MAL 100: Calculus

Lecture Notes
3 Innite Series
3.1 Denitions & convergence
Denition 3.1.1. Let {a
n
} be a sequence of real numbers.
a) An expression of the form
a
1
+ a
2
+ . . . + a
n
+ . . .
is called an innite series.
b) The number a
n
is called as the n
th
term of the series.
c) The sequence {s
n
}, dened by s
n
=
n

k=1
a
k
, is called the sequence of partial sums of the
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) If 0 < x < 1, then

n=0
x
n
converges to
1
1 x
.
Solution. Let us consider the sequence of partial sums {s
n
}, where s
n
=
n

k=1
x
k
. Here
s
n
=
n

k=1
x
k
=
1 x
n+1
1 x
=
1
1 x

x
n+1
1 x
, n N.
As, 0 < x < 1, x
n+1
0 as n . Hence s
n

1
1 x
. Thus the given series converges
to
1
1 x
. ///
2) The series

n=1
1
n
diverges.
1
Solution. Consider the sequence of partial sums {s
n
}, where s
n
=
n

k=1
1
k
. Now, let us
examine the subsequence s
2
n of {s
n
}. Here
s
2
= 1 + 1/2 = 3/2,
s
4
= 1 + 1/2 + 1/3 + 1/4 > 3/2 + 1/4 + 1/4 = 2.
Suppose s
2
n > (n + 2)/2, then
s
2
n+1 = s
2
n +
2
n

k=1
1
2
n
+ k
>
n + 2
2
+
2
n

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

n=1
1
n
diverges. ///
3) (Telescopic series:) Show that the series

n=1
1
n(n + 1)
converges to 1.
Solution. Consider the sequence of partial sums {s
n
}. Then
s
n
=
n

k=1
1
k(k + 1)
=
k

k=1
_
1
k

1
k + 1
_
= 1
1
n + 1
1.
Summarizing this observation, one has the following theorem on Telescopic series
Theorem 3.1.3. Suppose {a
n
} is a sequence of non-negative real numbers such that a
n
L.
Then the series

(a
n
a
n+1
) converges to a
1
L.
Lemma 3.1.4.
1) If

k=1
a
n
converges to L and

k=1
b
n
converges to M, then the series

k=1
(a
n
+ b
n
) converges
to L + M.
2) If

k=1
a
n
converges to L and if c R, then the series

k=1
ca
n
converges to cL.
2
Lemma 3.1.5. If

k=1
a
n
converges, then lim
n
a
n
= 0.
Proof. Suppose

k=1
a
n
= L. Then the sequence of partial sums {s
n
} also converges to L. Now
a
n
= s
n
s
n1
L L = 0. ///
Example 3.1.6. If x > 1, then the series

k=1
x
n
diverges.
Solution. Assume to the contrary that the series

k=1
x
n
converges. Then the n
th
term, i.e.,
x
n
0. But as x > 1, x
n
1 for all n N and hence lim
n
x
n
1, which is a contradiction.
Hence the series

k=1
x
n
diverges. ///
As a rst result we have the following comparison theorem:
Theorem 3.1.7. Let {a
n
}, {b
n
} be sequences of positive reals such that a
n
b
n
. If

b
n
converges then

a
n
converges.
Proof. Let s
n
= a
1
+ a
2
+ .... + a
n
and t
n
= b
1
+ b
2
+ .... + b
n
be the partial sum of

a
n
,

b
n
respectively. Then s
n
t
n
. Since

b
n
converges, we have {t
n
} converges and is bounded. Now
since {s
n
} is monotonically increasing sequence that is bounded above, we get the convergence
of {s
n
} and hence the convergence of

a
n
. ///
Theorem 3.1.8. Let {a
n
}

1
be an decreasing sequence of positive numbers. Then

n=1
a
n
con-
verges if and only if

n=0
2
n
a
2
n converges.
Proof. Let s
n
and t
n
be the sequence of partial sums of

a
n
and

2
n
a
2
n respectively. Then
proof follows from the observation that
s
2
n =
2
n

k=1
a
n
= a
1
+ a
2
+ (a
3
+ a
4
) + (a
5
+ a
6
+ a
7
+ a
8
) + .... + (a
2
n1
+1
+ .... + a
2
n)
a
1
+ a
2
+ 2a
4
+ 4a
8
+ 8a
16
+ ....2
n1
a
2
n
= a
1
+
1
2
t
n
. (3.1)
Therefore, if {s
n
} converges then {s
2
n} converges and by 3.1, {t
n
} converges.
On the other hand,
s
2
n
1
= a
1
+ (a
2
+ a
3
) + (a
4
+ a
5
+ a
6
+ a
7
) + (a
8
+ ....a
17
) + (a
2
n1 + .... + a
2
n
1
)
a
1
+ 2a
2
+ 4a
4
+ 8a
8
+ .... + 2
n1
a
2
n1 = a
1
+ t
n1
3
So if {t
n
} converges, then {s
2
n
1
} converges. Now the conclusion follows from s
n
s
2
n+1
1
and
the fact that {s
n
} is monotonically increasing sequence. ///
Examples 3.1.9.
1) Consider the series

n=1
1
n
p
, p > 0. Then, we have

n=1
2
n
1
(2
n
)
p
=

n=1
1
(2
n
)
p1
which con-
verges for p > 1 and diverges for p 1.
2) Consider the series

n=2
1
nlog n
. Here

n=2
2
n
1
2
n
log 2
n
=
1
log 2

n=2
1
n
which diverges. Hence
the given series diverges.
3.2 Absolute convergence
Denition 3.2.1. a) Let

n=1
a
n
be a series of real numbers. If

n=1
|a
n
| converges, we say that

n=1
a
n
converges absolutely.
b) If

n=1
a
n
converges but

n=1
|a
n
| diverges, we say that

n=1
a
n
converges conditionally.
Examples 3.2.2.
1) The series

n=1
(1)
n
n!
converges absolutely.
2) The series

n=1
(1)
n
n
2
converges absolutely.
Theorem 3.2.3. If

n=1
a
n
converges absolutely, then

n=1
a
n
converges.
Proof. Let t
n
=
n

k=1
|a
k
|. As the series converges absolutely, the sequence {t
n
}

1
is Cauchy. Thus,
given > 0, there exists N N such that
|t
m
t
n
| < m, n N.
Let m > n. Then
|s
m
s
n
| =

i=n+1
a
i

i=n+1
|a
i
| = |t
m
t
n
| < .
Thus the sequence {s
n
}

1
is Cauchy and hence converges. Thus

1
a
n
converges. ///
4
Theorem 3.2.4. Let

1
a
n
be a series of real numbers. Let p
n
= max{a
n
, 0} and q
n
=
min{a
n
, 0}.
a) If

a
n
converges absolutely, then both

p
n
and

q
n
converges.
b) If

a
n
converges conditionally then both

p
n
and

q
n
diverges.
Proof.
a) Observe that p
n
= (a
n
+|a
n
|)/2 and q
n
= (a
n
|a
n
|)/2. Thus the convergence of the two
series follows from the hypothesis.
b) From the observation that p
n
= (a
n
+ |a
n
|)/2, we have |a
n
| = 2p
n
a
n
. As the series

a
n
converges and

|a
n
| diverges, the series

p
n
diverges. Similarly, the series

q
n
diverges. ///
Tests for absolute convergence
Theorem 3.2.5 (Comparison test). Let

a
n
be a series of real numbers.
a)

a
n
converges if there is an absolutely convergent series

c
n
with |a
n
| |c
n
| for all
n N, N N.
b)

a
n
diverges if there is a divergent series

c
n
of nonnegative terms, with c
n
|a
n
| for
all n N, N N.
Proof follows as in Theorem 3.1.7
Examples 3.2.6.
1) The series

n=1
7
7n 2
diverges because
7
7n 2
=
1
n 2/7

1
n
for all n N and
1
n
diverges.
2) The series

n=0
1
n!
converges because
1
n!

1
2
n
and

n=0
1
2
n
converges.
Theorem 3.2.7 (Limit comparison test). Let {a
n
} and {b
n
} be two sequences of positive num-
bers. Then
a) if lim
n
a
n
b
n
= c > 0,

a
n
and

b
n
both converge or diverge together;
b) if lim
n
a
n
b
n
= 0 and

b
n
converges absolutely,

a
n
converges absolutely.
c) if lim
n
a
n
b
n
= and

|b
n
| diverges,

a
n
diverges.
5
Proof. (a) As lim
n
a
n
b
n
= c > 0, for =
c
2
> 0, there exists N N such that
n N =

a
n
b
n
c

<
c
2
.
Thus, for n N,
c
2

a
n
b
n
c
c
2
or equivalently
cb
n
2
a
n

3cb
n
2
.
Hence the conclusion follows from the comparison test.
b) Given that lim
n
a
n
b
n
= 0. Hence for =
1
2
, there exists N N such that
n N =
a
n
b
n
<
1
2
or equivalently,
n N = a
n

b
n
2
.
Thus the desired conclusion follows from the comparison test.
c) Here we are given that lim
n
a
n
b
n
= . Hence for any real number M > 0, there exists
N N such that
n N =
a
n
b
n
M
or equivalently,
n N = a
n
Mb
n
.
Thus if

|b
n
| diverges, then

|a
n
| diverges by comparison test. ///
Examples 3.2.8.
1) Consider the series

n=1
2n + 1
(n + 1)
2
. Here a
n
=
2n + 1
(n + 1)
2
. Let b
n
=
1
n
. Then
a
n
b
n
=
_
2n + 1
(n + 1)
2
_
1
n
=
2n
2
+ n
n
2
+ 2n + 1
2 as n . Further,

1
n
diverges. Thus by limit comparison theorem,
the given series diverges.
2) Consider the series

1
1
2
n
1
. Here a
n
=
1
2
n
1
. Let b
n
=
1
2
n
. Then
a
n
b
n
=
2
n
2
n
1
1.
Further,
1
2
n
converges and hence the given series converges.
6
3) Consider the series

e
n
n
2
. Here a
n
=
e
n
n
2
and b
n
=
1
n
2
. Then
an
bn
= e
n
0 as n .
Further,

1
n
2
converges and hence the given series converges.
Theorem 3.2.9 (Ratio test). Let

1
a
n
be a series of real numbers. Let
a = liminf
n

a
n+1
a
n

and A = limsup
n

a
n+1
a
n

.
Then
a)

1
a
n
converges absolutely if A < 1;
b)

1
a
n
diverges if a > 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
B = A + and also N N such that

a
n+1
an

B for all n N. Further, for any k N,

a
N+k
a
N

=
k

i=1

a
N+i
a
N+i1

i=1
B = B
k
.
Thus |a
N+k
| B
k
|a
N
|, k N. But

k=0
|a
N
|B
k
< as B < 1. Thus by comparison test, the
series

1
a
n
converges.
b) If a > 1, choose b such that 1 < b < a. There exits N N such that

a
n+1
an

b for all
n N. Further, for any k N,

a
N+k
a
N

=
k

i=1

a
N+i
a
N+i1

i=1
b = b
k
.
Thus |a
N+k
| |a
N
|, k N. But, as b > 1,

k=0
a
N
b
k
diverges. Thus, again, by the comparison
test, the series

1
a
n
diverges.
c) Consider the series
1
n
. Here lim
n
a
n+1
a
n
= 1. But
1
n
diverges. For the series
1
n
2
,
which converges, again lim
n
a
n+1
a
n
= 1. ///
Examples 3.2.10.
7
a) Consider the series

1
n
n
n!
. Here
a
n+1
a
n
=
(n + 1)
n+1
(n + 1)!
n!
n
n
=
_
n + 1
n
_
n
=
_
1 +
1
n
_
n
e,
which is greater than 1. So a = A = e > 1. Thus the given series diverges.
b) Consider the series

0
x
n
n!
, x R. Here
a
n+1
a
n
=
x
n+1
(n + 1)!
n!
x
n
=
x
n + 1
0.
Therefore a = A = 0 < 1. Thus, for all x R, the given series converges.
Theorem 3.2.11 (Root test). Let

1
a
n
be a series of real numbers. Let A = limsup
n
n
_
|a
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. a) If A < 1, choose B such that A < B < 1. Then there exists N N such that
n
_
|a
n
| < B for all n N. This implies |a
n
| < B
n
for all n N. As B < 1, the series converges
by comparison test.
b) If A > 1, there exists innitely many n N such that
n
_
|a
n
| > 1. But this implies that
|a
n
| > 1 for innitely many values of n and hence a
N
0, i.e.,

a
n
diverges.
c) Consider the series
1
n
. Here A = 1 and the series diverges. On the other hand, for the
series
1
n
2
, again A = 1, but the series converges. ///
Examples 3.2.12.
1) Consider the series

1
x
n
n
, x R. Here a
n
=
x
n
n
. Therefore,
n

x
n
n

x
n

0. Thus
the series converges for |x| < 1 and diverges for |x| > 1.
2) Consider the series

1
x
n
n
n
, x R. Here a
n
=
x
n
n
n
. Then,
n
_
|a
n
| =

x
n

0. Thus the series

converges for x R.
8
3) Consider the series

a
n
, where a
n
=
_
_
_
n
2
n
n is odd
1
2
n
n is even
. Then limsup
n
n

a
n
=
1
2
. There-
fore the series converges.
Alternating series:
Denition 3.2.13. An alternating series is an innite series whose terms alternate in sign.
Theorem 3.2.14. Suppose {a
n
} is a sequence of positive numbers such that
(a) a
n
a
n+1
for all n N and
(b) lim
n
a
n
= 0,
then the alternating series

n=1
(1)
n+1
a
n
converges.
Proof. Consider the partial sums with odd index, s
1
, s
3
, s
5
, . . . . Now, for any n N,
s
2n+1
= s
2n1
a
2n
+ a
2n+1
s
2n1
(by (a)).
Thus the sequence {s
2n1
}

1
forms a non-increasing sequence. Also, notice that
s
2n1
=
n1

i=1
(a
2i1
a
2i
) + a
2n1
.
Since each quantity in the parenthesis is non-negative and a
2n1
> 0, the sequence {s
2n1
} is
bounded below by 0. Hence {s
2n1
}

1
is convergent.
Now, consider the partial sums with even index, s
2
, s
4
, s
6
, . . . . For any n N,
s
2n+2
= s
2n
+ a
2n+1
a
2n+2
s
2n
(by (a)).
Thus the sequence {s
2n
}

1
forms a non-decreasing sequence. Further,
s
2n
= a
1

n1

i=1
(a
2i
a
2i+1
) a
2n
a
1
,
which means that s
2n
is bounded above by a
1
. Therefore, {s
2n
} is convergent.
Let L = lim s
2n
and M = lim S
2n1
. By ((b)),
0 = lim a
2n
= lim (s
2n
s
2n1
) = L M.
Thus L = M and hence the alternating series

n=1
(1)
n+1
a
n
converges. ///
9
Examples 3.2.15.
1) Consider the series

n=1
(1)
n+1
2
1/n
. Here a
n
= 2
1/n
1 as n . Hence the above
theorem does not apply. Anyhow, one can show that the series diverges.
2) Consider the series

n=1
(1)
n+1
n
. The a

n
s of this series satises the hypothesis of the above
theorem and hence the series converges.
Examples 3.2.16.
1) The series

n=1
(1)
n+1
n
converges conditionally.
2) The series

n=1
(1)
2n1
2n 1
converges conditionally.
3.3 Error estimation in Innite Series
Here we discuss the methods of estimating the sum of a innite series which converges by the
tests discussed in the previous sections. First we have the series of positive real numbers.
Theorem 3.3.1. Suppose {a
n
} is a positive decreasing sequence and lim
n
a
n+1
a
n
= L < 1.
(a) If
a
n+1
an
decreases to the limit L, then
a
n
_
L
1 L
_
< S S
n
<
a
n+1
1
a
n+1
an
(b) If
a
n+1
an
increases to the limit L, then
a
n+1
1
a
n+1
an
< S S
n
< a
n
_
L
1 L
_
.
Proof. It is obvious from the statement that the series converges by Ratio test. Let r =
a
n+1
an
< 1.
Then
a
k+1
a
k
< r for all k n. Hence
a
n+k
< a
n+k1
r < a
n+k2
r
2
< a
n+k3
r
3
..... < a
n
r
k
Therefore,
S S
n
=

k=n+1
a
k
=

k=1
a
n+k
<

k=1
a
n
r
k
=
a
n
r
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 <
a
n+1
an
, k n, we see
S S
n
=

k=1
a
n+k
>

k=1
a
n
L
k
= a
n
L
1 L
.
A similar argument with L and r reversed proves (b).
Remark: Let A
n
=
a
n+1
1
a
n+1
an
and B
n
=
Lan
1L
= 0. Suppose (A
n
B
n
)/2 , then the error
S S
n
< .
Example: Consider the series S =

n=1
1
n
2
5
n
. Then
a
n+1
a
n
=
n
2
5
n
(n + 1)
2
5
n+1
=
_
n
n + 1
_
2
1
5
increases to
1
5
. So by the above inequality
1
(n+1)
2
5
n+1
1
1
5
(n/n + 1)
2
< S S
n
<
1
4n
2
5
n
.
That is
1
(4n
2
+ 10n + 5)5
n
< S S
n
<
1
4n
2
5
n
.
so if we take n = 5, we get 0.2110037 < S < 0.2110049.
Problem : Find the sum of the series

n=1
n
2
n!
with error < 10
6
.
Here A
n
=
n+1
n!(1
n+1
n
2
)
and B
n
= 0. By trial and error we can see that (A
n
B
n
)/2 10
6
if
n 11.
Theorem 3.3.2. Consider the series

n=1
(1)
n+1
a
n
such that {a
n
} positive decreasing se-
quence that converges to 0 and b
n
= a
n
a
n+1
also decreases to 0. Then
a
n+1
2
< |S S
n
| <
a
n
2
.
Proof. Note that
S = S
n
+ (1)
n
(b
n+1
+ b
n+3
+ ....) or S = S
n1
+ (1)
n+1
(b
n
+ b
n+2
+ ...).
Since {b
n
} is decreasing, we obtain
|S S
n
| = b
n+1
+ b
n+3
+ .... < b
n
+ b
n+2
+ ... = |S S
n1
|.
11
Similarly, |S S
n+1
| < |S S
n
|. But S lies between two successive partial sums, so
a
n
= |S
n
S
n1
| = |S S
n
| +|S S
n1
| > 2|S S
n
|
and
a
n+1
= |S
n+1
S
n
| = |S S
n+1
| +|S S
n
| < 2|S S
n
|.
These two estimates immediately gives the required estimate.
Problem: Find the sum of the series S =

(1)
n+1 4
2n1
with error less than 0.0001.
Solution: It is easy to check b
n
= a
n
a
n+1
is decreasing!. So for odd n, the above inequality
implies that
S
n

2
2n 1
< S < S
n

2
2n + 1
(3.2)
By the remark above, typically n is such that a
n+1
=
4
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 {a
n
}

n=0
, the series

n=0
a
n
(x c)
n
is called power series with
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.
Remark: If

a
n
x
n
converges at x = r, then

a
n
x
n
converges for |x| < |r|.
Proof: We can nd C > 0 such that |a
n
x
n
| C for all n. Then
|a
n
x
n
| |a
n
r
n
||
x
r
|
n
C|
x
r
|
n
.
Conclusion follows from comparison theorem.
Theorem 3.4.1. Consider the power series

n=0
a
n
x
n
. Suppose = limsup
n
_
|a
n
| and R =
1

(We dene R = 0 if = and R = if = 0). Then

1.

n=0
a
n
x
n
converges for |x| < R
2.

n=0
a
n
x
n
diverges for |x| > R.
3. No conclusion if |x| = R.
Proof. Proof of (i) follows from the root test. For a proof, take
n
(x) = a
n
x
n
and =
limsup
n
_
|
n
|. For (ii), one can show that if |x| > R, then there exists a subsequence {a
n
} such
12
that a
n
0. Notice that = |x|. For (iii), observe as earlier that the series with a
n
=
1
n
and
b
n
=
1
n
2
will have R = 1.
Similarly, we can prove:
Theorem 3.4.2. Consider the power series

n=0
a
n
x
n
. Suppose = limsup

a
n+1
an

and R =
1

(We dene R = 0 if = and R = if = 0). Then

1.

n=0
a
n
x
n
converges for |x| < R
2.

n=0
a
n
x
n
diverges for |x| > R.
3. No conclusion if |x| = R.
Denition 3.4.3. The real number R in the above theorems is called the Radius of convergence
of power series.
Examples: Find the interval of convergence of (i)

x
n
n
(ii)

x
n
n!
(iii)

2
n
x
3n
1. = limsup |
a
n+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. = limsup |
a
n+1
an
| = 0. Hence the series converges everywhere.
3. To see the subsequent non-zero terms, we write the series as

2
n
(x
3
)
n
=

2
n
y
n
. For
this series
y
= limsup
n
_
|a
n
| = 2
1
. Therefore,
x
= 2
1/3
and R = 2
1/3
.
Taylor Series Revisited:
The following theorem is very useful in identifying the domain of convergence of some Taylor
series.
Theorem 3.4.4. (Term by term dierentiation and integration): Suppose f(x) =

n=0
a
n
x
n
converges for |x| < R. Then
1.

n=0
na
n
x
n1
converges in |x| < R and is equal to f

(x).
2.

n=0
1
n+1
a
n
x
n+1
converges in |x| < R and is equal to
_
f(x)dx.
From this theorem one concludes that a power series is innitely dierentiable 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) =

a
n
x
n
, then
a
n
=
f
(n)
n!
. The above theorem is useful to nd 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.
tan
1
x =
_
dx
1 + x
2
=
_
1 x
2
+ x
4
+ ...
=x
x
3
3
+
x
5
5
+ ...
Taking x = 1 we get interesting sum
1
1
3
+
1
5

1
7
+ .... = tan
1
(1) =

4
.
Though the function tan
1
x is dened on all of IR, we see that the power series converges on
(1, 1). We can apply Abels 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 s
n
(x), then the maximum
error is
|x|
n+1
n+1
. ///
We note that the power series may converge to a function on small interval, even though
the function is dened 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 dened on (1, ). The domain
of convergence of power series is symmetric about the centre but the domain of deniton of
function. For instance, for a function dened 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,
(1) erf(x) =
_
x
0
e
t
2
dt =
_
x
0
(1
t
2
1!
+
t
4
2!
+ ... = x
x
3
3
+
x
5
10
+ ...
(2)
_
x
0
sin t
t
=
_
x
0
1
t
2
3!
+
t
4
5!
= x
x
3
3! 3
+
x
5
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 Innite Series, The American Mathematical Monthly, Bart Braden,
Vol.99, 7, (1992), 649-655.
14