C H A P TE R 3 0

Applications of
Integration II: Volume
A solid of revolution is obtained by revolving a region in a plane about a line that does not intersect the
region. The line about which the rotation takes place is called the axis of revolution.
Let f be a continuous function such that f (x) ≥ 0 for a ≤ x ≤ b. Consider the region under the graph of f,
above the x axis, and between x = a and x = b. (See Fig. 30-1.) If is revolved about the x axis, the resulting
solid is a solid of revolution. The generating regions for some familiar solids are shown in Fig. 30-2.

Fig. 30-1

Disk Formula
The volume V of the solid of revolution obtained by revolving the region of Fig. 30-1 about the x axis is
given by

b b
V = π ∫ ( f ( x))2 dx = π ∫ y 2dx (disk formula)
a a

Fig. 30-2

244


   ! "#%$ & ' '(
)%"* +,- * .' / 01
21 )+3
04+,.
CHAPTER 30 Applications of Integration II: Volume 245

See P roblem 9 for a sketch of the proof of this formula.

Similarly, when the axis of rotation is the y axis and the region that is revolved lies between the y axis
and a curve x = g( y) and between y = c and y = d (see Fig. 30-3), then the volume V of the resulting solid of
revolution is given by the formula

d b
V = π ∫ (g( y))2 dy = π ∫ x 2dy (disk formula)
c a

Fig. 30-3

E X A M P L E 3 0 .1 : Consider the solid of revolution obtained by revolving about the x axis the region in the first q uad-
rant bounded by the parabola y2 = 8x and the line x = 2. (See Fig. 30-4.) B y the disk formula, the volume is

2
V = π ∫ y 2 dx = π ∫ 8 x dx = π (4 x 2 )  = π (16 − 0) = 16π
2 2

0 0 0

Fig. 30-4

E X A M P L E 3 0 .2 : Consider the solid of revolution obtained by revolving about the y axis the region bounded by the
parabola y = 4x2 and the lines x = 0 and y = 16. (See Fig. 30-5.) To find its volume, we use the version of the disk formula
in which we integrate along the y axis. Thus,

π π
16

dy = y 2  = (256 − 0) = 32π
16 16 y
V = π ∫ x 2 dy = π ∫
0 0 4 8 0 8
 246 CHAPTER 30 Applications of Integration II: Volume

Fig. 30-5

W ash e r M e th od
Assume that 0 ≤ g(x) ≤ f (x) for a ≤ x ≤ b. Consider the region between x = a and x = b and lying between
y = g(x) and y = f (x). (See Fig. 30-6.) Then the volume V of the solid of revolution obtained by revolving this
region about the x axis is given by the formula

V = π ∫ [ ( f ( x))2 − (g( x))2 ] dx

b
(washer formula)†
a

Fig. 30-6
b
The justification is clear. The desired volume is the difference of two volumes, the volumes π ∫ ( f ( x))2 dx
a
of the solid of revolution generated by revolving about the x axis the region under y = f (x) and the volume
b
π ∫ (g( x))2 dx of the solid of revolution generated by revolving about the x axis the region under y = g(x).
a
A similar formula

V = π ∫ [ ( f ( y))2 − (g( y))2 ] dy

d
(washer formula)
c

holds when the region lies between the two curves x = f ( y) and x = g( y) and between y = c and y = d, and it
is revolved about the y axis. (It is assumed that 0 ≤ g( y) ≤ f ( y) for c ≤ y ≤ d.)
†
The word “ washer” is used because each thin vertical strip of the region being revolved produces a solid that resembles a plumbing
part called a washer (a small cylindrical disk with a hole in the middle).
CHAPTER 30 Applications of Integration II: Volume 247

Fig. 30-7

E X A M P L E 3 0 .3 : Consider the solid of revolution obtained by revolving about the x axis the region bounded by the
curves, y = 4x2, x = 0, and y = 16. (The same region as in Fig. 30-5.) H ere the upper curve is y = 16 and the lower curve
is y = 4x2. H ence, by the washer formula,

0 0
2 2
(
V = π ∫ [ 162 − (4 x 2 )2 ] dx = π ∫ [ 256 − 16 x 4 ] dx = π ( 256 x − 165 x 5 )  0 = π 512 −
2 512
5 )=
2048π
5

C y lin d ric al S h e ll M e th od
Consider the solid of revolution obtained by revolving about the y axis the region in the first q uadrant be-
tween the x axis and the curve y = f (x), and lying between x = a and x = b. (See Fig. 30-7.) Then the volume
of the solid is given by

b b
V = 2π ∫ xf ( x) dx = 2π ∫ xy dx (cylindrical shell formula)
a a

See P roblem 10 for the justification of this formula.

A similar formula holds when the roles of x and y are reversed, that is, the region in the first q uadrant
between the y axis and the curve x = f ( y), and lying between y = c and y = d, is revolved about the x axis

d d
V = 2π ∫ yf ( y) dy = 2π ∫ yx dy
c c

E X A M P L E 3 0 .4 : R evolve about the y axis the region above the x axis and below y = 2x2, and between x = 0 and x = 5. B y the
cylindrical shell formula, the resulting solid has volume

2π ∫ xy dx = 2π ∫ x(2 x 2 ) dx = 4π ∫ x 3dx = π ( x 4 ) ]0 = 625π

5 5 5 5
0 0 0

N ote that the volume could also have been computed by the washer formula, but the calculation would
have been somewhat more complicated.

Diffe re n c e of S h e lls Formula

Assume that 0 ≤ g(x) ≤ f (x) on an interval [a, b] with a ≥ 0. Let be the region in the first q uadrant between
the curves y = f (x) and y = g(x) and between x = a and x = b. Then the volume of the solid of revolution
obtained by revolving about the y axis is given by

b
V = 2π ∫ x( f ( x) − g( x)) dx (difference of shells formula)
a
 248 CHAPTER 30 Applications of Integration II: Volume

This obviously follows from the cylindrical shells formula because the req uired volume is the difference of
two volumes obtained by the cylindrical shells formula. N ote that a similar formula holds when the roles of
x and y are reversed.

E X A M P L E 3 0 .5 : Consider the region in the first q uadrant bounded above by y = x2, below by y = x3, and lying
between x = 0 and x = 1. W hen revolved about the y axis, this region generates a solid of revolution whose volume, ac-
cording to the difference of shells formula, is

( 41 x − 15 x ) = 2π ( 41 − 15 ) = 10π
1
1 1
2π ∫ x( x 2 −x 3 ) dx = 2π ∫ ( x 3 − x 4 ) dx = 2π 4 5
0 0
0

C ross-S e c tion Formula (S lic in g Formula)

Assume that a solid lies entirely between the plane perpendicular to the x axis at x = a and the plane per-
pendicular to the x axis at x = b. For each x such that a ≤ x ≤ b, assume that the plane perpendicular to the x
axis at that value of x intersects the solid in a region of area A(x). (See Fig. 30-8.) Then the volume V of the
solid is given by

b
V = ∫ A( x) dx (cross-section formula)†
a

For justification, see P roblem 11.

Fig. 30-8

E X A M P L E 3 0 .6 : Assume that half of a salami of length h is such that a cross-section perpendicular to the axis of the
salami at a distance x from the end O is a circle of radius x . (See Fig. 30-9.) H ence, the area A(x) of the cross-section
is π ( x )2 = π x. So, the cross-section formula yields

π 2  π h2
h
h h
V = ∫ A( x) dx = ∫ π x dx = x =
0 0 2  0 2

Fig. 30-9

†
This formula is also called the slicing formula because each cross-sectional area A(x) is obtained by slicing through the solid.
CHAPTER 30 Applications of Integration II: Volume 249

S O LV E D P R O B L E M S

1. Find the volume of a cone that has height h and whose base has radius r.
The cone is generated by revolving about the x axis the region between the line y = r x and the x axis,
h
between x = 0 and x = h. (See Fig. 30-2(a).) B y the disk formula, the volume of the cone is

h
π ∫ y 2 dx = π ∫ r 2 x 2 dx = π r2 ( 13 x 3 )  = π r2 ( 13 h 3 ) = 13 π r 2 h
h h 2 2 2

0 0 h h  0 h

2. Find the volume of the cylinder of height h and radius r.

The cylinder is generated by revolving about the x axis the region between the line y = r and the x axis,
between x = 0 and x = h. (See Fig. 30-2(b).) B y the disk formula, the volume of the cylinder is

h
V = π ∫ y 2 dx = π ∫ r 2 dx = π r 2 x  = π r 2 h.
h h

0 0 0

3. Find the volume of a sphere of radius r.

The sphere is generated by revolving about the x axis the region between the semicircle y = r 2 − x 2 and
the x axis, between x = −r and x = r. (See Fig. 30-2(c).) B y the symmetry with respect to the y axis, we can use
the part of the given region between x = 0 and x = r and then double the result. H ence, by the disk formula, the
volume of the sphere is

0 0
r r
( 3 )
V = 2π ∫ y 2 dx = 2π ∫ (r 2 − x 2 ) dx = 2π ( r 2 x − 13 x 3 )  0 = 2π r 3 − r = 2π ( 23 r 3 ) = 43 π r 3
r 3

4. Let be the region between the x axis, the curve y = x3, and the line x = 2. (See Fig. 30-10.)
(a) Find the volume of the solid obtained by revolving about the x axis.
(b) Find the volume of the solid obtained by revolving about the y axis.

Fig. 30-10

(a) The disk formula yields the volume

2
V = π ∫ y 2 dx = π ∫ ( x 3 )2 dx = π ∫ x 6 dx = π x 7  = 128π
2 2 2

0 0 0 7 0 7
 25 0 CHAPTER 30 Applications of Integration II: Volume

(b) (First solution) The cylindrical shells formula yields the volume
2
V = 2π ∫ xy dx = 2π ∫ x( x 3 ) dx = 2π ∫ x 4 dx = 2π ( 15 x 5 )  = 64π
2 2 2

0 0 0 0 5

(Second solution) Integrating along the y axis and using the washer formula yields the volume

( y )  dy = π ∫ [ 4 − y ] dy = π ( 4 y − 35 y ) = π (32 − ( 35 )32) = 645π

8
V = π ∫  22 −
8 2 8 2 5
3 3 3

0  0
0

5. Find the volume of the solid obtained by revolving about the y axis the region in the first q uadrant inside the
circle x2 + y2 = r2, and between y = a and y = r (where 0 < a < r). See Fig. 30-11. (The solid is a “ polar cap” of a
sphere of radius r.)

Fig. 30-11

Integrating along the y axis, the disk formula yields the volume

V = π ∫ x 2 dy = π ∫ (r 2 − y 2 ) dy = π ( r 2 y − 13 y3 )  a = π ( )
r 3 − ( r 2 a − 13 a 3 ) = π (2r 3 − 3r 2 a + a 3 )
r r r
2
3
a a 3

6. Find the volume of the solid obtained by revolving about the y axis the region in the first q uadrant bounded
above by the parabola y = 2 − x2 and below by the parabola y = x2. (See Fig. 30-12.)

Fig. 30-12
CHAPTER 30 Applications of Integration II: Volume 25 1

The curves intersect at (1,1). B y the difference of cylindrical shells formula, the volume is

1
V = 2π ∫ x((2 −x 2 ) − x 2 ) dx = 4π ∫ ( x − x 3 ) dx = 4π ( 12 x 2 − 14 x 4 )  = 4π ( 12 − )=π
1 1
1
0 0 0 4

7. Consider the region bounded by the parabola y = 4x2 and the lines x = 0 and y = 16. (See Fig. 30-5.) Find the
volume of the solid obtained by revolving about the line y = −2.
To solve this problem, we reduce it to the case of a revolution about the x axis. R aise the region vertically
upward through a distance of 2 units. This changes into a region * that is bounded below by the parabola
y = 4x2 + 2, on the left by the y axis, and above by the line y = 18. (See Fig. 30-13.) Then the original solid of
revolution has the same volume as the solid of revolution obtained by revolving R* around the x axis. The latter
volume is obtained by the washer formula:

2 2
V = π ∫ (182 − (4 x 2 + 2)2 ) dx = π ∫ (256 − 16 x 4 − 16 x 2 − 4) dx
0 0

= π ( 252 x − 165 x 5 − 163 x 3 )  0 = π ( 504 − 5384π

3 )=
2
512
5 − 128
15

8. As in P roblem 7, consider the region bounded by the parabola y = 4x2 and the lines x = 0 and y = 16.
(See Fig. 30-5.) Find the volume of the solid obtained by revolving about the line x = −1.

Fig. 30-13

To solve this problem, we reduce it to the case of a revolution about the y axis. M ove the region to the right
through a distance of 1 unit. This changes into a region * that is bounded on the right by the parabola
y = 4(x − 1)2, above by y = 16, and on the left by x = 1. (See Fig. 30-14.) The desired volume is the same as that
obtained when we revolve * about the y axis. The latter volume is got by the difference of cylindrical shells
formula:

3 3
V = 2π ∫ x(16 − 4( x − 1)2 ) dx = 2π ∫ x(16 − 4 x 2 + 8 x − 4) dx
1 1

= 2π ∫ (16 x − 4 x 3 + 8 x 2 − 4 x) dx = 2π (8 x 2 − x 4 + 83 x 3 − 2 x 2 ) 1
3 3

= 2π (72 − 81 + 72 − 18) − (8 − 1 + 83 − 2 )  = 112

3
 25 2 CHAPTER 30 Applications of Integration II: Volume

Fig. 30-14

b
9 . J ustify the disk formula: V = π ∫ ( f ( x))2 dx.
D ivide the interval [ a, b] into n eq ual subintervals, each of length ∆x = b − a . (See Fig. 30-15.)
a

n
Consider the volume Vi obtained by revolving the region i above the ith subinterval about the x axis.
If mi and Mi are the absolute minimum and absolute maximum of f on the ith subinterval, then Vi lies
between the volume of a cylinder of radius mi and height ∆ x and the volume of a cylinder of radius Mi
V
and height ∆ x. Thus, π mi2 ∆x ≤ Vi ≤ π M i2 ∆x and, therefore, mi2 ≤ i ≤ M i2 . (W e have assumed that the
π∆x
volume of a cylinder of radius r and height h is πr2h.) H ence, by the intermediate value theorem for the
V
continuous function ( f (x))2, there exists xi* in the ith subinterval such that i = ( f ( xi* ) ) and, therefore,
2
π ∆ x
Vi = π ( f ( xi* ) ) ∆ x. Thus,
2

n n
V = ∑ Vi = π ∑ ( f ( xi* ) ) ∆ x Letting n → +∞ , we obtain the disk formula.
2

i =1 i =1

Fig. 30-15
CHAPTER 30 Applications of Integration II: Volume 25 3

b
10 . J ustify the cylindrical shells formula: V = 2π ∫ xf ( x) dx.
a
D ivide [a, b] into n eq ual subintervals, each of length ∆ x. (See Fig. 30-16.) Let i be the region above the
x +x
ith subinterval. Let xi* be the midpoint i −1 i of the ith interval. The solid obtained by revolving the region
2
t about the y axis is approximately the solid obtained by revolving the rectangle with base ∆ x and height
yi* = f ( xi* ). The latter solid is a cylindrical shell, that is, it lies between the cylinders obtained by revolving the
rectangles with the same height f ( xi* ) and with bases [0, xi−1] and [0, xi]. H ence, it has volume

π xi2 f ( xi* ) − π xi2−1 f ( xi* ) = π f ( xi* )( xi2 − xi2−1 )

= π f ( xi* )( xi − xi −1 )( xi + xi −1 ) = π f ( xi* )(2 xi* )( ∆ x) = 2π xi* f ( xi* )( ∆ x)

n b
Thus, the total V is approximated by 2π ∑ x1* f ( x1* )∆ x which approaches 2π ∫ x f ( x)dx as n → + ∞ .
a
i =1

b
11. J ustify the cross-section formula: V = ∫ A( x) dx.
a

D ivide [a, b] into n eq ual subintervals [xi−1, xi], and choose a point xi* in [xi−1, xi]. If n is large, ∆ x is small
and the piece of the solid between xi−1 and xi, will be close to a (noncircular) disk of thickness ∆ x and base area
n
A( xi* ). (See Fig. 30-17.) This disk has volume A( xi* ) ∆ x. So V is approximated by ∑ A( xi* ) ∆ x, which approaches
b i =1
∫a
A( x) dx as n → + ∞ .

Fig. 30-16

Fig. 30-17
 25 4 CHAPTER 30 Applications of Integration II: Volume

12 . A solid has a circular base of radius 4 units. Find the volume of the solid if every plane section perpendicular to a
particular fixed diameter is an eq uilateral triangle.
Take the circle as in Fig. 30-18, with the fixed diameter on the x axis. The eq uation of the circle is x2 + y2 = 16.
The cross-section ABC of the solid is an eq uilateral triangle of side 2y and area A(x) = 3 y2 = 3 (16 − x2). Then,
by the cross-section formula,

V = 3 ∫ (16 − x 2 ) dx = 3 (16 x − 13 x 3 )  −4 =
4 4
256
3 3
−4

13 . A solid has a base in the form of an ellipse with major axis 10 and minor axis 8. Find its volume if every section
perpendicular to the major axis is an isosceles triangle with altitude 6.
2
Take the ellipse as in Fig. 30-19, with eq uation x + y = 1. The section ABC is an isosceles triangle of base
2

25 16
(
2y, altitude 6, and area A( x) = 6 y = 6 4 25 − x 2 . H ence,
5 )
5
V= 24
5 ∫
−5
25 − x 2 dx = 60π

Fig. 30-18
5
(N ote that ∫ 25 − x 2 dx is the area of the upper half of the circle x2 + y2 = 25 and, therefore, is eq ual to 25π/2.)
−5

Fig. 30-19

S U P P LE M E N TA R Y P R O B LE M S

14 . Consider the region bounded by the parabola y2 = 8x and the line x = 2. (See Fig. 30-4.)
(a) Find the volume of the solid generated by revolving about the y axis.
(b) Find the volume of the solid generated by revolving about the line x = 2.

Ans. (a) 128π ; (b) 256π

5 15
CHAPTER 30 Applications of Integration II: Volume 25 5

15 . Find the volume of the solid generated by revolving the region between the x axis and the parabola y = 4x − x2
about the line y = 6.

Ans. 1408π
15

16 . Find the volume of the torus (doughnut) generated by revolving the circle (x − a)2 + y2 = b2 about the y axis,
where 0 < b < a.

Ans. 2π2ab2

17 . Consider the region bounded by y = −x2 − 3x + 6 and x + y = 3. Find the volume of the solid generated by
revolving about:
(a) the x axis; (b) the line x = 3.

Ans. (a) 1792π ; (b) 256π

15 3

In P roblems 18– 26, find the volume generated when the given region is revolved about the given line. U se the disk
formula.
18 . The region bounded by y = 2x2, y = 0, x = 0, x = 5, about the x axis.

Αns. 25 0 0 π

19 . The region bounded by x2 − y2 = 16, y = 0, x = 8, about the x axis.

Ans. 256π
3

2 0 . The region bounded by y = 4x2, x = 0, y = 16, about y = 16. (See Fig. 30-5.)

Ans. 4096π
15

2 1. The region bounded by y2 = x3, y = 0, x = 2, about the x axis.

Αns. 4 π

2 2 . The region bounded by y = x3, y = 0, x = 2, about x = 2.

Ans. 16π
5

2 3 . The region within the curve y2 = x4(l − x2), about the x axis.

Ans. 4π
35

2 4 . The region within the ellipse 4x2 + 9y2 = 36, about the x axis.

Ans. 16π

2 5 . The region within the ellipse 4x2 + 9y2 = 36, about the y axis.

Αns. 24 π
 25 6 CHAPTER 30 Applications of Integration II: Volume

2 6 . The region within the parabola x = 9 − y2 and between y = x − 7 and the y axis, about the y axis.

Ans. 963π
5

In P roblems 27– 32, find the volume of the solid generated by revolving the given region about the given line. U se the
washer formula.
2 7 . The region bounded by y = 2x2, y = 0, x = 0, x = 5, about the y axis.

Αns. 6 25 π

2 8 . The region bounded by x2 − y2 = 16, y = 0, x = 8, about the y axis.

Ans. 1 28 3 π

2 9 . The region bounded by y = x3, x = 0, y = 8, about x = 2.

Ans. 144π
5

3 0 . The region bounded by y = x2, y = 4x − x2, about the x axis.

Ans. 32π
3

3 1. The region bounded by y = x2, y = 4x − x2, about y = 6.

Ans. 64π
3

3 2 . The region bounded by x = 9 − y2, y = x − 7, about x = 4.

Αns. 153π
5

In P roblems 33– 37, find the volume of the solid generated by revolving the given region about the given line. U se the
cylindrical shells formula.

3 3 . The region bounded by y = 2x2, y = 0, x = 0, x = 5, about x = 6.

Αns. 3 7 5 π

3 4 . The region bounded by y = x3, y = 0, x = 2, about y = 8.

Αns. 320π
7

3 5 . The region bounded by y = x2, y = 4x − x2, about x = 5.

Ans. 64π
3

3 6 . The region bounded by y = x2 − 5x + 6 and y = 0, about the y axis.

Ans. 5π
6
CHAPTER 30 Applications of Integration II: Volume 25 7

3 7 . The region bounded by x = 9 − y2, y = x − 7, x = 0, about y = 3.

Ans. 369π
2

In P roblems 38– 42, find the volume generated by revolving the given region about the given line. U se any appropriate
method.
3 8 . The region bounded by y = e− x , y = 0, x = 0, x = 1, about the y axis.
2

Αns. π(1 − e−1 )

3 9 . The region bounded by y = 2x2, y = 2x + 4, about x = 2.

Αns. 27 π

4 0 . The region bounded by y = 2x, y = 0, x = 0, x = 1, about the y axis.

Αns. 4π
3

4 1. The region bounded by y = x2, x = y2, about the x axis.

Αns. 3π
10

4 2 . The region bounded by xy = 4, y = (x − 3)2, about the x axis.

Αns. 27π
5

4 3 . Find the volume of the frustum of a cone whose lower base is of radius R, upper base is of radius r, and
altitude is h.

Ans. 1 π h(r 2 + rR + R 2 )
3

4 4 . A solid has a circular base of radius 4 units. Find the volume of the solid if every plane perpendicular to a
fixed diameter (the x axis of Fig. 30-18) is: (a) a semicircle; (b) a sq uare; (c) an isosceles right triangle with the
hypotenuse in the plane of the base.

Ans. (a) 128π ; (b) 1024 ; (c) 256

3 3 3

4 5 . A solid has a base in the form of an ellipse with major axis 10 and minor axis 8. Find its volume if every section
perpendicular to the major axis is an isosceles right triangle with one leg in the plane of the base.

Ans. 640
3

4 6 . The base of a solid is the first-q uadrant region bounded by the line 4x + 5y = 20 and the coordinate axes. Find its
volume if every plane section perpendicular to the x axis is a semicircle.

Ans. 10π
3

4 7 . The base of a solid is the circle x2 + y2 = 16x, and every plane section perpendicular to the x axis is a rectangle
whose height is twice the distance of the plane of the section from the origin. Find its volume.

Αns. 1 0 24 π
 25 8 CHAPTER 30 Applications of Integration II: Volume

4 8 . The section of a certain solid cut by any plane perpendicular to the x axis is a circle with the ends of a diameter
lying on the parabolas y2 = 4x and x2 = 4y. Find its volume.

Ans. 6561π
280

4 9 . The section of a certain solid cut by any plane perpendicular to the x axis is a sq uare with the ends of a diagonal
lying on the parabolas y2 = 4x and x2 = 4y. Find its volume.

Ans. 144
35

5 0 . A hole of radius 1 unit is bored through a sphere of radius 3 units, the axis of the hole being a diameter of the
sphere. Find the volume of the remaining part of the sphere.

Ans. 64π 2
3
 C H A P TE R 3 1

Techniques of Integration I:
Integration by P arts
If u and v are functions, the product rule yields

Dx (uv) = uv′ + vu ′

which can be rewritten in terms of antiderivatives as follows:

uv = ∫ uv′ dx + ∫ vu ′ dx

N ow, ∫ uv′ dx can be written as ∫ u dv, and ∫ vu ′ dx can be written as ∫ v du.† Thus, uv = ∫ u dv + ∫ v du and,
therefore,

∫ u dv = uv − ∫ v du (integration by parts)

The purpose of integration by parts is to replace a “ difficult” integration ∫ u dv by an “ easy” integration

∫ v du.
E X A M P L E 3 1 .1 : Find ∫ x ln x dx .
In order to use the integration by parts formula, we must divide the integrand x ln x dx into two “ parts” u and dv so
that we can easily find v by an integration and also easily find ∫ v du. In this example, let u = ln x and dv = x dx. Then
we can set v = 12 x 2 and note that du = 1 dx . So, the integration by parts formula yields:
x

∫ x ln x dx = ∫ u dv = uv − ∫ v du = (ln x)( 1
2
x( )
x 2 ) − ∫ 12 x 2 1 dx

= 12 x 2 ln x − 1
2 ∫ x dx = 1
2 x 2 ln x − 14 x 2 + C

= 14 x 2 (2 ln x − 1) + C

Integration by parts can be made easier to apply by setting up a rectangle such as the following one for
E xample 1.

u = ln x dv = x dx
1
du = dx v = 12 x 2
x

∫ uv ′ dx = ∫ u dv, where, after the integration on the right, the variable v is replaced by the corresponding function of x. In fact, by
†

( ) ( )
the Chain R ule, Dx ∫ u dv = Dv ∫ u dv ⋅ Dx v = u ⋅ v ′. H ence, ∫ u dv = ∫ uv ′ dx. Similarly, ∫ v du = ∫ vu ′ dx .

25 9
56789 : ;< =>? @@ABCAAABCAA@BCAD?E 8F<GHI!J9 KL%M N: O O(56P%7KQ: GR,BS Q IT5O : IU<G9 GV1692=1G9 PR36VWR,GT
 26 0 CHAPTER 31 Techniques of Integration I: Integration by P arts

In the first row, we place u and dv. In the second row, we place the results of computing du and v. The
desired result of the integration parts formula uv − ∫ v du can be obtained by first multiplying the upper-left
corner u by the lower-right corner v, and then subtracting the integral of the product v du of the two entries
v and du in the second row.

E X A M P L E 3 1 .2 : Find ∫ xe x dx.
Let u = x and dv = ex dx. W e can picture this in the box below.

u = x dv = e x dx
du = dx v = e x

Then, ∫ xe dx = uv − ∫ v du = xe
x x
− ∫ e x dx = xe x − e x + C

= e x ( x − 1) + C

E X A M P L E 3 1 .3 : Find ∫ e x cos x dx.

Let u = ex and dv = cos x dx. Then we get the box

u = ex dv = cos x dx
du = e x dx v = sin x

So, ∫e x
cos x dx = uv − ∫ v du = e x sin x − ∫ e x sin x dx (1)

N ow we have the problem of finding ∫ e x sin x dx, which seems to be just as hard as the original integral ∫ e x cos x dx.
H owever, let us try to find ∫ e x sin x dx by another integration by parts. This time, let u = ex and dv = sin x dx.

u = ex dv = sin x dx
du = e x dx v = − cos x

Then, ∫e x
sin x dx = −e x cos x − ∫ −e x cos x dx

= −e x cos x + ∫ e x cos x dx
Substituting in formula (1) above, we get:

∫e x
(
cos x dx = e x sin x − − e x cos x + ∫ e x cos x dx )
= e x sin x + e x cos x − ∫ e x cos x dx

Adding ∫ e x cos x dx to both sides yields 2 ∫ e x cos x dx = e x sin x + e x cos x. So,

∫e x
cos x dx = 12 (e x sin x + e x cos x)

W e must add an arbitrary constant:

∫e x
cos x dx = 12 (e x sin x + e x cos x) + C

N otice that this example req uired an iterated application of integration by parts.
CHAPTER 31 Techniques of Integration I: Integration by P arts 26 1

S O LV E D P R O B L E M S

Find ∫ x 3e x dx.
2
1.
Let u = x2 and dv = xe x dx. N ote that v can be evaluated by using the substitution w = x2. (W e get
2

v= ∫e dw = 12 ew = 12 e x .)
1 w 2
2

u = x2 dv = xe x dx
2

du = 2 x dx v = 12 e x
2

H ence,
∫x e
3 x2
dx = 12 x 2 e x − ∫ xe x dx
2 2

= 12 x 2 e x − 12 e x + C
2 2

= 12 e x ( x 2 − 1) + C
2

2. Find ∫ ln( x 2 + 2) dx.

Let u = ln (x2 + 2) and dv = dx.

u = ln( x 2 + 2) dv = dx
du = 22 x dx v = x
x +2

x 2 dx
So, ∫ ln( x 2
+ 2) dx = x ln( x 2 + 2) − 2 ∫
x2 + 2

= x ln( x 2 + 2) − 2 ∫ 1 − ( 2
x2 + 2 )
dx

 
= x ln( x 2 + 2) − 2 x + 4 tan −1  x  + C
2  2

 
= x (ln( x 2 + 2) − 2) + 2 2 tan −1  x  + C
 2

3. Find ∫ ln x dx.
Let u = ln x and dv = dx.

u = ln x dv = dx
du = 1 dx v = x
x

So, ∫ ln x dx = x ln x − ∫ 1 dx = x ln x − x + C
= x (ln x − 1) + C

4. Find ∫ x sin x dx.

W e have three choices: (a) u = x sin x, dv = dx; (b) u = sin x, dv = x dx; (c) u = x, dv = sin x dx.
(a) Let u = x sin x, dv = dx. Then du = (sin x + x cos x) dx, v = x, and

∫ x sin x dx = x ⋅ x sin x − ∫ x(sin x + x cos x) dx

The resulting integral is not as simple as the original, and this choice is discarded.
 26 2 CHAPTER 31 Techniques of Integration I: Integration by P arts

(b) Let u = sin x, dv = x dx. Then du = cos x dx, v = 12 x 2, and

∫ x sin x dx = 1
2 x 2 sin x − ∫ 12 x 2 cos x dx

The resulting integral is not as simple as the original, and this choice too is discarded.
(c) Let u = x, dv = sin x dx. Then du = dx, v = − cos x, and

∫ x sin x dx = − x cos x − ∫ − cos x dx = − x cos x + sin x + C

5. Find ∫ x 2 ln x dx. 3
Let u = ln x, dv = x2 dx. Then du = dx , v = x , and
x 3
x3 x 3 dx x 3 x 3 ln x − 1 x 3 + C
∫ x 2 ln x dx = 3 ln x − ∫ 3 x = 3 ln x − 1
3 ∫ x dx =
2
3 9

6. Find ∫ sin −1 x dx.

Let u = sin−1 x, dv = dx.

u = sin −1 x dv = dx
du = 1 dx v = x
1 − x2

x
So, ∫ sin −1
x dx = x sin −1 x − ∫
1 − x2
dx

= x sin −1 x + 1
2 ∫ (1 − x 2 −1/ 2
) (−2 x )dx

= x sin −1 x + 12 (2(1 − x 2 )1/ 2 ) + C (by Quick Formula I)

= x sin −1 x + (1 − x 2 )1/ 2 + C = x sin −1 x + 1 − x 2 + C

7. Find ∫ tan −1 x dx.

Let u = tan−1x, dv = dx.

u = tan −1 x dv = dx
du = 1 dx v = x
1 + x2

x dx = x tan −1 x − 2 x dx
So, ∫ tan −1
x dx = x tan −1 x − ∫
1 + x2
1
2 ∫ 1+ x2

= x tan −1 x − 12 ln(1 + x 2 ) + C (by Quick Formula II)

8. Find ∫ sec3 x dx.

Let u = sec x, dv = sec2x dx.

u = sec x dv = sec 2 x dx
du = sec x tan x dx v = tan x
CHAPTER 31 Techniques of Integration I: Integration by P arts 26 3

Thus, ∫ sec 3
x dx = sec x tan x − ∫ sec x tan 2 x dx

= sec x tan x − ∫ sec x(sec 2 x − 1) dx

= sec x tan x − ∫ sec3 x dx + ∫ sec x dx

= sec x tan x − ∫ sec3 x dx + ln |sec x + tan x |

Then, 2 ∫ sec3 x dx = sec x tan x + ln |sec x + tan x |

H ence, ∫ sec 3
x dx = 12 (sec x tan x + ln |sec x + tan x |) + C

9. Find ∫ x 2 sin x dx.

Let u = x2, dv = sin x dx. Thus, du = 2x dx and v = − cos x. Then

∫x 2
sin x dx = − x 2 cos x − ∫ −2 x cos x dx

= − x 2 cos x + 2 ∫ x cos x dx
N ow apply integration by parts to ∫ x cos x dx, with u = x and dv = cos x dx, getting

∫ x cos x = x sin x − ∫ sin x dx = x sin x + cos x

H ence, ∫x 2
sin x dx = − x 2 cos x + 2( x sin x + cos x) + C

10 . Find ∫ x 3e2 x dx .
Let u = x3, dv = e2x dx. Then du = 3x2 dx, v = 12 e2 x, and

∫x e 3 2x
dx = 12 x 3e 2 x − 3
2 ∫x e
2 2x
dx

For the resulting integral, let u = x2 and dv = e2x dx. Then du = 2x dx, v = 12 e2 x, and

∫x e
3 2x
dx = 12 x 3e2 x − 3
2 ( 1
2 )
x 2 e2 x − ∫ xe2 x dx = 12 x 2 e2 x − 43 x 2 e2 x + 3
2 ∫ xe 2x
dx

For the resulting integral, let u = x and dv = e2x dx. Then du = dx, v = 12 e2 x, and

∫x e
3 2x
dx = 12 x 3e2 x − 43 x 2 e2 x + 3
2 ( 1
2 xe2 x − 1
2 ∫e 2x
)
dx = 12 x 3e2 x − 43 x 2 e2 x + 43 xe2 x − 83 e2 x + C

11. D erive the following reduction formula for ∫ sin m x dx.

x dx = − sin x cos x + m − 1 sin m − 2 x dx

m −1
∫ sin m
m m ∫

Let u = sinm−1 x and dv = sin x dx.

u = sin m −1 x dv = sin x dx
du = (m − 1)sin m − 2 x dx v = − cos x
 26 4 CHAPTER 31 Techniques of Integration I: Integration by P arts

Then ∫ sin m
x dx = − cos x sin m −1 x + ( m − 1) ∫ sin m − 2 x cos2 x dx

= − cos x sin m −1 x + ( m − 1) ∫ sin m − 2 x(1 − sin 2 x) dx

= − cos x sin m −1 x + (m − 1) ∫ sin m − 2 x dx − (m − 1) ∫ sin m x dx

H ence, m ∫ sin m x dx = − cos x sin m −1 x + (m − 1) ∫ sin m − 2 x dx

and division by m yields the req uired formula.

12 . Apply the reduction formula of P roblem 11 to find ∫ sin 2 x dx.

W hen m = 2, we get

sin x cos x
∫ sin 2
x dx = −
2
+ 1
2 ∫ sin 0
x dx

sin x cos x
=−
2
+ 1
2 ∫ 1 dx
sin x cos x x x − sin x cos x
=− + +C= +C
2 2 2

13 . Apply the reduction formula of P roblem 11 to find ∫ sin 3 x dx.

W hen m = 3, we get
sin 2 x cos x
∫ sin 3
x dx = −
3
+ 2
3 ∫ sin x dx
sin 2 x cos x 2
=− − 3 cos x + C
3

= − cos x (2 + sin 2 x) + C
3

S U P P LE M E N TA R Y P R O B LE M S

In P roblems 14 – 21, use integration by parts to verify the specified formulas.

14 . ∫ x cos x dx = x sin x + cos x + C

15 . ∫ x sec 2
3x dx = 13 x tan 3x − 19 ln |sec x | +C

16 . ∫ cos −1
2 x dx = x cos −1 2 x − 1
2 1 − 4x2 + C

17 . ∫ x tan −1
x dx = 12 ( x 2 + 1) tan −1 x − 12 x + C

18 . ∫x e 2 −3 x
dx = − 13 e−3 x ( x 2 + 23 x + 92 ) + C

19 . ∫x 3
sin x dx = − x 3 cos x + 3x 2 sin x + 6 x cos x − 6 sin x + C
CHAPTER 31 Techniques of Integration I: Integration by P arts 26 5

20. ∫ x sin −1
( x 2 ) dx = 12 x 2 sin −1 ( x 2 ) + 1
2 1 − x4 + C

ln x dx = − ln x + 1 + C
2 1. ∫ x2 x

2 2 . Show that ∫ x sin nx dx = − 2π for any positive integer n.

2π

0 n

2 3 . P rove the following reduction formula: ∫ sec n x dx = tan x sec x + n − 2 ∫ sec n− 2 x dx.
n−2

n −1 n −1

2 4 . Apply P roblem 23 to find ∫ sec 4 x dx.

Ans. 1
3 tan x(sec 2 x + 2) + C

2 5 . P rove the reduction formula:


∫ (a
x2 dx = 1  − 2 x 2 n−1 + ∫ 2 dx 2 n−11 
2
+ x 2 )n 2n − 2  (a + x ) (a + x ) 

2 6 . Apply P roblem 25 to find ∫ x2 dx.

(a 2 + x 2 ) 2

Ans. 1 − x
( + 1 tan −1 x + C
2 a2 + x 2 a a )
2 7 . P rove ∫ x n ln x dx = x n+1 [(n + 1)ln x − 1)] + C for n ≠ −1.
(n + 1)2

2 8 . P rove the reduction formula: ∫ x n eax dx = 1 x n eax − n ∫ x n−1eax dx.

a a

2 9 . U se P roblem 28 and E xample 2 to show that: ∫ x 2 e x dx = e x ( x 2 − 2 x + 2) + C .

 C H A P TE R 3 2

Techniques of Integration II:

Trigonometric Integrands and
Trigonometric Substitutions
T rig on ome tric In te g ran d s
1. Let us consider integrals of the form ∫ sin k x cosn x dx , where k and n are nonnegative integers.
T yp e 1 . At least one of sin x and cos x occurs to an odd p ower: Then a substitution for the other func-
tion works.

E X A M P L E 3 2 .1 : ∫ sin 3
x cos 2 x dx .
Let u = cos x. Then du = −sin x dx. H ence,

∫ sin 3
x cos 2 x dx = ∫ sin 2 x cos 2 x sin x dx

= ∫ (1 − cos 2 x)cos 2 x sin x dx

= − ∫ (1 − u 2 )u 2 du = ∫ (u 4 − u 2 ) du

= 15 u 5 − 13 u 3 + C = 15 cos5 x − 13 cos3 x + C

E X A M P L E 3 2 .2 : ∫ sin 4
x cos 7 x dx .
Let u = sin x. Then du = cos x dx, and

∫ sin 4
x cos 7 x dx = ∫ sin 4 x cos6 x cos x dx

= ∫ u 4 (1 − u 2 )3 du = ∫ u 4 (1 − 3u 2 + 3u 4 − u 6 ) du

= ∫ (u 5 − 3u 6 + 3u 8 − u10 ) du

= 16 u 6 − 37 u 7 + 13 u 9 − 1
11 u11 + C

= 16 sin 6 x − 37 sin 7 x + 13 sin 9 x − 1

11 sin11 x + C

E X A M P L E 3 2 .3 : ∫ sin 5
x dx .
Let u = cos x. Then du = −sin x dx and

26 6
56789 : ;< =>? @@ABCAAABCAA@BCAD?E 8F<GHI!J9 KL%M N: O O(56P%7KQ: GR,BS Q IT5O : IU<G9 GV1692=1G9 PR36VWR,GT
CHAPTER 32 Techniques of Integration II 26 7

∫ sin 5
x dx = ∫ sin 4 x sin x dx = ∫ (1 − cos 2 x)2 sin x dx

= − ∫ (1 − u 2 )2 du = − ∫ (1 − 2u 2 + u 4 ) du

= −(u − 23 u 3 + 15 u 5 ) + C

= − 15 cos5 x + 23 cos3 x − cos x + C

T yp e 2. Both p owers of sin x and cos x are even: This always involves a more tedious computation, using
the identities
1 + cos 2 x 1 − cos 2 x
cos 2 x = and sin 2 x =
2 2

E X A M P L E 3 2 .4 :

∫ cos 2
x sin 4 x dx = ∫ (cos 2 x)(sin 2 x)2 dx

( )( ) dx
2
= ∫ 1 + cos 2 x 1 − cos 2 x
2 2

= ∫ ( 1 + cos 2 x ) ( 1 − 2 cos 2 x + cos 2 x ) dx

2

2 4

= 1 ∫ (1(1 − 2 cos 2 x + cos 2 2 x) + (cos 2 x)(1 − 2 cos 2 x + cos 2 2 x)) dx

8

= 1 ∫ (1 − 2 cos 2 x + cos 2 2 x + cos 2 x − 2 cos 2 2 x + cos3 2 x) dx

8

= 1 ∫ (1 − cos 2 x − cos 2 2 x + cos3 2 x) dx

8

=1
8 ( ∫ 1 dx − ∫ cos 2x dx − ∫ cos 2x dx + ∫ cos 2x dx )
2 3

8 (
= 1 x − sin 2 x − ∫ 1 + cos 4 x dx + ∫ (cos 2 x)(1 − sin 2 2 x) dx
2 2 )
8 ( 2 2 ( 4 )
= 1 x − sin 2 x − 1 x + sin 4 x + ∫ cos 2 x dx − 1 ∫ u 2 du
2 ) [ letting u = sin 2 x]

= 1 ( x − sin 2 x − x − sin 4 x + sin 2 x − 1 sin 2 x ) + C

3

8 2 2 8 2 2 3

= 1 ( x − sin 4 x − sin 2 x ) + C
3

8 2 8 6
3
= x − sin 4 x − sin 2 x + C
16 64 48

2 . Let us consider integrals of the form ∫ tan k x sec n x dx, where k and n are nonnegative integers. R ecall
that sec2 x = 1 + tan2 x.
T yp e 1. n is even: Substitute u = tan x.

E X A M P L E 3 2 .5 : ∫ tan 2
x sec 4 x dx
Let u = tan x, du = sec2 x dx. So,

∫ tan 2
x sec 4 x dx = ∫ tan 2 x(1 + tan 2 x)sec 2 x dx = ∫ u 2 (1 + u 2 ) du

= ∫ (u 4 + u 2 ) du = 15 u 5 + 13 u 3 + C = 15 tan 5 x + 13 tan 3 x + C
 26 8 CHAPTER 32 Techniques of Integration II

T yp e 2. n is odd and k is odd: Substitute u = sec x.

E X A M P L E 3 2 .6 : ∫ tan 3
x sec x dx
Let u = sec x, du = sec x tan x dx. So,

∫ tan 3
x sec x dx = ∫ tan 2 x sec x tan x dx = ∫ (sec 2 x − 1)sec x tan x dx

= ∫ (u 2 − 1) du = 13 u 3 − u + C = 13 sec3 x − sec x + C

T yp e 3. n is odd and k is even: This case usually req uires a tedious calculation.

E X A M P L E 3 2 .7 :

∫ tan 2
x sec x dx = ∫ (sec 2 x − 1)sec x dx = ∫ (sec3 x − sec x) dx

1
= (sec x tan x + ln |sec x + tan x |) − ln |sec x + tan x | + C (by P roblem 8 of Chapter 31)
2
= 12 (sec x tan x − ln |sec x + tan x |) + C

3 . Let us consider integrals of the form ∫ sin Ax cos Bx dx , ∫ sin Ax sin Bx dx, and ∫ cos Ax cos Bx dx.
W e shall need the identities

sin Ax cos Bx = 12 (sin( A + B) x + sin( A − B) x)

sin Ax sin Bx = 12 (cos( A − B) x − cos( A + B) x)

cos Ax cos Bx = 12 (cos(( A − B) x + cos( A + B) x)

E X A M P L E 3 2 .8 :

∫ sin 7 x cos 3x dx = ∫ 1
2 (sin(7 + 3) x + sin(7 − 3) x) dx = ∫ 12 (sin 10 x + sin 4 x) dx

= 12 (− 101 cos10 x − 14 cos 4 x) + C = − 401 (2 cos10 x + 5 cos 4 x) + C

E X A M P L E 3 2 .9 :

∫ sin 7 x sin 3x dx = ∫ 1
2 (cos(7 − 3) x − cos(7 + 3) x) dx = ∫ 12 (cos 4 x − cos10 x) dx

= 12 ( 14 sin 4 x − 1
10 sin 10 x) + C = 1
40 (5 sin 4 x − 2 sin 10 x) + C

E X A M P L E 3 2 .1 0 :

∫ cos 7 x cos 3x dx = ∫ 1
2 (cos(7 − 3) x + cos(7 + 3) x) dx = ∫ 12 (cos 4 x + cos10 x) dx

= 12 ( 14 sin 4 x + 1
10 sin 10 x) + C = 1
40 (5 sin 4 x + 2 sin 10 x) + C

T rig on ome tric S ub stitution s

There are three principal kinds of trigonometric substitutions. W e shall introduce each one by means of a
typical example.
dx
E X A M P L E 3 2 .1 1 : Find ∫x 2
4 + x2
.

Let x = 2 tan q, that is, q = tan−1 (x/2). Then

dx = 2 sec 2 θ dθ and 4 + x 2 = 4 + 4 tan 2 θ = 2 1 + tan 2 θ = 2 sec 2 θ = 2 |sec θ |
CHAPTER 32 Techniques of Integration II 26 9

B y definition of the inverse tangent, −π/2 < q < π/2. So, cos q > 0 and, therefore, sec q > 0. Thus,
sec θ = |sec θ | = 4 + x 2 / 2 . H ence,

dx 2 sec 2 θ dθ
∫z 2
4 + x2
= ∫ 4 tan 2 θ (2 sec θ )

sec θ dθ 1 cosθ dθ 1
=1∫ = ∫ = ∫ (sin θ )−2 cosθ dθ
4 tan 2 θ 4 sin 2 θ 4

= 14 (−(sin θ )−1 ) + C = − 1 +C
4 sin θ

N ow we must evaluate sin q.

Analytic method: sin θ = tan θ = x/2 = x .

sec θ 4 + x2 /2 4 + x2
G eometric method: D raw the right triangle shown in Fig. 32-1. From this triangle we see that sin θ = x / 4 + x 2 .
(N ote that it follows also for q < 0.)

dx 4 + x2
H ence, ∫x 2
4 + x2
=−
4x
+C

Fig. 32 -1

This example illustrates the following general rule:

S tr a te g y I . If a 2 + x 2 occurs in an integrand, try the substitution x = a tan q.

dx
E X A M P L E 3 2 .1 2 : Find
9 − x2
. ∫x 2

Let x = 3 sin q, that is, q = sin−1 (x/3). Then dx = 3 cos q dq and

9 − x 2 = 9 − 9 sin 2 θ = 3 sin 2 θ = 3 cos 2 θ = 3 |cosθ |

B y definition of the inverse sine, −π/2 < q < π/2 and, therefore, cos q > 0. Thus, cosθ = |cosθ | = 9 − x 2 / 3 . N ow,

dx 3 cos θ dθ =1
∫x 2
9 − x 2 ∫ 9 sin θ (3 cos θ ) 9
= 2 ∫ csc 2
θ dθ

9 − x2 /3 9 − x2
= − 1 cot θ + C = − 1 cos θ + C = − 1 +C= − 1 +C
9 9 sin θ 9 x /3 9 x

This example illustrates the following general method:

S tr a te g y I I . If a 2 − x 2 occurs in an integrand, try the substitution x = a sin q.

x2
E X A M P L E 3 2 .1 3 : Find ∫ x2 − 4
dx .
 27 0 CHAPTER 32 Techniques of Integration II

Let x = 2 sec q, that is, q = sec−1 (x/2). Then dx = 2 sec q dq and

x 2 − 4 = 4 sec 2 θ − 4 = 2 sec 2 θ − 1 = 2 tan 2 θ = 2 |tan θ |

B y definition of the inverse secant, q is in the first or third q uadrant and, therefore, tan q > 0. So,
tan θ = |tan θ | = x 2 − 4 / 2 . N ow,

x2 4 sec 2 θ (2 sec θ tan θ )

∫ x −4
2
dx = ∫
2 tan θ
dθ

= 4 ∫ sec3 θ dθ = 2(sec θ tan θ + ln |sec θ + tan θ | + C ) ( by P roblem 8 off Chapter 31)

 x2 − 4 x2 − 4 
= 2 x + ln x +  +C
2 2 2 2 

x x2 − 4 x + x2 − 4
= + 2 ln +C
2 2

x x2 − 4
= + 2 ln x + x 2 − 4 + K where K = C − 2 ln 2
2

This example illustrates the following general method:

S tr a te g y I I I . If x 2 − a 2 occurs in an integrand, try the substitution x = a sec q.

S O LV E D P R O B L E M S

In P roblems 1– 23, verify the given solutions. R ecall the identities

sin 2 u = 12 (1 − cos 2u ) cos 2 u 12 (1 + cos 2u ) sin 2 x = 2 sin x cos x

1. ∫ sin 2
x dx = ∫ 12 (1 − cos 2 x) dx = 12 ( x − 12 sin 2 x) + C = 12 ( x − sin x cos x) + C .

2. ∫ cos (3x) dx = ∫
2 1
2 (1 + cos 6 x) dx = 12 ( x + 16 sin 6 x) + C .

3. ∫ sin 3
x dx = ∫ sin 2 x sin x dx = ∫ (1 − cos 2 x)sin x dx

= ∫ sin x dx + ∫ cos 2 x(− sin x) dx

= − cos x + 13 cos3 x + C (by Quickk Formula I)

4. ∫ sin 2
x cos3 x dx = ∫ sin 2 x cos 2 x cos x dx

= ∫ sin 2 x(1 − sin 2 x)cos x dx

= ∫ sin 2 x cos x dx − ∫ sin 4 x cos x dx

= 13 sin 3 x − 15 sin 5 x + C (by Quick Formula I)

CHAPTER 32 Techniques of Integration II 27 1

5. ∫ sin (3x)cos (3x) dx = ∫ (1 − cos

3 5 2
(3x))cos5 (3x)sin (3x) dx

= ∫ cos5 (3x)sin(3x) dx − ∫ cos 7 (3 x)sin 3x dx

= − 1 ∫ cos5 (3x)(−3 sin(3x)) dx + 1 ∫ cos 7 (3x)(−3 sin(3x)) dx

3 3
= − 13 16 cos6 (3x) + 1 1
3 8 cos8 (3x) + C (by Quick Formula I)

= 1
72 (3 cos (3x) − 4 cos (3x)) + C
8 6

6. ∫ cos 3
( 3x ) dx = ∫ (1 − sin ( 3x )) cos 3x dx
2

= ∫ (1 − sin ( x )) cos x dx = ∫ cos x dx − ∫ sin ( x ) cos x dx

2 2
3 3 3 3 3

= 3 sin x − 3 ∫ sin ( x ) ( 1 cos x ) dx

2
3 3 3 3

= 3 sin x − 3 sin ( x ) + C (by Quick Formula I)

1
3
3
3 3

= 3 sin x − sin ( x ) + C 3
3 3

7. ∫ sin 4
x dx = ∫ (sin 2 x)2 dx = 1 ∫ (1 − cos(2 x))2 dx
4

= 1 ∫ 1 dx − 1 ∫ cos(2 x) dx + 1 ∫ cos 2 (2 x)dx

4 2 4

= 1 x − 1 sin(2 x) + 1 ∫ (1 + cos 4 x)) dx

4 4 8
= 14 x − 14 sin (2 x) + 81 ( x + 14 sin (4 x)) + C

= 83 x − 14 sin (2 x) + 1
32 sin (4 x) + C

8. ∫ sin x cos
2 2
x dx = 1 ∫ sin 2 (2 x) dx = 1 ∫ (1 − cos(4 x)) dx
4 8
= 81 ( x − 14 sin(4 x)) + C = 81 x − 1
32 sin(4 x) + C

9. ∫ sin 4
(3x)cos 2 (3x) dx = ∫ (sin 2 (3x)cos 2 (3x))sin 2 (3x) dx

= 1 ∫ sin 2 (6 x)(1 − cos(6 x)) dx

8

= 1 ∫ sin 2 (6 x) dx − 1 ∫ sin 2 (6 x)cos(6 x) dx

8 8

= 1 ∫ (1 − cos(12 x)) dx − 1 ∫ sin 2 (6 x)(6 cos(6 x)) dx

16 48
= 1
16 (x − 1
12 sin(12 x)) − 1
144 sin 3 (6 x) + C ( by Quick Formula I)

= 1
16 x− 1
192 sin(12 x)) − 1
144 sin 3 (6 x) + C

10 . ∫ sin 3x sin 2 x dx = ∫ 1
2 (cos(3x − 2 x) − cos(3x + 2 x)) dx

= 1
2 ∫ (cos x − cos 5x)) dx = 1
2 (sin x − 15 sin 5x) + C

= 12 sin x − 1
10 sin 5x + C
 27 2 CHAPTER 32 Techniques of Integration II

11. ∫ sin 3x cos 5x dx = ∫ 1

2 (sin(3x − 5x) + sin(3x + 5x)) dx

= 1 ∫ (sin(−2 x) + sin(8 x)) dx = 1 ∫ (− sin(2 x) + sin(8 x))) dx

2 2
= 12 ( 12 cos(2 x) − 81 cos(8 x)) + C = 14 cos(2 x) − 1
16 cos(8 x) + C

1
12 . ∫ cos 4 x cos 2 x dx = 2 ∫ (cos(2 x) + cos(6 x)) dx
= 12 ( 12 sin(2 x) + 16 sin(6 x)) + C = 14 sin(2 x) + 1
12 sin(6 x) + C

13 . ∫ 1 − cos x dx = 2 ∫ sin x dx
2 ( ) ( by sin ( 2x ) = 1 − cos
2
2
x
)
= 2 ( −2 cos ( x )) + C = −2 2 cos ( x ) + C
2 2

14 . ∫ (1 + cos 3x) 3/ 2
dx = 2 2 ∫ cos3 ( 32x ) dx since cos ( 32x ) = 1 + cos(2 3x)
2

= 2 2 ∫  1 − sin ( ) cos ( ) dx

 2
3x  3x
2 2

= 2 2  ∫ cos ( ) dx − ∫ sin ( ) cos ( ) dx 

 3x 3x
2
3x 
 2 2 2 

= 2 2  sin ( ) − ∫ sin ( )  cos ( ) dx 

2 3x 2 2
3x  3 3x  
3 2 3 2 2 2 

= 2 2  sin ( ) − sin ( )  + C
2 3x 21 3
3x 
3 2 33 2 

3 sin ( ) − sin ( )  + C
4 2 3x 3x 
= 3
9  2 2 

dx dx
15 . ∫ 1 − sin 2 x ∫
=
(
1 − cos π − 2 x
2 )
2 dx

 sin
( ) 2 π
1 − cos π − 2 x
2 ( ) 
= ∫ π −x =
( )
nce sin
2  4 2 
sin −x  
4

csc ( π − x ) dx = − ln csc ( π − x ) − cot ( π − x ) + C

2 2
2 ∫
=
4 2 4 4

16 . ∫ tan 4
x dx = ∫ tan 2 x tan 2 x dx = ∫ tan 2 x(sec 2 x − 1) dx

= ∫ tan 2 x sec 2 x dx − ∫ tan 2 x dx

= 13 tan 3 x − ∫ (sec 2 x − 1)dx ( by Quick Formula I)

= 13 tan 3 x − (tan x − x) + C

= 13 tan 3 x − tan x + x + C
CHAPTER 32 Techniques of Integration II 27 3

17 . ∫ tan 5
x dx = ∫ tan 3 x tan 2 x dx = ∫ tan 3 x(sec 2 x − 1) dx

= ∫ tan 3 x sec 2 x dx − ∫ tan 3 dx

= 14 tan 4 x − ∫ tan x(sec 2 x − 1) dx ( by Quick Formula I)

= 14 tan 4 x − ∫ tan x sec 2 x dx + ∫ tan x dx

= 14 tan 4 x − 12 tan 2 x + ln |sec x | +C ( by Quick Foormula I)

18 . ∫ sec 4
(2 x) dx = ∫ sec 2 (2 x)sec 2 (2 x) dx

= ∫ sec 2 (2 x)(1 + tan 2 (2 x)) dx

= ∫ sec 2 (2 x) dx + ∫ sec 2 (2 x) tan 2 (2 x)) dx

= 12 tan(2 x) + 1
2 ∫ tan 2
(2 x)(2 sec 2 (2 x)) dx

= 12 taan(2 x) + 1 1
2 3 tan 3 (2 x) + C (by Quick Formula I )

= 12 tan(2 x) + 16 tan 3 (2 x) + C

19 .
∫ tan (3x)sec
3 4
(3x) dx = ∫ tan 3 (3x)(1 + tan 2 (3x))sec 2 (3x) dx

= ∫ tan 3 (3x)sec 2 (3x) dx + ∫ tan 5 (3x)sec 2 (3x) dx

= 11
34 tan 4 (3x) + 11
36 tan 6 (3x) + C
= 1
12 tan 4 (3x) + 1
18 tan 6 (3x) + C

20. ∫ cot (2 x) dx = ∫ cot(2 x)(csc (2 x) − 1) dx

3 2

= − 14 cot 2 (2 x) + 12 ln |csc(2 x)| + C

2 1. ∫ cot 4
(3x) dx = ∫ cot 3 (3x)(csc 2 (3x) − 1) dx

= ∫ cot 2 (3x) csc 2 (3x) dx − ∫ cot 2 (3x) dx

= − 19 cot 3 (3x) − ∫ (csc 2 (3x) − 1) dx

= − 19 cot 3 (3x) + 13 cot(3x) + x + C

22.
∫ csc 6
x dx = ∫ csc 2 x(1 + cot 2 x)2 dx

= ∫ csc 2 x dx + 2 ∫ cot 2 x csc 2 x dx + ∫ cot 4 x csc 2 x dx

= − cot x − 23 cot 3 x − 15 cot 5 x + C

23.
∫ cot 3
x csc5 x dx = ∫ cot 2 x csc 4 csc x cot x dx

= ∫ (csc 2 x − 1) csc 4 x csc x cot x dx

= ∫ csc6 x csc x cot x dx − ∫ csc 4 x csc x cot x dx

= − 17 csc 7 x + 15 csc5 x + C
 27 4 CHAPTER 32 Techniques of Integration II

9 − 4x2
2 4 . Find ∫ dx .
x

9 − 4x2 = 2 9
4 − x 2 . So, let x = 32 sin θ . Then

dx = 32 cosθ dθ and 9 − 4 x 2 = 9 − 9 sin 2 θ = 3 cos 2 θ = 3 |cosθ | = 3 cosθ

H ence,
9 − 4x2 3 cosθ ( 32 cosθ ) dθ
= 3 ∫ cos θ dθ = 3 ∫ 1 − sin θ dθ
2 2
∫ x
dx = ∫ 3
2 sin θ sin θ sin θ

= 3 ∫ (cscθ − sin θ ) dθ = 3 ln|csccθ − cot θ | +3 cosθ + C

B ut

9 − 4 x 2 /3 9 − 4x2
csc θ = 1 = 3 and cot θ = cosθ = =
sin θ 2x sin θ 2x /3 2x

So,

9 − 4x2 3 − 9 − 4x2
∫ x
dx = 3 ln
x
+ 9 − 4x2 + K where K = C − 3 ln 2

dx
2 5 . Find ∫x 9 − 4x2
.

Let x = 32 tan θ . (See Fig. 32-2.) Then dx = 32 sec 2 θ and 9 − 4 x 2 = 3 sec θ . H ence,

dx
3
sec 2 θ dθ
∫x 9 + 4x 2
=∫ 3 2
( 2 tan θ )(3 sec θ )

= 1 ∫ csc θ dθ = 13 ln|csc θ − cot θ | + C

3

9 + 4x2 − 3
= 13 ln +K
x

Fig. 32 -2

(16 − 9 x 2 )3/ 2
2 6 . Find ∫ x6
dx .

Let x = 43 sin θ . (See Fig. 32-3.) Then dx = 43 cosθ dθ and 16 − 9 x 2 = 4 cosθ . H ence,

(16 − 9 x 2 )3/ 2 (64 cos3 θ )( 43 cosθ dθ )

∫ x 6 dx = ∫
4096 sin 6 θ
729

= 243 ∫ cot 4 θ csc 2 θ dθ = − 243 cot 5 θ + C

16 80
(16 − 9 x 2 )5/ 2 (16 − 9 x 2 )5/ 2
= − 243 +C=− 1 +C
80 243x 5 80 x5
CHAPTER 32 Techniques of Integration II 27 5

Fig. 32 -3

x 2 dx = x 2 dx
2 7 . Find ∫ 2x − x 2 ∫ 1 − ( x − 1)2
.

Let x −1 = sin q. (See Fig. 32-4.) Then dx = cos q dq and 2 x − x 2 = cosθ . H ence,

x 2 dx = (1 + sin θ )2 cosθ dθ
∫ 2 x − x 2 ∫ cosθ

= ∫ (1 + sin θ )2 dθ = ∫ ( 32 + 2 sin θ − 12 cos 2θ ) dθ

= 32 θ − 2 cosθ − 14 sin 2θ + C

= 32 sin −1 ( x − 1) − 2 2 x − x 2 − 12 ( x − 1) 2 x − x 2 + C

= 32 sin −1 ( x − 1) − 12 ( x + 3) 2 x − x 2 + C

Fig. 32 -4

2 8 . Find ∫ dx dx
(4 x 2 − 24 x + 27)3/ 2 ∫ (4( x − 3)2 − 9)3/ 2
= .

Let x − 3 = 32 sec θ . (See Fig. 32-5.) Then dx = 32 sec θ tan θ dθ and 4 x 2 − 24 x + 27 = 3 tan θ . So,

2 sec θ tan θ dθ
3

∫ (4 x
dx
3/ 2 = ∫ = 1 ∫ csc θ cot θ dθ
2
− 24 x + 27) 27 tan 3 θ 18

= − 1 csc θ + C = − 1 x−3 +C (ffrom Fig. 32-5)

18 9 4 x 2 − 24 x + 27

Fig. 32 -5

S U P P LE M E N TA R Y P R O B LE M S

29. ∫ cos 2
x dx = 12 x + 14 sin 2 x + C

30. ∫ sin 3
2 x dx = 16 cos3 2 x − 12 cos 2 x + C
 27 6 CHAPTER 32 Techniques of Integration II

3 1. ∫ sin 4
2 x dx = 83 x − 81 sin 4 x + 1
64 sin 8 x + C

32. ∫ cos 4 1
2 x dx = 83 x + 12 sin x + 1
16 sin 2 x + C

33. ∫ sin 7
x dx = 17 cos 7 x − 35 cos5 x + cos3 x − cos x + C

34. ∫ cos 6 1
2 x dx = 5
16 x + 12 sin x + 3
32 sin 2 x − 1
24 sin 3 x + C

35. ∫ sin 2
x cos5 x dx = 13 sin 3 x − 25 sin 5 x + 17 sin 7 x + C

36. ∫ sin 3
x cos 2 x dx = 15 cos5 x − 13 cos3 x + C

37. ∫ sin 3
x cos3 x dx = 1
48 cos3 2 x − 1
16 cos 2 x + C

38. ∫ sin 4
x cos 4 x dx = 1
128 (3x − sin 4 x + 81 sin 8 x + C

39. ∫ sin 2 x cos 4 x dx = 1

4 cos 2 x − 1
12 cos 6 x + C

40. ∫ cos 3x cos 2 x dx = 1

2 sin x + 1
10 sin 5x + C

4 1. ∫ sin 5x sin x dx = 1
8 sin 4 x − 1
12 sin 6 x + C

cos3 x dx
42. ∫ 1 − sin x = sin x + 12 sin 2 x + C

cos 2 / 3 x dx = − 3 cot 5/ 3 x + C
43. ∫ sin8/3
x 5

cos3 x dx = csc x − 1 csc3 x + C

44. ∫ sin4
x 3

45. ∫ x(cos 3
x 2 − sin 3 x 2 ) dx = 1
12 (sin x 2 + cos x 2 )(4 + sin 2 x 2 ) + C

46. ∫ tan 3
x dx = 12 tan 2 x + ln|cos x | + C

47. ∫ tan 3
3x sec 3x dx = 19 sec3 3x − 13 sec 3x + C

48. ∫ tan 3/ 2
x sec 4 x dx = 25 tan 5/ 2 x + 92 tan 9 / 2 x + C

49. ∫ tan 4
x sec 4 x dx = 17 tan 7 x + 15 tan 5 x + C
CHAPTER 32 Techniques of Integration II 27 7

50. ∫ cot 3
x dx = − 12 cot 2 x − ln|sin x | + C

5 1. ∫ cot 3
x csc 4 x dx = − 14 cot 4 x − 16 cot 6 x + C

52. ∫ cot 3
x csc3 x dx = − 15 csc5 x + 13 csc3 x + C

53. ∫ csc 4
2 x dx = − 12 cot 2 x − 16 cot 3 2 x + C

∫ ( tan x ) dx = − 3 tan x − tan x + C

4
54. sec x 1 1
3

cot 3 x dx = − sin x − csc x + C

55. ∫ csc x

56. ∫ tan x sec x dx = 2 sec x + C

dx x
57. ∫ (4 − x ) 2 3/ 2 =
4 4 − x2
+C

25 − x 2 5 − 25 − x 2
58. ∫ x
dx = 5 ln
x
+ 25 − x 2 + C

= − a 2− x + C
2 2
dx
59. ∫x 2
a −x
2 2 a x

60. ∫ x 2 + 4 dx = 12 x x 2 + 4 + 2 ln( x + x 2 + 4 ) + C

6 1. ∫ (a 2
x 2 dx
− x 2 )3 / 2
= x
a2 − x2 a()
− sin −1 x + C

62. ∫ x 2 − 4 dx = 12 x x 2 − 4 − 2 ln x + x 2 − 4 + C

x2 + a2 a2 + x2 − a
63. ∫ dx = x 2 + a 2 + a ln 2 +C
x 2 a + x2 + a

x 2 dx x3
64. ∫ (4 − x ) 2 5/ 2 =
12(4 − x 2 )3/ 2
+C

dx x
65. ∫ (a 2 =
+ x 2 )3 / 2 a 2 a 2 + x 2
+C

dx 9 − x2
66. ∫x 2
9− x 2
=−
9x
+C

x 2 dx = 1 x x 2 − 16 + 8 ln x + x 2 − 16 + C
67. ∫ x 2 − 16 2
 27 8 CHAPTER 32 Techniques of Integration II

2
68. ∫x 3
a 2 − x 2 dx = 15 (a 2 − x 2 )5/ 2 − a (a 2 − x 2 )3/ 2 + C
3

dx
69. ∫ x 2 − 4 x + 13
= ln( x − 2 + x 2 − 4 x + 13 ) + C

dx x−2 +C
70. ∫ (4 x − x ) 2 3/ 2 =
4 4x − x2

7 1. ∫ (9 + x )
dx
2 2 = 1
54 ()
tan −1 x + x
3 18(9 + x 2 )
+C

In P roblems 72 and 73, first apply integration by parts.

72. ∫ x sin −1
x dx = 14 (2 x 2 − 1)sin −1 x + 14 x 1 − x 2 + C

73. ∫ x cos −1
x dx = 14 (2 x 2 − 1)cos −1 x − 14 x 1 − x 2 + C
 C H A P TE R 3 3

Techniques of Integration III:

Integration by P artial Fractions
N ( x)
W e shall give a general method for finding antiderivatives of the form ∫ dx, where N(x) and D(x) are
D( x)
N ( x)
polynomials. A function of the form is called a rational function. (N(x) is the numerator and D(x) is
D( x)
the denominator.) As examples, consider

x −1 x3 − x
∫x 3
+8
dx and ∫ x+2
dx

Two restrictions will be assumed, neither of which limits the applicability of our method: (i) the leading
coefficient (the coefficient of the highest power of x) in D(x) is +1; (ii) N(x) is of lower degree than D(x). A
q uotient N(x)/D(x) that satisfies (ii) is called a p rop er rational function. Let us see that the restrictions (i)–(ii)
are not essential.
N ( x) 2x3
E X A M P L E 3 3 .1 : Consider the case where is . H ere, our first restriction is not satisfied. H owever,
D( x) 5 x 8 + 3 x − 4
note that

2x3 dx = 1 ∫ 8 23x
3
∫ 5x 8
+ 3x − 4 5 x + 5x− 4
5
dx

The integral on the right side satisfies restrictions (i) and (ii).

is 2 x2 + 7 . H ere, our second restriction is not satisfied. B ut we can

N ( x) 5
E X A M P L E 3 3 .2 : Consider the case where
D( x) x +3
divide N(x) by D(x):

2 x 5 + 7 = 2 x 3 − 6 x + 18 x + 7
x2 + 3 x2 + 3

2 x 5 + 7 dx = 1 x 4 − 3x 2 + 18 x + 7 dx
H ence, ∫ x2 + 3 2 ∫ x2 + 3

and the problem is reduced to evaluating ∫ 182x + 7 dx, which satisfies our restrictions.
x +3

A polynomial is said to be irreducible if it is not the product of two polynomials of lower degree.
Any linear polynomial f (x) = ax + b is automatically irreducible, since polynomials of lower degree than
f (x) are constants and f (x) is not the product of two constants.
N ow consider any q uadratic polynomial g(x) = ax2 + bx + c. Then

g(x) is irreducible if and only if b2 − 4ac < 0

27 9
XYZ[\ ] ^_ `ab ccdefdddefddcefdgbh [i_jkl!m\ no%p q] r r(XYs%Znt] ju,ev t lwXr ] lx_j\ jy1Y\2`1j\ su3Yyzu,jw
 28 0 CHAPTER 33 Techniques of Integration III

To see why this is so, assume that g(x) is reducible. Then g(x) = (Ax + B)(Cx + D). H ence, x = −B/A and
x = −D/C are roots of g(x). The q uadratic formula

x = −b ± b 2 − 4 ac
2a

should yield these roots. Therefore, b2 − 4ac cannot be negative. Conversely, assume b2 − 4ac ≥ 0. Then the
q uadratic formula yields two roots of g(x). B ut, if r is a root of g(x), then g(x) is divisible by x − r.† H ence,
g(x) is reducible.

E X A M P L E 3 3 .3 :
(a) x2 + 4 is irreducible, since b2 − 4ac = 0 − 4(1)(4) = −16 < 0.
(b) x2 + x − 4 is reducible, since b2 − 4ac = 1 − 4(l)(−4) = 17 ≥ 0.

W e will assume without proof the following fairly deep property of polynomials with real coefficients.

T H E O R E M 3 3 .1 : Any polynomial D(x) with leading coefficient 1 can be expressed as a product of linear factors of
the form x − a and of irreducible q uadratic factors of the form x2 + bx + c. (R epetition of factors is permitted.)

EXA M P L E 3 3 .4 :
(a) x3 − 4x = x(x2 − 4) = x(x − 2)(x + 2)
(b) x3 + 4x = x(x2 + 4) (x2 + 4 is irreducible.)
(c) x − 9 = (x − 3)(x + 3) = (x − 3 )(x + 3 )(x2 + 3)
4 2 2
(x2 + 3 is irreducible.)
(d) x3 − 3x2 − x + 3 = (x + l)(x − 2)2

M e th od of P artial Frac tion s

N ( x) N ( x)
Assume that we wish to evaluate ∫ D( x) dx , where
D( x)
is a proper rational function and D(x) has leading
coefficient 1. First, write D(x) as a product of linear and irreducible q uadratic factors.
O ur method will depend on this factoriz ation. W e will consider various cases and, in each case, we will
first explain the method by means of an example and then state the general procedure.

C ase I
D(x) is a product of distinct linear factors.

E X A M P L E 3 3 .5 : Find ∫ 2dx .
x −4

In this case, D(x) = x2 − 4 = (x − 2)(x + 2). W rite

1 = A + B
( x − 2)( x + 2) x − 2 x + 2

It is assumed that A and B are certain constants, that we must now evaluate. Clear the denominators by multiplying
both sides by (x − 2)(x + 2):

1 = A(x + 2) + B(x − 2) (1)

First, substitute −2 for x in (1): 1 = A(0) + B(−4) = −4B. Thus, B = − 14 .

Second, substitute 2 for x in (1): 1 = A(4) + B(0) = 4A. Thus, A = 14 . H ence,

1 =1 1 −1 1
( x − 2)( x + 2) 4 x − 2 4 x + 2

†
In general, if a polynomial h(x) has r as a root, then h(x) must be divisible by x − r.
CHAPTER 33 Techniques of Integration III 28 1

So, ∫x 2 (
−4 ∫ 4 x−2 4 x+2 4 )
dx = 1 1 − 1 1 dx = 1 ln | x − 2 | − 1 ln | x + 2 | + C
4

= 14 (ln | x − 2 | − ln | x + 2 |) + C

= 14 ln x − 2 + C
x+2

( x + 1) dx
E X A M P L E 3 3 .6 : Find ∫ .
x3 + x 2 − 6x
Factoring the denominator yields x(x2 + x − 6) = x(x − 2)(x + 3). The integrand is x +1 .
x ( x − 2)( x + 3)
R epresent it in the following form:

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

Clear the denominators by multiplying by x(x − 2)(x + 3):

x + 1 = A(x − 2)(x + 3) + Bx(x + 3) + Cx(x − 2) (2)

Let x be 0 in (2): 1 = A(−2)(3) +B(0)(3) + C(0)(−2) = − 6A. So, A = − 16 .

Let x be 2 in (2): 3 = A(0)(5) + B(2)(5) + C(2)(0) = 10B. So, B = 3
10 .
Let x be −3 in (2): −2 = A(−5)(0) + B(−3)(0) + C(−3)(−5) = 15C. So, C = − 152 .

H ence, ∫x
( x + 1) dx
3 (
= − 1 1 + 3 1 − 2 1 dx
+ x 2 − 6 x ∫ 6 x 10 x + 2 15 x + 3 )
= − 16 ln | x | + 103 ln | x + 2 | − 152 ln | x + 3 | + C

G e n e ral R ule for C ase I

A
R epresent the integrand as a sum of terms of the form for each linear factor x − a of the denomina-
x−a
tor, where A is an unknown constant. Solve for the constants. Integrating yields a sum of terms of the form
A ln |x − a|.

Remark: W e assume without proof that the integrand always has a representation of the req uired kind. For
every particular problem, this can be verified at the end of the calculation.

C ase II
D(x) is a product of linear factors, some of which occur more than once.
(3x + 5) dx
E X A M P L E 3 3 .7 : Find ∫x
− x2 − x + 1
3 .
First factor the denominator:†
x3 − x2 − x + l = (x + l)(x − l)2

Then represent the integrand 3x + 5 as a sum of the following form:

x 3 − x2 − x + 1

3x + 5 = A + B + C
x 3 − x2 − x + 1 x + 1 x − 1 ( x − 1)2

†
In trying to find linear factors of a denominator that is a polynomial with integral coefficients, test each of the divisors r of the constant
term to see whether it is a root of the polynomial. If it is, then x − r is a factor of the polynomial. In the given example, the constant
term is 1. B oth of its divisors, 1 and −1, turn out to be roots.
 28 2 CHAPTER 33 Techniques of Integration III

N ote that, for the factor (x − 1) that occurs twice, there are terms with both (x − 1) and (x − 1)2 in the denominator.
N ow clear the denominators by multiplying both sides by (x + l)(x − 1)2:

3x + 5 = A(x − l)2 + B(x + l)(x − 1) + C(x + 1) (1)

Let x = 1. Then 8 = (0)A + (2)(0)B + (2)C = 2C. Thus, C = 4.

Let x = −1. Then 2 = (4)A + (0)(−2)B + (0)C = 4A. Thus, A = 12 .
To find B, compare the coefficients of x2 on both sides of (1). O n the left it is 0, and on the right it is A + B.
H ence, A + B = 0. Since A = 12 , B = − 12 . Thus,

3x + 5 = 1 1 − 1 1 +4 1 2
x3 − x 2 − x + 1 2 x + 1 2 x − 1 ( x − 2)

Therefore,

(3x + 5) dx
∫x = 1 ln | x + 1 | − 12 ln | x − 1| + 4 ∫ dx 2
3
− x2 − x + 1 2 ( x − 1)

B y Quick Formula I,

dx
∫ ( x − 1) = ∫ ( x − 1)−2 dx = −( x − 1)−1 = − 1
2
x −1

(3x + 5) dx
So, ∫x = 1 ln | x + 1| − 12 ln | x − 1| − 4 1 + C
3
− x2 − x + 1 2 x −1

= 12 ln | x + 1| − 4 + C
| x − 1| x − 1

( x + 1) dx
E X A M P L E 3 3 .8 : Find ∫ .
x 3 ( x − 2)2

( x + 1)
R epresent the integrand in the following form:
x 3 ( x − 2)2

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

Clear denominators by multiplying by x3(x − 2)2:

x + 1 = Ax2(x − 2)2 + Bx(x − 2)2 + C(x − 2)2 + Dx3(x − 2) + Ex3

Let x = 0. Then 1 = 4C. So, C = 14 .

Let x = 2. Then 3 = 8E. So, E = 83 .
Compare coefficients of x. Then 1 = 4B − 4C. Since C = 14 , B = 12 .
Compare coefficients of x2. Then 0 = 4A − 4B + 4C. Since B = 12 and C = 14 , A = 14 .
Compare coefficients of x4. 0 = A + D. Since A = 14 , D = − 14 .

( x + 1)
So, =11+1 1 +1 1 −1 1 +3 1
x 3 ( x − 2)2 4 x 2 x 2 4 x 3 4 x − 2 8 ( x − 2)2

( x + 1) dx
and
∫ x ( x − 2) = 1 ln | x | − 1 1 − 1 12 − 1 ln | x − 2 | − 3 1 +C
3 2
4 2x 8x 4 8
x−2

= 1 ln x − 4 x +2 1 − 3 1 + C
4 x−2 8x 8 x−2
CHAPTER 33 Techniques of Integration III 28 3

G e n e ral R ule for C ase II

A A2 Ak
For each repeated linear factor (x − r) that occurs k times in the denominator, use 1 + +. . .+
x − r ( x − r )2 ( x − r)k
as part of the representation of the integrand. E very linear factor that occurs only once is handled as in Case I.

C ase III
D(x) is a product of one or more distinct irreducible q uadratic factors and possibly also some linear factors
(that may occur more than once).

G e n e ral R ule for C ase III

Linear factors are handled as in Cases I– II. For each irreducible q uadratic factor x2 + bx + c, place a term
Ax + B
in the representation of the integrand.
x + bx + c
2

( x − 1) dx
E X A M P L E 3 3 .9 : Find ∫ .
x( x 2 + 1)( x 2 + 2)
R epresent the integrand as follows:

( x − 1)
= A + Bx2 + C + Dx2 + E
x( x 2 + 1)( x 2 + 2) x x +1 x +2

Clear the denominators by multiplying by x(x2 + l)(x2 + 2).

x − 1 = A(x2 + l)(x2 + 2) + (Bx + C)x(x2 + 2) + (Dx + E)x(x2 + 1)

M ultiply out on the right:

x − 1 = (A + B + D)x4 + (B+ E)x3 + (3A + C + D)x2 + (2C + E)x + 2A

Comparing coefficients, we get:

2A = −1, 2C + E= 1, 3A + C + D = 0, B + E = 0, A+B+D=0

H ence, A = − 12 and, therefore, C + D = 32 , B + D = 12 . From the latter two eq uations, C − B = 1. From 2C + E = 1 and
B + E = 0, we get 2C − B = 1. N ow, from C − B = 1 and 2C − B = 1, we get C = 0. H ence, from C − B= 1, B = −1.
Then, from B + D = 12 , it follows that D = 32 . Finally, from B + E = 0, E = 1.

Thus,

( x − 1)
= − 1 1 − 2 x + 1 3x2 + 2
x( x 2 + 1)( x 2 + 2) 2 x x +1 2 x + 2

( x − 1) dx
H ence,
∫ x( x = − 12 ln | x | − ∫ 2 x dx + 1 ∫ 3x2 + 2 dx
2
+ 1)( x 2 + 2) x +1 2 x +2

x dx = 1 2 x dx = 1 ln( x 2 + 1)
N ow, ∫x 2
+1 2 ∫ x2 + 1 2 (by Quick Formula II)

3x + 2 dx = 3x 2
Also, ∫x 2
+2 ∫ x 2 + 2 dx + ∫ x 2 + 2 dx
   
= 3 ∫ 22 x dx + 2 tan −1  x  = 32 ln( x 2 + 2) + 2 tan −1  x 
2 x +2  2  2

( x − 1) dx 2  
Therefore, ∫ x( x = − 12 ln | x | + 14 ln( x 2 + 1) + tan −1  x  + C
2
+ 1)( x 2 + 2) 2  2
 28 4 CHAPTER 33 Techniques of Integration III

C ase IV
D(x) is a product of z ero or more linear factors and one or more irreducible q uadratic factors.

G e n e ral R ule for C ase IV

Linear factors are handled as in Cases I– II. For each irreducible q uadratic factor x2 + bx + c that occurs to
the kth power, insert as part of the representation of the integrand.

A1 x + B1 A x + B2
+ 2 2 2 +
. . . + 2Ak x + Bk k
x + bx + c ( x + bx + c)
2
( x + bx + c)

Find ∫ 2 x2 + 32 dx.
2
E X A M P L E 3 3 .1 0 :
( x + 1)

Let 2 x2 + 32 = Ax2 + B + Cx2 + D2 . Then

2

( x + 1) x + 1 ( x + 1)

2x2 + 3 = (Ax + B)(x2 + 1) + Cx + D = Ax3 + Bx2 + (A + C)x + (B + D)

Compare coefficients: A = 0, B = 2, A + C = 0, B + D = 3. H ence, C = 0, D = 1. Thus,

2 x 2 + 3 dx = 2 1
∫ (x2
+ 1)2 ∫ x 2 + 1 dx + ∫ ( x 2 + 1)2 dx
= 2 tan −1 x + ∫ 1 dx
( x 2 + 1)2

In the second integral, let x = tan q. Then

1 sec 2 θ dθ
∫ (x 2
+ 1) 2 dx = ∫
sec 4 θ
= ∫ cos 2θ dθ = 12 (θ + sin θ cosθ )

(
= 1 θ + tan
2
θ
) (
= 1 tan −1 x + 2 x
tan 2 θ + 1 2 x +1 )
2 x 2 + 3 dx = 5 tan −1 x + 1 x + C
Thus,
∫ (x 2
+ 1)2 2
2 x2 + 1

S O LV E D P R O B L E M S

Find ∫ x − 3x − 2x − 1 dx.
4 3
1.
x −x
The integrand is an improper fraction. B y division,

x 4 − x3 − x − 1 = x − x + 1 = x − x + 1
x3 − x 2 x3 − x 2 x 2 ( x − 1)

W e write x + 1 = A + B + C and obtain

x 2 ( x − 1) x x 2 x − 1

x + 1 = Ax(x − 1) + B(x − 1) + Cx2

CHAPTER 33 Techniques of Integration III 28 5

For x = 0, 1 = −B and B = −1. For x = 1, 2 = C. For x = 2, 3 = 2A + B + 4C and A = −2. Thus,

x 4 − x 3 − x − 1 dx = x dx + 2 dx + dx − 2 dx
∫ x3 − x 2 ∫ ∫ x ∫ x2 ∫ x − 1
= 12 x 2 + 2 ln | x | − 1 − 2 ln | x − 1| + C = 12 x 2 − 1 + 2 ln x + C
x x x −1

x dx
2. Find ∫ ( x + 2)( x + 3) .
Let x = A + B . Clear the denominators:
( x + 2 ) ( x + 3) x+2 x+3
x = A(x + 3) + B(x + 2)

Let x = −2. Then −2 = A. Let x = −3. Then −3 = −B. So, B = 3.

x dx 1 1
∫ ( x + 2)( x + 3) = −2 ∫ x + 2 dx + 3∫ x + 3 dx
= −2 ln | x + 2 | + 3 ln | x + 3 | + C = − ln(( x + 2)2 ) + ln(| x + 3 |)3 + C

( x + 3)3
= ln +C
( x + 2)2

x2 + 2
3. Find
∫ x( x + 2)( x − 1) dx.
Let x2 + 2 = A + B + C . Clear the denominators:
x( x + 2)( x − 1) x x + 2 x − 1

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

Let x = 0. Then 2 = −2A. So, A = −1. Let x = −2. Then 6 = 6B. So, B = 1. Let x = 1. Then 3 = 3C.
So, C = 1. H ence,

x2 + 2 1 1 1
∫ x( x + 2)( x − 1) dx = − ∫ x dx + ∫ x + 2 dx + ∫ x − 1 dx
( x + 2)( x − 1)
= − ln | x | + ln | x + 2 | + ln | x − 1| + C = ln +C
x

4. Find x3 + 1 dx.
( x + 2)( x − 1)3

Let x3 + 1 = A + B + C + D . Clear the denominators:

( x + 2)( x − 1)3 x + 2 x − 1 ( x − 1)2 ( x − 1)3

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

Let x = −2. Then −7 = −27A. So, A = 277 . Let x = 1. Then 2 = 3D. So, D = 23 . Compare coefficients of x3. Then 1 = A + B.
27 . Compare coefficients of x . 0 = −3A + C. Since A = 27 , C = 9 .
Since A = 277 , B = 20 2 7 7

x3 + 1 dx = 7 ∫ 1 dx + 20 ∫ 1 dx + 7 ∫ 1 2 dx + 2 ∫ 1 3 dx
Thus, ∫ ( x + 2)( x − 1) 3
27 x + 2 27 x − 1 9 ( x − 1) 3 ( x − 1)

= 7 ln | x + 2 | + 20 ln | x − 1| − 7 1 − 1 1 +C
27 27 9 x − 1 3 ( x − 1)2
 28 6 CHAPTER 33 Techniques of Integration III

x 3 + x 2 + x + 2 dx.
5. Find ∫ x 4 + 3x 2 + 2
x4 + 3x2 + 2 = (x2 + l)(x2 + 2). W e write x 4+ x +2 x + 2 = Ax2 + B + Cx2 + D and obtain
3 2

x + 3x + 2 x +1 x +2
x3 + x2 + x + 2 = (Ax + B)(x2 + 2) + (Cx + D)(x2 + 1)
= (A + C)x3 + (B + D)x2 + (2A + C)x + (2B + D)

H ence A + C = 1, B + D = 1, 2A + C = 1, and 2B + D = 2. Solving simultaneously yields A = 0, B = 1, C= l, D = 0.

Thus,
x 3 + x 2 + x + 2 dx = 1 x
∫ x 4 + 3x 2 + 2 ∫ x 2 + 1 dx + ∫ x 2 + 2 dx
= tan −1 x + 12 ln( x 2 + 2) + C

x 5 − x 4 + 4 x 3 − 4 x 2 + 8 x − 4 dx.
6. Find
∫ ( x 2 + 2)3
W e write x 5 − x 4 + 4 x 3 − 4 x 2 + 8 x − 4 = Ax + B + Cx + D + Ex + F . Then
( x 2 + 2)3 x 2 + 2 ( x 2 + 2)2 ( x 2 + 2)3

x5 − x4 + 4x3 − 4x2 + 8x − 4 = (Ax + B)(x2 + 2)2 + (Cx + D)(x2 + 2) + Ex + F

= Ax5 + Bx4 + (4A + C)x3 + (4B + D)x2 + (4A + 2C + E)x
+ (4B + 2D + F)

from which A = 1, B = −1, C = 0, D = 0, E = 4, F = 0. Thus the given integral is eq ual to

( x − 1) dx x dx x dx dx + 4 x dx
∫ x2 + 2
+ 4∫ 2
( x + 2)3 ∫ x 2 + 2 ∫ x 2 + 2
= − ∫ ( x 2 + 2)3
B y Quick Formula II,
x dx 2 x dx 1
∫x =1
+ 2 2 ∫ x2 + 2 2
2 = ln( x 2 + 2)

and by Quick Formula I,

x dx
∫ (x = 1 ( x 2 + 2)−3 (2 x) dx = 12 (− 12 )( x 2 + 2)−2 = − 1 2 1 2
2
+ 2)3 2 ∫ 4 ( x + 2)

x 5 − x 4 + 4 x 3 − 4 x 2 + 8 x − 4 dx = 1 ln( x 2 + 2) − 2 taan −1  x  − 1
So, ∫ ( x 2 + 2)3 2
2  2  ( x 2 + 2)2 + C

S U P P LE M E N TA R Y P R O B LE M S

In P roblems 7– 25, evaluate the given integrals.

dx = 1 ln x − 3 + C
7. ∫x 2
−9 6 x+3

x dx
8. ∫x 2 = 1 ln ( x + 1)( x − 4)4 + C
− 3x − 4 5

x 2 − 3x − 1 dx = ln x1 2 ( x + 2)3 2 + C
9. ∫x 3
+ x2 − 2x x −1
CHAPTER 33 Techniques of Integration III 28 7

10 . ∫x
dx = 1 ln x + 1 + C
2
+ 7x + 6 5 x + 6

x 2 + 3x − 4 dx = x + ln ( x + 2)( x − 4)4 + C
11. ∫x 2
− 2x − 8

x dx
12 . ∫ ( x − 2) = ln | x − 2 | − 2 + C
2
x−2

x4 4 + 1
13 .
∫ (1 − x) 3 dx = − 12 x 2 − 3x − ln(1 − x)6 −
1 − x 2(1 − x)2
+C

dx = ln x
14 . ∫x 3
+x x2 + 1
+C

x 3 + x 2 + x + 3 dx = ln x 2 + 3 + tan −1 x + C
15 .
∫ (x 2
+ 1)( x 2 + 3)

x 4 − 2 x 3 + 3x 2 − x + 3 dx = 1 x 2 + ln x
16 . ∫ x 3 − 2 x 2 + 3x 2
x2 − 2x + 3
+C

2 x 3 dx
17 .
∫ (x = ln( x 2 + 1) + 2 1 + C
2
+ 1)2 x +1

18 .
∫ ( x 2 + 4) 2 2
2 ()
2 x 3 + x 2 + 4 dx = ln( x 2 + 4) + 1 tan −1 x + 4 + C
x2 + 4

x 3 + x − 1 dx = ln x 2 + 1 − 1 tan −1 x − 1 x + C
19 .
∫ ( x 2 + 1)2 2
2 x2 + 1

x 4 + 8 x 3 − x 2 + 2 x + 1 dx = ln x 3 − x 2 + x − 3 + 2 tan −1  2 x − 1  + C
20. ∫ ( x 2 + 3)( x 3 + 1) ( x + 1)2 x +1 3  3 

x 3 + x 2 − 5x + 15 dx = ln x 2 + 2 x + 3 + 5 taan −1  x + 1  − 5 tan −1  x  + C
2 1. ∫ (x 2
+ 5)( x 2 + 2 x + 3) 2  2   5 

x 6 + 7 x 5 + 15x 4 + 23x 2 + 25x − 3 dx = 1 − 3 + ln 2 x + 1 + C

2
22.
∫ ( x 2 + x + 2)2 ( x 2 + 1)2 x2 + x + 2 x2 + 1 x +x+2

= 1 + 1 ln e −x 3 + C
x
dx
23. ∫e 2x
− 3e x 3e x 9 e
(H int: Let ex = u.)

sin x dx 1 + cos 2 x
24. ∫ cos x(1 + cos 2
x)
= ln
cos x
+C (H int: Let cos x = u.)

(2 + tan 2 θ )sec 2 θ  
25. ∫ dθ = ln |1 + tan θ | + 2 tan −1  2 tan θ − 1  + C
1 + tan 3 θ 3  3 
 C H A P TE R 3 6

Applications of Integration III:

Area of a Surface of R evolution
If an arc of a curve is revolved about a line that does not intersect the arc, then the resulting surface is called
a surface of revolution. B y the surface area of that surface, we mean the area of its outer surface.
Let f be a continuous function on [a, b] that is differentiable in (a, b) and such that f (x) ≥ 0 for a ≤ x ≤ b.
Then the surface area S of the surface of revolution generated by revolving the graph of f on [a, b] about the
x axis is given by the formula
2

S = 2π ∫ y 1 +   dx = 2π ∫ f ( x) 1 + ( f ′( x))2 dx
b dy b
(36.1)
a  dx  a

For a justification of this formula, see P roblem 11.

There is another formula like (36.1) that is obtained when we exchange the roles of x and y. Let g be a
continuous function on [c, d ] that is differentiable on (c, d) and such that g( y) ≥ 0 for c ≤ y ≤ d. Then the
surface area S of the surface of revolution generated by revolving the graph of g on [c,d] about the y axis is
given by the formula:
2

S = 2π ∫ x 1 +   dy = 2π ∫ g( y) 1 + (g ′( y))2 dy
d dx d
(36.2)
c  dy  c

Similarly, if a curve is given by parametric eq uations x = f (u), y = g(u) (see Chapter 37), and, if the arc
from u = u1 to u = u2 is revolved about the x axis, then the surface area of the resulting surface of revolution
is given by the formula
2 2

S = 2π ∫ y   +   du
u2 dx dy
(36.3)
u1  du   du 

H ere, we have assumed that f and g are continuous on [u1, u2] and differentiable on (u1, u2), and that y = g(u) ≥ 0
on [u1, u2]. Another such formula holds in the case of a revolution around the y axis.

S O LV E D P R O B L E M S

1. Find the area S of the surface of revolution generated by revolving about the x axis the arc of the parabola y2 = 12x
from x = 0 to x = 3.
B y implicit differentiation,

2
y 2 + 36
1+   =
dy 6 dy
= and
dx y  dx  y2

301
XYZ[\ ] ^_ `ab ccdefdddefddcefdgbh [i_jkl!m\ no%p q] r r(XYs%Znt] ju,ev t lwXr ] lx_j\ jy1Y\2`1j\ su3Yyzu,jw
 302 CHAPTER 36 Applications of Integration III

B y (36.1),
3 y 2 + 36 3
S = 2π ∫ y dx = 2π ∫ 12 x + 36 dx
0 y 0

= 2π (8(12 x + 36) 2 )]30 = 24(2 2 − 1)π

3

2. Find the area S of the surface of revolution generated by revolving about the y axis the arc of x = y3 from y = 0 to
y = 1.
2
and 1 +  dx  = 1 + 9 y 4. So, by (36.2),
dx = 3y 2
dy  dy 
1 1
S = 2π ∫ x 1 + 9 y 4 dy = 2π ∫ y3 1 + 9 y 4 dy
0 0

= π ∫ (1 + 9 y 4 ) (36 y3 ) dy
1
1 2

18 0

= π 2 (1 + 9 y 4 ) 2 ]10
3

18 3
= π (10 10 − 1)
27

3. Find the area of the surface of revolution generated by revolving about the x axis the arc of y2 + 4x = 2 ln y from
y = 1 to y = 3.
2
3 1 + y2
S = 2π ∫ y 1 +  dx  dy = 2π ∫ y
d 3
dy = π ∫ (1 + y 2 ) dy = 32 π
e  dy  1 2y 1 3

4. Find the area of the surface of revolution generated by revolving a loop of the curve 8a2y2 = a2x2 − x4 about the x
axis. (See Fig. 36-1.)

Fig. 36 -1

2
dy a 2 x − 2 x 3 (a 2 − 2 x 2 ) 2 (3a 2 − 2 x 2 )2
H ere = and 1 +  dx  = 1 + 2 2 2 =
dx 8a 2 y  dy  8a (a − x ) 8a 2 (a 2 − x 2 )

2
a x a2 − x2
S = 2π ∫ 2 1 +   dx = 2π ∫ 3a 2 − 2 x 2 dx
a dy
H ence
0  
dx 0 2a 2 2a 2 a 2 − x 2

= π2
a
∫ (3a 2 − 2 x 2 ) x dx = 1 π a 2
4a 0 4
2 y2
5. Find the area of the surface of revolution generated by revolving about the x axis the ellipse x + = 1.
16 4
16 y 2 + x 2
dx = π
4 4
S = 2π ∫ y ∫ 64 − 3x 2 dx
−4 4y 2 −4

4
x 3  x 3   4 3 
= π  64 − 3x 2 + 32 sin −1   = 8π  1 + 9 π 
2 3 2  8     
−4
CHAPTER 36 Applications of Integration III 303

6. Find the area of the surface of revolution generated by revolving about the x axis the hypocycloid x = a cos3 q,
y = a sin3 q.
The req uired surface is generated by revolving the arc from q = 0 to q = π. W e have

( )
2 2
dx = −3a cos 2 θ sin θ , dy = 3a sin 2 θ cosθ , and dx +  dy  = 9a 2 cos 2 θ sin 2 θ . Then
dθ dθ dθ  dθ 

( ddxθ ) +  ddyθ  dθ = 2(2π )∫

π 2 2 π

S = 2(2π ) ∫ y (a sin 3 θ )3a cos θ sin θ dθ

2 2

0 0

= 12a π
2
(sq uare units)
5

7. Find the area of the surface of revolution generated by revolving about the x axis the cardioid
x = cos 3q − cos 2q, y = 2 sin q − sin 2q.
The req uired surface is generated by revolving the arc from q = 0 to q = π. (See Fig. 36-2.) W e have

dx = −2 sin θ + 2 sin 2θ , dy
= 2 cosθ − 2 cos 2θ ,
dθ dθ

Fig. 36 -2

and

( ddxθ ) +  ddyθ 
2 2
= 8 (1 − sin θ sin 2θ − cosθ cos 2θ ) = 8 (1 − cosθ )

Then
π
S = 2π ∫ (2 sin θ − sin 2θ )(2 2 1 − cosθ ) dθ
0

π
π  16 2 
= 8 2π ∫ sin θ (1 − cosθ ) 2 dθ =  π (1 − cosθ )5 2  
3

0  5  0

= 128π (sq uare units)

5
8. Show that the surface area of a cylinder of radius r and height h is 2πrh.
dy
The surface is generated by revolving about the x axis the curve y = r from x = 0 to x = h. Since = 0,
2 dx
1 +   = 1. Then, by (36.1),
dy
 dx 
h h
S = 2π ∫ r dx = 2π ( rx )  = 2π rh
0 0

9. Show that the surface area of a sphere of radius r is 4πr2.

The surface area is generated by revolving about the x axis the semicircle y = r 2 − x 2 from x = − r to x = r.
B y symmetry, this is double the surface area from x = 0 to x = r. Since y2 = r2 − x2,

2
x 2 + y2 r 2
1 +   = 1 + x2 =
dy dy dy 2
2y = −2 x and therefore =−x and = 2
dx dx y  dx  y y2 y
 304 CHAPTER 36 Applications of Integration III

H ence, by (36.1),

r
S = 2 ⋅ 2π ∫ y r 2 dx = 4π r ∫ 1 dx = 4π r x  = 4π r 2
r 2 r

0 y 0 0

10 . (a) Show that the surface area of a cone with base of radius r and with slant height s (see Fig. 36-3) is πrs.
(b) Show that the surface area of a frustum of a cone having bases of radius r1 and r2 and slant height u (see
Fig. 36-4) is π(r1 + r2)u. (N ote that the frustum is obtained by revolving the right-hand segment of the slant
height around the base of the triangle.)

Fig. 36 -3 Fig. 36 -4

(a) Cut open the cone along a slant height and open it up as part of a circle of radius s (as shown in Fig. 36-5).
N ote that the portion of the circumference cut off by this region is 2pr (the circumference of the base of the
cone.) N ow the desired area S is the difference between ps2 (the area of the circle in Fig. 36-5) and the area
A1 of the circular sector with central angle q. This area A1 is θ (π s 2 ) = 1 θ s 2. Since the arc cut off by q is
2π 2
2ps − 2pr, we get θ = 2π s − 2π r . Thus, A1 = p(s − r)s. H ence, S = ps2 − p(s − r)s = prs sq uare units.
s

Fig. 36 -5

u1 u1 + u ru
(b) From the similar triangles in Fig. 36-4, we get = . Then r2 u1 = r1 ul + r1u. So, u1 = r 1− r . N ow, by
r1 r2 2 1
part (a), the surface area of the frustum is pr2(u1 + u) − prl ul = p(r2 - rl)ul + pr2 u = pr1 u + pr2 u = p(r1 + r2)u
sq uare units.
CHAPTER 36 Applications of Integration III 305

11. Sketch a derivation of formula (36.1).

Assume that [a, b] is divided into n eq ual subintervals, [xk-1, xk], each of length ∆x = b − a . The total surface
n
area S is the sum of the surface areas Sk generated by the arcs between the points (xk-1, f (xk-1)) and (xk, f (xk)), each
of which is approximated by the surface area generated by the line segment between (xk-1, f (xk-1)) and (xk, f (xk)).
The latter is the area of a frustum of a cone. In the notation of Fig. 36-6, this is, by virtue of P roblem 10(b):

 f ( xk −1 ) + f ( xk ) 
(
π f ( xk −1 ) + f ( xk ) ) ( ∆x )2 + ( ∆y )2 = 2π 
 2 

( ∆x )2 + ( ∆y )2

N ow, f ( xk −1 ) + f ( xk ), being the average of f (xk-1) and f (xk), is between those two values and, by the
2 2
 ∆y 
intermediate value theorem, is eq ual to f ( xk* ) for some xk* in (xk-1, xk). Also, ( ∆x ) + ( ∆y ) = 1 +  ∆x  ∆x . B y
2 2

∆y
the mean value, theorem ∆ x = f ′ ( xk# ) for some xk# in (xk-1, xk). Thus, S is approximated by the sum

( )
n

∑ 2π f ( x ) 1 + f ′ ( xk# ) ∆x
2
*
k
k =1

and it can be shown that this sum can be made arbitrarily close to 2π ∫ f ( x ) 1 + ( f ′ ( x )) dx.† H ence, the latter is
b 2

a
eq ual to S.

Fig. 36 -6

S U P P LE M E N TA R Y P R O B LE M S

In P roblems 12– 20, find the area of the surface of revolution generated by revolving the given arc about the given axis:

12 . y = mx from x = 0 to x = 2; x axis Ans. 4 mπ 1 + m 2

13 . y = 1 x 3 from x = 0 to x = 3; x axis Ans. p(82√82 - 1)/9

3

†
In general, the following result can be proved:
B lis s ’s T h e o r e m : Assume f and g are continuous on [a, b]. D ivide [a, b] into subintervals [xk-1, xk] with a = x0 < x1 < · · · < xn < b,
n
and let Dk x = xk − xk-l. In each [xk-1, xk], choose xk* and xk#. Then the approximating sum
b
∑ f ( x ) g ( x ) ∆ xcan be made arbitrarily close
k =1
*
k
#
k k

to ∫ f ( x ) g ( x ) dx by letting n → + ∞ and making the maximum lengths of the subintervals approach 0.

a
 306 CHAPTER 36 Applications of Integration III

14 . y = 13 x 3 from x = 0 to x = 3; y axis Ans.

2  (
1 π 9 82 + ln 9 + 82 
 )
15 . O ne loop of 8y2 = x2(l − x2); x axis Ans. 1π
4

16 . y = x3/6 + l/2x from x = 1 to x = 2; y axis Ans. ( 154 + ln 2)π

17 . y = ln x from x = 1 to x = 7; y axis Ans.  (
34 2 + ln 3 + 2 2  π
 )
18 . O ne loop of 9y2 = x(3 − x)2; y axis Ans. 28p √3/5

19. An arch of x = a(q − sin q ), y = a(1 − cos q ); x axis Ans. 64p a2/3

20. x = e t cos t, y = et sin t from t = 0 to t = 1 π ; x axis Ans. 2p √2(2ep + l)/5

2

21. F ind the su rface are a of a z one cu t from a sp he re of rad iu s r b y tw o p aralle l p lane s, e ach at a d istance 1 a from
2
the ce nte r.
Ans. 2par

22. F ind the su rface are a of a toru s (d ou g hnu t) g e ne rate d b y re v olv ing the circle x2 + (y −b)2 = a2 ab ou t the x axis.
Assu me 0 < a < b.
Ans. 4π2ab
 C H A P TE R 3 7

Parametric Representation
of Curves
Parametric Equations
If the coordinates (x, y) of a point P on a curve are given as functions x = f (u), y = g(u) of a third variable or
p aram eter, u, the equations x = f (u) and y = g(u) are called p aram etric eq uatio ns of the curve.

EXAMPLE 37.1:
(a) x = cos θ, y = 4 sin2 θ are parametric equations, with parameter θ, of the parabola 4x2 + y = 4, since
4x2 + y = 4 cos2 θ + 4 sin2 θ = 4.

(b) x = 12 t , y = 4 − t2 is another parametric representation, with parameter t, of the same curve.

It should be noted that the first set of parametric equations represents only a portion of the parabola (F ig. 3 7 -1(a)),
whereas the second represents the entire curve (F ig. 3 7 -1(b)).

y
q = 1p
2

q = 3p q = 1p
4 4

q= p x
O q= 0

(a)

Fig. 37-1

EXAMPLE 37.2 :
(a) The equations x = r cos θ, y = r sin θ represent the circle of radius r with center at the origin, since
x 2 + y 2 = r 2 cos2 θ + r 2 sin2 θ = r 2(cos2 θ + sin2 θ) = r 2. The parameter θ can be thought of as the angle from the
positive x axis to the segment from the origin to the point P on the circle (Fig. 37-2).

(b) The equations x = a + r cos θ, y = b + r sin θ represents the circle of radius r with center at (a, b), since
(x − a)2 + (y − b)2 = r 2 cos2 θ + r 2 sin2 θ = r 2 (cos2 θ + sin2 θ) = r 2.

307
{|}~ € ‚ ƒ„… ††‡ˆ‰‡‡‡ˆ‰‡‡†ˆ‰‡Š…‹ ~Œ‚Ž! ‘’%“ ”€ • •({|–%}‘—€ ˜,ˆ™ — š{• € ›‚ œ1|2ƒ1 –˜3|œ˜,š
 308 CHAPTER 37 Parametric Representation of Curves

Fig. 37-2

A ssume that a curve is specified by means of a pair of parametric equations x = f (u) and y = g(u). Then
dy d2y
the first and second derivatives and 2 are given by the following formulas.
dx dx
(37.1) F irst D eriv ativ e
dy  dy   dx 
=
dx  du   du 

dy dy dx
This follows from the C hain R ule formula = ⋅ .
du dx du
(37.2 ) S econd D eriv ativ e
d 2 y  d  dy   dx
=
dx 2  du  dx   du

d  dy  d 2 y dx
This follows from the C hain R ule formula = ⋅ .
du  dx  dx 2 du

Arc Leng th for a Parametric C urv e

If a curve is given by parametric equations x = f (t), y = g(t), then the length of the arc of the curve between
the points corresponding to parameter values t1 and t2 is
2 2
t2  dx   dy 
L=∫ + dt
t1  dt   dt 

This formula can be derived by an argument similar to that for the arc length formula (29.2).

S O LV ED PR O B LEMS
dy d2y
1. Find and 2 if x = t − sin t, y = 1 − cos t.
dx dx
dx = 1 − cos t dy dy sin t
and = sin t. B y (37.1), = . Then
dt dt dx 1 − cos t

d  dy  = (1 − cos t ) (cos t ) − (sin t ) (sin t )

dt  dx  (1 − cos t )2

cos t − (cos 2 t + sin 2 t ) cos t − 1 1

= = =
(1 − cos t )2 (1 − cos t )2 cos t − 1
CHAPTER 37 Parametric Representation of Curves 309

H ence, by (37.2),

d2y 1 1
= (1 − cos t ) = −
dx 2 cos t − 1 (1 − cos t )2

dy d2y
2. Find and if x = et cos t, y = et sin t.
dx dx 2
dx = et (cos t − sin t ) and dy dy cos t + sin t
dt = et (cos t + sin t ). B y (37.1), dx = cos t − sin t . Then,
dt

d  dy  = (cos t − sin t ) − (cos t + sin t )(− sin t − cos t )

2

dt  dx  (cos t − sin t )2

(cos t − sin t )2 + (cos t + sin t )2 2(cos 2 t + sin 2 t )

= =
(cos t − sin t )2 (cos t − sin t )2
2
=
(cos t − sin t )2

S o, by (37.2),

d2y 2 2
= et (cos t − sin t ) = t
dx 2 (cos t − sin t )2 e (cos t − sin t )

3. Find an equation of the tangent line to the curve x = t , y = t − 1 at the point where t = 4.
t
dx = 1 dy 1 dy 1
and = 1 + 3/ 2 . B y (37.1), = 2 t + . S o, the slope of the tangent line when t = 4 is
dt 2 t dt 2t dx t
2 4 + 4 = 4 . W hen t = 4, x = 2 and y = 2 . A n equation of the tangent line is y − 72 = 174 ( x − 2).
1 17 7

4. The position of a particle that is moving along a curve is given at time t by the parametric equations x =
2 − 3 cos t, y = 3 + 2 sin t, where x and y are measured in feet and t in seconds. (S ee Fig. 37-3.) N ote that
9 ( x − 2) + 4 ( y − 3) = 1, so that the curve is an ellipse. Find: (a) the time rate of change of x when t = π/3;
1 2 1 2

(b) the time rate of change of y when t = 5π/3; (c) the time rate of change of the angle of inclination θ of the
tangent line when t = 2π/3.
dy dy dy 2
= 3sin t and = 2cos t . Then tan θ = = cot t .
dt dt dx 3

(a) W hen t = π , dx =
3 3
ft/sec
3 dt 2
(b) W hen t = 5π ,
dy
3 dt = 2( 2 ) = 1 ft/sec
1

− 23 csc 2 t −6 csc 2 t
(c) θ = tan −1 ( 23 cot t ). S o, dθ = = .
dt 1 + 9 cot t t 9 + 4 cot 2 t
4 2

y
q
t = 1p
3 t = 2p
3
t= 0 • (2, 3)

t = 5p
3
x
O

Fig. 37-3
 31 0 CHAPTER 37 Parametric Representation of Curves

W hen t = 2π , dθ = −6(2 / 3 )
2
24. Thus, the angle of inclination of the tangent line is decreasing at the
3 dt 9 + 4(−1/ 3 )2 = − 31
rate of 24
31 radians per second.

5. Find the arc length of the curve x = t2, y = t3 from t = 0 to t = 4.

( )
2 2
dx = 2t , dy = 3t 2 and dx +  dy  = 4t 2 + 9t 4 = 4t 2 (1 + 9 t 2 ).
dt dt dt  dt  4

Then
4 4
L = ∫ 2t 1 + 94 t 2 dt = 4
9 ∫ (1 + 94 t 2 )1/ 2 ( 92 t ) dt
0 0

(1 + 94 t 2 )3/ 2 ]0 =
4
= 4 2
9 3
8
27 (37 37 − 1)

6. Find the length of an arch of the cy cloid x = θ − sin θ, y = 1 − cos θ between θ = 0 and θ = 2π.

( ) ()
2 2
dx = 1 − cosθ , dy = sin θ and dx +  dy  = (1 − cosθ )2 + sin 2 θ = 2(1 − cosθ ) = 4 sin 2 θ . Then
dθ dθ dθ  dθ  2

() ()
2π
L = 2 ∫ sin θ dθ = − 4 cos θ  = −4(cos π − cos 0) = 8
4

0 2 2  0

S U PPLEMEN T AR Y PR O B LEMS

dy d2y
In P roblems 7–11, find: (a) ; (b) .
dx dx 2

7. x = 2 + t, y = 1 + t2 Ans. (a) 2t; (b) 2

8. x = t + 1/t, y = t + 1 Ans. (a) t2/(t2 − 1 ); (b) − 2t3/(t 2 − 1 )3

9. x = 2 sin t, y = cos 2t Ans. (a) −2 sin t; (b) −1

10 . x = cos3 θ, y = sin3 θ Ans. (a) −tan θ ; (b) 1/(3 cos4 θ sin θ)

11. x = a(cos φ + φ sin φ ), y = a(sin φ − φ cos φ ) Ans. (a) tan φ ; (b) 1/(aφ cos3 φ )

12 . Find the slope of the curve x = e–t cos 2t, y = e–2t sin 2t at the point t = 0.

Ans. −2

13 . Find the rectangular coordinates of the highest point of the curve x = 96t, y = 96t − 16t2. (H int: Find t for
maximum y.)

Ans. (28 8 , 1 44)

14 . Find equations of the tangent line and normal line to the following curves at the points determined by the given
value of the parameter:
(a) x = 3et, y = 5e– t at t = 0
(b) x = a cos4 θ, y = a sin4 θ at θ = π
4
Ans. (a) 3y + 5 x = 30 , 5 y − 3x = 1 6 ; (b) 2x + 2y = a, y = x
CHAPTER 37 Parametric Representation of Curves 31 1

15 . Find an equation of the tangent line at any point P(x, y) of the curve x = a cos3 t, y = a sin3 t. S how that the length
of the segment of the tangent line intercepted by the coordinate axes is a.

Ans. x sin t + y cos t = a sin 2t

2

16 . For the curve x = t2 − 1, y = t3 − t, locate the points where the tangent line is (a) horiz ontal, and (b) vertical. S how
that, at the point where the curve crosses itself, the two tangent lines are mutually perpendicular.

3
Ans. (a) t = ± ; (b) t = 0
3

In P roblems 17–20, find the length of the specified arc of the given curve.

17 . The circle x = a cos θ, y = a sin θ from θ = 0 to θ = 2π.

Ans. 2πa

18 . x = et cos t, y = et sin t from t = 0 to t = 4.

Ans. 2 (e4 − 1)

19 . x = ln 1 + t 2 , y = tan–1 t from t = 0 to t = 1.

Ans. ln(1 + 2 )

2 0 . x = 2 cos θ + cos 2θ + 1, y = 2 sin θ + sin 2θ.

Ans. 16

2 1. The position of a point at time t is given as x = 12 t 2, y = 19 (6t + 9)3/ 2. Find the distance the point travels from
t = 0 to t = 4.

Ans. 20

2 2 . Identify the curves given by the following parametric equations and write equations for the curves in terms of x and y:
(a) x = 3t + 5, y = 4t − 1 Ans. S traight line: 4x − 3y = 23
(b) x = t + 2, y = t2 Ans. P arabola: y = (x − 2)2
(c) x = t − 2, y = t Ans. H y perbola: y = 2 + 1
t−2 x
(d) x = 5 cos t, y = 5 sin t Ans. C ircle: x2 + y2 = 25

2 3 . (G C ) U se a graphing calculator to find the graphs of the following parametric curves:

(a) x = θ + sin θ, y = 1 − cos θ (cy cloid)
(b) x = 3 cos3 θ, y = 3 sin3 θ (hy pocy cloid)
(c) x = 2 cot θ, y = 2 sin2 θ (witch of A gnesi)
(d) x = 3θ 3 , y = 3θ 3
2
(folium of D escartes)
(1 + θ ) (1 + θ )
 C H A P TE R 3 8

Curvature
D eriv ativ e of Arc Leng th
L et y = f(x) have a continuous first derivative. L et A(x0, y0) be a fixed point on its graph (see Fig. 38-1) and denote
by s the arc length measured from A to any other point P(x, y) on the curve. W e k now that, by formula (29.2),
2
x  dy 
s=∫ 1+ dx
x0  dx 

if s is chosen so as to increase with x. L et Q(x + ∆x, y + ∆y) be a point on the curve near P. L et ∆ s denote
the arc length from P to Q. Then
2
ds ∆s  dy 
= lim = ± 1+
dx ∆x→0 ∆ x  dx 

and, similarly ,
2
ds ∆s  dx 
= lim = ± 1+
dy ∆y→0 ∆y  dy 

The plus or minus sign is to be tak en in the first formula according as s increases or decreases as x increases,
and in the second formula according as s increases or decreases as y increases.

Fig. 38 -1

W hen a curve is given by parametric equations x = f (u), y = g(u),

2 2
ds ∆s  dx   dy 
= lim =± +
du ∆u→0 ∆u  du   du 

H ere the plus or minus sign is to be tak en according as s increases or decreases as u increases.

31 2
{|}~ € ‚ ƒ„… ††‡ˆ‰‡‡‡ˆ‰‡‡†ˆ‰‡Š…‹ ~Œ‚Ž! ‘’%“ ”€ • •({|–%}‘—€ ˜,ˆ™ — š{• € ›‚ œ1|2ƒ1 –˜3|œ˜,š
CHAPTER 38 Curvature 31 3

To avoid the repetition of ambiguous signs, we shall assume hereafter that the direction on each arc has
been established so that the derivative of arc length will be positive.

C urv ature
The curvature K of a curve y = f (x) at any point P on it is defined to be the rate of change of the direction of
the curve at P, that is, of the angle of inclination of the tangent line at P, with respect to the arc length s.
(S ee Fig. 38-2.) Intuitively , the curvature tells us how fast the tangent line is turning. Thus, the curvature is
large when the curve bends sharply .

Fig. 38 -2
A s formulas for the curvature, we get:
d2y
dτ ∆τ dx 2
K= = lim = (38.1)
ds ∆s→0 ∆s  2 3/ 2
 dy  
1 +
  dx  

or, in terms of y,
d 2x
−
dy 2
K= 3/ 2 (38.2)
  dx  2 
 1 +  dy  

For a derivation, see P roblem 13.

K is sometimes defined so as to be positive. If this is assumed, then the sign of K should be ignored in
what follows.

T h e R ad ius of C urv ature

1
The radius of curvature R at a point P on a curve is defined by R = , provided that K ≠ 0.
K
T h e C ircle of C urv ature
The circle of curvature, or o sculating circle of a curve at a point P on it, is the circle of radius R ly ing on the
concave side of the curve and tangent to it at P. (S ee Fig. 38-3.)

Fig. 38 -3
 31 4 CHAPTER 38 Curvature

To construct the circle of curvature, on: the concave side of the curve, construct the normal line at P and
on it lay off a segment PC of length R. The point C is the center of the required circle.

T h e C enter of C urv ature

The center of curvature for a point P(x, y) of a curve is the center C of the circle of curvature at P. The co-
ordinates ( , ) of the center of curvature are given by

dy   dy  
2 2
1 +  dy 
dx   dx   1+
 dx 
α = x− β = y+ 2
d 2 y / dx 2 d y / dx 2

or by

dx   dx  
2 2
 dx  1 +
1+
 dy  dy   dy  
α = x+ 2 β = y−
d x / dy 2 d 2 x / dy 2

S ee P roblem 9 for details.

T h e Ev olute
The evolute of a curve is the locus of the centers of curvature of the given curve. (S ee P roblems 11–12.)

S O LV ED PR O B LEMS

1. Find ds at P(x, y) on the parabola y = 3x2.

dx
2
ds = 1 +  dy  = 1 + (6 x)2 = 1 + 36 x 2
dx  dx 

2. Find ds and ds at P(x, y) on the ellipse x2 + 4y2 = 8.

dx dy
dy dy 4y
S ince 2 x + 8 y = 0, = − x and dx = − . Then
dx dx 4y dy x

2
x 2 + 16 y 2 32 − 3x 2
1+   = 1+ x 2 = 32 − 3x 2
dy 2
= and ds =
 dx  16 y 16 y 2 32 − 4 x 2 dx 32 − 4 x 2

2
16 y 2 x 2 + 16 y 2 2 + 3y 2 2 + 3y 2
1 +  dx  = 1 + 2 = = and ds =
 dy  x x2 2 − y2 dy 2 − y2

3. Find ds at P( ) on the curve x = sec , y = tan .

dθ

( ddxθ ) +  ddyθ 
2 2
ds = = sec 2 θ tan 2 θ + sec 4 θ = |secc θ | tan 2 θ + sec 2 θ
dθ
CHAPTER 38 Curvature 31 5

4. The coordinates (x, y) in feet of a moving particle P are given by x = cos t – 1, y = 2 sin t + 1, where t is the time
in seconds. A t what rate is P moving along the curve when (a) t = 5π/6, (b) t = 5π/3, and (c) P is moving at its
fastest and slowest?

( )
2 2
+   = sin 2 t + 4 cos 2 t = 1 + 3 cos 2 t
ds = dx dy
dt dt  dt 

(a) W hen t = 5π/6, ds/dt = 1 + 3( 43 ) = 13 / 2 ft/sec.

(b) W hen t = 5π/3, ds/dt = 1 + 3( 14 ) = 7 /2 ft/sec.
(c) L et S = ds = 1 + 3 cos 2 t . Then dS = −3cos t sin t . S olving dS/dt = 0 gives the critical numbers t = 0,
dt dt S
π/2, π, 3π/2.

W hen t = 0 and π, the rate ds/dt = 1 + 3(1) = 2 ft/sec is fastest. W hen t = π/2 and 3π/2, the rate
ds/dt = 1 + 3(0) = 1 ft/sec is slowest. The curve is shown in Fig. 38-4.

Fig. 38 -4

5. Find the curvature of the parabola y2 = 12x at the points: (a) (3, 6); (b) ( 43, –3); (c) (0, 0).
2
dy 6  dy  36 d2y dy
= ; so 1 +  dx  = 1 + 2 and = − 62 = − 363
dx y y dx 2 y dx y

2
−1/ 6
(a) A t (3, 6): 1 +   = 2 and
dy d2y 2
= − 1, so K = 23/ 2 = − 24 .
 dx  dx 2 6
2
d2y
(b) A t ( 43 , − 3): 1 +   = 5 and 2 = 4 , so K = 43//32 =
dy 4 5
.
 dx  dx 3 5 75
2
is undefined. B ut dx = = 0, 1 +  dx  = 1, d x2 = 1 , and K = − 1 .
dy y 2
(c) A t (0, 0),
dx dy 6  dy  dy 6 6

6. Find the curvature of the cy cloid x = – sin , y = 1 – cos at the highest point of an arch. (S ee Fig. 38-5.)

Fig. 38 -5
 31 6 CHAPTER 38 Curvature

To find the highest point on the interval 0 < x < 2π; dy/d = sin , so that the critical number on the interval is
x = π. S ince d2y/d 2 = cos < 0 when = π, the point = π is a relative maximum point and is the highest point
of the curve on the interval.
To find the curvature,

dx = 1 − cosθ ,
dθ
dy
dθ
= sin θ ,
dy
= sin θ ,
dx 1 − cosθ
d2y d
= (sin θ dθ = −
dx 2 dθ 1 − cosθ dx ) 1
(1 − cosθ )2

A t = π, dy/dx = 0, d 2 y / dx 2 = − 14 , and K = − 14 .

7. Find the curvature of the cissoid y2(2 – x) = x3 at the point (1, 1). (S ee Fig. 38-6.)

Fig. 38 -6

D ifferentiating the given equation implicitly with respect to x, we obtain

− y 2 + (2 − x)2 yy ′ = 3x 2 (1)

and

−2 yy ′ + (2 − x)2 yy ′′ + (2 − x)2( y ′)2 − 2 yy ′ = 6 x (2)

From (1), for x = y = 1, –1 + 2y′ = 3 and y′ = 2. S imilarly , from (2), for x = y = 1 and y′ = 2, we find y′′ = 3.
Then K = 3 /(1 + 4)3/ 2 = 3 5 / 25.

8. Find the point of greatest curvature on the curve y = ln x.

dy 1 d2y −x dK = 2 x 2 − 1
= and = − 12 . S o, K= and
dx x dx 2 x (1 + x 2 )3/ 2 dx (1 + x 2 )5/ 2

 
The critical number is, therefore, x = 1 . The required point is  1 , − ln 2  .
2  2 2 

9. Find the coordinates of the center of curvature C of the curve y = f(x) at a point P(x, y) at which y′ ≠ 0. (S ee Fig. 38-3.)
The center of curvature C( , ) lies: (1) on the normal line at P and (2) at a distance R from P measured
toward the concave side of the curve. These conditions give, respectively ,

[1 + ( y ′) 2 ] 3
β − y = − 1 (α − x) and (α − x)2 + (β − y)2 = R 2 =
y′ ( y ′′)2

From the first, – x = – y′( – y). S ubstitution in the second y ields

[1 + ( y ′ ) 2 ] 3 1 + ( y ′) 2
( β − y) 2 [ 1 + ( y ′ ) 2 ] = and, therefore, β−y=±
( y ′′)2 y′′
CHAPTER 38 Curvature 31 7

To determine the correct sign, note that, when the curve is concave upward, y′′ > 0 and, since C then lies above
P, – y > 0. Thus, the proper sign in this case is +. (Y ou should show that the sign is also + when y′′ < 0.) Thus,

1 + ( y ′) 2 y ′[1 + ( y ′)2 ]
β = y+ and α = x −
y′′ y′′

10 . Find the equation of the circle of curvature of 2xy + x + y = 4 at the point (1, 1).

D ifferentiating y ields 2y + 2xy′ + 1 + y′ = 0. A t (1, 1), y′ = –1 and 1 + (y′)2 = 2. D ifferentiating again y ields
4y′ + 2xy′′ + y′′ = 0. A t (1, 1), y ′′ = 43. Then

−1(2) 5
K = 4 /3 , R= 3 2 , α = 1− = , β = 1+ 2 = 5
2 2 2 4 /3 2 4 /3 2

The required equation is (x – )2 + (y – )2 = R2 or ( x − 52 )2 + ( y − 52 )2 = 92.

11. Find the equation of the evolute of the parabola y2 = 12x.

A t P(x, y):

2
1 +   = 1 + 362 = 1 + 3 ,
dy 6 3 dy d2y 3
= = , = − 363 = − 3/ 2
dx y x  dx  y x dx 2 y 2x

Then

3 / x (1 + 3/x) 2 3 ( x + 3)
α = x− = x+ = 3x + 6
− 3/2 x 3/ 2 3

and

1 + 36/y 2 y3 + 36 y y3
β = y+ = y− =−
−36/y3 36 36

The equations = 3x + 6, = – y3/36 may be regarded as parametric equations of the evolute with x and
y, connected by the equation of the parabola, as parameters. H owever, it is relatively simple in this problem
to eliminate the parameters. Thus, x = ( – 6)/3, y = − 3 36β , and substituting in the equation of the parabola,
we have

(36β )2 / 3 = 4(α − 6) or 81β 2 = 4(α − 6)3

The parabola and its evolute are shown in Fig. 38-7.

Fig. 38 -7
 31 8 CHAPTER 38 Curvature

12 . Find the equation of the evolute of the curve x = cos + sin , y = sin – cos .
A t P(x, y):

dx = θ cosθ , dy dy d 2 y sec 2 θ sec3 θ

= θ sin θ , = tan θ , = =
dθ dθ dx dx 2 θ cosθ θ

Then

α = x − tan θ3sec θ = x − θ sin θ = cosθ

2

(sec θ )/θ

and

β = y+ sec 2 θ = y + θ cosθ = sin θ

(sec3 θ )/θ

and = cos , = sin are parametric equations of the evolute (see Fig. 38-8).

Fig. 38 -8

13 . D erive formula (38.1).

tan is the slope of the tangent line and, therefore,

dy
= tan τ . S o, d  dy  = d  dy  ⋅ dτ
dx ds  dx  dτ  dx  ds

H ence

d  dy  ⋅ dx = sec 2 τ ⋅ dτ
dx  dx  ds ds

This y ields

d2y 1   dy  2  dτ
⋅ =  1 +  dx   ⋅ ds
dx 2 2
1+  dy 
 dx 

from which

d2y
dτ = dx 2
ds  2 3/ 2
 dy  
 1 +  dx  
CHAPTER 38 Curvature 31 9

S U PPLEMEN T AR Y PR O B LEMS

In P roblems 14–16, find ds and ds .

dx dy

14 . x2 + y2 = 25 Ans. ds = 5 , ds = 5
dx 25 − x 2 dy 25 − y 2

ds = 4 + 9 y2/3
15 . y2 = x3 Ans. 1
2 4 + 9 x , ds =
dx dy 3y1/ 3

1/ 3
16 . x2/3 + y2/3 = a2/3 Ans. ds = (a / x)1/ 3, ds =  a 
dx dy  y 

In P roblems 17 and 18, find ds .

dx

17 . 6xy = x4 + 3 Ans. ds = x 4 + 1
dx 2x2

18 . 27ay2 = 4(x – a)3 Ans. ds = ( x + 2a) /3a

dx

In P roblems 19–22, find ds.

dt

19 . x = t2, y = t3 Ans. t 4 + 9t 2

2 0 . x = 2 cos t, y = 3 sin t Ans. 4 + 5 cos 2 t

2 1. x = cos t, y = sin t Ans. 1

2 2 . x = cos3 t, y = sin3 t Ans. 3

2 sin 2t

2 3 . Find the curvature of each curve at the given points:

(a) y = x3/3 at x = 0, x = 1, x = –2 (b) x2 = 4ay at x = 0, x = 2a
(c) y = sin x at x = 0, x = 12 π (d) y = e− x at x = 0
2

Ans. (a) 0 , 2 / 2, −4 17 / 289; (b) 1 /2a, 2 / 8a; (c) 0 , −1 ; (d) −2

2 4 . S how (a) the curvature of a straight line is 0; (b) the curvature of a circle is numerically the reciprocal
of its radius.

2 5 . Find the points of maximum curvature of (a) y = ex; (b) y = 13 x 3.

Ans. (a) x = − 12 ln 2; (b) x = 11/ 4

5

2 6 . Find the radius of curvature of

(a) x3 + xy2 − 6y2 = 0 at (3, 3).
(b) x = 2a tan , y = a tan2 at (x, y).
(c) x = a cos4 , y = a sin4 at (x, y).

Ans. (a) 5 5 ; (b) 2a | sec3 θ |; (c) 2a(sin4 + cos4 )3/2

 32 0 CHAPTER 38 Curvature

2 7 . Find the center of curvature of (a) P roblem 26(a); (b) y = sin x at a maximum point.

Ans. ( )
(a) C(−7 , 8 ); (b) C π , 0
2

2 8 . Find the equation of the circle of curvature of the parabola y2 = 12x at the points (0, 0) and (3, 6).

Ans. (x − 6 )2 + y2 = 36 ; (x − 1 5 )2 + ( y + 6 )2 = 28 8

2 9 . Find the equation of the evolute of (a) b2x2 + a2y2 = a2b2; (b) x2/3 + y2/3 + a2/3; (c) x = 2 cos t + cos 2t,
y = 2 sin t + sin 2t.

Ans. (a) (a )2/3 + (b )2/3 = (a2 − b2)2/3; (b) ( + )2/3 + ( − )2/3 = 2a2/3; (c) α = 13 (2 cos t − cos 2t ),
β = 13 (2 sin t − sin 2t )
 C H A P TE R 4 1

Polar Coordinates
The position of a point Ρ in a plane may be described by its coordinates (x, y) with respect to a given rectan-
gular coordinate system. Its position may also be described by choosing a fixed point O and specifying the
directed distance ρ = OΡ and the angle θ that OΡ makes with a fixed half-line OX . (See Fig. 41-1.) This is
the p o lar co o rdinate system . The point O is called the p o le, and OX is called the p o lar axis.

Fig. 4 1-1
To each number pair (ρ, θ) there corresponds one and only one point. The converse is not true. For ex-
ample, (1, 0 ) and (1, 2π) describe the same point, on the polar axis and at a distance 1 from the pole. That
same point also corresponds to (−1, π). (W hen ρ is negative, the point corresponding to (ρ, θ) is obtained as
follows: R otate the polar axis OX through θ radians (counterclockwise if θ is positive and clockwise if θ is
negative) to a new position OX ′ and then move | ρ| units on the half-line opposite to OX ′.)
In general, a point Ρ with polar coordinates (ρ, θ ) also can be described by (ρ, θ ± 2nπ) and (−ρ, θ ± (2n + 1)π),
where n is any nonnegative integer. In addition, the pole itself corresponds to (0 , θ ), with arbitrary θ.
EXAMPLE 4 1.1: In Fig. 41-2, several points and their polar coordinates are shown. N ote that point C has polar co-
( )
ordinates 1, 3π .
2
A p o lar eq uatio n of the form ρ = f (θ ) or F(ρ, θ ) = 0 determines a curve, consisting of those points cor-
responding to pairs (ρ, θ ) that satisfy the eq uation. For example, the eq uation ρ = 2 determines the circle
with center at the pole and radius 2. The eq uation ρ = −2 determines the same set of points. In general, an
eq uation ρ = c, where c is a constant, determines the circle with center at the pole and radius |c|. A n eq uation
θ = c determines the line through the pole obtained by rotating the polar axis through c radians. For example,
θ = π /2 is the line through the pole and perpendicular to the polar axis.
(2, p )
2

(1,3p )
4 (1, p )
4

x
(1, 0)

C (1,3p )
2

Fig. 41-2
339
žŸ ¡¢ £ ¤¥ ¦§¨ ©©ª«¬ªªª«¬ªª©«¬ª­¨® ¡¯¥°±²!³¢ ´µ%¶ ·£ ¸ ¸(žŸ¹% ´º£ °»,«¼ º ²½ž¸ £ ²¾¥°¢ °¿1Ÿ¢2¦1°¢ ¹»3Ÿ¿À»,°½
 340 CHAPTER 41 Polar Coordinates

Polar and Rectangular Coordinates

Given a pole and polar axis, set up a rectangular coordinate system by letting the polar axis be the positive
x axis and letting the y axis be perpendicular to the x axis at the pole. (See Fig. 41-3.) Then the pole is the origin
of the rectangular system. If a point Ρ has rectangular coordinates (x, y) and polar coordinates (ρ, θ ), then

x = ρ cos θ and y = ρ sin θ (41.1)

These equations entail

y
ρ 2 = x 2 + y2 and tan θ = (41.2)
x

Fig. 41-3

E X A M PL E 4 1 .2 : C onsider the polar curve ρ = cos θ.

M ultiplying by ρ, w e get ρ2 = ρ cos θ. H ence, x2 + y2 = x holds for the rectangular coordinates of points on the
curve. That is equivalent to x2 − x + y2 = 0 and completion of the square w ith respect to x yields ( x − 12 )2 + y 2 = 14 .
H ence, the curve is the circle w ith center at ( 12 , 0) and radius 12 . N ote that, as θ varies from 0 to π /2, the upper semi-
circle is traced out from (1, 0) to (0, 0), and then, as θ varies from π to π, the low er semicircle is traced out from (0, 0)
2
back to (1, 0). This w hole path is retraced once more as θ varies from π to 2π. Since cos θ has a period of 2π, w e have
completely described the curve.

E X A M PL E 4 1 .3 : C onsider the parabola y = x2. In polar coordinates, w e get ρ sin θ = ρ2 cos2 θ, and, therefore,
ρ = tan θ sec θ, w hich is a polar equation of the parabola.

S om e T y p ical Polar Curv es

(a) C ardioid: ρ = 1 + sin θ. See Fig. 41-4(a).
(b) L imaç on: ρ = 1 + 2 cos θ. See Fig. 41-4(b).
(c) R ose w ith three petals: ρ = cos 3θ. See Fig. 41-4(c).
(d) L emniscate: ρ2 = cos 2θ. See Fig. 41-4(d).
A t a point Ρ on a polar curve, the angle ψ from the radius vector OΡ to the tangent Ρ T to the curve (see
Fig. 41-5) is given by

dθ ρ dρ
tan ψ = ρ = , w here ρ′ = (41.3)
d ρ ρ′ dθ

For a proof of this equation, see P roblem 1. Tan ψ plays a role in polar coordinates similar to that of the
slope of the tangent line in rectangular coordinates.
CHAPTER 41 Polar Coordinates 341

Fig. 41-4

A ngle of Inclination
The angle of inclination τ of the tangent line to a curve at a point Ρ (ρ, θ) on it (see Fig. 41-5) is given by

ρ cos θ + ρ ′ sin θ
tan τ = (41.4)
− ρ sin θ + ρ ′ cos θ

For a proof of this equation, see P roblem 4.

Points of Intersection
Some or all of the points of intersection of tw o polar curves ρ = f1(θ ) and ρ = f2(θ) (or equivalent equations)
may be found by solving

f1 (θ ) = f2 (θ ) (41.5)
 342 CHAPTER 41 Polar Coordinates

Fig. 41-5

E X A M PL E 4 1 .4 : Find the points of intersection of ρ = 1 + sin θ and ρ = 5 − 3 sin θ.

Setting 1 + sin θ = 5 − 3 sin θ, w e obtain sin θ = 1. Then ρ = 2 and θ = π /2. The only point of intersection is
(2, π /2). N ote that w e need not indicate the infinite number of other pairs that designate the same point.
Since a point may be represented by more than one pair of polar coordinates, the intersection of tw o curves
may contain points that no single pair of polar coordinates satisfies (41.5).

E X A M PL E 4 1 .5 : Find the points of intersection of ρ = 2 sin 2θ and ρ = 1.

Solution of the equation 2 sin 2θ = 1 yields sin 2θ = 12 , and, therefore, w ithin [0, 2π), θ = π /12, ⋅5π /12, 13π /12,
17π /12. W e have found four points of intersection: (1, π /12), (1, 5π /12), (1, 13π /12), and (1, 17π /12). B ut the circle
ρ = 1 also can be represented as ρ = −1. N ow solving 2 sin 2θ = −1, w e get sin 2θ = − 12 and, therefore, θ = 7π /12,
11π /12, 19π /12, and 23π /12. H ence w e get four more points of intersection (−1, 7π /12), (−1, 11π /12), (−1, 19π /12),
and (−1, 23π /12).
W hen the pole is a point of intersection, it may not appear among the solutions of (41.5). The pole is a
point of intersection w hen there exist θ1 and θ2 such that f1(θ1) = 0 = f2(θ2).

E X A M PL E 4 1 .6 : Find the points of intersection of ρ = sin θ and ρ = cos θ.

From the equation sin θ = cos θ, w e obtain the points of intersection ( 2 / 2, π /4) and ( − 2 / 2, 5π /4). H ow ever, both
curves contain the pole. O n ρ = sin θ, the pole has coordinates (0, 0), w hereas, on ρ = cos θ, the pole has coordinates
(0, π /2).

E X A M PL E 4 1 .7 : Find the points of intersection of ρ = cos 2θ and ρ = cos θ.

Setting cos 2θ = cos θ and noting that cos 2θ = 2 cos2 θ − 1, w e get 2 cos2 θ − cos θ − 1 = 0 and, therefore, (cos θ − 1)
(2 cos θ + 1) = 0. So, cos θ = 1 or cosθ = − 12 . Then θ = 0, 2π /3, 4π /3, yielding points of intersection (1, 0), (− 12 , 2π /3),
and ( − 12 , 4π /3). B ut the pole is also an intersection point, appearing as (0, π /4) on ρ = cos 2θ and as (0, π /2) on ρ = cos θ.

A ngle of Intersection
The angle of intersection, φ , of tw o curves at a common point Ρ (ρ, θ), not the pole, is given by
tan ψ 1 − tan ψ 2
tan φ = (41.6)
1 + tan ψ 1 tan ψ 2

w here ψ1 and ψ2 are the angles from the radius vector OΡ to the respective tangent lines to the curves at Ρ.
(See Fig. 41-6.) This formula follow s from the trigonometric identity for tan(ψ1 − ψ2), since φ = ψ 1 − ψ 2 .

Fig. 41-6
CHAPTER 41 Polar Coordinates 343

E X A M PL E 4 1 .8 : Find the (acute) angles of intersection of ρ = cos 2θ and ρ = cos θ.

The points of intersection w ere found in E xample 7. W e also need tan ψ1 and tan ψ2. For ρ = cos θ, formula (41.3)
yields tan ψ1 = − cot θ. For ρ = cos 2θ, formula (41.3) yields tan ψ 2 = − 12 cot 2θ .
A t the point (1, 0), tan ψ1 = − cot 0 = ∞ and, lik ew ise, tan ψ2 = ∞. Then ψ1 = ψ2 = π /2 and, therefore, φ = 0.

( )
A t the point − 1 , 2π , tan ψ 1 = 3 / 3 and tan ψ 2 = − 3 / 6 . So, by (41.6),
2 3

( 3 / 3) + ( 3 / 6) 3 3
tan φ = =
1 − (1/ 6) 5

and, therefore, the acute angle of intersection φ ≈ 46° 6′ . B y symmetry, this is also the acute angle of intersection at
the point (− 12 , 4π /3).
A t the pole, on ρ = cos θ, the pole is given by θ = π /2. O n ρ = cos 2θ, the pole is given by θ = π /4 and θ = 3π /4.
Thus, at the pole there are tw o intersections, the acute angle being π /4 for each.

T h e D eriv ativ e of th e A rc L ength

The derivative of the arc length is given by
ds
= ρ 2 + ( ρ ′) 2 (41.7)
dθ
dρ
w here ρ ′ = and it is understood that s increases w ith θ.
dθ
For a proof, see P roblem 20.

Curv ature
The curvature of a polar curve is given by
ρ 2 + 2(ρ ′)2 − ρρ ′′
K= (41.8 )
[ ρ 2 + ( ρ ′) 2 ] 3 / 2
For a proof, see P roblem 17.

S O L V E D PRO B L E M S
ρ dρ
1. D erive formula (41.3): tan ψ = ρ dθ = , w here ρ ′ .
dρ ρ′ dθ
In Fig. 41-7, Q(ρ + ∆ρ, θ + ∆θ) is a point on the curve near P. From the right triangle PS Q,

ρ sin ∆θ ρ sin ∆θ ρ sin ∆θ

tan λ = S P = SP = = = ∆θ
S Q OQ − OS ρ + ∆ρ − ρ cos ∆θ ρ(1 − cos ∆θ ) + ∆ρ
ρ 1 − cos ∆θ + ∆ρ
∆θ ∆θ
N ow as Q → P along the curve, ∆θ → 0, OQ → OP, PQ → PT, and ∠ λ → ∠ ψ .

Fig. 41-7
 344 CHAPTER 41 Polar Coordinates

A s ∆θ → 0, sin ∆θ → 1 and 1 − cos ∆θ → 0. Thus,

∆θ ∆θ
ρ
tan ψ = lim tan λ = = ρ dθ
∆θ → 0 dp /dθ dρ

In P roblems 2 and 3, use formula (41.3) to find tan ψ for the given curve at the given point.

2. ρ = 2 + cosθ at θ = π . (See Fig. 41-8 .)

3 ρ
A t θ = π , ρ = 2 + 12 = 52 , ρ ′ = − sin θ = −
3
, and tan ψ = =− 5 .
3 2 ρ′ 3

Fig. 41-8

3. ρ = 2 sin 3θ at θ = π . (See Fig. 41-9.)

4
ρ
A t θ = π , ρ = 2 1 = 2 , ρ ′ = 6 cos 3θ = 6  − 1  = −3 2 and tan ψ = = − 13 .
4 2  2 ρ ′

Fig. 41-9

ρ cosθ + ρ ′ sin θ
4. D erive formula (41.4): tan τ = .
− ρ sin θ + ρ cosθ
From Fig. 41-7, τ = ψ + θ and

ρ dθ + sin θ
tan ψ + tan θ d ρ cosθ
tan τ = tan(ψ + θ ) = =
1 − tan ψ tan θ 1 − ρ dθ sin θ
d ρ cosθ
dρ
ρ cosθ + sin θ ρ cosθ + ρ ′ sin θ
= dθ =
dρ − ρ sin θ + ρ ′ cosθ
cosθ − ρ sin θ
dθ
5. Show that, if ρ = f (θ) passes through the pole and θ1 is such that f (θ1) = 0, then the direction of the tangent line to
the curve at the pole (0, θ1) is θ1. (See Fig. 41-10.)

Fig. 41-10
CHAPTER 41 Polar Coordinates 345

A t (0, θ1), ρ = 0, and ρ′ = f ′(θ1). If ρ′ ≠ 0

ρ cosθ + ρ ′ sin θ 0 + f ′(θ1 )sin θ1

tan τ = = = tan θ1
− ρ sin θ + ρ ′ cosθ 0 + f ′(θ1 )cosθ1

If ρ′ = 0,

f ′(θ )sin θ
tan τ = lim = tan θ1
θ →θ1 f ′(θ )cosθ

In P roblems 6– 8 , find the slope of the given curve at the given point.

6. ρ = 1 − cos θ at θ = π . (See Fig. 41-11.)

2

Fig. 41-11

A tθ = π,
2

sin θ = 1, cosθ = 0, ρ = 1, ρ ′ = sin θ = 1

ρ cosθ + ρ ′ sin θ
and tan τ = = 1 ⋅ 0 + 1 ⋅ 1 = −1
− ρ sin θ + ρ ′ cosθ −1 ⋅ 1 + 1 ⋅ 0

7. ρ = cos 3θ at the pole. (See Fig. 41-12.)

W hen ρ = 0, cos 3θ = 0. Then 3θ = π /2, 3π /2, 5π /2, and θ = π /6, π /2, 5π /6. B y P roblem 5, tan τ = 1/ 3 , ∞,
and −1 3 .

Fig. 41-12

8. ρθ = a at θ = π .
3
A t θ = π /3: sin θ = 3 / 2 , cosθ = 12 , ρ = 3a/π, and ρ′ = −a/θ 2 = −9a/π 2. Then

ρ cosθ + ρ ′ sin θ π −3 3
tan τ = =−
− ρ sin θ + ρ ′ cosθ 3π + 3
 346 CHAPTER 41 Polar Coordinates

9. Investigate ρ = 1 + sin θ for horiz ontal and vertical tangents. (See Fig. 41-13.)

Fig. 41-13

A t P(ρ, θ):

(1 + sin θ )cosθ + cosθ sin θ cosθ (1 + 2 sin θ )

tan τ = =−
−(1 + sin θ )sin θ + cos 2 θ (sin θ + 1)(2 sin θ − 1)

W e set cos θ (1 + 2 sin θ) = 0 and solve, obtaining θ = π /2, 3π /2, 7π /6, and 11π /6. W e also set (sin θ + 1)
(2 sin θ − 1) = 0 and solve, obtaining θ = 3π /2, π /6, and 5π /6.
For θ = π /2: There is a horiz ontal tangent at (2, π /2).
For θ = 7π /6 and 11π /6: There are horiz ontal tangents at ( 12 , 7π /6) and ( 12 , 11π /6).
For θ = π /6 and 5π /6: There are vertical tangents at ( 32 , π /6) and ( 32 , 5π /6).
For θ = 3π /2: B y P roblem 5, there is a vertical tangent at the pole.

1 0 . Show that the angle that the radius vector to any point of the cardioid ρ = a(1 − cos θ) mak es w ith the curve is
one-half that w hich the radius vector mak es w ith the polar axis.
A t any point P( ρ, θ ) on the cardioid,

ρ 1 − cosθ
ρ ′ = asin θ and tan ψ = = = tan θ .
ρ′ sin θ 2

So ψ = 12 θ .

In P roblems 11– 13, find the angles of intersection of the given pair of curves.

1 1 . ρ = 3 cos θ, ρ = 1 + cos θ. (See Fig. 41-14.)

Fig. 41-14

Solve 3 cos θ = 1 + cos θ for the points of intersection, obtaining (3/2, π /3) and (3/2, 5π /3). The curves also
intersect at the pole.
For ρ = 3 cos θ : ρ ′ = −3 sin θ and tan ψ 1 = − cot θ
For ρ = 1 + cos θ : ρ ′ = − sin θ and tan ψ 2 = − 1 + cosθ
sin θ
CHAPTER 41 Polar Coordinates 347

A t θ = π /3, tan ψ 1 = −1 3 , tan ψ 2 = − 3 , and tan φ = 1/ 3 . The acute angle of intersection at ( 32 , π /3) and, by
symmetry, at ( 32 , 5π /3) is π /6.
A t the pole, either a diagram or the result of P roblem 5 show s that the curves are orthogonal.

1 2 . ρ = sec 2 12 θ , ρ = 3 csc 2 12 θ .
Solve sec 2 12 θ = 3 csc 2 12 θ for the points of intersection, obtaining (4, 2π /3) and (4, 4π /3).
For ρ = sec 2 12 θ : ρ ′ = sec 2 12 θ tan 12 θ and tan ψ 1 = cot 12 θ
For ρ = 3 csc 2 12 θ : ρ ′ = −3 csc 2 12 θ cot 12 θ and tan ψ 2 = − tan 12 θ
A t θ = 2π /3, tan ψ 1 = 1/ 3 , and tan π 2 = − 3 , and φ = 12 π ; the curves are orthogonal. L ik ew ise, the curves are
orthogonal at θ = 4π /3.

1 3 . ρ = sin 2θ, ρ = cos θ. (See Fig. 41-15.)

The curves intersect at the points ( 3 / 2 , π /6) and ( − 3 /2 , 5π /6) and the pole.
For ρ = sin 2θ : ρ ′ = 2 cos 2θ and tan ψ 1 = 12 tan 2θ
For ρ = cos θ : ρ ′ = − sin θ and tan ψ 2 = − cot θ

A t θ = π /6, tan ψ 1 = 3 / 2 , tan ψ 2 = − 3 , and tan φ1 = −3 3 . The acute angle of intersection at the point ( 3 / 2 ,
π /6) is φ = tan −1 3 3 = 79° 6′. Similarly, at θ = 5π /6, tan ψ 1 = − 3 / 2 , tan ψ 2 = 3 , and the angle of intersection
is tan −1 3 3 .
A t the pole, the angles of intersection are 0 and π /2.

Fig. 41-15

ds
In P roblems 14– 16, find at the point P (ρ, θ ).
dθ

1 4 . ρ = cos 2θ.

ρ ′ = −2 sin 2θ and ds = ρ 2 + ( ρ ′)2 = cos 2 2θ + 4 sin 2 2θ = 1 + 3 sin 2 2θ

dθ

1 5 . ρ(1 + cos θ ) = 4.
D ifferentiation yields −ρ sin θ + ρ′(1 + cos θ ) = 0. Then
ρ sin θ
ρ′ = = 4 sin θ and ds = ρ 2 + ( ρ ′)2 = 4 2
1 + cosθ (1 + cosθ )2 dθ (1 + cosθ )3/ 2
 348 CHAPTER 41 Polar Coordinates

()
1 6 . ρ = sin 3 θ . (A lso evaluate ds at θ = π .)
3 dθ 2

ρ ′ = sin 2 13 θ cos 13 θ and ds = sin 6 1 θ + sin 4 1 θ cos 2 1 θ = sin 2 1 θ

dθ 3 3 3 3

A t θ = 12 π , ds /dθ = sin 2 16 π = 14 .

ρ 2 + 2(ρ ′)2 − ρρ ′′
1 7 . D erive formula (41.8 ): K = .
[ ρ 2 + ( ρ ′) 2 ] 3 / 2
B y definition, K = dτ . N ow , τ = θ + ψ and, therefore,
ds

dτ = dθ + dψ = dθ + dψ dθ = dθ  1 + dψ 
ds ds ds ds dθ ds ds  dθ 

 ρ
w here ψ = tan −1   . A lso,
 ρ′ 
dψ [( ρ ′)2 − ρρ ′′] / ( ρ ′)2 ( ρ ′)2 − ρρ ′′ dψ ( ρ ′)2 − ρρ ′′ ρ 2 + 2( ρ ′)2 − ρρ ′′
= = 2 ; so 1+ =1+ 2 =
dθ 1 + ( ρ /ρ ′)2 ρ + ( ρ ′) 2 dθ ρ + ( ρ ′) 2 ρ 2 + ( ρ ′) 2

 dψ  1 + dψ /dθ 1 + dψ /dθ ρ 2 + 2(ρ ′)2 − ρρ ′′

K = dθ  1 + = = =
dθ 
Thus,
ds  ds /dθ ρ + ( ρ ′)
2 2 [ ρ 2 + ( ρ ′) 2 ] 3 / 2

1 8 . L et ρ = 2 + sin θ. Find the curvature at the point P(ρ, θ)

ρ 2 + 2( ρ ′)2 − ρρ ′′ (2 + sin θ )2 + 2 cos 2 θ + (sin θ )(2 + sin θ ) 6(1 + sin θ )

K= = =
[ ρ 2 + ( ρ ′) 2 ] 3 / 2 [(2 + sin θ )2 + cos 2 θ ] 3/ 2 (5 + 4 sin θ )3/ 2

1 9 . L et ρ(1 − cos θ) = 1. Find the curvature at θ = π and θ = 4π .

2 3

ρ′ = − sin θ and ρ ′′ = − cosθ + 2 sin 2 θ ; so K = sin 3 θ

(1 − cosθ )2 (1 − cosθ )2 (1 − cosθ )3 2

A t θ = π /2, K = (1/ 2 )3 = 2 / 4 ; at θ = 4π /3, K = ( 3 / 2)3 = 3 3 /8 .

2 0 . D erive formula (41.7): ds = ρ 2 + ( ρ ′)2 .

dθ
C onsider ρ as a function of θ. From x = ρ cos θ and y = ρ sin θ, w e get dx/dθ = −ρ sin θ + (cos θ )ρ′ and dy/dθ =
ρ cos θ + (sin θ)ρ′. H ence,

( ddxθ ) = [ ρ sin θ + (ρ′) cos θ − 2ρρ′ sinθ cosθ ]

2
2 2 2 2

2
and  dy  = [ ρ 2 cos 2 θ + ( ρ ′)2 sin 2 θ + 2 ρρ ′ sin θ cosθ ]
 dθ 

( ddsθ ) = ( ddxθ ) +  ddyθ 

2 2 2
Thus, = ρ 2 + ( ρ ′) 2

Since s increases w ith θ, ds > 0 and w e obtain formula (41.7).

dθ

2 1 . For ρ = cos 2θ, find ds at θ = π . (A ssume as usual that s increases w ith θ.)
dθ 4
dρ
ρ′ = = −2 sin 2θ . B y Formula (41.7),
dθ
ds = cos 2 (2θ ) + 4 sin 2 (2θ ) = 1 + 3 sin 2 (2θ )
dθ

= 1 + 3 sin 2 (π /2) = 2
CHAPTER 41 Polar Coordinates 349

S U PPL E M E N T A RY PRO B L E M S

In P roblems 22– 25, find tan ψ for the given curve at the given points.

2 2 . ρ = 3 − sin θ at θ = 0, θ = 3π /4 Ans. −3; 3 2 − 1

2 3 . ρ = a(1 − cos θ) at θ = π /4, θ = 3π /2 Ans. 2 − 1; −1

2 4 . ρ(1 − cos θ ) = a at θ = π /3, θ = 5π /4 Ans. − 3 / 3; 1 + 2

2 5 . ρ2 = 4 sin 2θ at θ = 5π /12, θ = 2π /3 Ans. −1 3 ; 3

In P roblems 26– 29, find tan τ for the given curve at the given point.

2 6 . ρ = 2 + sin θ at θ = π /6 Ans. −3 3

2 7 . ρ2 = 9 cos 2θ ; at θ = π /6 Ans. 0

2 8 . ρ = sin3 (θ/3) at θ = π /2 Ans. − 3

2 9 . 2ρ(1 − sin θ) = 3 at θ = π /4 Ans. 1+ 2

3 0 . Investigate ρ = sin 2θ for horiz ontal and vertical tangents.

Ans. horiz ontal tangents at θ = 0, π, 54° 44′, 125° 16′, 234° 44′, 305° 16′; vertical tangents at θ = π /2, 3π /2,
35° 16′, 144° 44′, 215° 16′, 324° 44′

In P roblems 31– 33, find the acute angles of intersection of each pair of curves.

3 1 . ρ = sin θ, ρ = sin 2θ Ans. φ = 79° 6 ′ at θ = π /3 and 5π /3; φ = 0 at the pole

3 2 . ρ = 2 sin θ , ρ2 = cos 2θ Ans. φ = π /3 at θ = π /6, 5π /6; φ = π /4 at the pole

3 3 . ρ2 = 16 sin 2θ, ρ2 = 4 csc 2θ Ans. φ = π /3 at each intersection

3 4 . Show that each pair of curves intersects at right angles at all points of intersection.
(a) ρ = 4 cos θ, p = 4 sin θ (b) ρ = eθ, ρ = e−θ
(c) ρ2 cos 2θ = 4, ρ2 sin 2θ = 9 (d) ρ = 1 + cos θ, ρ = 1 − cos θ

3 5 . Find the angle of intersection of the tangents to ρ = 2 − 4 sin θ at the pole.

Ans. 2π /3
 35 0 CHAPTER 41 Polar Coordinates

3 6 . Find the curvature of each of these curves at P(ρ, θ): (a) ρ = eθ; (b) ρ = sin θ ; (c) ρ2 = 4 cos 2θ; (d) ρ = 3 sin θ +
4 cos θ.

Ans. (a) 1/( 2 eθ ) ; (b) 2; (c) 3

2 cos 2θ ; (d) 2
5

3 7 . Find ds for the curve ρ = a cos θ.

dθ

Ans. a

3 8 . Find ds for the curve ρ = a(1 + cos θ).

dθ

Ans. a 2 + 2 cosθ

3 9 . Suppose a particle moves along a curve ρ = f (θ) w ith its position at any time t given by ρ = g(t), θ = h(t).

( ) ( )
(a) M ultiply the equation ds = ρ 2 + ( ρ ′)2 obtained in P roblem 20 by dθ to obtain
2 2

dθ dt

( ) ( ) ρ
2
ds
2
dθ
2
 d 
v2 = = ρ2 +  .
dt dt  dt 
θ ρ dρ
(b) From tan ψ = ρ d = ρ θ /dt , obtain sin ψ = dθ and cosψ = 1
d .
dρ d ρ /dt v dt v dt

In P roblems 40– 43, find all points of intersection of the given equations.

4 0 . ρ = 3 cos θ, ρ = 3 sin θ Ans. (0, 0), ( 3 2 / 2, π /4)

4 1 . ρ = cos θ, ρ = 1 − cos θ Ans. (0, 0), ( 12 , π /3), ( 12 , −π /3)

4 2 . ρ = θ, ρ = π Ans. (π, π), (−π, −π)

 (2n + 1)π 
4 3 . ρ = sin 2θ, ρ = cos 2θ Ans. (0, 0),  2 ,  for n = 0, 1, 2, 3, 4, 5
 2 6

4 4 . (GC ) Sk etch the curves in P roblems 40– 43, find their graphs on a graphing calculator, and check your answ ers to
P roblems 40– 43.

4 5 . (GC ) Sk etch the graphs of the follow ing equations and then check your answ ers on a graphing calculator:
(a) ρ = 2 cos 4θ (b) ρ = 2 sin 5θ (c) ρ2 = 4 sin 2θ
(d) ρ = 2(1 − cos θ) (e) ρ = 2 (f ) ρ 2 = 1
1 + cosθ θ
(g) ρ = 2 − sec θ (h) ρ = 2
θ
(In parts (g) and (h), look for asymptotes.)

4 6 . C hange the follow ing rectangular equations to polar equations and sk etch the graphs:
(a) x2 − 4x + y2 = 0 (b) 4x = y2 (c) xy = 1
(d) x = a (e) y = b (f ) y = mx + b

Ans. (a) ρ = 4 cos θ ; (b) ρ = 4 cot θ csc θ ; (c) ρ2 = sec θ csc θ ; (d) ρ = a sec θ ; (e) ρ = b csc θ ;
(f ) ρ = b
sin θ − m cosθ
CHAPTER 41 Polar Coordinates 35 1

4 7 . (GC ) C hange the follow ing polar equations to rectangular coordinates and then sk etch the graph. (V erify on a
graphing calculator.) (a) ρ = 2c sin θ ; (b) ρ = θ ; (c) ρ = 7 sec θ

Ans. (a) x2 + ( y − c)2 = c2; (b) y = x tan( x 2 + y 2 ) ; (c) x = 7

4 8 . (a) Show that the distance betw een tw o points w ith polar coordinates (ρ1, θ1) and (ρ2, θ2) is

ρ12 + ρ22 − 2 ρ1ρ2 cos(θ1 − θ 2 )

(b) W hen θ1 = θ2, w hat does the distance simplify to? E xplain w hy this is so.

Ans. |ρ1 − ρ2|

(c) W hen θ1 − θ 2 = π , w hat does the formula yield? E xplain the signficance of the result.
2

Ans. ρ12 + ρ22

( )
(d) Find the distance betw een the points w ith the polar coordinates (1, 0) and 1, π .
4

Ans. 2− 2

4 9 . (a) L et f be a continuous function such that f (θ) ≥ 0 for α < θ < β. L et A be the area of the region bounded by
β β
the lines θ = α and θ = β, and the polar curve ρ = f (θ). D erive the formula A = 1 ∫ ( f (θ ))2 dθ = 1 ∫ ρ 2 dθ .
2 α 2 α
(H int: D ivide [α, β] into n equal parts, each equal to ∆θ. E ach resulting subregion has area approximately
equal to 12 ∆θ ( f (θi* ))2 , w here θi* is in the ith subinterval.)
(b) Find the area inside the cardioid ρ = 1 + sin θ.
π π
(c) Find the area of one petal of the rose w ith three petals, ρ = cos 3θ. ( H int: Integrate from − 6 to 6 .)
 C H A P TE R 4 2

Infinite Sequences
Infinite S eq uences
An infinite sequence 〈 sn 〉 is a function whose domain is the set of positive integers; sn is the value of this
function for a given positive integer n. Sometimes we indicate 〈 sn 〉 just by writing the first few terms of the
sequence s1, s2, s3, . . . , sn, . . . . We shall consider only sequences where the values sn are real numbers.

E X A M PL E 4 2 .1 :
(a) 1 is the sequence 1, 1 , 1 , . . . , 1 , . . ..
n 2 3 n

( 12 )
n
(b) is the sequence 1 , 1 , 1 , . . . , 1n , . . . .
2 4 8 2
(c) 〈n 2 〉 is the sequence of squares 1, 4, 9 , 16 , . . ., n2, . . . .
(d) 〈2n 〉 is the sequence of positive even integers 2, 4, 6 , 8, . . . , 2n, . . ..
(e) 〈2n − 1〉 is the sequence of positive odd integers 1, 3, 5 , 7 , . . ..

L im it of a S eq uence
If 〈 sn 〉 is an infinite sequence and L is a number, then we say that lim sn = L if sn gets arbitrarily close to L
n→+∞
as n increases without bound.
F rom a more precise standpoint, lim sn = L means that, for any positive real number  > 0 , there ex ists
n→+∞
a positive integer n0 such that, whenever n ≥ n0, we have |sn − L| < . T o illustrate what this means, place
the points L, L − , and L +  on a coordinate line (see F ig. 42-1), where  is some positive real number.
N ow, if we place the points s1, s2, s3, . . . on the coordinate line, there will eventually be an index n0 such that
sn0, sn0 +1, sn0 + 2 , sn0 +3 , . . . and all subsequent terms of the sequence will lie inside the interval ( L − , L + ).

s1 s2 sm sm + 1

L–  L L+ 

Fig. 42-1

If lim sn = L , then we say that the sequence 〈 sn 〉 co nv erg es to L. If there is a number L such that 〈 sn 〉
n→+∞
converges to L, then we say that 〈 sn 〉 is co nv erg ent. When 〈 sn 〉 is not convergent, then we say that 〈 sn 〉 is
div erg ent.

E X A M PL E 4 2 .2 : 1 is convergent, since lim 1 = 0. T o see this, observe that 1/n can be made arbitrarily close to
n n→+∞ n
0 by mak ing n large enough. T o get an idea of why this is so, note that 1/10 = 0.1, 1/100 = 0.01, 1/1000 = 0.001, and
so on. T o check that the precise definition is satisfied, let  be any positive number. T ak e n0 to be the smallest positive
integer greater than 1/. So, 1/ < n0. H ence, if n ≥ n0, then n > 1/  and, therefore, 1/n < . T hus, if n ≥ n0, |1/n − 0| < .
T his proves lim 1 = 0 .
n→+∞ n

E X A M PL E 4 2 .3 : 〈2n 〉 is a divergent sequence, since lim 2n ≠ L for each real number L. In fact, 2n gets arbitrarily
n→+∞
large as n increases.

35 2
ÁÂÃÄÅ Æ ÇÈ ÉÊË ÌÌÍÎÏÍÍÍÎÏÍÍÌÎÏÍÐËÑ ÄÒÈÓÔÕ!ÖÅ ×Ø%Ù ÚÆ Û Û(ÁÂÜ%Ã×ÝÆ ÓÞ,Îß Ý ÕàÁÛ Æ ÕáÈÓÅ Óâ1ÂÅ2É1ÓÅ ÜÞ3ÂâãÞ,Óà
CHAPTER 42 Infinite S equences 35 3

We write lim sn = +∞ if sn gets arbitrarily large as n increases. In such a case, we say that 〈 sn 〉 diverges to
n→+∞
+∞. M ore precisely, lim sn = +∞ if and only if, for any number c, no matter how large, there ex ists a positive
n→+∞
integer n0 such that, whenever n ≥ n0, we have sn > c.
L ik ewise, we write lim sn = −∞ if sn gets arbitrarily small as n increases. In such a case, we say that 〈 sn 〉
n→+∞
diverges to −∞. M ore precisely, lim sn = −∞ if and only if, for any number c, no matter how small, there
n→+∞
ex ists a positive integer n0 such that, whenever n ≥ n0, we have sn < c.
We shall write lim sn = ∞ if lim | sn| = +∞ , that is, the magnitude of sn gets arbitrarily large as n increases.
n→+∞ n→+∞

E X A M PL E 4 2 .4 : (a) lim 2n = +∞ ; (b) lim (1 − n)3 = −∞ ; (c) lim (−1)n (n 2 ) = ∞ . N ote that, in case (c), the sequence
n→+∞ n→+∞ n→+∞
converges neither to +∞ nor to −∞.

E X A M PL E 4 2 .5 : T he sequence 〈(−1)n 〉 is divergent, but it diverges neither to +∞, nor to −∞, nor to ∞. Its values
oscillate between 1 and −1.
A sequence 〈 sn 〉 is said to be bo unded abo v e if there is a number c such that sn ≤ c for all n, and 〈 sn 〉 is said
to be bo unded belo w if there is a number b such that b ≤ sn for all n. A sequence 〈 sn 〉 is said to be bo unded if
it is bounded both above and below. It is clear that a sequence 〈 sn 〉 is bounded if and only if there is a number
d such that |sn | ≤ d for all n.

E X A M PL E 4 2 .6 : (a) T he sequence 〈2n 〉 is bounded below (for ex ample, by 0) but is not bounded above. (b) T he
sequence 〈(−1)n 〉 is bounded. N ote that 〈(−1)n 〉 is −1, 1, −1, . . . So, |(−1)n| ≤ 1 for all n.

T h eorem 4 2 .1 : E very convergent sequence is bounded.

F or a proof, see P roblem 5 .
T he converse of T heorem 42.1 is false. F or ex ample, the sequence 〈(−1)n 〉 is bounded but not convergent.
Standard arithmetic operations on convergent sequences yield convergent sequences, as the following
intuitively obvious results show.

T h eorem 4 2 .2 : Assume lim sn = c and lim tn = d . T hen:

n→+∞ n→+∞

(a) lim k = k , where k is a constant.

n→+∞
(b) lim ksn = kc , where k is a constant.
n→+∞
(c) lim (sn + tn ) = c + d .
n→+∞
(d) lim (sn − tn ) = c − d .
n→+∞
(e) lim (sn tn ) = cd .
n→+∞
(f ) lim (sn /tn ) = c /d provided that d ≠ 0 and tn ≠ 0 for all n.
n→+∞

F or proofs of parts (c) and (e), see P roblem 10.

T he following facts about sequences are intuitively clear.

T h eorem 4 2 .3 : If lim sn = ∞ and sn ≠ 0 for all n, then lim 1 = 0 .

n→+∞ n→+∞ sn

F or a proof, see P roblem 7 .

T h eorem 4 2 .4 :
(a) If |a| > 1, then lim a n = ∞ .
n→+∞

In particular, if a > 1, then lim a n = +∞ .

n→+∞
(b) If |r | < 1, then lim r n = 0 .
n→+∞

F or proofs, see P roblem 8.

T h eorem 4 2 .5 (S q ueez e T h eorem ): If lim sn = L = lim un , and there is an integer m such that sn ≤ tn ≤ un for all n ≥ m,
n→+∞ n→+∞
then lim tn = L .
n→+∞

F or a proof, see P roblem 11.

 35 4 CHAPTER 42 Infinite S equences

Corollary 4 2 .6 : If lim un = 0 and there is an integer m such that |tn| ≤ |un| for all n ≥ m, then lim t n = 0.
n→+∞ n→+∞

T his is a consequence of T heorem 42.5 and the fact that lim an = 0 is equivalent to lim |an| = 0 .
n→+∞ n→+∞

E X A M PL E 4 2 .7 : lim (−1)n 12 = 0. T o see this, use C orollary 42.6 , noting that (−1)n 12 ≤ 1 and lim 1 = 0 .
n→+∞ n n n n→+∞ n

T h eorem 4 2 .7 : Assume that f is a function that is continuous at c, and assume that lim sn = c , where all the terms
n→+∞
sn are in the domain of f. T hen lim f (sn ) = f (c) .
n→+∞

See P roblem 33.

It is clear that whether or not a sequence converges would not be affected by deleting, adding, or altering a finite
number of terms at the beginning of the sequence. C onvergence depends on what happens “ in the long run.”
We shall ex tend the notion of infinite sequence to the case where the domain of a sequence is allowed to
be the set of nonnegative integers or any set consisting of all integers greater than or equal to a fix ed integer.
F or ex ample, if we tak e the domain to be the set of nonnegative integers, then 〈2n + 1〉 would denote the
sequence of positive odd integers, and 〈1/ 2n 〉 would denote the sequence 1, 12 , 14 , 81 , . . ..

M onotonic S eq uences
(a) A sequence 〈 sn 〉 is said to be no ndecreasing if sn ≤ sn+ 1 for all n.
(b) A sequence 〈 sn 〉 is said to be increasing if sn < sn+ 1 for all n.
(c) A sequence 〈 sn 〉 is said to be no nincreasing if sn ≥ sn+ 1 for all n.
(d) A sequence 〈 sn 〉 is said to be decreasing if sn > sn+ 1 for all n.
(e) A sequence is said to be mo no to nic if it is either nondecreasing or nonincreasing.
C learly, every increasing sequence is nondecreasing (but not conversely), and every decreasing sequence
is nonincreasing (but not conversely).

E X A M PL E 4 2 .8 : (a) T he sequence 1, 1, 2, 2, 3, 3, 4, 4, . . . is nondecreasing, but not increasing. (b) −1, −1, −2, −2,
−3, −3, − 4, − 4, . . . is nonincreasing, but not decreasing.
An important basic property of the real number system is given by the following result. Its proof is beyond
the scope of this book .

T h eorem 4 2 .8 : E very bounded monotonic sequence is convergent.

T here are several methods for showing that a given sequence 〈 sn 〉 is nondecreasing, increasing, nonin-
creasing, or decreasing. L et us concentrate on the property that 〈 sn 〉 is increasing.
M eth o d 1: S h o w th at sn+ 1 − sn > 0.

E X A M PL E 4 2 .9 : C onsider sn = 3n . T hen s = 3(n + 1) = 3n + 3 . So,

4n + 1 n+1 4(n + 1) + 1 4 n + 5

(12n 2 + 15 n + 3) − (12n 2 + 15 n)
sn+1 − sn = 3n + 3 − 3n =
4n + 5 4n + 1 (4 n + 5 )(4 n + 1)

= 3 >0
(4 n + 5 )(4 n + 1)

since 4n + 5 > 0 and 4n + 1 > 0

Method 2: When all sn > 0, show that sn+1/sn > 1.

EXAMPLE 42.10: Using the same example sn = 3n as ab o v e,

4n + 1

sn+1
sn (
= 3n + 3
4n + 5 ) ( 4n3n+ 1 ) = 3n3+n 3 44nn ++ 15 = 12n12n n+ 15+ 15n n+ 3 > 1,
2
2

since 12n2 + 15n + 3 > 12n2 + 15n > 0.

CHAPTER 42 Infinite Sequences 355

Method 3: F ind a differentiable function f (x) such that f (n) = sn for all n, and show that f ′(x) > 0 for all x ≥ 1
(and, hence, that f is an increasing function for x ≥ 1).

EXAMPLE 42.11: C o nsider sn = 3n again. L et f ( x) = 3x . T hen f ′( x) = 3 > 0 fo r all x.

4n + 1 4x + 1 (4 x + 1)2

S O LV ED PR O B LEMS

1. F o r each o f the fo llo w ing seq u ences, w rite a fo rmu la fo r the nth term and determine the limit (if it exists). It is
assu med that n = 1, 2, 3, . . ..
(a) 1 , 1 , 1 , 1 , . . .. (b ) 1 , 2 , 3 , 4 , . .
2 4 6 8 2 3 4 5
(c) 1, − 1 , 1 , − 1 , 1 , − 1 , . . . (d) 0.9, 0.99, 0.999, 0.9999, . . .
2 3 4 5 6
(e) sin π , sin π, sin 3π , sin 2π, sin 5π , . . . ()()()
2 3 4
(f ) 2 , 3 , 4 , 5 , . . .
2 2 2 1 2 3 4
(a) sn = 1 ; lim 1 = 0 .
2n n→ +∞ 2n

n + 1 n→ +∞ n + 1 n→ +∞ ( n +1 )
(b ) sn = n ; lim n = lim 1 − 1 = 1 − lim 1 = 1 − 0 = 1.
n→ +∞ n + 1

(−1)n+1 (−1)n+1
(c) sn = ; lim = 0 . T his is intu itiv ely clear, b u t o ne can also apply T heo rem 42.3 to the seq u ence
n n→ +∞ n
〈(−1)n+1 n 〉 , since lim (−1)n+1 n = ∞.
n→ +∞

10 n→ +∞ ( 10 )
(d) sn = 1 − 1 n ; lim 1 − 1 n = 1 − lim 1 n = 1 − 0 = 1.
n→ +∞ 10

N o te that lim 1 = 0 b y v irtu e o f T heo rem 42.4(b).

n→ +∞ 10 n
(e) sn = sin nπ . N o te that the seq u ence co nsists o f repetitio ns o f the cy cle 1, 0, −1, 0 and has no limit.
2

( ) ( ) = lim (1 + n1 ) = e b y (26.17).
n n n
(f ) sn = n + 1 ; lim n + 1
n n→ +∞ n n→ +∞

2. E v alu ate lim sn in the fo llo w ing cases:

n→ +∞

(a) sn = 5n2 − 4 n + 13 (b ) sn = 8n − 3 3n + 7
2 2
(c)
3n − 95n − 7 2n + 5 n 3 − 2n − 9

(a) R ecall that lim 5x 2 − 4 x + 13 = 5 b y C hapter 7, P ro b lem 13. T herefo re, lim 5n2 − 4 n + 13 = 5 . A similar
2 2

x → +∞ 3 x − 95 x − 7 3 n→ +∞ 3n − 95n − 7 3
resu lt ho lds w henev er sn is a q u o tient o f po ly no mials o f the same degree.
(b ) R ecall that lim 8 x − 3 = +∞ b y C hapter 7, P ro b lem 13. T herefo re, lim 8n − 3 = +∞. A similar resu lt ho lds
2 2

x → +∞ 2 x + 5 n→ +∞ 2 n +5
w henev er sn is a ratio nal fu nctio n w ho se nu merato r has greater degree than the deno minato r (and w ho se
leading co efficients hav e the same sign).
(c) R ecall that lim 3 3x + 7 = 0 b y C hapter 7, P ro b lem 13. T herefo re, lim 3 3n + 7 = 0 . T he same resu lt
x → +∞ x − 2 x − 9 n→ +∞ n − 2n − 9
ho lds w henev er sn is a ratio nal fu nctio n w ho se deno minato r has greater degree than the nu merato r.

3. F o r each o f the fo llo w ing seq u encies, determine w hether it is no ndecreasing, increasing, no nincreasing,
decreasing, o r no ne o f these. T hen determine its limit, if it exists.
(a) sn = 5n − 2 (b ) sn = nn (c) 1n
7n + 3 2 3

(7 x + 3)(5) − (5x − 2)(7)

(a) L et f ( x) = 5x − 2 . T hen f ′( x) = = 29 > 0.
7x + 3 (7 x + 3)2 (7 x + 3)2
H ence, f(x) is an increasing fu nctio n and, therefo re, 〈 sn 〉 is an increasing seq u ence.
2 x − x(ln x)2 x 1 − x(ln 2)
(b ) L et f ( x) = xx . T hen f ′( x) = = .
2 22 x 2x
 356 CHAPTER 42 Infinite Sequences

S ince ln 2 > 12 (b y (2 5 .1 2 )), x(ln2) > x/2 ≥ 1, w hen x ≥ 2. T hu s, 1 − x(ln2) < 0 w hen x ≥ 2 and, therefo re,
f ′(x) < 0 w hen x ≥ 2. S o , f(x) is decreasing fo r x ≥ 2 and this implies that sn is decreasing fo r n ≥ 2. N o te that
s1 = 12 = s2 . H ence, 〈 sn 〉 is no nincreasing. N o w let u s find the limit. B y L ’H ô pital’s R u le,

lim xx = lim 1 n =0
x = 0 and, therefo re, lim
2
x → +∞ x → +∞ (ln 2)2 n→ +∞ 2
n

(c)
sn+1
sn ( ) ( 31 ) = 13 < 1. H ence, 〈s 〉 is decreasing.
= 1n+1
3 n n

T heo rem 42.4(b) tells u s that lim 1 = lim ( 1 ) = 0 .

n

3 n→ +∞ 3n
n→ +∞

1 ⋅ 3 ⋅ 5 ⋅ 7 . . . (2n − 1)
4. S ho w that the seq u ence sn = is co nv ergent.
2 ⋅ 4 ⋅ 6 ⋅ 8 . . . (2n)
L et u s u se T heo rem 42.8. 〈 sn 〉 is b o u nded, since 0 < sn < 1. L et u s sho w that 〈 sn 〉 is decreasing. N o te that

1 ⋅ 3 ⋅ 5 ⋅ 7 ⋅⋅⋅ (2n + 1)
sn+1 = = s 2n + 1 < s
2 ⋅ 4 ⋅ 6 ⋅ 8 ⋅⋅⋅ (2n + 2) n 2n + 2 n

5. P ro v e T heo rem 42.1: E v ery co nv ergent seq u ence 〈 sn 〉 is b o u nded.

L et lim sn = L . T ak e  = 1. T hen there exists a po sitiv e integer n0 su ch that, w henev er n ≥ n0, w e hav e
n→ +∞
|sn − L| < 1 . H ence, fo r n ≥ n0, u sing the triangle ineq u ality , w e get

|sn| = |(sn − L ) + L | ≤ |sn − L| + |L | < 1 + | L |

S o , if w e tak e M to b e the maximu m o f 1 + |L | and |s1 |, |s2 |, |s3 |, . . . , |sn0| , then |sn| ≤ M fo r all n. T hu s, 〈 sn 〉 is
b o u nded.

6. S ho w that the seq u ence n! is div ergent.

2n
...
S ince nn! = 1 ⋅ 2 ⋅ 3 . . . n = 1 3 4 . . . n > n fo r n > 4, the seq u ence is no t b o u nded. S o , b y T heo rem 42.1,
2 2⋅2⋅2 2 2 2 2 2 2
the seq u ence canno t b e co nv ergent.

7. P ro v e T heo rem 42.3: If lim sn = ∞ and sn ≠ 0 fo r all n, then lim 1n = 0 .

n→ +∞ n→ +∞ s
C o nsider any  > 0 . S ince lim sn = ∞ , there exists so me po sitiv e integer m su ch that, w henev er n ≥ m,
n→ +∞

|sn | > 1 and, therefo re, 1 − 0 = 1 < . So, lim 1 = 0.

 sn sn sn
n→ +∞

8. P ro v e T heo rem 42.4: (a) if |a| > 1, then lim a n = ∞ ; (b ) If |r | < 1 , then lim r n = 0 .
n→ +∞ n→ +∞

(a) L et M > 0, and let |a | = 1 + b . S o , b > 0. N o w , |a|n = (1 + b)n = 1 + nb + . . . > 1 + nb > M w hen n ≥ M .
b
(b ) L et a = 1/r. S ince |r | < 1, |a | > 1 . B y part (a), lim a n = ∞ . H ence, lim (1/r n ) = ∞ . S o , b y T heo rem 42.3,
n→ +∞ n→ +∞
lim r n = 0 .
n→ +∞

9. P ro v e: lim 1n = 0 .
n→ +∞ 2

lim 2n = ∞ b y T heo rem 42.4(a). H ence, lim 1n = 0 b y T heo rem 42.3.

n→ +∞ n→ +∞ 2

10 . P ro v e T heo rem 42.2(c) and (e).

A ssu me lim sn = c and lim tn = d .
n→ +∞ n→ +∞
CHAPTER 42 Infinite Sequences 357

(c) lim (sn + tn ) = c + d . L et  > 0 . T hen there exist integers m1 and m2 su ch that |sn − c| <  / 2 fo r n ≥ m1 and
n→ +∞
|tn − d | <  / 2 fo r n ≥ m2. L et m b e the maximu m o f m1 and m2. S o , fo r n ≥ m, |sn − c| <  / 2 and |tn − d | <  / 2.
H ence, fo r n ≥ m,

|(sn + tn ) − (c + d )| = |(sn − c) + (tn − d )| ≤ |sn − c| + |tn − d | <  +  = 

2 2

(e) lim (sn tn ) = cd . S ince 〈 sn 〉 is co nv ergent, it is b o u nded, b y T heo rem 42.1 and, therefo re, there is a po sitiv e
n→ +∞
nu mb er M su ch that |sn | ≤ M fo r all n. L et  > 0 . If d ≠ 0, there exists an integer m1 su ch that |sn − c| <  / 2 |d |
fo r n ≥ m1 and, therefo re, |d ||sn − c| <  / 2 fo r n ≥ m1. If d = 0, then w e can cho o se m1 = 1 and w e w o u ld
still hav e |d ||sn − c| <  / 2 fo r n ≥ m1. T here also exists m2 su ch that |tn − d | <  / 2 M fo r n ≥ m2. L et m b e the
maximu m o f m1 and m2. If n ≥ m,

|sn tn − cd | = |sn (tn − d ) + d (sn − c)| ≤ |sn (tn − d )| + |d (ssn − c)|

= |sn ||tn − d | + |d ||sn − c| ≤ M ( 2M ) + 2 = 

11. P ro v e the S q u eez e T heo rem: If lim sn = L = lim un , and there is an integer m su ch that sn ≤ tn ≤ un fo r all n ≥ m,
n→ +∞ n→ +∞
then lim tn = L .
n→ +∞
L et  > 0. T here is an integer m1 ≥ m su ch that |sn − L| <  / 4 and |un − L| <  / 4 fo r n ≥ m1. N o w assu me
n ≥ m1. S ince sn ≤ tn ≤ un, |tn − sn| ≤ |un − sn|. B u t

|un − sn| = |(un − L ) + ( L − sn )| ≤ |un − L| + |L − sn| <  +  = 

4 4 2
T hu s, |tn − sn | <  / 2. H ence,

|tn − L| = |tn − sn | + (sn − L )| ≤ |tn − sn | + |sn − L| <  +  < 

2 4

S U PPLEMEN T AR Y PR O B LEMS

In each o f P ro b lems 12– 29, determine fo r each giv en seq u ence 〈 sn 〉 w hether it is b o u nded and w hether it is
no ndecreasing, increasing, no nincreasing, o r decreasing. A lso determine w hether it is co nv ergent and, if po ssib le, find
its limit. (N ote: If the seq u ence has a finite limit, it mu st b e b o u nded. If it has an infinite limit, it mu st b e u nb o u nded.)

12 . n+ 2 Ans. no ndecreasing; increasing fo r n ≥ 2; limit +∞

n

13 . sin nπ Ans. b o u nded; no limit

4

14 . 〈 3 n 2 〉 Ans. increasing; limit +∞

15 . n! Ans. increasing fo r n ≥ 10; limit +∞

10 n

16 . ln n Ans. decreasing fo r n ≥ 3; limit 0

n

17 . 〈 12 (1 + (−1)n+1 )〉 Ans. b o u nded; no limit

18 . ln n + 1 Ans. decreasing; limit 0

n
 358 CHAPTER 42 Infinite Sequences

19 . 2n Ans. no nincreasing; decreasing fo r n ≥ 2; limit 0

n!

2 0 . 〈n n 〉 Ans. decreasing fo r n ≥ 3; limit 1

2 1. 3n Ans. increasing; limit 3

n+2

22. co s π Ans. increasing, limit 1

n

23. 4n + 5 Ans. decreasing; limit 0

n 3 − 2n + 3

24. sin n Ans. limit 0

n

2 5 . 〈 n +1 − n〉 Ans. decreasing; limit 0

26. 2n Ans. decreasing; limit 0

3n − 4

27. n sin π Ans. increasing, limit π

n

28. 1 Ans. increasing; limit +∞

n2 + 1 − n

29. nn Ans. increasing; limit +∞

n!

In each o f P ro b lems 30– 32, find a plau sib le fo rmu la fo r a seq u ence w ho se first few terms are giv en. F ind the limit
(if it exists) o f y o u r seq u ence.

3 0 . 1, 3 , 9 , 27 , 81 , . . . Ans. sn = 3n−1 ; limit is +∞

2 4 6 8 2(n − 1)

3 1. −1, 1, −1, 1, −1, 1, . . . Ans. sn = (−1)n; no limit

3 2 . 3 , 7 , 11 , 3 , 19 , . . . Ans. sn = 4 n − 1 ; decreasing, limit is 4

1 4 7 2 11 3n − 2 3

3 3 . P ro v e T heo rem 42.7. (H int: L et  > 0 . C ho o se δ > 0 su ch that, fo r x in the do main o f f fo r w hich |x − c| < δ , w e
hav e | f ( x ) − f (c)| <  . C ho o se m so that n ≥ m implies |sn − c| < δ .)

3 4 . S ho w that lim n
1/n p = 1 fo r p > 0. (H int: np/n = e(p ln n)/n.)
n→ +∞
CHAPTER 42 Infinite Sequences 359

3 5 . (G C ) Use a graphing calcu lato r to inv estigate sn = n 2 + 5 fo r n = 1 to n = 5. T hen determine analy tically the
b ehav io r o f the seq u ence. 4n 4 + n

1
Ans. decreasing; limit is 2

5
3 6 . (G C ) Use a graphing calcu lato r to inv estigate sn = nn fo r n = 1 to n = 10. T hen determine analy tically the
2
b ehav io r o f the seq u ence.
Ans. decreasing fo r n ≥ 7 ; limit is 0

3 7 . P ro v e that lim an = 0 is eq u iv alent to lim |an | = 0 .

n→ +∞ n→ +∞

3 8 . If sn > 0 fo r all n and lim sn2 = c , pro v e that lim sn = c .

n→ +∞ n→ +∞

 
3 9 . (G C ) D efine sn b y recu rsio n as fo llo w s: s1 = 2 and sn+1 = 1  sn + 2  fo r n ≥ 1.
2 sn 
(a) Use a graphing calcu lato r to estimate sn fo r n = 2, . . . , 5.
(b ) S ho w that, if lim sn exists, then lim sn = 2 .
n→ +∞ n→ +∞
(c) P ro v e that lim sn exists.
n→ +∞

4 0 . D efine sn b y recu rsio n as fo llo w s: s1 = 3, and sn+1 = 12 ( sn + 6) fo r n ≥ 1.

(a) P ro v e sn < 6 fo r all n.
(b ) S ho w that < sn > is increasing.
(c) P ro v e that lim sn exists.
n→ +∞
(d) E v alu ate lim sn .
n→ +∞

Ans. (d) 6

4 1. P ro v e T heo rem 42.2, parts (a), (b ), (d ), (f ).

 C H A P TE R 4 3

Infinite Series
Let 〈sn〉 be an infinite sequence. We can form the infinite sequence of partial sums 〈Sn 〉 as follows:
S1 = s1

S2 = s1 + s2

S3 = s1 + s2 + s3

Sn = s1 + s2 + ⋅⋅⋅ + sn

We usually will designate the sequence 〈Sn 〉 by the notation

∑s n = s1 + s2 + ⋅⋅⋅ + sn + ⋅⋅⋅
The numbers s1, s2, . . ., sn, . . . will be called the terms of the series.
If S is a number such that lim Sn = S , then the series ∑ sn is said to conv erg e and S is called the sum of
n→ +∞ +∞
the series. We usually designate S by ∑ sn .
n=1
If there is no number S such that lim Sn = S , then
+∞
the series ∑ sn is said to div erg e. If lim Sn = +∞ , then
n→ +∞ n→ +∞
the series is said to div erge to +∞
+∞
and we write ∑ sn = +∞ . Similarly, if lim Sn = −∞ , then the series is said
n→ +∞
to div erge to −∞ and we write ∑ sn = −∞ .
n =1

n =1

EXAMPLE 43 .1: C onsider the sequence 〈(– 1) n+1〉. The terms are s1 = 1, s2 = − 1, s3 = 1, s4 = − 1, and so on. Hence,
the p artial sums begin with S1 = 1, S2 = 1 + (− 1) = 0, S3 = 1 + (−1) + 1 = 1, S4 = 1 + (−1) + (1) + (−1) = 0, and continue
with alternating 1s and 0s. So, lim Sn does not ex ist and the series div erges (but not to +∞ or −∞).
n→ +∞

G e o m e tric S e rie s
C onsider the sequence 〈ar n−1〉, which consists of the terms a, ar, ar2, ar3, . . . .
The series ∑ ar n−1 is called a g eometric series with ratio r and first term a. Its nth p artial sum Sn is giv en by

Sn = a + ar + ar 2 + ⋅⋅⋅ + ar n−1

Multip ly by r: rSn = ar + ar 2 + ⋅⋅⋅ + ar n−1 + ar n

Subtract: Sn − rSn = a − ar n

Hence,, (1 − r ) Sn = a(1 − r n )

a(1 − r n )
Sn =
1− r

36 0
äåæçè é êë ìíî ïïðñòðððñòððïñòðóîô çõëö÷ø!ùè úû%ü ýé þ þ(äåÿ%æú
é ö
,ñ
 øäþ é øëöè ö1åè2ì1öè ÿ 3å

,ö

CHAPTER 43 Infinite Series 36 1

E v erything now dep ends on the ratio r. If |r| < 1, then lim r n = 0 (by Theorem 42.4(b)) and, therefore,
n→ +∞
lim Sn = a / (1 − r ) . If |r| > 1, then lim r n = ∞ (by Theorem 42.4(a)) and, therefore, lim Sn = ∞. (A triv ial
n→ +∞ n→ +∞ n→ +∞
ex cep tion occurs when a = 0. In that case, all terms are 0, the series conv erges, and its sum is 0.) These results
are summariz ed as follows:

T h e o re m 43 .1: G iv en a geometric series : ∑ ar n−1

a
(a) If |r| < 1, the series conv erges and has sum 1 − r .
(b) If |r| > 1 and a ≠ 0, the series div erges to ∞.

EXAMPLE 43 .2: Tak e the geometric series ∑( 1 n−1

2) with ratio r = 1
2 and first term a = 1:

1 + 12 + 14 + 81 + ⋅⋅⋅
+∞
1 = 1 =2
B y Theorem 43.1(a), the series conv erges and has sum ∑ ( ) = 2. 1 − ( 12 ) 12
. Thus,
n =1
1 n −1
2

We can multip ly a series ∑ s by a constant c to obtain a new series ∑ cs , and we can add two series ∑ s
n n n

and ∑ t to obtain a new series ∑ (s + t ) .

n n n

T h e o re m 43 .2: If c ≠ 0, then ∑ csn conv erges if and only if ∑ sn conv erges. Moreov er, in the case of conv ergence,

+∞ +∞

∑ cs
n =1
n = c ∑ sn
n =1

To obtain this result, denote by Tn = cs1 + cs2 + ⋅⋅⋅ + csn the nth p artial sum of the series ∑ csn . Then
Tn = cSn, where Sn is the nth p artial sum of ∑ sn . So, lim Tn ex ists if and only if lim Sn ex ists, and when the
n→ +∞ n→ +∞
limits ex ist, lim Tn = c lim Sn . This yields Theorem 43.2.
n→ +∞ n→ +∞

T h e o re m 43 .3 : A ssume that two series ∑ sn and ∑ tn both conv erge. Then their sum ∑ (sn + tn ) also conv erges and

+∞ +∞ +∞

∑ (s
n =1
n + t n ) = ∑ sn +
n =1
∑t
n =1
n

To see this, let Sn and Tn be the nth p artial sums of ∑ sn and ∑ t n, resp ectiv ely. Then the nth p artial sum
Un of ∑ (sn + t n ) is easily seen to be Sn + Tn. So, lim U n = lim Sn + lim Tn . This yields Theorem 43.3.
n→ +∞ n→ +∞ n→ +∞

C o ro lla ry 43 .4: A ssume that two series ∑s n and ∑t n both conv erge. Then their difference ∑ (s n − tn ) also
conv erges and

+∞ +∞ +∞

∑ (s
n =1
n − t n ) = ∑ sn − ∑ t n
n =1 n =1

This follows directly from Theorems 43.2 and 43.3. J ust note that ∑ (s n − t n ) is the sum of ∑ sn and the
series ∑ (−1) t n .

T h e o re m 43 .5 : If ∑s n conv erges, then lim sn = 0 .

n→ +∞
+∞
To see this, assume that ∑ sn = S . This means that lim Sn = S, where, as usual, Sn is the nth p artial sum of
n→ +∞
n =1
the series. We also hav e lim Sn−1 = S. B ut, sn = Sn − Sn−1. So, lim sn = lim Sn − lim Sn−1 = S − S = 0 .
n→ +∞ n→ +∞ n→ +∞ n→ +∞
 36 2 CHAPTER 43 Infinite Series

C o ro lla ry 43 .6 (T h e D iv e rg e n c e T h e o re m ): If lim sn does not ex ist or lim sn ≠ 0, then ∑ sn div erges.

n→ +∞ n→ +∞

This is an immediate logical consequence of Theorem 43.5.

EXAMPLE 43 .3 : The series 13 + 25 + 37 + 94 + ⋅⋅⋅ div erges.

Here, sn = n . Since lim n = 1 ≠ 0 , the D iv ergence Theorem imp lies that the series div erges.
2n + 1 n→ +∞ 2n + 1 2
The conv erse of Theorem 43.5 is not v alid: lim sn = 0 does not imp ly that ∑ sn conv erges. This is shown
n→ +∞
by the following ex amp le.

1 1 1 1 1
EXAMPLE 43 .4: C onsider the so-called harmonic series ∑ = 1 + + + + + ⋅⋅⋅ . Let us look at the following
n 2 3 4 5
p artial sums of this series:

S2 = 1 + 1
2

S4 = 1 + 1 + 1 + 1 > 1 + 1 + 1 + 1 = 1 + 1 + 1 = 1 + 2
2 3 4 2 4 4 2 2 2

S8 = S4 + 1 + 1 + 1 + 1 > S4 + 1 + 1 + 1 + 1 = S4 + 4 = S4 + 1
5 6 7 8 8 8 8 8 8 2

> 1+ 3
2

S16 = S8 + 1 + 1 + 1 + 1 + 1 + 1 + 1 + 1
9 10 11 12 13 14 15 16

> S8 + 1 + 1 + 1 + 1 + 1 + 1 + 1 + 1 = S8 + 1
16 16 16 16 16 16 16 16 2

> 1+ 4
2

C ontinuing in this manner, we would obtain S32 > 1 + 52 , S64 > 1 + 62 , and, in general, S2k > 1 + k / 2 when k > 1. This imp lies
that lim Sn = +∞ and, therefore, the harmonic series div erges. B ut notice that lim sn = lim 1/n = 0.
n→ +∞ n→ +∞ n→ +∞

R emark: C onv ergence or div ergence is not affected by the addition or deletion of a finite number of
terms at the beginning of a series. F or ex amp le, if we delete the first k terms of a series and the sum of the
deleted terms is c, then each new p artial sum Tn has the form Sn+k − c. (F or ex amp le, T1 is Sk+1 − c.) B ut
lim (Sn+ k − c) ex ists if and only if lim Sn+ k ex ists, and lim Sn+ k ex ists if and only if lim Sn ex ists.
n→ +∞ n→ +∞ n→ +∞ n→ +∞

N otation: It will often be useful to deal with series in which the terms of 〈sn 〉 are index ed by the non-
negativ e integers: s0, s1, s2, s3, . . . . Then the p artial sums Sn would also begin with S0 = s0, and the sum of a
+∞
conv ergent series would be written as ∑ sn .
n= 0

S O LV ED PR O B LEMS

1. E x amine the series 1 + 12 + 13 + ⋅⋅⋅ for conv ergence.

5 5 5
This is a geometric series with ratio r = 15 and the first term a = 15 . Since |r | = 1
5 < 1, Theorem 43.1 (a) tells us
that the series conv erges and that its sum is a = 1/ 5 = 1/ 5 = 1 .
1 − r 1 − (1/ 5) 4 / 5 4

2. E x amine the series 1 + 1 + 1 + 1 + ⋅⋅⋅ for conv ergence.

1⋅ 2 2 ⋅ 3 3 ⋅ 4 4 ⋅ 5
The nth term is 1 . This is equal to 1 − 1 . Hence, the nth p artial sum
n ⋅ (n + 1) n n +1
CHAPTER 43 Infinite Series 36 3

Sn = 1 + 1 + 1 + 1 + ⋅ + 1
1⋅ 2 2 ⋅ 3 3 ⋅ 4 4 ⋅ 5 n ⋅ (n + 1)

( )( )( )(
= 1 − 1 + 1 − 1 + 1 − 1 + 1 − 1 + ⋅⋅⋅ + 1 − 1
1 2 2 3 3 4 4 5 n n +1 ) ( )
= 1− 1
n +1
Thus, lim Sn = lim 1 −
n→ +∞ n→ +∞
(1
n +1 )
= 1 − 0 = 1. Hence, the series conv erges and its sum is 1.

3. We k now that the geometric series 1 + 12 + 14 + 81 + 161 + ⋅⋅⋅ conv erges to S = 2. E x amine the series that results
when: (a) its first four terms are drop p ed; (b) the terms 3, 2, and 5 are added to the beginning of the series.
(a) The resulting series is a geometric series 161 + 321 + ⋅⋅⋅ with ratio 12 . It conv erges to 1/16 = 1/16 = 1 . N ote
1 − (1/ 2) 1/ 2 8
that this is the same as S − (1 + 12 + 14 + 81 ) = 2 − ( 158 ) = 81 .
(b) The new series is 3 + 2 + 5 + 1 + 2 + 4 + 8 + 16 + ⋅⋅⋅ . The new p artial sums are the old ones p lus (3 + 2 + 5).
1 1 1 1

Since the old p artial sums conv erge to 2, the new ones conv erge to 2 + 10 = 12. Thus, the new series is
conv ergent and its sum is 12.

4. Show that the series 12 + 43 + 87 + 16

15
+ ⋅⋅⋅ div erges.
Here, sn = 2 n
− 1 = 1 − n . Since lim 1n = 0 , it follows that lim sn = 1 − 0 = 1 ≠ 0 . So, by the D iv ergence
1
2n 2 n→ +∞ 2 n→ +∞
Theorem, the series div erges.

5. E x amine the series 9 − 12 + 16 − 64

3 + 256
9 − ⋅⋅⋅ for conv ergence.
This is a geometric series with ratio r = − 43 . Since |r | = 43 > 1, Theorem 43.1(b) tells us that the series div erges.

+∞
(−1) n
6. E v aluate ∑ = 1 − 1 + 1 − 1 − 1 − ⋅⋅⋅.
n=0
2n 2 4 8 16
This is a geometric series with ratio r = − 12 and first term a = 1. Since |r | = 12 < 1, the series conv erges and its
sum is a = 1 = 1 = 2.
1 − r 1 − (−1/ 2) 3/ 2 3

7. Show that the infinite decimal 0.999 . . . is equal to 1.

0.999 ⋅⋅⋅ = 9 + 9 + 9 + ⋅⋅⋅ . This is a geometric series with first term a = 9
10 and ratio r = 1
10 .
10 100 1000
Hence, it conv erges to the sum a = 9 /10 = 9 /10 = 1.
1 − r 1 − (1/10) 9 /10

8. E x amine the series 1 + 1 + 1 + 1 + ⋅⋅⋅ .

1⋅ 3 3 ⋅ 5 5 ⋅ 7 7 ⋅ 9
Here, sn = 1
(2n − 1)(2n + 1)
. N ote that 1 =1 1 − 1
(2n − 1)(2n + 1) 2 2n − 1 2n + 1 ( )
. Hence, the nth p artial sum Sn is

( ) ( ) (
1 1 − 1 + 1 1 − 1 + 1 1 − 1 + ⋅⋅⋅ + 1
2 1 3 2 3 5 2 5 7 )
1 − 1
(
= 1 1− 1
2 2n − 1 2n + 1 2 2n + 1 ) ( )
So, lim Sn = 12 . Thus, the series conv erges to 12 .
n→ +∞

9. E x amine the series 3 + 3 + 3 3 + 4 3 + ⋅⋅⋅.

sn = n 3 = 31/ n = e(ln 3) / n . Then lim sn = e 0 = 1 ≠ 0 . B y the D iv ergence Theorem, the series div erges.
n→ +∞

10 . E x amine the series 1

10 + 111 + 121 + 131 + ⋅⋅⋅ .
This series is obtained from the harmonics series by deleting the first nine terms. Since the harmonic series
div erges, so does this series.
 36 4 CHAPTER 43 Infinite Series

11. (Z e n o ’s P a r a d o x ) A chilles (A) and a tortoise (T ) hav e a race. T gets a 1000 ft head start, but A runs at 10 ft/sec,
whereas T only does 0.01 ft/sec. When A reaches T ’s starting p oint, T has mov ed a short distance ahead. When A
reaches that p oint, T again has mov ed a short distance ahead, etc. Z eno claimed that A would nev er catch T. Show
that this is not so.
When A reaches T ’s starting p oint, 100 seconds hav e p assed and T has mov ed 0.01 (100) = 1 ft. A cov ers that
additional 1 ft in 0.1 seconds, but T has mov ed 0.01(0.1) = 0.001 ft further. A needs 0.0001 seconds to cov er that
distance, but T meanwhile has mov ed 0.01(0.0001) = 0.000001 ft, etc. The limit of the distance between A and
T ap p roaches 0. The time inv olv ed is 100 + 0.1 + 0.0001 + 0.0000001 + ⋅⋅⋅ , which is a geometric series with first
term a = 100 and ratio r = 1/1000. Its sum is

a = 100 = 100 = 100000

1 − r 1 − (1/1000) 999 /1000 999

which is a little more than 100 seconds. The seeming p aradox arises from the artificial div ision of the ev ent into
infinitely many shorter and shorter step s.

S U PPLEMEN T AR Y PR O B LEMS

12 . E x amine each of the following geometric series. If the series conv erges, find its sum.
(a) 4 − 1 + 14 − 161 + ⋅⋅⋅ Ans. S = 165
(b) 1 + 32 + 94 + 278 + ⋅⋅⋅ Ans. D iv erges
(c) 1 − 13 + 19 − 271 + ⋅⋅⋅ Ans. S = 43
(d) 1 + e −1 + e −2 + e −3 + ⋅⋅⋅ Ans. S= e
e −1

13 . A rubber ball is drop p ed from a height of 10 ft. Whenev er it hits the ground, it bounces straight up three-fourths
of the p rev ious height. What is the total distance trav eled by the ball before it stop s?

Ans. 70 ft

14 . E x amine the series ∑ n(n1+ 4) = 11⋅ 5 + 21⋅ 6 + 31⋅ 7 + ⋅⋅⋅ .

Ans. S= 25
48

15 . E x amine the series ∑ n(n + 11)(n + 2) = 1 ⋅ 21 ⋅ 3 + 2 ⋅ 31 ⋅ 4 + 3 ⋅ 41 ⋅ 5 + ⋅⋅⋅

Ans. S= 1
4

+∞
16 . E v aluate ∑ sn when sn is the following:
n=1

(a) 3 −n
(b) 1 (c) 1 (d) n
n(n + 2) n(n + 3) (n + 1)!
1 3 11
Ans. (a) 2 ; (b) 4 ; (c) 18 ; (d) 1

17 . Show that each of the following series div erges:

(a) 3 + 52 + 73 + 94 + ⋅⋅⋅ (b) 2 + 2 + 3 2 + 4 2 + ⋅⋅⋅ (c) 1 + 1 + 31 + 41 + ⋅⋅⋅
2 2 2 2
(d) e+ e2
+ e3
+ e4
+ ⋅⋅⋅ (e) ∑ 1
8 27 64
n + n −1

18 . E v aluate the following

( )
+∞ +∞ +∞
(a) ∑ 1n + 1n
n=0
2 7
(b) ∑ 41n
n=1
(c) ∑ n 2(nn ++ 11)
n =1
2 2
CHAPTER 43 Infinite Series 36 5

+∞ +∞ +∞

∑ 2 5+ 3
n −1
∑ 23 ∑ 53
n n n
(d) n (e) n (f ) n
n= 0 n =1 n=1

+∞ +∞ +∞
(−1) n
∑ n n+ −n +3 2
2 3n
(g)
n =1
2 (h) ∑
n =1
52 n
(i) ∑ 32
n=1
2n

+∞
( j) ∑1
n=1

Ans. (a) 19
6 ; (b) +∞; (c) 1; (d) 25
6 ; (e) 1; (f ) +∞; (g) +∞; (h) − 261 ; (i) 8; ( j) +∞

19 . (G C ) In P roblems 1 and 6, use a calculator to comp ute the first 10 p artial sums and determine to how many
decimal p laces the 10th p artial sum is a correct estimate of the sum of the series.

+∞
2 0 . (G C ) (a) If |x| < 1, what function is rep resented by ∑ x n = 1 + x + x 2 + x 3 + ⋅⋅⋅?
n=0
(b) U se a grap hing calculator to grap h 1 + x + x 2 + x 3 + ⋅⋅⋅ + x 9 on the interv al (−1, 1) and comp are the grap h
with that of the function in (a).

Ans. (a) 1
1− x

2 1. In each of the following, find those v alues of x for which the giv en series conv erges, and then find the function
rep resented by the sum of the series for those v alues of x.

() ( )
+∞ +∞ +∞ n +∞ n
(a) ∑ (3x) n (b) ∑ ( x − 2) n (c) ∑ x (d) ∑ x − 1
n= 0 n= 0 n=0
2 n= 0
2

Ans. (a) |x| < 1 , 1 ; (b) 1 < x < 3, 1 ; (c) |x| < 2 , 2 ; (d) − 1 < x < 3, 2
3 1 − 3x 3− x 2− x 3− x
 C H A P TE R 4 4

Series w ith P ositive Term s.The

IntegralTest.C om parison Tests

S e rie s o f Po s itiv e T e rm s
If all the terms of a series ∑ s are p ositiv e, then the series is c alled a positiv e series.
n

F or a p ositiv e series ∑ s , the seq u en c e of p artial su ms 〈S 〉 is an in c reasin g

n n
seq u en c e, sin c e
Sn + 1 = Sn + sn + 1 > Sn . T his y ield s the follow in g u sefu l resu lt.

T h e o re m 44.1: A p ositiv e series ∑s n c on v erg es if an d on ly if the seq u en c e of p artial su ms 〈 sn 〉 is b ou n d ed .

T o see this, n ote first that, if ∑ sn c on v erg es, then , b y d efin ition , 〈 Sn 〉 c on v erg es an d , therefore, b y T heorem 42.1,
〈 Sn 〉 is b ou n d ed . C on v ersely , if 〈 Sn 〉 is b ou n d ed , then , sin c e 〈 Sn 〉 is in c reasin g , T heorem 42.8 imp lies that
〈 Sn 〉 c on v erg es, that is, ∑ sn c on v erg es.

T h e o re m 44.2 (In te g ra l T e s t): L et ∑

sn b e a p ositiv e series an d let f (x) b e a c on tin u ou s, p ositiv e d ec reasin g fu n c -
tion on [1, + ∞) su c h that f (n) = sn for all p ositiv e in teg ers n. T hen :
+∞
∑s n c on v erg es if an d on ly if ∫1
f ( x) dx c on v erg es
n
F rom F ig . 44-1 w e see that ∫ f ( x) dx < s1 + s2 + ⋅⋅⋅ + sn− 1 = Sn− 1 . If ∑ sn c on v erg es, then 〈 Sn 〉 is b ou n d ed ;
1
u +∞
so, ∫ f ( x) dx w ill b e b ou n d ed for all u ≥ 1 an d , therefore, ∫ f ( x) dx c on v erg es. C on v ersely , from F ig . 44-1
1 1
n n +∞
w e hav e s2 + s3 + ⋅⋅⋅ + sn <
+∞
∫ 1
f ( x) dx an d , therefore, Sn <∫
1 ∫
f ( x) dx + s1. T hu s, if
1
f ( x) dx c on v erg es, then
Sn < ∫ f ( x) dx + s, an d so 〈 Sn 〉 w ill b e b ou n d ed . H en c e, b y T heorem 44.1, ∑ s n c on v erg es. T his p rov es
1

T heorem 44.2.

Sn
1 2 3 4 5 n– 2 n– 1 n

Fig. 44-1

36 6
äåæçè é êë ìíî ïïðñòðððñòððïñòðóîô çõëö÷ø!ùè úû%ü ýé þ þ(äåÿ%æú
é ö
,ñ
 øäþ é øëöè ö1åè2ì1öè ÿ 3å

,ö

CHAPTER 44 Series w ith P ositive Term s 36 7

ln n
EXAMPLE 44.1: ∑ n
d iv erg es.
ln x
L et f ( x) = . N ow ,
x

+∞ u
ln x dx = lim u ln x dx = lim 1 (ln x) 2  = lim 1 ((ln u ) 2 − 0) = +∞
∫1 x u → +∞ ∫1 x u → +∞
2 1 u→ +∞ 2

H en c e, b y the in teg ral test, ∑ lnnn d iv erg es.

EXAMPLE 44.2: ∑ n1 2
c on v erg es.
1
L et f ( x) = 2 . N ow ,
x

( )
+∞ u
1 dx = lim u
1 dx = lim − 1  = liim − 1 − 1 = 1
∫
1 x2 u → +∞ ∫
1 x2 u → +∞ x 1 u→ +∞ u

1
H en c e, b y the in teg ral test, ∑n 2 c on v erg es.
R emark: T he in teg ral test c an b e easily b e ex ten d ed to the c ase w here the low er limit of the in teg ral is
c han g ed from 1 to an y p ositiv e in teg er.

T h e o re m 44.3 (C o m p a ris o n T e s t): L et ∑ an an d ∑b n b e tw o p ositiv e series su c h that there is a p ositiv e in teg er m

for w hic h ak ≤ bk for all in teg ers k ≥ m. T hen :
(1) If ∑b n c on v erg es, so d oes ∑ an ;
(2) If ∑ a n d iv erg es, so d oes ∑ bn .
W e may assu me in the d eriv ation of T heorem 44.3 that m = 1, sin c e c on v erg en c e is n ot affec ted b y d ele-
tion of a fin ite n u mb er of terms at the b eg in n in g of a series. N ote also that (2) is a log ic al c on seq u en c e of
(1). T o p rov e (1), assu me that ∑ bn c on v erg es. L et Bn = b1 + b2 + ⋅⋅⋅ + bn b e the nth p artial su m for ∑ bn an d
let An = a1 + a2 + ⋅⋅⋅ + an b e the nth p artial su m ∑ an . T hen An ≤ Bn , sin c e ak ≤ bk for all k. F rom the fac t
that ∑ bn c on v erg es, it follow s, b y T heorem 44.1, that the seq u en c e 〈 Bn 〉 is b ou n d ed . S in c e An ≤ Bn for all
n, it follow s that the seq u en c e 〈 An 〉 is b ou n d ed . S o, b y T heorem 44.1, ∑ an c on v erg es. T his p rov es T heo-
rem 44.3 .

EXAMPLE 44.3 : ∑n 1 c on v erg es.

+52

an d bn = 12 . T hen an < bn for all n. B y E x amp le 2, ∑ 2 c on v erg es. S o, b y the c omp arison test,
1 1
L et an = 2
n +5 n n
∑ n 2 1+ 5 c on v erg es.
EXAMPLE 44.4: ∑ 3 n1+ 5 d iv erg es.
1
L et an = 4 n an d bn = 1 . N ow , an ≤ bn for n ≥ 5. (T o see this, ob serv e that 1 ≤ 1 is eq u iv alen t to 3 n +
3n+5 4n 3 n + 5
5 ≤ 4n, w hic h is eq u iv alen t to 5 ≤ n.) R ec all that the harmon ic series ∑ 1 d iv erg es (b y C hap ter 43 , E x amp le 4).
n
H en c e, ∑ 1 d iv erg es b y T heorem 43 .2. T he c omp arison test imp lies that ∑ 1 d iv erg es.
4n 3n+5
S ometimes, as in E x amp le 4, c omp lic ated man eu v ers are n eed ed in ord er to ap p ly the c omp arison test.
T he follow in g resu lt offers a mu c h more flex ib le tool.
an
T h e o re m 44.4 (Lim it C o m p a ris o n T e s t): L et ∑ an an d ∑ bn b e tw o p ositiv e series su c h that L = nlim
→ +∞ bn
ex ists
an d 0 < L < +∞. T hen ∑ an c on v erg es if an d on ly if ∑ bn c on v erg es.
 36 8 CHAPTER 44 Series w ith P ositive Term s

Assume that ∑ bn converges. Let c be a positive number such that L < c. Then there exists a positive inte-
ger m such that an /bn < c for all n ≥ m. Hence, an < cbn for all n ≥ m. But, since ∑ bn converges, so does ∑ cbn .
Therefore, by the comparison test, ∑ an converges. Conversely, if ∑ an converges, then ∑ bn converges.
b 1
(In fact, lim n = > 0 and w e can use the same k ind of argument that w as just given.)
n→ +∞ an L

∑ 3n7n− 5+n2+ 4 diverges.

2
EXAMPLE 44.5: 3

W hen dealing w ith q uotients of polynomials, a good rule of thumb is to ignore everything except the leading
2
terms. In this case, w e have 3n 3 = 3 1 . Let us try a limit comparison w ith 1 . N ow
7n 7n n

n→ +∞ (
lim  3n −3 5n + 4
2

 7n + 2 ) 1  = lim 3n 3 − 5n 2 + 4 n = 3 .
n  n→ +∞ 7n 3 + 2 7

∑ 3n7n− 5+n2+ 4 diverges.

2
S ince ∑ 1 diverges, the limit comparison test tells us that 3
n
5n − 2
EXAMPLE 44.6 : ∑ n6 − 4n 2 + 7
converges.
5n = 5n = 5 .
U sing the rule of thumb given in E xample 5, w e should look at 3
n 2 S o, let us try a limit comparison
1 n6 n
w ith 2 :
n

  1  = lim 5n 3 − 2n 2
lim  6 5n − 22
n→ +∞
  n − 4 n + 7  n 2  n→ +∞ n 6 − 4 n 2 + 7

Let us divide the numerator and denominator by n3. N ote that, in the denominator, w e w ould get

1 n6 − 4n 2 + 7 = 1 n 6 − 4 n 2 + 7 = 1 − 44 + 66
n3 n6 n n

S o, the result w ould be

5− 2
lim n = 5 =5
n→ +∞ 4 7 1
1− 4 + 6
n x

5n − 2
Hence, since w e k now , by E xample 2, that ∑ n1 2 converges, the limit comparison test implies that ∑ n6 − 4n 2 + 7
converges.

S O LV ED PR O B LEMS

1. Consider the series ∑ 1p , w here p is constant. This is called a p-series. Then:

n
(a) If p > 1, the series ∑ 1p converges.
n
(b) If p ≤ 1, the series ∑ 1p diverges.
n
W e may assume that p ≠ 1, since w e already k now that the harmonic series ∑ 1 diverges. W e may also assume
n
that p > 0; if p ≤ 0, lim 1p ≠ 0 and the D ivergence Theorem implies that the series diverges. Let us apply the
n→ +∞ n
integral test w ith f ( x) = 1/ x p . (f (x) is positive and decreasing in [1, +∞).) N ow ,
u
+∞
1 dx = lim u
1 dx = lim x1− p 
∫ 1 xp u → +∞ ∫
1 xp u → +∞ 1 − p 
1

= lim  u − 1 .
1− p

u → +∞  1 − p 1− p
CHAPTER 44 Series with Positive Terms 369

(a) p > 1. Then p − 1 > 0 and lim u1− p = lim 1p−1 = 0. S o, lim  u − 1  = 1 . By the integral test,
1− p

u → +∞ u → +∞ u u → +∞  1 − p 1− p p − 1
∑ np1 converges.

(b) p < 1. Then 1 − p > 0 and lim u1− p = +∞. S o, lim  u − 1  = +∞ and, by the integral test, ∑ 1p
1− p

u → +∞ u → +∞  1 − p 1− p n
diverges.

In P roblems 2– 7, examine the given series for convergence.

2. 1 + 1 + 1 + 1 + ⋅ ⋅ ⋅.
3 5 7

sn = 1 . Let f ( x) = 1 . O n [1, +∞), f ( x) > 0 and f is decreasing.

2n − 1 2x − 1
+∞ u
1 dx = lim 1 u (2 x − 1)−1/22 (2) dx
∫1 2x − 1
dx = lim ∫
u → +∞ 1 2 x − 1 u→ +∞ 2 ∫1
u
= lim 1
(2)(2 x − 1)1/ 2  = lim ((2u − 1)1/ 2 − 1) = +∞
u → +∞
2
1 u→ +∞

Hence, the series diverges by the integral test.

3. 1 + 1 + 1 + ⋅ ⋅ ⋅ + 1 + ⋅ ⋅ ⋅.
3 10 29 n3 + 2
1 < 1.
n3 + 2 n3 ∑ n1 3 is convergent, since it is a p-series w ith p = 3 > 1. Thus, by the comparison test, ∑ n 1+ 2
3

is convergent.

4. 1 + 1 + 1 + 1 + ⋅ ⋅ ⋅.
2! 3! 4!

sn = 1 . N ote that 1 = 1 ≤ 1 for n ≥ 2. S ince ∑ 1n−1 is a convergent geometric series (w ith

n! n! n(n − 1) ⋅⋅⋅⋅⋅ 3 ⋅ 2 2n−1 2
ratio 12 ), ∑ 1 is convergent by the comparison test.
n!

5. 2 + 33 + 43 + 53 + ⋅ ⋅ ⋅.
2 3 4
sn = n +3 1 . U se limit comparison w ith n3 = 12 .
n n n

lim n +3 1 12 = lim n +3 n = 1
3 2

n→ +∞n n n→ +∞ n

W e k now that ∑ 12 converges. S o, by the limit comparison test, ∑ n +3 1 converges.

n n

6. 1 + 12 + 13 + 14 + ⋅ ⋅ ⋅
2 3 4
sn = 1n . N ow , 1n = 1 ≤ 1 and ∑ 1n−1 is a convergent geometric series (r = 12 ). S o, by the
n n n ⋅ n ⋅⋅⋅⋅⋅ n 2n−1 2
comparison test, ∑ 1n converges.
n

1 + 23 + 1 + 33 + 1 + 4 3 + 1 + ⋅ ⋅ ⋅.
2 2 2
7.
2 +1 3 +1 4 +1

sn = n 3 + 1 . U se limit comparison w ith n 3 = 1 :

2 2

n +1 n n

n→ +∞ n(
lim  n 3 + 1
2

+1 ) 1  = lim n 3 + n = 1
n  n→ +∞ n 3 + 1
 37 0 CHAPTER 44 Series with Positive Terms

2
+ 1 diverges.
W e k now that the harmonic series ∑ n1 diverges. S o, by the limit comparison test, ∑ nn 3
+1

8. 1 + 1 + 1 + ⋅ ⋅ ⋅.
2 ln 2 3 ln 3 4 ln 4

sn = 1 is defined for n ≥ 2.
n ln n

+∞ u
dx = lim u
dx = lim ln(ln u )  = lim (ln (ln u ) − ln (ln 2)) = +∞ .
∫2 x ln x u→ +∞ ∫
2 x ln x u→ +∞  2 u→ +∞

Hence, the series diverges by the integral test.

9. How many terms of ∑ 12 suffice to obtain tw o-decimal place accuracy (that is, an error < 5/103)?
n
If w e use k terms, then w e req uire that the error

( )
u
1 dx = lim − 1  = lim − 1 − 1
+∞ k +∞
1 ≤ +∞ 1 dx = lim u
∑
n =1
1 −
n 2 ∑
n =1
1 =
n 2 ∑
n = k +1
n 2 ∫k x 2 u → +∞ ∫ k x2 u → +∞ x  u→ +∞ u k
k

= 1 < 53 = 1
k 10 200

Hence, 200 < k. Thus, it suffices to use 201 terms of the series. (The graphing calculator can be used to find
201

∑
n =1
1 ≈ 1.64.)
n 2

10 . A ssume ∑ sn converges by virtue of the integral test applied to f (x) and, for each n, the error (or remainder) Rk
after k terms is defined to be

+∞ k +∞ +∞
∑s − ∑s .
n =1
n
n =1
n Then Rk = ∑ s <∫
n = k +1
n k
f ( x) dx.

+∞
F ind a bound on the error w hen ∑ 12 is approximated by the first five terms: 1 + 1 + 1 + 1 + 1 = 5269 ≈ 1.4636.
n=1
n 4 9 16 25 3600
+∞
The error R5 < ∫ 12 dx = 15 = 0.2.
5 x

11. A ssume ∑ sn and ∑c n are positive series, ∑ cn converges, and sn ≤ cn for all n. Then the error Rk after k terms is

+∞ k +∞ +∞

∑s − ∑s
n =1
n
n =1
n = ∑s
n = k +1
n ≤ ∑c.
n = k +1
n

+∞
A t least how many terms w ill suffice to estimate ∑ n 1+ 1 w
n =1
5 ith an error < 0.00001?
+∞ +∞ +∞
In this case, sn = 51 and cn = 15 . It suffices to have ∑ 15 < 0.00001. N ow , ∑ 15 < ∫ 15 dx = 1 4 .
n +1 n n = k +1
n n = k +1
n k x 4k
S o, w e need 1 4 < 0.00001 = 1 . E q uivalently, 100,000 < 4k4, 25,000 < k4, k ≥ 13.
4k 100, 000
CHAPTER 44 Series with Positive Terms 37 1

S U PPLEMEN T AR Y PR O B LEMS

F or P roblems 12– 43, determine w hether the series converges.

12 . ∑ n(n3+ 1) Ans. converges; comparison w ith ∑ 32

n

13 . ∑ (n + 1)(nn + 2) Ans. diverges; limit comparison w ith ∑ 1

n

14 . ∑ n n+ 12 Ans. diverges; limit comparison w ith ∑ n1

15 . ∑ en n Ans. converges; integral test

16 . ∑ (n + 1)(n2+n2)(n + 3) Ans. converges; limit comparison w ith ∑ 12

n

17 . ∑ (2n 1+ 1) 2 Ans. converges; limit comparison w ith ∑ n1 2

18 . ∑ n 1− 13 Ans. converges; limit comparison w ith ∑ 13

n

19 . ∑ nn− 2 3 Ans. converges; limit comparison w ith ∑ n1 2

20. ∑ nln+n22 Ans. converges; limit comparison w ith ∑ n1 3/ 2

2 1. ∑ n sin ( n1 ) Ans. diverges; D ivergence Theorem

22. ∑ 3
1
n
Ans. diverges; p-series, p = 13 < 1

23. ∑ n1 n−1 Ans. converges; comparison w ith ∑ 1n−1 , n ≥ 2

2

24. ∑ lnnn Ans. diverges; comparison w ith ∑ ln n

n

25. ∑ 1 +1ln n Ans. diverges; comparison w ith ∑ 1

n

n +1
26. ∑n 3n − 2
Ans. diverges; limit comparison w ith ∑ 1
n

27. ∑ n ln n ln1 (ln n) (for n ≥ 3) Ans. diverges; integral test

28. ∑ n ln n (ln1 (ln n)) 2 (for n ≥ 3) Ans. converges, integral test

 37 2 CHAPTER 44 Series with Positive Terms

29. 1 + 1 + 1 + 1 + ⋅ ⋅ ⋅.
4 2 72 10 2 132

Ans. sn = 1
(3n + 1)2
; converges; limit comparison w ith ∑ n1 2

3 0 . 3 + 31/ 3 + 31/ 3 + 31/ 3 + ⋅ ⋅ ⋅.

2 3 4

Ans. sn = 31/ 3 ; diverges; p-series, p = 13 < 1

n

3 1. 1 + 1 + 1 + 1 + ⋅ ⋅ ⋅.
5 9 13

Ans. sn = 1 ; diverges; limit comparison w ith

4n − 3 ∑ n1

32. 1+ 1 + 1 + 1 + ⋅ ⋅ ⋅.
2 3⋅ 4 4 ⋅5⋅6 5⋅6 ⋅ 7 ⋅8

Ans. sn = 1
(n + 1)(n + 2) ⋅ ⋅ ⋅ (2n)
; converges; limit comparison w ith ∑ n1
2

3 3 . 2 + 3 2 + 4 3 + 5 4 + ⋅ ⋅ ⋅.
3 2⋅3 3⋅3 4 ⋅3

Ans. sn = n + 1n ; converges; limit comparison w ith ∑ 31

n ⋅3 n

3 4 . 1 + 1 2 + 1 3 + 1 4 + ⋅ ⋅ ⋅.
2 2⋅2 3⋅2 4⋅2

Ans. sn = 1 n ; converges; comparison w ith

n2 ∑ 21 n

35. 2 + 3 + 4 + 5 + ⋅ ⋅ ⋅.
1⋅ 3 2 ⋅ 4 3 ⋅ 5 4 ⋅ 6

n + 1 ; diverges; limit comparison w ith

Ans. sn =
n(n + 2) ∑ n1
3 6 . 1 + 22 + 33 + 44 + ⋅ ⋅ ⋅.
2 3 4 5

Ans. sn = n ; converges; comparison w ith

(n + 1)n ∑ 21 n−1

3 7 . 1 + 12 + 15/ 2 + 13 + ⋅ ⋅ ⋅.
2 3 4

Ans. sn = 1 ; converges; comparison w ith

n ( n+ 2)/ 2 ∑ n1 2

3 8 . 1 + 3 + 4 + 5 + ⋅ ⋅ ⋅.
5 10 17

Ans. sn = n2 + 1 ; diverges; limit comparison w ith ∑ n1

n +1

3 9 . 2 + 2 ⋅ 4 + 2 ⋅ 4 ⋅ 6 + 2 ⋅ 4 ⋅ 6 ⋅ 8 + ⋅ ⋅ ⋅.
5 5 ⋅ 8 5 ⋅ 8 ⋅ 11 5 ⋅ 8 ⋅ 11 ⋅ 14

2 ⋅ 4 ⋅ ⋅ ⋅ ⋅ ⋅ (2n)
Ans. sn = ; converges; comparison w ith ∑ ( 23 )n
5 ⋅ 8 ⋅ ⋅ ⋅ (2 + 3n)
CHAPTER 44 Series with Positive Terms 37 3

40. 3 + 5 + 7 + 9 + ⋅ ⋅ ⋅.
2 10 30 68

Ans. sn = 23n + 1 ; c o n v e rg e s , lim it c o m p a ris o n w ith ∑ n1

n +n 2

41 . 3 + 10 + 29 + 66 + ⋅ ⋅ ⋅.
2 24 108 320

sn = n4 + 23 ; d iv e rg e s ; lim it c o m p a ris o n w ith

3
Ans.
n +n ∑ n1
42 . 1 + 2 + 3 + 4 + ⋅ ⋅ ⋅.
2 2 − 1 32 − 2 4 2 − 3 52 − 4

Ans. sn = n
( n + 1) 2 − n
; d iv e rg e s ; lim it c o m p a ris o n w ith ∑ n1
43 . 1 + 1 + 1 + 1 + ⋅ ⋅ ⋅.
23 − 12 33 − 22 4 3 − 32 53 − 4 2

1
Ans. sn =
( n + 1) 3 − n 2
; c o n v e rg e s ; lim it c o m p a ris o n w ith ∑ n1 3

44. (G C ) E s tim a te th e e rro r w h e n :

+∞
(a ) ∑ n 1 is a p p ro x im a te d b y th e s u m o f its firs t s ix te rm s .
n =1
3 +1
+∞
1 is a p p ro x im a te d b y th e s u m o f its firs t s ix te rm s .
(b ) ∑4
n =1
n
+3

Ans. (a ) 0.0007; (b ) 0.00009

45 . (G C ) (a ) E s tim a te th e e rro r w h e n th e g e o m e tric s e rie s ∑ 23 n is a p p ro x im a te d b y th e s u m o f its firs t s ix te rm s .

(b ) H o w m a n y te rm s s u ffic e to c o m p u te th e s u m if th e a llo w a b le e rro r is 0.00005?

Ans. (a ) 0 .0 4 7 ; (b ) 1 6
+∞
46 . (G C ) (a ) H o w m a n y te rm s s u ffic e to a p p ro x im a te ∑ n1
n=1
4 w ith a n e rro r < 0.001?
+∞
(b ) F in d a b o u n d o n th e e rro r if w e a p p ro x im a te ∑ 14 b y th e s ix th p a rtia l s u m .
+∞ n=1
n
(c ) W h a t is y o u r a p p ro x im a tio n to ∑ 14 b y th e s ix th p a rtia l s u m , c o rre c t to fo u r d e c im a l p la c e s ?
n=1
n
Ans. (a ) 7 ; (b ) 0 .0 0 1 5 ; (c ) 1 .0 8 1 1

47 . (G C ) L e t Sn b e th e nth p a rtia l s u m 1 + 1 + ⋅ ⋅ ⋅ + 1 o f th e d iv e rg e n t h a rm o n ic s e rie s .

2 n
(a ) P ro v e ln ( n + 1) ≤ Sn ≤ 1 + ln n.
(b ) L e t En = Sn − ln n . P ro v e th a t 〈 En 〉 is b o u n d e d a n d d e c re a s in g .
(c ) P ro v e th a t 〈 En 〉 c o n v e rg e s . Its lim it is d e n o te d γ a n d is c a lle d Euler’s co nstant.
(d ) U s e a g ra p h in g c a lc u la to r to a p p ro x im a te E999 to e ig h t d e c im a l p la c e s .

Ans. (d ) 0.57771608 (in fa c t, g ~ 0.57721566.)

48 . (E x te n s io n o f th e lim it c o m p a ris o n te s t.) A s s u m e ∑ sn a n d ∑t n a re p o s itiv e s e rie s . P ro v e :

sn
(a ) If lim = 0 a n d ∑ t n c o n v e rg e , s o d o e s ∑ sn .
n→ +∞ t n
sn
(b ) If lim = +∞ a n d ∑ t n d iv e rg e s , s o d o e s ∑ sn .
n→ +∞ t n
 37 4 CHAPTER 44 Series with Positive Terms

(ln n)4
4 9 . U se the extension of the limit comparison test to determine w hether ∑ converges.
n3

Ans. converges; use ∑ n1 2 and P roblem 48 (a)

5 0 . A ssume ∑ sn is a positive series and lim nsn exists and is positive. P rove that ∑ s diverges. (Hint: Limit
n
n→ +∞
comparison w ith ∑ (1/ n).)

5 1. A ssume ∑s n and ∑t n are convergent positive series. P rove that ∑s t

n n converges.
 C H A P TE R 4 6

Power Series
Po w e r S e rie s
An infinite series
+∞

∑ a ( x − c)
n= 0
n
n
= a0 + a1 ( x − c) + a2 ( x − c)2 + ⋅⋅⋅ (46.1)

is called a po w er series in x about c with coefficients 〈 an 〉. An important special case

+∞

∑a x
n= 0
n
n
= a0 + a1 x + a2 x 2 + ⋅⋅⋅ (46.2)

is a power series about 0.

For a given value of x, the series (46.1) either converges or diverges. Hence, (46.1) determines a function
f whose domain is the set of all x for which (46.1) converges and whose corresponding value f (x) is the sum
of the series.
Note that (46.1) converges when x = c.

EXAMPLE 46 .1 : The power series about 0

+∞

∑x
n=0
n
= 1 + x + x2 + ⋅ ⋅ ⋅

is a geometric series with ratio x. Thus, it converges for |x| < 1, and its sum is 1 . S o, the domain of the
1− x
corresponding function is an interval around 0.

+∞
T h e o re m 46 .1 : Assume that the power series ∑ a ( x − c)
n=0
n
n
converges for x0 ≠ c. Then it converges absolutely for all
x such that |x - c| < |x0 - c| (that is, for all x that are closer to c than x0).
For a proof, see P roblem 4.

+∞
T h e o re m 46 .2 : For a power series ∑ an ( x − c)n, one of the following three cases holds:
n=0
(a) it converges for all x; or
(b) it converges for all x in an open interval (c - R1, c + R1) around c, but not outside the closed interval
[c - R1, c + R1]; or
(c) it converges only for x = c.
+∞
B y the interv al o f co nv erg ence of ∑ a ( x − c)
n=0
n
n
we mean:
In case (a): (- ∞, +∞)
In case (b): (c - R1, c + R1)
In case (c): { c}

38 3


  















 !#"
$ %&('# )
*,+ - . ./ 
0,
)
1
 $
23
4 1&5
.  &6#
$
 $78
98$
 0 2:
7;
23$
5
 38 4 CHAPTER 46 Power Series

+∞
B y the radius o f co nv erg ence of ∑ a ( x − c)
n
n we mean:
In case (a): ∞ n=0

In case (b): R1
In case (c): 0
N o te: In case (b), whether the power series converges at neither endpoint of its interval of convergence or
at one or both of those endpoints depends upon the given series.
For a sk etch of a proof of Theorem 46.2, see P roblem 5.

EXAMPLE 46 .2 : The power series

+∞
( x − 2)n ( x − 2)2 ( x − 2)3
∑
n =1
n
= ( x − 2) +
2
+
3
+ ⋅⋅⋅

is a power series about 2. L et us use the ratio test to find the interval of convergence.

= |x − 2| |x − 2|n = n |x − 2|. Thus,

sn+1 n +1 sn+1
lim = |x − 2|.
sn n +1 n n +1 n→+∞ sn

S o, by the ratio test, the series converges absolutely for |x − 2| < 1. The latter ineq uality is eq uivalent to −1 < x − 2 < 1,
which, in turn, is eq uivalent to 1 < x < 3. Hence, the interval of convergence is (1, 3) and the radius of convergence is 1.
+∞
At the endpoint x = 1, the series becomes
+∞
∑ [(−1) / n], which converges by
n =1
n
the Alternating S eries Theorem. At the end-
point x = 3, the series becomes ∑ (1/ n), the divergent harmonic series. Thus, the power series converges for 1 ≤ x < 3.
n=1

EXAMPLE 46 .3 : The power series

+∞ n 2 3
∑ xn! = 1 + x + x2! + x3! + ⋅ ⋅ ⋅
n= 0

is a power series about 0. (R ecall that 0! = 1.) L et us use the ratio test:

sn+1 n +1
|x|n = |x| . sn+1
= | x| S o, lim = 0.
sn (n + 1)! n! n +1 n→+∞ sn

Hence, by the ratio test, the series converges (absolutely ) for all x. Its interval of convergence is (- ∞, +∞) and its
radius of convergence is ∞.

EXAMPLE 46 .4: The power series

+∞

∑ n! x
n= 0
n
= 1 + x + 2! x 2 + 3! x 3 + ⋅ ⋅ ⋅

is a power series about 0. L et us use the ratio test again:

sn+1 (n + 1)! |x|n+1 sn+1

= = (n + 1) |x|. S o, lim = +∞ .
sn n! |x|n n→+∞ sn

ex cept when x = 0. Thus, the series converges only for x = 0. Its (degenerate) “ interval” of convergence is { 0} and its
radius of convergence is 0.
CHAPTER 46 Power Series 38 5

U n ifo rm C o n v e rg e n c e
L et 〈 fn 〉 be a seq uence of functions, all defined on a set A. L et f be a function defined on A. Then 〈 fn 〉 is said
to co nv erg e unifo rmly to f on A if, for every  > 0, there ex ists a positive integer m such that, for each x in A
and every n ≥ m, |fn(x) − f(x)| < .

+∞
T h e o re m 46 .3 : If a power series
k
∑ a ( x − c)
n=0
n
n
converges for x0 ≠ c and d < |x0 − c|, then the seq uence of partial sums
+∞
〈 Sk ( x)〉, where Sk ( x) = ∑ an ( x − c)n, converges uniformly to ∑ a ( x − c)
n
n
on the interval consisting of all x such that
n=0 n=0
|x − c| < d. Hence, the convergence is uniform on any interval strictly inside the interval of convergence.
The reader is referred to more advanced book s on analy sis for a proof of this result.

T h e o re m 46 .4: If 〈 fn 〉 converges uniformly to f on a set A and each fn is continuous on A, then f is continuous on A.

For a proof, see P roblem 6.

+∞
C o ro lla ry 46 .5: The function defined by a power series ∑ a ( x − c)
n=0
n
n
is continuous at all points within its interval of
convergence.

This follows from Theorems 46.3 and 46.4.

+∞
T h e o re m 46 .6 (In te g ra tio n o f Po w e r S e rie s ): L et f be the function defined by a power series ∑ a ( x − c)
n=0
n
n
on
its interval of convergence (with radius of convergence R1). Then:

+∞
( x − c)n+1
(a) ∫ f ( x) dx = ∑ a n= 0
n n +1
+K for | x − c| < R1 (46.3)

where the interval of convergence of the power series on the right side of formula (46.3) is the same as that of the
original series. K is an arbitrary constant of integration. Note that the antiderivative of f is obtained by term-by -
term integration of the given power series.
(b) If a and b are in the interval of convergence, then:
b
( x − c)n+1  
+∞
f ( x)dx = ∑  an
b
∫
a
n= 0
 n + 1  
(46.4)
a

b
Thus, ∫
a
f ( x)dx is obtained by term-by -term integration.

For a proof of Theorem 46.6, the reader should consult a more advanced book on analy sis.

+∞
T h e o re m 46 .7 (D iffe re n tia tio n o f Po w e r S e rie s ): L et f be the function defined by a power series ∑ an ( x − c)n
n=0
on its interval of convergence (with radius of convergence R1). Then f is differentiable in that interval and

+∞
f ′( x) = ∑ nan ( x − c)n−1 for | x − c | < R1 (46.5)
n=0

Thus, the derivative f ′ is obtained by term-by -term differentiation of the power series. The interval of convergence of
the power series on the right side of formula (46.5) will be the same as for the original power series.

For a proof, the reader is referred to more advanced tex ts in analy sis.
 38 6 CHAPTER 46 Power Series

EXAMPLE 46 .5: W e already k now by E x ample 1 that, for |x| < 1,

+∞

1− x ∑
1 = x n = 1 + x + x 2 + x3 + ⋅ ⋅ ⋅ + x n + ⋅ ⋅ ⋅ (46.6)
n= 0

Now, Dx ( 1 −1 x ) = (1 −1x) . S o, by Theorem 46.7,

2

1 = 1 + 2 x + 3x 2 + ⋅ ⋅ ⋅ + nx n−1 + ⋅ ⋅ ⋅ for |x| < 1

(1 − x)2
+∞ +∞
= ∑ nx n−1 = ∑ (n + 1) x n
n =1 n= 0

EXAMPLE 46 .6 : W e k now already that

+∞

1− x ∑
1 = x n = 1 + x + x 2 + x3 + ⋅ ⋅ ⋅ + x n + ⋅ ⋅ ⋅ for |x| < 1
n=0

R eplace x by −x. (This is permissible, since |−x| = |x| < 1.) The result is
+∞ +∞

1+ x ∑ ∑
1 = (− x)n = (−1)n x n = 1 − x + x 2 − x 3 + ⋅ ⋅ ⋅ (46.7)
n=0 n=0

B y Theorem 46.6(a), we can integrate term by term:

+∞ +∞
x n+1 + K = (−1)n−1 x n + K
∫ 1 + x = ∑ (−1) ∑
dx n
for |x| < 1
n= 0
n +1 n =1
n
+∞ n
ln 1 + x = ∑ (−1)n−1 x + K for |x| < 1
n =1
n

L etting x = 0 and noting that ln 1 = 0, we find that K = 0.

Note also that, for |x| < 1, we have −1 < x < 1, 0 < 1 + x < 2, and, therefore, |1 + x| = 1 + x. Hence,

+∞ n
ln (1 + x) = ∑ (−1)n−1 x for |x| < 1
n =1
n

= x − 1 x2 + 1 x3 − 1 x 4 + ⋅ ⋅ ⋅ (46.8 )
2 3 4

The ratio test shows that this series converges.

If we replace x by x − 1, we obtain:

+∞
( x − 1)n
ln x = ∑ (−1)n−1 for |x − 1| < 1 (46.9 )
n =1
n

Note that |x − 1| < 1 is eq uivalent to 0 < x < 2.

Thus, ln x is definable by a power series within (0, 2).
+∞
T h e o re m 46 .8 (Ab e l’s T h e o re m ): Assume that the power series ∑ an ( x − c)n has a finite interval of convergence
n=0
|x − c| < R1 and let f be a function whose values in that interval are given by that power series. If the power series also
converges at the right-hand endpoint b = c + R1 of the interval of convergence, then lim− f ( x) ex ists and is eq ual to the
x →b
sum of the series at b. The analogous result holds at the left-hand endpoint a = c − R1.

The reader is referred to advanced book s on analy sis for a proof.

CHAPTER 46 Power Series 38 7

EXAMPLE 46 .7 : This is a continuation of E x ample 6. B y formula (46.8 ),

n +∞
ln (1 + x) = ∑ (−1)n−1 x for | x | < 1
n =1
n

At the right-hand endpoint x = 1 of the interval of convergence, the power series becomes the convergent
alternating harmonic series

+∞

∑ (−1)
n =1
n −1 1 =1− 1 + 1 + 1 + ⋅⋅⋅
n 2 3 4

B y Abel’s Theorem, this series is eq ual to lim− ln (1 + x) = ln 2 . S o,

x →1

ln 2 = 1 − 12 + 13 − 14 + ⋅ ⋅ ⋅ (46.10)

EXAMPLE 46 .8 : S tart again with

+∞

1− x ∑
1 = x n = 1 + x + x 2 + x 3 + ⋅ ⋅ ⋅ + x n + for |x | < 1
n=0

R eplace x by −x2, obtaining

+∞

1 + x2 ∑
1 = (−1)n x 2 n = 1 − x 2 + x 4 − x 6 + ⋅ ⋅ ⋅ (46.11)
n=0

S ince |−x2| < 1 is eq uivalent to |x| < 1, (46.11) holds for |x| < 1.
Now, by Theorem 46.6(a), the antiderivative tan−1 x of 1 2 can be obtained by term-by -term integration:
1+ x

+∞2 n +1
tan −1 x = ∑ (−1)n x +K for |x| < 1
n= 0
2 n +1

= K + x − 13 x 3 + 15 x 5 − 17 x 7 + ⋅ ⋅ ⋅

Here K is the constant of integration. If we let x = 0 and note that tan−1 0 = 0, it follows that K = 0. Hence,

+∞ 2 n +1
tan −1 x = ∑ (−1)n x = x − 13 x 3 + 15 x 5 − 17 x 7 + ⋅ ⋅ ⋅ (46.12)
n= 0
2 n +1

At the right-hand endpoint x = 1 of the interval of convergence, the series in (46.12) becomes

+∞

∑ (−1)
n= 0
n 1 =1− 1 + 1 − 1 + ⋅⋅⋅
2n + 1 3 5 7

which converges by virtue of the Alternating S eries Theorem. S o, by Abel’s Theorem,

1 − 13 + 15 − 17 + ⋅ ⋅ ⋅ = lim− tan −1 ( x) = tan −1 1 = π (46.13)

x →1 4
 38 8 CHAPTER 46 Power Series

+∞ n +∞ n
EXAMPLE 46 .9 : W e k now already , by E x ample 3, that ∑ xn! converges for all x. L et f ( x) = ∑ x for all x.
B y term-by -term differentiation (Theorem 46.7), n= 0 n= 0
n!

+∞ n −1 +∞ n
f ′( x ) = ∑ x = ∑ x = f ( x)
n =1
( n − 1 )! n= 0
n !

Note that f (0) = 1. Therefore, by formula (28 .2), f (x) = ex. Thus,

+∞ n
ex = ∑ x for all x (46.14)
n= 0
n!

S O LV ED PR O B LEMS

1. Find the interval of convergence of the power series

+∞
( x − 2)n ( x − 2)2 ( x − 2)3
∑
n =1
n
= ( x − 2) +
2
+
3
+ ⋅⋅⋅

and identify the function represented by this power series.

U se the ratio test:

= |x − 2| |x − 2|n = n |x − 2| .
sn+1 n +1 sn+1
S o, lim = | x − 2|
sn n +1 n n +1 n→+∞ sn

Hence, the interval of convergence is |x − 2| < 1. (This is eq uivalent to −1 < x − 2 < 1. which, in turn, is
eq uivalent to 1 < x < 3.) At the right endpoint x = 3, the series is the divergent harmonic series, and, at the left
endpoint x = 1, the series is the negative of the convergent alternating harmonic series. S o, the series converges
for 1 ≤ x < 3.
+∞ +∞
( x − 2)n
L et h ( x) = ∑ . By Theorem 46.7, h ′( x) = ∑ ( x − 2)n−1 . This series is a geometric series with
n =1
n n =1
first term 1 and ratio (x − 2); so its sum is 1 = 1 . Thus, h ′( x) = 1 . H ence,
1 − ( x − 2) 3 − x 3− x
h( x) = ∫ dx = − ln |3 − x| + C . N ow,
3− x
+∞
(2 − 2)n
h(2) = ∑ =0 and − ln |3 − 2| + C = 0. S o, C= 0
n =1
n

M oreov er, since x < 3 in the interv al of conv ergence, 3 − x > 0 and, therefore, |3 − x| = 3 − x. Thus,
h(x) = −ln (3 − x).

In P roblems 2 and 3, find the interv al of conv ergence of the giv en series and the behav ior at the endp oints (if any).

+∞ n 2 3
2. ∑ nx
n =1
2 = x + x + x + ⋅ ⋅ ⋅.
4 9

U se the ratio test:

( )
2
sn+1 n +1
|x|n = n sn+1
= |x| 2 |x| . H ence, lim = |x|.
sn (n + 1) n2 n +1 n→+∞ sn

H ence, the interv al of conv ergence is | x | < 1. The radius of conv ergence is 1. A t x = 1, we obtain the conv ergent
p-series with p = 2. A t x = −1, the series conv erges by the alternating series test. Thus, the series conv erges for
−1 ≤ x ≤ 1.
CHAPTER 46 Power Series 389

+∞
( x + 1)n ( x + 1)2 ( x + 1)3
3. ∑
n =1 n
= ( x + 1) +
2
+
3
+ ⋅ ⋅ ⋅.

U se the ratio test:

= |x + 1| |x + 1|n =
sn+1 n +1
n |x + 1|. sn+1
H ence, lim = |x + 1| .
sn n +1 n n +1 n→+∞ sn

H ence, the interv al of conv ergence is |x + 1| < 1. This is eq uiv alent to −1 < x + l < 1, which, in turn, is eq uiv alent
+∞
to −2 < x < 0. The radius of conv ergence is 1. A t the right endp oint x = 0, we get the div ergent p-series ∑ 1
+∞
(−1)n n
(with p = 12 ). A t the left endp oint x = −2, we get the alternating series ∑
n=1
, which conv erges by the
n =1 n
A lternating S eries Theorem. Thus, the series conv erges for −2 ≤ x < 0.

4. P rov e Theorem 46.1.

S ince ∑ an ( x0 − c)n conv erges, lim an ( x0 − c)n = 0 by Theorem 43.5. H ence, there is a p ositiv e number M
n→+∞
such that |an| |x0 − c|n < M for all n, by Theorem 42.1. A ssume |x − c| < |x0 − c|. L et

r = |x − c| < 1. Then, |an ||x − c|n = |an ||x0 − c|n r n < Mr n.

|x0 − c|

Therefore, ∑ |an ( x − c)n | is conv ergent by comp arison with the conv ergent geometric series ∑ Mr n . Thus,
∑ an ( x − c)n is absolutely conv ergent.

5. P rov e Theorem 46.2.

O nly a v ery intuitiv e argument is p ossible here. A ssume that neither case (a) nor case (c) holds. S ince case (a)
does not hold, the p ower series does not conv erge for some x ≠ c. S ince case (c) does not hold, the series does
conv erge for some x ≠ c. Theorem 46.1 imp lies that there is an interv al (c − K, c + K) around c in which the series
conv erges. The interv al of conv ergence is the max imal such interv al. (U sing Theorem 46.1, one tak es the “ least
up p er bound” R1 of all K such that the series conv erges in (c − K, c + K). Then, (c − R1, c + R1) is the desired
interv al.)

6. P rov e Theorem 46.4.

A ssume x is in A and  > 0. S ince 〈 fn 〉 conv erges uniformly to f on A, there is a p ositiv e integer m such that,
if n ≥ m, then | fn ( y) − f ( y)| < /3 for all y in A. S ince fm is continuous at x, there ex ists d > 0 such that, for any x*
in A, if |x* − x| < d, then | fm(x*) − fm(x)| < /3. H ence if |x* − x| < d,

| f ( x * ) − f ( x)| = |( f ( x * ) − fm ( x * )) + ( fm ( x * ) − fm ( x)) + ( fm ( x) − f ( x))|

≤ | f ( x * ) − fm ( x * )| + | fm ( x * ) − fm ( x) | + | fm ( x) − f ( x)|

<  +  +  =
3 3 3

This p rov es the continuity of f at x.

b b
7. If 〈 fn 〉 conv erges uniformly to f on [a, b] and each fn is continuous on [a, b], then
n→+∞ a ∫
f ( x) dx = lim ∫ fn ( x) dx .
a
A ssume  > 0. There is a p ositiv e integer m such that, if n ≥ m, then | fn ( x) − f ( x)| <  for all x in [a, b].
b b−a
Therefore, ∫ a | fn ( x) − f ( x)| dx < . Then

b b b b
∫a
f ( x) dx − ∫ fn ( x) dx =
a ∫a
( fn ( x) − f ( x)) dx ≤ ∫ | fn ( x) − f ( x)| dx < 
a
for n ≥ m
 390 CHAPTER 46 Power Series

8. P rov e that the function f defined by a p ower series is continuous within its interv al of conv ergence (C orollary 46.5).
f ( x) = lim Sn ( x) and the conv ergence is uniform by Theorem 46.3. E ach Sn(x), being a p olynomial, is
n→+∞
continuous. H ence, f is continuous by Theorem 46.4.

9. F ind a p ower series about 0 that rep resents the function x 2 . In what interv al is the rep resentation v alid?
+∞ 1+ x
By formula (46.11), 1 2 = ∑ (−1)n x 2 n for |x| < 1. H ence
1+ x n= 0
+∞

1 + x2 ∑
x = (−1)n x 2 n+1 for |x| < 1
n=0

The series div erges at both endp oints x = 1 and x = −1.

In P roblems 10 and 11, use the ratio test to find the interv al of conv ergence and indicate what hap p ens at the
endp oints (if any).

10. ∑ 10n n xn .

sn+1
sn
=
(n + 1) |x|n+1
10 n+1 10 n ( )
n |x|n = n + 1 |x| .
n 10
H ence, lim
n→+∞
sn+1
sn
= |x| .
10

W e get conv ergence when |x|/10 < 1, that is, when |x| < 10. That is the interv al of conv ergence. The series
div erges at both endp oints ± 10.

11. ∑ 3n ( x − π ) .
n
n

(n + 1) |x − π |n+1 n | x − π |n = n + 1 | x − π | = |x − π | .
sn+1 sn+1
= H ence, lim
sn 3n+1 3n n 3 n→+∞ sn 3

S o, the interv al of conv ergence is |x − π| < 3. The series div erges at both endp oints.

(n!)2
1 2. F ind the interv al of conv ergence of ∑ (2n)!x .n

A p p ly the ratio test:

sn+1 ((n + 1)!)2 |x|n+1 (n!)2 |x|n (n + 1)2 sn+1

= = |x| H ence, lim = |x| .
sn (2n + 2)! (2n)! (2n + 2)(2n + 1) n→+∞ sn 4

S o, the interv al of conv ergence is |x| < 4.

1 3 . F ind a p ower series about 0 that rep resents x 3 .

+∞ 1− x
S tart with 1 = ∑ x n for |x| < 1. R ep lace x by x3:
1 − x n=0
+∞

1 − x3 ∑
1 = x 3n for |x| < 1
n= 0

(since |x3| < 1 is eq uiv alent to |x| < 1). M ultip ly by x:

+∞

1 − x3 ∑
x = x 3n+1 for |x| < 1
n= 0

In P roblems 14– 16, find simp le formulas for the function f (x) rep resented by the giv en p ower series.
CHAPTER 46 Power Series 391

14. x + x 2 + x 3 + ⋅ ⋅ ⋅.
2! 3! 4! +∞ n
L et f ( x) = ∑ x .
n =1
( n + 1)!
+∞ n +1 +∞
xf ( x) = ∑ x = ∑ xn! − 1 − x = e x − 1 − x
n

n =1
( n + 1 )! n= 0

H ence, f ( x) = e − 1 − x .
x

15. 1
3 x 3 + 16 x 6 + 19 x 9 + ⋅ ⋅ ⋅ .
+∞ 3 n +∞
L et f ( x) = ∑ x . Then f ′( x) = ∑ x 3n−1 = x 2 + x 5 + x 8 + ⋅ ⋅ ⋅.
n =1
3n n =1
This is a geometric series with ratio x3. S o, it conv erges for |x3| < 1, which is eq uiv alent to |x| < 1. H ence,
2 2
f ′( x) = x 3 for |x| < 1. Therefore, f ( x) = ∫ x 3 dx = − 13 ln |1 − x 3| + C . But f (0) = 0. H ence, C = 0. A lso,
1− x 1− x
1 − x3 > 0 for |x| < 1. Therefore,

f ( x) = − 13 ln (1 − x 3 ) for |x| < 1.

1 6 . x + 2 x 3 + 3 x 5 + 4 x 7 + ⋅ ⋅ ⋅.
The ratio test shows that the series conv erges for |x| < 1. L et
+∞
g( x) = x + 2 x 3 + 3x 5 + 4 x 7 + ⋅ ⋅ ⋅ = ∑ nx 2 n−1
n =1
+∞
Then 2 g( x) = ∑ 2nx 2 n −1
H ence, tak ing antideriv ativ es,
n =1
+∞ +∞
2 ∫ g( x)dx = K + ∑ x 2 n = K + x2 (since ∑ x 2 n is a geometric series with ratio x 2 )
n =1
1 − x2 n =1

N ow differentiate:

2 g( x ) = D x ( 1 −x x ) = (1 −2xx ) ,
2
2 2 2 g( x ) = x
(1 − x 2 )2
for |x| < 1

ln (1 + x)
1/ 2
1 7 . (G C ) A p p rox imate ∫0 x
dx to two-decimal-p lace accuracy (that is, with an error < 5/103).
By formula (46.8), ln (1 + x) = x − 12 x 2 + 13 x 3 − 14 x 4 + ⋅ ⋅ ⋅ for |x| < 1, S o

+∞
ln (1 + x) (−1)n x n
= 1 − 12 x + 13 x 2 − 14 x 3 + ⋅ ⋅ ⋅ = ∑
x n= 0
n +1

By Theorem 46.6(b),
1/ 2
ln (1 + x) +∞
(−1)n x n+1  +∞
(−1)n 1
dx = ∑ =∑
1/ 2
∫ 0 x n= 0
n + 1 n + 1 
0 n= 0
( n + 1)2 2n+1

which is a conv ergent alternating series.

In order to get an ap p rox imation with an error less than 5/103, we must find n such that the first omitted term
1 1 is ≤ 5 = 1 . S o, we must hav e 200 ≤ (n + 1)2 2n+ 1. Trial and error shows that n ≥ 3. H ence, we
(n + 1)2 2n+1 10 3 200
can use the terms corresp onding to n = 0, 1, 2:

1 − 1 + 1 = 65 ~ 0.45
2 16 72 144

This answer can be confirmed by a grap hing calculator (which yields 0.44841421 as an ap p rox imation).
 392 CHAPTER 46 Power Series

+∞
1 8 . F ind the function defined by ∑2 n=0
n
xn .
This is a geometric series with ratio 2x and first term 1. H ence, it conv erges for |2x| < 1, that is, for |x| < 12 , and
its sum is 1 .
1 − 2x
+∞ n
1 9 . F ind the interv al of conv ergence of ∑ ln (xn + 1).
n =1
A p p ly the ratio test:

sn+1 n +1
|x|n ln (n + 1)
= |x| = |x|
sn ln (n + 2) ln (n + 1) ln (n + 2)

s
By L ’H ô p ital’s rule, lim n+1 = |x|. H ence, the interv al of conv ergence is giv en by |x| < 1. (F or x = 1, we get
n→+∞ sn
+∞ +∞
(−1)n
∑
n =1
ln (
1
n + 1)
, which we k now is div ergent. F or x = −1, we get the conv ergent alternating series ∑
n =1
ln (n + 1)
.)

20 . A p p rox imate 1 with an error less than 0.0001.

e
By formula (46.14),

+∞ +∞
n (−1)n
ex = ∑ x for all x. H ence, 1 = e−1 =
∑
n= 0
n! e n=0
n!

By the A lternating S eries Theorem, we seek the least n such that 1/n! ≤ 0.0001 = 1/10,000, that is, 10,000 ≤ n!.
Trial and error shows that n ≥ 8. S o, we must use the terms corresp onding to n = 0, 1, . . . ,7:

1 − 1 + 1 − 1 + 1 − 1 + 1 − 1 = 103 ~ 0.3679
2 6 24 120 720 5040 280

(A grap hing calculator yields the answer 0.367 8794412, correct to 10 decimal p laces.)

1
21 . A p p rox imate ∫e
0
− x2
dx to two-decimal-p lace accuracy, that is, with an error less than 5/103 = 0.005.
By formula (46.14),

+∞ +∞
n (−1)n 2 n
ex = ∑ x e− x = ∑
2
for all x. H ence, x for all x
n= 0
n! n=0
n!

By Theorem 46.6(b),
1
(−1)n x 2 n+1 
+∞ +∞
(−1)n 1
e− x dx = ∑ =∑
1
∫
2

+ 
0
n= 0
n ! 2 n 1  0 n = 0 n! 2n + 1

W e can ap p ly the A lternating S eries Theorem. The magnitude of the first term omitted 1 should be
(2n + 1)n!
≤ 0.005 = 1/200. S o, 200 ≤ (2n + 1)n!. Trial and error shows that n ≥ 4. H ence, we should use the first four terms,
that is, those corresp onding to n = 0, 1, 2, 3:

1 − 1 + 1 − 1 = 26 ~ 0.743
3 10 42 35

(A grap hing calculator yields the ap p rox imation 0.74682413, correct to eight decimal p laces.)

22. F ind a p ower series ex p ansion for 1 about 0.

x+3 +∞
1 =1 1 . By formula (46.7), 1 = ∑ (−1)n x n = 1 − x + x 2 − x 3 + ⋅ ⋅ ⋅ for |x| < 1.
x + 3 3 ( x /3) + 1 1 + x n= 0
CHAPTER 46 Power Series 393

H ence,

( ) = ∑ (−1) 3x
+∞ n +∞ n

( x /3) + 1 ∑
1 = (−1)n x n
forr x <1
n
n= 0
3 n=0
3
+∞
1 = (−1)n x n
x+3 ∑
Thus, for |x| < 3
n= 0
3n+1

The series div erges at x = ± 3.

23 . F ind a p ower series ex p ansion for 1 about 1.

x +∞
1= 1 . By formula (46.7), 1 = ∑ (−1)n x n for |x| < 1. H ence,
x 1 + ( x − 1) 1 + x n= 0
+∞

x 1 + ( x − 1) ∑
1= 1 = (−1)n ( x − 1)n for |x − 1| < 1
n= 0

SUPPLEMENTARY PROBLEMS

In P roblems 24– 31, find the interv al of conv ergence of the giv en p ower series.

24 . ∑ nx n
Ans. −1 < x < 1

n
25 . ∑ n(nx+ 1) Ans. −1 ≤ x ≤ 1

n
26 . ∑ nx5 n Ans. −5 ≤ x < 5

2n
27 . ∑ n(n + x1)(n + 2) Ans. −1 ≤ x ≤ 1

n+1
28 . ∑ (ln (xn + 1)) 2 Ans. −1 ≤ x < 1

n
29 . ∑ 1 +x n 3 Ans. −1 ≤ x ≤ 1

( x − 4) n
30. ∑ n2
Ans. 3≤x≤5

(3x − 2)n
31. ∑ 5n
Ans. −1 < x < 7
3

3 2. E x p ress e−2x as a p ower series about 0.

+∞
(−1)n 2n n
Ans. ∑
n=0
n!
x

3 3 . R ep resent e x/2 as a p ower series about 2.

+∞
Ans. ∑ 2 (en!) ( x − 2)
n=0
n
n
 394 CHAPTER 46 Power Series

3 4 . R ep resent ln x as a p ower series about 2.

+∞
(−1)n+1
Ans. ln 2 + ∑ ( x − 2)n
n =1
n2n

3 5 . (G C ) F ind ln (0.97) with sev en-decimal-p lace accuracy. (Hint: U se the p ower series for ln (1 − x) about 0.)

Ans. −0.0304592

3 6 . H ow many terms in the p ower series for ln (1 + x) about 0 must be used to find ln 1.02 with an error
≤ 0.00000005?

Ans. Three

3 7 . (G C ) U se a p ower series to comp ute e−2 to four-decimal-p lace accuracy.

Ans. 0.1353

1/ 2
dx to four-decimal-p lace accuracy.
3 8 . (G C ) E v aluate ∫ 0 1 + x4

Ans. 0.4940

In P roblems 39 and 40, find the interv al of conv ergence of the giv en series.

+∞ n
39. ∑ nx
n=1
n Ans. (-∞, + ∞)

+∞
40. ∑ 10n! x
n= 0
n
n
Ans. x=0

4 1 . R ep resent cosh x = e + e as a p ower series about 0.

x −x

2
+∞ 2n
Ans. ∑ (2xn)!
n= 0

x
∫ e− t / 2 dt .
2
4 2. F ind a p ower series about 0 for the normal distribution function
0
+∞
(−1)n 2 n +1
Ans. ∑ n!(2 ) 2xn + 1
n= 0
n

4 3 . F ind a p ower series ex p ansion about 0 for ln 1 + x .

1− x
+∞ 2 n +1
Ans. 2∑ x
n=0
2 n +1

4 4 . (G C ) A p p rox imate tan −1 12 to two-decimal-p lace accuracy.

Ans. 0.46
CHAPTER 46 Power Series 395

+∞
4 5 . S how that the conv erse of A bel’s Theorem is not v alid, that is, if f ( x) = ∑ an x n for |x| < r, where r is the radius
+∞ n= 0
of conv ergence of the p ower series, and lim− f ( x) ex ists, then
x→r
∑a r n
n
need not conv erge. (Hint: L ook at
f ( x) = 1 .)
n= 0

1+ x

+∞
4 6 . F ind a simp le formula for the function f (x) rep resented by ∑ n 2 x n .
n =1
x( x + 1)
Ans.
(1 − x)3

+∞
4 7 . F ind a simp le formula for the function f (x) rep resented by ∑ xn .
n=2
( n − 1)n

Ans. x + (1 − x) ln (1 − x)

+∞

(1 − x)2 ∑
4 8 . (a) S how that x = nx n for |x| < 1. (Hint: U se E x amp le 5.)
n =1
+∞
2 x 2 = n(n − 1) x n for |x| < 1. (Hint: F irst div ide the series by x, integrate, factor out x, use
(1 − x)3 ∑
(b) S how that
n=2
p art (a), and differentiate.)
+∞
x( x + 1)
(1 − x)3 ∑
(c) S how that = n 2 x n for |x| < 1.
n =1
+∞ +∞
(d) E v aluate ∑ 2n
n =1
n and ∑
n =1
n2
2n
.

Ans. (d) 2 and 6

 C H A P TE R 4 7

Taylor and M aclaurin Series.

Taylor’s Form ula with R em ainder

Ta y lo r a n d Ma c la u rin Se rie s
L et f be a function that is infinitely differentiable at x = c, that is, the deriv ativ es f (n)(c) ex ist for all p ositiv e
integers n.
The T aylo r series fo r f abo ut c is the p ower series
+∞

∑ a ( x − c)
n=0
n
n
= a0 + a1 ( x − c) + a2 ( x − c)2 + ⋅⋅⋅

f ( n ) ( c)
where an = for all n. N ote that f (0) is tak en to mean the function f itself, so that a0 = f (c).
n!
The Maclaurin series fo r f is the Taylor series for f about 0, that is, the p ower series
+∞

∑a x
n= 0
n
n
= a0 + a1x + a2 x 2 + ⋅⋅⋅

f ( n ) (0)
where an = for all n.
n!
EX AMPLE 4 7 .1 : T h e M a c la u r in s e r ie s fo r s in x
L et f (x) = sin x. Then
f ′( x) = cos x,
f ′′( x) = − sin x,
f ′′′( x) = − cos x,
S ince f (x) = sin x, further deriv ativ es rep eat this cycle of four functions. S ince sin 0 = 0 and cos 0 = 1, f (2k)(0) = 0 and
(4)

(−1) k
f (2k+1)(0) = (−1)k. H ence, a2k = 0 and a2 k +1 = . S o, the M aclaurin series for sin x is
(2 k + 1)!
+∞
(−1) k 3 5 7
∑ (2k + 1)! x
k =0
2 k +1
= x − x + x − x + ⋅⋅⋅
3! 5! 7!

A n ap p lication of the ratio test shows that this series conv erges for all x. W e do not k now that sin x is eq ual
to its M aclaurin series. This will be p rov ed later.

EX AMPLE 4 7 .2 : L et us find the M aclaurin series for f ( x) = 1 .

1− x
f ′( x ) = 1 , f ′′( x) = 2 , f ′′′( x) = 3⋅2 ,
(1 − x)2 (1 − x)3 (1 − x)4

f 4 ( x) = 4 ⋅ 3 ⋅ 25 , f 5 ( x) = 5 ⋅ 4 ⋅ 3 ⋅62
(1 − x) (1 − x)

396
<=
>
?@ A B
CDEFG
GH
I
J
H
H
H
I
J
H
H
GI
J
H
K
F L?#MC
N OP(Q#@ R
S,T UA V V/<=
W,>
R
X
A N
Y3I
Z XP[
<V A P\#C
N
@ N]8=
@9D8N
@ W Y:=
]^
Y3N
[
CHAPTER 47 Taylor and M aclaurin Series 397

W e can see the p attern: f ( n ) ( x) = n! . H ence, a = f ( n ) (0) = 1 for all n, and the M aclaurin series for 1 is
+∞ (1 − x)n+1 n n! 1− x
∑
n= 0
x . In this case, we already k now that
n 1
1− x
is eq ual to its M aclaurin series for |x| < 1.

+∞
f ( n ) ( c)
Th e o re m 4 7 .1 : If f ( x ) = ∑ bn ( x − c)n for some x ≠ c, then this series is the Taylor series for f, that is, bn = for
n=0
n!
+∞
all n. In p articular, if f ( x ) = ∑ bn x n for some x ≠ 0, then this series is the M aclaurin series for f.
n= 0
+∞
A ssume f ( x ) = ∑ bn ( x − c)n for some x ≠ c. Then f (c) = b0. By term-by-term differentiation (Theorem 46.7),
+∞ n=0 +∞
f ′( x ) = ∑ nbn ( x − c)n−1 in the interv al of conv ergence of ∑ bn ( x − c)n . H ence f ′(c) = b1. D ifferentiating again, we
n=0 +∞ n= 0
f ′′(c)
get f ′′( x ) = ∑ n(n − 1)bn ( x − c)n− 2. S o, f ′′(c) = 2b2 and, therefore, b2 = .
n=0
2!
+∞
f ′′′(c)
D ifferentiating again, we get f ′′′( x ) = ∑ n(n − 1)(n − 2)bn ( x − c)n−3. S o, f ′′′(c) = 3! b3 and, therefore, b3 = .
n=0
3!
Iterating this p rocedure, we obtain

f ( n ) ( c)
bn = for all n ≥ 0
n!

Thus, the series is the Taylor series for f.

EX AMPLE 4 7 .3 : W e already k now by formula (46.8) that

+∞ n
ln(1 + x) = ∑ (−1)n−1 x for | x | < 1
n =1
n

n +∞
H ence, by Theorem 47.1, the series ∑ (−1)n−1 x must be the M aclaurin series for ln (1 + x). It is not necessary to
n =1
n
go through the laborious p rocess of comp uting the M aclaurin series for In (1 + x) directly from the definition of
M aclaurin series.

EX AMPLE 4 7 .4 : If f ( x) = 1 , find f (47) (0).

1− x
+∞
f ( n ) (0)
1− x ∑
W e k now that 1 = x for |x| < 1. H ence, by Theorem 47.1, the coefficient of xn, namely 1, is eq ual to
n
.
n=0
n!
( 47 )
f (0)
S o, for n = 47, 1 = and, therefore, f (47)(0) = (47)!.
(47)!

Th e o re m 4 7 .2 (Ta y lo r’s F o rm u la w ith Re m a in d e r): L et f by a function such that its (n + 1)st deriv ativ e f (n+1) ex ists
in (a, b ). A ssume also that c and x are in (a, b). Then there is some x* between c and x such that

f ′′(c) f ( n ) ( c) f ( n+1) ( x* )
f ( x) = f (c) + f ′(c)( x − c) + ( x − c) 2 + ⋅ ⋅ ⋅ + ( x − c) n + ( x − c)n+1
2! n! (n + 1)!
n
f ( k ) ( c)
=∑ ( x − c) k + Rn ( x) (47.1)
k =0
k!

f ( n+1) ( x * )
H ere, Rn ( x) = ( x − c)n+1 is called the remainder term or the erro r.
(n + 1)!
Theorem 47.2 can be deriv ed from Theorem 13.6 (the H igher-O rder L aw of the M ean).
 398 CHAPTER 47 Taylor and M aclaurin Series

Ap p lic a tio n s o f Ta y lo r’s F o rm u la w ith Re m a in d e r

(I ) S h o w in g t h a t c e r t a in f u n c t io n s a r e r e p r e s e n t e d b y t h e ir T a y lo r s e r ie s b y p r o v in g t h a t
lim R n (x ) = 0
n→+∞

F rom Taylor’s formula (47.1),

n
f ( k ) ( c)
Rn ( x) = f ( x) − ∑ ( x − c) k
k =0
k!

If lim Rn ( x) = 0 then
n→+∞

n +∞
f ( k ) ( c) f ( k ) ( c)
f ( x) = lim
n→+ ∞
∑
k =0
k!
( x − c) k = ∑
k =0
k!
( x − c) k

that is, f (x) is eq ual to its Taylor series.

+∞
dn xn
Remark: lim = 0 for any d. To see this, recall that ∑ conv erges for all x. H ence, by Theorem
x n n→+∞ n ! n= 0
n!
43.5, lim = 0 for any x.
n→+∞ n !

EX AMPLE 4 7 .5 : sin x is eq ual to its M aclaurin series.

W hen f (x) = sin x, then ev ery deriv ativ e f (n)(x) is either sin x, cos x, − sin x, or − cos x, and, therefore, | f (n)(x)| ≤ 1.
S o,

f ( n+1) ( x* ) |( x − c)n+1|
|Rn ( x)| = ( x − c)n+1 ≤
(n + 1)! (n + 1)!

x − c)n+1 |
By the R emark abov e, lim | ( = 0 . H ence, nlim Rn ( x) = 0. Therefore, sin x is eq ual to its M aclaurin series:
n→+ ∞ (n + 1)! →+∞

+∞
(−1) k 3 5 7
sin x = ∑ x 2 k +1 = x − x + x − x + ⋅ ⋅ ⋅ (47.2)
k =0
(2 k + 1)! 3! 5! 7!

(I I ) A p p r o x im a tin g v a lu e s o f fu n c tio n s o r in te g r a ls
U se a bound on Rn(x) to get a bound on the error when we ap p rox imate the sum of an infinite series by
a p artial sum.

EX AMPLE 4 7 .6 L et us ap p rox imate e to four decimal p laces, that is, with an error less than 0.00005.
+∞ n
P reliminary result: e < 3. To see this, note that, since e x = ∑ x ,
n=0
n!

+∞
e = e1 = ∑ 1 = 1 + 1 + 1 + 1 + 1 + 1 + ⋅ ⋅ ⋅
n=0
n! 2! 3! 4! 5!

< 1 + 1 + 1 + 12 + 13 + 14 + ⋅ ⋅ ⋅
2 2 2 2
+∞
= 1 + ∑ 1n = 1 + 1 =1+ 2 = 3
n= 0
2 1 − (1/ 2)

N ow, for the function f (x) = ex, we wish to mak e the magnitude of the error Rn(1) < 0.00005. By Taylor’s formula
with remainder, with x = 1,

f ( n+1) ( x* )
|Rn (1)| = , where 0 < x* < 1
(n + 1)!
CHAPTER 47 Taylor and M aclaurin Series 399

S ince D x (e x ) = e x, f ( n+1) ( x) = e x for all x. S o, f ( n+1) ( x * ) = e x . S ince ex is an increasing function, e x < e1 = e < 3. Thus,
* *

| Rn (1)| < 3 . S ince we wish to mak e the error <0.00005, it suffices to hav e
(n + 1)!

3 ≤ 0.00005, that is, 3 ≤ 1 , 60, 000 ≤ (n + 1)!.

(n + 1)! (n + 1)! 20, 000
8
Trial and error shows that this holds for n ≥ 8. S o, we can use the p artial sum ∑ 1 ~ 1.7183.
n=0
n!

Th e o re m 4 7 .3 (Th e Bin o m ia l Se rie s ): A ssume r ≠ 0. Then

+∞
r (r − 1)(r − 2) ⋅ ⋅ ⋅ (r − n + 1) n
(1 + x)r = 1 + ∑ x for | x |< 1
n =1
n!

r (r − 1) 2 r (r − 1)(r − 2) 3
= 1 + rx + x + x + ⋅⋅⋅ (47.3)
2! 3!

A p p ly the ratio test to the giv en series:

sn+1 r (r − 1)(r − 2) ⋅ ⋅ ⋅ (r − n) x n+1 r (r − 1)(rr − 2) ⋅ ⋅ ⋅ (r − n + 1) x n

=
sn (n + 1)! n!

S o,

sn+1 (r − n) x
lim = lim = | x|
n→+∞ sn n→+∞ n +1

H ence, the series conv erges for |x| < 1. F or a sk etch of the p roof that this series is eq ual to (1 + x)r, see
P roblem 31.
N ote that, if r is a p ositiv e integer k, then the coefficients of xn for n > k are 0 and we get the binomial
formula
k
(1 + x) k = ∑ k! xn
n= 0
n!( k − n)!

EX AMPLE 4 7 .7 : L et us ex p and 1 + x as a p ower series about 0. This is the binomial series for r = 12 .

(1/ 2)(−1/ 2) 2 (1/ 2)(−1/ 2)(−3/ 2) 3

1 + x = 1 + 1/ 2 x + x + x
1! 2! 3!
(1/ 2)(−1/ 2)(−3/ 2)(−5/ 2) 4
+ x + ⋅⋅⋅
4!

= 1 + 1 x − 1 x 2 + 1 x3 − 5 x 4 + ⋅ ⋅ ⋅ (47.4)
2 8 16 128

EX AMPLE 4 7 .8 : L et us find a p ower series ex p ansion about 0 for 1

1− x
Tak e the binomial series for r = − , and then rep lace x by − x:
1
2

1 = 1 + −1/ 2 (− x) + (−1/ 2)(−3/ 2) (− x)2 + (−1/ 2)(−3/ 2)(−5/ 2) (− x)3 + ⋅ ⋅ ⋅

1− x 1! 2! 3!

1 ⋅ 3 ⋅ 5 ⋅ ⋅ ⋅ (2n − 1) n
+ x + ⋅⋅⋅
n! 2 n
+∞
1 ⋅ 3 ⋅ 5 ⋅ ⋅ ⋅ (2n − 1) n
=1+ ∑ x (47.5)
n =1
2 ⋅ 4 ⋅ 6 ⋅ ⋅ ⋅ (2n)
 400 CHAPTER 47 Taylor and M aclaurin Series

+∞ +∞ +∞
Th e o re m 4 7 .4 : If f ( x) = ∑ an x n for |x| < R1 and g( x) = ∑ bn x n for |x| < R2, then f ( x) g( x) = ∑ cn x n for |x| < minimum
n n= 0 n= 0 n=0
(R1, R2), where cn = ∑ ak bn− k.
k =0

The reader is referred to more adv anced treatments of analysis for a p roof. Theorem 47.4 guarantees that,
if f and g hav e p ower series ex p ansions, then so does their p roduct.

SOLV ED PROBLEMS

1. F ind a p ower series ex p ansion about 0 for cos x.

W e k now by E x amp le 5 that
+∞
(−1) k 3 5 7
sin x = ∑ x 2 k +1 = x − x + x − x + ⋅ ⋅ ⋅ for all x
k =0
(2 k + 1)! 3! 5! 7!

Then, by Theorem 46.7, we can differentiate term by term:

+∞
(−1) k 2 k 2 4 6
cos x = ∑ x = 1 − x + x − x + ⋅ ⋅ ⋅ for all x
k =0
(2 k )! 2! 4! 6!

2. F ind a p ower series about π for sin x.

2
( )
U se the identity sin x = cos x − π . Then, by P roblem 1,
2

( ) ( ) + 41! ( x − π2 ) − ⋅ ⋅ ⋅
+∞
(−1) k
x−π =1− 1 x − π
2k 2 4
sin x = ∑
k =0
(2 k )! 2 2! 2

3. If f (x) = tan−1 x, ev aluate f (38)(0).

W e k now by formula (46.12) that
+∞ 2 n +1
tan −1 x = ∑ (−1)n x = x − 13 x 3 + 15 x 5 − 17 x 7 + ⋅ ⋅ ⋅ for | x | < 1
n= 0
2 n +1
f (38) (0)
H ence, by Theorem 47.1, the coefficient of x38 in this p ower series is eq ual to . But the coefficient of x38 is 0.
(38)!
S o, f (0) = 0.
(38)

4. F ind p ower series ex p ansions about 0 for the following functions:

(a) cos (x2) (b) xe−2x (c) 1/ 3 1 + x
+∞ +∞
(−1) k 2 k (−1) k 4 k
(a) cos x = ∑ x by P roblem 1. Therefore, cos( x 2 ) = ∑ x .
k =0
(2 k )! k =0
(2 k )!
+∞ +∞
k (−1) k 2 k k
(b) W e k now that e x = ∑ x . S o, e−2 x = ∑ x . H ence,
k =0
k ! k =0
k!
+∞
(−1) k 2 k k +1 +∞ (−1)n−1 2n−1 n
xe−2 x = ∑ x =∑ x
k =0
k! n =1
(n − 1)!

(c) This is the binomial series for r = − 13 .

(−1/ 3)(−4 / 3) 2 (−1/ 3)(−4 / 3)(−

−7 / 3) 3
1/ 3 1 + x = 1 − 1 x + x + x
3 2! 3!
(−1/ 3)(−4 / 3)(−7 / 3)(−10 / 3) 4
+ x + ⋅⋅⋅
4!
+∞
(−1)n (1 ⋅ 4 ⋅ 7 ⋅ ⋅ ⋅ (3n − 2)) n
=1+ ∑ x
n =1
3n n !
CHAPTER 47 Taylor and M aclaurin Series 401

5. F ind the first fiv e terms of the M aclaurin series for ex(sin x).
Metho d 1: L et f (x) = ex(sin x). Then

f ′( x) = e x (sin x + cos x), f ′′( x) = 2e x (cos x), f ′′′( x) = 2e x (cos x − sin x)

f ( 4 ) ( x) = −4 e x (sin x), and f (5) ( x) = −4 e x (sin x + cos x)

f ( n ) (0)
H ence, since an = , we get a0 = 0, a1 = 1, a2 = 1, a3 = 13 , a4 = 0 , and a5 = − 301 . Thus
n!
3 5
e x (sin x) = x + x 2 + x − x + ⋅ ⋅ ⋅
3 30

( 2 3 3 5
)( )
Metho d 2: e x (sin x) = 1 + x + x + x + ⋅ ⋅ ⋅ x − x + x − ⋅ ⋅ ⋅ . If we multip ly out according to the rule in
2! 3! 3! 5!
Theorem 47.4, we get the same result as abov e. F or ex amp le, c5 = 241 − 121 + 120
1
= − 301 .

3 5
6. W e k now that sin x = x − x + x − ⋅ ⋅ ⋅. F or what v alues of x will ap p rox imating sin x by x p roduce an error < 0.005?
3! 5!
f (3) ( x* ) 3 | x |3
| R2 ( x)| = x ≤ | f (3) ( x)| ≤ 1 since f (3) is −cos x.) S o, we req uire | x |3/6 < 0.005, which is
3! 6 . (H ere,
eq uiv alent to |x|3 < 0.03. S o, we want | x | < 3 0.03 ~ 0.31.

3
7. If we ap p rox imate sin x by x − x for |x| < 0.5, what is a bound on the error?
3!
S ince sin x is eq ual to an alternating series for any x, the error will be less than the magnitude of the first term
omitted, in this case |x|5/5!. W hen |x| < 0.5, the error will be less than 1 (0.5)5 ~ 0.00026.
120
1
sin x dx with an error less than 0.005.
8. A p p rox imate ∫
0 x
+∞
(−1) k 3 5 7
sin x = ∑ x 2 k +1 = x − x + x − x + ⋅ ⋅ ⋅
k =0
( 2 k + 1 )! 3! 5! 7!

+∞
(−1) k 2 4 6
H ence, sin x =
x ∑
k =0
( 2 k + 1 )!
x2k = 1 − x + x − x + ⋅ ⋅ ⋅
3! 5! 7!

1
+∞
(−1) k +∞
(−1) k x 2 k +1 
∑ ∑
1 1
sin x dx =
∫0 x k =0
( 2 k + 1 )! ∫0
x 2k
dx =
k =0
( 2 k + 1)! 2 k + 1 
0
Therefore,
+∞
(−1) k
=∑ 1
k =0
(2 k + 1)! 2 k + 1

This is an alternating series. W e must find k so that 1 1 ≤ 0.005, or, eq uiv alently, 200 ≤ (2k + 1)!(2k + 1).
(2 k + 1)! 2 k + 1
It is true for k ≥ 2. S o, we need 1 − 181 = 17
18 ~ 0.944.

9. F ind a p ower series about 0 for sin−1 x.

By formula (47.5),
+∞
1 = 1 + 1 ⋅ 3 ⋅ 5 ⋅ ⋅ ⋅ (2n − 1) x n
1− x
∑
n =1
2 ⋅ 4 ⋅ 6 ⋅ ⋅ ⋅ (2n)
forr | x | < 1

R ep lace x by t2.
+∞
1 ⋅ 3 ⋅ 5 ⋅ ⋅ ⋅ (2n − 1) 2 n
1 =1+ ∑ t for | t | < 1
1− t 2
n =1
2 ⋅ 4 ⋅ 6 ⋅ ⋅ ⋅ (2n)
 402 CHAPTER 47 Taylor and M aclaurin Series

S o, for |x| < 1,

+∞
1 dt = x + 1 ⋅ 3 ⋅ 5 ⋅ ⋅ ⋅ (2n − 1) x 2 n+1
∑
x
sin −1 x = ∫
0 1 − t2 n =1
2 ⋅ 4 ⋅ 6 ⋅ ⋅ ⋅ (2n) 2n + 1

1 0 . F ind M aclaurin series for the following functions: (a) sin(x3); (b) sin2 x.
R ecall that, if a function has a p ower series ex p ansion in an interv al about 0, then that p ower series is the
M aclaurin series of the function.
+∞ +∞
(−1) k (−1) k
(a) sin x = ∑ x 2 k +1 for all x. H ence, sin( x 3 ) = ∑ x 6 k +3 and this series is the M aclaurin series for
k =0
( 2 k + 1 )! k =0
( 2 k + 1)!
sin (x3).
1 − cos(2 x) 1  +∞
(−1) k 22 k 2 k  +∞ (−1) k +1 22 k −1 2 k
(b) sin 2 x = = 1 − ∑ x  =∑ x by P roblem 1. S o, the M aclaurin series for
2 2 k =0
(2 k )!  k =1 (2 k )!
+∞
(−1) k +1 22 k −1 2 k
sin2 x is ∑ x .
k =1
(2 k )!

1 1 . F ind the first four nonz ero terms of the M aclaurin series for f (x) = sec x.
It would be v ery tedious to comp ute the successiv e deriv ativ es. Instead, since sec x cos x = 1, we can p roceed
+∞
differently. W e assume sec x = ∑ an x n . Then
n= 0

 +∞   +∞ (−1) k 2 k 
 ∑ an x   ∑ (2 k )! x  = 1
n

n=0 k =0

2
( 4 6
(a0 + a1 x + a2 x 2 + a3 x 3 + ⋅ ⋅ ⋅) 1 − x + x − x + ⋅ ⋅ ⋅ = 1
2 24 720 )
W e then “ multip ly out,” comp are coefficients on the two sides of the eq uation, and solv e for the an.

a0 = 1, a1 = 0, a2 = 12 ; a3 = 0; a4 = 5
24 ; a5 = 0; a6 = 61
720

Thus,

sec x = 1 + 1 x 2 + 5 x 4 + 61 x 6 + ⋅ ⋅ ⋅
2 24 270
2 4 6
A n alternativ e method would be to carry out a “ long div ision” of 1 by 1 − x + x − x + ⋅ ⋅ ⋅
2 24 720

SUPPLEMENTARY PROBLEMS

1 2. F ind the M aclaurin series for the following functions:

(a) sin (x5); (b) 1 5 ; (c) cos2 x.
1+ x
+∞ +∞ +∞
(−1) k (−1) k 22 k −1 2 k
Ans. (a) ∑ (2k + 1)! x
k =0
10 k + 5
; (b) ∑ (−1)
k =0
n
x 5n ; (c) 1 + ∑
k =1
(2 k )!
x

1 3 . F ind the Taylor series for ln x about 2.

+∞
( x − 2)n
Ans. ln 2 + ∑ (−1)n−1
n =1
n2n

1 4 . F ind the first three nonz ero terms of the M aclaurin series for (a) sinx x ; (b) ex cos x.
e

Ans. (a) x − x 2 + 13 x 3 + ⋅ ⋅ ⋅ ; (b) 1 + x − 13 x 3 + ⋅ ⋅ ⋅

CHAPTER 47 Taylor and M aclaurin Series 403

1 5 . C omp ute the first three nonz ero terms of the M aclaurin series for tan x.

Ans. x + 13 x 3 + 2
15 x5 + ⋅ ⋅ ⋅

1 6 . C omp ute the first three nonz ero terms of the M aclaurin series for sin−1 x.

Ans. x + 16 x 3 + 3
40 x5 + ⋅ ⋅ ⋅

3 3 3 ( ( ))
1 7 . F ind the Taylor series for cos x about π . [ Hint: U se an identity for cos π + x − π .]

( ) ( )
+∞ +∞ 2 k +1
(−1) (−1)
x−π x−π
k 2k k
3
Ans. 1 ∑
2 ∑
−
2 k = 0 (2 k )! 3 k =0
(2 k + 1)! 3

1 8 . (G C ) U se p ower series to ap p rox imate ∫

1/ 2
tan −1 x dx
0 x

Ans. 0.4872

1/ 2 ln(1 + x)
1 9 . (G C ) U se p ower series to ap p rox imate ∫ dx correctly to four decimal p laces.
0 x

Ans. 0.4484

1
20 . (G C ) U se p ower series to ap p rox imate ∫ 3
1 + x 2 dx correctly to four decimal p laces.
0

Ans. 1.0948

21 . (G C ) W hat is a bound on the error if we ap p rox imate ex by 1 + x + 12 x 2 for |x| ≤ 0.05? (Y ou may use e0.05 < 1.06.)

Ans. 0.0000221

22. (G C ) W hat is a bound on the error if we ap p rox imate ln (1 + x) by x for |x| ≤ 0.05?

Ans. 0.00125

23 . (G C ) U se the Taylor series for sin x about π to ap p rox imate sin 62° correctly to fiv e decimal p laces.
3

Ans. 0.88295

24 . (G C ) In what interv al can you choose the angle if the v alues of cos x are to be comp uted using three terms of its
Taylor series about π and the error must not ex ceed 0.000 05?
3

Ans. x − π ≤ 0.0669
3

25 . (G C ) U se p ower series to comp ute to four-decimal-p lace accuracy: (a) e−2; (b) sin 32° ; (c) cos 36° .

Ans. (a) 0.1353; (b) 0.5299; (c) 0.8090

26 . (G C ) F or what range of x can:

(a) ex be rep laced by 1 + x + 12 x 2 if the allowable error is 0.0005?
(b) sin x be rep laced by x − 16 x 3 + 1201
x 5 if the allowable error is 0.00005?

Ans. (a) |x| < 0.1; (b) |x| < 47°

 404 CHAPTER 47 Taylor and M aclaurin Series

27 . U se p ower series to ev aluate: (a) lim e − e3 ; (b) lim e − e2

sin x cos x
.
x→0 x x→0 x

Ans. (a) 1 ; (b) e

6 2

28 . (G C ) U se p ower series to ev aluate:

π /2
(a) ∫
0
(1 − 12 sin 2 x)−1/ 2 dx (to three-decimal-p lace accuracy).
1
(b) ∫ cos
0
x dx (to fiv e-decimal-p lace accuracy).
1/ 2
dx (to four-decimal-p lace accuracy).
(c) ∫
0 1 + x4

Ans. (a) 1.854; (b) 0.76355; (c) 0.4940

29 . (G C ) U se p ower series to ap p rox imate the length of the curv e y = 13 x 3 from x = 0 to x = .5, with four-decimal-
p lace accuracy.

Ans. 0.5031

3 0 . (G C ) U se p ower series to ap p rox imate the area between the curv e y = sin (x2) and the x ax is from x = 0 to x = 1,
with four-decimal-p lace accuracy.

Ans. 0.3103

3 1 . P rov e that the binomial series ex p ansion in Theorem 47.3 is correct.

+∞
Hint: L et y = 1 + ∑ r (r − 1)(r − 2) ⋅ ⋅ ⋅ (r − n + 1) x n . U se term-by-term differentiation to find the series for
dy
and
n =1
n ! dx
dy ry dy r dx
. Then deriv e y = (1 + x)r. [U se “ sep aration of v ariables” ; ∫
y ∫ 1+ x
show that = = .]
dx 1 + x

3.2
3 2. E x p and the p olynomial f (x) = x4 − 11x3 + 43x2 − 60x + 14 as a p ower series about 3, and find ∫ f ( x) dx.
3

Ans. 1.185
 C H A P TE R 4 8

PartialD erivatives

F u n c tio n s o f Se v e ra l V a ria b le s
If a real number z is assigned to each p oint (x, y) of a p art of the xy p lane, then z is said to be giv en as a
function, z = f (x, y), of the indep endent v ariables x and y. The set of all p oints (x, y, z) satisfying z = f (x, y)
is a surface in three-dimensional sp ace. In a similar manner, functions w = f (x, y, z, . . .) of three or more
indep endent v ariables may be defined, but no geometric p icture is av ailable.
There are a number of differences between the calculus of one and two v ariables. H owev er, the calculus
of functions of three or more v ariables differs only slightly from that of functions of two v ariables. The study
here will be limited largely to functions of two v ariables.

Lim its
By an o pen disk with center (a, b) we mean the set of p oints (x, y) within some fix ed distance d from (a, b),
that is, such that ( x − a)2 + ( y − b)2 < δ . By a deleted disk around (a, b) we mean an op en disk without its
center (a, b).
L et f be a function of two v ariables and assume that there are p oints in the domain of f arbitrarily close to
(a, b). To say that f (x, y) has the limit L as (x, y) ap p roaches (a, b) means intuitiv ely that f (x, y) can be made
arbitrarily close to L when (x, y) is sufficiently close to (a, b). M ore p recisely,

lim f ( x , y) = L
( x , y )→( a ,b )

if, for any  > 0, there ex ists d > 0 such that, for any (x, y) in the domain of f and in the deleted disk of radius
d around (a, b), | f (x, y) − L| < . This is eq uiv alent to saying that, for any  > 0, there ex ists d > 0 such that
0 < ( x − a)2 + ( y − b)2 < δ imp lies | f (x, y) − L| <  for any (x, y) in the domain of f. N ote that it is not as-
sumed that f (a, b) is defined.
L aws for limits analogous to those for functions of one v ariable (Theorems 7.1– 7.6) also hold here and
with similar p roofs.

EX AMPLE 4 8 .1 : U sing these standard laws for limits, we see that

 3xy 2 + 1 xy = 3(3)(1) + 1 (3)(1) = 9 + 3 = 21

lim 
( x , y )→( 3,1)  7 + y 2  7 +1 2 8 2 8

EX AMPLE 4 8 .2 : In some cases, these standard laws do not suffice.

3xy 2 0 , which is indeterminate. S o, we need a
L et us show that lim 2 = 0. O ur usual limit rules would yield
( x , y )→ ( 0 ,0 ) x + y
2
0
more inv olv ed argument. A ssume  > 0. N ow,

3xy 2 3xy 2 y2
2 − 0 = 2 = 3 | x| ≤ 3 | x | ≤ 3 x 2 + y 2 < 3δ = 
x +y
2
x +y
2
x + y2
2

if we choose d =  /3 and we assume that 0 < x 2 + y 2 < δ .

405
<=
>
?@ A B
CDEFG
GH
I
J
H
H
H
I
J
H
H
GI
J
H
K
F L?#MC
N OP(Q#@ R
S,T UA V V/<=
W,>
R
X
A N
Y3I
Z XP[
<V A P\#C
N
@ N]8=
@9D8N
@ W Y:=
]^
Y3N
[
 406 CHAPTER 48 Partial D erivatives

x 2 − y2
EX AMPLE 4 8 .3 : L et us show that lim 2 does not ex ist.
( x , y )→ ( 0 ,0 ) x + y
2

x 2 − y2 x 2
L et ( x, y) → (0, 0) along the x ax is, where y = 0. Then 2 = = 1. S o, the limit along the x ax is is 1. N ow
x + y2 x 2
x −y
2 2
y 2
let ( x, y) → (0, 0) along the y ax is, where x = 0. Then 2 = − 2 = −1. S o, the limit along the y ax is is −1. H ence,
x + y2 y
there can be no common limit as one ap p roaches (0, 0), and the limit does not ex ist.

2
 x 2 − y2 
EX AMPLE 4 8 .4 : L et us show that lim  2
( x , y )→ ( 0 , 0 )  x + y 
2  does not ex ist.
2
x 2 − y2 
H ere, we cannot use the same argument as in E x amp le 3, since  2 2  ap p roaches 1 as (x, y) ap p roaches
x +y 
(0, 0) along both the x ax is and the y ax is. H owev er, we can let (x, y) ap p roach (0, 0) along the line y = x. Then

( )
2 2
 x 2 − y 2  = x 2 − x 2 = 0. S o,  x 2 − y 2  → 0 along y = x. S ince this is different from the limit 1 ap p roached along
2

 x 2 + y 2  x2 + x2  x 2 + y 2 
the x ax is, there is no limit as ( x, y) → (0, 0).

C o n tin u ity
L et f be a function of two v ariables and assume that there are p oints in the domain of f arbitrarily close
to (a, b). Then f is co ntinuo us at (a, b) if and only if f is defined at (a, b), lim f ( x, y) ex ists, and
( x , y )→ ( a , b )
lim f ( x, y) = f (a, b)
( x , y )→ ( a , b )

W e say that f is continuous on a set A if f is continuous at each p oint of A.

This is a generaliz ation to two v ariables of the definition of continuity for functions of one v ariable. The
basic p rop erties of continuous functions of one v ariable (Theorem 8.1) carry ov er easily to two v ariables.
In addition, ev ery p olynomial in two v ariables, such as 7 x 5 − 3xy 3 − y 4 + 2 xy 2 + 5, is continuous at all p oints.
E v ery continuous function of one v ariable is also continuous as a function of two v ariables.
The notions of limit and continuity hav e obv ious generaliz ations to functions of three or more v ariables.

Pa rtia l D e riv a tiv e s

L et z = f (x, y) be a function of two v ariables. If x v aries while y is held fix ed, z becomes a function of x. Then
its deriv ativ e with resp ect to x

f ( x + ∆ x , y) − f ( x , y)
lim
∆ x→0 ∆x
∂z ∂f
is called the (first) partial deriv ativ e of f with resp ect to x and is denoted fx (x, y) or or .
∂x ∂x
S imilarly, if y v aries while x is held fix ed, the (first) partial deriv ativ e of f with resp ect to y is

∂z ∂f f ( x, y + ∆y) − f ( x, y)
f y ( x , y) = = = lim
∂y ∂y ∆y→0 ∆y

EX AMPLE 4 8 .5 : L et f (x, y) = x2 sin y. Then fx(x, y) = 2x sin y and fy(x, y) = x2 cos y.

N ote that, when fx is comp uted, y is temp orarily treated lik e a constant, and, when fy is comp uted, x is temp orarily
treated lik e a constant.
The p artial deriv ativ es hav e simp le geometric interp retations. C onsider the surface z = f (x, y) in F ig. 48-1. Through
the p oint P(x, y, z), there is a curv e APB that is the intersection with the surface of the p lane through P p arallel to the
xz p lane (the p lane determined by the x ax is and the z ax is). S imilarly, CPD is the curv e through P that is the intersec-
tion with the surface z = f (x, y) of the p lane through P that is p arallel to the yz p lane. A s x v aries while y is held fix ed,
P mov es along the curv e APB , and the v alue of ∂z at (x, y) is the slop e of the tangent line to the curv e APB at P.
∂x
S imilarly, as y v aries while x is held fix ed, P mov es along the curv e CPD , and the v alue of ∂z at (x, y) is the slop e of
∂y
the tangent line to the curv e CPD at P.
CHAPTER 48 Partial D erivatives 407

Fig. 48-1

Pa rtia l D e riv a tiv e s o f H ig h e r Ord e r

∂z
W e can tak e the p artial deriv ativ es with resp ect to x and y of , yielding
∂x
∂2 z ∂  ∂z  ∂2 z ∂  ∂z 
2 = f xx ( x , y) =  = fyx ( x, y) =
 ∂y  ∂x 
and
∂x ∂x  ∂x  ∂y ∂x
∂z
S imilarly, from we obtain
∂y
∂2 z ∂  ∂z  ∂2 z ∂  ∂z 
2 = f yy ( x , y) =  = fxy ( x, y) =
 ∂x  ∂y 
and
∂y ∂y  ∂y  ∂x ∂y

Th e o re m 4 8 .1 : A ssume that fxy and fyx ex ist and are continuous in an op en disk . Then fxy = fyx at ev ery p oint of the disk .
F or a p roof, see P roblem 30.

EX AMPLE 4 8 .6 : L et us v erify Theorem 48.1 for f (x, y) = x2(sin yx).

fx ( x, y) = x 2 (cos yx)( y) + 2 x(sin yx) = x[ xy(cos yx) + 2 sin yx]

fy ( x, y) = x 2 (cos yx) x + x 3 (cos yx)

fyx ( x, y) = x[ x( y(− sin yx)( x) + cos yx) + 2(cos yx)( x)]

= x 2 [− xy sin yx + 3 cos yx]

fxy ( x, y) = x 3 (− sin yx)( y) + 3x 2 cos yx = x 2 [− xy sin yx + 3 cos yx]

P artial deriv ativ es also can be defined for functions of three or more v ariables. A n analogue of Theorem 48.1
holds for any two orderings of giv en subscrip ts.
N ote that p artial deriv ativ es may fail to ex ist when the req uired limits do not ex ist.

SOLV ED PROBLEMS
x − y
1. E v aluate: (a) lim (2 xy 4 − 7 x 2 y 2 ); (b) x cos  .
lim
( x , y )→ ( 3 ,2 )  4 
( x , y )→ ( π ,0 )

S ince the standard limit laws ap p ly, the limits are:

(a) 2(3)(2)4 − 7(3)2(2)2 = 96 − 252 = −156; (b) π cos π = π 2
4 2

2. E v aluate lim x2 .
( x , y )→ ( 0 ,0 ) x + y
2 2

A s ( x, y) → (0, 0) along the y ax is, x = 0 and x 2 = 0 → 0.

x 2 + y2
 408 CHAPTER 48 Partial D erivatives

2 2
A s ( x, y) → (0, 0) along the x ax is, y = 0 and 2 x 2 = x 2 = 1 → 1.
x +y x
H ence, the limit does not ex ist.

xy
3. E v aluate lim .
( x , y )→ ( 0 ,0 ) x 2 + y2
xy
S ince |x | = x 2 ≤ x 2 + y 2 , ≤ | y| → 0 as ( x, y) → (0, 0). S o,
x 2 + y2
xy
lim = 0.
( x , y )→ ( 0 , 0 ) x 2 + y2

sin( x + y)
4. The function f ( x, y) = is continuous ev erywhere ex cep t at (0, 0) and on the line y = −x, where it is not
x+y
defined. C an f (0, 0) be defined so that the new function is continuous?
sin( x + y)
A s ( x, y) → (0, 0), x + y → 0 and, therefore, → 1, since lim sin u = 1. S o, if we let f (0, 0) = 1, the
x+y u→0 u
new function will be continuous at (0, 0). Thus, the original discontinuity was remov able.

In P roblems 5– 9, find the first p artial deriv ativ es.

5. z = 2x2 − 3xy + 4y2.

Treating y as a constant and differentiating with resp ect to x yields ∂z = 4 x − 3y .
∂x
Treating x as a constant and differentiating with resp ect to y yields ∂z = −3x + 8 y .
∂y
2 y2
6. z= x + .
y x
Treating y as a constant and differentiating with resp ect to x yields ∂z = 2 x − 2 .
y2
∂x y x
Treating x as a constant and differentiating with resp ect to y yields ∂z = − x 2 + .
2 2y
∂y y x

7. z = sin (2x + 3y).

∂z = 2 cos(2 x + 3y) and ∂z = 3 cos(2 x + 3y)
∂x ∂y

8. z = tan −1 ( x 2 y) + tan −1 ( xy 2 ).

∂z = 2 xy + y 2 and ∂z = x 2 + 2 xy
∂x 1 + x 4 y 2 1 + x 2 y 4 ∂y 1 + x 4 y 2 1 + x 2 y 4

z = ex
2 + xy
9.
∂z = e x 2 + xy (2 x + y) and ∂z = xe x 2 + xy
∂x ∂y

1 0 . The area of a triangle is giv en by K = 12 ab sin C . W hen a = 20, b = 30, and C = 30° , find:
(a) The rate of change of K with resp ect to a, when b and C are constant.
(b) The rate of change of K with resp ect to C, when a and b are constant.
(c) The rate of change of b with resp ect to a, when K and C are constant.

(a) ∂K = 12 b sin C = 12 (30)(sin 30° ) = 152

∂a
(b) ∂K = 12 ab cos C = 12 (20)(30)(cos 30° ) = 150 3
∂C
(c) b = 2 K and ∂b = − 22 K = − 2 2
2( 1 ab sin C )
=−b =−3
a sin C ∂a a sin C a sin C a 2
CHAPTER 48 Partial D erivatives 409

In P roblems 11– 13, find the first p artial deriv ativ es of z with resp ect to the indep endent v ariables x and y.

1 1 . x2 + y2 + z2 = 25. [This is the eq uation of a sp here of radius 5 and center (0, 0, 0).]
D ifferentiate imp licitly with resp ect to x, treating y as a constant, to obtain:

2 x + 2 z ∂z = 0. H ence, ∂z = − x
∂x ∂x z

D ifferentiate imp licitly with resp ect to y, treating x as a constant:

2 y + 2 z ∂z = 0. H ence, ∂z = − y
∂y ∂y z

1 2. x2(2y + 3z) + y2(3x − 4z) + z2(x − 2y) = xyz.

D ifferentiate imp licitly with resp ect to x:

2 x(2 y + 3z) + 3x 2 ∂z + 3y 2 − 4 y 2 ∂z + 2 z( x − 2 y) ∂z + z 2 = yz + xy ∂z
∂x ∂x ∂x ∂x

4 xy + 6 xz + 3y 2 + z 2 − yz
S olv ing for ∂z yields: ∂z = − 2 .
∂x ∂x 3x − 4 y 2 + 2 xz − 4 yz − xy
D ifferentiate imp licitly with resp ect to y:

2 x 2 + 3x 2 ∂z + 2 y(3x − 4 z) − 4 y 2 ∂z + 2 z( x − 2 y) ∂z − 2 z 2 = xz + xy ∂z
∂y ∂y ∂y ∂y

2 x 2 + 6 xy − 8 yz − 2 z 2 − xz
S olv ing for ∂z yields: ∂z = − 2 .
∂y ∂y 3x − 4 y 2 + 2 xz − 4 yz − xy

1 3 . xy + yz + zx = 1.
D ifferentiating with resp ect to x yields y + y ∂z + x ∂z + z = 0 , whence ∂z = − y + z .
∂x ∂x ∂x x+y
D ifferentiating with resp ect to y yields x + y + z + x ∂z = 0 , whence
∂z ∂z = − x + z .
∂y ∂y ∂y x+y

1 4 . C onsidering x and y as indep endent v ariables, find ∂r , ∂r , ∂θ , ∂θ when x = e2r cos q, y = e3r sin q.
∂x ∂y ∂x ∂y
F irst differentiate the giv en relations with resp ect to x:

1 = 2e2 r cosθ ∂r − e2 r sin θ ∂θ and 0 = 3e3r sin θ ∂r + e3r cosθ ∂θ

∂x ∂x ∂x ∂x

Then solv e simultaneously to obtain ∂r = 2 r cosθ 2 and ∂θ = − 2 r 3 sin θ 2 .

∂x e (2 + sin θ ) ∂x e (2 + sin θ )
N ow differentiate the giv en relations with resp ect to y:

0 = 2e2 r cosθ ∂r − e2 r sin θ ∂θ and 1 = 3e3r sin θ ∂r + e3r cosθ ∂θ

∂y ∂y ∂y ∂y

Then solv e simultaneously to obtain ∂r = 3r sin θ 2 and ∂θ = 3r 2 cosθ 2 .

∂y e (2 + sin θ ) ∂y e (2 + sin θ )

1 5 . F ind the slop es of the tangent lines to the curv es cut from the surface z = 3x2 + 4y2 − 6 by p lanes through the
p oint (1, 1, 1) and p arallel to the xz and yz p lanes.
The p lane x = 1, p arallel to the yz p lane, intersects the surface in the curv e z = 4y2 − 3, x = 1. Then
∂z = 8 y = 8(1) = 8 is the req uired slop e.
∂y
The p lane y = 1, p arallel to the xz p lane, intersects the surface in the curv e z = 3x2 + 2, y = 1. Then ∂z = 6 x = 6
∂x
is the req uired slop e.
 410 CHAPTER 48 Partial D erivatives

In P roblems 16 and 17, find all second p artial deriv ativ es of z and v erify Theorem 48.1.

1 6 . z = x2 + 3xy + y2.
∂z = 2 x + 3y,
∂x ( )
∂ 2 z = ∂ ∂z = 2,
∂x 2 ∂x ∂x ∂y ∂x ∂y ∂x ( )
∂ 2 z = ∂ ∂z = 3

∂z = 3x + 2 y, ∂ 2 z = ∂  ∂z  = 2, ∂ 2 z = ∂  ∂z  = 3
∂y ∂y 2 ∂y  ∂y  ∂x ∂y ∂x  ∂y 

N ote that ∂ z = ∂ z .
2 2

∂y ∂x ∂x ∂y

1 7 . z = x cos y − y cos x.

∂z = cos y + y sin x,
∂x ( )
∂ 2 z = ∂ ∂z = y cos x
∂x 2 ∂x ∂x

( )
∂ 2 z = ∂ ∂z = − sin y + sin x
∂y ∂x ∂y ∂x

∂z = − x sin y − cos x, ∂ 2 z = ∂  ∂z  = − x cos y

∂y ∂y 2 ∂y  ∂y 

∂ 2 z = ∂  ∂z  = − sin y + sin x
∂x ∂y ∂x  ∂y 

N ote that ∂ z = ∂ z .
2 2

∂y ∂x ∂x ∂y

1 8 . L et f (x, y, z) = x cos (y z). F ind all p artial deriv ativ es of the first, second, and third order.
fx = cos (yz), fxx = 0, fyx = −z sin (yz), fzx = −y sin (yz)
fy = −xz sin (yz), fyy = −xz cos (yz),
2
fxy = −z sin (yz)
fzy = −x(zy cos (yz) + sin( yz))
fz = −xy sin (yz), fzz = −xy2 cos (yz), fxz = −y sin (yz)
fyz = −x(zy cos (yz) + sin (yz))
N ote that fxy = fyx and fxz = fzx and fyz = fzy.
fxxx = 0, fxxy = fxyx = 0, fxxz = fxzx = 0
fxyy = −z2 cos (yz), fxyz = fxzy = −(zy cos (yz) + sin (yz))
fxzz = −y2 cos (yz)
fyyy = xz3 sin (yz), fyxx = 0, fyxy = fyyx = −z2 cos (yz)
fyxz = fyzx = −(yz cos (yz) + sin (yz))
fyyz = fyzy = −x(−z2y sin (yz) + z cos (yz) + z cos (yz))
= xz(zy sin (yz) − 2 cos (yz))
fyzz = −x(−y2z sin (yz) + 2y cos (yz))
= xy(z sin (yz) − 2 cos (yz))
fzzz = xy3 sin (yz), fzxx = 0, fzxy = fzyx = −(zy cos (yz) + sin (yz))
fzxz = fzzx = −y cos (yz)
2

fzyy = −x(−z2y sin (yz) + 2z cos (yz)) = xz(zy sin (yz) − 2 cos (yz))
fzyz = fzzy = −x(−zy2 sin (yz) + y cos (yz) + y cos (yz))
= xy(zy sin (yz) − 2 cos (yz))
N ote that, in the third order, any two rearrangements of subscrip ts will be eq ual. F or ex amp le, fxyz = fxzy = fyxz =
fyzx = fzxy = fzyx = −(zy cos (yz) + sin (yz)).
CHAPTER 48 Partial D erivatives 411

1 9 . D etermine whether the following functions are solutions of L ap lace’s eq uation ∂ z2 + ∂ z2 = 0 :

2 2

∂x ∂y
(a) z = ex cos y (b) z = 12 (e x + y ) (c) z = x2 − y2

(a) ∂z = e x cos y, ∂ 2 z = e x cos y

∂x ∂x 2
∂z = − e x sin y, ∂ 2 z = −e x cos y
∂y ∂y 2

Then ∂ z2 + ∂ z2 = 0.
2 2

∂x ∂y
(b) ∂z = 1 (e x + y ), ∂ 2 z = 1 (e x + y )
∂x 2 ∂x 2 2
∂z = 1 (e x + y ), ∂ 2 z = 1 (e x + y )
∂y 2 ∂y 2 2

S o, ∂ z2 + ∂ z2 = e x + y ≠ 0 .
2 2

∂x ∂y
∂z ∂2 z = 2
(c) ∂x = 2 x, ∂x 2
∂z = −2 y, ∂ 2 z = −2
∂y ∂y 2

S o, ∂ z2 + ∂ z2 = 0.
2 2

∂x ∂y

SUPPLEMENTARY PROBLEMS

In P roblems 20– 24, ev aluate the giv en limit.

x − 2y
20 . lim Ans. − 53
( x , y )→( −1, 2 ) x2 + y

x−y
21 . lim Ans. no limit
( x , y )→ ( 0 , 0 ) x 2 + y2

3xy
22. lim Ans. no limit
( x , y )→ ( 0 , 0 ) 2 x 2 + y2

xy 2
23 . lim Ans. no limit
( x , y )→ ( 0 , 0 ) x 2 + y4

x 2 + y2
24 . lim Ans. 4
( x , y )→ ( 0 , 0 ) x + y2 + 4 − 2
2

25 . D etermine whether each of the following functions can be defined at (0, 0) so as to be continuous:
y2 x−y x 3 + y3 x+y
(a) (b) (c) (d)
x 2 + y2 x+y x 2 + y2 x 2 + y2

Ans. (a) no; (b) no; (c) yes; (d) no

 412 CHAPTER 48 Partial D erivatives

26 . F or each of the following functions z, find ∂z and ∂z .

∂x ∂y
(a) z = x2 + 3xy + y2 Ans. ∂z = 2 x + 3y; ∂z = 3x + 2 y
∂x ∂y
y
(b) z = x2 − 2 Ans. ∂z = 1 + 2 y ; ∂z = − 2 x − 1
y x ∂x y 2 x 3 ∂y y3 x 2
(c) z = sin 3x cos 4y Ans. ∂z = 3 cos 3x cos 4 y ; ∂z = −4 sin 3x sin 4 y
∂x ∂y
(d) z = tan −1  
y
Ans. ∂z = − y ; ∂z = x
 x ∂x x 2 + y 2 ∂y x 2 + y 2
(e) x2 − 4y2 + 9z2 = 36 Ans. ∂z = − x ; ∂z = 4 y
∂x 9 z ∂y 9 z

(f ) z3 − 3x2y + 6xyz = 0 Ans. ∂z = 2 y( x − z) ; ∂z = x( x − 2 z)

∂x z 2 + 2 xy ∂y z 2 + 2 xy
(g) yz + xz + xy = 0 Ans. ∂z = − y + z ; ∂z = − x + z .
∂x x + y ∂y x+y

27 . F or each of the following functions z, find ∂ z2 , ∂ z , ∂ z , and ∂ z2 .

2 2 2 2

∂x ∂y ∂x ∂x ∂y ∂y
(a) z = 2x2 − 5xy + y2 Ans. ∂ 2 z = 4; ∂ 2 z = ∂ 2 z = −5; ∂ 2 z = 2
∂x 2 ∂x ∂y ∂y ∂x ∂y 2
y
(b) z = x2 − 2 ∂ z = − 6 y ; ∂ z = ∂ z = 2  1 − 1  ; ∂2 z = 6 x
2 2 2

y x
Ans.
∂x 2 x 4 ∂x ∂y ∂y ∂x  x 3 y3  ∂y 2 y 4

(c) z = sin 3x cos 4y Ans. ∂ 2 z = −9 z ; ∂ 2 z = ∂ 2 z = −12 cos 3x sin 4 y ; ∂ 2 z = −16 z

∂x 2 ∂x ∂y ∂y ∂x ∂y 2
(d) z = tan −1  
y ∂ z =−∂ z =
2 2 2 xy ∂ z = ∂ z = y − x2
2 2 2
Ans. 2 2;
 x ∂x 2
∂y 2
( x + y ) ∂x ∂y ∂y ∂x ( x 2 + y 2 )2
2

, show that x 2 ∂ z2 + 2 xy ∂ z + y 2 ∂ z2 = 0 .
xy 2 2 2
28 . (a) If z =
x−y ∂x ∂x ∂y ∂y
αx ∂ 2
z ∂ 2
(b) If z = e cos β y and β = ± α , show that 2 + z2 = 0 .
∂x ∂y
(c) If z = e−t(sin x + cos y), show that ∂ z2 + ∂ z2 = ∂z .
2 2

∂x ∂y ∂t
 2 
(d) If z = sin ax sin by sin kt a 2 + b 2 , show that ∂ z2 = k 2  ∂ z2 + ∂ z2  .
2 2

∂t  ∂x ∂y 

( )
29 . F or the gas formula p + a2 (v − b) = ct , where a, b, and c are constants, show that
v
∂p 2a(v − b) − ( p + a /v 2 )v 3 ∂v = cv 3
= ,
∂v v 3 (v − b) ∂t ( p + a /v 2 )v 3 − 2a(v − b)

∂t = v − b , ∂p ∂v ∂t
= −1
∂p c ∂v ∂t ∂p

3 0 . F ill in the following sk etch of a p roof of Theorem 48.1. A ssume that fxy and fyx ex ist and are continuous in an op en
disk . W e must p rov e that fxy(a, b) = fyx(a, b) at ev ery p oint (a, b) of the disk . L et ∆h = ( f (a + h, b + h) − f (a + h, b)) −
( f (a, b + h) − f (a, b)) for h sufficiently small and ≠ 0. L et F(x) = f (x, b + h) − f (x, b). Then ∆ h = F (a + h) − F (a).
A p p ly the M ean-V alue Theorem to get a* between a and a + h so that F(a + h) − F(a) = F′(a*)h = [fx(a*, b + h) −
fx(a*, b)]h, and ap p ly the M ean-V alue Theorem to get b* between b and b + h so that fx(a*, b + h) − fx(a*, b) =
fxy(a*, b*)h. Then,

∆h
∆ h = h 2 fxy (a* , b* ) and lim = lim f (a*, b* ) = fxy (a, b)
h→ 0 h 2 ( a* , b* )→( a , b ) xy
CHAPTER 48 Partial D erivatives 413

By a similar argument using ∆h = ( f (a + h, b + h) − f (a, b + h)) − ( f (a + h, b) − f (a, b)) and the M ean-V alue
Theorem, we get
∆h
lim = fyx (a, b)
h→ 0 h2

3 1 . S how that Theorem 48.1 no longer holds if the continuity assump tion for fxy and fyx is drop p ed. U se the following
function:
 xy( x 2 − y 2 )
 if (x, y) ≠ (0, 0)
f ( x , y) =  x 2 + y 2
 0 if ( x, y) = (0, 0)

[F ind formulas for. fx(x, y) and fy(x, y) for (x, y) ≠ (0, 0); ev aluate fx(0, 0) and fy(0, 0), and then fxy(0, 0) and fyx(0, 0).]