∈ N such that
 
fn (x) − f (x) <  for all x ∈ A,

whenever n ≥ N .

Here is a figure that graphically depicts the definition:

3
y

y = f (x) + 

y = fn (x), n ≥ N

y = f (x)

y = f (x) − 

x
| {z }
A

fn → f uniformly on A

4
Example of Non-Uniform Convergence

1
nx

 if 0 ≤ x ≤ n
gn (x) = 2 − nx if n1 ≤ x ≤ 2
n


0 otherwise.
lim gn (x) = 0.
n→∞

If g(x) := 0, then (gn ) → g pointwise.

1
Let  = 2 and xn = n1 . Then
   
gn (xn ) − g(xn ) = 1 − 0 = 1 >  = 1 .
2

So, it is not true that for all  > 0, there exist an N ∈ N large enough such
that n ≥ N implies |gn (x) − g(x)| <  for all x. So, (gn ) does not converge to
g uniformly.

( n1 , 1)
1
y = gn (x)

0 1 2
n n 1

limn→∞ gn (x) = 0 for all x

5
Another Example of Non-Uniform Convergence

enx
hn (x) = nx
, x ∈ R, n ∈ N.
1
 + e
0 if x < 0,


lim hn (x) = 12 if x = 0,
n→∞ 

1 if x > 0.

Plot of h1 (x), h3 (x), and h10 (x).

1.0

0.8

0.6

0.4

0.2

-3 -2 -1 1 2 3

6
Example of Uniform Convergence
q
fn (x) = x2 + n12
√ √
lim fn (x) = x2 + 0 = x2 = |x|.
n→∞

So, fn (x) → f (x) = |x| pointwise.

Let  > 0 be given. Choose N ∈ N large enough such that N1 < . Then for any x ∈ R and
n ≥ N we have
q 
 
|fn (x) − f (x)| =  x2 + n12 − |x|
q 
q  x 2 + 1 + |x|
  n2
=  x2 + n12 − |x|  q 
2 1
x + n2 + |x|
1
n2
=q
1
x2 + n2
+ |x|
1
2 1
≤q n = < .
0 + n12 + 0 n

This shows that (fn ) → f uniformly on R.

Note that each fn (x) is both continuous and differentiable on R, but f (x) = |x| is continuous
on R and not differentiable at x = 0.

Plot of f1 (x), f3 (x), f10 (x), and |x|.

3.0

2.5

2.0

1.5

1.0

0.5

-3 -2 -1 1 2 3

7
 Cauchy Criterion for Uniform Convergence
A sequence of functions (fn ) defined on a set A ⊆ R converges uniformly on A
if and only if for every  > 0 there exists an N ∈ N such that |fn (x)−fm (x)| < 
whenever m, n ≥ N and x ∈ A.

Proof. (⇒) Assume the sequence (fn ) converges uniformly on A to a limit func-
tion f . Let  > 0 be given. Then there exists an N ∈ N such that
fn (x) − f (x) <  ,
 
2
whenever n ≥ N and x ∈ A. Then if n, m ≥ N and x ∈ A, we have
   
fn (x) − fm (x) = fn (x) − f (x) + f (x) − fm (x)
   
≤ fn (x) − f (x) + fm (x) − f (x)
 
< + = .
2 2

(⇐) Conversely, assume that for every  > 0 there exists an N ∈ N such that
 
fn (x) − fm (x) whenever m, n ≥ N and x ∈ A. This hypothesis implies that, for
each x ∈ A, (fn (x)) is a Cauchy sequence. By Cauchy’s Criterion, this sequence
converges to a point, which we will call f (x). So, the uniformly Cauchy sequence
converges pointwise to the function f (x). We must show that the convergence is
also uniform. For the value of  given above, we use the corresponding N . Then
for n, m ≥ N and all x ∈ A,
 
fn (x) − fm (x) ≤ .

Taking the limit as m → ∞ gives

 
fn (x) − f (x) ≤  for all x ∈ A,

which shows that (fn ) converges uniformly to f on A. This completes the proof.

Continuous Limit Theorem

Let (fn ) be a sequence of functions defined on A ⊆ R that converges uniformly
on A to a function f . If each fn is continuous at c ∈ A, then f is continuous
at c.

8
Proof. Let  > 0 be given. Fix c ∈ A. Since fn → f uniformly, there exists an
N ∈ N such that

|fN (x) − f (x)| < for all x ∈ A.
3
Since fN is continuous at c, there exists δ > 0 such that

|fN (x) − fN (c)| <
3
whenever |x − c| < δ.

If |x − c| < δ, then
   
f (x) − f (c) = f (x) − fN (x) + fN (x) − fN (c) + fN (c) − f (c)
     
≤ f (x) − fN (x) + fN (x) − fN (c) + fN (c) − f (c)
  
≤ + + = .
3 3 3
The first and third /3 are due to uniform convergence and the choice of N . The
second /3 is due to the choice of δ. This shows that f is continuous at c, as
desired.

Remark about Previous Theorem

In the previous theorem, we did not assume that each fn was continuous at
every c ∈ A. Let D = {c ∈ A : fn is continuous at c for all n ∈ N}. Then the
limit function f is guaranteed to be continuous at every c ∈ D. If E = {c ∈
A : f is differentiable at c}, then E ⊆ D ⊆ A.

9
Lecture 31

§6.3 Uniform Convergence and Differentiation

Differentiable Limit Theorem
Let (fn ) → f pointwise on the closed interval [a, b] and assume each fn is
differentiable. If (fn0 ) → g uniformly on [a, b], then f is differentiable and
f 0 = g.

Proof. Fix c ∈ [a, b] and let  > 0. We’ll show there exists δ > 0 such that
 
 f (x) − f (x) 
 − g(c) < ,
 x−c 

whenever 0 < |x − c| < δ and x ∈ [a, b].

For x 6= c, consider the following:

   
 f (x) − f (x)   f (x) − f (c) fn (x) − fn (c) 
 − g(c) ≤  − 
 x−c x−c x−c 
| {z }
(iii)

 fn (x) − fn (c)
 (†)
− fn0 (c) + |fn0 (c) − g(c)|

+ 
x−c | {z }
| {z } (i)
(ii)

Since limn→∞ fn0 (c) = g(c), there exists N1 ∈ N such that


|fn0 (c) − g(c)| < (1)
3
for all n ≥ N1 .

From Cauchy’s Criterion for uniform convergence, since the sequence (fn0 ) converges
uniformly to g, there exists an N2 ∈ N such that

|fn0 (x) − fm
0
(x)| <
3
whenever m, n ≥ N and x ∈ [a, b]. Set N = max{N1 , N2 }.

10
The function fN is differentiable at c. So there exists δ > 0 such that
 
 fN (x) − fN (c) 0
 
 − f N (c)< (2)
 x−c  3

whenever 0 < |x − c| < δ and x ∈ [a, b]. We’ll show this δ will suffice.

Suppose 0 < |x − c| < δ and m ≥ N . By the Mean Value Theorem applied to

fm − fN on the interval [c, x] (if x < c the argument is the same) there exists
α ∈ (c, x) such that

0 [fm (x) − fN (x)] − [fm (c) − fN (c)]

fm (α) − fN0 (α) = .
x−c
By our choice of N ,
0 
|fm (α) − fN0 (α)| <
3
and so  
 [fm (x) − fN (x)] − [fm (c) − fN (c)]  
 < .
 x−c  3
Since fm → f as m → ∞, by the Algebraic Order Limit Theorem,
 
 f (x) − f (c) fN (x) − fN (c)  

 x−c − ≤ . (3)
x−c  3

Combining inequalities (†), (1), (2), and (3), we obtain for 0 < |x − c| < δ and
x ∈ [a, b]  
 f (x) − f (c)    

 x−c − g(c) < + + = .
 3 3 3
This proves that f = limn→∞ fn is differentiable and that f 0 = g = limn→∞ fn0 .

11
Example
In the previous lecture we examined the example
r
1
hn (x) = x2 +
n
We showed that (hn (x)) → h(x) = |x| uniformly on R. However, since the function h(x) = |x|
is not differentiable at x = 0, by the previous theorem, we know that h0n (x) does not converge
uniformly to a limit function on R. Note that

1 if x > 0,


0 x
lim hn (x) = lim q = 0 if x = 0,
n→∞ n→∞
x2 + n1 

−1 if x < 0.

Here is a plot of h1 (x), h3 (x), h10 (x), and |x|.

3.0

2.5

2.0

1.5

1.0

0.5

-3 -2 -1 1 2 3

Here is a plot of h01 (x), h03 (x), and h010 (x).

1.0

0.5

-3 -2 -1 1 2 3

-0.5

-1.0

12
 Example

x x2
gn (x) = +
2 2n
x x2
 
x x
g(x) = lim gn (x) = lim + = +0= (pointwise)
n→∞ n→∞ 2 2n 2 2
1
g 0 (x) =
2
On the other hand,
1 x 1
gn0 (x) = + → h(x) = = g 0 (x) (pointwise for all x ∈ R.)
2 n 2

We’ll now examine how the previous theorem applies to this example. Consider
the interval [−M, M ] where M > 0. Let h(x) = 21 for all x ∈ R. Let  > 0 be
given. Let N ∈ N be large enough such that M
N < . Then if x ∈ [−M, M ] and
n ≥ N , we have
    
 0   1 x 1  x M
gn (x) − h(x) =  + − = ≤ < .
 2 2 2 n N
This shows that gn0 converges uniformly to h on [−M, M ]. Because we verified
that (gn0 ) converges uniformly on [−M, M ], the theorem tells us that

lim gn0 (x) = h(x) = g 0 (x) for x ∈ [−M, M ].

n→∞

Since M is arbitrary we can conclude that

lim gn0 (x) = h(x) = g 0 (x) for all x ∈ R.

n→∞

Theorem Related to Differentiable Limit Theorem

Let (fn ) be a sequence of differentiable functions defined on the closed interval
[a, b], and assume (fn0 ) converges uniformly on [a, b]. If there exists a point


x0 ∈ [a, b] where the sequence fn (x0 ) converges, then (fn ) converges uniformly
on [a, b].

Proof. (This is Exercise 6.3.7) Let x ∈ [a, b] where x 6= x0 . Both x and x0 will

13
be fixed real numbers throughout the proof. Without loss of generality, we may
assume x > x0 . (If x < x0 , the argument is the same.) By the Mean Value Theorem
applied the function fn − fm on the interval [x0 , x], there exists some α ∈ (x0 , x)
(α depends on m and n) such that
[fn (x) − fm (x)] − [fn (x0 ) − fm (x0 )]
fn0 (α) − fm
0
(α) =
x − x0
which implies
[fn (x) − fm (x)] − [fn (x0 ) − fm (x0 )] = fn0 (α) − fm
0


(α) (x − x0 ).

Let  > 0 be given. Since (fn0 ) converges uniformly, by Cauchy’s Criterion for
uniformly convergent sequences of functions, there exists some N1 ∈ N such that

|fn (c) − fm (c)| < for all n, m ≥ N1 and c ∈ [a, b].
2(b − a)


By hypothesis, the sequence fn (x0 ) converges. So there exists an N2 ∈ N such
that

|fn (x0 ) − fm (x0 )| < for all m, n ≥ N2 .
2
Let N = max{N1 , N2 }. Then if m, n ≥ N , we have


  
|fn (x) − fm (x)| ≤  fn (x) − fm (x) − fn (x0 ) − fm (x0 )  + fn (x0 ) − fn (x0 )
=  fn0 (α) − fm0

  
(α) (x − x0 ) + fn (x0 ) − fm (x0 )
   
< (x − x0 ) + ≤ + = .
2(b − a) 2 2 2


Since the choice of N was independent of x, this proves that fn converges uni-
formly on [a, b].

Combining the previous two theorems gives a stronger version of the Differentiable
Limit Theorem.
Better Version of Differentiable Limit Theorem
Let (fn ) be a sequence of differentiable functions defined on the closed interval
[a, b], and assume that the sequence (fn0 ) converges uniformly to a function g on
[a, b]. If there exists a point x0 ∈ [a, b] for which (fn (x0 )) converges, then (fn )
converges uniformly. Moreover, the limit function f = lim fn is differentiable
and satisfies f 0 = g.

14
Lecture 32

§6.4 Series of Functions

Examples
P∞ xn
1. f (x) = n=0 n! (This equals ex for all x ∈ R.)
P∞ (x−1)n
2. g(x) = n=1 n (This equals − ln(2 − x) for |x − 1| < 1.)

0 if x = 0,


P∞ sin(nπx)
3. h(x) = n=1 nπ/2 = 1−x if 0 < x < 2,


h(x + 2n) for all x ∈ R and n ∈ Z.

Definitions
Let f and fn for n ∈ N be functions defined on a set A ⊆ R.

(a) The infinite series

∞
X
fn (x) = f1 (x) + f2 (x) + f3 (x) + · · ·
n=1

converges pointwise on A to f (x) if the sequence of partial sums

sk (x) = f1 (x) + f2 (x) + · · · + fk (x)

converges pointwise to f (x) on A.

(b) The infinite series converges uniformly on A to f (x) if the sequence

of partial sums converges uniformly on A to f (x)

We already know lots of theorems!

Since an infinite series of functions is defined in terms of the limit of a sequence
of partial sums, everything we already know about sequences applies to series.
For the sum ∞
P
n=1 fn (x), we merely restate all of the previous theorems for the
sequence of kth partial sums sk (x) = f1 (x) + · · · + fk (x).

15
 Term-by-term Continuity Theorem
Let fn be continuous functions defined on a set A ⊆ R, and assume ∞
P
n=1 fn
converges uniformly on A to a function f . Then f is continuous on A.

Proof. Apply the Continuous Limit Theorem to the partial sums sk = f1 + f2 +

· · · + fk .

Term-by-term Differentiability Theorem

Suppose the following three statements:

(i) Let fn be differentiable functions defined on an interval A = [a, b].

(ii) Assume ∞ 0
P
n=1 fn (x) converges uniformly to a limit g(x) on A.

(iii) There exists a point x0 ∈ [a, b] where ∞

P
n=1 fn (x0 ) converges.

Then the series ∞

P
n=1 fn (x) converges uniformly to a differentiable function f (x)
satisfying f 0 (x) = g(x) on A. In other words,
∞
X ∞
X
0
f (x) = fn (x) and f (x) = fn0 (x).
n=1 n=1

Proof. Apply the stronger version of the Differentiable Limit Theorem to the partial
sums sk = f1 + f2 + · · · + fk .

Cauchy Criterion for Uniform Convergence of Series

A series ∞
P
n=1 fn converges uniformly on A ⊆ R if and only if for every  > 0
there exists an N ∈ N such that
 
fm+1 (x) + fm+2 (x) + · · · + fn (x) < 

whenever n > m ≥ N and x ∈ A.

16
 Corollary (Weierstrass M -Test)
For each n ∈ N , let fn be a function defined on a set A ⊆ R, and let Mn > 0
be a real number satisfying
 
fn (x) < Mn

for all x ∈ A. If ∞
P P∞
n=1 M n converges, then n=1 fn converges uniformly on A.

P∞
Proof. Let  > 0. Since n=1 Mn converges, there exists N ∈ N such that n >
m ≥ N implies
Mm+1 + Mm+2 + · · · + Mn < .
Since
       
fm+1 (x) + fm+2 + · · · + fn (x) ≤ fm+1 (x) + fm+2 (x) + · · · + fn (x)
< Mm+1 + Mm+2 + · · · + Mn < ,

the series converges uniformly by Cauchy’s Criterion.

Example
cos(2n x)
The continuous function 2n for n ∈ {0, 1, 2, 3, . . . } satisfy
 cos(2n x) 
 
  ≤ Mn = 1
 2n  2n
for all x ∈ R and n ∈ {0, 1, 2, 3, . . . }. Since
∞ ∞
X X 1 1
Mn = n
= = 2 < ∞,
n=0 n=0
2 1 − (1/2)
P∞ cos(2n x)
the series n=0 2n converges uniformly to a continuous function
∞
X cos(2n x)
g(x) =
n=0
2n

on R.

cos(2n x)
The function g(x) = ∞
P
n=0 2n is also an example of an everywhere continuous
nowhere differentiable function. Its plot is given below:

17
 2.0

1.5

1.0

0.5

-4 π -3 π -2 π -π π 2π 3π 4π

-0.5

-1.0

Example
Define ∞
X x2n
g(x) = .
n=0
1 + x2n
Find the values of x where the series converges and show that we get a contin-
uous function on this set.

Solution. If |x| < 1, then by the Comparison Test, the series converges as follows:
∞ ∞
X x2n X 1
0≤ 2n
≤ x2n = 2
< ∞.
n=0
1+x n=0
1 − x

If |x| ≥ 1, then the series diverges by the Divergence Test since

(
2n 1
x 2 if x = ±1,
lim =
n→∞ 1 + x2n 1 if |x| > 1.

Now let 0 < K < 1. Then on the interval [−K, K] we have

 2n 
 x  2n 2n
 1 + x2n  ≤ x ≤ Mn = K .
 

Since ∞ ∞
X X 1
Mn = K 2n = < ∞,
n=0 n=0
1 − K2

18
the series converges uniformly on [−K, K] to a continuous function. Since K was
arbitrary, the series
∞
X x2n
g(x) =
n=0
1 + x2n
is a continuous function on (−1, 1).

Example
Let
1 1 1 1 1
f (x) = − + − + − ··· .
x x+1 x+2 x+3 x+4
Show that f is defined for all x > 0. Is f continuous on (0, ∞)? How about
differentiable?

Hint. The series f (x) converges at all x > 0 by the Alternating Series Test.
Then show that on any interval of the form [M, ∞) with 0 < M , the series
converges uniformly. Use the Weierstrass M -test. Apply the Weierstrass M -
test again to the series of derivatives.

19
Lecture 33

§6.5 Power Series

Definition of Power Series
Let (an )∞
n=0 be a sequence of real numbers and let c ∈ R. The power series
centered at c with coefficients an is the series
∞
X
an (x − c)n .
n=0

Examples

∞
X xn
centered at 0
n=0
n!
∞
X (x − 1)n
centered at 1
n=1
n
∞
X
nxn centered at 0
n=0
∞
X (x + 2)n
centered at −2
n=1
n2

Theorem
If a power series ∞ n
P
n=0 an x converges at some nonzero point x0 ∈ R, then it
converges absolutely for any x satisfying |x| < |x0 |.

Proof. If ∞ n n
P
n=0 an x0 converges, then limn→∞ an x0 = 0. So there exists some M > 0
such that |an x20 | < M for all n ≥ 0. If x satisfies |x| < |x0 |, then
 n  n
n x
n
  x
|an x | = |an x0 |   ≤ M   .
x0 x0

20
Then by the Comparison Test the series ∞ n
P
n=0 an x converges absolutely as follows:
 
∞ ∞ ∞  n
X  X
n n
X x M
an x  ≤ |an x | ≤ M   =   < ∞.

x0 1 −  xx 

 
n=0 n=0 n=0 0

P∞ n
Thus n=0 an x converges absolutely for x satisfying |x| < |x0 |.

Definition (Radius of Convergence)

P∞ n
Let R = sup{|x0 | : n=0 aP
n x0 converges}. Then R is called the radius of
convergence of the series ∞ n
n=0 an x .

Note: From the previous theorem and the definition of the radius of convergence,
it is clear that if 0 < R < ∞, then the series converges for |x| < R and diverges for
|x| > R.

Corollary
If the series ∞ n
P
n=0 an x has radius of convergence R, then the set of all x for
which the series converges is one of the following intervals:

(i) If R = 0, the series converges only for x ∈ {0}.

(ii) If 0 < R < ∞, the series converges for all x in one of the following four
intervals:
(−R, R), [−R, R), (−R, R], [−R, R].

(iii) If R = ∞, then the series converges for all x ∈ R = (−∞, ∞).

Proof. If you understood the previous theorem and the definition of the radius of
convergence, then this corollary is obvious. If the corollary is not obvious to you,
go back and figure out the previous theorem and definition.

Theorem
If a power series ∞ n
P
n=0 an x converges absolutely at a point x0 , then it converges
uniformly on the closed interval [−|x0 |, |x0 |].

21
 P∞
Proof. For each n ≥ 0, let Mn = |an xn0 |. By hypothesis the series n
n=0 an x0
P∞ n
P ∞
converges absolutely and so n=0 |an x0 | = n=0 Mn converges. Then for any
x ∈ [−|x0 |, |x0 |], we have
∞  ∞ ∞ ∞
X  X X X
n n n
an x  ≤ |an x | ≤ |an x0 | = Mn < ∞.


 
n=0 n=0 n=0 n=0

By the Weierstrass M -test, the series converges uniformly on the closed interval
[−|x0 |, |x0 |].

If a series centered at 0 converges at a point x = R 6= 0, then the convergence

on the closed interval [0, R] is especially nice, as described by Abels’s Theoreom
below:
Abel’s Theorem about Uniform Convergence
Let g(x) = ∞ n
P
n=0 an x be a power series that converges at the point x = R > 0.
Then the series converges uniformly on the interval [0, R]. A similar result
holds if the series converges at x = −R.

Proof. The proof is quite interesting, but the calculation if fairly long and not very
useful to go through in class. If you’re interested, read the proof in the book and
do several of the exercises that contain parts of the proof.

Example
The series ∞
X (−1)n+1 xn
f (x) =
n=1
n
converges conditionally at x = 1 but diverges at x = −1. So the radius of
convergence is R = 1 and the interval of convergence is (−1, 1]. By Abel’s
Theorem, the series converges uniformly on the closed interval [0, 1].

22
 Example
It can be verified that
∞
X (x − 1)n
− ln(2 − x) = for x ∈ [0, 2).
n=1
n

The series is centered at c = 1 and has radius of convergence R = 1. Since the

series converges if and only if x ∈ [c − R, c + R) = [0, 2), by Abel’s Theorem,
the series converges uniformly on the closed interval [c − R, c] = [0, 1]. This
is illustrated in the figure below. The graph of y = − ln(2 − x), which has a
vertical asymptote at x = 2, is in black. We don’t expect the partial sums to
converge uniformly near the asymptote. The partial sum s2 (x) is in blue, and
the partial sum s5 (x) is in magenta. The uniform convergence of the sequence
of partial sums on the closed interval [0, 1] is strongly suggested by the plots.

-1 1 2 3

-2

-4

Theorem
If a power series converges pointwise on the set A ⊆ R, then it converges
uniformly on any compact subset K ⊆ A.

Proof. A compact set K contains both a minimum x0 and a maximum x1 , which

by hypothesis belong to A. By Abel’s Theorem, the series converges uniformly on
the closed interval [x0 , x1 ] and hence also on K.

23
 Theorem
P∞
If a xn converges for all x ∈ (−R, R), then the differentiated series
P∞ n=0 nn−1
n=1 nan x converges at each x ∈ (−R, R) as well. Consequently, the con-
vergence is uniform on compact sets contained in (−R, R).

Proof. Let x0 ∈ (−R, R). From an earlier theorem, ∞ an xn0 converges absolutely.
P
n=0
Let |x0 | < t < R. Similarly, ∞ n n
P
n=0 an t converges absolutely. Hence limn→∞ |an t | =
0. Then,
∞ ∞    
X X 1  x0 n−1
|nan x0n−1 | = n  |an tn |.
n=1 n=1
t t
 n−1
Since limn→∞ n  xt0  = 0, the nth term on the right hand side of the sum in the
last displayed formula is < |an tn | for all sufficiently large n. By the Comparison
Test, the sum ∞
P n−1
P∞ n−1
n=1 |na x
n 0 | converges, and so n=1 nan x0 converges absolutely
at all x0 ∈ (−R, R). By the previous theorem, the series also converges uniformly
on any compact subset of (−R, R).

Theorem (The Derivative of a Power Series)

Assume ∞
X
f (x) = an x n
n=0
converges on an interval A ⊆ R. The function is continuous on A and differen-
tiable on any open interval (−R, R) ⊆ A. The derivative is given by
∞
X
0
f (x) = nan xn−1 .
n=1

Moreover, f is infinitely differentiable on (−R, R), and the successive derivatives

can be obtained via term-by-term differentiation of the appropriate series.

Proof. On any closed subinterval [a, b] ⊂ (−R, R), the series f (x) = ∞ n
P
n=0 an x
converges uniformly. Since each an xn is continuous, the series represents a con-
tinuous function f (x) on [a, b]. Since [a, b] ⊂ (−R, R) was arbitrary, the series is
continuous on the whole interval (−R, R).

24
 P∞ n−1
By the previous theorem, this same argument applies to the series n=1 nan x .
So, by the Term-by-Term Differentiability Theorem, we have
∞
X
0
f (x) = nan xn−1
n=1

for all x ∈ (−R, R).

Since the derivative series f 0 (x) satisfies the same hypothesis as the series f (x),
all derivative of all orders exist on (−R, R) and are obtained by differentiating the
original series the appropriate number of times.

Term-by-term Antidifferentiation
Assume f (x) = ∞ n
P
n=0 an x converges on (−R, R), then
∞
X an n+1
F (x) = x
n=0
n + 1

is defined on (−R, R) and satisfies F 0 (x) = f (x).

Proof. Exercise.

25
Lecture 35

§6.6 Taylor Series

Recall a few theorems from §6.5 on power series:

Theorem 6.5.7 (The Derivative of a Power Series)

Assume ∞
X
f (x) = an x n
n=0
converges on an interval A ⊆ R. The function is continuous on A and differen-
tiable on any open interval (−R, R) ⊆ A. The derivative is given by
∞
X
0
f (x) = nan xn−1 .
n=1

Moreover, f is infinitely differentiable on (−R, R), and the successive derivatives

can be obtained via term-by-term differentiation of the appropriate series.

Exercise 6.5.4 (Term-by-term Anti-differentiation)

Assume f (x) = ∞ n
P
n=0 an x converges on (−R, R), then
∞
X an n+1
F (x) = x
n=0
n + 1

is defined on (−R, R) and satisfies F 0 (x) = f (x).

26
Example

1
f (x) = = 1 + x + x2 + x3 + · · · x ∈ (−1, 1)
1−x
1
f 0 (x) = 2
= 1 + 2x + 3x2 + 4x3 + · · · x ∈ (−1, 1)
(1 − x)
x2 x3
F (x) = − ln(1 − x) = C + x + + + ··· x ∈ [−1, 1)
2 3
Evaluate the last formula at x = 0 to determine that the constant of integration
is C = 0.

Question
Let the function f (x) be n-times differentiable on the interval (c − R, c + R).
What is the “best” polynomial

pn (x) = a0 + a1 (x − c) + a2 (x − c)2 + · · · + an (x − c)n

of degree ≤ n for approximating f (x) in the interval (c − R, c + R) subject to

the conditions

f (c) = pn (c)
f 0 (c) = p0n (c)
f 00 (c) = p00n (c)
..
.
f (n) (c) = p(n)
n (c)?

pn (x) = a0 + a1 (x − c)1 + a2 (x − c)2 + · · · + an (x − c)n f (c) = pn (c) f (c) = a0

p0n (x) 1 2
= 1 · a1 + 2 · a2 (x − c) + 3 · a3 (x − c) + · · · + n · an (x − c) n−1 0
f (c) = p0n (c) f 0 (c) = 1!a1
p00
n (x)
1 2 n−2
= 2 · 1 · a2 + 3 · 2 · a3 (x − c) + 4 · 3 · a4 (x − c) + · · · n · (n − 1) · an (x − c) 00
f (c) = p00
n (c) f 00 (c) = 2!a2
..
.
(n) (n)
pn (x) = n · (n − 1) · (n − 2) · · · 3 · 2 · 1 · an f (n) (c) = pn (c) f (n) (c) = n!an

27
 nth Taylor Polynomial
n
X f (k) (c)
pn (x) = (x − c)k .
k!
k=0

Error term
The error term is En (x) where

f (x) = pn (x) + En (x)

Can we take a limit?

Assume f (x) is infinitely differentiable on A = (c − R, c + R). If

lim En (x) = 0,
n→∞

then
n ∞
X f (k) (c) k
X f (k) (c)
f (x) = lim pn (x) = lim (x − c) = (x − c)k
n→∞ n→∞ k! k!
k=0 k=0

for x ∈ A.

Theorem 6.6.3 (Lagrange Remainder Theorem)

Let f be N + 1 times differentiable on A = (−R, R). Let
N
X f (k) (0)
pN (x) = xk .
k!
k=0

Given x 6= 0 in A, there exists a point c satisfying |c| < |x| where the error
function
EN (x) = f (x) − pN (x)
satisfies
f (N +1) (c) N +1
EN (x) = x .
(N + 1)!

Proof. Due to our choice of coefficients of the Taylor polynomial,

(k)
f (k) (0) = pN (0) k = 0, 1, 2, . . . , N

28
and so
(N )
EN (0) = EN0 (0) = EN00 (0) = · · · = EN (0) = 0.
Consider the case x > 0 where x ∈ A. (The case x < 0 is similar.) Apply the
Generalized Mean Value Theorem to the function EN (x) and xN +1 on the interval
[0, x]. There exists x1 ∈ (0, x) such that
EN (x) − EN (0) EN (x) EN0 (x1 )
= N +1 = .
xN +1 − 0N +1 x (N + 1)xN 1

Then apply the Generalized Mean Value Theorem to the last expression on the
interval [0, x1 ]. There exists x2 ∈ (0, x1 ) such that
EN (x) EN0 (x1 ) EN00 (x2 )
= = −1
.
xN +1 (N + 1)xN 1 (N + 1)N xN 2

Similarly, there is x3 ∈ (0, x2 ) such that

EN (x) EN0 (x1 ) EN00 (x2 ) E (3) (x3 )

= = N −1
= N −2
.
xN +1 (N + 1)xN 1 (N + 1)N x 2 (N + 1)N (N − 1)x 3

Repeatedly applying the Generalized Mean Value Theorem produces numbers

0 < c = xN +1 < xN < · · · < x2 < x1 < x

such that
EN (x) E (N +1) (xN +1 )
= · · · = .
xN +1 (N + 1)!
If c = xN +1 , then
E (N +1) (c) N +1
EN (x) = x .
(N + 1)!

29
An extreme example
Consider the function
(
0 if x = 0,
f (x) :=
1


exp − x2 if x 6= 0.

A plot is given below

0.8

0.6

0.4

0.2

-4 -2 2 4

It can be shown that f (k) (0) = 0 for all k = 0, 1, 2, 3, . . .. Hence, every Taylor
polynomial is identically zero. Thus

f (x) = pn (x) +En (x).

| {z }
=0

For x 6= 0,
lim En (x) = f (x) 6= 0.
n→∞

So, the Taylor series for f (x) centered at 0 equals f (x) if and only if x = 0.

30
Example
Let f (x) = cos(x). Then

f (0) = cos(0) = 1
f 0 (0) = − sin(0) = 0
f 00 (0) = − cos(0) = −1
f 000 (0) = sin(0) = 0
..
.
f (k+4) (0) = f (k) (0).

So, the 2nth and (2n + 1)st Taylor polynomials are

x2 x4 n x
2n
p2n (x) = p2n+1 (x) = 1 − + + · · · + (−1)
2! 4! (2n)!
Then for x > 0, there is some c with c ∈ (0, x) such that

|f (2n+1) (c)| x2n+1 x2n+1

|E2n (x)| = |E2n+1 (x)| = ≤ → 0 as n → ∞.
(2n + 1) (2n + 1)
The case x < 0 is similar. Thus, for all x ∈ R,
∞
X (−1)k x2k
cos(x) = .
(2k)!
k=0

31
Example
1
f (x) =
x
centered at c = 1.
f (x) = x−1 f (1) = 1
0 −2
f (x) = (−1)x f 0 (1) = −1
f 00 (x) = (−1)(−2)x−3 f 00 (1) = (−1)2 2!
..
.
f (k) (x) = (−1)(−2) · · · (−k)x−k−1 f (k) (1) = (−1)k k!

nth Taylor polynomial

n n
X (−1)k k! X
pn (x) = (x − 1)k = (−1)k (x − 1)k
k!
k=0 k=0

Error term: for some d between x and 1

f (n+1) (d)
En (x) = (x − 1)n+1 = (−1)n+1 d−n (x − 1)n+1 → 0 as n → ∞
(n + 1)!

if and only if |x − 1| < 1.

32