7.

6 IMPROPER INTEGRALS 385

Strengthen Your Understanding

In Problems 49–52, explain what is wrong with the state- In Problems 57–67, decide whether the statements are true
ment. for all continuous functions, 𝑓 . Give an explanation for your
answer.
49. The midpoint rule never gives the exact value of a def-
inite integral. 57. If 𝑛 = 10, then the subdivision size is Δ𝑥 = 1∕10.
50. TRAP(𝑛) → 0 as 𝑛 → ∞.
58. If we double the value of 𝑛, we make Δ𝑥 half as large.
51. For any integral, TRAP(𝑛) ≥ MID(𝑛).
52. If, for a certain integral, it takes 3 nanoseconds to im- 59. LEFT(10) ≤RIGHT(10)
prove the accuracy of TRAP from one digit to three dig- 60. As 𝑛 approaches infinity, LEFT(𝑛) approaches 0.
its, then it also takes 3 nanoseconds to improve the ac-
curacy from 8 digits to 10 digits. 61. LEFT(𝑛) − RIGHT(𝑛) = (𝑓 (2) − 𝑓 (6))Δ𝑥.

In Problems 53–54, give an example of: 62. Doubling 𝑛 decreases the difference LEFT(𝑛) −
RIGHT(𝑛) by exactly the factor 1∕2.
53. A continuous function 𝑓 (𝑥) on the interval [0, 1] such
1
that RIGHT(10) < ∫0 𝑓 (𝑥)𝑑𝑥 < MID(10). 63. If LEFT(𝑛) = RIGHT(𝑛) for all 𝑛, then 𝑓 is a constant
function.
54. A continuous function 𝑓 (𝑥) on the interval [0, 10] such
that TRAP(40) > TRAP(80). 64. The trapezoid estimate TRAP(𝑛) = (LEFT(𝑛) +
6
RIGHT(𝑛))∕2 is always closer to ∫2 𝑓 (𝑥)𝑑𝑥 than
In Problems 55–56, decide whether the statements are true LEFT(𝑛) or RIGHT(𝑛).
or false. Give an explanation for your answer. 6
65. ∫2 𝑓 (𝑥) 𝑑𝑥 lies between LEFT(𝑛) and RIGHT(𝑛).
1
55. The midpoint rule approximation to ∫0 (𝑦2 − 1) 𝑑𝑦 is 𝑏
always smaller than the exact value of the integral. 66. If LEFT(2) < ∫𝑎 𝑓 (𝑥) 𝑑𝑥, then LEFT(4) <
𝑏
56. The trapezoid rule approximation is never exact. ∫𝑎 𝑓 (𝑥) 𝑑𝑥.
67. If 0 < 𝑓 ′ < 𝑔 ′ everywhere, then the error in approxi-
The left and right Riemann sums of a function 𝑓 on the 𝑏
interval [2, 6] are denoted by LEFT(𝑛) and RIGHT(𝑛), re- mating ∫𝑎 𝑓 (𝑥) 𝑑𝑥 by LEFT(𝑛) is less than the error in
𝑏
spectively, when the interval is divided into 𝑛 equal parts. approximating ∫𝑎 𝑔(𝑥) 𝑑𝑥 by LEFT(𝑛).

7.6 IMPROPER INTEGRALS

𝑏
Our original discussion of the definite integral ∫𝑎 𝑓 (𝑥) 𝑑𝑥 assumed that the interval 𝑎 ≤ 𝑥 ≤ 𝑏 was
of finite length and that 𝑓 was continuous. Integrals that arise in applications do not necessarily have
these nice properties. In this section we investigate a class of integrals, called improper integrals, in
which one limit of integration is infinite or the integrand is unbounded. As an example, to estimate
the mass of the earth’s atmosphere, we might calculate an integral which sums the mass of the air
up to different heights. In order to represent the fact that the atmosphere does not end at a specific
height, we let the upper limit of integration get larger and larger, or tend to infinity.
We consider improper integrals with positive integrands since they are the most common.
One Type of Improper Integral: When the Limit of Integration Is Infinite
Here is an example of an improper integral:
∞
1
𝑑𝑥.
∫1 𝑥2
𝑏
To evaluate this integral, we first compute the definite integral ∫1 (1∕𝑥2 ) 𝑑𝑥:
𝑏
1 |𝑏
−1 | 1 1
𝑑𝑥 = −𝑥 | = −𝑏 + 1.
∫1 𝑥2 |1
Now take the limit as 𝑏 → ∞. Since
𝑏
1 1
( )
lim 𝑑𝑥 = lim − + 1 = 1,
𝑏→∞ ∫1 𝑥2 𝑏→∞ 𝑏
∞
we say that the improper integral ∫1 (1∕𝑥2 ) 𝑑𝑥 converges to 1.
386 Chapter 7 INTEGRATION

∞
If we think in terms of areas, the integral ∫1 (1∕𝑥2 ) 𝑑𝑥 represents the area under 𝑓 (𝑥) = 1∕𝑥2
from 𝑥 = 1 extending infinitely far to the right. (See Figure 7.17(a).) It may seem strange that this
region has finite area. What our limit computations are saying is that
10 10
1 1 || 1
When 𝑏 = 10: 𝑑𝑥 = − = − + 1 = 0.9
∫1 𝑥2 𝑥 |1
| 10
100
1 1
When 𝑏 = 100: 𝑑𝑥 = − + 1 = 0.99
∫1 𝑥 2 100
1000
1 1
When 𝑏 = 1000: 𝑑𝑥 = − + 1 = 0.999
∫1 𝑥2 1000
and so on. In other words, as 𝑏 gets larger and larger, the area between 𝑥 = 1 and 𝑥 = 𝑏 tends to 1.
∞
See Figure 7.17(b). Thus, it does make sense to declare that ∫1 (1∕𝑥2 ) 𝑑𝑥 = 1.
(a) 𝑦 (b) 𝑦
1 1
𝑦= 𝑥2
𝑦= 𝑥2

∞ 1 𝑏 1
Area = ∫1 𝑑𝑥 Area = ∫1 𝑑𝑥 ;
𝑥2 𝑥2
now let 𝑏 → ∞

✠ ✠
𝑥 𝑥
1 1 𝑏

Figure 7.17: Area representation of improper integral

Of course, in another example, we might not get a finite limit as 𝑏 gets larger and larger. In that
case we say the improper integral diverges.

Suppose 𝑓 (𝑥) is positive for 𝑥 ≥ 𝑎.

𝑏 ∞
If lim 𝑓 (𝑥) 𝑑𝑥 is a finite number, we say that 𝑓 (𝑥) 𝑑𝑥 converges and define
𝑏→∞ ∫𝑎 ∫𝑎
∞ 𝑏
𝑓 (𝑥) 𝑑𝑥 = lim 𝑓 (𝑥) 𝑑𝑥.
∫𝑎 𝑏→∞ ∫𝑎

∞ 𝑏
Otherwise, we say that 𝑓 (𝑥) 𝑑𝑥 diverges. We define 𝑓 (𝑥) 𝑑𝑥 similarly.
∫𝑎 ∫−∞

Similar definitions apply if 𝑓 (𝑥) is negative.

∞
1
Example 1 Does the improper integral √ 𝑑𝑥 converge or diverge?
∫1 𝑥

Solution We consider
𝑏
1
𝑏
−1∕2
|𝑏
√ 𝑑𝑥 = ∫ 𝑥 𝑑𝑥 = 2𝑥1∕2 || = 2𝑏1∕2 − 2.
∫1 𝑥 1 |1
𝑏 √
We see that ∫1 (1∕ 𝑥) 𝑑𝑥 grows without bound as 𝑏 → ∞. We have shown that the area under the
 7.6 IMPROPER INTEGRALS 387

∞ √
curve in Figure 7.18 is not finite. Thus we say the integral ∫1 (1∕ 𝑥) 𝑑𝑥 diverges. We could also
∞ √
say ∫1 (1∕ 𝑥) 𝑑𝑥 = ∞.
∞
Notice that 𝑓 (𝑥) → 0 as 𝑥 → ∞ does not guarantee convergence of ∫𝑎 𝑓 (𝑥) 𝑑𝑥.
𝑦
1
𝑦= √
𝑥

Area representing
∞
∫1 √𝑑𝑥 not finite
𝑥

✠
𝑥
1 2 3
∞ 1
Figure 7.18: ∫1 √ 𝑑𝑥 diverges
𝑥

√
What is the difference between the functions 1∕𝑥2 and 1∕ 𝑥 that makes the area under the graph
√
of 1∕𝑥2 approach 1 as 𝑥 → ∞, whereas the area under 1∕ 𝑥 grows very large? Both functions
approach 0 as 𝑥 grows, so as 𝑏 grows larger, smaller bits of area are being added√to the definite
integral. The difference between the functions is subtle: the values of the function 1∕ 𝑥 don’t shrink
√ for the integral to have a finite value. Of the
fast enough two functions, 1∕𝑥2 drops to 0 much faster
2
than 1∕ 𝑥, and this feature keeps the area under 1∕𝑥 from growing beyond 1.

∞
Example 2 Find 𝑒−5𝑥 𝑑𝑥.
∫0
𝑏
Solution First we consider ∫0 𝑒−5𝑥 𝑑𝑥:

𝑏
1 |𝑏 1 1
𝑒−5𝑥 𝑑𝑥 = − 𝑒−5𝑥 || = − 𝑒−5𝑏 + .
∫0 5 |0 5 5

1 ∞
Since 𝑒−5𝑏 = , this term tends to 0 as 𝑏 approaches infinity, so ∫0 𝑒−5𝑥 𝑑𝑥 converges. Its value
𝑒 5𝑏
is
∞ 𝑏
1 1 1 1
( )
𝑒−5𝑥 𝑑𝑥 = lim 𝑒−5𝑥 𝑑𝑥 = lim − 𝑒−5𝑏 + =0+ = .
∫0 𝑏→∞ ∫0 𝑏→∞ 5 5 5 5
Since 𝑒5𝑥 grows very rapidly, we expect that 𝑒−5𝑥 will approach 0 rapidly. The fact that the area
approaches 1∕5 instead of growing without bound is a consequence of the speed with which the
integrand 𝑒−5𝑥 approaches 0.

∞
1
Example 3 Determine for which values of the exponent, 𝑝, the improper integral 𝑑𝑥 diverges.
∫1 𝑥𝑝
Solution For 𝑝 ≠ 1,
𝑏 |𝑏
( )
−𝑝 1 1 1
𝑥 𝑑𝑥 = 𝑥−𝑝+1 || = 𝑏−𝑝+1 − .
∫1 −𝑝 + 1 |1 −𝑝 + 1 −𝑝 + 1
The important question is whether the exponent of 𝑏 is positive or negative. If it is negative, then
as 𝑏 approaches infinity, 𝑏−𝑝+1 approaches 0. If the exponent is positive, then 𝑏−𝑝+1 grows without
bound as 𝑏 approaches infinity. What happens if 𝑝 = 1? In this case we get
∞
1 |𝑏
𝑑𝑥 = lim ln 𝑥|| = lim ln 𝑏 − ln 1.
∫1 𝑥 𝑏→∞ |1 𝑏→∞
388 Chapter 7 INTEGRATION

Since ln 𝑏 becomes arbitrarily large as 𝑏 approaches infinity, the integral grows without bound. We
∞
conclude that ∫1 (1∕𝑥𝑝 ) 𝑑𝑥 diverges precisely when 𝑝 ≤ 1. For 𝑝 > 1 the integral has the value
∞ 𝑏 ( ) ( )
1 1 1 1 1 1
𝑑𝑥 = lim 𝑑𝑥 = lim 𝑏−𝑝+1 − =− = .
∫1 𝑥𝑝 𝑏→∞ ∫1 𝑥𝑝 𝑏→∞ −𝑝 + 1 −𝑝 + 1 −𝑝 + 1 𝑝−1

Application of Improper Integrals to Energy

The energy, 𝐸, required to separate two charged particles, originally a distance 𝑎 apart, to a distance
𝑏, is given by the integral
𝑏 𝑘𝑞 𝑞
1 2
𝐸= 𝑑𝑟
∫𝑎 𝑟2
where 𝑞1 and 𝑞2 are the magnitudes of the charges and 𝑘 is a constant. If 𝑞1 and 𝑞2 are in coulombs,
𝑎 and 𝑏 are in meters, and 𝐸 is in joules, the value of the constant 𝑘 is 9 ⋅ 109 .

Example 4 A hydrogen atom consists of a proton and an electron, with opposite charges of magnitude 1.6⋅10−19
coulombs. Find the energy required to take a hydrogen atom apart (that is, to move the electron from
its orbit to an infinite distance from the proton). Assume that the initial distance between the electron
and the proton is the Bohr radius, 𝑅𝐵 = 5.3 ⋅ 10−11 meter.

Solution Since we are moving from an initial distance of 𝑅𝐵 to a final distance of ∞, the energy is represented
by the improper integral
∞ 𝑞1 𝑞2 𝑏
1
𝐸= 𝑘 𝑑𝑟 = 𝑘𝑞1 𝑞2 lim 𝑑𝑟
∫𝑅𝐵 𝑟2 𝑏→∞ ∫𝑅
𝐵
𝑟 2

𝑏 ( )
1| 1 1 𝑘𝑞 𝑞
= 𝑘𝑞1 𝑞2 lim − || = 𝑘𝑞1 𝑞2 lim − + = 1 2.
𝑏→∞ 𝑟 |𝑅 𝑏→∞ 𝑏 𝑅𝐵 𝑅𝐵
𝐵

Substituting numerical values, we get

(9 ⋅ 109 )(1.6 ⋅ 10−19 )2

𝐸= ≈ 4.35 ⋅ 10−18 joules.
5.3 ⋅ 10−11
This is about the amount of energy needed to lift a speck of dust 0.000000025 inch off the ground.
(In other words, not much!)

What happens if the limits of integration are −∞ and ∞? In this case, we break the integral at
any point and write the original integral as a sum of two new improper integrals.

For a positive function 𝑓 (𝑥), we can use any (finite) number 𝑐 to define
∞ 𝑐 ∞
𝑓 (𝑥) 𝑑𝑥 = 𝑓 (𝑥) 𝑑𝑥 + 𝑓 (𝑥) 𝑑𝑥.
∫−∞ ∫−∞ ∫𝑐

If either of the two new improper integrals diverges, we say the original integral diverges.
Only if both of the new integrals have a finite value do we add the values to get a finite value
for the original integral.

It is not hard to show that the preceding definition does not depend on the choice for 𝑐.
 7.6 IMPROPER INTEGRALS 389

Another Type of Improper Integral: When the Integrand Becomes Infinite

There is another way for an integral to be improper. The interval may be finite but the function may
1 √
be unbounded near some points in the interval. For example, consider ∫0 (1∕ 𝑥) 𝑑𝑥. Since the graph
√
of 𝑦 = 1∕ 𝑥 has a vertical asymptote at 𝑥 = 0, the region between the graph, the 𝑥-axis, and the
lines 𝑥 = 0 and 𝑥 = 1 is unbounded. Instead of extending to infinity in the horizontal direction as
in the previous improper integrals, this region extends to infinity in the vertical direction. See Fig-
1 √
ure 7.19(a). We handle this improper integral in a similar way as before: we compute ∫𝑎 (1∕ 𝑥) 𝑑𝑥
for values of 𝑎 slightly larger than 0 and look at what happens as 𝑎 approaches 0 from the positive
side. (This is written as 𝑎 → 0+ .)
First we compute the integral:
1
1 1∕2
|1 1∕2
∫𝑎 √ 𝑑𝑥 = 2𝑥 || = 2 − 2𝑎 .
𝑥 | 𝑎

(a) (b)
1 𝑑𝑥 1 𝑑𝑥
Area = ∫0 √ Area = ∫𝑎 √ ;
𝑥 𝑥
now let 𝑎 → 0
✠ ✠
1 1
√ √
𝑥 𝑥

𝑥 𝑥
1 𝑎 1
Figure 7.19: Area representation of improper integral

Now we take the limit:

1
1 1∕2
lim+ √ 𝑑𝑥 = lim+ (2 − 2𝑎 ) = 2.
𝑎→0 ∫𝑎 𝑥 𝑎→0

Since the limit is finite, we say the improper integral converges, and that
1
1
𝑑𝑥 = 2.
∫0 √𝑥

Geometrically, what we have done is to calculate the finite area between 𝑥 = 𝑎 and 𝑥 = 1 and take the
limit as 𝑎 tends to 0 from the right. See Figure 7.19(b). Since the limit exists, the integral converges
to 2. If the limit did not exist, we would say the improper integral diverges.

2
1
Example 5 Investigate the convergence of 𝑑𝑥.
∫0 (𝑥 − 2)2

Solution This is an improper integral since the integrand tends to infinity as 𝑥 approaches 2 and is undefined
at 𝑥 = 2. Since the trouble is at the right endpoint, we replace the upper limit by 𝑏, and let 𝑏 tend
to 2 from the left. This is written 𝑏 → 2− , with the “−” signifying that 2 is approached from below.
See Figure 7.20.

2 𝑏 |𝑏
( )
1 1 1 1
𝑑𝑥 = lim− 𝑑𝑥 = lim− (−1)(𝑥 − 2)−1 || = lim− − − .
∫0 (𝑥 − 2)2 𝑏→2 ∫0 (𝑥 − 2)2 𝑏→2 |0 𝑏→2 (𝑏 − 2) 2
1
( )
Therefore, since lim− − does not exist, the integral diverges.
𝑏→2 𝑏−2
390 Chapter 7 INTEGRATION
𝑦 𝑦
1
1 𝑦= 𝑥4
𝑦= (𝑥−2)2

𝑥 𝑥
2 −1 2
2 1 Figure 7.21: Shaded area
Figure 7.20: Shaded area represents ∫0 (𝑥−2)2
𝑑𝑥 2
represents ∫−1 𝑥14 𝑑𝑥

Suppose 𝑓 (𝑥) is positive and continuous on 𝑎 ≤ 𝑥 < 𝑏 and tends to infinity as 𝑥 → 𝑏.

𝑐 𝑏
If lim− 𝑓 (𝑥) 𝑑𝑥 is a finite number, we say that 𝑓 (𝑥) 𝑑𝑥 converges and define
𝑐→𝑏 ∫𝑎 ∫𝑎
𝑏 𝑐
𝑓 (𝑥) 𝑑𝑥 = lim− 𝑓 (𝑥) 𝑑𝑥.
∫𝑎 𝑐→𝑏 ∫𝑎
𝑏
Otherwise, we say that 𝑓 (𝑥) 𝑑𝑥 diverges.
∫𝑎

When 𝑓 (𝑥) tends to infinity as 𝑥 approaches 𝑎, we define convergence in a similar way. In

addition, an integral can be improper because the integrand tends to infinity inside the interval of
integration rather than at an endpoint. In this case, we break the given integral into two (or more)
improper integrals so that the integrand tends to infinity only at endpoints.

Suppose that 𝑓 (𝑥) is positive and continuous on [𝑎, 𝑏] except at the point 𝑐. If 𝑓 (𝑥) tends to
infinity as 𝑥 → 𝑐, then we define
𝑏 𝑐 𝑏
𝑓 (𝑥) 𝑑𝑥 = 𝑓 (𝑥) 𝑑𝑥 + 𝑓 (𝑥) 𝑑𝑥.
∫𝑎 ∫𝑎 ∫𝑐

If either of the two new improper integrals diverges, we say the original integral diverges.
Only if both of the new integrals have a finite value do we add the values to get a finite value
for the original integral.

2
1
Example 6 Investigate the convergence of 𝑑𝑥.
∫−1 𝑥4

Solution See Figure 7.21. The trouble spot is 𝑥 = 0, rather than 𝑥 = −1 or 𝑥 = 2. We break the given improper
integral into two improper integrals each of which has 𝑥 = 0 as an endpoint:
2 0 2
1 1 1
𝑑𝑥 = 𝑑𝑥 + 𝑑𝑥.
∫−1 𝑥4 ∫−1 𝑥4 ∫0 𝑥4
 7.6 IMPROPER INTEGRALS 391

We can now use the previous technique to evaluate the new integrals, if they converge. Since
2 |2
( )
1 1 1 1 1
( )
𝑑𝑥 = lim+ − 𝑥−3 || = lim+ − −
∫0 𝑥4 𝑎→0 3 |𝑎 𝑎→0 3 8 𝑎3
2
the integral ∫0 (1∕𝑥4 ) 𝑑𝑥 diverges. Thus, the original integral diverges. A similar computation shows
0
that ∫−1 (1∕𝑥4 ) 𝑑𝑥 also diverges.
It is easy to miss an improper integral when the integrand tends to infinity inside the interval.
2 |2
For example, it is fundamentally incorrect to say that ∫−1 (1∕𝑥4 ) 𝑑𝑥 = − 13 𝑥−3 | = − 24
1
− 13 = − 38 .
|−1

6
1
Example 7 Find 𝑑𝑥.
∫0 (𝑥 − 4)2∕3

Solution Figure 7.22 shows that the trouble spot is at 𝑥 = 4, so we break the integral at 𝑥 = 4 and consider
the separate parts.
𝑦
1
𝑦= (𝑥−4)2∕3

𝑥
4 6
6 1
Figure 7.22: Shaded area represents ∫0 (𝑥−4)2∕3
𝑑𝑥

We have
4
1 |𝑏
1∕3 | 1∕3 1∕3
= 3(4)1∕3 .
( )
𝑑𝑥 = lim 3(𝑥 − 4) = lim 3(𝑏 − 4) − 3(−4)
∫0 (𝑥 − 4)2∕3 𝑏→4− |
|0 𝑏→4 −

Similarly,
6
1 |6
𝑑𝑥 = lim+ 3(𝑥 − 4)1∕3 || = lim+ 3 ⋅ 21∕3 − 3(𝑎 − 4)1∕3 = 3(2)1∕3.
( )
∫4 (𝑥 − 4) 2∕3 𝑎→4 |𝑎 𝑎→4

Since both of these integrals converge, the original integral converges:

6
1
𝑑𝑥 = 3(4)1∕3 + 3(2)1∕3 = 8.542.
∫0 (𝑥 − 4)2∕3

Finally, there is a question of what to do when an integral is improper at both endpoints. In this
case, we just break the integral at any interior point of the interval. The original integral diverges if
either or both of the new integrals diverge.

∞
1
Example 8 Investigate the convergence of 𝑑𝑥.
∫0 𝑥2
392 Chapter 7 INTEGRATION

Solution This integral is improper both because the upper limit is ∞ and because the function is undefined at
∞
𝑥 = 0. We break the integral into two parts at, say, 𝑥 = 1. We know by Example 3 that ∫1 (1∕𝑥2 ) 𝑑𝑥
1
has a finite value. However, the other part, ∫0 (1∕𝑥2 ) 𝑑𝑥, diverges since:

1
1 |1 (
1
)
𝑑𝑥 = lim+ −𝑥−1 || = lim+ −1 .
∫0 𝑥2 𝑎→0 |𝑎 𝑎→0 𝑎
∞
1
Therefore 𝑑𝑥 diverges as well.
∫0 𝑥2

Exercises and Problems for Section 7.6 Online Resource: Additional Problems for Section 7.6
EXERCISES
4 𝜋∕2
1. Shade the area represented by: 1 sin 𝑥
17. √ 𝑑𝑥 18. 𝑑𝑥
∞ 1 √ ∫0 𝑥 ∫𝜋∕4 √cos 𝑥
(a) ∫1 (1∕𝑥2 ) 𝑑𝑥 (b) ∫0 (1∕ 𝑥) 𝑑𝑥
∞ 1 1
2. Evaluate the improper integral ∫0 𝑒−0.4𝑥 𝑑𝑥 and sketch 1 𝑥4 + 1
19. 𝑑𝑣 20. 𝑑𝑥
the area it represents. ∫0 𝑣 ∫0 𝑥
3. (a) Use a calculator or computer to estimate ∞
1
∞
1
𝑏
∫0 𝑥𝑒−𝑥 𝑑𝑥 for 𝑏 = 5, 10, 20. 21. 𝑑𝑥 22. 𝑑𝑥
∫1 𝑥2 + 1 ∫1
√
𝑥2 + 1
(b) Use your answers to part (a) to estimate the value
∞
of ∫0 𝑥𝑒−𝑥 𝑑𝑥, assuming it is finite. ∞ 𝑑𝑡 1
𝑑𝑡
23. 24.
4. (a) Sketch the the area represented by the improper in- ∫0 √ ∫−1 √𝑡 + 1
∞ 2 𝑡+1
tegral ∫−∞ 𝑒−𝑥 𝑑𝑥.
(b) Use a calculator or computer to estimate 4
−1
∞
𝑦
𝑎 2
∫−𝑎 𝑒−𝑥 𝑑𝑥 for 𝑎 = 1,2,3,4,5. 25. 𝑑𝑢 26. 𝑑𝑦
∫0 𝑢2 − 16 ∫1 𝑦4 + 1
(c) Use the answers to part (b) to estimate the value of
∞ 2
∫−∞ 𝑒−𝑥 𝑑𝑥, assuming it is finite. ∞ 𝑑𝑥 1 ln 𝑥
27. 28. 𝑑𝑥
∫2 𝑥 ln 𝑥 ∫0 𝑥
In Exercises 5–39, calculate the integral if it converges. You
may calculate the limit by appealing to the dominance of one 20 𝜋
1 1 −√𝑥
function over another, or by l’Hopital’s rule. 29. 𝑑𝑦 30. 𝑒 𝑑𝑥
∫16 𝑦2 − 16 ∫0 √𝑥
∞ ∞
1 1
5. 𝑑𝑥 6. 𝑑𝑥 ∞ 𝑑𝑥 2
1
∫1 5𝑥 + 2 ∫1 (𝑥 + 2)2 31. 32. 𝑑𝑥
∫3 𝑥(ln 𝑥)2 ∫0 √4 − 𝑥2
1 ∞ √
7. ln 𝑥 𝑑𝑥 8. 𝑒− 𝑥
𝑑𝑥
∫0 ∫0 ∞ 𝑑𝑥 ∞ 𝑑𝑥
∞ ∞
33. 34.
2
∫4 (𝑥 − 1)2 ∫4 𝑥2 −1
9. 𝑥𝑒−𝑥 𝑑𝑥 10. 𝑒−2𝑥 𝑑𝑥
∫0 ∫1
∞ 𝑑𝑦 3
𝑦 𝑑𝑦
∞ 𝑥 ∞ 35. 36.
𝑥 ∫7 ∫0
√ √
11. 𝑑𝑥 12. 𝑑𝑥 𝑦−5 9 − 𝑦2
∫0 𝑒𝑥 ∫1 4 + 𝑥2
∞ 𝜋∕2
0 𝑒𝑥 ∞ 𝑑𝑧 37. 𝑡𝑒−2𝑡 𝑑𝑡 38. tan 𝜃 𝑑𝜃
∫0 ∫0
13. 𝑑𝑥 14.
∫−∞ 1 + 𝑒𝑥 ∫−∞ 𝑧2 + 25
6
𝑑𝜃
∞ ∞
39.
𝑧 𝑧 ∫3 (4 − 𝜃)2
15. 𝑑𝑧 16. 𝑑𝑧
∫1 (1 + 𝑧2 )3 ∫0 3 + 𝑧2
 7.6 IMPROPER INTEGRALS 393

PROBLEMS
40. Find a formula (not involving integrals) for 50. The probability that a light bulb manufactured by a
company lasts at least 𝑎 hundred hours is
𝑥
𝑓 (𝑥) = 𝑒𝑡 𝑑𝑡. ∞
∫−∞ 0.012𝑒−0.012𝑡 𝑑𝑡.
∫𝑎
41. In statistics we encounter 𝑃 (𝑥), a function defined by The CEO claims that 90% of the company’s light bulbs
𝑥
last at least 1000 hours. Is this statement accurate?
1 2 ∞
𝑒−𝑡 𝑑𝑡. 2 √
𝑃 (𝑥) = √ 51. Given that ∫−∞ 𝑒−𝑥 𝑑𝑥 = 𝜋, calculate the exact value
𝜋 ∫0 of ∞
2 ∕𝑏
Use a calculator or computer to evaluate 𝑒−(𝑥−𝑎) 𝑑𝑥.
∫−∞
(a) 𝑃 (1) (b) 𝑃 (∞)
52. Assuming 𝑔(𝑥) is a differentiable function whose val-
42. Find the area under the curve 𝑦 = 𝑥𝑒−𝑥 for 𝑥 ≥ 0. ues are bounded for all 𝑥, derive Stein’s identity, which
is used in statistics:
43. Find the area under the curve 𝑦 = 1∕ cos2 𝑡 between
∞ ∞
𝑡 = 0 and 𝑡 = 𝜋∕2. 2 ∕2 2 ∕2
𝑔 ′ (𝑥)𝑒−𝑥 𝑑𝑥 = 𝑥𝑔(𝑥)𝑒−𝑥 𝑑𝑥.
∫−∞ ∫−∞
In Problems 44–47, evaluate 𝑓 (3).

∞ ∞ 53. Given that

44. 𝑓 (𝑥) = 𝑥−𝑡 𝑑𝑡 45. 𝑓 (𝑥) = 𝑡−𝑥 𝑑𝑡 ∞
∫0 ∫1 𝑥4 𝑒𝑥 4𝜋 4
𝑑𝑥 =
∞ ∞ ∫0 (𝑒𝑥 − 1) 2 15
2
46. 𝑓 (𝑥) = 𝑥𝑒−𝑥𝑡 𝑑𝑡 47. 𝑓 (𝑥) = 2𝑡𝑥𝑒−𝑡𝑥 𝑑𝑡
∫0 ∫0 evaluate ∞
𝑥4 𝑒2𝑥
𝑑𝑥.
∫0 (𝑒2𝑥 − 1)2
48. The rate, 𝑟, at which people get sick during an epidemic 54. Let 𝑥0 = 𝑎 = ln 3, 𝑥𝑛 = 𝑏, Δ𝑥 = (𝑏 − 𝑎)∕𝑛, 𝑥𝑖 =
of the flu can be approximated by 𝑟 = 1000𝑡𝑒−0.5𝑡 , 𝑥0 + 𝑖Δ𝑥. Evaluate
where 𝑟 is measured in people/day and 𝑡 is measured
in days since the start of the epidemic.
( 𝑛
)
∑
−𝑥𝑖
lim lim 𝑒 Δ𝑥
(a) Sketch a graph of 𝑟 as a function of 𝑡. 𝑏→∞ 𝑛→∞
𝑖=1
(b) When are people getting sick fastest?
(c) How many people get sick altogether?
55. Let 𝑥0 = 𝑎 = 4, 𝑥𝑛 = 𝑏, Δ𝑥 = (𝑏 − 𝑎)∕𝑛, 𝑥𝑖 = 𝑥0 + 𝑖Δ𝑥.
49. Find the energy required to separate opposite electric Evaluate ( )
𝑛
charges of magnitude 1 coulomb. The charges are ini- ∑ 1
lim lim Δ𝑥
tially 1 meter apart and one is moved infinitely far from 𝑏→∞ 𝑛→∞ 2
𝑖=1 𝑥𝑖
the other. (The definition of energy is on page 388.)

Strengthen Your Understanding

In Problems 56–57, explain what is wrong with the state- In Problems 60–65, decide whether the statements are true
ment. or false. Give an explanation for your answer.
∞ ∞
56. If both ∫1 𝑓 (𝑥) 𝑑𝑥 and ∫1 𝑔(𝑥) 𝑑𝑥 diverge, then so ∞
60. If 𝑓 is continuous for all 𝑥 and ∫0 𝑓 (𝑥) 𝑑𝑥 converges,
∞
does ∫1 𝑓 (𝑥)𝑔(𝑥) 𝑑𝑥. ∞
then so does ∫𝑎 𝑓 (𝑥) 𝑑𝑥 for all positive 𝑎.
∞
57. If ∫1 𝑓 (𝑥) 𝑑𝑥 diverges, then lim𝑥→∞ 𝑓 (𝑥) ≠ 0. ∞
61. If 𝑓 (𝑥) is a positive periodic function, then∫0 𝑓 (𝑥) 𝑑𝑥
In Problems 58–59, give an example of: diverges.

58. A function 𝑓 (𝑥), continuous for 𝑥 ≥ 1, such that 62. If 𝑓 (𝑥) is continuous and positive for 𝑥 > 0 and if
∞ ∞
lim𝑥→∞ 𝑓 (𝑥) = 0, but ∫1 𝑓 (𝑥)𝑑𝑥 diverges. lim𝑥→∞ 𝑓 (𝑥) = 0, then ∫0 𝑓 (𝑥) 𝑑𝑥 converges.
59. A function 𝑓 (𝑥), continuous at 𝑥 = 2 and 𝑥 = 5, such 63. If 𝑓 (𝑥) is continuous and positive for 𝑥 > 0 and if
5 ∞
that the integral ∫2 𝑓 (𝑥) 𝑑𝑥 is improper and divergent. lim𝑥→∞ 𝑓 (𝑥) = ∞, then ∫0 (1∕𝑓 (𝑥)) 𝑑𝑥 converges.
394 Chapter 7 INTEGRATION

∞ ∞
64. If ∫0 𝑓 (𝑥) 𝑑𝑥 and ∫0 𝑔(𝑥) 𝑑𝑥 both converge, then which are false. Give an explanation for your answer.
∞
∫0 (𝑓 (𝑥) + 𝑔(𝑥)) 𝑑𝑥 converges.
∞
∞ ∞
65. If ∫0 𝑓 (𝑥) 𝑑𝑥 and ∫0 𝑔(𝑥) 𝑑𝑥 both diverge, then 66. ∫0 𝑎𝑓 (𝑥) 𝑑𝑥 converges.
∞ ∞
∫0 (𝑓 (𝑥) + 𝑔(𝑥)) 𝑑𝑥 diverges. 67. ∫0 𝑓 (𝑎𝑥) 𝑑𝑥 converges.
∞
Suppose that 𝑓 is continuous for all real numbers and that 68. ∫0 𝑓 (𝑎 + 𝑥) 𝑑𝑥 converges.
∞
∫0 𝑓 (𝑥) 𝑑𝑥 converges. Let 𝑎 be any positive number. De- ∞
cide which of the statements in Problems 66–69 are true and 69. ∫0 (𝑎 + 𝑓 (𝑥)) 𝑑𝑥 converges.

7.7 COMPARISON OF IMPROPER INTEGRALS

Making Comparisons
Sometimes it is difficult to find the exact value of an improper integral by antidifferentiation, but it
may be possible to determine whether an integral converges or diverges. The key is to compare the
given integral to one whose behavior we already know. Let’s look at an example.

∞
1
Example 1 Determine whether 𝑑𝑥 converges.
∫1
√
𝑥3 + 5

Solution First, let’s see what this integrand does as 𝑥 → ∞. For large 𝑥, the 5 becomes insignificant compared
with the 𝑥3 , so
1 1 1
√ ≈ √ = 3∕2 .
𝑥3 + 5 𝑥3 𝑥
Since
∞
1
∞
1
𝑏
1 |𝑏
𝑑𝑥 = lim −2𝑥−1∕2 || = lim 2 − 2𝑏−1∕2 = 2,
( )
√ 𝑑𝑥 = 𝑑𝑥 = lim
∫1 ∫1 𝑥3∕2 𝑏→∞ ∫1 𝑥3∕2 𝑏→∞ 𝑏→∞
𝑥3 |1
∞
the integral ∫1 (1∕𝑥3∕2 ) 𝑑𝑥 converges. So we expect our integral to converge as well.
In order to confirm this, we observe that for 0 ≤ 𝑥3 ≤ 𝑥3 + 5, we have

1 1
√ ≤√ .
𝑥3 + 5 𝑥3

and so for 𝑏 ≥ 1,
𝑏 𝑏
1 1
𝑑𝑥 ≤ √ 𝑑𝑥.
∫1 √ ∫
𝑥3 + 5 1 𝑥3

𝑦
1
𝑦= √ ∞ 𝑑𝑥
𝑥3 Total shaded area = ∫1 √
𝑥3
∞ 1
Dark shaded area = ∫1 √ 𝑑𝑥
1 𝑥3 +5
𝑦= √
𝑥3 +5

𝑥
1
∞ 1 ∞ 𝑑𝑥
Figure 7.23: Graph showing ∫1 √ 𝑑𝑥 ≤ ∫1 √
𝑥3 +5 𝑥3
 7.7 COMPARISON OF IMPROPER INTEGRALS 395

𝑏
√
(See Figure 7.23.) Since ∫1 (1∕ 𝑥3 + 5) 𝑑𝑥 increases as 𝑏 approaches infinity but is always smaller
∞ ∞
𝑏
√
than ∫1 (1∕𝑥3∕2 ) 𝑑𝑥 < ∫1 (1∕𝑥3∕2 ) 𝑑𝑥 = 2, we know ∫1 (1∕ 𝑥3 + 5) 𝑑𝑥 must have a finite value
less than 2. Thus,
∞
𝑑𝑥
converges to a value less than 2.
∫1 √𝑥3 + 5

Notice that we first looked at the behavior of the integrand as 𝑥 → ∞. This is useful because
the convergence or divergence of the integral is determined by what happens as 𝑥 → ∞.

∞
The Comparison Test for 𝒇 (𝒙) 𝒅𝒙
∫𝒂
Assume 𝑓 (𝑥) is positive. Making a comparison involves two stages:
1. Guess, by looking at the behavior of the integrand for large 𝑥, whether the integral con-
verges or not. (This is the “behaves like” principle.)
2. Confirm the guess by comparison with a positive function 𝑔(𝑥):
∞ ∞
• If 𝑓 (𝑥) ≤ 𝑔(𝑥) and ∫𝑎 𝑔(𝑥) 𝑑𝑥 converges, then ∫𝑎 𝑓 (𝑥) 𝑑𝑥 converges.
∞ ∞
• If 𝑔(𝑥) ≤ 𝑓 (𝑥) and ∫𝑎 𝑔(𝑥) 𝑑𝑥 diverges, then ∫𝑎 𝑓 (𝑥) 𝑑𝑥 diverges.

∞
𝑑𝑡
Example 2 Decide whether converges or diverges.
∫4 (ln 𝑡) − 1

Solution Since ln 𝑡 grows without bound as 𝑡 → ∞, the −1 is eventually going to be insignificant in comparison
to ln 𝑡. Thus, as far as convergence is concerned,
∞ ∞
1 1
𝑑𝑡 behaves like 𝑑𝑡.
∫4 (ln 𝑡) − 1 ∫4 ln 𝑡
∞
Does ∫4 (1∕ ln 𝑡) 𝑑𝑡 converge or diverge? Since ln 𝑡 grows very slowly, 1∕ ln 𝑡 goes to zero very
slowly, and so the integral probably does not converge. We know that (ln 𝑡) − 1 < ln 𝑡 < 𝑡 for all
positive 𝑡. So, provided 𝑡 > 𝑒, we take reciprocals:

1 1 1
> > .
(ln 𝑡) − 1 ln 𝑡 𝑡
∞
Since ∫4 (1∕𝑡) 𝑑𝑡 diverges, we conclude that
∞
1
𝑑𝑡 diverges.
∫4 (ln 𝑡) − 1

How Do We Know What to Compare With?

In Examples 1 and 2, we investigated the convergence of an integral by comparing it with an easier
integral. How did we pick the easier integral? This is a matter of trial and error, guided by any
information we get by looking at the original integrand as 𝑥 → ∞. We want the comparison integrand
to be easy and, in particular, to have a simple antiderivative.
396 Chapter 7 INTEGRATION

Useful Integrals for Comparison

∞
1
• 𝑑𝑥 converges for 𝑝 > 1 and diverges for 𝑝 ≤ 1.
∫1 𝑥𝑝
1
1
• 𝑑𝑥 converges for 𝑝 < 1 and diverges for 𝑝 ≥ 1.
∫0 𝑥𝑝
∞
• 𝑒−𝑎𝑥 𝑑𝑥 converges for 𝑎 > 0.
∫0

Of course, we can use any function for comparison, provided we can determine its behavior.

∞
(sin 𝑥) + 3
Example 3 Investigate the convergence of 𝑑𝑥.
∫1 √
𝑥

Solution Since it looks difficult to find an antiderivative of this function, we try comparison. What happens
to this integrand as 𝑥 → ∞? Since sin 𝑥 oscillates between −1 and 1,

2 −1 + 3 (sin 𝑥) + 3 1 + 3 4
√ = √ ≤ √ ≤ √ =√ ,
𝑥 𝑥 𝑥 𝑥 𝑥
√ √
the integrand oscillates between 2∕ 𝑥 and 4∕ 𝑥. (See Figure 7.24.)
∞ √ ∞ √
What do ∫1 (2∕ 𝑥) 𝑑𝑥 and ∫1 (4∕ 𝑥) 𝑑𝑥 do? As far as convergence is concerned, they cer-
tainly do the
√ same thing, and whatever that is, the original
√ integral does it too. It is important to
notice that 𝑥 grows very slowly. This means that 1∕ 𝑥 gets small slowly, which means that con-
vergence is unlikely. Since 𝑥 = 𝑥1∕2 , the result in the preceding box (with 𝑝 = 12 ) tells us that
√
∞ √
∫1 (1∕ 𝑥) 𝑑𝑥 diverges. So the comparison test tells us that the original integral diverges.

𝑦
𝑏 (sin 𝑥)+3
Total shaded area = ∫1 √ 𝑑𝑥
𝑥
𝑏
Dark shaded area = ∫1 √2𝑥 𝑑𝑥
4
𝑦= √
𝑥 (sin 𝑥)+3
𝑦= √
𝑥 2
𝑦= √
𝑥
✠
✠ ✠
𝑥
1 𝑏
𝑏 2 𝑏 (sin 𝑥)+3
Figure 7.24: Graph showing ∫1 √
𝑥
𝑑𝑥 ≤ ∫1 √
𝑥
𝑑𝑥, for 𝑏 ≥ 1

Notice that there are two possible comparisons we could have made in Example 3:
2 (sin 𝑥) + 3 (sin 𝑥) + 3 4
√ ≤ √ or √ ≤√ .
𝑥 𝑥 𝑥 𝑥
∞ √ ∞ √
Since both ∫1 (2∕ 𝑥) 𝑑𝑥 and ∫1 (4∕ 𝑥) 𝑑𝑥 diverge, only the first comparison is useful. Knowing
that an integral is smaller than a divergent integral is of no help whatsoever!
 7.7 COMPARISON OF IMPROPER INTEGRALS 397

The next example shows what to do if the comparison does not hold throughout the interval of
integration.

∞
2 ∕2
Example 4 Show 𝑒−𝑥 𝑑𝑥 converges.
∫1
2
Solution We know that 𝑒−𝑥 ∕2 goes very rapidly to zero as 𝑥 → ∞, so we expect this integral to con-
verge. Hence we look for some larger integrand which has a convergent integral. One possibility
∞ ∞
is ∫1 𝑒−𝑥 𝑑𝑥, because 𝑒−𝑥 has an elementary antiderivative and ∫1 𝑒−𝑥 𝑑𝑥 converges. What is the
2
relationship between 𝑒−𝑥 ∕2 and 𝑒−𝑥 ? We know that for 𝑥 ≥ 2,

𝑥2 𝑥2
𝑥≤ so − ≤ −𝑥,
2 2
and so, for 𝑥 ≥ 2
2 ∕2
𝑒−𝑥 ≤ 𝑒−𝑥 .
Since this inequality holds only for 𝑥 ≥ 2, we split the interval of integration into two pieces:
∞ 2 ∞
2 ∕2 2 ∕2 2 ∕2
𝑒−𝑥 𝑑𝑥 = 𝑒−𝑥 𝑑𝑥 + 𝑒−𝑥 𝑑𝑥.
∫1 ∫1 ∫2
2 2 ∞ 2 ∞
Now ∫1 𝑒−𝑥 ∕2 𝑑𝑥 is finite (it is not improper) and ∫2 𝑒−𝑥 ∕2 𝑑𝑥 is finite by comparison with ∫2 𝑒−𝑥 𝑑𝑥.
∞ 2
Therefore, ∫1 𝑒−𝑥 ∕2 𝑑𝑥 is the sum of two finite pieces and therefore must be finite.

The previous example illustrates the following general principle:

If 𝑓 is positive and continuous on [𝑎, 𝑏],

∞ ∞
𝑓 (𝑥) 𝑑𝑥 and 𝑓 (𝑥) 𝑑𝑥
∫𝑎 ∫𝑏

either both converge or both diverge.

∞
In particular, when the comparison test is applied to ∫𝑎 𝑓 (𝑥) 𝑑𝑥, the inequalities for 𝑓 (𝑥) and
𝑔(𝑥) do not need to hold for all 𝑥 ≥ 𝑎 but only for 𝑥 greater than some value, say 𝑏.

Exercises and Problems for Section 7.7

EXERCISES
∞ ∞
In Exercises 1–9, use the box on page 396 and the behavior 𝑥 𝑥2 − 6𝑥 + 1
5. 𝑑𝑥 6. 𝑑𝑥
of rational and exponential functions as 𝑥 → ∞ to predict ∫1 𝑥2 + 2𝑥 + 4 ∫1 𝑥2 + 4
whether the integrals converge or diverge.
∞ ∞
5𝑥 + 2 1
∞ 2 ∞ 3
7. 𝑑𝑥 8. 𝑑𝑡
𝑥 𝑥 ∫1 𝑥4 + 8𝑥2 + 4 ∫1 𝑒5𝑡 + 2
1. 𝑑𝑥 2. 𝑑𝑥
∫1 𝑥4 + 1 ∫2 𝑥4 − 1

∞ ∞ ∞
𝑥2 + 1 1 𝑥2 + 4
3. 𝑑𝑥 4. 𝑑𝑥 9. 𝑑𝑥
∫1 𝑥3 + 3𝑥 + 2 ∫1 𝑥2 + 5𝑥 + 1 ∫1 𝑥4 + 3𝑥2 + 11
398 Chapter 7 INTEGRATION

1 1 5 𝑑𝑡
In Exercises 10–13, what, if anything, does the comparison
18. 𝑑𝑥 19.
tell us about the convergence of the integral? ∫0 𝑥19∕20 ∫−1 (𝑡 + 1)2
∞
sin2 𝑥 1
10. 𝑑𝑥, compare with ∞ 𝑑𝑢 ∞ 𝑑𝑢
∫1 𝑥2 𝑥2 20. 21.
∞ ∫−∞ 1 + 𝑢2 ∫1 𝑢 + 𝑢2
sin2 𝑥 1
11. 𝑑𝑥, compare with
∫1 𝑥 𝑥
∞ 𝑑𝜃 ∞ 𝑑𝜃
1
sin2 𝑥 1 22. 23.
12. 𝑑𝑥, compare with 2 ∫1 √ ∫2 √
∫0 𝑥2 𝑥 𝜃2 + 1 𝜃3 + 1
1
sin2 𝑥 1
13. √ 𝑑𝑥, compare with √ 1 𝑑𝜃 ∞ 𝑑𝑦
∫0 𝑥 𝑥 24. 25.
∫0 √
𝜃3 + 𝜃 ∫0 1 + 𝑒𝑦
In Exercises 14–29, decide if the improper integral con-
verges or diverges. ∞ 2 + cos 𝜙 ∞ 𝑑𝑧
26. 𝑑𝜙 27.
∫1 𝜙2 ∫0 𝑒𝑧 + 2𝑧
∞ 𝑑𝑧 ∞ 𝑑𝑥
14. 15.
∫50 𝑧3 ∫1 1+𝑥 𝜋 2 − sin 𝜙 ∞ 3 + sin 𝛼
28. 𝑑𝜙 29. 𝑑𝛼
∞ 8 ∫0 𝜙2 ∫4 𝛼
𝑑𝑥 6
16. 17. 𝑑𝑡
∫1 𝑥3 + 1 ∫5 √𝑡 − 5

PROBLEMS
30. The graphs of 𝑦 = 1∕𝑥, 𝑦 = 1∕𝑥2 and the functions
𝑓 (𝑥), 𝑔(𝑥), ℎ(𝑥), and 𝑘(𝑥) are shown in Figure 7.25.
(a) Is the area between 𝑦 = 1∕𝑥 and 𝑦 = 1∕𝑥2 on the
interval from 𝑥 = 1 to ∞ finite or infinite? Explain.
(b) Using the graph, decide whether the integral of
𝑓 (𝑥)
each of the functions 𝑓 (𝑥), 𝑔(𝑥), ℎ(𝑥) and 𝑘(𝑥) on
✠ 𝑔(𝑥)
the interval from 𝑥 = 1 to ∞ converges, diverges, ✠
or whether it is impossible to tell. 𝑥
𝑎

Figure 7.26

𝑘(𝑥) ∞
32. Suppose ∫𝑎 𝑓 (𝑥) 𝑑𝑥 converges. What does Figure 7.27
ℎ(𝑥) ∞
suggest about the convergence of ∫𝑎 𝑔(𝑥) 𝑑𝑥?

𝑓 (𝑥) ✛ 1∕𝑥 𝑔(𝑥)

✛ 1∕𝑥2
𝑔(𝑥)

1 𝑓 (𝑥)
Figure 7.25 ✠
𝑥
𝑎
∞
31. For 𝑓 (𝑥) in Figure 7.26, both 𝑓 (𝑥) 𝑑𝑥 and ∫𝑎
𝑎
∫0 𝑓 (𝑥) 𝑑𝑥 converge. Decide if the following integrals Figure 7.27
converge, diverge, or could do either. Assume that 0 <
𝑔(𝑥) < 𝑓 (𝑥) for 𝑥 > 𝑎 and 0 < 𝑓 (𝑥) < 𝑔(𝑥) for 𝑥 < 𝑎.
(a)
∞
∫𝑎 𝑔(𝑥) 𝑑𝑥
𝑎
(b) ∫0 𝑔(𝑥) 𝑑𝑥 For what values of 𝑝 do the integrals in Problems 33–34 con-
∞ ∞ verge or diverge?
(c) ∫0 𝑓 (𝑥) 𝑑𝑥 (d) ∫0 𝑔(𝑥) 𝑑𝑥
∞
∞ 1 ∞
𝑑𝑥
2
𝑑𝑥
(e) ∫𝑎 (𝑓 (𝑥) − 𝑔(𝑥)) 𝑑𝑥 (f) 𝑑𝑥 33. 34.
∫𝑎 𝑓 (𝑥) ∫2 𝑥(ln 𝑥)𝑝 ∫1 𝑥(ln 𝑥)𝑝
∞
(g) ∫𝑎 (𝑓 (𝑥))2 𝑑𝑥
 7.7 COMPARISON OF IMPROPER INTEGRALS 399

35. (a) Find an upper bound for (a) Explain why a graph of the tangent line to 𝑒𝑡 at
∞
2
𝑡 = 0 tells us that for all 𝑡
𝑒−𝑥 𝑑𝑥.
∫3
2
[Hint: 𝑒−𝑥 ≤ 𝑒−3𝑥 for 𝑥 ≥ 3.] 1 + 𝑡 ≤ 𝑒𝑡 .
(b) For any positive 𝑛, generalize the result of part (a)
to find an upper bound for (b) Substituting 𝑡 = 1∕𝑥, show that for all 𝑥 ≠ 0
∞
2
𝑒−𝑥 𝑑𝑥
∫𝑛 1
𝑒1∕𝑥 − 1 > .
by noting that 𝑛𝑥 ≤ 𝑥2 for 𝑥 ≥ 𝑛. 𝑥
36. In Planck’s Radiation Law, we encounter the integral
∞ (c) Use the comparison test to show that the original
𝑑𝑥
. integral converges.
∫1 𝑥5 (𝑒1∕𝑥 − 1)

Strengthen Your Understanding

In Problems 37–40, explain what is wrong with the state- In Problems 43–48, decide whether the statements are true
ment. or false. Give an explanation for your answer.
∞ ∞
37. ∫1 1∕(𝑥3 + sin 𝑥) 𝑑𝑥 converges by comparison with 1
∞ 43. The integral 𝑑𝑥 converges.
∫1 1∕𝑥3 𝑑𝑥. ∫0 𝑒𝑥 + 𝑥
∞
√ 1
38. ∫1 1∕(𝑥 2
+ 1) 𝑑𝑥 is divergent. 1
44. The integral 𝑑𝑥 diverges.
∞ ∫0 𝑥2 − 3
39. If 0 ≤ 𝑓 (𝑥) ≤ 𝑔(𝑥) and 𝑔(𝑥) 𝑑𝑥 diverges then by
∫0
∞
the comparison test ∫0 𝑓 (𝑥) 𝑑𝑥 diverges. 45. If 𝑓 (𝑥) < 𝑔(𝑥) for all 𝑥 and 𝑎 < 𝑏, then
∞ 𝑏 𝑏
40. Let 𝑓 (𝑥) > 0. If ∫1 𝑓 (𝑥) 𝑑𝑥 is convergent then so is
∞ 𝑓 (𝑥) 𝑑𝑥 < 𝑔(𝑥) 𝑑𝑥.
∫1 1∕𝑓 (𝑥) 𝑑𝑥. ∫𝑎 ∫𝑎

In Problems 41–42, give an example of: ∞

46. If 𝑓 (𝑥) < 𝑔(𝑥) for all 𝑥 and ∫𝑎 𝑔(𝑥) 𝑑𝑥 converges, then
41. A continuous function 𝑓 (𝑥) for 𝑥 ≥ 1 such that the im- ∞
∞
∫𝑎 𝑓 (𝑥) 𝑑𝑥 converges.
proper integral ∫1 𝑓 (𝑥)𝑑𝑥 can be shown to converge
∞ 47. If |𝑓 (𝑥)| < |𝑔(𝑥)| for all 𝑥 and 𝑎 < 𝑏, then
by comparison with the integral ∫1 3∕(2𝑥2 ) 𝑑𝑥.
𝑏 𝑏
42. A positive, continuous function 𝑓 (𝑥) such that
∞ 𝑓 (𝑥) 𝑑𝑥 < 𝑔(𝑥) 𝑑𝑥.
∫1 𝑓 (𝑥)𝑑𝑥 diverges and ∫𝑎 ∫𝑎
3
𝑓 (𝑥) ≤ , for 𝑥 ≥ 1. ∞
7𝑥 − 2 sin 𝑥 48. If |𝑓 (𝑥)| < |𝑔(𝑥)| for all 𝑥 and if ∫𝑎 |𝑔(𝑥)| 𝑑𝑥 con-
∞
verges, then ∫𝑎 |𝑓 (𝑥)| 𝑑𝑥 converges.

Online Resource: Review problems and Projects