AppucATtoNS oF ThtE

t}ERIVATIVE

Applications of the derivative abound. ln this chapter we first

develop the theoretical underpinnings for applications o{ the
derivative. Then we show how the derivative is used in
solving a variety of numerical, geometrical, and physical
problems.

4
.:l
- -
TOOLS FOR APPLTCATIONS OF THE DERIVATIVE

As we learned earlier, a closed interval along the x axis is an interval which

includes its endpoints, an open inteiva! is one which exciudes the endpoints,
and a half-open intervai contains one endpoint but not the other'
The basL theorern of this section is the Extreme Value Theorem, a result
which is usually proved in a more advanced mathematics coursc. We shall
show how thiq theorern can be used to obtain important applications for the
differentiation process we studied in Chapter 3.

THEOREM 1 (Extreme Value Theorem) If -f is a continuous Junction defined on the closed

interual fa,bl, there is ( at least) one point in la, b) ( call it xy ) where f has a
laygest ualue, and there is (at least) one point (call it x2) where f has a
smallest ualue.

This theorem is fairly clear intuitively if we think of a continuous

function as one with no breaks or gaps. As we move along the curve from the
: :
point corresponding to x a to the point corresponding to x b' there must
be a place where the curve has a high point (called the maximum value) and
there must also be a place where it has a low point (called the minimum
value). In spite of the simplicity of the situation from the intuitive point of

125
126
CHAPTEN,4 APPTICATIONS OT THE DIRIVATIVE

a view, the proof of Theorem ! is hard. Consequently we shali restrict ourselves

{ to a discussion of the meaning of this theorem:
1 A good way to examine a theorem critically is to sce what happens if
: some of the hypotheses are altered, In Theorem I there are two principal
I

J(r.\ t /r.
I hypotheses: (l) the interval [a,b] is closed,'and (2) the lunction -f is
continuous. We shall show by example that if we tamper rvith eithcr
1 hypothesis, the conclusion of the theorem may be false.
Suppose thc assumption that the interval is closed is replaced by the
I

J
assumption that the interval is open. The function /(x): l/x is continuous
1
I

1 '-..+'--+--,+, on the open interval 0<,r < 1 (Fig l), and it has no maximum value in this
-t--' ; I I open interval. A morc subtle example is given by the function I\x\: x2
TIGURE 1 defined in the open interval0 < -x < 2 (Fig. 2). This function has no maximum
or minimum value on the open interval but does irave a maximum of 4 an<i a
mininrum of 0 on the closed interval. This situation comes about because
/'(x) : x2 is continuous on thdclosed interval 0 < x < 2. In the first examplc
the function l/x is continuous on 0 < x < I but not on the closed interval
0(x< l. Even making the interval half-open is not good enough, sincc the
function l/x is continuous on the half-open interval 0 < x < I and still has no
maximum value there. The second example, f {x): x2, is continuous on the
half-open interval 0 < .x < 2 and does not have a minimum in this half-open
interval.
The second hypothesis, that of continuity, is also essential. The function

,f(..):J,, o<xS 2, x*l

FIGURI 2

l0\:z
: I (Fig. 3). As the graph clearly shows, lhe function
is continuous except at -x
has no maximum and no minimum on the closed interval [0,2]. A second,
more sophisticated, example is the function
[x2+1. 0<x<l
,/(.x):1r_r. t<x<2
whose graph is shown in Fig. 4. This lunction is continuous except at x l, :
where its minimum occurs with the value zero. There is no point on the closed

-f* r
,

/(r:[];:0,='=2'x*l -
J(r):
fx'+ l' 0<-t<l
tx _ I. l<x<2
TIGURE 3 TIGUTE 4
 127

3€CTloN 4.1 Toots FoR APPLlcATloNs or rHE DtRlvATlvt

the function gets

nb point where it

M at xl and its

instead of M and f (x2)

(or
THEOREM 2 SuP:Pose thut J is continuous on an interual and takes on its maximum
minimum) at some point xs which is in the interior of the interual. If f'(xo)
exists, then
/'(ro) :0.

The proolfor a
proof we prove the theorem for the case where jf (xo) is a maximum.
minimum is similar. If /(xo) is the maximum value' then

f (xo + h) <,f(xo)
is that
lor every possible h, both positive and negative. The only restriction
x6*hmustbeintheintervalinorderfor/(x6+lr)tohaveameaning.We
can also write (see Fig. 5)
(1)
f (x, + h) -/(xo) < 0.
O I ts c6*ft tf h is positive we may divide by h to get
h>0 {-9s7h)-:!.9J- 3s, h > o.
(a)

Taking the one-sided limit as h --+ 0*, we conclude* that

,,- .f(xo + h) -/(xo)

,iT. --- I . o. (2)

If h is negat ive, the inequaility (l) reuerses when we divide by h' so that

/(xo+Ir_)-/(xo) rO, h<0.

L)l c6*h. rs
h-
I

[<0 Taking the one-sided limit as h -'0-, we conclude* that

(b)
mcuRt 5
,,*
itftl
,f(xo t1)_:{Io)- >
-----.--:- o. (3)
i-o- h

-w. the following (intuitively clear) theorem on limits: if F{x'l <0 for
".".rploying
all x anil rf F(x) - L as x - a, tien L<a.The same result holds for one-sided limits;
case L 2 0'
,t. analogous resrrlt holds for functions l.(x) > 0' in which
"iro
CHAPTER 4 APPTTCATTONS OT THE DCIIVATIVE

Since the.ordinary limit (two-sided limit) exists as h tends to zero, the one-
sided limits exist and are equal (see Theorem 1l of chapter 2). Examining (2)
and (3), we see that they can only be equal if /'(xe) :0, which is what we
wished to prove. tr
Discussion. The important hypotheses in Theorem 2 are (l) that xo is an
interior point, and (2) that / has a derivative at xo. If the first hypothesis is
neglected the theorem is false, as is shown by the example f (x'.: x2 on the
intirval I < x < 2 (Fig. 6). The maximum occurs at x:2 (not an interior
point) and the minimum occurs at x: 1 (not an interior point). The
ierivative of /(.r) = x2 is f'(x\ :2x, and this is different from zero through'
out the interval [1, 2]. The fact that xo is interior to the interval is important;
rrcURE 6
whether the interval is open, closed, or half-open is irrelevant.
using the methods described in Chapter 2, Section 4, we now show that
the function
,f(x):x3-x, -l<xcl
has a maximum at the Point

(-+,;,fi)
and a minimum at

(-f , a":) (+,- i'fr)

(see Fig. 7). To find these points we first get the derivative of /:
|$)=3x2-l'
Then, setting this equal to zero and solving for x, we obtain 3x2 - I :0,
*:ifi, -1J, Substitution of the x values into the original expression
yieldsihe Aeslrla points. We shall prove in Section 3 that these points are the
GraPh of Y: x3 - x true maximum and minimum points of the function.
TIGURE 7 The second hypothesis in Theorem 2, namely, tbat f has a derivative at
x6, is also essential, as is shown by the following example. Let

r$):{_il.,il,l' i::11
This function is continuous on the interval I < x S 3, since both expressions
approach the same value at x: 2. As Fig. 8 shows, however, there is a
..corner" at x:2,and the derivative does not exist at this point. Nevertheless,
the point P(2,1) is exactly where the maximum occurs.
The derivative may not exist by virtue of being infinite at a maximum or
minimum point. The relation x2t3 +y2t3: I for -1<x51, has the graph
shown. in Fig. 9. The portion of the graph above the x axis represents a
function. The maximum occurs at x:0 and at x:0 we have y = 1. We can
compute the derivative implicitly and obtain
,1rr,, dy yrts
*1r-,,'0-:o or
dx xlll '
 129

sEcrloN r.1 Toots toR AmtcATloNs or rHE DEllvATlv[

A maximum occurs at a point where Derivative is infinite at maximum point

the function is not differentiable and at minimum point
FIGURE I FIGUf,E 9

and we see that this tends to infinity as x tends to zero. Therefore there is no
derivative at the maximum Point.

1 PROBTEMS

I Show that the continuous function /

the graph. Show that has the range [-
l, 1), and that the
function never assumes its maximum value. Why does
( x, 0<xcl Theorem 1 not aPPIY?
I
f(x):\2- x, I <x<2 7 After studying the function in Problem 6, construct a
continuous function defined for - oo < x < 6 with range
I
Lx-2,2<x<3
( - 2, 3] and which never assumes its minimum value.
has its maximum value at two different points. Also show
8 Given the continuous function
that the minimum value is taken on at two points. Sketch'
2 Construct by formula (as in Problem 1) a continuous : l(x2+ll.
(x2 0<x< +@
/(r) _' xz +
function on the interval [ - 2, 5] which assumes its max- t 11x2 ry. -m<x<0.
imum value at two different points and its minimum value Show that assumes its maximum or minimum
./ never
at threedifferent points. Sketch the graph. value. Sketch.
3 Construct a continuous function (by formula) defined on
the interval [1,6] which assumes its maximum value at In Problems 9 through 14, use the method described in
Chapter 2, Section 4. These methods will be justified in Section 3
every point of the subinterval [2,4.5]. Sketch the graph.
of this chapter.
4 Construct a continuous function (by formula) defined on
the interval [0, 5] which assumes its minimum at x I and : 9 Given the function f(xl=3x2 *6x+ 1 on the interval
at every point ofthe interval [2, 3] but at no other points. -2<x <0, find the minimum Point.
Sketch.the graph. l0 Given the function f(x):3x2 *6x* I on the interval
0 < x < 3, find the minimum Point.
5 Construct a continuous function (by formula) which is
defined on [-1,4], which has its maximum at x:0, its 1l Find the maximum value of
minimum at x:2 and is such th.t f' does not exist at (x+1. 0<x<l
x : 0 and at x :2. Sketch the graph.' f(x): {
l--x-+J, 1<-x<3
6 The function
on the interval 0 < x < 3. Where is the minimum value?
-l
x2
12 Find the maximum point of /(x) : $x3 - x + 2 on the
"f(x):V+r What is the maximum point on the
interval l-2,2).
is continuous for all values ofx from - @ to + co. Sketch interval t-3,3l?
130
CHAPTER { APPTICATIONS OT THE DERIVATIVE

13 Given the function

*18 Prove the following result: If F(x) > 0 for all x on an
(-x2+4x-2. interval [c, b] containing the point c and if F(x) - L as
l<x<2
/(x):t-1.x2+ 2r. 2<x<3, x - c, then L> O. (Hint: Assume that L is negative and use
the <lefinition of limit to show there must be a value of x
near c where F(.x) < 0, thereby reaching a contradiction'
find the maximum point. Does the function have a dert- (See the footnote on Page 127.))
vative at this point? If so, find the value of the derivative'
14 Examine the relation 19 Write out a proof of Theorem 2 (page 127) for the case
where /'takes on its minimum value.
xrts+y'i3:2, -8<x-<8
with regard to maximum and minimum points' 20 Suppose that a function / defined on the closed interval
[a, b] assumes its maximum and minimum values on this
l5 Given the function
interval. Is / continuous on [4, b]? Justify your answer'
I
/(x):;r*1' -)o<'x<oo' *21 A function / defined on [a, b] has a left-hand derivative at
l-l if
show that / has a maximum value, but no minimum value'
+ -/(b)
l.m /(b
Sketch the graPh. h_)

16 Draw the graph of y = 5in ,, * co < 'x < 'i;, and iden- h-o - h

tify the inlinite collection of maximum and minrmum

points. exists. The right-hand derivative at a is defined similarly'
*17 The function -y: arctan ,r has domain - rD' * oo) and
( Show that the Extreme Value Theorem holds for functions
(
range - nll,tl2). Sketch the graph. Does the function which are differentiable on (a, b) and have a right-hand
have maximum and minimum values? derivative at c and a left-hand derivative at b'

",

FURTHER TOOLS: ROILE',S THEOREM; MrnN VATUE THEOREM

Theorems and 2 of Section I are the basis for the next result, known as
I
Rolle's Theorem. While Rolle's Theorem is quite special, it has the advantage
of being intuitively clear, especially when viewed geometrically. Also, the very
next theorem, known as the Mean Value Theorem, is then an easy
consequence even though it does not appear to be as elementary as Rolle's
Theorem.

THEORTM 3 (Rolle's Theorem) Suppose that f is continuous.for a<x<b and that -f'(x)
exists for each x between a and b. lf
f(a)=/(b): Q'

then there must be (at least) one point, call it xo, between a and b such that

/'(ro) :0.

Proof There are three possibilities:

Case 1.(The trivial case.) /(x) :0 for all x between a and b; then /'(x) : (
for all x, and xo can be chosen to be any value between a and b.
 131
stcTtoN 4.2 Funrxrr roots: RouEs THECIRrM; Mr,lx vltut rHEoRrM

case 2. /(x) is positive somewhere between c and 6. Then the maximum of

/ is positive, and we choose xo (Theorem l) to be a place *,here this
maximum occurs. (see Fig. t0a.) According to Theorem 2 of the previous
section,
"/'(-ro)
:0, since xo must be interior to the interval.

case 3. /(x) is negative somewhere between a and h. Then the minimum of

/ is negative, and we choose xo to be a place where this minimum occurs. (See
.l-,(xo) :0.
Fig. l0b.) According to Theorem 2 of the previous section.
Since everjr function which is zero at a and b must fall into one of the
three cases, the theorem is provcd.

Figure l0(c) shows that a lunction may fall into both case 2 and Case 3,
and Fig. l0(d) illustrates the possibility of several choices for._xo even though
it satisfies only Case 2.

(c) (d)
FIGURE 10

we can state Rolle's Theorem in a simple way: I! a differentiable curue crosses

the xaxis twice there must be a point between successiue crossings at which the
line tangent to the curue is parallel to the x axis.

we have been building up to the proof of the theorem known as the

Mean value Theorem, one which is used over and over again throughout the
branch of mathematics known as analysis. Every mathematician wio works
in analysis encounters it often, uses it, and feels at home with it. we shall see
that, important though it is, the Mean value Theorem is merely a variation
of
Rolle's Theorem.

THEOREM 4 (Mean Yalue Theorem) Suppose that is continuousfor

f aSx<O oni tnot
f'(x) existsfoi each x between a and b. Then there isLn xo i"ri"n a and b
(that is, a1 xo<b) such that

f'(xd:ry+
CHAPTER { APPtIcATtoNs or THE DEnIvAnvT

a Before proving this theorem we shall discuss its meanrng lrom a

*-..., g"o^etric standpoint. Figure 11 shows a typical function / between the
(ro,.f(rd) (o'J(o)) points a and b. The point P has coordinates (a, f (a)) and e has coordinates
(b, f {b)).We construct the straight line through pe and calculate its slope.
we know that the slope n is the difference of the y values over the difference
l/../
of the x values, that is,
',

P(a,l@)

m- f @ -r*,
h-a
'rhis is cxactly the sanre expression that occurs in the statement of the Mean
Value Theorem. The theorem says there is a point (xo,/(xq))on the curve
where the slope has the value m; that is, the tangent line at (.xo,/(xo)) is
parallel to the line through PQ. Glancing at the figure, we see that there must
be such a point. In fact, by meaus of the following device, we can see that there
must be one: We look at the figure and tilt it so that the lini through pp
appears horizontally as the .x axis. Then the Mean value Theorem resembles
Rollc's Theorem.

Proof of Theorem + The equation ol the line thrt>ugh PQ (according tcr the two-point formula for
the equation of a straight iine) is

y f (ar _ "f tb,t_J(r)

_
1 *_ o).
b-tt
We construct the function

F(r):/(x) -f(ht-J@t(x -c) _ ftat.

h-a
By straight substitution with r : a and then r : b, we find

F(at :f(a\ - I lb)-1tal (a - at - f(a) : o.

D-d

F(b):f (b) *ru)-* tb - at *.f tat:0.

b-a
Therefore F(x) satisfies all the hypotheses of Rolle's Theorem. There must be
a value ,xo such that F'(.xn) :0. But (by differentiation) we see that

This implies that

which is what we wished to prove.

EX,AMPLE 1 Given that

/r11 :'I1]_ and a: t.b:2.

rfl
find aii values xo in the interval I < x < 2 such that
 133
sEcTtoN 4.2 Funrxm Toots: Roru's THEoTEA4 Max velur THEoREM

I'(xo): lL:J!!
D-A

Solution We differentiate .f , getting

(r+ l)'l -1-x-+ 2l'l:
'f'('x): (.r + t)2 (.r '
+ 1)2'
and
-l

f ot : ). r(2t -
4
. t:llr
-lV\
: ;
We solve the equation l''(.to) : - |, which yields

=-1,,.:-1
+ l)2
(_xo 6
or xj+2xo-5:0,
and

'- -z+-:"^,Eq :-l+-/6.

The value Xo : - t + yG is in the interval (1, 2), ,"nlt" the value

ro - --,rE- V'6 rs relected since it is outside this interval. tr
EXAMPI-E 2 Given that /(x) : x3 * 2x2 + 3x -- 2 and u:0, b: 2, find all possible values
for xn in the interval 0 < x < 2 such that

I'{xo: : f t2:[kt
Solution We differentiate f,getting
f'(x) :3x2 - 4x + 3. Also, /(0) - -2, f (2) : 4,
so that

f(b)_l@):-r--:'.,'
4+2
b-,
We solve the equation

3xf,-4xo-r3:3,
which has two solutionSi xo : 0 and xo : $. But xo : 0 is not between 0 and 2,
and therefore the only answer is xo - f . tr
DGMPTE 3 Given the function

,f{x;:x2-4_x+3
x-l
and

a:l,b:3
discuss the validity of the Mean Value Theorem.
7y
CHAPTER 4 AmtcATtoNs or rHE Dt$vATIvt

Solution If we proceed formally, we see that /(l) 0 and[(3):0, so we must find an

=
.re in the interval (1, 3) such that /'(xo) :0. Compuiing the derivative by the
quotient rule, we write

+ 3) _ x2 - 4x *
f'(x): {x - - -!x
2)(2x 4) - (x2 5
f:1r-21r-'
Setting this equal to zero, we obtain
xf;-4xot5:0
and

.. 4 t,fr610:?4 L,/ 4
,o:-2
which is impossible! we look once again at the function / and see that it
becomes infinite at x : 2. Thereflore the hypotheses of the
Mean value Theorem
are not satisfied, and the theorem is not applicable. There is no value xo. I

2 Pnonums

In each of Problems I through 12, find all numbers xo between 14 Same as Problem 13 for /(x) = x3 + 2x2 - l lx - 12.
a and b which satisfy the equativrr
15 Given that /(x) = (2x * 3)lQx + 2), a: - l, b : 0, show
-f(a) that there is no number xo between a and b which satisfies
['(x,):f(b)D-a ' the Mean Yalue Theorem. Sketch the graph.
I /(.x) : - 3x - 2, a:0, b:2
x2 16 Given that /(x)'= 2xzt3, a: - l, b: 1, show that there is
no number xo between a and, b which satisfies the Mean
2 f(x):x3-5x: +4x-2, a:l,b:3
Value Theorem. Sketch the graph.
3 /(.t):.x3-7.r2 + 5x, a:l,b:5
17 Given that
J'G\:xr+2x2-x, a: -3,b:2
v
s /(x) : --j.
f(x):UQ-\', a: -t,b=2.
-)
a:0, b: i
x+2 Is there a number xo between a and b which satisfies the
: 2x+3 equation
6
"f(x) 3x-2'
a:l,b:5
(b - a) f '(xo):f (b) -t@)1
7 f (xl- !6s-7, a: -3,b:4 Can the Mean Value Theorem be used?
8 Itx) :v4'+ 81, a: 12,b :40
9 f{x): xa -2x3 + x2 -2x, a:
18 Given that / is a quadratic function of x, and a and b are
-l,b:2 any numbers, shorr that xo=+(a+b) is the value that
10,f(x):xo+x3 -3x2 +2x, a: -2,b: -l satisfies the Mean'1fu[u9 Theorem. (.[Iint: Assume that
x2-3,x-4 f (x\ : cx2 + ilx* e, where_c, d, and e are any constants.)
l1 /(.x) = a: -l,b:4
x*5 *19 Given that
/ is a general equation ofthe third degree. That
2+(tx is, suppose f(x) : axl + px2 + ?x + 6, with a * 0. Show
12f(xl-xt+6x-*5 a:l,b:5
x-6 that there are at most two values of xo which satisfy the
Mean Value Theorem for any given numbers a and b,
13 Given ,f(x) : x3 -3x2 - 6x -r 8, find all values of x such
Prove that these are given by the formula
that /(-r) :0. Then find the values of x between each two
zeros of / such that /'(x) :0, thus verifying Rolle's 8l
Theorem.
Xo: - *.x-Jp,+3ar(p@.
3d )d,'
 135
sEcTloN {.2 FURTHET roors: Roru's rHEor,E t Mr,lx vnrur rHEotE/vr

20 Given the function / defined for 0 < x <2 by the formula +23 Use the Mean Value Theorem to shorv that for any two
real numbers x and -y,
(xfor 0<x<l
/(*):{g,,+tr ror l<x<2. lsinx-sinyl <lx-yl.
+24 Use the Mean Value Theorem to show that if /(x) :
Does the Mean Yalue Thaorem hold for 0 < x < 2? Justify x2 +2x+ 3, then for any values a and b in the interval
your answer. (.1, 3),

2l Same as Problern l8 for the function lf (b) -f (a)l < 8lb - ai

(2x3-t for 0<x<l *25 Use the Mean Value Theorem to show tnat if / is
Jlx): \. for a1x<b and /'(x) exists for each x
[6x-5- for l<x<2. continuous
between a and b, and if l/'(-r)l < K for all x between a and
22 Same as Problem 18 for the function b, where K is a positive constant, then

for 0<x<l ll$\ -f 0)l < Klx-yl

t@:{';::'-' for l<x<2. for all x, y between a and b.

3
AppuclnoNs ro cRApHs oF ruNcnoNs

Knowing how the derivative of a lunction behaves helps us in obtaining an

accurate idea of the graph of a function. We shall see that the derivative tells
us when a graph is rising and falling, as well as when it has peaks and troughs.
In this and the following section we establish some ntles (in the form of
theorems) which are useful for graphing functions; we also give examples to
illustrate the technique.

DEFINITION Afunctionf issaidtobeincretsingontheinterval Irf f(*r) >"f(r,) wheneaer

x2) X1, so long as both x, and xz ore in I. It is decreasing if f (xr) <"f(r,)
wheneuer xz)Xt (Fig. l2). The interual I may contain one endpoint, both
endpoints, or neither endpoint.

Increasing function 'I)ecreasing function

HGt lt 12
t36
CHAPTTR 4 AP?trcAnONs oF THT DERTAnVE

DEFINITION A function f is said to haoe a relative maximum at x, if there is some interoal

with xo as an interior point, such that /(xo) is the true maximum of f in this
interual. Similarly, it has a relative minimum at x, if there is some interual with
x I as an interior point, such that f (x ,) is the true minimum of f in this interual.

Maximum
The latter definition needs some explanation, as provided in Fig. 13,
Relative Relative which shows a typical situation. The function / has its maximum on fa, b) at
xo, a point which is a relative maximum. However, / has a relative maximum
at -rr, although f (xz) is not the maximum of the function on [a, b]. The
Minimum function
/ has a relative nlinimum at xr. The true minimum on [c, b] occurs
L at b. However, / does not have a relative minimum at b, since the definition
for relative minimum is not satisfied there. In Section 5 we investigate further
FIGI}RE 13 the distinction between true maxima and minima and relative maxima and
minima.

THEORE/Vi 5 If f iscontinuousonaninterusl I andif f'(x)>0 for eachxintheinteriorof I,

then f is increasing on I.

Proof We apply the Mean Value Theorem to two points xr, x, in I. We find that
f (xr) -f (xr):f'(xr)(x2- xr),
where xo is between x, and x, and hence interior to 1. For Xz) Xt and f'
positive, we obtain l''$o\$r-xr)>0, and so /(xr)
-f(xr) >0. This
means, by definition, that f is increasing. tl

coRotl-ARY If f is continuous on I and f'(x)<O for each x interiar to I, then f is

decreasing on I.

Theorem 5 is useful for graphing; we can find out whether the graph of a
function is rising or falling by observing whether the derivative is positive or
negative. This theorem (together with Theorem 2 of Section 1, which says
that at a relative maximum or minimum the derivative of a function is zero)
helps us find the peaks and troughs.
There is a complication in using Theorem 2 because, although the
derivative of a function is zero at a. maximum or minimum point (assuming
the derivative exists there), the conuerse moy not hold true. For example, the
function f (x): x3 1Fig. 14) has derivative f'(x):3x2, which is zero at x:0.
Example showing that -f'(0) :0, yet / But (0, 0) is neither a maximum nor a minimum point of the function. Thus
does not have a relative minimum or the knowledge that the derivative of a function vanishes at a certain point is
relative maximum at x:0 not enough to guarantee that the function has a maximum or minimum there.
FTCURI 14
.')
DEFINITIONS ,4 critical value of afunction f is a ualue of x where f (x) : 0. ,4 critical point o/
,

a function f is the point (x, f(x)) on the graph corresponding to thi qitical
ualue x.
 137
sEcTroN {.3 AmrcAfloNs To GRA?}IS of FUNcTtoNs

A critical point occurs at any relative maximurn or minimum point

of a function which has a derivative at that point. However, the function
I$):.x3 has a critical point at (0,0), but this point is neither a maximum
nor a minimum of the function.
Suppose a function / and its derivative /' are continuous. If the equation
f '(x) :0 has only a finite number of solutions, we may use Theorem 5 and its
corollary as an aid in graphing. To do so, we let .x,, X2, ... t x,be all the values
of x for which /'(x) : 0, arranged in order of increasing size. In each of the
intervals (-:co,xr), (xr,xz), (xr-,xr), ..., (x,-r,xo), (x,, *oo), the quantity
/'(x) must remain either posifiue throughout or negatioe throughout. This fact
is a result of Thcorem 6, stated below without proof. As a result, to determine
the sign of /'(x) in any one of the intervals, we need only find its sign at a
single interior point in that interval.

THEOREM t Suppose f is continuous on anintert)al I anil f isnbt zero at any point of l. Then
f is either positive on all of I or negatiue on all of l.

This theorem is intuitively clear when we graph a continuous function. If

/ starts out positive, (i.e., above the x axis) and ttren becomes negative, (i.e.,
goes below the x axis) it can do so only by crossing the axis. The place where
it crosses the x axis is a point where / is zero.
Theorem 6 is usually proved in a course in advanced calculus. It is a
consequence of the Intermediate Value Theorem, stated in Chapter 5
(Theorem ll,
page 206).
We give an example showing how the determination of the zeros of /'can
be used as an aid in graphing.

EXAMPLE 1 Study the derivative of the function /(x) : *x, - x2 - 3x* 3 and use the
resulting knowledge to sketch the graph of the function.

Solution The derivative is /'(x) = x2 - 2x - 3: (x * 1)(, -

for x, - -l and xz:3. We consider the intervals (-o,
3). The derivative is zero
-1), (-1,3), and
(3, +oo). We pick a single point in each interval, selecting the point
-3 in
(-o, - l), the point 0 in (- 1,3), and the point 5 in (3, +co). We determine
the sign of f'(x) at each ofthese points and the value of/ at each ofthem, as
shown in the table below. With these aids we are ready to graph the function
(Fig. ls).

x -3 -l 0 3 5

-3 i-i 0 f'(x) + 0 0 +
-1
f(x\ -6 4 3
-6
.i
-3
-4 The critical value x :
* I corresponds to a relative maximum, and the critical
-5 value x: 3 corresponds to a relative minimum. D
-6 The theorems and definitions presented in this section give us a method
The graph of y : jx3
- xz - 3x + 3 for finding relative maxima and minima. we state the method in the form of a
EGTJRE 15 test.
 CHAPTET A,?uc^noNs or THE D,BIVAIIVE

THEoRtlt z (First Derivative Test) t If is increasing t > 01 on

f 7 some intensal to tlw
left of xq with xo as endpoint ofthis interuot, oia if is decreasing (f,
f
some interoal to the right of xo fuith xo as endpoint), then
< 0) on
maximum at xo if it is continuous there.
iai"a reratiue f
it f
If is deueasing (/, < 0) in some interval to the left of xo with xo as
endpoint of this interual , and if is increasing ( ' > 0) in
f ,i*i
f inirruol to the
right of xo (with xo as endpoint), then hai a reratioe minimum at
f xo if it is
continuous there.

Remarks. If / has a derivative at xo, then the derivative will be zero.

However, we include the possibility that there may be a cusp or
a corner.
Figure 16 shows various possibilities.

Relative maximum Relative minimum

l,@d:o l'@o):o
(a) (b)

Relative maximum Belative minimum

//(e6) does not eist; corner
/(es) infinite; cusp
ncum rG (c)
(d)
Proof of Theorem 7 To prove part (i), we denote by I'the intervar to the left (including
xo) and by
I I"-the interval to the right of x6; then we combine them into
oneintJrvar and
call.this interval r (Fig. l7). That is" I' v I": r. Then xo is
an interioi point ol
I. Since / is increasing on / and decreasing on r,, and Loth contain
the point
t---Y---l-----Y----J x6, w€ must have f (xd > f $) for all x in I.
I/"Iu The proof of (ii) is the same.
D
FIGURT 17
EXAtmt 2 Discuss the function / defined by

f (x): xst3 + 5x2tl

for relative maxima and minima, and determine the intervd
in which / is
increasing and those in which is decreasing. Sketch
/ tf,, gr"pt.
Solution The derivative is

f'(x) = tyztt + rfx- r/3

- *x- r/3( x + 2).
The critical varue is x:
-2. The derivative is not defined at
construct the following table (r means undefined):
x:0. w€
 139
sEcTtoN 4.3 ApH.rc^noNs ro cr nfi or rt NcnoNs

x -5 1
-l 0 I

f'(x) + 0 t +
,f (-x) 0 3(27zrt 4 0 6

We conclude that

f incre_ases for x <

-2;
/ decreases for -2< x < 0;
/ increases for x > 0.
rile now apply the First Derivative
Test and conclude that [here is a relative
maximum at x: -2 and a relative minimum at x : 0. The graph is
sketched
The graPh of y: as/r * 5x2tt in Fig. 18. tr
:-.
HG.'RE 18
ilAMpr-E I Discuss the function / defined by

.r(-x):tP-, *r-1,2'
J2x+t'
for relative maxima and minima, and determine where is increasing
/. and
where it is decreasing. Sketch the graph.
.
Sotution The derivative is

f'(x):
(2x + l)ttz(2x + l) - (r1 . + l)lex + t)- ttz(z)
+ t) (2x
We simplify by multiplying numerator and denominator through by
(2x + l\rrz and obtain

+ t)z - (x2 + x + 7) _3(x +2)(x _.1)

r'^t-@
r,( yt :(2x
1r+lpt-.
we set this equal to zero and find that x : l is a criticar value. The apparent
solution x: -2 is excluded, since the function is defined only for r,
We easily discover that -i.
/(x) hecreases for - 1< x < l; /(x) increases for x Z l.
The First Derivative Test now applies, and we find that the value
rcURE 19
x: I yields
a relative minimum. The graph is sketched in Fig. 19. tr

3 PROBTEMS

In Problems 1 through 36, discuss each function for relative 5 2xt + 3x2 - l2x
maxima and minima, and determine those intervals in 6 x3-6x2+9x+5
which
the function is increasing and those in which it is decreasing. 7 x1 +2x2 -3x-2 t x3-3x2+6x-3
Sketch the graphs. 9x3+6x2+l2x-5 l0 x3+3x2+6x-3
I x2 +3x+4 2 x2 _4x+5 ll -x3+2x2-x+l 12 -x3-2x2+3x-6
3 -Zx2*x-6 4 -3x2-3x*2 13 xa-fx3 -4x2 +l l4 xa + !x3 - llx2
1ll{t
CIIAPTEI 4 AmrcAfloNs oF THE DnryAnvt

15 xa + 2x3 16 x4+ 4x3 +6x2 45 I=3sinx+cosx 6 y=sin3x-3sinx

I 2r
17 x+lx 18 47 y=rfrsin2x-cos2x y=4sin3x:3sinx
4E
xt+l 49 The function
19 cos x 20 sin x
/(x)=(tanx),/, is defined for O<x<n12.
Show that f'{xl>0 for 0<x <a/2. ls it true that
2l sin x cos x 22 cos2 x f"(x)>0for0<x<n/2?
23rl
r+l
3x-2
2x*3
fl) Find all the relative maximum and minimum points of the
function
^
-2
25 -+,
.r/x + I
,x> -l 25 3xtt2 - x3l2
/(x) - sin 1
27 x2t3g+3)lt3 2E xlg + t) defined for 0 < x < l. (There are infinitely many.)
29 xJT4, x<3 fi x215-i, x(5 *51 Given

3l xJz-x,, lrl<,fz U x2 -2x+ 1

x*l : argo
r
'f(x)
33 f_-tl
x-2 34 lxz -Zxl+ I
x
for 0 < x < l. Show that if p is any positive number, there
are always infinitely many relative maxima and minima. If
35 x2 +l}x+21 K -:i:2-
lx'-2xl+ l
these points are labeled (xr,yr), (xz,yt),... in order of
decreasing x,, then it follows that x, + 0, .Ii *
0 as i + o.
In each of Problems 37 through 44, find the relative maxima *52 Given the function
and minima of the given function / for 0 < x 3 22. Determine I$): ax3 + bxz + cx + d which is the
most general'equation of the third degree. Figure 15
where the function is increasing where decreasing, and sketch the
exhibits the shape of this function when a:1,'b:
graph. -1,
c: *3, d:3. Figure 14 exhibits the shape when a: l,
37 f (x): cos.x * sin x 3S
- ,f(x) : cos x sin x b:0, c =Q d:0. Show that there are basically four
39 f(x):x-sinx afi I$)=2x+sos.x possible shapes for a third-degree equation and sketch the
al f $):2 cos x + sin 2x a2 f (.xl:2 cos x + cos 2x remaining two shapes. (Assume a # 0.)
*53 The general equation
a3 f$):3x+sin3x U f(x)- -4x-cos3x of the fourth degree has the form
f{x):axo +bx3 + cx2 +dx*e. Assuming that a#0, de-
In each of Problems 45 througlr48, find the relative maximum cide how many different general shapes the graph ofsuch
and minimum points of the graph. an equation can exhibit. Sketch one ofeach type.

L
APPI.ICATIONS USING THE SECOND DERIVATIVE

T'he more information we have about a function the more-accurately we can

construct its graph. In Section 3 we saw that a knJwledge of ihe first
derivative helps us decide when the graph is increasing and when it is
decreasing. Furthermore, we learned that the First Derivative Test enables us
to locate the relative maxima and minima. The second derivative yields
additional factswhich are helpful_in determining the nature of the graph. The
first of these facts is a test for relative maxima and minima which empl,oys the
second derivative-
 141
sEcTtoN 4.a Amtceroxs usrNc rHE sEcoND DERtvATtvE

DEFINITION If, at each point of an interual, the graph of a function f always remains aboue
the line tangent to the curue at this point, we say that the curoe is concave
. upward on the interttal (see Fig. 20\. If the curue always remains below its
tangent line, we say it is concaye downward (see Fig.2l).

Coneave uprvard on [a, b] Concave dorvnward on [a, bj

FIGURE 20 FICURE 21

Before stating and proving the basic theorem of this section we recall the
equation of the line tangent to a curve. At any value xo the function / has
value /(xr) and the slope of the curve at this point is /,(xo). The equation of
a line through the point (xo,,f(xo)) with slope _f,Ud is, according to the
point-slope formula,

This is the tangent line.

THEOREM 8 (Concavity Tes0 Assume that f has a second d.erioatiue on an interyal l.

Proof To prove (a), we must show that the curve lies above the tangent line at any
point. Let xo be any (fixed) point in / (see Fig. 22).The tangent line at xo has
equation
y:f (xi */'(x6)(x - xo),
Case (1): cr > cs and we therefore have to establish (for all x in /) the inequality
(a)
,f(x) >"f(xo) */'(xs)(x - xr).
:
If x xo, this becomes f (xi
> f Go), which is true. If x is some other value,
say xr, then we can apply the Mean value Theorem to the function between
/
the two points x6 and xr. This theorem says that there is ydue such that i
f'(i):f(x')-f(xd,
xr-xo
Case (2): r1<rs where i is between xo and xr. We have two possibilities: (l) x, >,c., and (2)
(b)
x1(x6. In Case (1) we write the Mean Value Theirreni in the form
ECLnt 22
f (xr):f (xo) +/'(i)(x, - xo). ( t)
CHAPTER 4 ApHrcATroNs or Tffi tmtvATtvE

Since /" > 0 by hypothesis, we know from Theorem 5 that /' is increasing
on /. If .X1 ) x6: then x, > i > x6 and
f'(i) >{'(xo\.
We multiply this last inequality through by the positive quantity (x, - xo) to
get
f'(i)(x, - xo) >,f'(xo)(x, - xo),

and therefore, by substitution in Eq. (1), we obtain

f (xt) >f (xo) +/'(xo)(x, - xo),

which is what we wished to show.
In Case (2), since x6)i, we must have /'(x6) >f'(r).But now we
multiply through by the negative number (x, - xo), which reverses the
inequality, giving
f '(*)(x, - xo) >,f'(xo)(x, - ,o)
as before, and therefore, using (l)
IG)>f (xr)+/'(xo)(x,
- xo).
The proof of part (b) of Theorem 8 is similar. iI
Theorem 8 shows that knowledge of the second derivative gives us an
even clearer picture ofthe appearance ofthe curve. In addition, it gives us the
following useful test, known as the Second Derivative Test, for relative
maxima and minima.

THEORTM 9 (Second Derivative Test) Assume that f has a second deriaatiue, that f" i5
continuous, and that x6 is c critical rtalue ( f '(xs) : 0\. Then

Proof To prove part (a), we see from Theorem 8 that the curve is concave upward
and must lie above the tangent line at xs. But this line is horizontal, since
f'{xi :0. Therefore /(xs) must be a minimum value. The proof of (b) is the
same. Part (c) is added for the sake of completeness. n

D(AI$PT"E 1 Discuss the function

,f(x) : *' * nL*'* 9x - 4

for relative mdxima and minima. Sketch the graph:

Solution The derivative is

f' (+\ = 3x2 - $x + 9 : 3{* * l1x + 31,

and the critical values are solutions of

xz -Zx* 3:0 or x=*,2.
sEcTtoN a.a AmJcAnoNs us,tNc rHt s[coND Dmry T]vE

We apply the Second Derivative Test.

f"(xl :6, -*; f"G) : -;<0, :*>0. I"(2)

Therefore I : ! corresponds to a relative maximum, and x : 2 corresponds to
a relative minimum. Also, the second derivative is positive if x > J and negative
if x < 1. That is, the curve is conchve upward for x > fr and concave downward
for x <J. We construct the following table.

x 0 i2 2
4 2 3

LL 1# I LL
T -4 16 4

f' + 0 0 +

f' 0 + +

This problem indicates the difficulty that may arise when a relative
maximurn and a relative minimum are close together. Without knowledge of
the derivatives given in the table, we would normally plot the points at x:0,
t l, 2,3, etc. This would miss the maximum at x : ! and give a misleading
picture of the curve. We know that the curve is concave downward to the left
of x:Iand concave upward to the right of J. These facts prevail indefinitely
far to the left and right. The graph is sketched in Fig. 23. In this example the
Graph of "f(x) : x3 - {x2 + 9x - 4 point x : J is important, as it separates a concave downward region from a
HGUTE 23 concave upward region. tr

DEflNtTtON A point on a curue is o point of inflection if f '(xd :0 at this point and d the
graph is concaoe upward on one side and concaue downward on the other.

We make two remarks which the reader should keep in mind: The first
derivative may or may not vanish at a point of inflection. tn Example I the
value x:] corresponds to a point of inflection,and f'(l): -t. On the
other hand, the function f (x):(r-2)' * I has a point of inflection (see
Fig.2q at x:2, since f"(x):6(x-2) and f"(2):0. However, we notice
that /'(x) : 3(x - 2)2 and f'(2):0, also.

a:@,-2\a+1

Graph of /(x) : (x - 2)3 +I

RGUNE 24

The second remark: It is not enough to know that /"(xo) :0 to

guarantce that xo corresponds to a point of inflection. We must also know
that /"(x) > 0 on one side and that f '(x) < 0 on the other. An example which
1U
cHAPrtt a APPUcATIoNS oF THE DERtvATtvE

shows this difficulty is the function /(x) : xn (see Fig. 25). In this case,
f (x) l2x2 and f (0) : 0. However, /"(x) is always positive for x * 0 and, in
" : "

fact, x : 0 is the value for a minimum. (Part (c) of Theorem 9 for the Second
Derivative Test is the rule which holds hefe.) The First Derivative Test,
however, works to give a minimum at x:0.

ExllMPtE 2 Discuss the function

lxa _ )xz f(+):

for relative maxima, relativd minima, and points of inflection. Sketch the
graPh.

Solution The first two derivatives are

FIGURE 25
f'(x) :
- 3x and f"(x):3xz - 3'
x3

The critical values are solutions of x3 - 3x :0, and so we get x:0,

,fr, -f. fne Second Derivative Test tells us that
x:0 is at a relative maximum;
*:.,[3, - rf; arc at relative minima.
The possible points of inflection are obtained from solutions of 3x2 - 3 :0;
thatis,x: *1, -1. Since f"(x)is negativefor -1<x< l andpositivefor
lxl> 1, both x: I and x: - 1 yield points of inflection. We construct the
table:

x a
=fi -l 0 { 2

I -2
9
4
5
4 0 _54 9
4 -2
f' 0 + 0 0 +

f' + + 0 0 + +
Graph of /(x1:txa -lx2
FIGURE 25 The graph is shown in Fig.26. n
E(AMPTE 3 Discuss ihe function

f (x): x4tt + Ax\tt

for relative maxima, relative minima, and points of inflection. Sketch the
graph.

Solution We have
' f'(x):$1r/r *tx-ztt:fy-2r3(x* 1);
f ' (x\ : $x- - 8x- : $x- (x - 2\.
2/3 s/3 5t3

The first and second derivatives do not exist at x:0. The value x: - I lsa
critical value and x:2 corresponds to a possible point of inflection. We
construct the table:
 14s

SECTlON.r.4 APPUCATTONS USING THE SECOND DIMVANVE

x -4 *1 -1
I 0 I 2 3

f 0 -.3 -2fi 0 5 6{. 7{

f' 0 + I + r* t-

+ + + * 0 +

From the table we see that x: - 1 yields a relative rlrinimurn and that x : 2
yields a point of inflection. Furthermore, x : 0 yields a point whcre the curve
ihanges from concave upward io concave downward, The graph is shown in
Fig.27.

GraPh of f (i: xttt * 4xrtx

f (x) : sin 2x, -n < x < z, for relative maxima, rela-

FTGUNE 27
EXAMPTE 4 Discuss the function
tive minima, and points of inflection. Sketch the graph.

We have

f'(x) :2cos2x, f'(x): -l.sin2x.

The critical points are solutions o[ 2 cos 2x:0. These occur when

3'.
2x: +!.
tn-.1, +3n
- 2 or ---4,1-'!- +-.4.
x:
Using the Second Derivative Test, we find

t'(X) <a, r"(?),0, /'(-i),0, r'(-?).n

Thus relative maxima occur at

n- 4'
n3n
4

and relatiye minima at

*: 3nn
+' -i'
To obtain the points of inflection" we set f"(x) :0 and find that these occur
when

2x:0, t n or x:0, lE
t ;.
zl
I

\
The points of inflection are

(o, o), (;,0), (-;,0

CHAPTTN AppucenoNs oF THE DERTvATtvE
'
The graph is shown* in Fig.28. I

4 PROBI-EMS

In Problems I through 35, discuss each of the functions for 2

relative maxima and minima, concavity, and points olinflec- 20 f(x):x-3* r*,
tion. Sketch the graphs.
4x
I f(x):x2-3x+4 2l [(x): --;--
x'+4.
2f(x)=-2x2+6x+5 22 f(x) - 5x2i3 - xsis
3 f(x):xr -27x+4 -/
23 f (x):x',E:/, lrl<.,4
af$):x3+3x2+6x-4
5 /(x) : xa + 4xl
2a [$):*zt3(x*2)-1
6 IG) - x4 - 4x3 - 2x2 + 12x - 8
25 f(xl=x2tEJ,', x> -5
26 f(x): x216;', lrl<..,5
1

7IG):x2+1
x- 27 f (x\ - x1l3(x *2)-zrz
2S f(x) = lx2 - 6xl+ 2
t
8 /(x) = 2x2 - -- 29 f (x):2 sin *x, -tt < x <tt
x'
-2x 30 /(x) : 3r. -! . r.\
'22
cos
gJ(x):xr+l
ft ,f(x) :4 sin 2xcos2x, -t < x <rI
l0 /(x) : xJx + 3
ll fG):x3-ix2 -6x*2
32 f(x):2coq3*x, -i.r.i
12f{x):x3+x2-x-l .7818
33/(x):tanx, -'i.*<,
tr"f(x):x3-4x2*4x-l
l4 f{x): x3 + 3x2 - 3x - 5 ll f(x) : sin x + cos x, -2n < x <2n.
15l'(x)-x3-x2*x-l lS,f(x): .,6 sin x + cos x' -tc<x <tt.
16 f lx): xn + 3x3 -4*'-t In Problems 36 and 37, find thc relative maxima and minima.
17 f 0) -- (x + 2)(x -2)3 N {(x): *
x3 l2x2 -zxl+ I
It /(x) : xa - 3x3 + 3x2 v- I

19 f U): xa + 5x3 + 6x2

37 f(x): i;-Eh,
.We provide a review on graphing of trigonometric functions in Scction 2 of
Appendix L
 147
sE€Tl.oN a.a Apruc^noNs ustNc Tltr stcoND DEflvAnvE

'3E Given the functioo.f(x) = {,

where n is a positive integer. not exist. Construct an example of a function which has
/
Show that has a minimum at x = 0 if n is an eyen integer such a point of inflection.
/
and that has a poiot of inflection at r:0
if z is an odd Show that if the function /(x) =aox3 + a1x2 + axx* a3
integer. has exactly one critical point, then it is a point of inflection.
*39 Usinii thc results of Problcm 3E as a gurdc, devise a
'Highcr Derivatirrc Test'for relative maxima and minima Show that the function
when both /'(xo) :0 and /"(x6) :0' Describe the situ-
ation if the first lc derivatiies of / vanish at x6 but the It')='{+1
' cx+d
derivative of order ,( + 1 is different from zero.
0 lt f is a given function, devisc a definition of a point of has no relative maxima or minima, where a, b, c, d are any
irtflection at a point at which the second derivative does numbers with ad bc *0.
-

5
THE MAXIMUM AND MINIMUM VATUES OF A FUNCTION
ON AN INTERVAT

In the preceding sections we studied ways of finding relative maxima and

minima of a [unction. One of these relative maxima may be the true
maximum o.[ the function, or it may not be. As Fig.29 shows, the function
exhibited has a relative maximum at P and a relative minimum at 0. As we go
offto the right, hoivever, the function gets larger than the value at P, and as
we go off to the left, the function gets smaller than the value at Q. Suppose
, that'we look at the above function only on some intervalla,b), as shown in
Fig.29. Then'the maximum occurs at b and is /(b), while the minimum
occurs at a andis /(a). lf fa, b']were the interval of interest, as shown, then
(o,lb)) the maximum of /on la,b'f is at P, while the minimum is at a. Further, on
FIGiJTE 29 the interval la',b'l the maximum is at P, the minimum at 0.

DEHNITTONS I*t f be a continuous fuwtion defineil on the closed intental a < x < b. The
riaximum oalue f takes on in this interual is called the absolute maximum o//
on fa, bf. The minimum ualue I takes on in the interual fa, b) is called the
absolute minimum of f on la,bf.

To obtain the maximum and minimum of a continuous function on an

interval fa,b), we employ the following procedure (which is based on
Theorem 5 and the First Derivative Test):

RULE a) Find the relatiue maximum and minimum ualues of f .

b) Find the oalue of the function at each of the endpoints.
The Lqrgest of the ualues of (a) and (b) is rhe absolute maximum. The
smallest oalue is the absolute minimum.
148
CHAPTTT { AmICATIONS OT THE DERIVATIVT

EXAMPLE Given the function

/(x):{x3 +lxz -2x,
find the absolute maximum and absolute minimum on the interval [ - 3, 4].

Solution We follow the Rule and find relative maxima and minima:
-f'(x): x2 + x=2,
and the critical values are x:
-2, l. Thc second Derivative Test gives us

.f "{x):2-x + 1; f'(-2) - -3'

making $ a relative maximum value, and

f"(l):3,
(1, :-?) is a
which makes -3 a relative minimum value. That is, the point
go on to part
relative minimum, and ( - 2, *) is a relative maximum' Now we

iii.r,tr" nurc: l(-3):|Lia t$l:$. since is larger than $

the f,
absolute miximum of the function occurs at x- 4, the right endpoint. Since

|, the absolute minimum of the function is at x: I' the

-f is smaller than Figure n
relitive minimum, 30 shows the various points'

*it (4, $), max-imum

,01
15

l0
(-2, 19) 5

-[), *ini-u*

fr6unE 30

5 PROBLEMS

In Problems I through 14, find the absolute maximum and

8,f(x): xa-2x2 +1, -53x<5
ui*fur. minimum of th. gi,.n function on the interval x,
indicated. e/(x):;i'-i<'<I
I /(x): x2+2x-4, -4<x33
2 f(x)= -2xz + x-3, -l <x<6 ro /(x) -:#, -3<x<r
3 /(x)= 2x3 - xz +2, *2<x<l
rtfi
a f@\:x3+2x2+ 18x, -l <x32 1l /(x):tanx, --.rS,
5 /(x): xx +3x2 -9x * 4' -5 < x < 6
6 f(x): xo -Zxz + l, -2Sx3l f\x)= 3x, O=,
7 f(x) : va -2x2 + l, 0 <x =< 5
12 sin
= I
 149
sEcTtoN 4.s Txt iuaxruurr,t AND MINIMUM vAtuES of A FUNciloN oN AN tNTtRvAL

ft +2
13 /(x) : ,6 sin x - cos x, --<x<-
7t
20 /('x) : x2
3- -4 x +2' -2<x<0
2n I
2l /(.x):
M f(x):cos22x, -; <x< 3 x+l :, 2<x--n
*_ I
l5 Give an example which shows that the Rule of thls section 22 f(x):^.r+l;, *<n <x<-2
may fail if the function considered is not continuous.
n5n
In each of Problems 16 through 24, find the absolute maxima 23 f(x) - -<x<-
sec x,
3- - 6
and minima of the functions on.the interval indicated. or show
that there are none- 24 l(x\:ran2x. -1.t.1
16 f(x):y'1, -2<x<l 6-'6
25 Given /(-x) :.r3 + ax2 + bx + c defined on an interval
l7 f(x) -- x3 --2x21-x- l, -3<x< J
ldr,drl. Show rhat if a2 <3b then the zibsolute maximum
and minimum of / must occur at the endpoints.
x-t i 26 Letrbearational number. Let f (x):(1+.x)'-(1+rx) be
defined for - 1 < x ( a. Locate the absolute maximum and
Y2
19l(x): -;. l<x<3 absolute minimum values of /. Show that if r > I, then the
x-l minimum must occur at x:0.

6
F

APPTICATIONS OF MAXIMA AND MTNIMA

Until now we have encountered polynomials, rational functions, trigono-

metric functions, and so on. once we know these functions, we set out to find
the various relative maxima and minima, maxima at endpoints, and such
other properties as concavity and points of inflection,
In applications, however, we usually encounter problems in which we
must discover the function itself before we can discuss its properties. Having
found the function, we then use the methods described in the previoui
sections of this chapter to determine the function's characteristics. In order to
be equal to the task, the reader should memorize the formulas for areas and
volumes of simple geometric shapes, as listed below. In many of the problems
these formulas turn our to be likely candidates for functions

Elementary Geometric Formulas

150
CHAPTET /r Arpucmons orrut t BivArvE

If a manufacturer wants to make a tin can at the lowest possible cost, he

faces the problem of finding aminimum. Making the strongest possible bridge
of a certain size, type and span presents a problenl of finding a maximum.
Whenever we use words such as largest, most, Ieast, smallest, best, etc., we can
translate them into mathematical language in terms of maxima and minima.
If we have a specific formula for the function in question (a situation which is
very often impossible in actual practice), then we may be able to use the
methods of the calculus to find the required maximum or minimum.
We start by giving several examplet of the,operating technique and then
outline the procedure by giving a set of rules consisting of five steps.

EXAMPLE 1 A man has a stone wall alongside a field. He has 1200 meters of fencing
material and he wishes to make a rectangular pen, using the wall as one side.
What should the dimensions of the pen be in order to enblose the largest
possible area?

Solution Here we see that the problem is to find the largest area; clearly this is a
maximum problem. If we can find tf,e/area as a function of something or
other and then differentiate it, we may be able to find the maximum of this
function and so get the answer. Let us draw a figure (Fig. 31), and call ihe area /
I riili:,i::ia:iirj:fliillaliii=ritliiirli,,'li*:],,,',' I Wall of the pen by some letter, say .4. The length and width of the pen are both /
unknown. Ho*ever, if x is the width, thenth€ length must be 1200 - 2x, since
TIGURE,31 there are 1200 meters of fencing to be used. Now we can get an expression for
the area, namely, length times width or

" A : A(x): x( 1200 - 2x) = l2CfJ; - 2x2.

We note that x must lie between 0and 600, since otherwise one side would have
negative length. That is, 0 S x < 600. The derivative gives us

A'(x)=1200-4x,
and this vanishes when x: 300.
:T The Second Derivative Test tells us that
I

A": -4,
l0
and so a relative maximum occurs at x: 300. Actually, it is the absolute
maximum, since z{(x) is a quadratic function which is concave downward
everywhere. In fact, ,4 : 0 when x:0 and x : 600, so the endpoints give a
minimum. We conclude that the width of the pen must b€ 300 meters and the
16 length 1200 - 600: 600 meters. n
FIGURE 32
EXAA{PLE 2 A rectangular box with an open top is to be made in the following way. A
piece oftin 10 cm by 16 cm has a small square cut from each corner.(shaded
portion in Fig. 32) and.then the edges (dashed lines) are folded vertically.
What shcruld be the size o{ the squares cut out if the box is to have as large a
volume as possible?

Solution We se,e that if we cut out exceptionally small squares the box.will have
practically no hdight at all and so wiil have a snnall volurne. Similarly; if the
squares are too large the base of the box will be tiny and again the volume
will be small. We seek a formula for the volume. We let x be theedge of oneof
the squares cut out. Then:
 t51
sEcTroN 4.6 ArnrcnnoNs oF M xtM AND MtNtMA

16
- 2x will be the length of the base of the box,
l0 - 2x will be the width of the base of the box,
x will be the height of the box.
The volurne" of course, is length times width times height. Therefore, calring the
volume l,', we have
V : V(x): (16 - 2x)(10 -2x)x and - l3x2 + xr), I/(.x) = 4(210x
which gives volume as a function of x. Now we must note that there are
restrictions on x. First of all, x must be positive in order to make any sense at
all. Secondly, x must be less than 5. For, once it gets to 5, the width of the base
is zero and we have no box at all. We write

0<x<5.
With this in mind we calculate the derivative:
V'(x) :4(40 - 26x + 3x2),

and by solving

40 - 26x * 3x2 :0,

we obtain

x=2,61.
But 6t is outside the interval of interest and we reject it. we note that
V"(x) : (-26 + 6x) and V',(21 <0,
which gives a relative maximum at x :2. The endpoints of the interval [0, 5]
both give zero volume. Therefore according to the Rule in section 5, x : 2 cm
yields the absolute maximum. The volume of the box is (2X6Xl2) : 144 cm3.
t_l

EXAMPLE 3 The sum of one number and three times a secdnd number is 60. Among the
possible numbers which satisfy this condition, find the pair whose product is
as large as possible.

Solution We start by letting x be one of the numbers and y the other. We want to find
the maximum of the product, which we shall call p. So we write
p: xt.
We have expressed P in terms not of one quantity but of two. We can overcome
this difficulty by recalling that the first sentence of the problem asserts that

x * 3Y: fQ'
We can eliminate either x or y; if we eliminate y, we obtain
. P: xi(60- x): 2ox-txz.
Since P is a function of x, we can differentiate. We write

P':P'(x):20-3x.
CHAPTEB /t APPUcaTtoNs or rHE DERIVATIVE

Setting this equal to zero, we find that x:30 and /:

10. It is easy to verify
that this valuc x
of gives the maximum value to P. The result is the same if we
eliminate x. express P as a function of y, and differentiate. n

using these three examples as a guide, we now list the steps to take in
attacking this type of problem in maxima and minima.

With this five-step procedure in mind. we now study two more examples.

EXAMPLE 4 Find the dimensions of the right circular cylinder of maximum volume which
can be inscribed in a sphere of radius 12.

Solution When a cylinder is inscribed in a sphere we mean that the upper and lower
bases of the cylindcr have their bounding circles on the surface of the sphere.
The axis of the cylinder is along a diameter of the sphere. We first draw a
1l
n(' cross section of the inscribed cylinder, as in Fig.33 (Step 1). The cylinder
i might be short and fat or tali and thin. We label the radius of the base r and
the height of the cylinder trr; the volume is denoted by Iz (Step 2), Then
(Step 3),

v : trzh.

HGU'iE 33 We hir-ve rrs\{ e xpIesSed the volunte i/ in terms of the two quantities r and h, so
wc must eliniinate one (Step 4). Referring to Fig' 33, we use the Pythagorean
Theorern to get
Lz
,' * :144'
;
Solving for fi, we get
h:2a,744 3,
and substituting this value of ft in the formula for 4 we find

Y:2nr2Jtu-i, 0<r<12.
 1s3
StCTlON,r.6 APPIICATIoNS or MAXIMA AND MINIMA

By differentiating (Srep 5). we obtain

V' (r) : 4nr r4+l -7 * 2xr2 . t1t++ -- r2) - t,2(

- 2r)

- r*..
_n/,\r' ,_r_, _Jii=l
A4- - -:y::
2rr(288-2r2 *12)
JT44-
Setting this equal to zero, we find that or r:0 r:
+4_/G. We reiect r:0
and the negative value of r. Since the Second Derivative iest is rather messy.
we reason as follows: Izis cofrtinuous for 0 <r < l2;when and r = 12, we r:0
get V:0. The volume is positive for r in the interval (0, l2). Z takes on its
maximum on [0, 12] at somero which must be interior to the interval. Thus
V'(r) :0. But Z'(r) : 0 for only one value of r in (0, l2). so that value mtrst
be 4r,?. Therelore the crirical value ro :4J6must give the maximum. From
the relation between r and h we find that h: gv6. [,] i

EXAMPI-E 5 A lighthouse is at point A,4kmoffshore from the nearest poinr o of a straift

beach; a store is at point B, 4 km down the beach from o. If the lighthouse
keeper can row 4 kmlhr and walk 5 km/hr, how shourd she'proceed in order
to get from the lighthouse to the store in the least possible time?

Solution clearly she will row to some point on the beach between o (the nearest point)
and the store at B and then walk the rest of the way. we drav; a figure such as
Fig. 34 (Step 1), where c is the spot on the beach where she landsind x is the
distance from o to c. we denote by r the time ol the trip from ,4 to B
(step 2); r is the quantity we wish to make a minimum. To get a forrnula for
T, we use the fact thatrate times time equals distance. We then have
* I: time of trip along ,4C + time of trip along CB
,1 tre"ctr
* di:glglg
| distance,4C
4 ' ..'-..*l
45
To express this in terms of one quan'"ity is not hard. since
CB:4-X,
and. using right triangle AOC.

AC:\f,C+?.
Therefore (Step 3),

r(r):+$iii)+|14-rr.
The only part of Step 4 we need is the determination ol the domain of the
function r. Since C is between o and B we have the restriction
0<x<4.
Taking the derivative (Step 5). we obtain
T',(x) : t( 16 + .x2)- r/2(2x) -+
 154
CHAPTER 4 APPTICATTONS OT THT DERIVATTVT

and. setting rhis equal to zero, we find that

_--:: -(

41rc + x2
We solve for x:

,: * t6, .

But these values of x are outside the intervar [0, a]. The concrusion is that
there are no relative maxima or minima in trre intJrvur o < , a, Th".efo.e
<
the absolute minimum must occur at one of the eniipoints. we have

r(01 : ?
and

'rg):.,/1.
Since yD<!, the fastest method for the Iighthouse keeper is to row directly
to the store B and do no walking. Ll

6 PROELEMS

N rectangle
has a perimeter of 120 meters. What length and 9 The top and bottom margins of a page are each I cm and
--l width yield the maximum area? What is the result when the I
the side margins are sach I cm. Ii the area of the printed
pefimeter is L units? material per page is fixed at 30 cm2, what are the dimen-
2 Find the dimensions of the rectangle of maximum area that sions of the page of least area?
can be inscribed in a circle of radius 6. What is the result for l0 A closed right circular cylinder (i.e., top and bottom
a circle of radius R? included) has a surface area of 100 cm2. What should the
3 Find the dimensions of the right circular cylinder of radius and altitude be in order to provide the largest
maxirnum volume which can be inscribed in a sphere of possible volume? Find the result if the surface area is
S cm2.

-4
radiu: R. I I A right circular cone has a volume of 120 cm3. What shape
A horizontal gutter is to be made from a long piece of sheet
. should it be in order to have the smallest lateral surface
jron 8 cm wide by turning up equal widths area? Find rhe result if the volume is I/cmr.
along the edges
into a vertical position. How many centimeters should be t2 At midnight,
turned up at each side to yield the- maximum carrying
ship B was 90 km due soulh of ship l. Ship I
s.ailed east at 15 km/hr and ship B sailed nqrth at 20 km/hr.
capacity? At what time were they closest together?
5 The difference between two numbers is 20. Select the 13 Suppose that in Example 5 on page 153, the lighthouse
is
numbers so that the product is as small as possible.
_ 5 km from shore, the store is 6km down the beach from O,
6 A box with a square base is to have an open top. The area of the lighthouSe keeper can row 2 km,/hr, and she can walk
the matirial in the box is to be l00cm2. What should the 4kmlhr. Where should she land in order to get from rhe
dimensions be in order to make the volume as large as lighthouse to the store in the shortest time?
p+ssible? What is the result for an area of S square !4 Find the dimerrsions of the rectangle having the largest area
cen timeters? which can be inscribed in the ellipse
7' A window is in the shape of a rectangle surmounted by
a semicircle.. Find the dimensions when the perimeter is
12 meters and the area is as large as possible.
Il* { :,
d'

8 Find the radius and central angle (in radians) oflhe circular 15 Find the coordinates of the poin! or points on the
sector of maximum area having a perimeter of l6 cm. curve
l:2x2 which are closest to the poinr i9, 0).
 155

SECTION 4.6 APPTICATIONS OF MAXIMA AND AAINIMA

16 Find the coordinates of the point or points on the curve 30 Find the dimensions of the right circular cone of minimum
x, - y': t6 which are nearest to the point (0,6). volume which can be circumscribed about a sphere of
radius R.
t7 (a) A right triangle has hypotenuse oflength 13 and one leg
of length 5. Find the dimensions of the rectangle of largest 3l A right circular cone is to have a volume Iz. Find the
area which has one side along the hypotenuse and the ends dimensions so that the lateral surface area is as small as
of the opposite side on the legs of this triangle' (b) What is possible.
the result for a hypotenuse oflength H with an altitude to it 32 A silo is to be built in the form of a right circular cylinder
of length ft? surmounted by a hemisphere. If the cost of the material pcr
18 A trough i! to be made from a long strip of sheet metal square meter is the same for floor, walls, and top, find the
12 cm wide bfturning up strips 4 cm wide on each side so most economical proportions for a given capacity Iz'
that they make the same angle with the bottom of the 33 Work Problem 32, given that the floor costs twice as mueh
trough (trapezoidal cross seption)' Find the width across per square meter as the sides and the hemispherical top
the top such that the trough will have maximum carrying costs three times as much per square meter as the sides'
capacity. 34 A tank is to havb a given volume 7 and is to be made in the
19 The sum of three positive numbers is 30. The first plus twice form of a right circular cylinder with hemispheres attached
the second plus three times the third add up to 60. Select the at,each end. The material for the ends costs twice as much
numbers so that the product of all three is as large as per square meter as that for the sides- Find the moft
possible. economical proportions. /
20 The stiffness of a given length of beam is proportional to the 35 Find the length of the longest rod which can be carried
product of the width and the cube of the depth. Find the horizontally around a corner from a corridor 8 m wide into
shape of the stiffest beam which can be cut from a one 4 m wide. (Ilinr:Observe that this length is the
cylindrical log (of the given. length) with cross-settional minimum value of certain lengths.)
diameter of 4 cm. :4
36 Suppose the velocity of light is in air and I/, in water. A
21 (a) A manufacturer makes aluminium cups of a given r"y of tigttt traveling from a point P, above the surface of
volume (16 cm3; in the form of right circular cylinders open the liquid to a point P2 below the surface will travel by the
at the top. Find the dimensions which use the least material. path which requires the least time. Show that the ray will
(b) What is the result for a given volurne Iz? cross the surface at the point Q in the vertical plane through
22 In Froblem 21, suppose that the material for the bottom is P, and F, so placed that
1j times as expensive as the material for the sides' Find the sin0, sin 0,
dimensions which give the lowest cost. vr=kt
23 Find the shortest segment with ends on the positive x and y
axes which passes thiough the point (1,8). where 0, aad 0, are the angles shown in Fig. 35.

24 The product of t*o positive numbers is 16. Determine them

so that the square of one plus the cube of the other is as
small as possible.
25 Find the dimensions of the cylinder of greatest lateral area
which can be inscribed in a sphere of given radius'R.
26 A. piece of wire of length L is cut into two partsj one of
which is bent into the shape ofa square and the other into
the shape of a circle. (a) How should the wire be cut so that
the sum of the enclosed areas is a minimum? (b) How
should it be cut to get the maximum enclosed areas?
27 Solve Problem 26 if one part is bent into the shape of an
HCURE 3s
equilateral triangle and the other into a circle'
28 Solve Problem 26 if one part is bent into the shape of an
equilateral triangle and the other into a square. 37 A manufacturer makes widgets td sel. so that an order oi
29 Find the dimensions of the right circular cone of maxi- 1000 costs $3 per hundred. For each additional' 100 in the
mum volume which can be inscribed in a sphere of given order, the price is reduced by 6l per hundred. What size
radius R. order will yieid the maximum dollar value?
156
CHAPTIR { Arrucmom or rHE DERrvATrvr

7
Tur orrrrRENlrrAr. AppRoxrMATloN

Until now we have been dealing with functions whose domain is a set of real
num'bers(i.e., a set in R1) and whose range is also a set of real numbers (a set
in R1)..Symbolically, if D is the set of numbers comprising the domain
of a function / and S is the set of numbers comprising the range, we may
wite f : D -* S. In the case of functions given by formulds, we write, for
example, f (x) : x2 - 2x * 7 to indicate the particular function.
We now enlarge'the notion of function by considering domains which are
ordered pairs of numbers instead of numbers in R 1. That is, we consider the
case in which an element in the domain is a point in R2. The range is the same
as before: a collection of real numbers (a set in R1). These functions are
defined as functions on R2, and we give the following precise statement.

DEHNlrtoNs Let D be a set bf orilered pairs oJ'real numbers (x, y), i.e., D is a set in Rz. Itt E
be a set of real nufiber* ,4 function f from D to E is a correspondence that
assigns to eoch element (x, y) o{ D a unique number z where z is a number in E.
The set D is the domain of f and the set E is therange of f.
We recognize the above definition as a straightforward extension of the
definition of function given in Chapter l, Section 5, page 20. We also have the
following equivalent definition, which extends the one given on page 20.

DEFINITIONS Consider a collection of ordered pairs (A,w), in which the elements A are
themsehses ordered pairs of real numbers and the elements w are real numbers. If
no twd members of the collection haue the same item A as afirst element-Le., if
it can neuer happen that there are two members (Arw) and (Ar,w2) with
wr*w2-then we call this collection c function on R2. The totality of pos-
sible ordered pairs A is called tfte domain of the function. The totality of
possible ualues for w is called the runge of the function.

We riow revise our terminololy and call a function as defined onpage 20

a fu4ction on Rr. We use the same letters /, F,0, g, H, elc. to represent
functions on R2 that we usgd for functions on R1. However, the formulas used
for describin g particular functions on R2 appear quite different from those we
have used until now. If the pair (x, y) is used to represent a typical element in
the domain of a function F on R2, a particular function may be written
F(x,y) - x2 -2y? +3x+4y.
If the letter z is used as a typical element in the range of a function F, we write
z: F(x, y).
Employing classical terminology, we call a function on R2 a function of
two variables, while a function on R 1 is called a function of one variaHe. Thcse
terms, although not quite precise, have considerable intuitive value, and we
shall use them from tirne to time.*
*It is not difficult to extbnd the definition of functicn to the iase in wliich the domain
consists of ordered triples, ordered quadruples, etc., of numbers and in wkich the
range also consists of ordered pairs, triples, etc., of numbers.
sEcrloN 4.7 THE DrFrExENTrAt. APPRoxtMATtoN

The letters used for elements of the domain in R2 are not restricted to x
and ;. Simple examples of functions of two variables are
z:u2 +2uz
j'' -2
+3ta. ^,:*' !l+.,r- - [ : I'11.' n t,
4_ 2u2- r2'
The domain in each of these formulas is not described explicitly. In each such
gnsg ui€ shall always assume that the domain consi.sfs oJ- all possible rtalues
which may be substituted in the-right side of the formula.
We are now ready to define a function which we call the dilfere;ntial.
/
Suppose that is a function of one variable.

DEFTNITTON lVe designate by dJ' the difrerential of f which is a function qf ttto uuriables
gitten by the formula

in this definition, the symbol d/ is considered to be a single entity- and

does rrof mean d iimes /. Since d/ is the symbol for a function on R2, lve
recognize df(.x, h) as the symbol for the value of the function when x and h are
substituted into the right side of the above formula.

EXAMPIEI Giventhat/(x):.r2(l-12.lt!2.findtheformulafordJ'.Also,findthevalue
of df(1.t!.

Solution By differcntiation. we havc

l"'("x) : 2x(l
* x2)tt2 *x3(1 - x')- ti'..
Therefore
d.l'\x.h): [2.x( | - x21ttz x3(l - x'\- tt2f 'h
-
: (l - r21- tt2l2x(.1- x2; - x31h

Substituting a :j and lr : 3 in this expression, we get

df (+,3) =i.,f;- Ll

Earlier (in Chapter 2, Section 2) we introduced the symbol Lf by

defining
--1
Q(a*h,l@*h))
2-
n'---T Al ,.r, h t
Lf :f (x + h) -.f (x).
We did not say so at the time, but the reader can now see that A / is a function

P(x,f(r))
dfti,ht I
on R2. That is, we may v,,rite A/:
A/(x, fi).
s----l-------a It wili help our understanditrg to compare the functions d/ and Al. We
can dc this geometrically. as shown in Fig. 36. v''hcre P reprcsents a poin'. on
the graph oi-l:7-(-x). trf ft has any value,*'.tren the point {x + h, !t,x + h)) is

Comparison of dJ' and Lf * Figure 36 is drav, n u ith h positive. lf'h is negative, the point x r- /r lies to the left of .x.
'I'he discussion is unchanged.
F6URE 35
CHAPTIR 4 APPLICATIONS Or THI DTnIvATIvE

on the graph and is denoted by p. We see that

Lf (x,h): height S0.

The tangent at P intersects the line through Q and S at R, and we know that
SR/PS is the slope of thrs line, or simply /'(x). Since PT: h, we have

df(x,h): height SR'

This tells us that the height RQ is just the diflerence between A/ and d!.That
is,
Lf - df : height R0'
Also we see from the graph that as ft tends to zero (thus implying that the
point p slides along the curve to P), the difference between A/ and d/ tends to
zero. In [act, we have

"\!J#:o
The proof of this statement is easy. We know from the definition of
derivative that

!h -f,t*l as h -.+ o.

On the other hand, lh is the derivative f'(x). Therefore

df

Lf-df Lt df
l,*-t-:lTf, -lT, :f'(x) -.f'(x):0.
If ir is small, A/ and d/ must be close together. The function A/ measures the
change in / as we go from one point to another. The differential df can be
used as a good approximation for A/ if the points are Rear each other. We
write

for this approximation.

EXAMPI-E 2' Given that /(x): l/x, compute df and. L,f when x: 1 and lr is 0.1, 0,01,
0.001. Do this to four significant figures.

Sotution ,f(1) : l; f'(x) - -Ux' and /'(1) : - l. For h:0.1,

/(1.t1 :6.99999, and A/:0.909091 - tr : -0.09091.
On the other hand, df :f'(l).h- -l .(0.1): -0.10000. In this way we
obtain the table:

hdfLf
0.1 -0.10000 -0.09091
0.01 -0.010000 -0.009901
0.001 -0.0010000 ' -0.0009990 D
sEcTloN 4.7 TIG DTENEMI/TL Af?TOXIMATION

The point of this example-as the reader can readily'see if she has
worked along with it-is that the quantities t!' require practically no
computing, *hil" finding A/ is quite a bit of work. Since the error made in
usine ar instead of a/ is small, the saving in computation time may make it
worthwhiie. especially in very complicated problems'

ExAMptE 3 Use the differential to comput" tX approximately'

JTi exut'tly
Solution To work this problem wb need an idea ;it is simply that we know
and 33 is "close to" 32. The function we are talking about is

/(x):..7 C : x''s.
If we think ol x ---32and h : 1, then / (33) is what we want to find. we know
that
Lf :f (33) -f {32).
while
dI : I', Q2l
. (t).
The derivative of / is

I'$):tx-ats,
and
' I'(321:*'*1: #' df :#'
Using the fact that d/ is almost the same as A/, we write

IG3):f (32\ + Lf xf (32) + df .

(The symbol means "nearly equal.") Therefore

=
Xn:f el) =2+#:2.0125,
approximately. The actual error in this computation is less than aabo'
Of course the fifth root of 33 can be found easily with a hand calculator.
The point of Example 3 and other problems of this tyOe is to illustrate the
po*".r of the differential in making approximations without intricate comput-
ations. There are many problems, especially where symbols rather than
numbers aqe manipulated, in which differentials are crucial in simplifying
approximate formulas.
A measurement of a length of pipe is made, and it is found that there is an
cm. Is this error large or small? If the length of the pipe is about
error df I-would
8 cm we say that the error is pretty large. If the total length measured
were j km, we would consider the error extremely small. The error is the
difference between the planned length and the actual length'
If a quantity is being measured and the true value is a but there is an error
of an amount h. we define the proportional error as

L
a

and the percentage error as

!.,oox.
a
 cltAPTtn a

In the above example of the ]-cm error, when a: g cm and i:

proportional error is S, and the percentage error is
! on, the
and ft:f cm, then the prLportion-ul
fi.100:Olilr. tt
':lkm error is 0.0001%-one ten-thousandth is I/1OO,OOO and the
percentage "rro,
of one percent.
suppose that / is a function and we wish to find
/(a),ihe varue a to be
determined by measurement. This measurement is not p.""ir. (measurements
never are) and there is an error Ir in measuring the ,uiue o. The error in the
function is L'f : f (a + h) f (a) and the proportional error in the function is
f -
simply
Lf f (a+ h) -I@)
f Ik)
However' in such computations, we conveniently use the approximation
to
the proportionrl erron

The approximst€ percentage error is then found by murtiplying by 100.

EXAMPLE 4 The of a sphere is found by measurement to be 3 cm, but there is a

.radius
possible error of *0.03 cm in the measurement. Find, approximately,
the
errol and percentage error in the value of the surfac. ur"u-of the sphere that
might occur because of the error in the radius.

Solution The surface area s is given by the formula s:4nr2,and for r:3 the
area
is 362 cm2. The error is approximated by ds:s'(r)h :gnrh. The quantity
;1: 10.03, and so the approximate error is
dS: 8z(3)( +0.03) : t0.72r cmz.
The approximate proportional error is

ds *0.72n
s 36n
: *0.02,

and the approximate percentage error is *2o%. tr

7 PROBTEMS

In Problems I through le find ilf(x,h). In Problems I I through 2e find df and Lf and evaluate them
for the quantities given.
I /(x)- xa+3x2-2x Z tt-l:fi ll f (x) = - 1, x: 1, i:0.01
x2 +x
3 /(x)= x'$i 4f(x):*#+ 12 f(x):x2 -2x--3, x= -1, h: -0.02
13 /(x) = x3 + 3x2 - 6x - 3, x:2, i:0.01
5 f (x) : (x2 + 2)tt2(2x + Dv3 U f(x):x1 -2x2*3x*4, x: -1, h:0.02
6 f(x)=(x2 + l!tt"1y,]- 21ts 15 f(x\ : 1116, x:2, ft :0.05
7 /(x) :5in 3;g S ,f (x) : x cos3 2x 16 f(x): ytrz, x: l, h:
-0.1
9 f (x) = tan +x + sec 2x l0 /(x) = sec3 4x + csc x l7 f(x) : a- ttz, x : l, ft :0.1
sEcTloN a.T TtE DIrIEnB\TTIAL AP?ROXIMATION

x _ ^_ computed. The diameter is 9 cm, with a possible error of

tt /(x) : , . ,, x:0,- h:0.1 *0.05cm. Find (approximately) the maximum possible
pcrcentage error in the volume.

t9 /(x) :xsin2x, r=1' ;r=g.5 . The radius of a sphere is to be measured and its volume
computed. If the diameter can be accuratcly measurcd to
within 0.1%, find (approximately) thc maximum perccn-
2Q f {x\: cos2 x + 2 sin 3x, ,:\' n:O'Z
tage error in the determination of the volume.

In Problems 2l through 28, calculate (approximately) the given 3l A coat of paint of thickness I cm is applied evenly to the
faces of? cubc of edge a cm. Use diffcrentials to find
quantity by means of the differential' Then use a hand cal'
approximately the number of cubic centimeters of paint
culator'to comPute the quantity to 2 decimal placcs and
used. Compare this with the exact amount used by compu-.
compare the results.
ting volumes before and after painting.
il J65 22 :@
32 Work Problem 3l with the cubc replaced by a sphcre of
2s (0.e8)-' 24 {8O radius R.
25 (lt;'r, % (tt1-'tn 33 State precisely the definition of a function when thc domain
consiJts of elements of R3 and the range is a set in Rt. Do
zt Juwzt n Jn+ln the same when the domain is a set in R3 and the range is a
29 The diameter of a sphere is to be measured and its volume set in R2.

DIFFERENTIAT NOTATION

Let / be a function of one variable. Writing y=f(x\, we recall that in

chapter 2, section 4 the symbor j| *", introduced as an alternate notation

for the derivative /'(x). At this point we identily dx with the numbcr lr used
in the definition of the differential, and we identily dy wilh the difrerential d/.
That is, setting
ilx: h and ily : ilf ,

we obtain the equivalent formula for the differential

We call dx the difrereatid of x and dy tb difierenthl of y. lt is important to

observe that dx is an independent uariabte and dy, which is defined by the above
formulq is not. As expcctd the ratio dyldxis the derivative whenever dx *0,,
aad the use of tliis notation for the derivative is justified in terms of differentials.
Suppose that y :/(x) is a function and that x = g(t) is a second function
with the range of g in the domain of /. Then we may consider y as a function
of r by writing
,- --f lctt)).
To get tbe derivative of /te(r)l we apply the Chain Rule to obtain

f'Ls$l)d$\.
162
CHAPTEf, { APPTICATIoNS or Tffi DERIVATIVT

We also have the following formulas for diflerentials:

dy:f'(x) dx and 7y: g,(t) dt.
By direct substitution, we find
dy:f'(x)s'(t) dt.
The chain Rule may now be expressed in terms of differentials. If dx * 0 and
dt +0, then

Also,

EXAMrLE 1 Given that y: (x2 + 2x * 1)3 and x:3t2 + 2t - l, find dyldt.

Sotution Here we have

dy dy ilx
dt dx ih
and

dx
,**' r 2x * t)2(2x + 2),
*,: *:6t+2.
Therefore

*:
dt
6(x + l)(x2 + 2x * t)2 . 2(3t + t).

since y is a function of x and x is a function of r, we see that y is actually a

function of r. By replacing x with 3t2 + 2t l in the expression ior y, we could
-
simply differentiate y with respect to r. If we substitute for x irr the above
dv
expresslon lor we get
*,
dv
+2t)l(3r2.+2.t-t)2 +2(3t2 +2t_ 1)+ 1l' .2gt+
i:6(?t, 1)

and x is eliminated from the final result.

 163
stcTtoN 4.8 DtFFERthmfl Nor noN

ExAMptE 2 Given that y: jVi 112*- 3), find dy.

Solution One way would be to find the derivative and multiply by dx. But we may also
- use the formula for the differential .of a quotient, and we do it here for
practice:

dy:
(zx-t1i($a7- x2+ 'd(2x - 3)
(2x - 3)

We see now that dQ? +T : +$2 + l)- Lt22x dx and d(2x - 3) :2 dx.We
obtain

x(2x - 3) (x2 + l\- 't' dx * 2

dy: (2x - 3)2

We multiply both the numerator and the denominator by to get

dy: -3x-2 dx. l

$ +r1zx-t1'?
The method of diflerentials is particularly helpful in implicit di{ferentiation.
Suppose two functions, each defined on Rr, satisfy a relation. We may take
differentials if one of the functions is expressible in terms of the other. However,
the actual process of solving for one quantity in terms of the other need not be
carried out. An example will illustrate the idea.

EXAMPLE 3 Suppose that u and p are functions defined on Rr and that they satisfy the
relation

u2+2uu2+u3-6:0.
Find duldu.

Solution Takedifferentials:
2u du * 2u'2u du + 2u2 du + 3u2 du :0.
Divide through by da and solve:
du 2(u + u2)
du: - 4lrui5n'
if du*0 and if 4uo +3u2 *0. n
ll y:f (x), there are now two symbols for the derivative: f'(x) and dyldx.
These are the most prevalent symbols, commonly used in texts and papers on
various related subjects. Another symbol, not quite so common but neverthe-
less used often, is D,/. The only notation we learned so far for the second
derivative is l"(x). The expression
dry
d?'
which is read: d second y by dx second, is a classical one for the seconci
derivative. The numerat or, d2 y, and the denominator, dx2, haue absolutely no
meaning by themselues. (In elementary calculus there is no such thing as the
 CHAPTEN APfl.lcATtoNs ot rHt DERtvATtvE

differential of a differential.) we just use d2 yldx2 as an equivalent for

/,,(x).
Similarly, third, fourth, and fifth derivatives are written
dty doy dty
N, 77, 7F,
and so on; In each case, the expression is to be thought of not as a fraction
dividing two quantities but is an inseparable ,yiruot representing the
appropriate derivative.

8 PROBTEMS

In Problems I through 18, find the differential dy. 29 Find, dsldt, given that s:x2+3x-6, x:r3-gr*5,
I y:(x3 -3x2 +2)s I y:(x2 q lltt3 r: v4t+ 5.
30 Find dzlds, given that
3 y=1x3+4)-5 n n=fi
' /+l
x3-8x+l
y:__::,
y-1' X:s3_gs+5.
5 y:x2J2x+3 6 y:(x- t)y!a 1 ./x+l
7 y=(x+ l)2(2x- l)3 E y:("x +21,1$- l;ur 3t Find dyldt, given that x3 + 2xy - y3 + g :0 and x : 13
2x x -Zt + l.
e Y:;7;J l0 y:--: 32 Find dyldt, given that
Jx'* I

tl Y:
x2ll
t2
,F*1 xa+2x2y-yo-3y:0, x=uET\.
GTJy/3 x 33 Find dzldt, given that
EN
13 v:Jx_r
x +Zyt -322 :1, y: +Zx-6, r:(r+
,
t4 tr:-- _ z3 x2 l)4.
1/x'+x+l 34 Find dyldx, given that
15 ,y =.x2 sin 2x
16 y:cos 3x*lan2x x3 + t2 -z$ - l=g, yo - t3 +Zy -7 :0.
17 y: x tan2 tx *35 Find d2yldx2, given that

18 /: sec(x + 4) + cot2 2.x x3-4y4*7x-6y-3:0.

In Problems 19 through 26, find the derivative dyldx 36 a) Given y -f (x), x : g(r), r = }(s). State appropriate
by the conditions which make the formula
method of differentials.

19 2x2 + xy - y2 + 2x- 3y + :0 5
dY
:4!.d* .!
ds dx dt ds
20 x3 + x2y -2y3 =Q n ,fr +,f,:2 valid.
22 x2l3 + y2t3 - a2i3, a const. b) Extend the above formula to the case of n functions
23 2x3 - xyz - y3 + 2x - -y:0 f ,, fr, ..., "f,.
24 x2cos2y*x2y3-l=0 *37 Civen the relations

!J xy cos2x + 2x2cos y - : 0 : l,
- t, *
3 x2 + 2xt + y, 3xy - 2t3
?5 x3 + 2r' y -3x cos x sin y - 4: 0
2x3 - rt + 2xy2* 3yt3 +.t = 3.
21 Find dyldt, given that y:1x2 + 2x + 5)3/a and x: Use differentials to find a formula (in terms of x, y, r) for
+I
"Grlt
Fld dzldr, given that z: (2r+t)1fu2 *u)
the derivative dx/dt. (Hint: Observe that theoreiically y
23 and u: may be eliminated between the two equations and x
(r2 + 5)a.
expressed in terms of r.)
 165
sEcrtoN 4.9 NEwrotls MtrHoD (omoNAt)

NEWTON'S METHOD (OPTIONAT)

In many calculus problems as well as in problems throughout mathematics

we often seek solutions of an equation f (x):0. The values of x which satisfy
such an equation are called its roots. These numbers are also designated the
zercs of the function f . We already had practice in finding roots of polynomia!
equations and equations involving trigonometric functions when we were
finding relative maxima and minima and points of inflection. In most
problems the roots of an equation cannot be found exactly. This is
particularly evident if the equation contains a polynomial of high degree or if
it involves combinations of trigonometric functions. Therefore it is important
to have a method which approximates a root to any desired degree of
accuracy. Newton's Method is a simple and powerful technique for accom-
plishing this task.
If we plot the graph of a function y : f (x) which we suppose is smooth,
thel each place where the curve crosses the x axis is a root of /(x) = 0, i.e., a
zero of f .If for some value a we have f (a) <0 and if for some value b we have
:
f (b) > 0, then we know there is at least one root of /(x) 0 between a and b.
Suppose we choose some value of x between a and b, designat'ed x, , as a first
guess'for the root r of f(x\:0, as shown in Fig.37. To get a closer
approximation to the root r we draw the line L tangent to the curve at the
point (xr,
"f(xr)). This line intersects the x axis at a point which we designate
xr. If the situation is actually as it appears in Fig. 37, then x2 will be a better
approximation to r than x, is. In many circumstances this process is entirely
correct. To determine xr, the point where L intersects the x axis, we use the
point-slope formula for the equation of L. We find
y -f (x r) :/'(x,)(x - x1).

('r, f trrl)

Finding a root by Newton's Method

FIGURE 37

Since x: rz when ):0, we obtain

-f(x):f'(x1)(x2-x1)
CHAPTII.

and, supposing that

.f'(xr) # O we can solve for x, to get

xr=x. -IG,)
,(x
f r)'
In other worts, given the function
(x) andknowing a first approximation
to the roor r' the above formura f x,
mation to r. Once *".11:".r,
iii; ;;;* ro find ,r, a b"tt". approxi-
,.1g ;;; ;;; the process. Drawing the line
L2 tangent to the curve at
1xi, {U,ii,*l Jrno," rhe point where-l,
sects the x axis bv x,. rr,"n
,r-i, a';;;;pp-ximation to ,ii", i,inter-
,r.
,lT,;il"iill;i1il,:,T;, ir,li';,V,;,;)L *" ,o,*ur, r", h#,,g x, is

G", tto))

FIGUTE 38

procedure is now clear

I i_:li:Hif*:lj:. and the general term, known

is valid so long as
f,(xlis never zero in the interual
aSx <, :;; ; ;;;;;
;;
:[fl']"I]',1lill':":lg:tl*;"-i" ),r,* the above process yierds
a
#.#.'ff"il;?;;;;,f #l,Xl,l.n'd';;;;;:';:*?;::l'ff :l'"'"ff ,:
EXIrA{pr'E r Find the cube root of 3 with
an accuracy of three decimar
praces.
solution The key to solving the problem
is the observation that the
the positive.root of tt" equation cube root of 3 is
y =f (x) x3 _ 3 by setting
rr_l:;;;..fiIr,, o.u* the graph of
= up a table of values:
 167
sEcTloN 4.e NEwroN's METHoD (opfl oNAt)

The graph is shown in Frg. .19, and we note that there is a root between
I and 2. 'rve also observe tha( f'( y) = l.r'r is re Ve r Zerr) between I and 2. wc
choose any number in this intcrval" sav 1.5 as a fir-cl guess for r. Then by
Newton's [:ormula

r:: r') ./(I5)

iii's)
We find .F( 1.5) : ( l.-s)3
- 3: 0.175 and .f ,(1.5) : 6.75. Therefore
.xz: 1.5 - 0.05556 :1.44444.
Continuing the process, we obtain for -x.,

xt: 1.44444 - fUlyll

'(1.444441
: 1.444d4* 0.0021e

:1.44225.
-We
FICURE 39 carry out one more step;

x*: 1.44225 - f[:,'.o,*,?'^t!

'tt.+tzzs1
: t.44225- 0.00000268

:1.44224.
we sec that the third srcp in Newton's Method leaves the first four decimal
places unchanged and we conclude that 1.442 is the cube root of 3 with an
accuracy of three decimal places. T-l

Generally, if we require an approximation to a fixed number of decimal

places the computation should be made with rwo more places than is actually
needed. with a hand calculator the additional work is usually not lengthy.
Then when the iterations yield no change in any of the digits drop the twc *i
extra digits to get the final result.
Newton's Method is quite [ast. In most polynomial equations and
combinations of polynomial and trigonometric equations eaih successive
step doubles the number of digits o[ accuracy from the preceding step. Thus
accuracy improves quickly with successive iterations.
If the first guess x, is poor then it may happen that Newton's Formula
will give a value of x, that is a worse approximation to r than x, is. Figure 40
shows how this can happen. If the slope
/'(xr) is very small, then the rangenr
line L may intersect the x axis at a point which is farther away from r than x,
is. In such a case it usually happens that the successive valuls X1, x21 x3, ...
become dispersed rather than clustering about a specific root r. To guarantee
that the sequence x1, X2,..., x, actually converges to a root, we rJquire not
only that /'(x) does not vanish but also that f "(x) is not roo larle in the
x, is not sufficiently close to r interval a<x<b. A typical theorem on the convergence of frewton's
Eculf .10 Method, which we statc without proof, is given in the filowing result.

THEoRElvt 10 supposethereisanumber M > 1 suththntlJ'ix)lzM qndlf,,(x)l<2lMfar x

onl: {x;-a S x <b}. tf f (a) antl f (b) hru, ipp,or;6" rfg:r,r'on) *'ri rri prtrr
oflwithl"f(xr)l<l,thenthesequencext,x2>....,xn,...tleterminedbyNewton's
Formula conerges to a numbew such that f (r) :0. That is, r is a root of the
equation.f(x) :0.
 CHAPTER 4

We illustrate Newton,s Method with a second

example.

EUWpl"E 2 Find the positiverogt o_f the equation 2xa

+ 2x1 _ 3x2 _5x _ :0
accuracy of two decimal places.
5 with an

Soturion graph of y:!(x) + * 5 bv construct-

,Y: lTl,1**-the
rng a table of values:
=2xa 2x3 - 3!x2- 5x

From the graph in Fig. 41 we observe that the positive

root r lies between
I and 2, and that it seems to ire- somewhat closer ;"-t,i;;
,; t.'ii, .rroo."
I : l.q as a first guess. With f,tx):8x3 + 6x2 _ 6, _
Formula and obtain
t;;;;iyNewton,s

f!.a _,1.6. o.6ts2:

xz: 1.6 - fu: - SIjrE 1.6 - 0.d185 : 1.5815.

We calculate the next approximation:

RGURE 41 x":1.5815-/(1'5815)
- I '( 1.5815)
and we find /(l'5815) :0.114. we next carcurate /'( r.58r5), getting 32.1623.
Hence

xs - 1.5815 - #: 1.5815
- 0.0035

: 1.5880,

and the root is r: 1.5g accuratery to two

decimar praces. In this problem we
can simplify some of the compuiation by using
the varue of /, obiained in
finding x2 rather than by comfuting y,1iry aswe
did above. Iiwe use f,(1.6)
instead of /'(1.5815) in finding r. *i g.f'

xr : 1.5815 * 0.114
:
33.528
1.5815
- 0.0034.
The answer accurate to two decimal places
is the same. In general, 1f is
large then it won't change much throuih
successive steps and it saves time to
f,
use the same value of
/'
throughout. iire resurting
,,
r"qu"o." *iti.olu".g.
the root only slightry more siowly than wourd
Newton Formula were used.-occasionalry,
b" ih"
"ur"
ii. prr"i.,
ii-"o*-pututioo
makes the simplification worthwhile.
the reduction li
n
 169
sccTtoN a.9 NEwroNs l'tErlloD (oPnoM[)

9 TIROBLEI\,IS

In each of Problcrns I through 6, find the appropriate root with divide d, into c, getting a quotient qr. We average d, and
an a@uracy of three decimal places. Use the methods of this qr; that is, we form dz:iUt* 41). We then divide ilrinto
'scction c, getting a quotient q2. Repcat the process. Show that dr,
and then check your rciults with a calculator.

t\fr 2lzs dr, ..., dn, ... converges to .[ by showing that we have
described Newton's Method for the equatioo x2 - c:0.
3 J67 4 ,/84 12 Find allthe roots of 2x3 - x2 - 5x + 3 : 0 with an accuracy
of two decimal places.
s{n 6yio 13 Find {l the roots of 3x3+ x2-llx*6=0 with an ac-
7 Find the positive root of 2x4-3x2-5=0 with an ac- curacy'of two decimal places.
curacy of three decimal places. t4 a) Find the positive root of 3x3 + l6x2 - $x - 16:0 with
8 Find the positive root of 3x3 + 6x2 - 7x - 14 : 0 with an an accuracy of four decimal places.
accuracy of three decimal places. b) Find the same root as in (a) by the modified Newton's
Find the ncgative root of xa - 2x! - 3x2 - 2x - 4 :0 with
Method in which /'(xr) is used in all iterations, as described
9 in Example 2. Compare the speed of convergence of (b) with
an accuracy of two decimal places.
that of(a).
l0 Find the root of cos x - x 0, 0 < x <
= |, *itfr an a@uracy 15 Find the positive roots of l2xs + l2xa -23x3 + lOx +
of two decimal places. l0:0with an accuracy of two decimal places.
ll Suppose we wish to find the positive square root of a l6 Find the positive roots of l0x5 + 20xa - 33x3 - 66x2 +
number c. We take any rea$onable guess and call it dr. We 27x + 54 = 0 with an accuracy of two decimal places.

10
REI.ATED MTES

In Chapter 2, Section 5, we discussed motion along a straight line and

developed the ideas of velocity and speed. We learned that if a particle moves
in a straight line so that the distance traveled, s, depends on the time t,
according to some law s:,f(r), the velocity is obtained by tnding the
derivative /'(t). We can also write

f '(t') : ds
dt'
The velocity may be thought of as the rate af change of distance with respect
to time. If there are several particles, each moving in a straight line according
to some law; then we can talk about the rate of change of each of the particles.
Suppose the motion of these particles is related in some way. (One may go up
as the other goes down, as in a lever, for example.) Then we say that we have a
problem in related rates.
We are not limited to particles moving in a straight line. If a tank is being
filled with water, the level of the surface is rising with time. We talk about the
rate of change of the depth of the water. If the depth is denoted by Jr, then
dhldt is the rate of change of the depth. Similarly, the volume lz is increasing;
dYldt measures the rate of this increase. Any quantity which grows or
diminishes with time is a candidate for a problem in related rates.
In the problems we shall consider, it is important to remember that every
 Appr_lcenoNs oF THE DERtvATIvE

quantity is a function of time. Therefore \rye

can take derivatives of each
quantity wirh respect to /, the time. This derivative
""iJ;l;;d rt;;;;.
i,
We first give two examples and then map
out gerreJ-rules
EXAMPLE 1 One airplane flew^over an airport at the rate
"f;;;;;;;:-
of 300 km/hr. Ten minutes later
another airprane flew over the airport at 240
km/hr. nthe first airplane was
flying west i:9 jh: second o.ying-sortr, (borh
at the same artitude), determine
the rate at.which they were separating i0
*inut", ur* L. second prane flew
over rhe airport. (we assume-the airfraner ur;;;;;i*g
at .onstulird.)
solution Draw a diagram, as shown in Fig. 42. Let s be the distance between the
50+br o: A;nr^*
o:Airporr :lT::": ::lT y be l,!"1!:
at time r, and let 1,il
;;i;;; ii,
distance the westbound
the distance the southbouna uirfirr. has
truu"ra
travered at
] lT: ,when
y" know ihat dxldt:i 1.7*in, dyldt:4 km/min. We wish to find
\ N I drldt r:20 min, measured from the time the J;;,i",ffi. ilj::
lo,
\l
over O. We have the relation

s2 : x2 I Y2'
aJ rake differentiars;
ISouth :
One airplane flying west and the
2s ds 2x dx * 2y dy.
Divide by dr:
other flying south
FtcuRr 42 "ds _ _dx ,
dy
'dt-odt-Ydt'
When r : 20, the second airplane has traveled (20)(4)km
and /: g0. The
:.:,,H:?il1.,Ii,ii:ix:"1,,#i?:1ffi1i,:1iHiJffi l..1r:'$#;;;
150.5+80.4 ds 107
E: n kmTmin.
<, Il
"o.f-
we begin the solution of such a problem by using
a letter to denote each
of the quantities which change. tn itre ubou.'e*ffili
each of these is a function of r. Some relation
*" uu. r, y, and s;
is ro'ura among the letters
(s2:x2 *.y2,.as we.saw), and differentiars by irpri"i,
methods are used
to get derivatives with respect to t. Then the numer'ca!
value for each quan_
tity is substituted to obtain the answer. Let us work
another example.
EXAMPLE 2 water is flowing at the rate of 5 cubic meters/min into a
tank (Fig. 43) in the
form oi a cone of altitude 20 meters and base ruOl"r-f
O meters and with its

] ',,
,L r&-
I lij.
li
)-'>|--'
Water ffowing. inio tank is at
height Ir

FIGURE 43
 171

sEcTloN d.10 REI-ATED RATES

vertex in the downward direction. How fast is the water level rising u'hen thc
water is 8 meters deeP?

Solution Let h be the depth, r the radius of the surface, and V the volume of the water
:
at an arbitrary time r. We wish to find dhl tlt. W e know dV I dt 5. The volume
of water is given bY
y:\nr2h,
where all quantities depend on f. By sirnilar triangles,
r10 or ,:ih.
h20
and so
,::"T:|",s
We take differentials:
7y : tanh2 dh,
and, Cividingby dt:
dv 1TthL.,dh
_
dr4dt
We want tci find tlh,'dr when i:8. and since dVldt':S, we obtain
dh5 u
dt l6n

We apply these rules in the following example.

EXAMPLE 3 An airplane at an altitude of 3000 meters, flying horizontally at 300 km/hr,

passes directly over an observer. Find the rate at which it is approaching the
observer when it is 5000 meters away.
 CHAPTER APPTIcATtoNs or THE DERIVATIVE

Solution We draw Fig. 44. The airplane is at point ,,{ flying

toward point B. The
distance oB is labered as 3000 (step i). Note that this
distance does not
change in the course of the problem. Lei x be the
distance of the airprane ,{
from the point B directly over the observer, o, ands be
the distance from the
airplane to the observ"i: W. want to find dsldt when s:
SOOO. W. ,re given
dxldt: -300 km7/hr. The negative sign is used becarrse x is decreasing. A
('T)jly)
Observer, O
change of units gives us dxldt
' : - 3600 m/sec : - 83$ m/sec.
We have (Step 2)
Observer sees airplane at ,4
fIGURE {4 s2:x2+(3000)2,
and, applying differentials (Step 3),

2s ds =2x ilx.
Dividing by dt, we get

ds dx
s-- : x--:-
dt dt
We now insert the values (Step 4), noting that x:4000 when s: 5000:

t*f :4000(-83+) e fr: -u;im/sec: -240km/hr. tr

1O PROBTEMS

I Water is flowing into a vertical cylindrical tank of radius

7 At a certain instant a small balloon is released (from ground
2 meters at the rat6 of 8 cubic meters per minute. How fast
is the water level rising? l"*ll t a point 75 m away from an observer (on ground
level). If the balloon goes straight up at a rate
of dm/sec,
2 A launch whose deck is 7 meters below the level of a wharf how rapidly will it be receding from the observer 30sec
is being pulled toward the wharf by a rope attached to a later?
ring on the deck. If a winch pulls in the rope at the rate of
l5 m/min, how fast is the launch moving _t.hrough the water
8 A street light is 5 meters above a sidewalk. A man 2 meters
tall walks away from the point under the light at the
when there are 25 m oi rope out? rate of
2 m/sec.How fast is his shadow lengtheninl wle" tre is im
3 Two automobiles start from a point z4 at the same time. away from the point under the light?
One travels west at 60 kmlhr and the other travels north at
35 fmlhri How fast is the distance between them increasing 9 A balloon is being inflated at the rate of l5 m37min.
At what
3 hr later? rate is the diameter increasing after 5 min? Assume
that the
diameter is zero at time zero.
4 At noon of a certain day, ship ,,{ is 60 km due north of
ship a. If ,4 sails east at lZ imlhr and g sails north at l0 A baseball diamond is 90 ft on a side. (It is really a square.)
9 km/hr, determine howrapidly the distance between theni A man runs from first base to second base at 25 ft/sec. Ai
is changing 2 hr later. Is it increasing or decreasing? what rate is his distance from third base decreasing when
he
is 30 ft from first base? At what rate is his distance from
S l-t u given instant the legs ofa right triangle are 16 cm and
t, crn, respeetively. The first leg decreases at cm/min and home plate increasing at the same instant?
{
the second increases at 2 cm/min. At what rite is the area l1 A woman starts walking eastward at 2 mlsec frorq a point,,{.
increasing after 2 min? Ten minutes later a man starts walking west at the rate
of
6 A rhombus with an acute angle of 45. has sides which are 2 m/sec from a point g, 3000 meters ,orih of
e. How fast are
growing at the rate of 2 cm/min, while always retaining they separating 10 min after the man starts?
the
same shape. How fast is the area changing when the
sides 12 A point moves along rhe curve J,,:
are 10 cm?
that dxldt:4. Find dyldt when x='3.
viTJl in such a way
 173
sEcTtoN 4.10 REI.ATED RATTS

13 A point moves along the upper half of the curve y2 :

*| 2x there are 65 m of rope out? (Assume that the rope forms a
in such a way that dxldl: !/r. + | Find dyldt when x:4' straight line from deck to anchor.)
14 A point moves along the curve / = sin2 3x in such a way 26 A man lifts a bucket of cement to a scaffold 30 m above his
that dxldt: 3. Find dyldt when x: nl4. head by means of a rope which passes over a pulley on the
scaffold. The rope is 60 m long. If he keeps his end of the
15 The variables x, y, and z are all functions of t and satisf-v the rope horizontal and walks away from beneath the pulley at
relation x3 -2xy * y2 +2xz -Zxzz + t=0. Find dzldt 4 m/sec, how fast is the bucket rising when he is 22j m
when x = l, y:2,if dxldt: 3 and dyldt= 4 for all times r.
away?
t5 The variables x and y are functions of t and satisfy the 27 Water isflowing into a tank in the form of a hemisphere
relation sin2x- 2tan3y+l:0. fina! when x:n/4 of radius l0 m with flat side up (Fig. 45) at the rate o[
4 m3/min. At any instant let h denote the depth of the water,
aod dxldt = 2. r the radius of the surface, and V the volume of the water.
t7 A ladder 5 meters long leans against a vertical wall of a Assuming dll : nr2 ilh,find how fast the water level is risrng
house. If the bottom of the ladder is pulled horizontally whenfi=5m.
away from the house at 4 m/sec, how fast is the top of the
ladder sliding down when the bottom is 3 meters from the
r- _r0_--1
wall?
18 A light is on the ground 40 m from a building. A man 2 m
tall walks from the light toward the building at 2 m/sec.
How rapidly is his shadow on the building growing shorter
when he is 20 m from the building?
IIGURE 45
19 A trough is m long and its ends are isosceles triangles
10
with altitude 2 m and base 2 m, their vertices being at the
28 A bridge is lom above a canal. A motorboat going 3 m/sec
bottom. If water is let into the trough at the rate of passes under the center of the bridge at the same instant
3 m3/min, how fast is the water level rising when it is I m
that a woriran walking 2 m/sec reaches that point. How
deep?
rapidly are they separating 3 sec later?
20 A tror.lgh 10 m long has as its ends isosceles trapezoids,
29 If in Problem 28 the woman reaches the center of the bridge
altitude 2 m, lower base 2 m, upper base 3 m. If water is let
5 sec before the boat passes under it, find the rate at which
in at the rate of 3 m3/min, how fast is the water level rising
the distance between them is changing 4 sec after the woman
when the water is I m deep?
crosses.
2l A swimming pool is 9 m wide, 15 m long, I m deep at the
30 When a gas expands or contracts adiabatically, it obeys the
shallow end, and 3 m deep at the deep end, the bottom
law ptsv: K where p is the pressure, u the volume, and y and
being an inclined plane. If water is pumped into the pool at
K are constants. At a certain instant a container of air is
the rate of 3 m3/min, how fast is the water level rising when
under pressure of 40 gram/cm2, the volume is 32 cm3, and
it is 2 m deep at the deep end? 5 cm3 per second.
the volume is increasing at the rate of
22 Sand is issuing from a spout at the rate of 3 m3/min and Assuming that 7 = 1.4, find how rapidly the pressure is
falling on a conical pile whose diameter at the base is decreasing at that instant.
always three times the altitude. At what rate is the aliitude
increasing when the altitude is 4 m?
3l A light is at the top of a pole which is h meters high, A ball is
dropped from a height jtr at a point which is at a horizontal
23 Water is leaking out of a conical tank (vertex down) at the distance d meters from the pole. Assume that the ball falls
rate of 0.5 m37min. The tank is 30 m across at the top and :
according to the law s gt2, where I is the time in seconds, s
10 m deep. If the water level is rising at the rate of 1] m/min, is the distance in meters, and g is a constant. Find how fast
at what rate is water being poured into the tank frorn the the tip of the shadow of the ball is moving along the ground
top? f, seconds after it is dropped.
24 In Example 2 on page 170, find the rate at which the 32 Ohm's law for a certain electrical circuit states that E: IR
uncovered surface of the conical tank is decreasing at the where E is the voltage in volts, I the current in amperes, and
instant rn question. R the resistance in ohms. If the circuit heats up and the
25 A boat is anchored in such a way that its deck is 25 m above voltage is kept constant, the resistance increases at the rate
the level of the anchor. If the boat drifts directly away from of 0.5 ohms per second. Find the rate at which the current
the point above the anchor at the rate of 5 m/min, how fast decreases when I:2 amps and E is kept constant at
does the anchor rope slip over the edge of the deck when t0 volts.
174
CHAPTEN 4
ApprrcenorusGffiilFil

11

ANilDERIVATIVES

the.probrem of nnding
H;,,,::ilX",n,;:T.,* the derivative of a function.
,ri*;;;;;ilT;:,,3i,:ffi f T,3,,,lf.porv-no.i"i,;;;;iiun.tioit,
gi ven a ru on / a *" r.J uiu n.
n ct i

That is, *. *"ni to nd

un
# I rlfn, i;;,";:::i I ::lTru:l;i
u
n
:,: :fi
fi
;#' ('r) : f (*).,r-.._".pl, r),
ruroor"
'ifri:?,i:.;['#;'*1T:::;1'ji::i*:',i-t;ffi frlil'i..'i,,*
runctions fxa + 8,
**o +i,
unilre..In Chapters jand g ;l:
answer to the problem
weposed ir not ^ii{.;;';JT#1XL;X.:tfiiil"}j
devetop systemaric methods
i;; #;; ffierivatives.
we will

DEFINITION n
C"*

;|,;1::3?:h:lX[l*.nndingtheantiderivativeis,,o,""o,oil*-
EXAMPI-I 1 Given the function
f (x) : 4x3 + 2x_ 5, find its antiderivative.
solution To find u'-1"'10:''y3rive,
we simpfy
xa:2x is the derivative;i;1"_] guess. fe see ,n", or, is rhe derivative of
anriderivarive F(x) is given " i, ir," a"*rrr*:f:"rr. Hence, a possibte
ty
* x2 - 5x t 7,
which we verify o,
5x * 4; stru another is
*-rr,ltJ'rl,*
- F(x): J]'l:r"'n?1."'tid::*tive is F(x) = xa + x2
no si ngre sor u tion to ti,.
given. That.is, the probreri
p'our"t'ii*;iff ,:r'$iff ::il:liJ:,,i
solution. It is easy to ,"" ilrui
,ririirr the antideriuatiue do
n", haue a unique
"ny
function .F of the ,orf
is a sorution::.'::
*,"-r:*;;X;,:::;r"r .

;::[:iJ [:: :.il IT: ?l'f:,,ffi

,; ;;; y, ; il ;#ilwhichF,,we
are gi,ei iv e
may have anv

rn rhe above iri,".",i*, jilffi,: ff#Hll{ii::33;,r," q,"i,jiyl.

F(2):4:16+4-10+c,
and conclude ihat c _ ._6.
The function f. is then

,u,.'n"
process o*"'o"oll1;;: l;, - r*arized into tr,. roro*inrg

-W.,n"ff
*" f" Clr"pter 5 that all solutions
are of this form.
 175
sEcTtoN 4.11 ANtlogttvlnvus

POI.YNOMIAI. ANTIDERIVATIVE RUIE Ict p(x) be the polynomial

p(x): arx'* ar-txn-l + .'' + olx + cc.
Then the antideriuatiue of p(x), d.enoted P(x), is

P(x r: o,fi * an-t; . ,,* * aox + c,

where c is a number which may haue any ualue.

-The validity of the Polynomial Antiderivative Rule is easily checked by

differentiating P(x) and noting that P'(x) - p(x). An important speciai case is
P(x) : x'. We have

We also observe that if r is any rational numtreq the derivative of x" is

1.
rx'- Hence the Polynomial Antiderivative Rule also works when x is raised
to any rational exponent, positive or negative.

TXAMPI.E 2 Find an antiderivative of

f (x) : xttt - -t2l3 + sin x.

Solution An antiderivative of x1l? is $xa[; an antiderivative of -x2l3 is -!xsl3: and

we easily guess that an antiderivative of sin x is - cos x. Adding these anti-
derivatives, we get F(x), an antiderivative of f:
F(x) : $xst7 -]*stt - cos x.
This F(x) is, however, only one antiderivative of f . Since the derivative of a
constant is zero, we can always add a constant to F(x) to get another
antiderivative. To indicate the freedom in the choice of a constant we write
the antiderivative of/ in the form
$x8/? - trslr : cos .x + c,

where c is an arbitrary constant. ft

Remark. In Chapter 3 we developed systematic methods for fincling the
derivatives of algebraic functions. So far, we can oniy find antiderivatives by
guessing at the result and then verifying the correctness of the guess by
differentiating the answer
Let us work another example by the guessing method.

EXAMPLE 3 Suppose that F is a function whose derivative is

*r'(x* l)- tt2 +3x2(x+ l)r/2.

Given that F(3):5, find the function F.
176
CHAPTTR 4 APPUCATIONS OT THT OERTVATIVT

Solution We observe that.ilu(x) = x3, then u,(x):3x2, and if u(x): (x + l)r/2, then
u'(x):l1v* l)-rtz. Guessing that the expression for r,1xy is in ihe-forrn
tto'+u'o, we observe that F(x):u(x).u(x). Therefore
F(x): x311 + x)tt2 + c.

Setting f (3):5, we find

5:27'r/++c + c:-4?.
We conclude that

f(x):x311+ x)'t2 --4g. u

The next example employs the definitions of velocity.4nd acceleration
as
given in Chapter 2, Section 5.

EXAMPIE e A ball is thrown upward, and we know that the acceleration is

- 10(m/sec)/sec. (a) What is the velocity at any instant of time if the ball was
launched with a velocity of 60 m/sec? (b) Find it. t right of the
ball above the
launching point at any time t

solution since the acceleration a is given, finding the velocity u is a probtem in

antiderivatives. We write a(r) : u,{t) and, using the polynomial Rule, we have
a= -10, o: -l0t + c,
r:0 when the balr is thrown). we
where , is the time (starting at
when r:0, which means that 60: - r0'0 + c, or c : oti. we have u: ,i.
60

formula "uoi,
u:-l0r+d0.
To get the distance traveled, we again have a problem in antiderivatives.
We
find that u: s'(f) and again using the polynomial Rule, we obtain
s: -5/2 +60t+k.
To evaluate k, we note that at time r:0 the distance traveled is also
zero.
This makes Ic:0, and we have the formula

s: _5r2 + 60r. D
_ Suppose F(x) is the antiderivative of
/(x), and we write y : F(x). Then
the equation F'(x):f(x) can be written in the forrn

dy
: {(x). (1)
dx
Here /(x) is a known function and we are seeking y as a function
of x. An
equation with derivatives, such as (r) above, is cailed a differential
equation.
we provide a'systematic study of differential equations in chapter lg.
Typically, in solving a differential equation such as (r), i.e., in finding
the
antidedvatr.ie, there is usua\\y an additional condition'imposed ,n order to
determine the constant c which arises. we call this added condition an initial
condition for y: y(x). The next example describes the method.
 177
stcTtoN 4.11 ANtrDEntvATrvEs

EXAMPTE 5 Solve the differential equation

{'' :6" t 3't - 5

dx
. with the initial condition y:4 when x :0.
Solution Since 6x2 * 3x - 5 is a polypomial, rve otltain thc antiderivative

^ 3.xl
rr:7x''+--.-<tlr.
.12

wherecisanyccnstant.The initialcon itionyielr.!-s4==U i ii+ 0+ r atirJthe

solution is
3x2
-t,: 2x3 -l- *{ - 5,x +,1 i:

"11 PROBLEMS

In each of Problems 1 through 16, find the antidcrivative F. Use In Problems I 7 tlrrcrugh 22, find the distarice s traveled along a
the additional condition to determine the solution uniquely. straight line by a particle if thr velocity , : D(r) follows the law
I /(x) = xa - 2x3 * 7.x - 6; at rJ,: , given. The additional fact required for the result is also given.

2 f (x) = -2xs + 7xt - Sxz + 2x; [-(21 : I 17 t:t2 -t.t-2anci-s:2whr.n t:-l

3,f(r):(x+2)7; r(-2) : 1
lll 1;: -1: +- 2t -t- I anrl .s:0 when r:2
4 + 2.x(x+ 1)3; r(3i: '1 19 1,: S - 2t .- }t: a1d .s:0 when r:0
f(x\=3x2(x+ 1)2
5 : 2(2x +'1131x + 2\ + 6\2x + 1)2(x 1 2)?; 20 1.: 5 *7rr and s= 5 uhcrr t:0
,f(*)
r(-+):0 2l r- cos2 2t siri 2: ir,i.J s -= I when t :0
22 r: =, .-.sin lr and .s.-. I u,lren t:0
7 f {x): xtts + 6x2t3 + x2; f (0) : 3 2-l A ball i:: thrown upu nid from the ground rvith a launching
8 11x; = x3'4 + .r3 + cos .r; F\9) : Z velocity of E0 nr/scc. [ts acceleration is constant and equal
to - l0(m/sec.liscc. Florv long does the ball continue to
g
f $):xria + sec
^-
tan x: .(;) = , rise'f I'iow long does the balt stay in the air?
24 A wonran Crivirig an automobile in a straight line at
lo/(.x):f *::u'-t tt F(r):l a speed
appiies the brakcs at a certain instant {which we
<lf 80 rirlsec
Y' take to be r i- 0). If the brakes frirnish a constant acceler-
ation of *20(m;;sec)/sec iactually a deceleration), how far
II /(r) :Lx cosr-x2sin-x; t(;) =, v,ill shc go be[ore she stops?

l.x; t(1) : In each oi' Problems 25 through i0, soivc the diflerential
12 f (x): sec 2x * 3r! sec 2r tan ,
equatiorr with thc givcrr irritial condition.

^;
13l(.x):cos2xi .(;):* 25
*-:
: 3x) - 5r r 2; .r'i0) = 6
r/.x

14 f (x): -sin 4.x: t(;) :, 26 {r :.ir- - (ir .r J: rr(}} .= 7

d.\

*_ dy
,,
15/(x):sin2xcosx; r(;): * 7;: 6 sin r'. j'(0) - .r

16 /(x) : sec2 2x; .(i) : t 284l -i4rr- jr,,7: ,,( ji-.rl

.l'.
CHAPTTR 4 APPttcATloNs oF THE DtRrvArtvr

dv 34 Find a solution y: y(x) of the differential equation

*:27x"
dx -3x" +4x+ 6; y(-l\=2
d2vr _1- 1
ilv
+:4cosx-3sinx:
dx l@):n dx2

+31
Suppo-se that /'(x):0 for a<x<b. Show that /(x) is 35 Find the solution y: y(x) of the differential equation
constant. (Hint: Use the Mean Value Theorem.)
*32 il2 v
If Fr(x) and Fr(x) are two solutions of the differential
equation *z:6x-3
y(0):6
ltd;:J $)' subject to the conditions
36 Find the solution y:
and ]l'(0):7.
y(x) of the differential equation
show that Fr(x) - F2(x) is constant..(Ilinr.' Use the result dry
of Problem 31.) :2 sin 3x * 4 cos 2x
*33 Let .F(x) be a solution of the differential equation
iF
subject to the conditions y(0) = 3 and y'(0) = 5.
dv
dx:I{xl 37 Show that -v(x) : sin x + cos x is a solution of

where :
F(0) a. lf /(x) is continuous show that the d2v
solution is unique. (Ilfnt: Use the result of Problem 32.) -*
elx'
+ v:0.

CHAPTER 4
REVIEW PROBTTMS

In Problems I through t0 find the relative mirxima and minima In Problems 21 through 30 find the absolute maximum and
and sketch the graph. absolute minimum values of the function on the interval given"
I f(x)- --x2 +2x+3 2 l'$):2x2 +x-5 2l f(x):x3+2x, -1<x<4
3 ,f(*):x3+4x2-2x a f(x\:x3-3x2 +2x*4 22 f(x\: *Zxr + x2, *2<x <3
5 :
"f(r) 3xa - 8x3 - 6x2 *
24x * 2 23 -f(rl -- x4 _ 2x2, -2< x S2
6 f(*\:xo-14x2+24x*4 2a f{x) : x) +2xz -3x.- 4, -1 <x<4
7 ft.,rt:sin2x- _ n<x<n , (
zs /(x): { -2x+5. -2<x< -1
8 /(x):cos3.{. -tt<x<ft [ .*'+3x+9. -l<x<4
9,ftr) :#
-2 1
l0 /(x)
-l
^r '=
= Y- 26 frxr:l
x2-3x+1. -4<x<l
+ /
l-x'+2x-2, l<x<3
In Problems I I through 20 find the relative maxima and
minima and the points of inflection. Find the intervals of 2i f(x\_ [:x + 5. --2< r < - 1

concavity both upward and downward.

ixr-3x*1, -l<x<2
ll f(x):-3x2+5 l2 f(x):.tr+4x o<x<nl2
28 /(x)- fsinx'
13 /(x) :2x3 -x2+4 l4 f(*):xi+4.x2-3,x-1 )lLz-.o.*, !.r.n
:
15 J'(x) xa - 14x2 *24x +4
29 fix): 3xa - l6x3 - l2x2 +96x+ 2, -3 <x <4
16 f(xl : 3xa -4x3 - 18x2 * 36x * 5
1,
17 f(*l:2sin 3x, -0<x<22 30 t(x): ,'o=, -l<xSZ
x'+ l
It /{-r) : tan 2,x, -7t < x <n 3l The product of two numbers is 51. Find the numbers if the
19 -f(rl:- 2o {(xt:5i sum of three times one number and the other is as large as
possible. Find them if the same sum is as small as possible.
 1il
CHAPTER 4 Rwrrw PnoBLEMS

32 A rectangular box with square base has the top and bottom rr-I c
made of one material while the sides are made of a different ^i'l- ,-t,- ix,;-.
material which costs twice as much. The volume is to be
100 cm3. Find the dimensions so that the cost is as small 45 A man at point I is walking east at 2 m/sec. A second man
as possible. who is 60 meters northeast of ,4 starts walking north at the
same instant at the rate of 3 m/sec. F.Iow fast are they
33 Find the point on the curve ! = x2 - 2x which is closest to
separating I minute later?
the point (1, -3).
3 A window is designed in the shape of a rectangle surmoun-
46 Two biildings, one 20 m tall and the other zt0 m tall, are
20 m apart. A ladder stretches from the roof of one to the
ted by an equilateral triangle. If the perimeter is 300 cm,
roof of the other. At a certain instant a man beginning at the
find the dimensions of the window which provides the most
smaller building climbs the ladder at the rate of 2 m/sec.
light.
Starting at the same time, a woman runs along the ground
In Problems 35 through 38 calculate the differential. from the smaller to the larger building at the rate of 3 m/sec.
w2 t)* _ 7, How fast is the distance between the two changing after
3s =7;;T;
/(x) t6 f(x) =(x2 * ztul -2x1t4 4 ssconds?

37 f (xl= sec2 2x 38 /(x) :2cotjx + sin 4x In each of Problems 47 through 5l find y: y(x) from the
39 Find dyldx grven that x2+22-2xz-4:O and information given.
y3+2y22+224-3:0.
In Problems 40 through 42 use Newton,s Method to find the * y(ot: -2.
*:2-3x+x2,
value accurately to three decimal places.

4s Xu $ trr:i 42 1q Of
*=x5+Zx3t2+1, t11)=2
43 To find the cube root of a number c, show that Newton's
Method leads to the formula * 2x + cos .x, y(nl4): t
*:sin
/2\ c
xr*r:(3,/xr+tI. tO.
*,:2 sin x cos x, y(0):0
44 To find the nth root of a number c, show that Newton's
Method leads to the formula
" *:cos
3x* sec22x, ,(il:,