Rb

• We have defined a f (x)dx under the conditions that f is defined and

bounded on the bounded interval [a, b].

•In this lecture, we will extend the theory of integration to bounded functions
defined on unbounded intervals and also to unbounded functions defined on
bounded or unbounded intervals.

R∞ R4 dx
• [Example] 0 sin x 2 dx, 0 x−3 .
 Rb
• [Definition:] The integral a f (x)dx is called improper integral if

1. a = −∞ or b = ∞ or both; that is one or both integration limit is infinite.

2. f (x) is unbounded at one or more points in [a, b].

R∞
• [Example:] 1. 0 sin x 2 dx is an improper integral.
R4 dx 1
2. 0 x−3 is an improper integral. Here f (x) = x−3 is unbounded at x = 3
• There are two types of improper integral.

1. First Kind: if a = −∞ or b = ∞ or both.

2. Second Kind: if f (x) is unbounded at one or more points in [a, b].

R∞
• [Example:] 1. 0 sin x 2 dx is an improper integral of first kind.
R 4 dx 1
2. 0 x−3 is an improper integral of second kind. Here f (x) = x−3 is
unbounded at x = 3
• There are three types of improper integral of the first kind.
R∞
i) a f (x)dx.
Rb
ii) −∞ f (x)dx.
R∞
iii) −∞ f (x)dx.

• [Remark:
R ∞Third type can
R c be written Ras∞sum of first type and second type.
That is, −∞ f (x)dx = −∞ f (x)dx + c f (x)dx, where c is a real number.
• [Convergence or Divergence of Improper Integrals of the First Kind:]

Rx
•Suppose f is (Riemann) integrable on [a, x] for all x > a, that is a f (t)dt
exists for all x > a. If Z x
lim f (t)dt = L
x→∞ a

for
R ∞ some L ∈ R, then we say that the improper
R∞ integral (of the first kind)
a f (x)dx converges to L and
R∞ we write a f (x)dx = L. Otherwise we say
that the improper integral a f (x)dx diverges.
Rb
Similar convergence or divergence definition works for −∞ f (x)dx, in this
case if Z b
lim f (t)dt = L
x→−∞ x

for some L ∈ R.
 dx
R∞
• [Example:] 1. 1 x2
. Check whether this improper integral is convergent
or not.
R∞ dx
Rx dx
1 x2
= limx→∞ 1 x2

= limx→∞ (1 − x1 )

= 1. Hence this integral is convergent.

R∞
2. 1 cos xdx. Check whether this improper integral is convergent or not.
R∞ Rx
1 cos xdx = limx→∞ 1 cos xdx

= limx→∞ (sin x − sin 1).

We know that limx→∞ sin x does not exist. Hence the integral is not
convergent.
• [The P integral of first kind:] a dx
R∞
x p where p is constant and a > 0.
This integral is convergent if p > 1 and divergent if p ≤ 1.
  
 1 1 1
R ∞ dx R x dt
1−p x p−1 − ap−1
if p 6= 1
Sol: a x p = limx→∞ a t p = limx→∞
ln x if p = 1

  
 1 1 1
1−p x p−1 − ap−1 if p 6= 1
limx→∞ exists if p > 1 and does not
ln x if p = 1

exist if p ≤ 1.

dx
R∞
Therefore a xp is convergent if p > 1 and divergent if p ≤ 1.
• [Convergence Test for Improper Integrals of the First Kind:]

R∞
• [Remark:] The following test are given for the case a f (x)dx. Similar
Rb R∞
tests are available for −∞ f (x)dx and −∞ f (x)dx.

• [Comparison Test:]

We assume that f and g are integrable on [a, x] for all x > a. Suppose that
0 ≤ f (x) ≤ g (x) for all x > a. Then the following are true.
R∞ R∞
1. If a g (x)dt converges, then a f (x)dt converges.
R∞ R∞
2. If a f (x)dt diverges, then a g (x)dt diverges.
 dx
R∞
• [Example 1:] By using comparison test we check whether 0 e x +1
converges or diverges.

Sol: Since e x + 1 > e x for all x > 0. Then e x1+1 < e1x . Now we can check
R ∞ −x R x −t
0 e Rdx converges or not. limx→∞ 0 e dt = lim
R ∞ x→∞
(1 − e −x ) = 1.
Hence 0 e dx converges. By comparison test 0 e xdx+1 is convergent.
∞ −x

dx
R∞
• [Example 2:] By using comparison test we check whether 2 ln x
converges or diverges.

Sol: Since ln x < x for all x ≥ 2. Then ln1x > x1 . Now we check that
R ∞ dx
2 x R converges or not. This one is P integral of first kind. Since p = 1.
Then 2 dx
∞ R ∞ dx
x diverges. By comparison test 2 ln x diverges.
• [Limit Ratio Test:] We assume that f and g are integrable on [a, x] for
all x > a. Suppose that f (x) ≥ 0 and g (x) > 0 for all x > a. Then the
following are true.

1. If limx→∞ gf (x)
R∞
R∞ (x) = A where A 6= 0, ∞, then bot the integrals a f (x)dx
and a g (x)dx converge or both the integral diverge.
R∞
R2.∞ If A = 0, then convergence of a g (x)dx implies convergence of
a f (x)dx.
R∞
R3.∞ If A = ∞, then divergence of a g (x)dx implies divergence of
a f (x)dx.
• [Example 1.] a x 2dx+1 . Check whether this improper integral is conver-
R∞

gent or not by using limit ratio test.

Sol: Let f (x) = x 21+1 . We consider g (x) = x12 . Since f and g both
are continuous functions in [a, x] for all x > a. Hence they are Riemann
integrable on [a, x] for all x > a.

limx→∞ gf (x)
2x
R∞ dx
(x) = limx→∞ x 2 +1 = 1. The limit A = 1, then a x 2 +1
and
R ∞ dx
a x 2 both diverge or converge.

dx dx
R∞ R∞
a x2
is convergent by using P integral of first kind. Hence a x 2 +1
is
convergent.
• Theorem: Let limx→∞ x µ f (x) = A. Then
R∞
i) a f (x)dx converges if µ > 1 and A is finite.
R∞
ii) a f (x)dx diverges if µ ≤ 1 and A 6= 0 (A may be infinite).

Sometimes it is called µ test.

 x 2 dx 2
converges since limx→∞ x 2 . 4x 4x+25 = 14 .
R∞
• [Example 1:] 0 4x 4 +25

√ xdx x
R∞
• [Example 2:] 0 x 4 +x 2 +1
diverges since limx→∞ x. √x 4 +x 2 +1
= 1.
• [Absolutely and Conditional Convergence:]
R∞ R∞
a f (x)dx is called absolutely convergent if a |f (x)|dx converges.
R∞ R∞ R∞
If a f (x)dx converges but a |f (x)|dx diverges, then a f (x)dx is called
conditionally convergent.

R∞ R∞
• [Theorem:] If a |f (x)|dx converges, then a f (x)dx converges. But the
converse is not true.
• [Improper Integrals of the second kind:]

There are three types of improper integral of the second kind.

i) If f (x) becomes unbounded only at the endpoint a of the interval [a, b],
Rb
then we define a f (x)dx is improper integral of second kind.

ii) If f (x) is unbounded at the endpoint b of the interval [a, b] then

Rb
a f (x)dx is an improper integral of second kind.
Rb
iii) If f (x) is unbounded at some point c in (a, b) then a f (x)dx is an
improper integral of second kind.
• [Convergence or divergence of the second type Improper Integral:]
Rb
i) Suppose x f (t)dt exists for all a < x ≤ b. The function f is unbounded
at a. If Z b
lim f (t)dt = M
x→a+ x
for some real number M, then we say that the improper integral (of the
Rb Rb
second kind) a f (t)dt converges to M and we write a f (t)dt = M.
Rb
Otherwise a f (t)dt diverges.
Rb
ii) Suppose x f (t)dt exists for all a < x ≤ b. The function f is unbounded
at b. If Z x
lim f (t)dt = M
x→b− a
for some real number M, then we say that the improper integral (of the
Rb Rb
second kind) a f (t)dt converges to M and we write a f (t)dt = M.
Rb
Otherwise a f (t)dt diverges.
 Rb
iii) Suppose x f (t)dt exists for all a < x ≤ b. The function f is unbounded
Rc Rb
at c in (a, b). Then we can write a f (t)dt + c f (t)dt.

If Z x
lim f (t)dt = M1
x→c − a
and Z b
lim+ f (t)dt = M2
x→c x
for some two real numbers RM1 and M2 , then we say that the improper
b
integral (of the second kind) a f (t)dt converges to M1 + M2 and we write
Rb Rb
a f (t)dt = M. Otherwise a f (t)dt diverges.
• [Special type of Integrals of the second type:
Rb dx
1. a (x−a)p converges if p < 1 and diverges if p ≥ 1.

Sol: The solution is same as the solution of a dx

R∞
R b dx xp .
2. a (b−x)p converges if p < 1 and diverges if p ≥ 1.

dx
R∞
Sol: The solution is same as the solution of a xp .
• [Convergence test for improper integrals of the second kind:]

• [Remark:] The following test are given for the case where f (x) is un-
bounded only at a in [a, b]. Similar tests are available if f (x) is unbounded
at x = b or at x = c where a < c < b.

• [Comparison Test:]

We assume that f and g are Riemann integrable on [x, b] for all x > a
and Suppose 0 ≤ f (x) ≤ g (x) for all a < x ≤ b. Then the following are true.
Rb Rb
1. If a g (t)dt converges, then a f (t)dt converges.
Rb Rb
2. If a f (t)dt diverges, then a g (t)dt diverges.
• [Limit Ratio Test:] Suppose that f (x) ≥ 0 and g (x) > 0 for all
a < x ≤ b. Then the following are true.

1. If limx→a gf (x)
Rb
(x) = A where A 6= 0, ∞, then bot the integrals a f (x)dx
Rb
and a g (x)dx converge or both the integral diverge.
Rb
2. If A = 0, then convergence of a g (x)dx implies convergence of
Rb
a f (x)dx.
Rb Rb
3. If A = ∞, then divergence of a g (x)dx implies divergence of a f (x)dx.
• [Theorem:] Let limx→a+ (x − a)µ f (x) = A. Then
Rb
i) a f (x)dx converges if µ < 1 and A is finite.
Rb
ii) a f (x)dx diverges if µ ≥ 1 and A 6= 0 (A may be infinite).

cm

• [Theorem:] Let limx→b− (b − x)µ f (x) = B. Then

Rb
i) a f (x)dx converges if µ < 1 and B is finite.
Rb
ii) a f (x)dx diverges if µ ≥ 1 and B 6= 0 (B may be infinite).

Sometimes these two theorems are called µ test.

 Rb Rb
• [Definition:] a f (x)dx is called absolutely convergent if a |f (x)|dx
converges.
Rb Rb Rb
If a f (x)dx converges but a |f (x)|dx diverges, then a f (x)dx is called
conditionally convergent.
Rb Rb
Theorem: If a |f (x)|dx converges, then a f (x)dx converges. But the
converse is not true.

• [Remark] Suppose f is unbounded at c1 and c2 in (a, b). In that case

you choose one point c in (c1 , c2 ). Then split the integral in the fol-
Rb Rc Rc Rc Rb
lowing way. a f (x)dx = a 1 f (x)dx + c1 f (x)dx + c 2 f (x)dx + c2 f (x)dx.

You have to check convergence

Rb of each of the integral. If all the integrals
are convergent then a f (x)dx is convergent.
• [Beta function:] The beta function defined by improper integral
Z 1
β(m, n) = x m−1 (1 − x)n−1 dx, m > 0, n > 0 (1)
0

•Since beta function is an improper integral, we check the convergence of

β(m, n).

• [Theorem:] The beta function is convergent for m > 0 and n > 0.

Sol: We can write β(x) in the following way.
R 1/2 R1
β(m, n) = 0 x m−1 (1 − x)n−1 dx + 1/2 x
m−1 (1 − x)n−1 dx
R 1/2
β(m, n) = I1 + I2 where I1 = 0 x m−1 (1 − x)n−1 dx and
R1
I2 = 1/2 x m−1 (1 − x)n−1 dx
Case I Check the convergence of I1 . If 0 < m < 1, then I1 is improper and
x = 0 is the point of infinite discontinuity.

We use limit comparison test.

Let f (x) = x m−1 (1 − x)n−1 and g (x) = x m−1

f (x)
lim+ = 1(check!)
t→0 g (x)
By limit comparison test both the integrals either converge or diverge.
R 1/2 R 1/2 1
0 g (x)dx = 0 x 1−m
, this one is convergent if 1 − m < 1 =⇒ m > 0.

Then I1 is convergent for m > 0.

Case II Check the convergence of I2 . This integral is improper if 0 < n < 1
and x = 1 is the point of infinite discontinuity.
We use limit comparison test.

Let f (x) = x m−1 (1 − x)n−1 and g (x) = (1 − x)n−1

f (x)
lim+ = 1(check!)
t→0 g (x)
By limit comparison test both the integrals either converge or diverge.
R1 R1 1
1/2 g (x)dx = 1/2 (1−x)1−n , this one is convergent if 1 − n < 1 =⇒ nn >
0.

The integral I2 is convergent for n > 0.

Since I1 is convergent for m > 0 and I2 is convergent for n > 0. Hence
β(m, n) is convergent for m > 0 and n > 0.
• [Properties of beta function]

• β(m, n) = β(n, m), beta function is symmetric.

R1
Proof: β(m, n) = 0 x m−1 (1 − x)n−1 dx.

Put x = 1 − t.
R0 R1
β(m, n) = − 1 t n−1 (1 − t)m−1 dt = 0 t n−1 (1 − t)m−1 dt = β(n, m).

R π/2
• β(m, n) = 2 0 sin2m−1 (θ) cos2n−1 (θ)dθ =
R π/2 2n−1
2 0 sin (θ) cos2m−1 (θ)dθ.
 Proof: Put x = sin2 θ.

x m−1
R∞
• β(m, n) = 0 (1+x)m+n dx

t
Proof: Put x = t+1 .

• β(m, n) = β(m + 1, n) + β(m, n + 1).

 (n−1)!
• β(m, n) = m(m+1)···(m+n−2)(m+n−1) , n being a positive integers.

(m−1)!
• β(m, n) = n(n+1)···(n+m−2)(n+m−1) , m being a positive integers.

(m−1)!(n−1)!
• β(m, n) = (m+n−1)! , both m and m are positive integers.
• The gamma function defined by improper integral
Z ∞
Γ(x) = t x−1 e −t dt for x > 0 (2)
0
Since gamma function is an improper integral, we check the convergence of
Γ(x).

• The gamma function is convergent for x > 0.

Proof: We can write Γ(x) in the following way.

Z 1 Z ∞
x−1 −t
Γ(x) = t e dt + t x−1 e −t dt
0 1
.
R1 R∞
So Γ(x) = I1 + I2 where I1 = 0 t x−1 e −t dt and I2 = 1 t x−1 e −t dt
Case I Convergence of I1 .

It is clear that I1 is an improper integral if 0 < x < 1 and t = 0 is the point

of infinite discontinuity. For x ≥ 1, I1 is proper integral.

We use limit comparison test to check convergence of I1 for 0 < x < 1.

Let f (t) = t x−1 e −t and g (t) = t x−1 .

f (t)
lim+ = 1(check!)
t→0 g (t)
By limit ratio test both the integrals either converge or diverge.
R1 R1 1
0 g (t)dt = 0 t 1−x dx, this integral is convergent only when
1 − x < 1 =⇒ x > 0.

Hence I1 is convergent only when x > 0.

Case II Convergence of I2 .

We use comparison test.

Since x − 1 (x > 0) is a real number, there exits a natural number n such

that x − 1 < n. Therefore, t x−1 < t n and t x−1 e −t < t n e −t , 1 < t < ∞.
R∞
We now check the convergence of 1 t n e −t dt
 b
Rb
limb→∞ 1 t n e −t = limb→∞ e −t P n (t) (by using integrating by parts,
1
Pn (t)) is a polynomial of degree n)
   
Therefore limb→∞ e Pn (b) − e Pn (1) = limb→∞ e Pn (b) −
−b −1 −b

e −1 Pn (1)
 
You can easily check that limb→∞ e −b Pn (b) = 0
 R∞ R∞
Therefore 1 t n e −t dt is convergence. Hence 1 t x−1 e −t dt convergence
for x > 0.

Since I1 and I2 both the integral converge for x > 0. Hence Γ(x) is conver-
gent for x > 0.
• [Properties of Gamma functions:]

• Γ(1) = 1

Proof: Very easy.

• Γ(x + 1) = xΓ(x)
R∞
Proof: Γ(x + 1) = t x e −t dt. By using integrating by parts, we have
0
 ∞
R ∞ x −t x
R∞
Γ(x + 1) = 0 t e dt = − t e −t + x 0 t x−1 e −t dt = xΓ(x).
0
• Γ(m + 1) = m!

Proof: Just apply the above result successively. That is Γ(m + 1) =

mΓ(m) = m(m−1)Γ(m−1) = m(m−1)(m−2)Γ(m−2) = · · · 1Γ(1) = m!.

√
• Γ(1/2) = π.
R∞ R∞ 2
Proof: Γ(1/2) = 0 t −1/2 e −t dt. Put t = u 2 , we have 2 0 e −u du.
 2   
R ∞ −u2 R ∞ −v 2 2 2
dv = 4 0 0 e −(u +v ) dudv
R∞R∞
Γ(1/2) = 2 0 e du 2 0 e
(they are independent).

Changing to polar coordinates u = r cos θ and v = r sin θ dudv = rdrdθ

 2
R π/2 R π/2 −(r )2
Γ(1/2) = 4 0 0 re drd θ = π.
 Γ(m)Γ(n)
• β(m, n) = Γ(m+n) .