POWER SERIES AND TAYLOR SERIES

(ADDITIONAL NOTES TO STEWART §11.8 – §11.10)

Theorem A
∞
X ∞
X
1. If cn xn converges at a 6= 0, then cn xn converges absolutely for all x such that |x| < |a|.
n=1 n=1

∞
X ∞
X
n
2. If cn x diverges at b 6= 0, then cn xn diverges for all x such that |x| > |b|.
n=1 n=1

Proof.
∞
X
1. Suppose that cn an converges. Then lim (cn an ) = 0, so that |cn an | ≤ 1 for all n ≥ N , say.
n→∞
n=1
∞
n n x n x n x
X x n
It follows that 0 ≤ |cn x | = |cn a | a
≤ a
. If |x| < |a|, then a
< 1, so that < ∞.
n=1
a
∞
X
Hence, using the comparison test, it follows that |cn xn | converges.
n=1

∞
X
2. Suppose that cn bn diverges. Also suppose that there exists an x such that |x| > |b| and
n=1
∞
X ∞
X
n
cn x converges. Then it follows from (1) that cn bn converges absolutely — contradiction.
n=1 n=1
Hence the result follows. 

Theorem B (Theorem 4, p.789 in Stewart)

∞
X
For a power series cn (x − a)n there are (only) three possibilities:
n=1

1. The series converges only for x = a.

2. The series converges for all x ∈ R.

3. There exists an R > 0 such that the series converges for all x such that |x−a| < R and diverges
for all x such that |x − a| > R.

Proof. We only prove the case where a = 0. Suppose that (1) and (2) do not hold. Then there
∞
X ∞
X
n
exist non-zero numbers b and d such that cn b converges and cn dn diverges. Hence the set
n=1 n=1
∞
X ∞
X
S = {|x| : cn xn < ∞} is not empty. By Theorem A (2) cn xn diverges for all x such that
n=1 n=1
|x| > |d|. Hence it follows that if |x| ∈ S, then |x| ≤ |d|, so that |d| is an upper bound for S. It
follows from the Completeness Property of R that R = sup S exists. If |x| > R, then |x| 6∈ S, so that
X∞
cn xn diverges. If |x| < R, then |x| is not an upper bound for S, so that there exists an element
n=1

1
 ∞
X
|a| ∈ S for which we have that |a| > |x|. Since |a| ∈ S, it follows that cn an converges, so that,
n=1
∞
X
by Theorem A (1), cn xn converges. 
n=1

The number R is called the radius of convergence of the series, and the interval I consisting of all x
where the series converges is called the interval of convergence of the series. In (1) R = 0 and in (2)
R = ∞.

Theorem C
∞
X
Suppose that cn (x − a)n is a power series with cn 6= 0 for all n and radius of convergence R. Let
n=1

cn+1
L = lim .
n→∞ cn

1. If L ∈ R\{0}, then R = L1 .

2. If L = 0, then R = ∞.

3. If L = ∞, then R = 0.

Proof. Let an = cn (x − a)n . Then

cn+1 (x − a)n+1
 
an+1 cn+1
lim = lim = lim |x − a| .
n→∞ an n→∞ cn (x − a)n n→∞ cn

an+1
It follows from this that if L = ∞, then lim = ∞, provided that x 6= a, so that the series
n→∞ an
diverges for all x 6= a, i.e. R = 0.
an+1
It also follows that if L ∈ R, then lim = L|x − a|. Hence, if L 6= 0, then, using the ratio
n→∞ an
test, it follows that the series converges if L|x − a| < 1, i.e. if |x − a| < L1 ; and diverges if L|x − a| > 1,
an+1
i.e. if |x − a| > L1 ; so that R = L1 . Finally, if L = 0, then it follows that lim = 0, so that the
n→∞ an
series converges for all x, i.e. R = ∞. 

Theorem D ∞
X p
Suppose that cn (x − a)n is a power series with radius of convergence R. Let L = lim n
|cn |.
n→∞
n=1

1. If L ∈ R\{0}, then R = L1 .

2. If L = 0, then R = ∞.

3. If L = ∞, then R = 0.

2
Prove this.

Lemma E
If  > 0, then |nxn−1 | < (|x| + )n for all n large enough.

1 1 1
Proof. Let  > 0. Since lim |nxn−1 | n = lim n n |x|1− n = |x|, it follows that there exists an N ∈ N
n→∞ n→∞
1 1
such that if n > N , then |x|− < |nxn−1 | n < |x|+. It follows that |nxn−1 | = (|nxn−1 | n )n < (|x|+)n
for all n > N . 

Theorem F ∞ ∞
X X
The power series an (x − a)n converges on (a − R, a + R) if and only if the power series nan (x −
n=0 n=1
a)n−1 converges on (a − R, a + R).

∞ ∞  
X
n−1
X d n
Note that nan (x − a) = (an (x − a) ) .
n=1 n=0
dx

∞
X
Proof of Theorem F. We only prove the case where a = 0. So suppose that the series an x n
n=0
converges on (−R, R). Then it follows from Theorem A that the series is absolutely convergent on
this interval. Let x ∈ (−R, R). Take an  > 0 such that |x| < |x| +  < R (e.g.  = 21 (R − |x|)).
∞
X
Since |x| +  ∈ (−R, R), it follows that |an (|x| + )n | converges. It follows from Lemma E that
n=0
0 ≤ |nan xn−1 | ≤ |an |(|x| + )n for all n large enough, say for n ≥ N . Using the comparison test it
X∞ X∞
n−1
then follows that |nan x | converges, and hence that nan xn−1 converges.
n=1 n=1
∞
X
Conversely, suppose that the series nan xn−1 converges on (−R, R). Then it follows from
n=1
∞
X ∞
X ∞
X
n−1 n−1
Theorem A that |nan x | converges on (−R, R) and hence that |x| |nan x |= |nan xn |
n=1 n=1 n=1
converges on (−R, R). Since 0 ≤ |an xn | ≤ n|an xn | = |nan xn | for all n ∈ N, using the comparison
∞
X X∞
n
test it follows that |an x | converges on (−R, R), and hence that an xn converges on (−R, R). 
n=1 n=0

It follows from Theorem F that the radii of convergence of the series

∞ ∞ ∞
X
n
X d X d2
an (x − a) , (an (x − a)n ), 2
(an (x − a)n ), ...
n=0 n=0
dx n=0
dx

are all the same.

3
Theorem G (Theorem 2, p.794 in Stewart)
∞
X
If the power series cn (x − a)n has radius of convergence R > 0, then the function
n=0

∞
X
f (x) = cn (x − a)n
n=0

is differentiable (and hence continuous, and hence integrable) on the interval (a − R, a + R) and
∞
X
0
1. f (x) = ncn (x − a)n−1 ,
n=1

∞
(x − a)n+1
Z X
2. f (x) dx = C + cn .
n=0
n+1

Both the power series in (1) and (2) have radius of convergence R.

The equations in (1) and (2) can be rewritten as follows:

"∞ # ∞  
d X n
X d n
cn (x − a) = (cn (x − a) ) ,
dx n=0 n=0
dx
and Z "X
∞
# ∞ Z 
X
n n
cn (x − a) dx = cn (x − a) dx .
n=0 n=0
∞
X
Proof of (2). We only prove the case where a = 0. Since cn xn converges on (−R, R), it follows
n=0
∞
X
cn xn
from Theorem A that |cn xn | converges on (−R, R). If n ∈ N, then 0 ≤ n+1
= 1
|c xn |
n+1 n
≤
n=0
∞
X cn x n
|cn xn |, and hence, using the comparison test, it follows that converges on (−R, R). Hence
n=0
n + 1
∞ ∞ ∞
X cn xn+1 X cn x n X cn xn+1
= |x| also converges on (−R, R), so that converges on (−R, R).
n=0
n + 1 n=0
n + 1 n=0
n + 1
It follows from (1) that
∞
! ∞  ∞
d X cn xn+1 cn xn+1
  X
X d
= = cn xn = f (x),
dx n=0 n + 1 n=0
dx n + 1 n=0

∞
cn xn+1
Z X
so that f (x) dx = C + . 
n=0
n+1

Theorem H (Theorems 5 and 6, p.800 in Stewart)

If f has a power series representation around a, i.e.
∞
X
f (x) = cn (x − a)n for |x − a| < R
n=0

4
then
f (n) (a)
cn = ,
n!
so that
∞
X f (n) (a) f 0 (a) f 00 (a) f 000 (a)
f (x) = (x − a)n = f (a) + (x − a) + (x − a)2 + (x − a)3 + · · · .
n=0
n! 1! 2! 3!
Prove this. The series above is called the Taylor series of f around a.

If a = 0, then the series becomes:

∞
X f (n) (0) n f 0 (0) f 00 (0) 2 f 000 (0) 3
x = f (0) + x+ x + x + ···,
n=0
n! 1! 2! 3!
which is called the Maclaurin series of f .

Suppose that f (n) (a) exists for all n ∈ N. Let

f 0 (a) f 00 (a) f (n) (a)
Tn (x) = f (a) + (x − a) + (x − a)2 + · · · + (x − a)n .
1! 2! n!
Then Tn is called the n-th degree Taylor polynomial of f at a.

Theorem I (Theorem 8, p.801 in Stewart)

If f (x) = Tn (x) + Rn (x), where Tn is the n-th degree Taylor polynomial of f at a, and lim Rn (x) = 0
n→∞
for |x − a| < R, then f is equal to its Taylor series for |x − a| < R.

Prove this. Rn is called the remainder of the Taylor series.

Rolle’s Theorem
If f is continuous on [a, b] and differentiable on (a, b), and f (a) = f (b), then there exists a d ∈ (a, b)
such that f 0 (d) = 0.

Taylor’s Theorem
Suppose that I is an open interval and the (n + 1)-st derivative of f exists at each point of I. If
a, b ∈ I with a < b, then there exists a dn ∈ (a, b) such that
f 00 (a)
f (b) = f (a) + f 0 (a)(b − a) + (b − a)2 + · · ·
2!
f (n) (a) n f (n+1) (dn )
··· + (b − a) + (b − a)n+1 .
n! (n + 1)!
00 (n) (a)
Proof. Let L = f (b)−f (a)−f 0 (a)(b−a)− f 2!(a) (b−a)2 −· · ·− f n!
(b−a)n , and define kn = (n+1)!L
(b−a)n+1
.
Define φ : I → R by
f 00 (x)
φ(x) = f (b) − f (x) − f 0 (x)(b − x) − (b − x)2 − · · ·
2!
f (n) (x) kn
··· − (b − x)n − (b − x)n+1 .
n! (n + 1)!

5
 n+1
Then φ(a) = L − kn (b−a)
(n+1)!
= 0 and φ(b) = 0. Furthermore, since f (n+1) (x) exists for each x ∈ I, it
follows that each of the functions f, f 0 , f 00 , . . . , f (n) is continuous and differentiable on I, so that φ is
continuous and differentiable on I. In particular, φ is continuous on [a, b] and differentiable on (a, b).
Hence it follows from Rolle’s Theorem that there exists a dn ∈ (a, b) such that φ0 (dn ) = 0. But

f 000 (x)
φ0 (x) = −f 0 (x) + f 0 (x) − f 00 (x)(b − x) + f 00 (x)(b − x) − (b − x)2 + · · ·
2!
f (n) (x) f (n+1) (x) kn
··· + (b − x)n−1 − (b − x)n + (b − x)n
(n − 1)! n! n!
n
(b − x)
= (kn − f (n+1) (x)),
n!
so that
(b − dn )n
φ0 (dn ) = (kn − f (n+1) (dn )) = 0.
n!
(n+1)!L
Since a < dn < b, it follows that b − dn 6= 0, and consequently f (n+1) (dn ) = kn . Since kn = (b−a)n+1
,
f (n+1) (dn ) 00
it follows that L = (n+1)!
(b − a)n+1 , but since L = f (b) − f (a) − f 0 (a)(b − a) − f 2!(a) (b − a) − · · · −
2

f (n) (a)
n!
(b − a)n , the result follows. 

Corollary (Taylor’s Theorem)

Suppose that I is an open interval and the (n + 1)-st derivative of f exists at each point of I. If
a ∈ I, then there exists for any x 6= a in this interval a number dn properly between a and x such
that
f 00 (a) f (n) (a) f (n+1) (dn )
f (x) = f (a) + f 0 (a)(x − a) + (x − a)2 + · · · + (x − a)n + (x − a)n+1 .
2! n! (n + 1)!
(n+1)
Hence Rn (x) = f (n+1)!
(dn )
(x − a)n+1 . The latter is called the Lagrange Formula for the remainder of
the Taylor series of f .