You are on page 1of 12

Spring 2020 Math 155 Professor Shneerson

1 Improper Integrals

Let y = f (x) be a function which is continuous on the closed interval [a; b].
Rb
Then the de…nite integral f (x)dx is a real number that represents the area
a
under the curve y = f (x) from x = a to x = b in the case when f (x) 0
on [a; b].
The number a is called the lower limit and b is called the upper limit of
Rb
the de…nite integral f (x)dx:
a
It follows from the Fundamental Theorems of Calculus (FTC 1 and FTC 2)
that f (x) has antiderivative F (x) on [a; b] and
Rb x=b
f (x)dx = F (x) jx=a = F (b) F (a) (1)
a

Example 1 The area under one lump of a sine curve is equal to


R
sin(x)dx = cos(x) jx=
x=0 = (cos ) ( cos 0) = 2. (see Fig.3 in the …le
0
with graphing illustrations)
1
Example 2 The area under the graph of f (x) = x from x = 1 to x = 2 is
R2 1
equal to x dx = ln(2) ln(1) = ln(2): (see Fig.5)
1

Example 3 The area under the graph of f (x) = x12 from x = 1 to x = 4 is


R4 R4 2+1
equal to x12 dx = x 2 dx = x 2+1 jx=4
x=1 =
1 x=4
x jx=1 =
1 3
4 + 1 = 4 . (Fig 4)
1 1

The main goal of this lecture is to develop the notion of the de…nite integral
and extend it for the case when at least one of the numbers a; b is in…nite and
also consider the case when f has an in…nite discontinuity at one of the numbers
a; b or at some point of the open interval (a; b). The new types of integrals will
be called improper integrals. The improper integral with non-negative
integrand f(x) will represent the area of the unbounded region with
respect to the graph of f on a certain interval which is not necessarily
…nite.

1.1 Improper Integrals with In…nite Upper or Lower Lim-


its
1
Consider the function f (x) = on the interval I = [1,1). Since f (x) is
x2
Rb
continuous on I; then for any real number b 1; the integral x12 dx exists and
1

1
coming back to Example 3, we see that
Rb 1 1 x=b 1
dx = j = ( 1) = b + 1: (2)
1 x
2 x x=1 b
Assume that b is getting larger and larger. For instance, starting with
b = 4; 10; 1000 we obtain the corresponding values of our integral, namely
R4 1 1 3
R10 1 1 9
R 1
1000
1
x2 dx = 4 + 1 = 4; x2 dx = 10 + 1 = 10 and x2 dx = 1000 + 1 =
1 1 1
999
1000 .
Rb 1
We are interested the behavior of the integral x2 dx as the upper
1
limit b is increasing without bound. In other words, we are interested in
the value of the limit
Rb 1
lim dx (3)
b!1 1 x2

The formula (2) shows that


Rb 1 1 1
lim dx = lim ( + 1) = lim ( ) + lim (1) = 0 + 1 = 1: (4)
b!1 1 x2 b!1 b b!1 b b!1

In other words, the area under the graph of y = x12 on the interval [1; b]
approaches 1 as b tends to in…nity. Thus we can assign to the unbounded
region under the graph of y = x12 on [1; 1) a number 1 as its area. Sym-
bolically, we will also write
R1 1
2
dx = 1. (5)
1 x

De…nition 4 Let f (x) be a function which is continuous on the interval (a; 1).
Then we de…ne
R1 Rb
f (x)dx = lim f (x)dx (6)
a b!1 a

provided that this limit exists as a …nite number.


Remark 5 We will often use the letter t instead of b because usually in
Calculus, the letter b represents some constant and t is a standard name of
some variable.
So, usually we will write
R1 Rt
f (x)dx = lim f (x)dx: (7)
a t!1 a
R 1
Example 6 Since 1+x2 dx = arctan(x) + C; and arctan(0) = 0; then
R1 1 Rt 1
2
dx = lim [ dx] = lim [(arctan(t) arctan(0))] = lim [arctan(t)] = :
0 1+x t!1 0 1 + x2 t!1 t!1 2
(8)

2
Rt
De…nition 7 If the limit lim f (x)dx exists as a …nite number, then we say
t!1 a
R1
that the integral f (x)dx is convergent. Otherwise, it is divergent.
a

Example 8 It follows from this de…nition and formulas (4) and (8) that
R1 R1 1
the integrals x12 dx = 1 and 1+x 2 dx (see Fig 4 and resp. Fig 6) are
1 0
convergent. .
R1 x
Example 9 Consider the improper integral e dx: Since
0

Rt
lim ex dx = lim (et e0 ) = lim (et 1) = 1; (9)
t!1 a t!1 t!1

R1 x
then e dx is divergent (see Figure 7 for geometric illustration).
0

Example 10 The improper integral


R1 1
dx (10)
1 x

is divergent. Indeed,
R1 1 Rt 1 x=t
dx = lim dx = lim ln(x) jx=1 = lim [ln(t) ln(1)] = lim ln(t) = 1:
1 x t!1 1 x t!1 t!1 t!1
(11)

Thus the improper integral (10) cannot be a …nite number and it diverges.
(Fig.8)
R1
Exercise 11 Show that the improper integral p1 dx is divergent (see Fig.9)
x
1
R R 1 1 p 1 +1
Solution. Since p1x dx = x 2 dx = x + C = 2x 2 + C = 2 x + C; 1
2

2 +1
R1 p p
then p1x dx = lim 2 x jx=t
x=1 = lim 2 x 2 = 1:
1 t!1 t!1

R1 3x
Exercise 12 Prove that the integral e dx is convergent (Fig.11-12). Eval-
1
uate the integral. R
Step I. Let us …nd the corresponding inde…nite integral e 3x dx. Put u =
3x:
R Then du =R ( 3)dx and dxR= ( 13 )du. Thus
1 u 1 1 u 1
e 3x dx = 3 e du =
u
3 e du = 3e + C = 3e
3x
+ C:
R1 3x
Step II. e dx = lim 11 e 3x jx=1
x=t
= lim ( 13 e 3t ( 13 e 3 ))
1 t!1 t!1
= lim [ 13 e13t + 13 e13 ] = 0 + 3e13 = 1
3e 3
t!1
Thus this integral is convergent.

3
Rb
Now suppose that the lower limit a of the integral f (x)dx is decreasing
a
without bound.
De…nition 13 Let f (x) be a function which is continuous on the interval (a,1).
Then we de…ne
Rb Rb
f (x)dx = lim f (x)dx (12)
1 a!1 a

provided that this limit exists as a …nite number.


It follows from the de…nition that
Rb Rb
f (x)dx = lim f (x)dx: (13)
1 t!1 t

R0
Example 14 Consider the integral e5x dx. (Fig. 13). Let us show that this
1
integral is convergent. Indeed,
R0 R0
e5x dx = lim e5x dx; (14)
1 t! 1 t
R 1
Let us …nd e5x dx …rst. Put u = 5x: Then du = 5dx and dx = 5 du:
Therefore
R R 1 1R u 1 1
e5x dx = eu ( du) = e du = eu + C = e5x + C: (15)
5 5 5 5
Combining the equalities (14) and (15) we obtain
R0 5x R0
e dx = lim e5x dx = lim 15 e5x jx=0 x=t
=
1 t! 1 t t! 1

lim ( 15 e0 1 5t 1 0
5e ) = 5e lim 1 e5t = 1
1 0= 1
t! 1 t! 1 5 5 5

Example 15 Thus
R0 1
e5x dx = (16)
1 5
and the integral is convergent and the area under the graph of the function
f (x) = e5x on the interval ( 1; 0] is …nite and equal to 15 .
R0 1
Example 16 The improper integral x2 +1 dx is convergent.
1

Example 17 Indeed,
R0 1 R0 1
dx = lim dx = lim [arctan(0) arctan(t)] = 0 ( )= :
1 x2 +1 t! 1 t x2 + 1 t! 1 2 2
(17)
See Fig.15, 14 for graphing illustration.

4
1.2 Improper Integrals With Both Limits In…nite.
De…nition 18 Let f(x) be a function which is continuous on ( 1; 1) and sup-
R1 R0
pose that the improper integrals f (x)dx and f (x)dx both converge. Then,
0 1
by de…nition

R1 R1 R0
f (x)dx = f (x)dx + f (x)dx: (18)
1 0 1

Example 19 (See the formulas (8), (17) and Fig. 16-17.)

R1 1 R1 1 R0 1
2
dx = 2
dx + dx = + = : (19)
1 1 + x 0 1 + x 1 1 + x2 2 2

R1
Example 20 The improper integral ex dx is divergent because the integral
1
R1 x
e dx is divergent (See Example 9).
0

R1 2
Example 21 The improper integral xe x dx is convergent (Fig.18).
1
R 2
Solution. Let us …nd xe x dx: Put u = x2 . Then du = 2xdx; xdx =
1
2 du and

R x2
R x2
R 1 1R u 1 u 1 x2
xe dx = e (xdx) = eu ( du) = e du = e +C = e +C
2 2 2 2
(20)
This implies that

R1 x2
Rt x2 1 x2 x=t 1 t2 1 1 1
xe dx = lim xe dx = lim e jx=0 = lim ( e + e0 ) = 0+ = :
0 t! 1 0 t! 1 2 t! 1 2 2 2 2
(21)
Then
R0 x2
R0 x2
Rt x2
R1 x2 1
xe dx = lim xe dx =(!!) lim ( xe dx) = xe dx =
1 t! 1 t t!1 0 0 2
(22)
R1 x2
R1 x2
R0 x2
Thus xe dx = xe dx + xe dx = ( 12 ) + ( 1
2 ) = 0:
1 0 1

1.3 Further Examples


R1 1
R0 1
R1 1
1. Evaluate the improper integrals x2 +9 dx; x2 +9 dx; x2 +9 dx (Fig.19)
0 1 1

5
Solution: Since
R 1 R 1 1 x
dx = dx = arctan( ) + C ; (23)
x2 +9 x2 + 3 2 3 3
Rt 1 1
then x2 +9 dx = 3 arctan( 3t ) 1
3 arctan(0) = 1 t
3 [arctan( 3 )] ! 1
3[ 2 ] = 6 as
0
t ! 1:
Thus
R1 1
dx = : (24)
0 x2 + 9 6
R0 1
R0
1
Similarly, x2 +9 dx = lim 2 dx = lim 31 arctan( x3 ) jx=0
x=t = lim (0
1 t! 1 t x +9 t! 1 t!1
1
3 arctan( 3t )) = lim 1 t
3 [arctan( 3 )] =
1
3 [ 2 ] = 6 : Hence
t!1

R0 1
2+9
dx = : (25)
1 x 6

It follows immediately from (24) and (25) that

R1 1 R1 1 R0 1
2
dx = 2
dx + 2
dx = + = : (26)
1x +9 0 x +9 1x +9 6 6 3

1.3.1 2. The p-test for " x1p improper integrals".


Let p be a real number. Consider the improper integral
R1 1
p
dx (27)
1 x

Our goal is to …nd all values of p for which this integral is convergent.
R1
It follows from our previous investigation that the integral x1p dx is conver-
1
gent when p = 2 (see the formula (5)), diverges when p = 1 (Example 10) and
p = 1=2 (Exercise 11).
See Fig. 2 for cases p = 12 ; 1; 2:

R1 1
Theorem 22 The improper integral xp dx converges for all p > 1 and diverges
1
for any p 1:
Proof. Case 1 ( p>1). Let = p 1: Since p>1, then > 0. Also, we have
that p = + 1:Consider the inde…nite integral

R 1 R p
R ( +1) x ( +1)+1 x
p
dx = x dx = x dx = +C = + C: (28)
x ( + 1) + 1

6
Thus
R1 1 Rt 1 x t 1
p
dx = lim p
dx = lim jx=t
x=1 = lim ( )= (29)
1 x t!1 1 x t!1 t!1

1 1 1 1 1
lim ( + ) =( >0 !!!)
0+ = = : (30)
t!1 t p 1
We see that
R1 1 1
p>1) p
dx = : (31)
1 x p 1
Case 2 (p<1). As in Case 1, put = p 1. Then < 0 and is positive.
Rt 1 R1
Therefore xp dx = ( t 1
) ! 1 as t ! 1. Hence the integral x1p dx is
1 1
divergent for this case.
R1 1
Case 3 (p = 1). As we noted above the integral x dx is divergent. This
1
completes the proof of the theorem.

Exercise. Which of the following improper integrals are convergent?


R1 1
(a) x100 dx
1
R1 1
(b) x0:998 dx
1
R1 1
(c) 3 4 dx
p
x
1
Exercise. Evaluate the integral
R1 6
x5 dx
2
R1 6 6
Rt x 4
6
Solution : x5 dx = lim 5 dx = lim 6 jx=t
x=2 ) = lim 4
x=t
jx=2 )
2 t!1 2 x t!1 4 t!1 ( 4)x
lim ( 6 4 + 6
( 4) 24 ) = lim ( ( 6
4)t4 ) + lim 6
=0 6 3
+ 64 = 32
t!1 ( 4)t t!1 t!1 4 16

1.4 The Comparison Test for Improper Integrals.


Theorem 23 Let f (x); g(x) be functions such that 0 f (x) g(x) on the
R1 R1
interval [a,1). If g(x)dx is convergent then so does f (x)dx:
a a

In other words, if the area under the graph of g(x) is …nite, then the area
under the graph of f (x) being smaller must also be …nite.
R1
Corollary 24 (Contrapositive statement). If f (x)dx diverges, then so does
a
R1
g(x)dx:
a

7
Example 25 Consider the improper integral
R1 1
dx: (32)
1 x5 + 3x + 1
1
Clearly, the integrand x5 +3x+1 is non-negative and smaller than x15 on the in-
terval of integration [1,1): So, we may use the Comparison Test with f (x) = x15
1
R1
and g(x) = x5 +3x+1 . Since the integral x15 dx is convergent by the p- test
1
with p = 5, so is the improper integral (32). (See Fig.20)
R1px+100
Exercise 26 Determine whether the improper integral x dx is conver-
1
gent or divergent.
p p p p
x
Solution: Since 0 < x < x + 100; then x+100 x > x =
p1 : Since
x
R 1
1
1
the improper integral px dx is divergent by the p- test (with p= 2 ), then by
1
R1px+100
Corollary 24 so does x dx:
1

R1 2
Exercise 27 Use the Comparison Test to show that the improper integral p
3
x 1
dx
8
is divergent.
R1 1 2
Hint: Compare this integral with 3 x dx
p . Use the inequality p
3
x 1
>
8
1 1
p
3
x 1
> p
3 x for all x>1. (See Fig. 21)

1.5 Improper Integrals With Unbounded Integrands.


Example 28 Consider the function f (x) = p11 x2 : It is clear that f (x) is con-
tinuous on [0; 1) and has an in…nite discontinuity at 1: (Fig.22) More precisely,
1
lim p = 1: (33)
x!1 1 x2
At the same time for any number t with 0 t < b; the integral

Rt 1 x=t
p dx = arcsin(x) jx=0 = arcsin(t) arcsin(0) = arcsin(t) (34)
0 1 x2

is a …nite number. It is natural to ask


Rt
What can we say about the behavior of the integral p 1 dx as t
1 x2
a
gets closer and closer to 1, but t is less than 1?

8
Rt
In other words, we are interested in the left-hand limit p 1 dx = arcsin(t)
1 x2
0
as t ! 1 :
Since lim arcsin(t) = 2 (also see Fig.23) then the equalities (34) show
t!1
that
Rt 1
lim p dx = lim arcsin(t) = : (35)
t!1 0 1 x2 t!1 2

De…nition 29 Let f (x) be a function which is continuous on the semi-open


Rt
interval [a; b). Then for any number t with a t < b; the integral f (x)dx
a
exists and so is a …nite number). Assume that f (x ) has discontinuity at b: Then
Rb
we de…ne the improper integral f (x)dx by
a

Rb Rt
f (x)dx = lim f (x)dx (36)
a t!b a

provided that the limit exists as a …nite number.

De…nition 30 Let f (x) be a function which is continuous on the semi-open


interval (a; b]. Assume that f (x ) has discontinuity at a: Then we de…ne the
Rb
improper integral f (x)dx by
a

Rb Rb
f (x)dx = lim f (x)dx (37)
a t!a+ t

provided that the limit exists as a …nite number.

De…nition 31 If the limit (36) ( or (37) ) exists as a …nite number, we say that
Rb
the improper integral f (x)dx is convergent; otherwise it is called divergent.
a

Examples.
In Examples 1 through 3 below evaluate the improper integral (if it con-
verges)
R1 1 1
I. p 3 x dx (See Fig.26 for the behavior of y = 3 x on the interval (0; 1])
p
0
R1 1
Solution. The integral p
3 x dx is improper because the integrand
0
1
p
3 x is unde…ned at the lower limit 0.
R 1
We begin by evaluating the corresponding inde…nite integral 3 x du
p using
the power rule for

9
1
R 1 R 1 x 3 +1 3 2 3p3
p dx = 1 dx = 1 + C = x3 + C = x2 + C (38)
3
x x 3
3 + 1 2 2
Thus
R1 1 R1 3 p 3 p
3 3 p
3
p
3 3 3
p dx = lim 3
(x)2 dx = lim [ x2 ) jx=1
x=t ] = lim [ 12 t2 ] j= (1 0) =
0
3
x t!0 +
0 2 t!0 + 2 t!0 + 2 2 2
(39)
R5
II. (x 13)2 dx
3
Solution. Here the integral is improper because the integrand
1
(x 3)2 is unde…ned at the lower limit 3:
We begin by using
R the substitution u = x 3 to evaluate the corresponding
1
inde…nite integral (x 3)2 dx
Using the substitution u = x 3 we obtain du = d(x 3) = (x 3)0dx =
1 dx = dx:
Then
R 1 R 1 R 1 1
2
dx = 2
du = u 2 du = +C = +C (40)
(x 3) u u (x 3)
Therefore,
R5 1 R5 1 1 1 1 1
2
dx = lim+ dx = lim+ jx=5
x=t = lim+ ( + )= +1 = 1:
3 (x 3) t!3 t (x 3)2 t!3 (x t 3 3) t!3 (5 3) 2
(41)
This shows that the improper integral (42) cannot be a …nite number and
so it diverges.
R2
III. pcos(x) dx (Fig.28).
sin(x)
0
Solution. This integral is improper because the integrand pcos(x) is unde-
sin(x)
…ned at the lower limit 0.
As before, let us …nd the corresponding inde…nite uintegral using a suitable
u- substitution.
since cos(x)dx = d sin(x);then the integrand pcos(x) dx can be rewritten
sin(x)
d sin(x)
as p . Put u = sin(x). Then
sin(x)

R cos(x) R du R 1 1 p p
p dx = p = u 2 du = 2u 2 + C = 2 u + C = 2 sin(x) + C (42)
sin(x) u
Thus
r
R2 cos(x) R2 cos(x) p x= 2 p p
p dx = lim+ p dx = lim+ 2 sin(x) jx=t = lim+ 2 sin( ) 2 sin(t) = 2 1 0 = 2
0 sin(x) t!0 t sin(x) t!0 t!0 2
(43)
Answer: The integral is convergent.

10
1.5.1 Challenging Problems
Evaluate the integrals
1.
R1
ln(x)dx (Fig.24)
0
Hint: The integral is improper because the limit lim+ ln(x) = 1: Since
R x!0
ln(x)dx = x(ln(x) 1) + C ( See Section 6.1, Integration by Parts, Example
2, page 319)), then
R1 R1
ln(x)dx = lim+ ln(x)dx = lim+ [x(ln(x) 1)] jx=1
x=t = lim+ [(1 ln(1) 1)
0 t!0 t t!0 t!0
(t(ln(t) 1)] =
lim [ 1 + t t ln(t)] = lim ( 1 + t) lim t ln(t) = 1 lim t ln(t)
t!0+ t!0+ t!0+ t!0+
Since the limit lim+ t ln(t) = lim+ x ln(x) = 0 (See Section 5.8 (Indetermi-
t!0 x!0
R1
nate Forms and L’Hospital’s Rule, Example 6, p.309), then ln(x)dx = 1:
0
R1 1
2. (x2 +2x+2) dx:
1
Hint: Since x2 + 2x + 2 = (x + 1)2 + 1; then
R1 1 R1 1
2 + 2x + 2)
dx = 2+1
dx (44)
1 (x 1 (x + 1)

Let f (x) = x21+1 and g(x) = (x+1)1


2 +1 :Then g(x) = f (x + 1) (!!) and the

graph of g can be obtained from the graph of f by a horizontal shif t 1 unit


to the left (See Fig.25) . This shows that the area under the graph of g(x)
on ( 1; 1) is exactly equal to the area under the graph of f (x) on the same
interval.
Thus
R1 1 R1 1
2
dx = 2
dx = (45)
1 (x + 1) + 1 1 (x + 1)

Combining (44) and (45) we obtain


R1 1 R1 1
dx: = dx = : (46)
1 (x2 + 2x + 2) 1 (x2 + 1)

Final Remark.
Proposition 32 Assume that the function f (x) is continuous on the interval
R1
[a,1) and c is a real number with c > a. f (x)dx is convergent if and only if
a
R1
so does f (x)dx.
c

11
Outline of the Proof :
R1 Rc R1
f (x)dx = f (x)dx + f (x)dx: (47)
a a c

Rc
Suppose that the left - hand side of (47) is …nite. Since f (x)dx is a …nite
a
R1 R1 Rc
number, then f (x)dx = f (x)dx f (x)dx must be …nite.
c a a
R1 R1
Conversely, if f (x)dx is …nite, then the equality (47) shows that f (x)dx
c a
is …nite.
R1 1
Exercise 33 Determine whether or not the integral x1:0001 dx is convergent.
7
R1 1
Solution: Consider the improper integral x1:0001 dx. This integral is conver-
1
gent by p series test with p=1.0001>1. Thus, by Proposition 31, the integral
R1 1
x1:0001 dx is such.
7

R1 1
R1 1
Exercise 34 Prove that the improper integrals x0:9998 dx and x0:9998 dx
8 0:1
both divergent.
R1 1
Solution. (i) Consider the improper integral x0:9998 dx: This integral is di-
1
vergent by the p- series test with p=0.9998<1 and represents in…nite area
1
under the curve y = x0:9998 on [1,1):
R1
1
R1 1
(ii) Since x 0:9998 dx is …nite and x0:9998 dx is in…nite, then
0:1 1
R1 1
R1 1
R1 1
x0:9998 dx = x0:9998 dx + x0:9998 dx cannot be a …nite number. Thus
0:1 0:1 1
R1 1
x0:9998 dx diverges.
0:1
Similarly,
R1 1 R8 1 R1 1
dx = dx + dx: (48)
1 x0:9998 1 x0:9998 8 x0:9998
R8 1
R1 1
Since x0:9998 dx is …nite and x0:9998 dx is in…nite, then the second summand
1 1
R1 1
x0:9998 dx from the right- hand side of (48) cannot be …nite.
8

12

You might also like