4.

6 Techniques of integration

(integration by parts, integration

by substitution, integration by

partial fractions)
November 15, 2011 18:35 c04 Sheet number 17 Page number 281 cyan magenta yellow black

4.3 Integration by Substitution 281

✔QUICK CHECK ANSWERS 4.2

! !
1 √
1. F (x) = f(x) 2. (a) √ dx = x + C (b) cos x dx = sin x + C
2 x
3. (a) 41 x 4 + 21 x 2 + 5x + C (b) tan x + csc x + C 4. 2x + 1; x 2 + x + 3 5. 0; − 23

4.3 INTEGRATION BY SUBSTITUTION

In this section we will study a technique, called substitution, that can often be used to
transform complicated integration problems into simpler ones.

u-SUBSTITUTION
The method of substitution can be motivated by examining the chain rule from the viewpoint
of antidifferentiation. For this purpose, suppose that F is an antiderivative of f and that
g is a differentiable function. The chain rule implies that the derivative of F (g(x)) can be
expressed as d
[F (g(x))] = F (g(x))g (x)
dx
which we can write in integral form as
!
F (g(x))g (x) dx = F (g(x)) + C (1)

or since F is an antiderivative of f ,
!
f(g(x))g (x) dx = F (g(x)) + C (2)

For our purposes it will be useful to let u = g(x) and to write du/dx = g (x) in the differ-
ential form du = g (x) dx. With this notation (2) can be expressed as
!
f(u) du = F (u) + C (3)

The process of evaluating an integral of form (2) by converting it into form (3) with the
substitution
u = g(x) and du = g (x) dx
is called the method of u-substitution. Here our emphasis is not on the interpretation of
the expression du = g (x) dx. Rather, the differential notation serves primarily as a useful
“bookkeeping” device for the method of u-substitution. The following example illustrates
how the method works.
!
Example 1 Evaluate (x 2 + 1)50 · 2x dx.

Solution. If we let u = x 2 + 1, then du/dx = 2x, which implies that du = 2x dx. Thus,
the given integral can be written as
! !
u51 (x 2 + 1)51
(x 2 + 1)50 · 2x dx = u50 du = +C = +C
51 51

It is important to realize that in the method of u-substitution you have control over the
choice of u, but once you make that choice you have no control over the resulting expres-
sion for du. Thus, in the last example we chose u = x 2 + 1 but du = 2x dx was computed.
November 15, 2011 18:35 c04 Sheet number 18 Page number 282 cyan magenta yellow black

282 Chapter 4 / Integration

Fortunately, our choice of u, combined with the computed du, worked out perfectly to pro-
duce an integral involving u that was easy to evaluate. However, in general, the method of
u-substitution will fail if the chosen u and the computed du cannot be used to produce an inte-
grand in which no expressions involving x remain, or if you cannot evaluate the resulting in-
tegral. Thus, for example, the substitution u = x 2 , du = 2x dx will not work for the integral
!
2x sin x 4 dx

because this substitution results in the integral

!
sin u2 du

which still cannot be evaluated in terms of familiar functions.

In general, there are no hard and fast rules for choosing u, and in some problems no
choice of u will work. In such cases other methods need to be used, some of which will be
discussed later. Making appropriate choices for u will come with experience, but you may
find the following guidelines, combined with a mastery of the basic integrals in Table 4.2.1,
helpful.

Guidelines for u-Substitution

Step 1. Look for some composition f(g(x)) within the integrand for which the substi-
tution
u = g(x), du = g (x) dx
produces an integral that is expressed entirely in terms of u and its differential
du. This may or may not be possible.
Step 2. If you are successful in Step 1, then try to evaluate the resulting integral in terms
of u. Again, this may or may not be possible.
Step 3. If you are successful in Step 2, then replace u by g(x) to express your final
answer in terms of x.

EASY TO RECOGNIZE SUBSTITUTIONS

The easiest substitutions occur when the integrand is the derivative of a known function,
except for a constant added to or subtracted from the independent variable.

Example 2
! !
sin(x + 9) dx = sin u du = − cos u + C = − cos(x + 9) + C

u=x+9
du = 1 · dx = dx
! !
u24 (x − 8)24
(x − 8) dx =
23
u23 du = +C = +C
24 24
u=x−8
du = 1 · dx = dx

Another easy u-substitution occurs when the integrand is the derivative of a known
function, except for a constant that multiplies or divides the independent variable. The
following example illustrates two ways to evaluate such integrals.
November 15, 2011 18:35 c04 Sheet number 20 Page number 284 cyan magenta yellow black

284 Chapter 4 / Integration

!
Example 6 Evaluate sin2 x cos x dx.

Solution. If we let u = sin x, then

du
= cos x, so du = cos x dx
dx
Thus,
! !
u3 sin3 x
sin2 x cos x dx = u2 du = +C = +C
3 3

! √
cos x
Example 7 Evaluate √ dx.
x
√
Solution. If we let u = x, then
du 1 1 1
= √ , so du = √ dx or 2 du = √ dx
dx 2 x 2 x x
Thus,

! √ ! !
cos x √
√ dx = 2 cos u du = 2 cos u du = 2 sin u + C = 2 sin x + C
x

! 
3
Example 8 Evaluate t 4 3 − 5t 5 dt.

Solution.
!  ! !
3 1 √ 1
t 4 3 − 5t 5 dt = − 3
u du = − u1/3 du
25 25 u = 3 − 5t 5
du = −25t 4 dt or − 25
1
du = t 4 dt

1 u4/3 3
4/3
=− +C =− 3 − 5t 5 +C
25 4/3 100

LESS APPARENT SUBSTITUTIONS

The method of substitution is relatively straightforward, provided the integrand contains
an easily recognized composition f(g(x)) and the remainder of the integrand is a constant
multiple of g (x). If this is not the case, the method may still apply but may require more
computation.

!
√
Example 9 Evaluate x 2 x − 1 dx.

√
Solution. The composition x − 1 suggests the substitution
u=x−1 so that du = dx (4)
From the first equality in (4)
x 2 = (u + 1)2 = u2 + 2u + 1
November 15, 2011 18:35 c04 Sheet number 73 Page number 337 cyan magenta yellow black

4.9 Evaluating Definite Integrals by Substitution 337

4.9 EVALUATING DEFINITE INTEGRALS BY SUBSTITUTION

In this section we will discuss two methods for evaluating definite integrals in which a
substitution is required.

TWO METHODS FOR MAKING SUBSTITUTIONS IN DEFINITE INTEGRALS

Recall from Section 4.3 that indefinite integrals of the form
!
f(g(x))g (x) dx

can sometimes be evaluated by making the u-substitution

u = g(x), du = g (x) dx (1)
which converts the integral to the form
!
f(u) du

To apply this method to a definite integral of the form

! b
f(g(x))g (x) dx
a

we need to account for the effect that the substitution has on the x-limits of integration.
There are two ways of doing this.

Method 1.
First evaluate the indefinite integral
!
f(g(x))g (x) dx

by substitution, and then use the relationship

! b ! b
f(g(x))g (x) dx = f(g(x))g (x) dx
a a

to evaluate the definite integral. This procedure does not require any modification of the
x-limits of integration.

Method 2.
Make the substitution (1) directly in the definite integral, and then use the relationship
u = g(x) to replace the x-limits, x = a and x = b, by corresponding u-limits, u = g(a)
and u = g(b). This produces a new definite integral
! g(b)
f(u) du
g(a)

that is expressed entirely in terms of u.

! 2
Example 1 Use the two methods above to evaluate x(x 2 + 1)3 dx.
0

Solution by Method 1. If we let

u = x2 + 1 so that du = 2x dx (2)
November 15, 2011 18:35 c04 Sheet number 74 Page number 338 cyan magenta yellow black

338 Chapter 4 / Integration

then we obtain
! !
u4 1(x 2 + 1)4
x(x 2 + 1)3 dx = +C = +C
u3 du =
8 2 8
Thus, ! 2 ! 2
x(x + 1) dx =
2 3
x(x + 1) dx
2 3
0 x=0
4 2
(x + 1)
2
625 1
= = − = 78
8 x=0 8 8

Solution by Method 2. If we make the substitution u = x 2 + 1 in (2), then

if x = 0, u=1
if x = 2, u=5
Thus, ! 2 ! 5
1
x(x + 1) dx =
2 3
u3 du
0 2 1
5
u4 625 1
= = − = 78
8 u=1 8 8
which agrees with the result obtained by Method 1.

The following theorem states precise conditions under which Method 2 can be used.

4.9.1 theorem If g is continuous on [a, b] and f is continuous on an interval

containing the values of g(x) for a ≤ x ≤ b, then
! b ! g(b)

f(g(x))g (x) dx = f(u) du
a g(a)

proof Since f is continuous on an interval containing the values of g(x) for a ≤ x ≤ b,

it follows that f has an antiderivative F on that interval. If we let u = g(x), then the chain
rule implies that
d d dF du du
F (g(x)) = F (u) = = f(u) = f(g(x))g (x)
dx dx du dx dx
for each x in [a, b]. Thus, F (g(x)) is an antiderivative of f(g(x))g (x) on [a, b]. Therefore,
by Part 1 of the Fundamental Theorem of Calculus (4.6.1)
! b b ! g(b)

f(g(x))g (x) dx = F (g(x)) = F (g(b)) − F (g(a)) = f(u) du ■
a a g(a)

The choice of methods for evaluating definite integrals by substitution is generally a

matter of taste, but in the following examples we will use the second method, since the idea
is new.

Example 2 Evaluate
! π/8 ! 5
(a) sin5 2x cos 2x dx (b) (2x − 5)(x − 3)9 dx
0 2
 012 34567895
4 985 2

 !"%#&&''(#!)#1
$
$ *#(+,-&).,!)#1
!"#
/01234515672480198:1;4<<152:=91>014?@8A2>0214026BA>24801267304C:617><<6=1DEFGHIJFDKELMN
„“"…ˆ‰‹Šˆ† OJIFPQR3451267304C:617>01S61>@@<46=1281>1;4=61T>A462918U1U:07248051>0=1451@>A247:<>A<9
:56U:<1U8A14026BA>0=5140T8<T40B1VWXYZ[\]8U1><B6SA>471>0=12A>05760=602><1U:0724805^1_8A
g g g
”•––—˜™š›œ–›˜™–•žŸ  6A61451> 4052>076`4026BA>24801S91@>A251;8Aa51;6<<1;42314026BA><515:731>5
=4UU6A6021>@@A8>731281@A8T40B1236
U8A?:<>1U8A14026BA>24801S91@>A25^
¡c6A745612>a601UA8?1¢£A88U j1<01j1Yj` jn1mj1Yj` >0= mj15401j1Yj^
¤4238:21¤8A=5¥/026BA>24801S9
£>A25¦S91§8B6A1b^1’6<560` /026BA>24801S91@>A251451S>56=18012361U8A?:<>1U8A12361=6A4T>24T618U1>1@A8=:72
¨©\ªm«©\¬[]­¨©®©¯¬°m`±²`’8^
n`³@A4<1́µµ´`@^1́¶·`S9
@6A?45548018U12361>:238A^ 1YYjkZelfZ1YYjehe1YYZj
Å 1fZeiheZi
¿1À1ďj Å1À1Áj ;36A61S8231> e A61=4UU6A6024>S<61U:072480518U1j^/U1Zi>0=1ei>A61780240:8:5`98:
Z 0=1>
gg
]­À1ÁÂ
7>014026BA>261S823154=6518U1234516C:>248012818S2>40

gg
W1À1ÁÃ
1Zef Zei1Yjh eZi1Yj
¿ 1f Z1Yeh e1YZ^
V1À1ďÐ ¼1À1ď
b91A6;A4240B1234516C:>2480`98:18S2>4012361U8<<8;40B12368A6?^
³A6>1¾h³A6>1¾f¼]VW
g g ¹º
] V »¼`]½ opqrsqtuvwxyzoq{s|oyrzu}~u|so€
g ¹º g
1
Z1Y eh 11
e1YZf Ze
/U1>
Z 0=1>
g g
e A61U:072480518U1>
j 0=13>T61780240:8:51=6A4T>24T65`2360
W ¼ »VẀ½
] »¼`]½ V
W1
Z1Yef Ze»VẀ½1¼1Y eZ Z1YefZe e1YZ^
¡c@<>40138;123451BA>@31@A8T651236
2368A6?^1¤34731082>248014012345
@A88U1451:0U>?4<4>A¸1¤3>21=8198:
2340a1421?6>05¸ R3451U8A?:<>16c@A65565123618A4B40><14026BA><140126A?518U1>08236A14026BA><^1d6@60=40B
8012361738476518U1>Z 0=1Ye`421?>91S616>546A12816T><:>261236156780=14026BA><123>01236
8A4B40><1806^1b67>:5612361738476518U1> Z 0=1Ye>A617A4247><140123614026BA>24801S91@>A25
@A87655`2361U8<<8;40B1B:4=6<40651>A61@A8T4=6=^
‚*ƒ„…†„‡%ˆ‰†Š„‚‰‹Šˆ†Œ"‹‰Š‡
ŽQRA91<62240B1YeS612361?852178?@<47>26=1@8A248018U123614026BA>0=123>21U4251>
S>54714026BA>24801A:<6^1R3601;4
Z <<1S612361A6?>4040B1U>728A518U123614026BA>0=^
‘QRA91<62240B1S
Z 612361@8A248018U123614026BA>0=1;38561=6A4T>24T61451>1U:072480
54?@<6A123>01Z^R3601Ye;4<<1S612361A6?>4040B1U>728A518U123614026BA>0=^
’826123>21Ye><;>951407<:=6512361Yj8U123618A4B40><14026BA>0=^

ÞÕÆÇ
ÉÏÔÈÇÉÎÊÏßËÌÓÍÏàÎÏÇÐÈÑËÒÓÑÝÏáÆÓ
ÏÛàÕÔÔÌÔÕÓÌÝÓáÏÏÖÓ
ÇÊÕÏÝÊâÔÈËÔÙËÌà×ÕÏØÙ
ÎÓÝÙÏáÚË
ÏËÔÌÏãÍ
ÍÎÛÇÏÚÓ
ÙÓÏÛÇÓÊÊÏÜËÔÓÏÝÈ×ÕÊÎ×
P<Q RS T9
Q UVWXYZ[W\]V^_X`aV\bcXde^fghij\W[kgd^lckXe^[Vm^UnjZ]jXZ^UVWXYZ[kd

M
0123450 6 789
 984 9
?N?1?/
98 !"#$%"&!%"'(!)##"*%"#"+#"#$%"+#,%-
M M M M
O012/+#%#%#'#.#%*!'""+'/
J?KJN?1?K( JN?KJ?1?K( JGKJ?N?1?K( J?N?KJ1?K
0 12 0 12 03 12 0 12
+#$)##'$#456?')$$#'""+#,%'""&#7)'#"+##%.".#,0E?'
'-#%"+?(12EN1?'"+#-'"7-7"#%","+#"#$%"+"

opq0 9rs-#G("#"+""'
0E?
MM
,"'&'7"#$%",%-)/
12EN?1? 2E 12E N?1?EN?
10E1?
"#7#''%!"7)#7'"",

M
"#$%"*+#'.$
2E N?1?EN?ILG/
M M
8*("#$%"&!%"'%)7#'

M M
012E02B 210
?N?1?E?N?B N?1?
9"#$%"&!%"',%-)
:)&'"")"#/
)?'"%"#"+'(%#7#2EN?&!
2EN ILG!"#$%"&! E?N?BN?IL/ 9"#$%"#/
%"'"'##"+"!)&""+#'-#
%#')"/ 7+#7;"+'(,,#%#""#?N?BN?IL"'##"+"!)&""+#%$"#$%/

M
0123450 < 789
 984 9
?5?1?/
98 9"+'7'#(?5 '-%##'!"#$%"#"+5?/)%"+#%-%#("+#
12E?51?
0E ?
M
#%.".#,?''-#%"+?/:(!)'+)#"12E?1?/
2E ?51?E ?AA
10E G?1?

M M
9"#$%"&!%"'%)7#'

M MFMHFH
012E02B 210 9"#$%"&!%"',%-)
?5?1?E ?AA?B ?AA G?1? :)&'"")"#/
E ?AA?B AG?51? :-,!/

M
q0Rxop5pyz %!$%+$ ?A?B ?AIL/
?A ?A E A C 9"#$%"#/
?5?1?  A?B C
=)77+#7;"+'%#')"&!,,#%#""$/
!)%$%+$)""!/t!)$#"
"+#'-#$%+uv+'*";#
*+#('&#"#"/w 1? A C A ? @ D FH
1 ?A?B ?A E ?A G I J?KJ?5KB ?5E?5?
A >

“Š{|
~„‰}|~ƒ„”€ˆ‚„•ƒ„|…}†€‡ˆ†’„–{ˆ
„•Š‰‰‰Šˆ’ˆ–„„‹ˆ
|Š„’—‰}€‰Ž€•ŒŠ„Ž
ƒˆ’Ž„–€
„€‰„˜‚
‚ƒ|„ˆ
Žˆ„|ˆ„‘€‰ˆ„’}ŒŠƒŒ
abc de!+#f ghijklminohpqjrshntujvwpxyz{|nim}yvp~u}jwpmhpg€|lo|jlpghijklm}v
0123567389
93 593933
733173567389
7516
97192


\
  !"#$%#&'#()!*+!,&)#-.#+!%
/56XY56XZX
0&12!,&) 33Y
4719XY
656X
933 
3
71567389
735163739873
3957
75731X Y34123523986393
733957
7573156 X 136170181
137P[X
Z_[56XZX
P[XY
_[ 56XZX[]41X
ZP[YXZX
\
\
9168567389
7516
979143
\
XY56XZX[]XY41X^ YX41XZX /59731567389
7516
97
3559731567389
7516
97
443335652556873195856
567389
8
773567389
1673958775136:757

54567389
7516933137

737
7

P[YX
_[ 41XZX[56X
ZP[YZX
\
567389
814
6
567389
7516
97
8
56357523837P[YX
Z_[41XZX

\
9168567389
7516
979143
YX41XZX[YX56X] Y56XZX
[YX56X^Y41X^`
\ 0341631567389
7516
97

\
;1256568733761937814
669573
XY56XZX[]XY41X^YX56X^Y41X^` <

=362
>568933
73
54
75161567389
7516
97816337134
93?
61771567394
68373757751656Y443573
54
7516/19567
643856
@A
23B87359775775166
P[X
6Z_[56XZXC856733416

54
751681
6574373757751671P[41X
6Z_[YX8161

‚(ƒ
\

7317
563
XY56XZX[]XY41X^ YX41XZX \\ \
[]XY41X^XY41X^ XY56XZX[ XY56XZX
\
397156
RX41YXZX 7393615687393751567389
7516
6937965687173DEFGFHIJ567389
=36
2
>568933
73
54
75161567389
7516
97811
16
741973
3775X68P[41YX
6
3
9
6431
KDHLMIHMNOPJMFQJR173195856
567X389
/19567
6438751449
Z_[RZX56735977577516 63613567389
7516
977137

73WR41YXZX8
6
1144956
/19733416757X7516837 @A
23S167363A7
83
P[56YX
6Z_[RZX 33567389
56@A
23S5
652197
67163C60347516TBU@A
23SV81
65337
75753715673
9436871

9
154382367

œ“„…
‡’†…‡Œˆ‰Š‘‹žŒ…Ž†‰‘›Ÿ„‘
™ž“’’Š’“‘Š›‘Ÿ”‘
…ˆ“›ˆ ’†‰’—‰Šž•“–—
Œ‘›—Ÿ˜‰
‰’Š¡‹
‹Œ™…˜‘
—‘™…‘ˆˆš‰’‘›†•“ˆŒ•
September 29, 2011 17:59 c07 Sheet number 8 Page number 495 cyan magenta yellow black

7.2 Integration by Parts 495

which is an equation we can solve for the unknown integral. We obtain

!
2 ex cos x dx = ex cos x + ex sin x

and hence !
ex cos x dx = 21 ex cos x + 21 ex sin x + C

A TABULAR METHOD FOR REPEATED INTEGRATION BY PARTS

Integrals of the form !
p(x)f(x) dx
More information on tabular integra-
tion by parts can be found in the ar-
where p(x) is a polynomial, can sometimes be evaluated using repeated integration by
ticles “Tabular Integration by Parts,”
College Mathematics Journal, Vol. 21,
parts in which u is taken to be p(x) or one of its derivatives at each stage. Since du is
1990, pp. 307–311, by David Horowitz computed by differentiating u, the repeated differentiation of p(x) will eventually produce
and “More on Tabular Integration by 0, at which point you may be left with a simplified integration problem. A convenient
Parts,” College Mathematics Journal, method for organizing the computations into two columns is called tabular integration by
Vol. 22, 1991, pp. 407–410, by Leonard
parts.
Gillman.

Tabular Integration by Parts

Step 1. Differentiate p(x) repeatedly until you obtain 0, and list the results in the first
column.
Step 2. Integrate f(x) repeatedly and list the results in the second column.
Step 3. Draw an arrow from each entry in the first column to the entry that is one row
down in the second column.
Step 4. Label the arrows with alternating + and − signs, starting with a +.
Step 5. For each arrow, form the product of the expressions at its tip and tail and then
multiply that product by +1 or −1 in accordance with the sign on the arrow.
Add the results to obtain the value of the integral.

"
This process is illustrated in Figure 7.2.1 for the integral (x 2 − x) cos x dx.

repeated repeated
differentiation integration

x2 − x + cos x
2x − 1 − sin x
2 + −cos x
0 −sin x

(x 2 − x) cos x dx = (x 2 − x) sin x + (2x − 1) cos x − 2 sin x + C

= (x 2 − x − 2) sin x + (2x − 1) cos x + C
Figure 7.2.1

" √
Example 6 In Example 9 of Section 4.3 we evaluated x 2 x − 1 dx using u-sub-
stitution. Evaluate this integral using tabular integration by parts.
September 29, 2011 17:59 c07 Sheet number 9 Page number 496 cyan magenta yellow black

496 Chapter 7 / Principles of Integral Evaluation

Solution.
repeated repeated
differentiation integration

x2 + (x − 1)1/2
2x − 2_ (x
3
− 1)3/2
2 + 4 (x
15
− 1)5/2
0 8 (x
105
− 1)7/2

The result obtained in Example 6 looks

quite different from that obtained in Ex- Thus, it follows that
!
ample 9 of Section 4.3. Show that the √
two answers are equivalent. x 2 x − 1 dx = 23 x 2 (x − 1)3/2 − 8
15
x(x − 1)5/2 + 16
105
(x − 1)7/2 + C

INTEGRATION BY PARTS FOR DEFINITE INTEGRALS

For definite integrals the formula corresponding to (3) is
! b b ! b
u dv = uv − v du (7)
a a a

REMARK It is important to keep in mind that the variables u and v in this formula are functions of x and that
the limits of integration in (7) are limits on the variable x . Sometimes it is helpful to emphasize this
by writing (7) as
! b b ! b
u dv = uv − v du (8)
x=a x=a x=a

The next example illustrates how integration by parts can be used to integrate the inverse
trigonometric functions.
! 1
Example 7 Evaluate tan−1 x dx.
0

Solution. Let
1
u = tan−1 x, dv = dx, du = dx, v=x
1 + x2
Thus, 1
! 1 ! 1 ! 1
−1 The limits of integration refer to x;
tan x dx = u dv = uv − v du that is, x = 0 and x = 1.
0 0 0 0
1 ! 1
x
= x tan−1 x − dx
0 0 1 + x2
But ! ! 1
1 1
x 1 2x 1 1
dx = dx = ln(1 + x 2 ) = ln 2
0 1+x 2 2 0 1+x 2 2 0 2
so
! 1 π  1
1
1 π √
tan−1 x dx = x tan−1 x − ln 2 = − 0 − ln 2 = − ln 2
0 0 2 4 2 4
September 29, 2011 17:59 c07 Sheet number 10 Page number 497 cyan magenta yellow black

7.2 Integration by Parts 497

REDUCTION FORMULAS
Integration by parts can be used to derive reduction formulas for integrals. These are
formulas that express an integral involving a power of a function in terms of an integral that
involves a lower power of that function. For example, if n is a positive integer and n ≥ 2,
then integration by parts can be used to obtain the reduction formulas
! !
1 n−1 n−1
sin x dx = − sin
n
x cos x + sinn−2 x dx (9)
n n
! !
1 n−1
cosn x dx = cosn−1 x sin x + cosn−2 x dx (10)
n n

To illustrate how such formulas can be obtained, let us derive (10). We begin by writing
cosn x as cosn−1 x · cos x and letting
u = cosn−1 x dv = cos x dx
du = (n − 1) cosn−2 x(− sin x) dx v = sin x
= −(n − 1) cos n−2
x sin x dx
so that
! ! ! !
cos x dx =
n
cos n−1
x cos x dx = u dv = uv − v du
!
= cos n−1
x sin x + (n − 1) sin2 x cosn−2 x dx
!
= cos n−1
x sin x + (n − 1) (1 − cos2 x) cosn−2 x dx
! !
= cos n−1
x sin x + (n − 1) cosn−2
x dx − (n − 1) cosn x dx

Moving the last term on the right to the left side yields
! !
n cos x dx = cos x sin x + (n − 1) cosn−2 x dx
n n−1

from which (10) follows. The derivation of reduction formula (9) is similar (Exercise 63).
Reduction formulas (9) and (10) reduce the exponent of sine (or cosine) by 2. Thus,
if the formulas are applied repeatedly, the exponent can eventually be reduced to 0 if n is
even or 1 if n is odd, at which point the integration can be completed. We will discuss this
method in more detail in the next section, but for now, here is an example that illustrates
how reduction formulas work.
!
Example 8 Evaluate cos4 x dx.

Solution. From (10) with n = 4

! !
cos4 x dx = 1
4
cos3 x sin x + 3
4
cos2 x dx Now apply (10) with n = 2.

 ! 
= 1
4
cos3 x sin x + 3
4
1
2
cos x sin x + 1
2
dx

= 1
4
cos3 x sin x + 38 cos x sin x + 38 x + C
7.4 Integration by Partial Fractions
The method of partial fractions is used to integrate rational functions. That is, we want to compute

P( x )
Z
dx where P, Q are polynomials.
Q( x )

R( x )
First reduce1 the integrand to the form S( x ) + Q( x )
where °R < °Q.

Example Here we write the integrand as a polynomial plus a rational function x+7 2 whose denom-
inator has higher degreee than its numerator. Thankfully, this expression can be easily integrated
using logarithms.

x2 + 3 x ( x + 2) − 2x + 3 −2( x + 2) + 4 + 3 7
= = x+ = x−2+
x+2 x+2 x+2 x+2
x2 + 3 7 1
Z Z
=⇒ dx = x−2+ dx = x2 − 2x + 7 ln | x + 2| + c
x+2 x+2 2

What if °Q ≥ 2?

If the denominator Q( x ) is quadratic or has higher degree, we need another trick:

R( x )
Theorem. Suppose that °R < °Q. Then the rational function Q( x )
can be written as a sum of fractions of the
form

A Ax + B
( ax + b)m ( ax2+ bx + c)n

where A, B, a, b, c are constants and m, n are positive integers.

Expressions such as the above can all be integrated using either logarithms or trigonometric substi-
tutions.

Example With a little experimenting, you should be convinced that

3x2 + 2x + 3 3 2
3
= +
x +x x 1 + x2

It follows that

3x2 + 2x + 3
Z
dx = 3 ln | x | + 2 tan−1 x + c
x3 + x

The burning question is how to find the expressions in the Therorem. The approach depends on
the form of the denominator Q( x ).
1 By Long Division or some other Torture. . .

1
Case 1: Distinct Linear Factors
Suppose that our denominator can be factorized completely into distinct linear factors. That is
Q( x ) = ( x − a1 )( x − a2 ) · · · ( x − an )
where the values a1 , . . . , an are all different.2
Theorem. For such a Q, there exist constants A1 , . . . , An such that
n
R( x ) Ai A1 An
=∑ = +···+ (∗)
Q( x ) i =1
x − a i x − a 1 x − an
whence the integral can be easily computed term-by-term:
n n
R( x ) Ai
Z Z
dx = ∑ dx = ∑ Ai ln | x − ai | + c
Q( x ) i =1
x − ai i =1

We find the constants Ai by putting the right hand side of (∗) over the common denominator Q( x )
R( x ) R( x ) A1 An
= = +···+
Q( x ) ( x − a1 ) · · · ( x − a n ) x − a1 x − an
and comparing numerators.

Examples
1. According to the Theorem, there exist constants A, B such that
x+8 x+8 A B
= = +
x2 +x−2 ( x − 1)( x + 2) x−1 x+2
Summing the right hand side, we obtain
x+8 A ( x + 2) + B ( x − 1)
=
( x − 1)( x + 2) ( x − 1)( x + 2)
Since the denominators are equal, it follows that the numerators are equal:
x + 8 = A ( x + 2) + B ( x − 1)
This is a relationship between A, B which holds for all3 x: every value of x gives a valid rela-
tionship between A and B. Evaluating at x = 1 and x = −2 gives two very simple expressions:
x=1: 9 = 3A =⇒ A = 3
x = −2 : 6 = −3B =⇒ B = −2
Putting it all together, we have
x+8 3 2
Z Z
2
dx = − dx = 3 ln | x − 1| − 2 ln | x + 2| + c
x +x−2 x−1 x+2
| x − 1|3
= ln +c
| x + 2|2
2 We assume for clarity that the leading term of Q( x ) is x n (coefficient 1). If not, absorb it into the numerator!
3 You might worry that it doesn’t when x = 1 or x = −2 because of the denominator. The fact fact that polynomials are
continuous combined with x + 8 = A( x + 2) + B( x − 1) everywhere else guarantees that we have equality everywhere.

2
 2. We know that there exist constants A, B, C such that

x2 + 2 x2 + 2 A B C
3
= = + +
x −x x ( x − 1)( x + 1) x x−1 x+1

Combining the right hand side yields

x2 + 2 = A( x − 1)( x + 1) + Bx ( x + 1) + Cx ( x − 1)

Now evaluate at x = 0, ±1:

x=0: 2 = − A =⇒ A = −1
3
x=1: 3 = 2B =⇒ B =
2
3
x = −1 : 3 = 2C =⇒ C =
2
It follows that
x2 + 2 −2 3 3
Z Z
dx = + + dx
x3 − x x 2( x − 1) 2( x + 1)
3
= −2 ln | x | + (ln | x − 1| + ln| x + 1|) + c
2
3
| x 2 − 1| 2
= ln +c
x2

Case 2: Repeated Linear Factors

Suppose that when we factorize Q( x ) we obtain a repeated linear factor. That is, some term of the
form ( x − a)m where m ≥ 2. In a partial fractions decomposition, such a factor produces m seperate
contributions:
A1 A2 Am
+ +···+
x − a ( x − a) 2 ( x − a)m
each of which can be integrated normally. One way to remember this is to count the constants:
( x − a)m has degree m and must therefore correspond to m distinct terms.

Examples
x −2
1. has a repeated factor of x in the denominator. The single factor of x − 1 behaves exactly
x 2 ( x −1)
as in Case 1. We therefore have constants A, B, C such that
x−2 A B C
= + 2+
x 2 ( x − 1) x x x−1

Combining the right hand side and cancelling the denominators yields4

x − 2 = Ax ( x − 1) + B( x − 1) + Cx2 (†)
4 Be careful: think about what each term is missing compared to the common denominator.

3
 There are only two nice places at which to evaluate this expression:

x=0: − 2 = − B =⇒ B = 2
x=1: −1 = C

To obtain A we have choices. Either evaluate (†) at another value of x, or compare coefficients.
For example, it is easy to see that the coefficient of x2 on the right side of (†) is A + C. This is
clearly zero, since ther is no x2 term on the left. We might write this as

coeff( x2 ) : 0 = A + C =⇒ A = −C = 1

Putting it all together, we have

x−2 1 2 1 |x| 2
Z Z
dx = + − dx = ln − +c
x 2 ( x − 1) x x2 x−1 | x − 1 | x

x3 + 3x + 1
2. Suppose we want to integrate . We have two repeated factors, whence there
( x + 1)2 ( x − 2)2
exist constants A, B, C, D such that

x3 + 3x + 1 A B C D
2 2
= + 2
+ +
( x + 1) ( x − 2) x + 1 ( x + 1) x − 2 ( x − 2)2

Combining the right hand side and cancelling the denominators yields

x3 + 3x + 1 = A( x + 1)( x − 2)2 + B( x − 2)2 + C ( x + 1)2 ( x − 2) + D ( x + 1)2

We evaluate at the two nice places then compare some coefficients and evaluate at x = 0:

5
x=2: 15 = 9D =⇒ D =
3
1
x = −1 : − 3 = 9B =⇒ B = −
3
coeff( x3 ) : 1 = A+C
1
x=0: 1 = 4A + 4B − 2C + D =⇒ 2A − C =
3
4
The last two equations can be solved to obtain A = 9 and C = 95 . The final integral is then

x3 + 3x + 1 4 1 5 5
Z Z
dx = − + + dx
( x + 1)2 ( x − 2)2 9( x + 1) 3( x + 1)2 9( x − 2) 3( x − 2)2
4 1 5 5
= ln | x + 1| + + ln | x − 2| − +c
9 3( x + 1) 9 3( x − 2)
1 1 5
= ln | x + 1|4 | x − 2|5 + − +c
9 3( x + 1) 3( x − 2)

4
Case 3: Quadratic Factors

Suppose that the denominator Q( x ) contains an irreducible quadratic term: a term of the form5

ax2 + bx + c where b2 − 4ac < 0

Each such factor generates a partial fraction of the form

Ax + B
ax2 + bx + c

which can be integrated using logarithms and/or tangent substitutions.6

x2 − x + 2 x2 − x + 2
Example The rational function = contains the irreduciuble quadratic x2 + 4
x3 + 4x x ( x 2 + 4)
in its denominator. We therefore know that there exist constants A, B, C such that

x2 − x + 2 A Bx + C
3
= + 2
x + 4x x x +4
Combining the right hand side and equating numerators yields

x2 − x + 2 = A( x2 + 4) + ( Bx + C ) x

which can be solved (try it!) to obtain

1 1
A= , B= , C = −1
2 2
It follows that

x2 − x + 2 1 x−2 1 x 1
Z Z Z
dx = + dx = ln | x | + − 2 dx
x3 + 4x 2
2x 2( x + 4) 2 2
2( x + 4) x + 4
1 1 1 x
= ln | x | + ln( x2 + 4) − tan−1 + c
2 4 2 2
We had to be a little creative with the quadratic term in order to find an anti-derivative.

Case 4: Repeated Quadratic Factors (very hard!)

If Q( x ) contains a repeated factor ( ax2 + bx + c)m where ax2 + bx + c is irreducible and m ≥ 2, then
each such expression yields the m terms

A1 x + B1 A2 x + B2 Am x + Bm
+ +···+
2 2
ax + bx + c ( ax + bx + c) 2 ( ax2 + bx + c)m

Each term may be integrated similarly to Case 3: part by inspection, part by completing the square.
5 Thus ax2 + bx + c cannot be factored (over R) into linear terms.
6 Warning: These examples are often very involved. Master Cases 1 and 2 first!

5
 x3 + 2x2 + 4
(Partial) Example To integrate we first seek a partial fraction de-
( x2 + 2x + 5)2 ( x − 3)4 ( x − 2)2
composition:

x3 + 2x2 + 4 Ax + B Cx + D
= 2 +
( x2 + 2x + 5)2 ( x − 3)4 ( x − 2)2 x + 2x + 5 ( x2 + 2x + 5)2
E F G H
+ + 2
+ 3
+
x − 3 ( x − 3) ( x − 3) ( x − 3)4
I J
+ +
x − 2 ( x − 2)2

This is long and messy. The first two terms may be integrated by completing the square and substi-
tuting u = x + 1

x2 + 2x + 5 = ( x + 1)2 + 4 = u2 + 1

The integral of these terms will then be a combination of expressions such as

u
tan−1 , ln(u2 + 1), ( u 2 + 1 ) −1
2
If you’re interested in the solution, ask a computer to help: the mathematician in you should be
comfortable believing that it could be done!

Rationalizing
A clever substitution can sometimes convert an irrational expression into a rational one, to which the
partial fractions method may be applied.
For example, the substitution u3 = x − 7 (dx = 3u2 du) gives
Z √
3
x−7 3u3 24
Z Z
dx = 3
du = 3 − du
x+1 u +8 (u + 2)(u2 − 2u + 4)
u2 − 2u + 4 √ u−1
= 3u + ln 2
− 2 3 tan−1 √ + c (partial fractions in here)
( u + 2) 3
( x − 7)2/3 − 2( x − 7)1/3 + 4 √ −1 ( x − 7 )
1/3 − 1
= 3( x − 7)1/3 + ln − 2 3 tan √ +c
(( x − 7)1/3 + 2)2 3
√
A similar approach (substituting u = x − 2) rationalizes the integral

1 2 du
Z Z
√ dx =
( x − 2)( x − 2 + x − 2) u2 ( u+ 1)

Suggested problems
1. Evaluate the integrals:
8
Z
(a) dx
( x − 2)( x + 6)

6
 x
Z
(b) dx
( x − 6)( x + 2)2
2. Evaluate the integrals:

8 − x2
Z 2
(a) 2
dx
1 x ( x + 5x + 8)
1
Z
(b) dy
y + 3y2 + 1
4

3. Evaluate x2dx in two ways: using partial fractions and using a trigonometric substitution.7
R
−1
Reconcile your two answers.

7 Look up the integral of csc θ if you need to. . .

7
4100 AWL/Thomas_ch08p553-641 8/20/04 10:07 AM Page 576

576 Chapter 8: Techniques of Integration

HISTORICAL BIOGRAPHY The Heaviside “Cover-up” Method for Linear Factors

Oliver Heaviside When the degree of the polynomial ƒ(x) is less than the degree of g(x) and
(1850–1925)
gsxd = sx - r1 dsx - r2 d Á sx - rn d
is a product of n distinct linear factors, each raised to the first power, there is a quick way
to expand ƒ(x)> g(x) by partial fractions.

EXAMPLE 6 Using the Heaviside Method

Find A, B, and C in the partial-fraction expansion
x2 + 1 A B C
= + + . (3)
sx - 1dsx - 2dsx - 3d x - 1 x - 2 x - 3

Solution If we multiply both sides of Equation (3) by sx - 1d to get

x2 + 1 Bsx - 1d Csx - 1d
= A + +
sx - 2dsx - 3d x - 2 x - 3
and set x = 1, the resulting equation gives the value of A:
s1d2 + 1
= A + 0 + 0,
s1 - 2ds1 - 3d
A = 1.
Thus, the value of A is the number we would have obtained if we had covered the factor
sx - 1d in the denominator of the original fraction

x2 + 1
(4)
sx - 1dsx - 2dsx - 3d

and evaluated the rest at x = 1:

s1d2 + 1 2
A = = = 1.
 sx - 1d  s1 - 2ds1 - 3d s -1ds -2d
y
Cover

Similarly, we find the value of B in Equation (3) by covering the factor sx - 2d in Equa-
tion (4) and evaluating the rest at x = 2:

s2d2 + 1 5
B = = = -5.
s2 - 1d  sx - 2d  s2 - 3d s1ds -1d
y
Cover

Finally, C is found by covering the sx - 3d in Equation (4) and evaluating the rest at
x = 3:

s3d2 + 1 10
C = = = 5.
s3 - 1ds3 - 2d sx - 3d  s2ds1d
y
Cover

Copyright © 2005 Pearson Education, Inc., publishing as Pearson Addison-Wesley

4100 AWL/Thomas_ch08p553-641 08/24/04 2:53 PM Page 577

8.3 Integration of Rational Functions by Partial Fractions 577

Heaviside Method
1. Write the quotient with g(x) factored:
ƒsxd ƒsxd
= .
gsxd sx - r1 dsx - r2 d Á sx - rn d
2. Cover the factors sx - ri d of gsxd one at a time, each time replacing all the
uncovered x’s by the number ri . This gives a number Ai for each root ri :
ƒsr1 d
A1 =
sr1 - r2 d Á sr1 - rn d
ƒsr2 d
A2 =
sr2 - r1 dsr2 - r3 d Á sr2 - rn d
o
ƒsrn d
An = .
srn - r1 dsrn - r2 d Á srn - rn - 1 d
3. Write the partial-fraction expansion of ƒsxd>gsxd as
ƒsxd A1 A2 An
= + + Á + .
gsxd sx - r1 d sx - r2 d sx - rn d

EXAMPLE 7 Integrating with the Heaviside Method

Evaluate
x + 4
dx.
L x + 3x - 10x
3 2

Solution The degree of ƒsxd = x + 4 is less than the degree of gsxd = x 3 + 3x 2

- 10x, and, with g(x) factored,
x + 4 x + 4
3 2
= .
x + 3x - 10x xsx - 2dsx + 5d
The roots of g(x) are r1 = 0, r2 = 2, and r3 = -5. We find
0 + 4 4 2
A1 = = = -

x
s0 - 2ds0 + 5d s -2ds5d 5
y
Cover
2 + 4 6 3
A2 = = =
2
sx - 2d
s2 + 5d s2ds7d 7
y
Cover
-5 + 4 -1 1
A3 = = = - .
s -5ds -5 - 2d
sx + 5d
s -5ds -7d 35
y
Cover

Copyright © 2005 Pearson Education, Inc., publishing as Pearson Addison-Wesley

4100 AWL/Thomas_ch08p553-641 8/20/04 10:07 AM Page 578

578 Chapter 8: Techniques of Integration

Therefore,

x + 4 2 3 1
= - + - ,
xsx - 2dsx + 5d 5x 7sx - 2d 35sx + 5d

and
x + 4 2 3 1
dx = - ln ƒ x ƒ + ln ƒ x - 2 ƒ - ln ƒ x + 5 ƒ + C.
L xsx - 2dsx + 5d 5 7 35

Other Ways to Determine the Coefficients

Another way to determine the constants that appear in partial fractions is to differentiate,
as in the next example. Still another is to assign selected numerical values to x.

EXAMPLE 8 Using Differentiation

Find A, B, and C in the equation
x - 1 A B C
= + + .
sx + 1d3 x + 1 sx + 1d2 sx + 1d 3

Solution We first clear fractions:

x - 1 = Asx + 1d2 + Bsx + 1d + C.

Substituting x = -1 shows C = -2. We then differentiate both sides with respect to x,

obtaining

1 = 2Asx + 1d + B.

Substituting x = -1 shows B = 1. We differentiate again to get 0 = 2A, which shows

A = 0. Hence,

x - 1 1 2
3
= 2
- .
sx + 1d sx + 1d sx + 1d3

In some problems, assigning small values to x such as x = 0, ;1, ;2, to get

equations in A, B, and C provides a fast alternative to other methods.

EXAMPLE 9 Assigning Numerical Values to x

Find A, B, and C in

x2 + 1 A B C
= + + .
sx - 1dsx - 2dsx - 3d x - 1 x - 2 x - 3

Solution Clear fractions to get

x 2 + 1 = Asx - 2dsx - 3d + Bsx - 1dsx - 3d + Csx - 1dsx - 2d.

Copyright © 2005 Pearson Education, Inc., publishing as Pearson Addison-Wesley

4100 AWL/Thomas_ch08p553-641 8/20/04 10:07 AM Page 579

8.3 Integration of Rational Functions by Partial Fractions 579

Then let x = 1, 2, 3 successively to find A, B, and C:

x = 1: s1d2 + 1 = As -1ds -2d + Bs0d + Cs0d
2 = 2A
A = 1
x = 2: s2d2 + 1 = As0d + Bs1ds -1d + Cs0d
5 = -B
B = -5
x = 3: s3d2 + 1 = As0d + Bs0d + Cs2ds1d
10 = 2C
C = 5.
Conclusion:
x2 + 1 1 5 5
= - + .
sx - 1dsx - 2dsx - 3d x - 1 x - 2 x - 3

Copyright © 2005 Pearson Education, Inc., publishing as Pearson Addison-Wesley