MAL 100: Calculus

Lecture Notes

3 Infinite Series
3.1 Definitions & convergence
Definition 3.1.1. Let {an } be a sequence of real numbers.

a) An expression of the form

a1 + a2 + . . . + an + . . .
is called an infinite series.

b) The number an is called as the nth term of the series.

n
P
c) The sequence {sn }, defined by sn = ak , is called the sequence of partial sums of the
k=1
series.

d) If the sequence of partial sums converges to a limit L, we say that the series converges
and its sum is L.

e) If the sequence of partial sums does not converge, we say that the series diverges.

Examples 3.1.2.
∞ 1
xn converges to
P
1) If 0 < x < 1, then .
n=0 1−x
n
xk . Here
P
Solution. Let us consider the sequence of partial sums {sn }, where sn =
k=1

n
X 1 − xn+1 1 xn+1
sn = xk = = − , n ∈ N.
k=1
1−x 1−x 1−x

1
As, 0 < x < 1, xn+1 → 0 as n → ∞. Hence sn → . Thus the given series converges
1−x
1
to . ///
1−x

P∞ 1
2) The series diverges.
n=1 n

1
 n
1
P
Solution. Consider the sequence of partial sums {sn }, where sn = k
. Now, let us
k=1
examine the subsequence s2n of {sn }. Here

s2 = 1 + 1/2 = 3/2,
s4 = 1 + 1/2 + 1/3 + 1/4 > 3/2 + 1/4 + 1/4 = 2.

Suppose s2n > (n + 2)/2, then

2 n
X 1
s2n+1 = s2n +
k=1
2n +k
2n
n+2 X 1
> +
2 k=1
2n+1
n+2 2n (n + 1) + 2
= + n+1 = .
2 2 2
Thus the subsequence {s2n } is not bounded above and as it is also increasing, it diverges.
P∞ 1
Hence the sequence diverges, i.e., the series diverges. ///
n=1 n

∞
P 1
3) (Telescopic series:) Show that the series converges to 1.
n=1 n(n + 1)

Solution. Consider the sequence of partial sums {sn }. Then

n k  
X 1 X 1 1 1
sn = = − =1− → 1.
k=1
k(k + 1) k=1 k k+1 n+1

Summarizing this observation, one has the following theorem on Telescopic series

Theorem 3.1.3. Suppose {an } is a sequence of non-negative real numbers such that an → L.
P
Then the series (an − an+1 ) converges to a1 − L.

Lemma 3.1.4.
∞
P ∞
P ∞
P
1) If an converges to L and bn converges to M, then the series (an + bn ) converges
k=1 k=1 k=1
to L + M.
∞
P ∞
P
2) If an converges to L and if c ∈ R, then the series can converges to cL.
k=1 k=1

2
 ∞
P
Lemma 3.1.5. If an converges, then lim an = 0.
k=1 n→∞
∞
P
Proof. Suppose an = L. Then the sequence of partial sums {sn } also converges to L. Now
k=1

an = sn − sn−1 → L − L = 0. ///
∞
xn diverges.
P
Example 3.1.6. If x > 1, then the series
k=1
∞
xn converges. Then the nth term, i.e.,
P
Solution. Assume to the contrary that the series
k=1
xn → 0. But as x > 1, xn ≥ 1 for all n ∈ N and hence lim xn ≥ 1, which is a contradiction.
n→∞
∞
n
P
Hence the series x diverges. ///
k=1

As a first result we have the following comparison theorem:

P
Theorem 3.1.7. Let {an }, {bn } be sequences of positive reals such that an ≤ bn . If bn
P
converges then an converges.
P P
Proof. Let sn = a1 + a2 + .... + an and tn = b1 + b2 + .... + bn be the partial sum of an , b n
P
respectively. Then sn ≤ tn . Since bn converges, we have {tn } converges and is bounded. Now
since {sn } is monotonically increasing sequence that is bounded above, we get the convergence
P
of {sn } and hence the convergence of an . ///
∞
Theorem 3.1.8. Let {an }∞
P
1 be an decreasing sequence of positive numbers. Then an con-
n=1
∞
n
P
verges if and only if 2 a2n converges.
n=0

2n a2n respectively. Then

P P
Proof. Let sn and tn be the sequence of partial sums of an and
proof follows from the observation that
2n
X
s2n = an = a1 + a2 + (a3 + a4 ) + (a5 + a6 + a7 + a8 ) + .... + (a2n−1 +1 + .... + a2n )
k=1

≥ a1 + a2 + 2a4 + 4a8 + 8a16 + ....2n−1 a2n

1
= a1 + tn . (3.1)
2
Therefore, if {sn } converges then {s2n } converges and by 3.1, {tn } converges.
On the other hand,

s2n −1 = a1 + (a2 + a3 ) + (a4 + a5 + a6 + a7 ) + (a8 + ....a17 ) + (a2n−1 + .... + a2n −1 )

≤ a1 + 2a2 + 4a4 + 8a8 + .... + 2n−1 a2n−1 = a1 + tn−1

3
So if {tn } converges, then {s2n −1 } converges. Now the conclusion follows from sn ≤ s2n+1 −1 and
the fact that {sn } is monotonically increasing sequence. ///

Examples 3.1.9.
∞ 1 ∞ ∞
n 1 1
P P P
1) Consider the series p
, p > 0. Then, we have 2 n p
= n p−1
which con-
n=1 n n=1 (2 ) n=1 (2 )
verges for p > 1 and diverges for p ≤ 1.
∞ 1 ∞ 1 1 P ∞ 1
2n n
P P
2) Consider the series . Here = which diverges. Hence
n=2 n log n n=2 2 log 2n log 2 n=2 n
the given series diverges.

3.2 Absolute convergence

∞
P ∞
P
Definition 3.2.1. a) Let an be a series of real numbers. If |an | converges, we say that
n=1 n=1
∞
P
an converges absolutely.
n=1

∞
P ∞
P ∞
P
b) If an converges but |an | diverges, we say that an converges conditionally.
n=1 n=1 n=1

Examples 3.2.2.
P∞ (−1)n
1) The series converges absolutely.
n=1 n!

(−1)n
2) The series ∞
P
n=1 n2 converges absolutely.
∞
P ∞
P
Theorem 3.2.3. If an converges absolutely, then an converges.
n=1 n=1
n
|ak |. As the series converges absolutely, the sequence {tn }∞
P
Proof. Let tn = 1 is Cauchy. Thus,
k=1
given  > 0, there exists N ∈ N such that

|tm − tn | <  ∀ m, n ≥ N.

Let m > n. Then  

Xm  m
X
|sm − sn | =  ai  ≤ |ai | = |tm − tn | < .
 
 
i=n+1 i=n+1
∞
Thus the sequence {sn }∞
P
1 is Cauchy and hence converges. Thus an converges. ///
1

4
 ∞
P
Theorem 3.2.4. Let an be a series of real numbers. Let pn = max{an , 0} and qn =
1
min{an , 0}.
P P P
a) If an converges absolutely, then both pn and qn converges.
P P P
b) If an converges conditionally then both pn and qn diverges.

Proof.

a) Observe that pn = (an + |an |)/2 and qn = (an − |an |)/2. Thus the convergence of the two
series follows from the hypothesis.

b) From the observation that pn = (an + |an |)/2, we have |an | = 2pn − an . As the series
P P P P
an converges and |an | diverges, the series pn diverges. Similarly, the series qn
diverges. ///

Tests for absolute convergence

P
Theorem 3.2.5 (Comparison test). Let an be a series of real numbers.
P P
a) an converges if there is an absolutely convergent series cn with |an | ≤ |cn | for all
n ≥ N, N ∈ N.
P P
b) an diverges if there is a divergent series cn of nonnegative terms, with cn ≥ |an | for
all n ≥ N, N ∈ N.

Proof follows as in Theorem 3.1.7

Examples 3.2.6.
∞
P 7 7 1 1 P1
1) The series diverges because = ≥ for all n ∈ N and
n=1 7n − 2 7n − 2 n − 2/7 n n
diverges.
P∞ 1 1 1 P∞ 1
2) The series converges because ≤ n and n
converges.
n=0 n! n! 2 n=0 2

Theorem 3.2.7 (Limit comparison test). Let {an } and {bn } be two sequences of positive num-
bers. Then
an P P
a) if lim = c > 0, an and bn both converge or diverge together;
n→∞ bn

an P P
b) if lim = 0 and bn converges absolutely, an converges absolutely.
n→∞ bn

an P P
c) if lim = ∞ and |bn | diverges, an diverges.
n→∞ bn

5
 an c
Proof. (a) As lim = c > 0, for  = 2
> 0, there exists N ∈ N such that
n→∞ bn

 
 an  c
n ≥ N =⇒  − c < .
bn 2

Thus, for n ≥ N,
−c an c
≤ −c≤
2 bn 2
or equivalently
cbn 3cbn
≤ an ≤ .
2 2
Hence the conclusion follows from the comparison test.
an
b) Given that lim = 0. Hence for  = 21 , there exists N ∈ N such that
n→∞ bn

an 1
n ≥ N =⇒ <
bn 2

or equivalently,
bn
n ≥ N =⇒ an ≤
.
2
Thus the desired conclusion follows from the comparison test.
an
c) Here we are given that lim = ∞. Hence for any real number M > 0, there exists
n→∞ bn
N ∈ N such that
an
n ≥ N =⇒ ≥M
bn
or equivalently,
n ≥ N =⇒ an ≥ M bn .
P P
Thus if |bn | diverges, then |an | diverges by comparison test. ///

Examples 3.2.8.
 
2n + 1
P∞ 2n + 1 2n + 1 1 an (n + 1)2
1) Consider the series . Here an = . Let bn = . Then = =
n=1 (n + 1)
2 (n + 1) 2 n bn 1
n
2n2 + n P1
→ 2 as n → ∞. Further, n
diverges. Thus by limit comparison theorem,
n2 + 2n + 1
the given series diverges.
∞
P 1 1 1 an 2n
2) Consider the series . Here a n = . Let b n = . Then = → 1.
1 2n − 1 2n − 1 2n bn 2n − 1
P 1
Further, converges and hence the given series converges.
2n

6
 P e−n −n
3) Consider the series n2
. Here an = en2 and bn = n12 . Then an
bn
= e−n → 0 as n → ∞.
P 1
Further, n2
converges and hence the given series converges.
∞
P
Theorem 3.2.9 (Ratio test). Let an be a series of real numbers. Let
1
   
 an+1   an+1 
a = lim inf   and A = lim sup 
 an  .

n→∞ an  n→∞

Then
∞
P
a) an converges absolutely if A < 1;
1

∞
P
b) an diverges if a > 1;
1

c) the test fails if a < 1 < A.

Proof. a) If A < 1, choose B such thata

 A < B < 1. Then there exists an  > 0 such that
 an+1 
B = A +  and also N ∈ N such that  an  ≤ B for all n ≥ N. Further, for any k ∈ N,

  k   k
 aN +k  Y  aN +i  Y

 aN 
=  ≤
 aN +i−1  B = Bk.
i=1 i=1

∞
Thus |aN +k | ≤ B k |aN |, k ∈ N. But |aN |B k < ∞ as B < 1. Thus by comparison test, the
P
k=0
∞
P
series an converges.
1  
b) If a > 1, choose b such that 1 < b < a. There exits N ∈ N such that  an+1 ≥ b for all
 
an 
n ≥ N. Further, for any k ∈ N,
  k   k
 aN +k  Y  aN +i  Y

 aN 
=  ≥
 aN +i−1  b = bk .
i=1 i=1

∞
aN bk diverges. Thus, again, by the comparison
P
Thus |aN +k | ≥ |aN |, k ∈ N. But, as b > 1,
k=0
∞
P
test, the series an diverges.
1
P1 an+1 P1 P 1
c) Consider the series . Here lim = 1. But diverges. For the series ,
n n→∞ an n n2
an+1
which converges, again lim = 1. ///
n→∞ an

Examples 3.2.10.

7
 ∞ nn
P
a) Consider the series . Here
1 n!

n n
(n + 1)n+1 n!
 
an+1 n+1 1
= = = 1+ → e,
an (n + 1)! nn n n

which is greater than 1. So a = A = e > 1. Thus the given series diverges.

∞ xn
P
b) Consider the series , x ∈ R. Here
0 n!

an+1 xn+1 n! x
= n
= → 0.
an (n + 1)! x n+1

Therefore a = A = 0 < 1. Thus, for all x ∈ R, the given series converges.

∞
P p
Theorem 3.2.11 (Root test). Let an be a series of real numbers. Let A = lim sup n |an |.
1 n→∞
Then

a) the series converges absolutely if A < 1;

b) the series diverges if A > 1;

c) the test fails if A = 1.

Proof.
p a) If A < 1, choose B such that A < B < 1. Then there exists N ∈ N such that
n
|an | < B for all n ≥ N. This implies |an | < B n for all n ≥ N. As B < 1, the series converges
by comparison test. p
b) If A > 1, there exists infinitely many n ∈ N such that n |an | > 1. But this implies that
P
|an | > 1 for infinitely many values of n and hence aN 9 0, i.e., an diverges.
P1
c) Consider the series . Here A = 1 and the series diverges. On the other hand, for the
n
P 1
series , again A = 1, but the series converges. ///
n2
Examples 3.2.12.
s  
∞ xn xn  xn   x 

P
, x ∈ R. Here an =  n  = √
1) Consider the series . Therefore, n     → 0. Thus
1 n n n
n

the series converges for |x| < 1 and diverges for |x| > 1.
∞ xn
P xn p x
2) Consider the series , x ∈ R. Here an = n . Then, |an | =   → 0. Thus the series
n  
n
1 n n n
converges for x ∈ R.

8
  n
 n is odd √
3) Consider the series
P
an , where an = 2n . Then lim sup n an = 21 . There-
1
 n is even n→∞
2n
fore the series converges.

Alternating series:
Definition 3.2.13. An alternating series is an infinite series whose terms alternate in sign.
Theorem 3.2.14. Suppose {an } is a sequence of positive numbers such that

(a) an ≥ an+1 for all n ∈ N and

(b) lim an = 0,
n→∞

∞
(−1)n+1 an converges.
P
then the alternating series
n=1

Proof. Consider the partial sums with odd index, s1 , s3 , s5 , . . . . Now, for any n ∈ N,

s2n+1 = s2n−1 − a2n + a2n+1 ≤ s2n−1 (by (a)).

Thus the sequence {s2n−1 }∞

1 forms a non-increasing sequence. Also, notice that

n−1
X
s2n−1 = (a2i−1 − a2i ) + a2n−1 .
i=1

Since each quantity in the parenthesis is non-negative and a2n−1 > 0, the sequence {s2n−1 } is
bounded below by 0. Hence {s2n−1 }∞1 is convergent.
Now, consider the partial sums with even index, s2 , s4 , s6 , . . . . For any n ∈ N,

s2n+2 = s2n + a2n+1 − a2n+2 ≥ s2n (by (a)).

Thus the sequence {s2n }∞

1 forms a non-decreasing sequence. Further,

n−1
X
s2n = a1 − (a2i − a2i+1 ) − a2n ≤ a1 ,
i=1

which means that s2n is bounded above by a1 . Therefore, {s2n } is convergent.

Let L = lim s2n and M = lim S2n−1 . By ((b)),

0 = lim a2n = lim (s2n − s2n−1 ) = L − M.

∞
(−1)n+1 an converges.
P
Thus L = M and hence the alternating series ///
n=1

9
Examples 3.2.15.
∞
(−1)n+1 21/n . Here an = 21/n → 1 as n → ∞. Hence the above
P
1) Consider the series
n=1
theorem does not apply. Anyhow, one can show that the series diverges.
∞ (−1)n+1
. The a0n s of this series satisfies the hypothesis of the above
P
2) Consider the series
n=1 n
theorem and hence the series converges.

Examples 3.2.16.
P∞ (−1)n+1
1) The series converges conditionally.
n=1 n
P∞ (−1)2n−1
2) The series converges conditionally.
n=1 2n − 1

3.3 Error estimation in Infinite Series

Here we discuss the methods of estimating the sum of a infinite series which converges by the
tests discussed in the previous sections. First we have the series of positive real numbers.
an+1
Theorem 3.3.1. Suppose {an } is a positive decreasing sequence and lim = L < 1.
n→∞ an
(a) If an+1
an
decreases to the limit L, then
 
L an+1
an < S − Sn <
1−L 1 − an+1
an

an+1
(b) If an
increases to the limit L, then
 
an+1 L
< S − Sn < an .
1 − an+1
an
1−L
an+1
Proof. It is obvious from the statement that the series converges by Ratio test. Let r = an
< 1.
Then ak+1
ak
< r for all k ≥ n. Hence

an+k < an+k−1 r < an+k−2 r2 < an+k−3 r3 ..... < an rk

Therefore,
∞ ∞ ∞
X X X an r
S − Sn = ak = an+k < an r k =
k=n+1 k=1 k=1
1−r

10
Now substituting the value of r we get the right hand estimate in (a). For the left hand side
inequality, using L < an+1
an
, ∀k ≥ n, we see
∞ ∞
X X L
S − Sn = an+k > an L k = an .
k=1 k=1
1−L

A similar argument with L and r reversed proves (b).

Remark: Let An = 1−an+1
an+1 and Bn =
Lan
1−L
= 0. Suppose (An − Bn )/2 ≤ , then the error
an
S − Sn < .
Example: Consider the series S = ∞ 1
P
n=1 n2 5n
. Then
2
n2 5n

an+1 n 1
= =
an (n + 1)2 5n+1 n+1 5

increases to 15 . So by the above inequality

1
(n+1)2 5n+1 1
< S − Sn < .
1 − 15 (n/n + 1)2 4n2 5n

That is
1 1
< S − S n < .
(4n2 + 10n + 5)5n 4n2 5n
so if we take n = 5, we get 0.2110037 < S < 0.2110049.
Problem : Find the sum of the series ∞ n2 −6
P
n=1 n! with error < 10 .
Here An = n!(1−n+1
n+1
)
and Bn = 0. By trial and error we can see that (An − Bn )/2 ≤ 10−6 if
n2
n ≥ 11.
Theorem 3.3.2. Consider the series ∞ n+1
P
n=1 (−1) an such that {an } positive decreasing se-
quence that converges to 0 and bn = an − an+1 also decreases to 0. Then
an+1 an
< |S − Sn | < .
2 2
Proof. Note that

S = Sn + (−1)n (bn+1 + bn+3 + ....) or S = Sn−1 + (−1)n+1 (bn + bn+2 + ...).

Since {bn } is decreasing, we obtain

|S − Sn | = bn+1 + bn+3 + .... < bn + bn+2 + ... = |S − Sn−1 |.

11
Similarly, |S − Sn+1 | < |S − Sn |. But S lies between two successive partial sums, so

an = |Sn − Sn−1 | = |S − Sn | + |S − Sn−1 | > 2|S − Sn |

and
an+1 = |Sn+1 − Sn | = |S − Sn+1 | + |S − Sn | < 2|S − Sn |.
These two estimates immediately gives the required estimate.
4
Problem: Find the sum of the series S = (−1)n+1 2n−1
P
with error less than 0.0001.
Solution: It is easy to check bn = an − an+1 is decreasing!. So for odd n, the above inequality
implies that
2 2
Sn − < S < Sn − (3.2)
2n − 1 2n + 1
4
By the remark above, typically n is such that an+1 = 2n+1
< 0.0001. This implies that we need
20, 000 terms.

3.4 Power series & Taylor Series revisited

∞
Given a sequence of real numbers {an }∞ an (x − c)n is called power series with
P
n=0 , the series
n=0
center c. It is easy to see that a power series converges for x = c. Power series is a function
of x provided it converges for x. If a power series converges, then the domain of convergence
is either a bounded interval or the whole of IR. So it is natural to study the largest interval
where the power series converges.
an xn converges at x = r, then an xn converges for |x| < |r|.
P P
Remark: If
Proof: We can find C > 0 such that |an xn | ≤ C for all n. Then
x x
|an xn | ≤ |an rn || |n ≤ C| |n .
r r
Conclusion follows from comparison theorem.
∞ p 1
an xn . Suppose β = lim sup n |an | and R =
P
Theorem 3.4.1. Consider the power series β
n=0
(We define R = 0 if β = ∞ and R = ∞ if β = 0). Then
∞
an xn converges for |x| < R
P
1.
n=0
∞
an xn diverges for |x| > R.
P
2.
n=0

3. No conclusion if |x| = R.
Proof. Proof
p of (i) follows from the root test. For a proof, take αn (x) = an xn and α =
lim sup |αn |. For (ii), one can show that if |x| > R, then there exists a subsequence {an } such
n

12
that an 6→ 0. Notice that α = β|x|. For (iii), observe as earlier that the series with an = n1 and
bn = n12 will have R = 1.
Similarly, we can prove:
∞  
an xn . Suppose β = lim sup  an+1 and R = β1
P
Theorem 3.4.2. Consider the power series
 
an 
n=0
(We define R = 0 if β = ∞ and R = ∞ if β = 0). Then
∞
an xn converges for |x| < R
P
1.
n=0

∞
an xn diverges for |x| > R.
P
2.
n=0

3. No conclusion if |x| = R.

Definition 3.4.3. The real number R in the above theorems is called the Radius of convergence
of power series.
P xn P n P −n 3n
Examples: Find the interval of convergence of (i) n
(ii) xn! (iii) 2 x

1. β = lim sup | an+1

an
| = 1, and we know that the series does not converge for x = −1, 1. So
the interval of convegence is (−1, 1).

2. β = lim sup | an+1

an
| = 0. Hence the series converges everywhere.
P −n 3 n P −n n
3. To see the subsequent non-zero
p terms, we write the series as 2 (x ) = 2 y . For
−1 −1/3
this series βy = lim sup |an | = 2 . Therefore, βx = 2
n
and R = 21/3 .

Taylor Series Revisited:

The following theorem is very useful in identifying the domain of convergence of some Taylor
series.
∞
an x n
P
Theorem 3.4.4. (Term by term differentiation and integration): Suppose f (x) =
n=0
converges for |x| < R. Then
∞
nan xn−1 converges in |x| < R and is equal to f 0 (x).
P
1.
n=0

∞
1
R
a xn+1
P
2. n+1 n
converges in |x| < R and is equal to f (x)dx.
n=0

From this theorem one concludes that a power series is infinitely differentiable with in its radius
of convergence. Now it is natural to ask weather this series coincides with the Taylor series of
the resultant function. The answer is yes and it is simple to prove that if f (x) = an xn , then
P
(n)
an = fn! . The above theorem is useful to find the domain of convergence of Taylor series of
some functions.

13
Example: The Taylor series of f (x) = tan−1 x and a domain of its convergence.
Z Z
−1 dx
tan x = = 1 − x2 + x4 + ...
1 + x2
x 3 x5
=x − + + ...
3 5
Taking x = 1 we get interesting sum

1 1 1 π
1− + − + .... = tan−1 (1) = .
3 5 7 4
Though the function tan−1 x is defined on all of IR, we see that the power series converges on
(−1, 1). We can apply Abel’s theorem on alternating series to show that the series converges
at x = 1, −1.
For approximation, we can use the error approximation of alternating series discussed in
the previous section. The total error if we approximate tan−1 x by sn (x), then the maximum
n+1
error is |x|n+1 . ///
We note that the power series may converge to a function on small interval, even though
the function is defined on a much bigger interval. For example the function log(1 + x) has
power series that converges on (−1, 1), but log(1 + x) is defined on (−1, ∞). The domain
of convergence of power series is symmetric about the centre but the domain of definiton of
function. For instance, for a function defined on (−1, 3) the radius of convergence of its power
series(about 0) cannot be more than 1.
Another interesting application is to integrate the functions for which we have no ”clue”. For
example, Z x Z x
−t2 t2 t4 x3 x5
(1) erf (x) = e dt = (1 − + + ... = x − + + ...
0 0 1! 2! 3 10
Z x Z x
sin t t2 t4 x3 x5
(2) = 1− + =x− + + ...
0 t 0 3! 5! 3! · 3 5! · 5

References
[1] Methods of Real Analysis, Chapter 2, R. Goldberg.

[2] Elementary Analysis: The Theory of Calculus, K. A. Ross.

[3] Calculating Sums of Infinite Series, The American Mathematical Monthly, Bart Braden,
Vol.99, 7, (1992), 649-655.

14