Improper Integrals

Introduction
Let f ( x) be a real valued function of a real variable 𝑥 defined over a closed interval [𝑎, 𝑏]. If 𝑓(𝑥) is
a bounded and integrable on [𝑎, 𝑏], the definite integral
𝑏
∫ 𝑓(𝑥)𝑑𝑥
𝑎

is called a proper integral.

Improper Integral
The abovementioned definite integral is called improper if one or both of the limits of integration
become infinite, or the integrand become unbounded over the interval [𝑎, 𝑏]. The improper integrals
are classified in three subcategories.

Improper integral of first kind

If the interval of integration is (−∞, 𝑏], [𝑎, ∞)𝑜𝑟(−∞, ∞) but the function 𝑓(𝑥) is bounded, the
definite integral is called improper integral of the first kind.
 1

∞ 1
Example 1. ∫0 𝑒 −𝑥 𝑑𝑥, ∫−∞ 𝑒 𝑥 𝑑𝑥, dx .
− 1 + x 2

In each of these integrals the integrand is bounded.

Improper integral of second kind

If the interval [𝑎, 𝑏] is finite but the function is unbounded in [𝑎, 𝑏], the definite integral is called
improper integral of the second kind.

1 1 ex

2
Example 2.  dx ,
1
0 1 − x dx ,
1
dx .
0 x 1 (2 x − 3)
Note that the integrand is unbounded at lower boundary in the first, upper boundary in the second and
at an interior point in the last improper integrals.

Improper integral of third kind

If the interval is not finite and the function is also not bounded in the interval, the definite integral is
called improper integral of the third kind.
 1 
Example 3. 1 1− x
dx or 
−
( x2 − 3x)dx .

In the first of these improper integrals, the integrand is unbounded at the lower boundary and also the
upper limit is infinite. In the second, both the limits are infinite and the integrand is also unbounded.
The improper integrals may have finite value, infinite value or indeterminate depending upon the
nature of the function 𝑓(𝑥) and the limits of integration. Although, special care is taken while
evaluating the improper integrals, theoretically these are not much distinct from the proper integrals
because by taking some suitable transformation of variable, a proper integral can be converted to a
convergent improper integral and vice-versa. The reader can verify the equality of the following
integrals:

1 1x

1.  2 e dx =  e x dx
1

1 x 0

 1 1
2.  dx = 0 dx
2 ( x − 1) 2

The integrals on the left hand side are improper while those on the right hand side are proper integrals.

Convergence of Improper Integrals

An improper integral is called convergent if it has a definite value, otherwise it is called divergent.
Thus, if an improper integral can be evaluated to have a definite finite value (with the known methods
of indefinite integrals and putting limits), the integral is convergent. However, many times the
improper integrals cannot be evaluated directly but may be convergent. For such improper integrals,
there are many tests to determine if the integral converge or not. Here are some simple tests to
determine the convergence of improper integrals.

Test for improper integrals of the first kind

Convergence in [𝒂, ∞)
If the function 𝑓(𝑥) is bounded and integrable for all 𝑥 ≥ 𝑎 for some real number 𝑎, the improper
integral in the interval [𝑎, ∞) is defined as:

 f ( x)dx = lim  f ( x)dx
b

a b → a

If the limit on the right hand side exists, the improper integral on the left hand side is convergent
otherwise divergent.
∞
Example 4. Consider the integral ∫0 𝑒 −𝑥 𝑑𝑥. It is easy to show that the integrand is bounded over the
interval [0, ∞) because lim 𝑒 −𝑥 = 0. Also
𝑥→∞

lim  e− x dx = lim(1 − e−b ) = 1 − 0 = 1

b

b → 0 b →

Therefore,

 e− x dx = lim  e − x dx = 1 .
b

0 b → 0

∞
Example 5. Consider the integral ∫0 cos(𝑥) 𝑑𝑥 . Again, the integrand is bounded over the interval
[0, ∞) because | cos(𝑥)| ≤ 1. However, the limit

lim  cos( x)dx = lim[sin(b) − 0] = lim sin(b)

b

b → 0 b → b →

∞
does not exist because 𝑠𝑖 𝑛(𝑏) fails to have a limit as 𝑏 → ∞. Therefore, the integral ∫0 cos(𝑥) 𝑑𝑥 is
not convergent.
A special Integral
Example 6. Show that
∞
2 √𝜋
∫ 𝑒 −𝑥 𝑑𝑥 =
0 2

Proof. Let

I =  e− x dx
2

I2 = (  e dx)(  e dx ) =   e


0
− x2


0
− x2


0


0
− x2 − y 2
e dxdy = 
0


0

e−( x
2
+ y2 )
dxdy

Taking x = r cos  , y = r sin  , we have x 2 + y 2 = r 2 , dxdy = rdrd and the limits of

integration become r = 0 to  for radial variable, and  = 0 to  / 2 . Thus, we have

I2 = 
0


0
 /2
e− r rd dr =
2

( 0
 /2
d )(  e 

0
−r2
rdr )
But, taking r 2 = z, and therefore, 2rdr = dz , we have

 1  −z 1  /2 
 e− r rdr =  e dz = , and  d =
2

0 2 0 2 0 2

 I2 = ( 0
 /2
d )(  e 

0
−r2
rdr = ) 1 
=
22 4


I = .
2
That is,
∞
2 √𝜋
∫ 𝑒 −𝑥 𝑑𝑥 = .
0 2

Convergence in (−∞, 𝒃]
If the function 𝑓(𝑥) is bounded and integrable for all 𝑥 ≤ 𝑏 for some real number 𝑏, the improper
integral in the interval (∞, 𝑏] is defined as:


b

b
f ( x)dx = lim f ( x)dx
− a →− a

If the limit on the right-hand side exists, the improper integral on the left hand side is convergent
otherwise divergent.


1
 e dx = lim (e − e ) = e − 0 = e .
1
Example 7. e x dx = lim x a
− a →− a a →−
Convergence in (−∞, ∞)
a 
If the improper integrals  −
f ( x)dx and 
a
f ( x)dx are convergent for all finite values of 𝑎, we say

that the improper integral  f ( x)dx is also convergent. In such case,
−

 a 
−
f ( x)dx =  f ( x)dx +  f ( x)dx
− a

Example 8. Let use examine the convergence of the following improper integral
 1
− 1 + x2
dx .

First, observe that the integrand is bounded for all real values of 𝑥. Now, let 𝑎 be some finite real
number, then by definition
 1 1  1
− 1 + x − 1 + x
a
2
dx = 2
dx + a 1 + x2 dx
 1 1  1
 dx = lim 
a
dx + lim  dx
− 1 + x 2  →−  1 + x 2  → a 1 + x 2

 1
 dx = lim (tan −1 a − tan −1  ) + lim(tan −1  − tan −1 a)
− 1 + x 2  →−  →

 1  
 dx = lim (− tan −1  ) + lim(tan −1  ) = + =  .
− 1 + x 2  →−  → 2 2

Exercise
Test the convergence of the following integrals

1
1. 
0 1 + x2
dx Convergent
2
 x
2.  dx Divergent
0 1 + x2

 x
3.  dx Convergent
0 1 + x2

2
 x
4.  4 dx Convergent
2 x −1


5. 0
x cos xdx Divergent
 1
6. 0 1+ x
dx Divergent

7. 0
eax dx Convergent if 𝑎 < 0, divergent if 𝑎 ≥ 0.

 x3e− x dx Convergent
2
8.
0

9. −
xdx Divergent
 
10. 
0
xe− x dx Convergent.

Some Standard Results

∞ 𝟏
1. ∫𝒂 𝒙𝒏
𝒅𝒙, 𝒂 >𝟎
The improper integral
 1
a xn
dx, a  0

is convergent for 𝑛 > 1 and divergent for 𝑛 ≤ 1.

Proof. Case 1: Let 𝑛 = 1. We have
 dx b dx
 = lim = lim ( log b − log a ) = 
a x b→ a x b→
Case 2. Let 𝑛 ≠ 1.

 dx b dx 1  1 1 
a x n
= lim  n = lim
b → a x

b → 1 − n b

n −1
− n −1  .
a 
n −1
Thus, if 𝑛 > 1, lim (1/ b ) → 0 , and we get
b →

 dx b dx 1
a x n
= lim
b → a x n
=−
(n − 1)a n−1
.

n −1 1− n
If 𝑛 < 1, lim (1/ b ) = lim (b ) →  , and we get
b → b →

( b1−n − a1−n ) = (  − a1−n ) =  .

 dx b dx 1 1
a x n
= lim n = lim
b → a x b → 1− n 1− n
Therefore, the integral converges if 𝑛 > 1 and diverges if 𝑛 ≤ 1.
∞ 1
Example 9. Show that the integral ∫2 (𝑥−1)3
𝑑𝑥 is convergent.

x =b
 dx dx  1 1
 = lim 
b
= lim  − 
2 ( x − 1) b→ 2 ( x − 1) b→  2  ( x − 1) 2
3 3
x=2

 1  1   1 1
=  −  lim  − 1 =  −  ( 0 − 1) = .
 2  (b − 1)   2
→ 2
b 2
∞ 𝟏
2. ∫𝒂 (𝒙−𝒂)𝒏
𝒅𝒙
The improper integral

1

b
dx
a ( x − a) n
is convergent only if 𝑛 < 1.
Proof. Let 𝑛 ≠ 1 and let 𝜖 > 0 be a sufficiently small real number. We can write,
x =b
1 1 1  1 1 
a− ( x − a)n dx = (n −1)( x − a)n−1
b
=  − n−1 
x = a −
(n − 1)  (b − a) n −1
 
1
Now, if 𝑛 < 1, 𝜖𝑛−1 = 𝜖 1−𝑛 → 0 as 𝜖 → 0, and we have

1 1 1  1  1 1

b
dx = lim 
b
dx = lim  −
−  1− n  = n −1
.
a ( x − a) n  → 0 a − ( x − a ) n  →0 ( n − 1) (b − a )

n 1
 (n − 1) (b − a)
1
But, if 𝑛 > 1, 𝜖𝑛−1 → ∞ as 𝜖 → 0, thus, we have

1 1 1  1 1 

b
dx = lim 
b
dx = lim  − n −1 
a ( x − a) n  → 0 a −  ( x − a) n  → 0 (n − 1)  (b − a ) n −1
 

1  1 
=  n −1
−  =  .
(n − 1)  (b − a) 
Finally, if 𝑛 = 1, the integral is not convergent because 𝑙𝑜 𝑔(𝑥 − 𝑎) is not defined at 𝑥 = 𝑎 on real
line.
Hence, the integral is convergent if and only if 𝑛 < 1.

Test for Improper Integrals of Second Kind

Integrand unbounded at the upper limit
If the function 𝑓(𝑥) be unbounded at 𝑥 = 𝑏 but bounded and continuous for all 𝑥 ∈ [𝑎, 𝑏), the
improper integral of second kind is defined as:
b −

b
f ( x)dx = lim  f ( x)dx
a  →0 a

If the limit on the right-hand side exists, the improper integral on the left hand side is convergent
otherwise divergent.
3 1
Example 10. Test the convergence of the integral ∫0 𝑑𝑥 .
√3−𝑥

We have,
3 dx 3− dx 3−

0
= lim 
3 − x  →0 0 3− x
dx = lim −2 3 − x 
 →0 0

3 dx
 = − [Link]   − 3  = 2 3 .
0
3− x  →0

1 1
Example 11. Test the convergence of the integral ∫0 𝑥(𝑥+1)
𝑑𝑥.

We have,

dx 1 1 1 
log x − log( x + 1)
1
 = lim   −  dx = lim
1
0 x( x + 1)  → 0 
 x x +1   → 0
 dx
= lim log x − log( x + 1) = lim  − log  − log 2 + log(1 +  )
1

1
0 x( x + 1)  →0  →0

dx
= lim log x − log( x + 1) = − log 2 + lim  − log  + log(1 +  )
1

1
0 x( x + 1)  →0  →0

Because the limit does not exist, the given integral is not convergent.

Integrand unbounded at the lower limit

If the function 𝑓(𝑥) be unbounded at 𝑥 = 𝑎 but bounded and continuous for all 𝑥 ∈ (𝑎, 𝑏], the
improper integral of second kind is defined as:


b
f ( x)dx = lim 
b
f ( x)dx
a  →0 a +

If the limit on the right hand side exists, the improper integral on the left hand side is convergent
otherwise divergent.
1 1
Example 12. Test the convergence of the integral ∫0 𝑑𝑥 .
√𝑥

We have,
1 dx 1 dx 1

0
= lim 
x  →0  x
dx = lim 2 x  = [Link] 1 −   = 2 .
 →0   →0

Integrand unbounded at both the ends

If only points of infinite discontinuity of the function 𝑓(𝑥) are 𝑥 = 𝑎 and 𝑥 = 𝑏, that is, 𝑓(𝑥) is
continuous for all 𝑥 ∈ (𝑎, 𝑏), the improper integral of second kind is defined as:


b
f ( x)dx =  f ( x)dx +  f ( x)dx ; a  c  b
c b

a a c
b − 2
= lim 
c
f ( x)dx + lim  f ( x)dx
1 →0 a +1  2 →0 c

If limits and the integrals on the right hand side exist, the improper integral on the left hand side is
convergent otherwise divergent.
5 1
Example 13. Examine the convergence of the improper integral ∫0 𝑥(𝑥−5)
𝑑𝑥.

Because the integrand has only infinite discontinuity at 𝑥 = 0,5, we can choose any 𝑎 ∈ (0,5), and
write
5 1 a 1 5 1
0 x( x − 5)
dx = 
0 x( x − 5)
dx + 
a x( x − 5)
dx

5 1 a 1 5− 2 1
 dx = lim  dx + lim  dx
0 x( x − 5) 1 →0 1 x( x − 5)  2 →0 a x( x − 5)

5 1 1 a 1 1 1 5 − 2  1 1
 dx = lim   −  dx + lim   −  dx
0 x( x − 5) 1 →0 5 1 x − 5

 x  2 →0 5 a
 x −5 x 
Note that
 1 1
 dx = lim  log( x − 5) − log x 
a a

1 x( x − 5) 1 → 0 5 1

1
= lim  log(a − 5) − log a − log(1 − 5) + log 1 
1 →0 5

1   a −5   
= lim log   + log  1   = 
1 →0 5
  a   1 − 5  
Therefore the given integral is not convergent.
It can be shown that the other integral also diverges.
5− 1 1
 dx = lim log( ) − log(5 −  ) − log(a − 5) + log a 
a x( x − 5)  →0 5

However, if one of the integrals on the right hand side diverges, the given integral will diverge, and
there is no need to test others.

Integrand unbounded at an interior point

If only points of infinite discontinuity of the function 𝑓(𝑥) are 𝑥 = 𝑐 ∈ (𝑎, 𝑏) and 𝑓(𝑥) is continuous
in each of the subintervals [𝑎, 𝑐) and (𝑐, 𝑏] , the improper integral of second kind is defined as:


b
f ( x)dx =  f ( x)dx +  f ( x)dx
c b

a a c
c −1
= lim  f ( x)dx + lim 
b
f ( x)dx
1 →0 a  2 →0 c +  2

If both the limits and the integrals on the right hand side exist, the improper integral on the left hand
side is convergent otherwise divergent.
1 1
Example 14. Examine the convergence of the improper integral ∫−1 𝑑𝑥.
𝑥2

The integrand has infinite discontinuity at 𝑥 = 0. So, we can write

1 0 1 1 1

1
2
dx =  2 dx +  2 dx .
−1 x −1 x 0 x

Now,

1 1 1 1 1 
0 x 2
dx = lim  2 dx = lim  − 1 = 
 → 0  x  → 0
 
Because one of the integrals on right hand side does not exist, the given integral also does not exist.
There is no need to test the other integral on the right side.

Cauchy’s Principal Value

If only points of discontinuity of the function 𝑓(𝑥) are 𝑥 = 𝑐 ∈ (𝑎, 𝑏), the improper integral of second
kind was defined as:


b
f ( x)dx =  f ( x)dx +  f ( x)dx
c b

a a c
c −1
= lim  f ( x)dx + lim 
b
f ( x)dx
1 →0 a  2 →0 c +  2
If right hand side exist for 𝜖1 = 𝜖2 , the value obtained on the right hand side is called Cauchy’s
Principal Value (CPV) of the improper integral on the left hand side. That is, the Cauchy’s principal
𝑏
Value of the improper integral ∫𝑎 𝑓(𝑥) 𝑑𝑥 is calculated as

b
CPV  f ( x)dx = lim
a  →0 ( c −

a
f ( x)dx + 
b

c +
f ( x)dx .)
𝑏
Note: If the improper integral ∫𝑎 𝑓(𝑥) 𝑑𝑥 converges, the Cauchy’s Principal Value also exists.
However, the Cauchy’s Principal Value does not guarantee the convergence of the improper integral.
1 1
Example 15. Show that the Cauchy’s Principal Value exists for the integral ∫−1 𝑥 𝑑𝑥; however, the
integral is divergent.
1
The integrand is unbounded and discontinuous at 𝑥 = 0. We can write
𝑥

1 0 1 11 − 1 11

1
dx =  dx +  dx = lim  dx + lim  dx
−1 x −1 x 0 x 1 →0 −1 x  2 →0  x

1
dx = lim  log | x |−11 + lim log | x |
−

1 1
−1 x 1 →0  2 →0 2

= lim  log | −1 | −0 + lim 0 − log |  2 |

1 →0  2 →0

1

1
dx = lim log 1 − lim log  2
−1 x 1 →0  2 →0

The limit on the right hand side does not exist and, therefore, the given improper integral diverges.
However, if we take 𝜖1 = 𝜖2 = 𝜖 (say), we get

1 1  
CPV  dx = lim log  − lim log  = lim log   = log1 = 0
−1 x  → 0  → 0  → 0
 
Therefore, the Cauchy’ Principal Value of the given improper integral is zero; however, the integral is
divergent.
1 1
Example 16. Show that the Cauchy’s Principal Value of the integral ∫−1 𝑥 2𝑛+1 𝑑𝑥 is zero if 𝑛 is a non-
negative integer; however, the integral is divergent.
1
The integrand 𝑥 2𝑛+1 is unbounded and discontinuous at 𝑥 = 0. We can write

1 − 1 1

1
dx = lim  dx + lim 
1
2 n +1 2 n +1 2 n +1
dx
−1 x 1 →0 −1 x  2 →0  x
− r
1  1  1  1 

r
dx = lim  − + lim  −
 2nx  − r  2 →0  2nx   2

−r 2 n +1 1 →0 2 n 2n
x
1  1 1  1  1 1 
=− lim  2 n − 2 n  − lim  2 n − 2 n 
2n 1 →0  1 r  2n  2 → 0  r 2 

The limits on the right hand side do not exist and, therefore, the given improper integral diverges.
However, if we take 𝜖1 = 𝜖2 = 𝜖 (say), we get

1 1  1 1  1
lim ( 0 ) = 0 .
r
CPV  dx = lim  2 n − 2 n  =
−r x 2 n +1
2n  → 0     2n  → 0

Some Important Tests for Improper Integrals

Absolutely Integrable Functions

A function 𝑓(𝑥) is called absolutely integrable over interval [𝑎, 𝑏] (finite or infinite) if the integral
𝑏
∫𝑎 |𝑓(𝑥)| exist.

Theorem
An absolutely integrable function over interval 𝐼 is also integrable over 𝐼.
Proof. We will give only an intuitive proof for this theorem.
Let a function 𝑓(𝑥) be absolutely integrable over an interval 𝐼 (finite or infinite). What does it mean?
The basic idea of the definite integrals says that the area under the curve 𝑓(𝑥) bounded by the 𝑥 −
𝑎𝑥𝑖𝑠 within the domain 𝐼 is finite and unique.
Now, we know that |𝑓(𝑥)| ≥ 𝑓(𝑥) wherever 𝑓(𝑥) is bounded. This implies that the area under the
curve |𝑓(𝑥)| is larger than the area under the curve 𝑓(𝑥).
Thus, if the larger area is finite, the smaller is obviously finite.
Hence, if the integral ∫𝐼|𝑓(𝑥)|𝑑𝑥 , ∫𝐼 𝑓(𝑥)𝑑 𝑥 must exist.
∞ sin 𝑥 ∞ sin 𝑥 ∞ 1
Example 17. The integral ∫1 𝑥2
𝑑𝑥 is convergent because the ∫1 | 𝑥2
| 𝑑𝑥 ≤ ∫1𝑥2
𝑑𝑥, and the
latter is convergent.
1
1 sin( )
Example 18. Show that the integral ∫0 𝑥 𝑝𝑥 𝑑𝑥 is convergent for 𝑝 < 1.
1
sin( )
𝑥
Let 𝑓(𝑥) = 𝑥𝑝
, 𝑝 > 0.
1
𝑥 = 0 is the only point of infinite discontinuity and sin (𝑥) changes sign in every neighbourhood of
1
𝑥 = 0. But |sin (𝑥)| ≤ 1, thus, we have
1
sin( ) 1
𝑥
|𝑓(𝑥)| = | | ≤ 𝑥𝑝 .
𝑥𝑝

1  0; p  1 1

1
Now, lim → shows that the integral dx is convergent only if 𝑝 < 1.
; p  1
p
x →0 x 0 xp
1
1 sin(𝑥)
Hence, by the absolute integrability test, ∫0 𝑥𝑝
𝑑𝑥 is also convergent for 𝑝 < 1.

Comparison test for improper integrals of first kind

Let the function 𝑓(𝑥) and 𝑔(𝑥) be bounded and continuous for 𝑥 ∈ [𝑎, ∞) such
|𝑓(𝑥) ≤ |𝑔(𝑥)| for all 𝑥 ∈ [𝑎, ∞) , then
∞ ∞
a. if ∫𝑎 𝑔(𝑥)𝑑𝑥 is convergent, ∫𝑎 𝑓(𝑥)𝑑𝑥 is also convergent.
 ∞ ∞
b. if ∫𝑎 𝑓(𝑥)𝑑𝑥 is divergent, ∫𝑎 𝑔(𝑥)𝑑𝑥 is also divergent.

Note: If you think in terms of area the Comparison Test makes a sense. If 𝑓(𝑥) is smaller than 𝑔(𝑥),
the area under the curve 𝑦 = 𝑓(𝑥) bounded by the 𝑥 − 𝑎𝑥𝑖𝑠 and the lines 𝑥 = 𝑎, 𝑥 = 𝑏 must be
smaller than the area under 𝑔(𝑥) bounded by the same boundaries. So, if the area under the larger
function is finite, the area under the smaller function must be finite. Similarly, if the area under the
smaller function is infinite, the area under the larger function must be infinite. However, be careful
that if the integral of smaller function converges there is no reason to say that the integral of the larger
will converge. Likewise, if the integral of the larger function diverges, there is no reason to conclude
that the integral of the smaller function will diverge.
3 𝑠𝑖𝑛𝑥
Example 19. Test the convergence of the improper integral ∫0 𝑑𝑥 .
√3−𝑥

3 sin x
0
3− x
dx .

We know that |𝑠𝑖𝑛𝑥| ≤ 1. If we take

sin x 1
f ( x) = , g ( x) = ,
3− x 3− x
we see that for all 𝑥 ∈ [0,3],

sin x | sin x | 1
| f ( x) |= =  =| g ( x) | .
3− x 3− x 3− x
3 1
But example 9 (page 6) shows that the integral ∫0 𝑑𝑥 is convergent, therefore, the integral
√3−𝑥
3 𝑠𝑖𝑛𝑥
∫0 𝑑𝑥 is convergent by the comparison test.
√3−𝑥

∞ cos2 𝑥
Example 20. The integral ∫2 (𝑥−1)3
𝑑𝑥 is convergent by the comparison test because |𝑐𝑜𝑠𝑥| ≤ 1
cos2 𝑥 1 ∞ 1
which implies |(𝑥−1)3 | ≤ | (𝑥−1)3 | for all 𝑥 ∈ [2, ∞), and the integral ∫2 (𝑥−1)3
𝑑𝑥 is convergent (see,
example 8, page 5).

+ 1
Example 21. Use a comparison test to show that the integral  1 xe x
dx converges.

Solution. We can see that

1 1
0 x
 x = e− x .
xe e

+ + 1
So, if  1
e− x dx converges, so does  1 xe x
dx .

+
To evaluate  1
e− x dx, , first rewrite it as a limit:

But,
 +
 e
b
e− x dx = lim −x
dx = lim (−e− x )∣1b = lim (−e−b + e) = e.
1 b →+ 1 b →+ t →+

+ + 1
That is,  1
e− x dx converges, therefore,  1 xe x
dx also converges.

Limit Comparison test for integrals of first kind

Let the function 𝑓(𝑥) ≥ 0 and 𝑔(𝑥) ≥ 0 for 𝑥 ∈ [𝑎, ∞) such that

f ( x)
lim =l.
x → g ( x)
∞ ∞
Case 1. Let 𝑙 be non-zero and finite, then both ∫𝑎 𝑓(𝑥)𝑑𝑥 and ∫𝑎 𝑔(𝑥)𝑑𝑥 converges or diverges
together.
∞ ∞
Case 2. For 𝑙 = 0, if ∫𝑎 𝑔(𝑥)𝑑𝑥 is convergent, ∫𝑎 𝑓(𝑥)𝑑𝑥 is also convergent.
∞ ∞
Case 3. For 𝑙 = ∞, if ∫𝑎 𝑔(𝑥)𝑑𝑥 is divergent, ∫𝑎 𝑓(𝑥)𝑑𝑥 is also divergent.

Example 22. Test the convergence of the integral

 x
1 ( x + 1)5/2
dx .

x x 1
Let f ( x) = , g ( x) = 5/2 = 3/2 . We can see that
( x + 1) 5/2
x x

f ( x) x x5/2
1. lim = lim .x = lim
3/2
= 1 (finite and non-zero).
x → g ( x) x → ( x + 1)5/2 x → ( x + 1)5/2

  dx
2.  g ( x)dx =  3/2 is convergent (see standard result 1., page 5)
1 1 x

∞
Therefore, by the limit comparison test, the given integral ∫1 𝑓(𝑥)𝑑𝑥 is convergent.

Example 23. Test the convergence of the integral

 x +5
1 x
dx .

x +5 x 1
Let f ( x) = , g ( x) = = 1/2 . We can see that
x x x

f ( x) x + 5 1/2  5 
1. lim = lim .x = lim 1 + =.
x → g ( x ) x → x x →
 x
  dx
2.  g ( x)dx =  1/2 is divergent (see standard result 1., page 5)
1 1 x

∞
Therefore, by the limit comparison test, the given integral ∫1 𝑓(𝑥)𝑑𝑥 is divergent.
Example 24. Test the convergence of the integral
 x sin x
1 (1 + x)3
dx .

x sin x x
Because  for all 𝑥 ≥ 1, we will first test the convergence of the integral
(1 + x) 3
(1 + x)3

 x
1 (1 + 2 x)3
dx .

x x 1
Let f ( x) = , g ( x) = 3 = 2 . We can see that
(1 + 2 x) 3
x x

f ( x) x 1
1. lim = lim .x 2 = (finite and non-zero).
x →
g ( x) x → (1 + 2 x) 3
2
  dx
2.  g ( x)dx =  2 is convergent (see standard result 1., page 5)
1 1 x

∞
Therefore, by the limit comparison test, the given integral ∫1 𝑓(𝑥)𝑑𝑥 is convergent.

Test for Improper Integrals of Third Kind

We know that the improper integrals of the third kind are the integrals with infinite domain of
integration as well as unbounded integrand, that is, these integrals can be thought of as some
combination of the first and second kind improper integrals and proper integrals. Due to this reason,
these integrals are written as a sum of integrals of the first and second kind, and analyzed with the
strategies discussed earlier.
For an illustration, consider the following example.
Example 25. Test the convergence of the integral
 sin x
0 x
dx

This is an improper integral of the third kind because the domain of integration is [0, ∞) and the
integrand is also discontinuous at the lower limit. Thus, we can re-write this integral as
 sin x 1 sin x  sin x
0 x
dx = 
0 x
dx + 
1 x
dx = I1 + I 2

Note that lim(sin x / x) = 1 , which means that the lower limit is not the point of infinite
x →0

discontinuity (function is bounded). Therefore, 𝐼1 is a proper integral, therefore, convergent. But

 sin x
I2 =  dx
1 x
1
is improper integral of the first kind. Again, note that |sin 𝑥| ≤ 1 and 𝑥
is bounded in [1, ∞),
therefore, 𝐼2 is convergent.
∞ sin 𝑥
Hence, as a sum of two convergent integrals ∫0 𝑥
𝑑𝑥 is also convergent.
Example 26. Test the convergence of the integral
 x
0 x sin 2 x + 1
4
dx

Motivation. Note that sin 𝑥 is a periodic function and vanishes at 𝑛𝜋(𝑛 = 0, ±1, ±2, ±3, … ).
Therefore, whenever 𝑥 = 𝑛𝜋(𝑛 = 0,1,2,3, … ), the denominator will remain 1+0=1, and the integrand
will become 𝑛𝜋. Because this will happen infinitely many times in the domain of integration, the area
under the curve will become infinite. Therefore, we can take a comparison function whose integral in
this domain diverges.
Let us take

x 1
f ( x) = and g ( x) = .
x sin x + 1
4 2
x
Now, we observe two things,

f ( x) x2
1. lim = lim 4 2 = 0.
x → g ( x) x → x sin x + 1

  dx
2.  g ( x)dx =  is divergent (see standard result 1., page 5)
0 0 x

∞ ∞
By the limit comparison test, convergence of ∫0 𝑓(𝑥)𝑑𝑥 implies the convergence ∫0 𝑔(𝑥)𝑑𝑥 , but
∞ ∞
∫0 𝑔(𝑥)𝑑𝑥 is divergent, therefore, ∫0 𝑓(𝑥)𝑑𝑥 is also divergent.

Hence, the given integral is divergent.

Other Solved Examples.

Example [27]. Determine if the following integral is convergent or divergent.
 1
 3 x + ex
dx

Let’s first take a guess about the convergence of this integral. As noted after the fact in the last section
about
 1
 a xp
dx

1
if the integrand goes to zero faster than x then the integral will probably converge. Now, we’ve got an
exponential in the denominator which is approaching infinity much faster than the xx and so it looks
like this integral should probably converge.
So, we need a larger function that will also converge. In this case we can’t really make the numerator
larger and so we’ll need to make the denominator smaller in order to make the function larger as a
whole. We will need to be careful however. There are two ways to do this and only one, in this case
only one, of them will work for us.
First, notice that since the lower limit of integration is 3 we can say that 𝑥 ≥ 3 > 0 and we know that
exponentials are always positive. So, the denominator is the sum of two positive terms and if we were
to drop one of them the denominator would get smaller. This would in turn make the function larger.
The question then is which one to drop? Let’s first drop the exponential. Doing this gives,
 1 1

x+e x
x
This is a problem however, since
 1
 3 x
dx

diverges by the fact. We’ve got a larger function that is divergent. This doesn’t say anything about the
smaller function. Therefore, we chose the wrong one to drop.
Let’s try it again and this time let’s drop the 𝑥.

1 1
 x = e− x
x+e x
e
Also,

e− x dx = lim  e− x dx = lim ( −e−t + e−3 ) = e−3



t

3 t → 3 t →


So,  3
e− x dx is convergent. Therefore, by the Comparison test, the given integral is also convergent.

Example 22. Determine if the following integral is convergent or divergent.

 1 + 3sin 4 ( 2x )
 1
x
dx

First notice that the numerator in this function is going to be bounded since the sine is never larger
than 1. Therefore, since the exponent on the denominator is less than 1 we can guess that the integral
will probably diverge. We will need a smaller function that also diverges.

We know that 0  3sin ( 2 x )  3 . In particular, this term is positive and so if we drop it from the
4

numerator the numerator will get smaller. This gives,

1 + 3sin 4 ( 2 x ) 1  1
x

x
and  1
x
dx

diverges so by the Comparison Test, the given integral also diverges.

Example 23. Determine if the following integral is convergent or divergent.

 e− x dx
2

We know that exponentials with negative exponents die down to zero very fast so it makes sense to
guess that this integral will be convergent. We need a larger function, but this time we don’t have a
fraction to work with so we’ll need to do something different.
2
We’ll take advantage of the fact that e−x is a decreasing function.
2 2 1
Now, note that for all 𝑥 ∈ [1, ∞), x 2 > 𝑥. That means ex > ex > x 2, that is, e−x < .
x2
∞ 1 ∞ 2
Because ∫1 dx is convergent, the integral ∫1 e−x dx is convergent by the comparison test..
x2
Example 23. Determine if the following integral is convergent or divergent.

 1 + cos 2 ( x )
 2
x  2 − sin 4 ( x ) 
dx

In this case we can notice that because the cosine in the numerator is bounded the numerator will
never get too large. Likewise, the sine in the denominator is bounded and so again that term will not
get too large or too small.
That leaves only the square root in the denominator and because the exponent is less than one we can
guess that the integral will probably diverge. Therefore, we will need a smaller function that also
diverges.

We know that 0  cos ( x )  1 . In particular, this term is positive and so if we drop it from the
2

numerator the numerator will get smaller. This gives,

1 + cos 2 ( x ) 1

x  2 − sin 4 ( x ) x  2 − sin 4 ( x )

Next, we also know that 0  sin ( x )  1 . Again, this is a positive term and so if we no longer
4

subtract this off from the 2 the term in the brackets will get larger and so the rational expression will
get smaller. This gives,

1 + cos2 ( x ) 1 1
 
x  2 − sin 4
( x ) x  2 − sin ( x ) 2 x
4

∞ 1
Finally, because ∫1 𝑑𝑥 is divergent, the given integral is also divergent by the comparison test.
√(𝑥)

Exercise
Test the convergence of the following integrals

1

1
1. dx Divergent
x0

x
2. 
2/3
dx Convergent
1/3
3x − 1
4 x2 + 2x + 1
3.  dx Convergent
2
( x − 2)(4 − x)
 1
4.  dx Convergent
1
( x + 1) x 2 + 1

5. 2
x cos2 xdx Divergent
x tan −1 x

6. 0 (1 + x4 )1/5 dx Divergent
 log x
7.  dx Divergent
1 x
 
log x
8.
1 x2
dx Convergent
b 1
9.  dx Convergent
a ( x − a)( x − b)

1 1
10.  2 dx Divergent.
0 x (1 + x 2 )

tan −1 x

11.  dx Divergent
1 1 + x2

−1
 tan x
12.  dx Divergent
0 1+ x 2

−1
1 sin x
13.  dx Convergent
2/3
1 − x2
 x
14. 0 sin x
dx Divergent
 /4 1
15. 0 tan x dx Convergent.
 1
16. 1 x3 + 1dx Convergent
2
 z
17.  3 z 3 − 1dz Divergent
−y
 e
18.  4 y dy Convergent
 w2 + 1

1
19. dw Divergent. [Hint. Take 𝑔(𝑥) = 𝑤 and apply limit comparison test.]
6
w ( cos ( w ) + 1)
3 2

 x −1
20.  1 x + 2 x2
4
dx Convergent.

Special Integrals
Theorem
∞
The integral ∫0 𝑥 𝑛−1 𝑒 −𝑥 𝑑𝑥 is convergent if 𝑛 > 0 (𝑛 is a real number, not necessarily an integer.)

Proof. Because 𝑒 𝑥 ≥ 1 for all 𝑥 ∈ [0, ∞), we have 0 < 𝑒 −𝑥 ≤ 1 for all 𝑥 ∈ [0, ∞). Thus, 0 ≤
𝑥 𝑛−1 𝑒 −𝑥 ≤ 𝑥 𝑛−1 for all 𝑥 ∈ [0, ∞).

Now, take 𝑓(𝑥) = 𝑥 𝑛−1 𝑒 −𝑥 and 𝑔(𝑥) = 𝑥 𝑛−1 . We see that

f ( x) x n−1e− x
1. lim = lim n −1 = 0 .
x → g ( x) x → x

   1
2. 0
g ( x)dx =  x n−1dx = 
0 0 x
dx is convergent for 1 − 𝑛 < 1 ⇒ 𝑛 > 0. (see standard
1− n

result 2, page 5)
 1
Therefore, by the limit comparison test, ∫0 𝑥 𝑛−1 𝑒 −𝑥 𝑑𝑥 is convergent for 𝑛 > 0.

The integral discussed in this theorem is of great importance, which we will study in detail at a
later point of time. This integral is also known as Gamma Function.

Definition (Gamma Function)

Let 𝑛 > 0 be a real number. The Gamma function of 𝑛 is defined as

n =  e− x x n−1dx .
0

Theorem
1
The integral ∫0 𝑥 𝑛−1 (1 − 𝑥)𝑚−1 𝑑𝑥 is convergent for 𝑛 > 0, 𝑚 > 0 only.

Proof. For 𝑚 ≥ 1, 𝑛 ≥ 1, the given integral is a proper integral.

However, if 𝑛 < 0, there is a discontinuity at 𝑥 = 0. And if 𝑚 < 1, there is a discontinuity at 𝑥 = 1.
Thus, we must check convergence at both the ends for 𝑛 < 1, 𝑚 < 1.
CASE 1: Convergence at 𝑥 = 0. [𝑚 ≥ 1].

Take 𝑓(𝑥) = 𝑥 𝑛−1 (1 − 𝑥)𝑚−1 and 𝑔(𝑥) = (1 − 𝑥)𝑚−1 .

Because 𝑔(𝑥) has no discontinuity and 0 ≤ 𝑔(𝑥) ≤ 1, we have 𝑓(𝑥) ≤ 𝑥 𝑛−1 ∀𝑥 ∈ [0,1].
1
The discontinuity at 𝑥 = 0 occurs if 𝑛 − 1 < 0 ⇒ 1 − 𝑛 > 0. Thus, 𝑥 𝑛−1 = 𝑥 1−𝑛 . By the standard
1 1 1
result 2 (page 5), the integral ∫0 𝑥 𝑛−1 𝑑𝑥 = ∫0 𝑥 1−𝑛
𝑑𝑥 is convergent for 1 − 𝑛 < 1 ⇒ 𝑛 > 0 only.
1 𝑛−1
Therefore, by the comparison test, the integral ∫0 𝑥 (1 − 𝑥)𝑚−1 𝑑𝑥 is convergent for 𝑛 > 0 only.

CASE 2: Convergence at 𝑥 = 1. [𝑛 ≥ 1].

Take 𝑓(𝑥) = 𝑥 𝑛−1 (1 − 𝑥)𝑚−1 and 𝑔(𝑥) = 𝑥 𝑛−1 .

Because 𝑔(𝑥) has no discontinuity and 0 ≤ 𝑔(𝑥) ≤ 1, we have 𝑓(𝑥) ≤ (1 − 𝑥)𝑚−1 ∀𝑥 ∈ [0,1].
1
The discontinuity at 𝑥 = 1 occurs if 𝑚 − 1 < 0 ⇒ 1 − 𝑚 > 0. Thus, (1 − 𝑥)𝑚−1 = (1−𝑥)1−𝑚 . By
1 1 1
the standard result 2 (page 5), the integral ∫0 (1 − 𝑥)𝑚−1 𝑑𝑥 = ∫0 (1−𝑥)1−𝑚
𝑑𝑥 is convergent for 1 −
1
𝑚 < 1 ⇒ 𝑚 > 0 only. Therefore, by the comparison test, the integral ∫0 𝑥 𝑛−1 (1 − 𝑥)𝑚−1 𝑑𝑥 is
convergent for 𝑚 > 0 only.
CASE 3: Convergence at both the ends
Let 𝑚 < 1, 𝑛 < 1. Take a real number 𝑐 ∈ (0,1) and we can write
1 𝑐 1
∫ 𝑥 𝑛−1 (1 − 𝑥)𝑚−1 𝑑𝑥 = ∫ 𝑥 𝑛−1 (1 − 𝑥)𝑚−1 𝑑𝑥 + ∫ 𝑥 𝑛−1 (1 − 𝑥)𝑚−1 𝑑𝑥
0 0 𝑐

Because the first integral on the right-hand side has no discontinuity at the upper limit and (1 −
𝑥)𝑚−1 is bounded in [0, 𝑐], therefore by the CASE 1, the integral is convergent only for 𝑛 > 0.

Similarly, the second integral on the right-hand side has no discontinuity at the lower limit and 𝑥 𝑛−1
is bounded in [𝑐, 1], therefore by the CASE 2, the integral is convergent only for 𝑚 > 0.
 1
Hence, the integral ∫0 𝑥 𝑛−1 (1 − 𝑥)𝑚−1 𝑑𝑥 is convergent for 𝑛 > 0, 𝑚 > 0 only.

The integral in this theorem is also of great importance. This integral is known as Beta Function.

Definition (Beta Function)

Let 𝑛 > 0, 𝑚 > 0 be a real numbers. The Beta function of 𝑛, 𝑚 is defined as
1
 (n, m) =  x n−1 (1 − x)m−1 dx .
0

∞
Example 28. The integral ∫0 𝑥𝑒 −𝑥 𝑑𝑥 is convergent. In fact, this is Gamma function for 𝑛 = 2.
∞1 ∞
Example 29. The integral ∫0 𝑒 −𝑥 𝑑𝑥 is divergent. (∫0 𝑥 𝑛−1 𝑒 −𝑥 𝑑𝑥 for 𝑛 = 0)
𝑥
1 3 5
Example 30. ∫0 √𝑥 (1 − 𝑥)3/2 𝑑𝑥 is convergent. (Beta function for 𝑛 = 2 , 𝑚 = 2.
1 𝑥 1
Example 31. ∫0 (1−𝑥)2
𝑑𝑥 is convergent. (∫0 𝑥 𝑛−1 (1 − 𝑥)𝑚−1 𝑑𝑥 for 𝑛 = 2, 𝑚 = −1).

Exercise
Test the convergence of the following integrals.

xn
0 1 − x dx Divergent
1
1.
n
1 x
2.  dx Convergent for 𝑛 ≻ 1
0 1+ x

3 x +1
2
3.  2 dx Divergent
2 x −4

n
1 x log x
4.  dx Convergent for 𝑛 ≻ 1
0 (1 + x) 2

 e−1/ x
5. 0 x6
dx Convergent
 /2
6. 0
sin p −1 x cosq −1 xdx Convergent for 𝑝 > 0, 𝑞 > 0. (Solved problem in Beta functions).
 x
7. 0 sin x
dx Divergent (Hint: infinite discontinuity at both the ends).
Gamma and Beta Functions
Introduction
In the last chapter, we studied the convergence and divergence of improper integral. We also defined
two special improper integrals as the gamma function and the beta function. In this chapter, we will
learn the properties, interrelation, and applications of these two functions.

Gamma Function
We have already defined the gamma function as

n =  e− x x n−1dx; n  0
0

Note: The definition of gamma function can be extended for all negative numbers except negative
integers and zero.

Theorem. Other form of gamma function

Statement. Gamma Function can be transformed to the following form:
n −1
  1 
n =  log   
1
dx
0
  x 

Proof. We know that


n =  e−t t n−1dt; n  0
0

Taking the transformation e−t = x, so that − t = log x or t = log(1/ x) and dt = −dx / x , and from
e−t = x, limit of integration become x = 0 to 1 . Thus, we have

=  log ( x −1 ) dx
n −1 dx n −1
n =  x log (1/ x )
1 1

0 x 0

Therefore,
n −1
  1 
n =  log   
1
dx .
0
  x 

Properties of gamma function

Basic properties
(i) (1) = 1
(ii) (n + 1) = n(n); n  0
Proof

(i) By the definition of gamma function, we have


n =  e− x x n−1dx; n  0
0
 Put n = 1 , we get
b
e− x
= lim (1 − e−b ) = 1 − 0 = 1

1=  e x dx = lim  e x dx = lim
b
−x 0 −x 0
0 b → 0 b → −1 b →
0

(ii) By the definition of gamma function, we have,

 
(n + 1) =  e − x x ( n+1)−1 dx =  e − x x n dx.
0 0

Integration by parts taking e − x as second function.


(n + 1) = {− x n e− x }0 −  nxn−1 (−e − x )dx
0

Now

xn xn
lim x = lim =0
x → e x → x 2 x3 xn
1 + x + + + ..... +
2! 3! n! ,
∞
Therefore, 𝛤(𝑛 + 1) = 𝑛 ∫0 𝑒 −𝑥 𝑥 𝑛−1 𝑑𝑥 = 𝑛𝛤(𝑛)

Integer property of gamma function

Γ(𝑛 + 1) = 𝑛! ; 𝑛 = 1,2,3, …
Proof. We know that,
(n + 1) = n(n); n  0

Put 𝑛 = 1,2,3 … in property we get,

 2 = 11 = 1 = 1!
3 = 2 2 = 2.1 = 2!
 4 = 33 = 3.2. 2 = 3.2.1 = 3!
Now, let
(k + 1) = k !

for some positive integer k, we shall show that the relation also holds for k+1.

(k + 1 + 1) = (k + 1)(k + 1) = (k + 1)k ! = (k + 1)! .

Therefore, by induction principle, the relation (k + 1) = k ! holds for all positive integers.

That is,
(n + 1) = n! ; n = 1, 2,3,

Hence Proved!
 1
Example 32. Evaluate
2
Solution. We know that

n =  e−t t n−1dt
0

1
Taking 𝑛 = , we get
2

 e−t
1/ 2 =  dt
0
t

Let us consider the transformation t = x so that t = x2 , dt = 2 xdx, and we have


1/ 2 = 2 e− x dx .
2

Now consider,

( 1/ 2)
2
( 
= 2 e− x dx 2 e− x dx = 4
0
2

)( 

0
2

) 0


0

e− x e− y dxdy = 4
2 2

0


0

e− ( x
2
+ y2 )
dxdy

Taking x = r cos  , y = r sin  , we have x 2 + y 2 = r 2 , dxdy = rdrd and the limits of

integration become r = 0 to  for radial variable, and  = 0 to  / 2 . Thus, we have

( 1/ 2)
2
= 4
0


 /2

0
e− r rd dr = 4
2

(
0
 /2
d )(  e 0

−r2
rdr )
But, taking r 2 = z, and therefore, 2rdr = dz , we have

 1  −z 1  /2 
 e− r rdr =  e dz = , and  d =
2

0 2 0 2 0 2
Thus,

   1 
( 1/ 2)
2
= 4    = 
 2  2 

1/ 2 =  .

Example 33. Show that

 /2
n 1 − n = 2 tan 2 n−1  d .
0

Proof. We know that


n =  e−t t n−1dt
0

Put 𝑡 = 𝑥 2 so that 𝑑𝑡 = 2𝑥𝑑𝑥, and we get

  
n =  e− x x 2 n−2 2 xdx = 2 x 2 n−1e− x dx
2 2

0 0

Now,
   
n m = 4 x 2n−1e− x dx  y 2m−1e− y dy = 4  x 2n−1 y 2m−1e− ( x + y2 )
2 2 2
dxdy
0 0 0 0

Put 𝑚 = 1 − 𝑛, so that
 
n 1 − n = 4  x 2 n−1 y1−2 n e−( x + y2 )
2
dxdy
0 0

Taking x = r cos  , y = r sin  , we have x 2 + y 2 = r 2 , dxdy = rdrd and the limits of

integration become r = 0 to  for radial variable, and  = 0 to  / 2 . Thus, we have
  /2
0 ( r sin  ) ( r cos  )
2 n −1 1− 2 n − r 2
n 1 − n = 4 e rdrd
0
  /2
r 2 n −1− 2 n +1 ( sin  ) ( cos  )
2 n −1 1− 2 n − r 2
=4   e rdrd
0 0
  /2
=2  2re − r dr  sin 2 n −1  cos 2 n −1  d
2

0 0

But, from Example 6 (page 3), we know that

 
 2re− r dr =  e− z dz = 1; (by taking r 2 = z )
2

0 0

 /2
 n 1 − n = 2 sin 2n−1  cos1−2 n  d
0

Proof. We know that


n =  e−t t n−1dt
0

Put 𝑡 = 𝑥 2 so that 𝑑𝑡 = 2𝑥𝑑𝑥, and we get

 
n =  e− x x 2 n−2 2 xdx = 2 x 2 n−1e− x dx
2 2

0 0

Now,
   
n m = 4 x 2n−1e− x dx  y 2m−1e− y dy = 4  x 2n−1 y 2m−1e− ( x + y2 )
2 2 2
dxdy
0 0 0 0

Put 𝑚 = 1 − 𝑛, so that
 
n 1 − n = 4  x 2 n−1 y1−2 n e−( x + y2 )
2
dxdy
0 0

Taking x = r cos  , y = r sin  , we have x 2 + y 2 = r 2 , dxdy = rdrd and the limits of

integration become r = 0 to  for radial variable, and  = 0 to  / 2 . Thus, we have
   /2
 ( r sin  ) ( r cos  ) e rdrd
2 n −1 1− 2 n − r 2
n 1 − n = 4
0 0
 
( sin  ) ( cos  ) e rdrd
2 n −1 1− 2 n − r 2
=4   r
/2
2 n −1− 2 n +1
0 0
 
=2  2re dr  sin  cos  d
/2
−r2 2 n −1 2 n −1
0 0

But, from Example 6 (page 3), we know that

 
 2re− r dr =  e− z dz = 1; (by taking r 2 = z )
2

0 0

 /2
 n 1 − n = 2 sin 2n−1  cos1−2 n  d
0

Example 34. Using example 33, show that (0) is not defined.

Solution. We know that

 /2
n 1 − n = 2 sin 2n−1  cos1−2 n  d
0

Put 𝑛 = 1,
 /2 sin 
d =  −2log cos  0
 /2
1 0 = 2
0 cos 
Which is not defined because 𝑙𝑜𝑔(0) is not defined quantity.

Example 35. Using example 33, show that (n) =  .

Solution. We know that

 /2  /2
n 1 − n = 2 sin 2n−1  cos1−2n  d = 2 tan 2n−1  d
0 0

Put 𝑛 = 1/2, we get

11  /2
= 2 d = 2  d 0 = 
 /2

22 0

1
 = 
2

1
Example 36. Evaluate −
2
Solution. We know
(n + 1) = n(n)

For n = −1/ 2 (see “note” to the definition of gamma function), we get

(1/ 2) = (−1/ 2)(−1/ 2)

(−1/ 2) = −2(1/ 2) = −2  .
Example 37. Evaluate the improper integral

0
xe−3 x dx

Solution. Let

I =  x1/2e−3 x dx
0

1 2 2
Take 3 x = t  x = t , dx = tdt , so that
9 9
 t −t 2 2  −t 2 2 2 4
I = e tdt =  e t dt = (3) = 2! =
0 3 9 27 0 27 27 27
Example 38. Evaluate the improper integral

 xe− x dx
3

Solution. Let

I = xe− x dx
3

Take 3
x = t  x = t 3 , dx = 3t 2dt , so that
  7531 315
I =  t 3/2e−t 3t 2 dt = 3 e−t t 7/2 dt = 3(9 / 2) = 3. (1/ 2) = .
0 0 2222 16
Example 39. Evaluate the improper integral

 0
xc c − x dx

where, 𝑐 is a constant.
Solution. Let

I =  xc c − x dx
0

1
Take c − x = e−t  x log c = t , dx = dt , so that
log c
c c +1 c
  t  −t 1  1    1 
I =   e dt =   
−t c
e t dt =   (c + 1); c  0.
0
 log c  log c  log c  0
 log c 
Example 40. Show that
 (n)
0
x n−1e− x dx =
n
Proof. Let
 
I =  x n−1e− x dx
0

Take  x = t so that x = t /  , dx = dt /  and we have

n −1
 t  dt 1   ( n)
I =   e−t =  t n −1e−t dt = .
    n
0 n 0

dxdy =  .
 x2
− y2 2 e
Example 41. Show that
0 ye
2y y  2 2

Solution. Let
 
I = ye− y dy  ye− y dy
2 2

0 0

1
Put 𝑦 2 = 𝑡 ⇒ 𝑦 = 𝑡1/2 , 𝑑𝑦 = 2 𝑡 −1/2 𝑑𝑡, so that

 1 −1/ 2  1
I =  t 1/ 4e−t t dt   e−t t −1/ 4 t −1/ 2 dt
0 2 0 2

1  −t −1/ 4 1 
= 
2 0
e t dt   e −t t −3 / 4 dt
2 0

1  −t (3 / 4−1)  −t 14 −1
=  e t 0 e t dt
4 0
1
= (3 / 4)  (1 / 4)
4
1
= (1 / 4)(1 − 1 / 4)
4
1 
=
4 4 sin  / 4
  
 n(1 − n) = sin n 
 

1  2 2
=  = 
4 1 4 2
2


= =
2 2 R.H.S.

Example 42. Evaluate the improper integral

 ∞
𝑥𝑑𝑥
∫ .
0 1 + 𝑥6
 xdx
I =
Let 0 1 + x6
1 −5 / 6
x 6 = y or x = y1/ 6  dx = y dy
Let 6

1  y1/ 6 y −5 / 6 1  y (1/ 3−1)

6 0 6 0 (1 + y)1/ 3+ 2 / 3)
I= dy = dy
1+ y

1 2 1  1
   1 − 
1 1 2 1
= B ,  = 3 3 = 1 3  3 = 1  
6 3 3 6 1 2 6 1 6 sin 
 + 
3 3 3

1  2 
=  = =
6 ( )
3/ 2 6 3 3 3

Beta Function
While discussing some special improper integrals, we have, we defined the beta function (Page 19) as
1
𝛽(𝑚, 𝑛) = ∫ 𝑥 𝑚−1 (1 − 𝑥)𝑛−1 𝑑𝑥 ; 𝑚 > 0, 𝑛 > 0.
0

In this section, we shall study properties and applications of the beta function.

Properties of Beta Function

Property of Symmetry
Statement. Beta function is symmetrical in m and n, that is,  (m, n) =  (n, m)

Proof: we have by definition

B(m, n) =  x m−1 (1 − x) n−1 dx,

1
m  0, n  0
0

1
 a f ( x)dx = a f (a − x)dx
=  x m−1{1 − (1 − x)}n−1 dx  0 0 
0

1
=  (1 − x) m−1 x n−1dx
0

1
=  x n−1 (1 − x)}m−1 dx = B(n, m)
0

 B(m, n) = B(n, m)
Relation between beta and gamma function
Statement. For any positive number m > 0, n > 0,
Γ(𝑚)Γ(𝑛)
𝛽(𝑚, 𝑛) = .
Γ(𝑚 + 𝑛
Proof : We know that

m 
m
=  e − zx x m−1 dx
z 0

 
m = z m  e − zx x m−1dx =  z me − zx x m−1 dx
 0 0

− z n −1
Multiplying both sides by e z , we get

m e− z z n−1 =  e− z (1+ x ) z m+ n−1 x m−1dx
0 (1)
Now integrate both sides of (2) with respect to z from 0 to  , we get

m e − z z n−1dz = 

  e − z (1+ x ) z m+ n−1 x m−1 dx dz
0 0 0 

mn = 

  e − z (1+ x ) z m+ n −1dz x m−1dx
or 0 0 

Using (1)
 (m + n) m−1  x m−1
mn =  x dx =  ( m + n ) 0 (1 + x)m+n dx
0 (1 + x) m+ n

= (m + n) B(m, n)

mn
B(m, n) =
 (m + n)

Another form of beta function

Statement. Beta function can be transformed to the following form
∞ ∞
𝑥 𝑚−1 𝑥 𝑛−1
𝛽(𝑚, 𝑛) = ∫ 𝑑𝑥 = ∫ 𝑑𝑥 .
0 (1 + 𝑥)𝑚+𝑛 0 (1 + 𝑥)
𝑚+𝑛

Proof: We know that

1
B(m, n) =  x m−1 (1 − x) n−1 dx
2 (1)
1 1
x= dx = − dy
Put (1 − y ) in (1) and (1 + y) 2

When x = 0, y =  when x = 1 , y = 0
 n −1
1  1   1 
B(m, n) = 
0
. 1−
m −1    (1 + y ) 2  dy

 (1 + y )  1+ y   

1 y n−1
=
0
. dy
 (1 + y ) m −1+ 2 (1 + y ) n −1

y n −1
=
0
dy
 (1 + y ) m+ n
 a a
f (t ) dt 
 
0
f ( x) dx = 
0 

x n−1
B(m, n) = 
1
dx
0 (1 + x) m + n
(2)
Since Beta function is symmetrical in m and n, we have

x m−1
B(m, n) = 
1
dx
0 (1 + x) m + n
.

Beta function as a sine & cosine integral

Statement. For any pair of positive numbers m > 0, n > 0,

B(m, n) = 2 sin 2m−1  cos2n−1  d
0

Proof. We know that

1
𝛽(𝑚, 𝑛) = ∫ 𝑥 𝑚−1 (1 − 𝑥)𝑛−1 𝑑𝑥 ; 𝑚 > 0, 𝑛 > 0.
0

Take the transformation of x as x = sin 2  so that 𝑑𝑥 = 2𝑠𝑖𝑛𝜃𝑐𝑜𝑠 𝜃𝑑𝜃. The limits are
changed to 𝜃 = 0 𝑡𝑜 𝜋/2, and we have
 /2
B(m, n) =  (sin 2  ) m−1 (cos2  ) n−1 (2 sin  cos ) d
0

 /2
 (m, n) = 2 sin 2 m−1  cos 2 n−1  d .
0

Evaluation of a special integral

Statement. For any pair of numbers p > -1, q > -1,

𝑝+1 𝑞+1
𝜋/2 Γ( 2 )Γ( 2 )
𝑝 𝑞
∫ 𝑠𝑖𝑛 𝜃 𝑐𝑜𝑠 𝜃 𝑑𝜃 = .
𝑝+𝑞+2
0 2Γ ( 2 )

Proof. We just proved,

 𝜋/2
𝛽(𝑚, 𝑛) = 2 ∫0 𝑠𝑖𝑛2𝑚−1 𝜃 𝑐𝑜𝑠 2𝑛−1 𝜃 𝑑𝜃.

Put p = 2m-1, q = 2n-1, we get

1
1 𝑝+1 𝑞+1
∫ 𝑠𝑖𝑛𝑝 𝜃 𝑐𝑜𝑠 𝑞 𝜃 𝑑𝜃 = 𝛽( , )
0 2 2 2

Using relation between beta and gamma function, we get

𝑝+1 𝑞+1
1 Γ( 2 )Γ( 2 )
∫ 𝑠𝑖𝑛𝑝 𝜃 𝑐𝑜𝑠 𝑞 𝜃 𝑑𝜃 = .
𝑝+𝑞+2
0 2Γ ( 2 )
Note. If p and q are the odd positive integers of the form p = 2M-1, q = 2N-1, this formula takes the
form
1
M! N!
∫ 𝑠𝑖𝑛𝑝 𝜃 𝑐𝑜𝑠 𝑞 𝜃 𝑑𝜃 = .
0 (M + N)!

An interesting formula
𝜋
Γ(𝑛)Γ(1 − 𝑛) =
𝑠𝑖𝑛 𝑛 𝜋
Proof. We know that
∞
𝑥 𝑛−1 Γ(m)Γ(n)
𝛽(𝑚, 𝑛) = ∫ 𝑑𝑥 =
0 (1 + 𝑥)𝑚+𝑛 Γ(m + n)

Put 𝑚 + 𝑛 = 1, so that 𝑚 = 1 − 𝑛, and we have

∞
𝛤(𝑛)𝛤(1 − 𝑛) 𝑥 𝑛-1 𝜋
=∫ 𝑑𝑥 =
𝛤(1) 0 (1 + 𝑥) 𝑠𝑖𝑛 𝑛 𝜋

Using Γ(1) = 1, and the standard result

∞
𝑥 𝑛-1 𝜋
∫ 𝑑𝑥 =
0 (1 + 𝑥) 𝑠𝑖𝑛 𝑛 𝜋

We get
𝜋
Γ(𝑛)Γ(1 − 𝑛) = .
𝑠𝑖𝑛 𝑛 𝜋
Example 43. Evaluate
𝜋/2
∫ √𝑡𝑎𝑛 𝑥 𝑑𝑥
0

Solution.
𝜋/2 𝜋/2
∫ √𝑡𝑎𝑛 𝑥 𝑑𝑥 = ∫ 𝑠𝑖𝑛1/2 𝑥 𝑐𝑜𝑠 −1/2 𝑥 𝑑𝑥
0 0

Using formula
 𝑝+1 𝑞+1
1 Γ( 2 )Γ( 2 )
𝑝 𝑞
∫ 𝑠𝑖𝑛 𝜃 𝑐𝑜𝑠 𝜃 𝑑𝜃 =
𝑝+𝑞+2
0 2Γ ( 2 )

with 𝑝 = 1/2, 𝑞 = −1/2, we get

3 1
𝜋/2 Γ (4) Γ (4) 1 3 1
∫ √𝑡𝑎𝑛 𝑥 𝑑𝑥 = = Γ( )Γ( )
0 2Γ(1) 2 4 4

Using
𝜋
Γ(𝑛)Γ(1 − 𝑛) =
𝑠𝑖𝑛 𝑛 𝜋
3 1 𝜋
with 𝑛 = 1/4, we get Γ ( ) Γ ( ) = = 𝜋√2. Therefore,
4 4 𝑠𝑖𝑛(𝜋/4)

𝜋/2
1 𝜋
∫ √𝑡𝑎𝑛 𝑥 𝑑𝑥 = 𝜋√2 = .
0 2 √2
Example 44. Evaluate
𝜋/2
∫ 𝑠𝑖𝑛3/2 𝑥 𝑑𝑥
0

Solution.
𝜋/2
∫ 𝑠𝑖𝑛3/2 𝑥 𝑑𝑥
0

Using formula
𝑝+1 𝑞+1
1 Γ( 2 )Γ( 2 )
𝑝 𝑞
∫ 𝑠𝑖𝑛 𝜃 𝑐𝑜𝑠 𝜃 𝑑𝜃 =
𝑝+𝑞+2
0 2Γ ( 2 )

with 𝑝 = 1/2, 𝑞 = 0, we get

3 1 3
𝜋/2 Γ (4) Γ (2) √𝜋 Γ (4)
3/2
∫ 𝑠𝑖𝑛 𝑥 𝑑𝑥 = = .
0
5 2 Γ (5)
2Γ (4) 4

Duplication Formula
Form 1.
1 𝛤(2𝑚)
𝛤(𝑚)𝛤 (𝑚 + ) = 2√𝜋 ( 2𝑚 ) , 𝑚 > 0.
2 2
Form 2.
1 1
𝛽(𝑚, 𝑛) = 21−2𝑚 𝛽 (𝑚, ) = 21−2𝑛 𝛽 (𝑛, ).
2 2
Proof : We know that
 (m)(n)
 (m, n) = , m  0, n  0
(m + n)

Taking m = n, we get

[(m) 2
B(m, m) =
(2m) …(1)
By the definition of Beta function, we have
1
B(m, m) =  x m−1 (− x) m−1 dx
0

x = sin 2   dx = 2 sin  cos d

1
= 
When x = 0 then  = 0 and when x = 1 then 2

 /2
B(m, m) =  sin 2( m−1)  . cos2( m−1)  .2 sin  cos d
Then 0

 /2 x/2
= 2 sin 2m−1  . cos2m−1  d = 2 (sin  cos ) 2m−1 d
0 0

2 m −1
 /2  sin 20  1 x/2
= 2   d = 2 m−2  sin 2 m−1 2 d
0
 2  2 0

1 d 1

x
= 2 m−2
sin 2 m−1 . , 20 =   d = d
2 0 2 Put 2

1  1  /2
=
2 2 m−1  0
sin 2 m−1  d =
2 2 m−1
 2
0
sin 2 m−1  d

1  /2
=
2 2 m−2  0
sin 2 m−1 . cos0  d

1 1 1
 (2m − 1 + 1) (0 + 1)  ( m )  
=
1
 2 2 1
= 2 m −1  2
2m−2
2 1
2 (2m − 1 + 0 + 2) 2  1
 m + 
2  2

1  ( m) 
= 2 m −1

2  1
 m + 
 2 …(2)

 1 
  2  =  
   
Comparing  (m, n) in equation (1) and (2), We get
 [(m)]2 1 (m). 
= 2 m −1 
 ( 2 m) 2  1
 m + 
 2

 1 
(m) m +  = 2 m −1 (2m).
or  2 2

For second part, we have

m 
=
 ( 2m)  2m + 1 
2 2 m −1  
 2 

Multiplying both sides by m , we obtain

m m 21− 2 m (1 / 2)m

=
 ( 2 m)  1
 m + 
 2

 1
B(m, n) = 21−2 m B m, .
Or  2

Applications of beta function

To evaluate the integrals of the form of beta function, the following form may be quick to apply:
1
0
xm (1 − x)n dx =  (m + 1, n + 1) .

Example 45. Evaluate the integral

1
0
x 2 (1 − x)3 dx

Solution. Using the formula for beta function

1
 0
xm (1 − x)n dx =  (m + 1, n + 1)

We get
1 (3)(4)
0
x 2 (1 − x)3 dx =  (2 + 1,3 + 1) =  (3, 4) =
(7)

2!  3! 1

1
x 2 (1 − x)3 dx = = .
0 6! 60
Example 46. Evaluate the integral

 (1 − x ) dx
1 3

Solution. Using the formula for beta function

Take x = t  x = t 2 and dx = 2tdt . We get

 (1 − x ) dx =  (1 − t ) 2tdt =2 t (1 − t ) dt
1 3 1 3 1 3
0 0 0

(2)(4)
= 2 (2, 4) = 2
(6)

1! 3! 1
( )
3
  1 − x dx = 2
1
= .
0 5! 10

Example 47. Evaluate the integral

 x (1 − x )
1 5
4
dx
0

Solution. Using the formula for beta function

Take x = t  x = t 2 and dx = 2tdt . We get

 ( )
x 4 1 − x dx =  t 8 (1 − t ) 2tdt =2 t 9 (1 − t ) dt
1 5 1 5 1 5
0 0 0

(10)(6)
= 2  (10, 6) = 2
(16)

9! 5!
( )
5
  x 4 1 − x dx = 2 
1
.
0 15!
Example 48. Evaluate the integral

1

1
dx
(1 − x2 )
0 3

Solution. Using the formula for beta function

Take x = t  x = t 2 and dx = 2tdt . We get

 x (1 − x ) dx =  t 8 (1 − t ) 2tdt =2 t 9 (1 − t ) dt
1 5 1 5 1 5
4
0 0 0

(10)(6)
= 2  (10, 6) = 2
(16)

9! 5!
( )
5
  x 4 1 − x dx = 2 
1
.
0 15!

1 x 2dx 2 
0 (1− x ) 2 y (1− x )
dx
Example 49. Show that 1/2 1/2
=
4 4 4 2
 x 2 dx

1

Solution. Let
0 (1 − x 4 )1/ 2

1
x 2 = (sin  )1/ 2  dx = (sin  ) −1/ 2 cos d
Let x = sin  or
2
2

 
 when x = 0   = 0, when x = 1   = 2 

 /2 sin  1
I1 =   (sin  ) −1/ 2 cos d
0 (1 − sin  )
2 1/ 2
2

1  /2
2 0
= sin  d

1 dx
I2 = 
Let
0 (1 + x 4 )1/ 2

1
x = (tan  )1/ 2  dx = (tan  ) −1/ 2 sec2  d
x = tan  or
2
2 .

Put in I 2 , we get

dx 1  / 4 (tan  ) −1/ 2 sec2  d

I2 = 
(1 + x 4 )1/ 2 2 0
1
=
0 (1 + tan 2  )1/ 2

1  /4 d 1  /4 d d
2 0 2 0
= = [ Put 2 =  and d = ]
sin  cos sin 2 2
1  /2 d 1  /2 d
2 sin  2 2 0
I2 = =
Then 2 0
sin 
 a g ( x)dx = a f (t )dt 
 0 0 

Put I1 , I 2 in (1)

x 2 dx dx
0 (1 − x 4 )1/ 2 0 (1 + x 4 )1/ 2 = I1  I 2
1 1

1  /2 1 d
2 0 2
= sin  d 
2 sin 

1   /2  /2 d 
4 2  0
= sin  d  
0
sin  
 1 
=  =
4 2 4 2.

Example 50. Show that

 ( p + 1, q) +  ( p, q + 1) =  ( p, q )

Solution.
( p + 1)(q) ( p)(q + 1)
 ( p + 1, q) +  ( p, q + 1) = +
( p + q + 1) ( p + q + 1)

( p + 1)(q) + ( p)(q + 1) p( p)(q) + ( p)q(q)

= =
( p + q + 1) ( p + q)( p + q)

( p)(q)( p + q) ( p)(q)
= = =  ( p, q) .
( p + q)( p + q) ( p + q)

Example 51. Show that

 ( p + 1, q)  ( p, q + 1)  ( p, q )
= =
p q p+q

Solution.
 ( p + 1, q) 1 ( p + 1)(q) 1 p( p)(q)
i. = =
p p ( p + q + 1) p ( p + q)( p + q)
1 ( p)(q)  ( p, q)
= = .
( p + q ) ( p + q ) ( p + q )
ii. Interchanging p and q (using symmetry of beta function)
 ( p, q + 1)  (q + 1, p)  (q, p)  ( p, q)
= = = .
q q ( p + q) ( p + q)

Example 52. Show that

1 1
24𝑛 𝑛𝛽(𝑛, 𝑛)𝛽(𝑛 + , 𝑛 + ) = 2𝜋
2 2
Solution.

(n)(n) (n + 12 )(n + 12 ) (n)2 (n + 12 )2

 (n, n)  (n + 12 , n + 12 ) =  = 
(2n) (2n + 1) (2n) 2n(2n)
Using Legendre’s duplication formula
1 𝛤(2𝑚)
𝛤(𝑚)𝛤 (𝑚 + ) = 2√𝜋 ( 2𝑚 ),
2 2
 2
1 2   2 1
2
1  (n)(n + 12 ) 
 (n, n)  (n + 12 , n + 12 ) =  =   =
2n  (2n)  2n  22 n  n 24 n

 24n n (n, n) (n + 12 , n + 12 ) = 2 .

Example 53. Show that

( x − a)n ( x − b)m dx =(−1)m ( b − a )
b n + m+1
a
 (m + 1, n + 1)
and hence, evaluate the integral
15
5
( x − 5)( x − 15)dx
Solution. Let us suppose that
b
I =  ( x − a)n ( x − b)m dx
a
Take 𝑥 − 𝑎 = 𝑡 so that 𝑑𝑥 = 𝑑𝑡, and we get

b−a
I = t n (t + a − b)m dt
0
Again, take 𝑡 = (𝑏 − 𝑎)𝑧 so that 𝑑𝑡 = (𝑏 − 𝑎)𝑑𝑧, and

I =  ( b − a ) z n (b − a) z + a − b (b − a)dz = ( b − a ) (a − b)m  z n (1 − z ) dt

1 n m n +1 1 m
0 0

 I = (−1) m ( b − a )
n + m +1
 (m + 1, n + 1) ,

  ( x − a)n ( x − b)m dx =(−1)m ( b − a )

b n + m +1
 (m + 1, n + 1)
a

Now, put 𝑎 = 5, 𝑏 = 15, 𝑚 = 𝑛 = 1/2 to get

15 ( 32 )( 32 )
5
( x − 5)( x − 15)dx = −1 102  ( 32 , 32 ) = 100i
(3)
1
( 12 ) 12 ( 12 ) 25

15
( x − 5)( x − 15)dx =100i 2
= i .
5 2 2
Example 54. Evaluate the integral

 ( x − 7)(10 − x)
10 1/3
dx
7
Solution. Let,
( x − 7)(10 − x)
10
I =
1/3
dx
7
𝑥−7
Put, 𝑧 = 3
, so that 𝑥 = 3𝑧 + 7 and 𝑑𝑥 = 3𝑑𝑧. We get

I =  (3z )(3 − 3z) 3dz = 35/3  z1/3 (1 − z)1/3 dz =35/3  ( 43 , 43 )

1 1/3 1

0 0

( 43 )( 43 )
I = 35/3  ( 43 , 43 ) = 35/3 .
( 83 )

Using duplication formula,

  ( n)  2   ( 43 ) 
=  2 n  = =
(2n)  2 (n + 12 ) n→ 4 ( 83 ) 2 ( 116 )
5/3

35/3  ( 43 )
7 ( x − 7)(10 − x) dx = 35/3  ( 43 , 43 ) =
10 1/3
.
25/3 ( 116 )

Example 55. Evaluate the integral

𝜋⁄2
∫ 𝑠𝑖𝑛3 𝑥 𝑐𝑜s 8 𝑥 𝑑𝑥
0

Solution. Using the standard integral

𝑝+1 𝑞+1
1 Γ( 2 )Γ( 2 )
𝑝 𝑞
∫ 𝑠𝑖𝑛 𝜃 𝑐𝑜𝑠 𝜃 𝑑𝜃 =
𝑝+𝑞+2
0 2Γ ( 2 )

We get
 /2 (2)(9 / 2) 1 (9 / 2)
 0
sin3 x cos8 x dx =
2(13 / 2)
= 11 9
2 2 2 (9 / 2)
 /2 2
0
sin3 x cos8 x dx =
99
.

Dirichlet’s Integrals
Dirichlet integral in two-dimension
Theorem: If 𝑙, 𝑚 are all positive, then the triple integral:
(l )(m)
 S
xl −1 y m−1dxdy =
(l + m + 1)

where the integral is taken over the area S of the surface in the first quadrant bounded by
the line 𝑥 + 𝑦 ≤ 1.

Dirichlet integral in three-dimension

Theorem: If 𝑙, 𝑚, 𝑛 are all positive, then the triple integral:
(l )(m)(n)
 V
xl −1 y m−1 z n−1dxdydz =
(l + m + n + 1)

where the integral is extended over the volume V bounded by a closed surface in the first
octant bounded by the plane 𝑥 + 𝑦 + 𝑧 ≤ 1.

Example 56. Find the area bounded by the axes and the line 𝑎𝑥 + 𝑏𝑦 = 1.
 Solution. The area bounded by the axes and the given line lies in the first quadrant,
therefore, for the required area 𝑥, 𝑦, 𝑧 ≥ 0.
Let 𝑢 = 𝑎𝑥, 𝑣 = 𝑏𝑦, so that 𝑢 + 𝑣 = 1, and 𝑑𝑥 = 𝑎𝑑𝑢, 𝑑𝑦 = 𝑏𝑑𝑣. We get

Area = ∫ ∫𝑑𝑥𝑑𝑦
𝑆

= ∫ ∫ 𝑎𝑏𝑑𝑢𝑑𝑣 = 𝑎𝑏 ∫ ∫ 𝑢1-1 𝑣 1-1 𝑑𝑢𝑑𝑣 ; 𝑢 + 𝑣 = 1

𝑆′ 𝑆′

Γ(1)Γ(1) 1
= 𝑎𝑏 = 𝑎𝑏
2Γ(3]) 2

Example 57. Find the mass of an octant of the ellipsoid

𝑥2 𝑦2 𝑧2
+ 𝑏2 + 𝑐 2 = 1
𝑎2

if the density at any point is given by 𝜌 = 𝑘𝑥𝑦.

1 
x2 a  −1
2
= u  x = au1/ 2  dx = u  2  du
Solution. Let a 2
1 
y2 b  2 −1
= v  y = bv1/ 2
 dy = v dv
b2 2
1 
z2 c  2 −1
= w  z = cw1/ 2
 dz = w dw
c2 2

=    R k xyz dx dy dz
The mass of octant of the ellipsoid
abc
=  R ka u b v c w du dv dw
8 u v w

k a 2b 2 c 2
=
8  R du dv dw

k a 2b 2 c 2
  u
1−1
= v1−1 w1−1 du dv dw
8

k a 2b 2c 2 111 k a 2b 2c 2
= =
8 4 48 .

Liouville’s Extension of Dirichlet’s Theorem

Theorem: If the variables x, y, z are all positive such that
ℎ1 ≤ 𝑥 + 𝑦 + 𝑧 ≤ ℎ2 ,
then
(l )(m)(n) h2 l + m+ n−1
 v
xl −1 y m−1 z n−1 f ( x + y + z )dxdyz =
(l + m + n) h1
u f (u )du .

Note: The integral is applicable to two variables as follows:

If the variables x, y are all positive such that
ℎ1 ≤ 𝑥 + 𝑦 ≤ ℎ2 ,
then
(l )(m) h2 l + m−1
 x y f ( x + y)dxdy = (l + m) h1
l −1 m−1
u f (u )du .

Problem. Find the value of  log ( x + y + z ) dxdydz , the integral extending over all
v

positive values of x, y, z subject to the condition x + y + z 1.

Solution. Here the integral is to be extended for all positive values of x, y, z, therefore,
0 < 𝑥 + 𝑦 + 𝑧 ≤ 1.

 the required integral =    log( x + y + z ) dx dy dz, where 0  x + y + z  1

=    log( x + y + z ) x1−1 y1−1 z1−1dx dy dz

(1)(1)(1) 1
(1 + 1 + 1) 0
= (log u )u1+1+1−1du

(by Liouville’s extension of Drichlet’s theorem)

1 1
=
3 
0
u 2 log u du,

1  
1
u3  11 u
3
=  log u.  −  . du,
2.1  3 0 0 u 3 
 
(integrating by parts taking 𝑢2 as the second function)

1 1 1 1 
=  0 − lim u 3 log u −  u 2 du
2  3 u →0 3 0

1
1 u3 
=−  
6  3 0 lim u log u = 0
u →0
3
 1
=− .
8