Department of Mathematics

Indian Institute of Technology Guwahati

MA 101: Mathematics I
Series of real numbers
July-December 2019

Let (an ) be a sequence of real numbers. An expression of the form

a1 + a2 + · · · + an + · · ·
∞
X
is called an infinite series. We use the notation: an .
n=1

(1) The number an is called the n-th term of the series.

n
X
(2) sn = ak is called the n-th partial sum of the series.
k=1

(3) If the sequence of partial sums (sn ) converges to a limit `, we say that the series
converges and its sum is `.

(4) If (sn ) diverges, we say that the series diverges.

Example 1. We have
∞
arn−1 (where a 6= 0) converges if and only if |r| < 1.
P
(1) The geometric series
n=1

∞
1
P
(2) The series n(n+1)
is convergent with sum 1.
n=1

(3) The series 1 − 1 + 1 − 1 + · · · is not convergent.

∞
X 1
(4) The series diverges.
n=1
n
∞
P ∞
P
Theorem 1 (Algebraic operations on series). Let xn and yn be convergent with
n=1 n=1
sums x and y respectively. Then
∞
P
(a) (xn + yn ) is convergent with sum x + y
n=1

∞
P
(b) αxn is convergent with sum αx, where α ∈ R
n=1

∞
X
Theorem 2 (Necessary condition for convergence). If an converges then lim an = 0.
n→∞
n=1
 n
X ∞
X
Proof. Let sn = ak . Then an = sn − sn−1 . Since an converges, so sn and sn−1 will
k=1 n=1
converge to the same limit, and hence an → 0.
∞
P
Hence if xn 6→ 0, then xn cannot be convergent.
n=1

∞
X
Remark 1. The condition lim an = 0 is not sufficient for the convergence of an .
n→∞
n=1
∞
1 X 1
For example, → 0 but diverges.
n n=1
n

Example 2. The following series are not convergent.

∞ ∞
n2 +1 n
(−1)n n+2
P P
(a) (n+3)(n+4)
(b) .
n=1 n=1

∞
P
Theorem 3 (Monotone criterion). A series xn of non-negative terms is convergent if
n=1
and only if the sequence (sn ) is bounded above.

Proof. Since xn ≥ 0, so the sequence (sn ) of partial sums is increasing. By Monotone

Convergence Theorem, (sn ) is convergent if and only if it is bounded above. Equivalently,
∞
P
xn is convergent if and only if (sn ) is bounded above.
n=1

∞
1
P
Example 3. (a) n2
is convergent.
n=1

Solution. We have
1 1 1
sn = 1 + + + · · · +
22 32 n2
1 1 1
≤1+ + + ··· +
1·2 2·3 (n − 1) · n
     
1 1 1 1 1
=1+ 1− + − + ··· + −
2 2 3 n−1 n
1
=2−
n
< 2.
∞
1
P
Thus, (sn ) is bounded above. Hence the sereis n2
is convergent.
n=1

∞
1
P
(b) n
is divergent.
n=1

n ∞
1 1
P P
Solution. We know that sn = n
is not bounded above. Hence n
is divergent.
k=1 n=1
 ∞
P
Theorem 4 (Cauchy criterion). A series xn is convergent if and only if for each ε > 0,
n=1
there exists n0 ∈ N such that

|xn+1 + · · · + xm | < ε for all m > n ≥ n0 .

Proof. We know that a sequence in R is convergent if and only if it is a Cauchy sequence.

P∞
Hence, the series xn is convergent if and only if the sequence (sn ) of partial sums is
n=1P
Cauchy. Now, sn = nk=1 xn is Cauchy if for given ε > 0, there is n0 ∈ N such that for
all m > n ≥ n0 ,
|sm − sn | = |xn+1 + · · · + xm | < ε.
This completes the proof of the theorem.
Tests for convergence:

Theorem 5 (Comparison test). Let (xn ) and (yn ) be sequences in R such that for some
n0 ∈ N, 0 ≤ xn ≤ yn for all n ≥ n0 . Then
∞
P ∞
P
(a) yn is convergent ⇒ xn is convergent.
n=1 n=1

∞
P ∞
P
(b) xn is divergent ⇒ yn is divergent.
n=1 n=1

Proof. Clearly, (a) ⇔ (b). So, we prove (a).

∞
P
(a) Suppose that yn is convergent. Then for given ε > 0, there is some n1 ∈ N
n=1
such that |yn+1 + yn+2 + · · · + ym | < ε for all m > n ≥ n1 . We also have 0 ≤ xn ≤ yn for
all n ≥ n0 . Let n2 = max{n0 , n1 }. Then for m > n ≥ n2 , we have

|xn+1 + xn+2 + · · · + xm | = xn+1 + xn+2 + · · · + xm

≤ yn+1 + yn+2 + · · · + ym
= |yn+1 + yn+2 + · · · + ym |
< ε.
∞
P
By Cauchy’s criterion, the sereis xn is convergent.
n=1
Alternative proof: Without loss of generality, we may assume that 0 ≤ xn ≤ yn for all
n ≥ 1 as we can always add or remove finitely manyPterms without affectingPn the nature
n
(convergence/divergence) of the series. Let sn = k=1 xn and tn = k=1 yn . Since
P∞
yn ≥ xn ≥ 0 for all n, so (sn ) and (tn ) are increasing sequences and sn ≤ tn . If yn is
n=1
convergent, then (tn ) is convergent and hence bounded above. This implies that (sn ) is
∞
P
bounded above, and since (sn ) is increasing so it is convergent. This proves that xn is
n=1
convergent. This completes the proof of (a).
∞
1+sin n
P
Example 4. (a) 1+n2
is convergent.
n=1
 1+sin n 2
P∞ 1
Solution. We have 0 ≤ 1+n2
≤ n2
for all n ∈ N and n=1 n2 converges.
∞
√ 1
P
(b) is not convergent.
n=2 n(n−1)

P∞
Solution. We have √ 1
> 1
n
> 0 for all n ≥ 2 and 1
n=1 n diverges.
n(n−1)

Theorem 6 (Limit comparison test). Let (xn ) and (yn ) be sequences of positive real
numbers such that xynn → ` ∈ R.
∞
P ∞
P
(a) If ` 6= 0, then xn is convergent if and only if yn is convergent.
n=1 n=1
∞
P ∞
P
(b) If ` = 0, then yn is convergent ⇒ xn is convergent.
n=1 n=1
∞
P ∞
P
(c) If ` = ∞ and yn is divergent ⇒ xn diverges.
n=1 n=1

Proof. Note that ` ≥ 0.

(a) If ` 6= 0, then ε = 2` > 0. Since xynn → `, so there is some n0 ∈ N such that 2` < xynn < 3`2
for all n ≥ n0 . This implies 2` yn < xn < 3`2 yn for all n ≥ n0 . Applying the Comparison
Test twice, we complete the proof.
(b) Let ε > 0. Since xynn → 0, so there is some n0 ∈ N such that 0 < xynn < ε for all n ≥ n0 .
This implies 0 < xn < ε · yn for all n ≥ n0 . We now complete the proof by applying
the Comparison Test.
(c) If xynn diverges to ∞, then for given M > 0 there is some n0 such that xynn > M for all
n ≥ n0 . This is, xn > M · yn for all n ≥ n0 . Now applying the Comparison Test, we
complete the proof.

∞
n
P
Example 5. 4n3 −2
is convergent.
n=1
n 1 xn n3
Solution. Let xn = 4n3 −2
and yn = n2
for all n ∈ N. Then lim = lim 3 =
n→∞ yn n→∞ 4n −2
lim 1 2 = 1
4
6= 0. By Limit Comparison Test, the given series is convergent.
n→∞ 4− n3

Theorem 7 (Cauchy’s condensation test). Let (xn ) be a decreasing sequence of nonneg-

∞ ∞
2n x2n is convergent.
P P
ative real numbers. Then xn is convergent if and only if
n=1 n=1
Pn Pn k
Proof. Let sn = k=1 xk and tn = k=1 2 x2k . Since (xn ) is decreasing, we have
X2n
s2n = xk
k=1
= x1 + x2 + (x3 + x4 ) + (x5 + x6 + x7 + x8 ) + · · · + (x2n−1 +1 + x2n−1 +2 + · · · + x2n )
≥ x1 + x2 + 2x4 + 4x8 + · · · + 2n−1 x2n
1 tn
= x1 + (2x2 + 22 x4 + 23 x8 + · · · + 2n x2n ) = x1 + .
2 2
 ∞
P
Now, if xn is convergent, then (sn ) is convergent. This implies that (s2n ) is bounded
n=1
∞
2n x2n is conver-
P
and hence (tn ) is bounded above. By Monotone criterion, the series
n=1
gent. On the other hand,
2n
X
s2 n = xk
k=1
= x1 + (x2 + x3 ) + (x4 + x5 + x6 + x7 ) + · · · + (x2n−1 + x2n−1 +1 + · · · + x2n −1 ) + x2n
≤ x1 + 2x2 + 4x4 + 8x8 + · · · + 2n−1 x2n−1 + x2n
≤ x1 + 2x2 + 4x4 + 8x8 + · · · + 2n−1 x2n−1 + 2n x2n
= x1 + tn .
∞
2n x2n is convergent then (tn ) is bounded and hence (s2n ) is bounded above.
P
Now, if
n=1
Since (sn ) is increasing, so sn ≤ s2n for all n ≥ 1. This proves that (sn ) is bounded above
P∞
and by Monotone criterion, the series xn is convergent.
n=1

Example 6. We have
∞
1
P
(a) p-series: np
is convergent if and only if p > 1.
n=1

Solution. If p ≤ 0, then clearly the series diverges as n1p 6→ 0. Let p > 0. Then
the sequence (1/np ) of non negative terms is decreasing. By Cauchy’s condensation
∞ ∞
1
2n (2n1 )p is convergent. Now, the geometric
P P
test, np
is convergent if and only if
n=1 n=1
∞ ∞
n
n 1 1 1
P P
series 2 (2n )p
= 2p−1
is convergent if and only if 2p−1
< 1, that is, if and
n=1 n=1
only if p > 1.
∞
1
P
(b) n(log n)p
is convergent if and only if p > 1.
n=2

1
Solution. Let f (x) = x(log x)p
for all x > 1. Then f : (1, ∞) → R is differentiable
p−1 (log x+p)
0
and f (x) = − (log x)
≤ 0 for all x > max{1, e−p } = a (say). Hence f is
x2 (log x)2p
decreasing on (a, ∞) and so f (n + 1) ≤f (n) for all n ≥ n0 , where n0 ∈ N is chosen
∞
1
to satisfy n0 > a. Thus the sequence n(log n)p
of non-negative real numbers
n=n0
∞ ∞
1 1
2n ·
P P
is decreasing. Since the series 2n (log 2n )p
= (log 2)p np
is convergent if and
n=n0 n=n0
∞
1
P
only if p > 1, by Cauchy’s condensation test, n(log n)p
is convergent if and only
n=n0
if p > 1. Consequently the given series is convergent if and only if p > 1.

Alternating series: An alternating series is an infinite series whose terms alternate in

sign.
 ∞ ∞
X X 1
Example 7. (a) (−1)n (b) (−1)n .
n=1 n=1
n

Theorem 8 (Leibniz’s test). Let (xn ) be a decreasing sequence of positive real numbers
∞
(−1)n+1 xn is convergent.
P
such that xn → 0. Then the alternating series
n=1

Proof. Since (xn ) is decreasing, we have

2n+1
X
s2n+1 = (−1)k+1 xk = x1 − x2 + x3 − · · · + x2n−1 − x2n + x2n+1
k=1
= s2n−1 − x2n + x2n+1 = s2n−1 − (x2n − x2n+1 ) ≤ s2n−1 .

Hence, the subsequence (s2n+1 ) is decreasing. Since (xn ) is decreasing and xn ’s are posi-
tive, we have

s2n+1 = x1 − x2 + x3 − · · · + x2n−1 − x2n + x2n+1

= (x1 − x2 ) + (x3 − x4 ) + · · · + (x2n−1 − x2n ) + x2n+1 ≥ 0.

Thus, the subsequence (s2n+1 ) is bounded below. Hence, it is convergent.

Similarly,

s2n+2 = x1 − x2 + x3 − · · · + x2n−1 − x2n + x2n+1 − x2n+2

= s2n + (x2n+1 − x2n+2 ) ≥ s2n

and

s2n = x1 − x2 + x3 − · · · − x2n−2 + x2n−1 − x2n

= x1 − (x2 − x3 ) − (x4 − x5 ) − · · · − (x2n−2 − x2n−1 ) − x2n ≤ x1 .

Thus, the subsequence (s2n ) is increasing and bounded above.

Let s2n−1 → `1 and s2n → `2 . We have x2n = s2n−1 − s2n . Since xn → 0, so `1 = `2 . This
proves that the subsequences (s2n−1 ) and (s2n ) converge to the same limit, and hence (sn )
∞
(−1)n+1 xn is convergent.
P
is convergent. That is, the alternating series
n=1

∞
X 1
Example 8. By Leibniz’s test, the alternating harmonic series (−1)n+1 converges.
n=1
n
∞
P ∞
P ∞
P
Definition 1. xn is called absolutely convergent if |xn | is convergent. xn is
n=1 n=1 n=1
∞
P ∞
P
called conditionally convergent if xn is convergent but |xn | is divergent.
n=1 n=1

∞ ∞ ∞
n1 1 1
X X X
Example 9. The series (−1) converges. But diverges. Hence, (−1)n
n=1
n n=1
n n=1
n
converges conditionally.
Theorem 9. Every absolutely convergent series is convergent.
Proof. Let ∞
P Pn
n=1 xn be absolutely convergent. Let tn = k=1 |xk |. Then (tn ) is convergent
and hence Cauchy. Let Pn ε > 0. Then there is some n 0 ∈ N such that |tm − tn | < ε for all
m, n ≥ n0 . Let sn = k=1 xk . Now, for m > n ≥ n0 ,

|sm − sn | = |xm + xm−1 + · · · + xn+1 |

≤ |xm | + |xm−1 | + · · · + |xn+1 |
= tm − tn
= |tm − tn |
< ε.

Hence, (sn ) is Cauchy and so it is convergent. Equivalently, ∞

P
n=1 xn is convergent.

Theorem 10 (Comparison test-II). Let (xn ) be a series of realPnumbers. Then ∞

P
n=1 xn
∞
converges absolutely if there is an absolutely convergent series n=1 yn and some n0 ∈ N
satisfying |xn | ≤ |yn | for all n ≥ n0 .

Theorem 11 (Limit  comparison

 test-II). Let (xn ) and (yn ) be sequences of nonzero real
 xn 
numbers such that   → ` ∈ R.
yn
∞
P ∞
P
(a) If ` 6= 0, then xn is absolutely convergent iff yn is absolutely convergent.
n=1 n=1

∞
P ∞
P
(b) If ` = 0, then yn is absolutely convergent ⇒ xn is absolutely convergent.
n=1 n=1

∞
X
Theorem 12 (Ratio Test). Let xn be a series of nonzero real numbers. Let
n=1

xn+1 xn+1
a = lim inf | | and A = lim sup | |.
xn xn
Then
∞
P
(1) If A < 1, then xn is absolutely convergent.
n=1

∞
P
(2) If a > 1, then xn is divergent.
n=1

Proof. (1) Let A < 1. Let B ∈ R be such that A < B < 1. Put ε = B − A. We have
xn+1
A = lim sup | | = lim yn ,
xn n→∞

n o
|xk+1 |
where yn = sup |xk |
: k ≥ n . Since ε = B − A > 0, so there is some n0 ∈ N such that

|xn+1 |
0 < yn < A + ε = B for all n ≥ n0 , that is, < B for all n ≥ n0 .
|xn |
 ∞
X ∞
X
This yields |xn0 +k | < |xn0 |B k for all k ≥ 1. Since 0 < B < 1, so |xn0 |B k = |xn0 | Bk
k=1 k=1
∞
X
is convergent. Therefore, by Comparison Test, |xn0 +k | is also convergent. This proves
k=1
∞
X
that xn converges absolutely.
n=1

(2) Suppose that a = lim inf | xxn+1

n
| > 1. Let b ∈ R such that 1 < b < a. Let ε = a − b.
We have
xn+1
a = lim inf | | = lim zn ,
xn n→∞
n o
where zn = inf |x|xk+1
k|
|
: k ≥ n . Since ε = a − b > 0, so there is some n0 ∈ N such that

|xn+1 |
a − ε < zn for all n ≥ n0 , that is, > b for all n ≥ n0 .
|xn |

This yields |xn0 +k | > |xn0 |bk > |xn0 | for all k ≥ 1. Thus, lim |xn | ≥ |xn0 |. This proves
n→∞
∞
P
that xn 6→ 0, and therefore xn is divergent.
n=1

Remark 2. If | xxn+1
n
| → `, then a = A = `.

Example 10. We have

∞
n
P
(a) The series 2n
is convergent.
n=1

∞
P (2n)!
(b) The series (n!)2
is not convergent.
n=1

∞
xn
P
(c) The series n!
is convergent for any x ∈ R.
n=1

Remark 3. If ` = lim | xxn+1n

| = 1, then the Ratio test is inconclusive. For example,
n→∞
∞
X1 ∞ ∞ ∞
X 1 X 1 X 1
for both the series and 2
, ` = 1. However, is divergent and is
n=1
n n=1
n n=1
n n=1
n2
convergent.

Theorem 13 (Root Test). Let ∞

P p
n=1 x n be a series of real numbers. Let A = lim sup n
|xn |.
Then
∞
P
(1) If A < 1, then xn is absolutely convergent.
n=1

∞
P
(2) If A > 1, then xn is divergent.
n=1

(3) The test is inconclusive if A = 1.

Proof. (1) Let A < 1. Let B ∈ R be such that A < B < 1. Put ε = B − A. We have
p
A = lim sup n |xn | = lim yn ,
n→∞
np o
where yn = sup k |xk | : k ≥ n . Since ε = B − A > 0, so there is some n0 ∈ N such
that
0 < yn < A + ε = B for all n ≥ n0 , that is, |xn | < B n for all n ≥ n0 .
∞
X ∞
X
n
Since 0 < B < 1, so B is convergent. Therefore, by Comparison Test, |xn | is also
n=1 n=1
∞
X
convergent. This proves that xn converges absolutely.
n=1

(2) Let A > 1. Let B ∈ R be such that A > B > 1. Put ε = A − B. Then there
exists some n0 ∈ N such that
B = A − ε < yn for all n ≥ n0 .
Thus, yn > 1 for all n ≥ n0 . From this, we find that |xn | > 1 for infinitely many values of
P∞
n. Hence xn does not converse to 0. This proves that xn is divergent.
n=1
∞
(3) Let xn = n1 . Then A = 1 and we know that 1 1
P
n
is divergent. Again, if xn = n2
then
n=1
∞
1
P
also A = 1. However, n2
is convergent.
n=1
p p
Remark 4. If n
|xn | → `, then A = lim sup n |xn | = `.
∞
P (n!)n
Example 11. (a) The series nn2
is convergent.
n=1

(n!)n 1 n!
Solution. Taking xn = nn2
for all n ∈ N, we have lim |xn | n = lim n = 0 (since
n→∞ n→∞ n
nn
lim (n+1)!
n+1 · = lim 1
1 n = e
1
< 1). Hence by the root test, the given series is
n→∞ (n+1) n! n→∞ (1+ n )
convergent.
∞
5n
P
(b) The series 3n +4n
is not convergent.
n=1

5n 1 5 5
Solution. Taking xn = 3n +4n
for all n ∈ N, we have lim |xn | n = lim 1 = 4
n→∞ n→∞ (3n +4n ) n
1
(since lim (3n + 4n ) n = 4, as shown earlier by using Sandwich theorem). Hence by
n→∞
the root test, the given series is not convergent.
Given a series ∞
P P∞
n=1 xn , we can construct many other series n=1 yn by leaving the
order of the terms xn fixed, but inserting parentheses that group together finite number
of terms. For example, the following series
     
1 1 1 1 1 1 1 1 1
1− + − + − + − + − ··· + − ···
2 3 4 5 6 7 8 9 13
is obtained by grouping the terms in the alternating harmonic series ∞ n+1 1
P
n=1 (−1) n
.
Theorem 14. Grouping of terms of a convergent series does not change the convergence
and the sum. However, a divergent series can become convergent after grouping of terms.
∞
X ∞
X ∞
X
Proof. Let xn be convergent. Suppose that the series yn is obtained from xn
n=1 n=1 n=1
by grouping the terms. Then we have
y1 = x1 + x2 + · · · + xn1 , y2 = xn1 +1 + xn1 +2 + · · · + xn2 , · · ·
∞
X ∞
X
Let (sn ) and (tn ) be the sequences of partial sums of xn and yn , respectively. Then
n=1 n=1

t1 = y1 = sn1 , t2 = y1 + y2 = sn2 , · · · .
∞
X
Thus, (tn ) is a subsequence of (sn ). Since xn is convergent, so (sn ) is convergent.
n=1
Therefore, (tn ) is also convergent and converges to the limit of (sn ). This proves that the
X ∞ X∞
grouped series yn is convergent and its sum is same as xn .
n=1 n=1
∞
X
It is clear that the converse to this theorem is not true. We know that the series (−1)n
n=1
diverges. However, the grouping
(−1 + 1) + (−1 + 1) + · · · + (−1 + 1) + · · ·
converges to 0.
P∞
P∞2 (Rearrangement of series). A series n=1 yn is called a rearrangement of
Definition
a series n=1 xn if there is a bijection f of N onto N such that yn = xf (n) for all n ∈ N.
Theorem 15. Rearrangement of terms does not change the convergence and the sum of
an absolutely convergent series.
Example 12. Let 1 − 21 + 13 − 41 + 15 − 16 + · · · = s. Then,
1 1 1 1 1 1 3
1+ − + + − + + · · · = s.
3 2 5 7 4 9 2
1 1 1
Solution. We first note that by Leibniz’s test, the series 1 − 2
+ 3
− 4
+ · · · converges.
Let
1 1 1
1− + − + · · · = s. (1)
2 3 4
1
Then the series 2
− 41 + 16 − · · · = 21 (1 − 12 + 13 − · · · ) converges to 12 s. It follows that the
series
1 1 1 1 1
0+ − 0 − + 0 + − 0 − + ··· = s (2)
2 4 6 8 2
Hence the series (1 + 0) + (− 21 + 12 ) + ( 31 − 0) + (− 14 − 14 ) + ( 15 + 0) + · · · , which is
the sum of the series (1) and (2), converges to s + 12 s = 32 s. Therefore it follows that
1 + 13 − 21 + 15 + 17 − 14 + 91 + · · · = 23 s.
 ∞
P
Theorem 16 (Riemann’s rearrangement theorem). Let xn be a conditionally conver-
n=1
gent series.
∞
P
(1) If s ∈ R, then there exists a rearrangement of terms of xn such that the rear-
n=1
ranged series has the sum s.
∞
P
(2) There exists a rearrangement of terms of xn such that the rearranged series
n=1
diverges.