Lrwrs

Tnr DERtvATTvE
AppucATroNs

l-e begin the study of calculus wirh the process known as

differentiation. To establish this procedure
we develop the
important idea of a limit, first in an intuitive
and later in a
precise way. We also provide both a geometric
and a
physical interpretation of differentiatio"n.

IIMITS ONTUITIVE)

The single most important idea in carcurus is that

of limit. The limit concept is
at the foundation of almost a[ of mathemaiicar
analysis, ura"rstand-
ing of it is absorutery essentiar. The reader "nJun
shourd ,oi b"'rurprir.Jio find the
discussion rather complex and perhaps
JimcUt. ff,i, .f,orfa no, pror.
discouraging, however,lor once a precise
understanding or trre limit concept
is achieved, the reward is a good g.*p
or uit the basic processes of calculus.
We begin with an intuitive discussion of limit
' description to section.6. [t may-seem strange that it
;J ;;;;;;;; a formal
is possible to work with
and applv an idea without aifining it precrsety,
concept evolved in just this way. tn ttre_ earty
;;riil;;aiv"the timit
deuliopmen, or.urcJru, precise
statements in the modern sense were serdom
made. iet progres. *u, i.r"a
because there was stilr a degree of understanaing.
wt eo'i*pr..i. "ct ,o ii"*na
trouble, a later, more carefur approach overcame
the difficurty,-"*I , ao.,
solid basis for carculus was attained. The same process
continues
today in
mathematical research.
we start by considering a function and its graph.
/
particular value of x, s&] r: s. For .*a.pl",
we concentrate on a
let the function be

49
I
cHA?rgr 2 Lmns. Trc DBryATlvE. APPucAnoNs

).:,f(x): -2x2 *8x-4

and take the particular value x : 3. A graph of this function is shown in
4 Fig. interested in this function not only when x = 3 but also when x
1. We are
tuk.r on values in various intervals coniaiuing 3'
Suppose that we select the ittterval from x = I to x = 4' A graph of the
3

2 function in this regi:n shows that the highest value is attained at x whel :2 I
is 4, and the lowist occurs at -x = 4 when y is (Fig- 2).
-4 In other
I 'alue by the
words, the graph of the function lies in the rectangle bounded lines
x:1, x:iand l:4, !: to a smaller interval
---f-+ f,
-4.-The next step is select
il : :
about x 3. Suppose *" tuk. the interval fiom x 2Ito x 3iand draw - the
graph in this reii,on (Fig. 3). The graph of the. function is now in the rectangfe
:
Iounaed by the llns5 -r 2i, * = 3i, ! : -*,
y =3], as shown' Proceedilg
further, we take a still smiller interval about x = 3, Say' fror-n x
:29
to
x:3 .x: 3.1. We draw the graph (enlarged) in this region, as shown in Fig. 4. The
function values are now situated in the rectangle bounded by the lines x:2:9,
x : 3.1, ): 1.58, I: 2.38.
The main point we wish to emphasize concerns the height of these
EGURE 1

rectangles. As the widths of the rectangles become narrower' the heights also
shrinkln size. If we go further and take.an x-interval from 2.99 to 3.01, the
corresponding rectangle containing the graph ofthe function is bounded by
the lines x:2.99, n-3'01, y: t.gSqg, I:2'0398' A width of 0'02 unit
containing the value 3leads io"a rectangl" rf t only 0.08 unit. In addition
and narrower "ight
as the widths squeeze in tothe
to the heiglts becoming narrower
value x:3, *. obserie that the heights cluster about the value'y=?.Tlr9
reader may wonder what all this fuss is about, since the formula.fo_r- the
equation yields, by direct substitution, the result that /: 2 when x: 3' Note'
howerer, itrat throughout the discussion we never made use of this fact (Iiig. 5).
Indeed, we carefully avoided any consideration at all of what happens when x
is 3. We are conqerned solely wiitr ttre behavior of y when x is in some interval
about the value 3.
For functions we have studieC up to this time, the behavior of a function
at a point, say at x : 3, and its behavior in a sequence of shrinking intervals
about this poirt *.r" never distinguished. Now, however, a striking change
r!GURE 2
occurs as we start to study functions whose behavior cannot be discovered by
straight substitution. For example, the function
sln x
v: f(x) : x
-
is defined for all values of x except x:0. A straight substitution at x = 0
would tell us that
y:/(0) = $,
which is completgly meaningless. However, we shall see that by studying a
sequence of intervals about x = 0 which get smaller and smaller, we find,th,a!
the corresponding rectangles containing the funbtion above get thigngr,ar{
thinner, uod th" trligtrts cluster about a particular value of y. Nothing iSever'
stated about the value of l,when x is zero. We just study the value of y when x
gets closer and closer to zero.
Returning to the first example of the function ,f (x) = -2x2 * 8x - 4, we
HGLNT 3 see that ug co*". closer and closer to the value 3, I$) gets closer and closer
 51
sEcTrON 2.1 l-tmns (txrumvB

to the value 2. we say, "f (x) approaches 2 as x approaches 3." This sentence
is abbreviated even further by the statement

(x):2.
l,* f
Here once again we have the shorthand so typical of mathematics. A complex
idea which takes several paragraphs to describe, even in a simple .rri, i,
boiied down to a brief syrnbolic expression. Furthermore, thii, symbolic
expression contains a shorthand symbol for the notion of function, developed
previously in chapter l, Section 5. As we proceed further, more concepts will
be built on this symbolism.
If a function / is defined for values of x about the fixed number c, and if,
as -x tends toward a, the values of .f(x) get closer and closer to some specific
number L, then we write
lim /(x): l,
and we read it, "The limit of /(x) as x approaches a is 1." Geometrically, this
means that the series of
rectangles which surround c and which have
narrower and narrower widths become smaller and smaller in height and
z:2.9 cluster about the point (a,I.).
EIGURE 4 of course all the above statements containing terms sqch as .,closer,,,
"nearer," "narrower," etc., are quite imprecise and are intended to give'only
an intuitive idea of what occurs. As mentioned, a precise definition of the
expression lim**o f G): l, is given in Section 6.
Now let us take an example in which the result is quite different from that
of the first example. We define the function
(x-2)
y =J \x):
14*:fr,
which is well determined for all values of x except x:2. Lt x : 2 the function
is not defined, since straight substitution yields y :8, a meaningless ex-
FIGURE 5 pression. The graph of the function, shown ln nig. e, is quite simpie. If x is
larger than 2, then lx 2l: x 2 and the functiol has the value i
- - |. lf x is
less than 2, then lx
- 2l - -(x'- 2), and the function is equal to - +. we wish
to study the behavior of the function as x tends to 2. we select in interval
containing x=2, say from'x:1.4 to x=2.3. we see that the function
is contained in the rectangle bounded by the lines x:1.4, x=2.3,
y: ++ (Fig. 7). In fact, no matter how narrow the interval about r:2 l: _+,
be-
comes, the height of the rectangle will always be I unit. There is no limit
as x ctpproaches 2. We say that

FtctrRf 5
t::1.4 t:2.3
FIGUf,E 7
CHAPTER 2 llMrrs. THT DERIVATIVE AP?TICATIONS

- _1
lim ^ " does not exist.
,-zZlx - 2l

We now examine a number of examples of functions, with a view toward

discovering what happens in the neighborhood of a particular value when the
function is not defined at that value by straight substitution.

EXAMPTE 1 The.function
- 2x2 -x-3
f(x): . + I
is defined for all values of x except x = - 1, since at x = - I both numerator
and denominator vanish. Does

/t*)
'liT'
exist and, if so, what is its value?

Solution To get an idea of what is happening, we construct a table of values:

I$)l -3 l-1 ll
Then we draw the graph, which appears to be a straight line with a "hole" at
the point (- l, -5) (Fig.8). From the geometric discussion of rectangles
--+-->' given at the beginning of this section we see that

/t') - -5'
'1T'
However, we would like a more systematic method of obtaining limits,
without relying on pictorial reprasentation and intuition. By means of
factoring, we can write /(x) in the form
t)
f (x):tzx-lt4 .
1+l
Now if x*-1, we are allowed to divide both the numerator and the
rIGUf,t 8 denominator by (x * l). Then
f(t):2x-3, if x# -l'
This function tends to - 5 as x tends to - 1, since simple substitution now
works. We conclude that
_ _5.
,lim, "f(x)
Note that we never substituted the value x = - I in the original expression,
D

ExAMptE 2 Find the limit of the function

 5&
sEeTtoN 2.1 Lmns oNTUrflvD

x-4
flx):=#=,
3(Jx-2)
x*4,
. as x tends to 4. (This function is defined for nonnegaiive values ofx except for
x:4,'which is not in its domain)

Solurion Note that straight substitution of x : 4 fails, siirce /(4) r: 0/0, which is
meaningless. We could proceed graphically, as in Example 1, but instead we
introduce a useful algebraic trick. We rationalize the denominalor by multiply-
ing both numerator and denominator by \/ x + 2. So long as x # 4, we can
write
(x-4) (r/i+z) :
3($-2teG+2)
and furthermore the common factor can be canceled if x * 4. We get

/-.t1
f (x) : Yi;-i' if x *4'
J

The limit of this expression can be found by straight substitution of x : 4. We

find
[Z.tt n
lim/(x):v';'':;' tr
x+4JJ

ExAMpLE 3 Find the limit of the function

f(xr:ff.a,xr"-r.
as x tends to - 1. (This ftrnction is defined only for values of i larger than or
equal to -2 except for x: -1, which is not in its domain.)
Solution Since straight substitution fails, we employ the device of rationalizing the
numerator. Multiplication ef numerator and denominator Ay jZ + x + t
yields

t.rrz+x-rlt.r4+x+rl _ *,4 , x* -1,

.f(x)
= (x+ t)(/+x++ t)
(x-r l)1y'?+x -
tx7f11la
1x7{11T+*+ r)

where wi have used the fact that (c b)(a+ b) az bz with c

- : - :,fiTi,
b: l. We now divide numerator and denominator by x 1 and find that *
-f(x): a: x* -1, xZ..-Z.

As x tends to - 1. f (.

I
Qi
+l 2'
 CHAPT.EB 2 Uurs. Tm DERrvATrvE. ApPucATroNs

1 PROBTEMS

In Problems l thrbugh tr4, in each cilse find the limit (if any) 22 tim'/6t+h- $,
2t
approached by f(x) as x tends to a: Note all values ofx where i-o h
/(x) is not defined. r
x2-4 lim Jo +4n--
'r 9 -JD
I IG):7_sx+6, a:2 h-1
t[- zE
(x): x2 -9 x6 -4w6
25 lim _
2f
F_7,. n, a:3 li*V^-V'
xn2 x-2 r--4 x* 4
x3 +2x-l
1}"f(x): x+2 , a:3 ,. x3+4x2+x-6
Ilm- :
x--2 x'+6-x-8
af@):ffi,,
-2_ --)
a=2 !
r-tr-\6
ti

-
x3+l
.-r. x - jt
5 f(x.t:;_r a: I
* J3 +. -V-l_l_
x la
,-o J4 *x2 - J+-zx
* _"- "r, : :--:--- " i[ = l
Let l(x)=(1-lxl)/(l+x) be given with domain all
x + - 1. Plot the graph of ,f. Find
-,i'ji,x,ir:+ a:\r> -Z .. 1-lxl
t,fid-#, a: -l
,--r 1*x
30 Given
) f{xl:A'::? a:3,x> - 3
, :
(t
{
for -oo<x<3J
"f(x)
[2 for 3<x<co J
ro /(x) :14#;', a: -l Plot the graph. Show that lim,-1/(x) does not exist.
v_o 3l Given /(x) : (sin x)lx lor all x * 0 where x is measured in
radians. Sketch the graph of for x between / and I by -l
Jr-3 plotting the values of for f x:
-1, -0.9, -0.8, ...,0.9, l.
t2 f(x) = --
,a3 - ^-- 1
Can any conclusion be suggested about the value of

Jx',+7 -4 sin x

t3 /(.x) =
x3-6.12+llx-6
x3 -7x*6
a= l. Whaf is lim /(x)?
li;'
x1l 32 Given
What is lirn /(x)?
r-3
x3-5x2+2x+8 /(x) :3n I
t4 l lx): a: - I' what is lim /(x)? .I
Flr;r=x - 6' where x is measured in radians. Use a calculator to
In Problems l5 through 28, in each case find the indicated limit. compute /(0. I ), /! 0.0U. /(0.001 ), and /(C.000 1 ). Then fi ntj
t. lT g2+4x-6) ,r(-0.1), /( -0.01), /{ -0.m1), and /( -0.0001). Make ar,
estimate of
16 lim (2x4-3x2+7x+5)
li* t"n'.
17 rim
x2 - 16
l8
r-o X

,-a, X-4 J3 Sarne as Problem 32 for

(iD ri, Js+x-{5 - I -cos-\
/(x):---
x-O lX x
 55
sEcrroN 2.2 CoMrurAT0N or uM,Is

2
COMPUTATION OF ttMITS

In evaluating limits, functional notation plays a convenient and important

part. Consider the function

J'(x): x2 + 3'

and suppose that we wish to evaluate the limit

, f ts + x)
l':l*#'-/(5) x*o'
A straight substituti.n rt, _i,,r.i"o)ll,rn"o*""ningless expression

o -- :;j'
and we must penetrate a little deeper. \!e have f(5'l:28 and /(5+x)
: (5 + x)2 + 3 : xz * 10x * 28.'fherefore
/(5 + xl -"1 (5) _ x2 + 10x +28 -28 : x * 10.
The cancellation of x frorn numerator and denominator is most fortunate,
since now
lim (.x-r l0)
r-O
may be found by direct substitution; the answer ii 10.
The best way of learning the technique [6r evaluating such limits is by a
careful study of the following four examples.

EXAMPIE 1 Find the value of

.. f(4+h)-f(4)
lrm"_:_#, h*0,
,r-O n

where

f(x):^+117,
(x + ll-
x* -1.

Solution Direct substitution of ft :0 gives

/(4) -/(4) :g
0 0'
which is rneaningless. However,
l-.lt
[l4t: ,i, fg + h): (4 + ,, +lj, = (siir.
f HAPTER 2 LlMns. THt ImlvArwE. AmlcAnoNs

Therefore

:---h-' '
h
Wemighttrydirectsubstitutionatthispoint,butagainwewouldfail'since
theresultisagain$.Weproceedbyfindingthelowestcommondenominator,
getting

f (4
--i-
+ h) - f (4)
=
2s - (5 + hY
- 2sh(s :
- --
10h + h1)(
=
-
+-hf Ehc +tr ,swt * h1z'
-!110
*,
ll .

Solongas}r*0,thisfactormaybecanceledinbothnumeratorand
denominator. Then

I*i'-ffi rJ
rnay be obtained by direct substitution' The answer is -#'

EXAMPTE 2 Find the value of

lirn /(x),

where
I
.,' ,+, ::T, + f
f(x)=G:lir,x*3.
Sketchingthegraphofthisfunctionaboutx:3'weseethatitincreases
notion of limit as
without bound as x tends to 3 (Fig. 9). According to the
in
gi"., i, Section 1, we take an inteival of x values about 3 and thenissee
it clear
irt ui o.,ungle the function values are contained. From the figure
thatnomatterhownarrowwemaketherectangle,wecannotmakeithigh
No rcctangte can contain the graPh ;;;G ,-o .onoin the graph of the function. In such a case we say that
ncaf x = 3. lim /(x)
fiotnt 9

does not exist'

n

EXAMPLE s Find the limit

*Mrle
where /(x) = x2'
as before to find that
Solution Direct substitution of t = 0 fails, and we proceed
' f(2\:4 and f(2+t\:(2+r\2 :4+ 4t+t2'

Therefore, if , *0,
f t2+t)-f{i\ -4 -4+t
---T--'- f
-4+4t+t2 t
 57

sEcTtoN 2.1 Comnnrnox or uMrrs

However, this function increases without bound as t tends

to zero through
positive values, and it decreases without bound as t tends to zero through
TI
negative values (Fig. l0). The limit does not exist'

FIGUNE 10

Aconvenierrtsymbolusedthroughoutcalculustodesignateareal
number is-Ax, wfricfr is read "delta e*prtssion is to be treated as a
x.l'thir
h in Example I (an{ ngt I
singte qruniiry such as r in the above Example 3 or
A and x). Later we shall see ihat such symbols actually help
in
the prod,uct oi
undirstanding lather complicated limiting processes'

EXAMPTE a Find the limit

lim
f(-r :l!:-1), ax>0,
Ax-0

where /(x) :13.

solution We have /(-1) - -1,.f('t +Ax) =1-1+Ax)3 and

/(- 1 + Ax) -/(- 1) - I + 3(Ax) - 3(Ax)2 +(Ax)3 -(r 1I
J^,

: Jtx:t- 3(ax) +(ax)21.

and thc
As Ax tends to.zero, this expression tends to {tl-ttO)+(0)21'
answer is zero.

2 PN.OBLEMS

Evaluate thc following limits' , */(t+? {(1), .f(x) =2xt-2

/(4 + I) -,f(4) , f G)= 3rf - I
, ,,,n
r-O
h)-f(-3), IG): xz + 2 .
[J_1+ mre*{, f{*t=z;,xrro
CHAPTTN 2 [mm. fxr DE{ryATwL ApptrcluoNs

s t^:\tJ!+19 , f$): rx2 - l8 I'lt+h)-F(t)

3 lirn
,r- O
-- i F(x):x:--L,
x-
x#o

6:\tl:rpr!) , r,,,: -:,, ,*o l9 lim

i-.0
sk+h)-sk)
---,, -' .9(x) : .,r2 + 3-x: c

sl2+ h) * e(2)
,l,Itt:l::l, g(x+3h)-g(x)
s(x):12-x2 20 lim .9(x):3x2+2,t-l
,-o -,,-
r y*ry4:st4), s(r:,,i 2l lim f(x + h) -[{x)
n-o h
/(x) :-_r
I ?l
I

-,
x*-l

s y^F(-?1!-:-\-2, F(.):,+, x*l ,. /'(x -f g)

+ ht
/r,t:1,'
lT- r-' -Y"
x>o

ro ri- Igj - + -f (2) , I(x\:, +../6., .x > 0

ft) F(a\
. F(x): I-, x>0, a>0 23 lim f
(2 Lx)
r,-O h
vx Ar- O Ax

It li_ F(, + r,) - F("x)

F(x): -rz * 5 24 lim /(.rr + Ax) -/(x) , i(*):rfi
,r+O h Ar- O

12 Lln
e(b!t-_9(b)
G(x): 25
..
llm
G(x + Ax) - G(x)
. Gtx) -r- I
x*-i
I..ri k
x2 -2x + 3 ax-o A-t x+1.
n u*{('3._1ll-I{11, x3 + 2x + 5 251i^[Jls!-[Jr)
-__ y'(x): x3 - I
r-'o t .:(r): -'Oi h

i4
Ft4+hl-F'(41
-___:__
llrn trlx):71+ 1
27 lim f
(.2 - x+ I
- 3h) -l{2) , 16l:
h-1, h hjO ra2h

F(2+h|-Ft2\ f(a+h)-f(a) l$l: tl

h-{r V,h
F(x):1+ -2x+5, (t,>0) lim
,!-O h x-l x*1._l
x+l.-_---rr
.2r
U3+h)-U3\ r
lim
f(v+k)-ftv) I$\:+, t+ -2
lo ltlll
r.o -
------t-ri---L(x): x'+
, l
r (tt>0) 29
t-O
h"'' --3,
nz2+l
I / lrrn
/'(x+/r)-/{.r)
.f(x):xtt', x>o * r.^f(z+pJ-fQ)
p-o 3P
I@: 2, ' z*o
i..t) --. h

THT DERIVATIVE

DEFINITION If f is afunction, the derivative of the fumtion f , ilenoted by f' (read "f
prime" ), is defined by the fonrula

In this definition, x remains fixed, wh,ile & tends to zero" tf the limit does not
exist for a particular value of x, the function has no derivative for that value.
/
we also say that a functiom is difrerentiable e6 x if the above limit exists. A
/
function is simply said to be differentiabte if it has a derivative for all x in its
domain.
stcTtoN 2.3 THE DERTVATIVE

Looking back at the previous section, we see that many of the problems
consisted of finding the derivative at a particular value of x. We now give a
systematic procedurre for obtaining derivaiives, a method which we call
the three-step rule. The technique-is illustrated by some examples. In the
following sections we provide interpretations and applications of the
derivative.

EXAMPIE 1 Given ,f (.t) : x2. Finri the derivative. /'(x}. by the three-step rule'

We have ,f (x) : x2 and J-$ + h): (x * Il2 : *z + 2xh + h2. Therefore

f$: h) -f ix\ : x2 t 2xh + hz - x2
2xh + h2.

Since h is a l'actor ol this difference, we get

l'(x + h) -J'(x) 2xh + hz :2x + h.

hh

(r, * h):2*.
l,*
Answer: ,f
't,x) : 2x. U
In the three-step rule, the first two steps are purely mechanical and are
carried out in a routine manner. It is the third and last step that frequently
requires ingenuity and algebraic manipulation.

EXAMPTE 2 Gii,en /(x) - llx, x# 0. Find /'txi by the three-step ruie.

Soiution We have the following.

l1
Slepl: .f(.x+fi) -./t.):.(+h -x

lt
Steo 2:
ttt+h)--f(.) ;i'i r
hh
It is at this point that the difficulty arises. since tetting h tend lo Tero
:Beferre
yields $. applying the third step, we manipulate the expression by
finding the lowest common denominator. We obtain

U:n_/(!_:I_ llih) :_-l

h hx(x h) x(x + hl' *
CIIAPTER 2 UMns. THE DERIVATIVE. Am:clnoxs

' 3:
Steo lim ----1 : - ---.
I
r,-o x(x + h) x'
I
Answer: f'(x\:- x--. l

EXAMPI-E 3 Given /(x) : uG, lor x > 0. Find ,f'(x) by the three-step rule.

Solution In the three-step rule, it is not essential that the letter Ir be used in computing
the limit. Any symbol, such as t k, [x,and so forth, may be used. We employ
Ax in solving this problem.

Step l: /(x + Ax) -f (x): r/i-+tx - $

Step 2:
/(x + Ax) -,f(x) _J;+ Lx -,/C
Ax Ax

Now the trick of rationalizing the numerator is used. The result is

J;+^.-J; .J;+*+"'7_ _ (x+Ax)*x

Ax .l*+ t* + J* ax(yG+ a, * r/il
1

Now taking the limit, we get

Step 3:
il'i"ffi4- ,i;+J;- z.Ji'
-1
Answer: ft(x\: n
2J*
-=.
Some problems require quite a bit of algebraic manipulation, as the next
example shows.

EXAMPLE 4 Given ,f(x) : xlk t l). Find f'(x) by the three-step rule.

Solution We have

StePl: [(x+h\-,f(x) =#-,

Step 2:
f(x+h)-f(x) :tl x*h* I x+ I
 s1
sEcTtoN 2.3 THE DERIVATIVT

At this point we get the lowest common denominator in the numerator so

that we can combine the expressio" ,:::j." fraction. We find
(x + h)(F+"1) - x(-x-+ h + 1l

f (x + h) -,f (x) (x+ft+l)(x'-\-t)

We now multiply out the nunterator and write the above expression as one
fraction:

f(x + h) -f(x)
-
x2 + hx + x + ft - (x2 + l:-x +-t)
h h(x+h + 1)(x+ l)
All terms but one in the numerator cancel, giving
f(x+h)-f(x)
h h(x+h+ l)(x+ 1)

pividing through by h, we apply the next step, getting

ll :
SteP 3: I- * 1, * 1X, *
1, 11 [r-;17
I
. rf'(x\:
Answer: \--t (x*l)2.

3 PROBLEMS

In each of Problems I through 33, find the derivative by the

2t f (x): !f4 --)/ 22 flxl:-:
three-step rule.
,t'4 -
t f{x):fvz 2f (x\: -2x2 + 1 "
3 f(x):2x2-3x+4 4 f(x):3xl -2x +2
23f(x\:-"x2x'+x+z 2a f{xl:2x-*.3
=
5 f(x): -aa 6 f(x): x3 -2x
t 25 f (x): !F+ xn 26 f(xl: x5 + 2x -I
8 "f(x) : -;
1
7 f(x): .x- ,
I x' I
2x+3
27 f{x): (x+ tF 2S,f(x):x2+?*r*
x
s /(r):1x- I0 /(x) :
5X- I
I t
It /(x): -I
t2 f (x\: -;-.
zg f(xl:i rtl .f(-r):
irar;;
r -j; x-+l
13/(x):?- x2-l x2 -2x 3t f (x): 32 /(x) : lrl, x *o
,,*
t5 IG)=+ 16 f(x)
I
: --=_ 33 /(x):lx-51, x*5
vx ./x+3 34 We define /(x): [x] as follows: /{x} is the largest integer
r7 f{x):tr[* ls /(.r) : rj* +t that is less than or equal to x. For exampl€, -f{lt) :3,
-,G-l i tll : I , /(8 2) : 8. We catl / the 'greatest integer func-
te f (xl: -:: f (x): tion." Find the derivative of /(x) when x is not an intepr.
Jx-l
zo
t_ What happens if x is an integer?
 62
CHAPTER 2 Lluns. Txr DrRlvATtvE. AppucAnoNs

CToUTTnIC INTERPRETATIoN oF THE DTRIvATIVE

It is easy to define a mngent line to a circle at a point p on the circle. A radius

from the center o to P is drawn and then, at p, a line L perpendicular to this
radius is constructed (Fig. 11). The line L is called the tangent to the circie at
the point P. We next draw the curve given by the equation !:x2 * 1, as
shown in Fig. 12. How do we define a tangent line to this curve at a point p?
If the point P is (0, 1), the lowest point on the curve, it seems natural to
Tangent line to a circle at p take the line y: 1, as shown, as the tangent line. However, at a point such as
n6unt 11
Q, it is not evident what line is tangent to the curve. intuitively, the definition
should state that a tangent line should "touch" the curve at one point only.
But this "definition" is not good enough, since there may be r.r..ui such lines
and, in fact, a vertical line through e has only one point in common with the
a:r.2 +7
curve. while the definition of tangent line to a circle is quite elementary, the
definition for other kinds of curves requires the sophisticated notion of limit.
Let a curve be given by a function y:f (x\, as shown in Fig. 13. Draw a
line, denoted 11, through Po and p, two points on the curve. Any line through
two points on a curve is called a secant line. The rine L, is such a secant line.
-yo). This means that yo :f $o).
Suppose that the coordinates of Po are (xo,
a:1 The coordinates of P are taken to be (xo * h, f (xo* lr)), where l, i, so.,
positive or negative number. Figure I3 is drawn with ft positive. If p were to the
'+-*--+*_++-lt--r=-+-+* o left of Po, then lr would be negative. The slope of the line L, is obtained from the
I I
usual formula for slope when two points are known:

'fangeut line trr y: x2 + I at P slope of L, : f(xn+h) -,f(ro) _ f (xo + h) -f (xo)

TSGURE 12 (xo+ft)-xo h

As h tends to zero, we see geometrically that the point p slides along the curve
y:f(x) and tends to the point po. The secant line I,, rotates about po.
Intuitively it seems that Lh.approiches a limiting line as /r -* 0, and this
limiting line should be the tangent line to the curve at po. Figure 14 shows
several positions for L, and the limiting position as h --+ 0. On the other hand,

P (rn * h, | (16 + 1

Po{.ro, I (

Graph of the secant line L, to f Secant lines tending to a limit line as h

tends to O
FIGURE 13 HGUNT 14
SECTION 2.4 GEoMErRlc TNTEnpRETATIoN or rHt DtRtvATtvE

if the function / possesses a derivatire, then

/(.xo + h.t-l't*nl __7,txo).
,.n-,
h-0 lt

In other words. the slopes of the lines L, tend to o lintirittg slope which is the
deriuatiue of thefunction f at Po. Suppose that we now construct, at Pr, the
line passing through Po and having slopc l''(r,,). This can be dc.ne by the
point-slope formula for the str4ight line.

DEFINITION lhe tangent line to the curue with equation y:.i(.x) al l'o(,x0,.1 (xn)) is the line
through Po(xo, ,f (ru)) with slope /'(.rr).
We recall that the point-slope form for the equation of a straight line is

)'- )o : m(-x - x6)'

If in this formula we take yo :"f (xo) and m : f' \ro), we obtain tlre cquation of
the line tangent to the curve at the point Pn.
.: ..:i.: : . :":i'
"''":
ThC: equation of the tangent lilrb a[,(x,,,,f

[XAMPI-[ 1 Find the equation of the line which is tangent to the curve that has equation
y: x2 'l I at the point on thc curve where x : 2.

Solution In this problem f (x): x2 + 1, xo:2. Therefore yo:,f(-ro) :l'(l): J. To

find /'(xo), we apply the three-step rule to f (x): x2-.t- l.

Step l: J$+h) -,f(x):x2 * 2xh+h2 + 1-. x2 - i.

Step 2: {8 ! a:J(-! :"!# :2x * h.

Step3: Taking the limrt, we get ['(x):2x.

Then /'(x6) :f '(2): 4 and m: 4. The desired equation is
y-5:4lx-2)

DEFINITION The normalline to the curue with equation y :J'$) at Ps(xs, (xo)) is the line
"f
through Po u,hich is perpendicular to the tangent line at Po.

EXAMPLE 2 In the problem of Example 1, find the equation of the line normal to the curve
at the point where x : 3.

Solution Since xo:3 and.f(x) : x2 + l,we have yo:-f (xo): 10. The tangent line at
:
xo 3 has slope /'(xo) :f '(3). We have already found that f'(x) :2xfor aity
value of x. Therefore the tangent line at x : 3 has slope 6. The normal line
must have a slope which is the negative reciprocal of the slope of the tangent
line, namely f . The equation of the normal line at (3, l0) is
y - 10: -*(, - 3)
g
CHAPTEN 2 lffiIrs" THE I,aNATIVE APnic noNs

We have just seen that the derivative is merely the slope of the tangent
line. In fact, the tangent line is defined in this way. The intuitive notion of the
nature of a line tangent to a curle is of great help in plotting graphs. If the
derivative /'(x) is positive, the slope is positive, showing that the curve must
be rising as we go from left to right. Similarly, if f '(x) < 0, the curve must be
falling as we go from left to right. These statements will be proved rigorously
in Chapter 4, Section 3' The fact that /'(x) :0
means that the tangent line is
horizontal, and the cvtye may have either a low point or a high point at
such a location. (However, other possibilities exist. as we shall see later in
Chapter 4, Section 3.)

EXAMPTE 3 Find the intervals in which the function y : f (x) - x3 - 3x + 2 is increasing

and those in which it is decreasing. Sketch the graph.

Solution We first find the derivative by the three-step rule.

Step 1: f(x + h)-"f(x) :(x * h)3 - 3(.x - (x' - 3x + 2)

+ /r) + 2

: x3 * 3xzh + 3xh2 + h3 - 3x - 3h + 2
-x3 + 3x-2
:3xzh*3xh2 +h3 *3h.

f (x + h) -,f(x) :3x2
Step 2: -3+3xh+h2.

Step3: Taking the limit as h-r0, we get I'(x):3x2 -3. We first find the
places at which /'(x) : 0. That is, we solve the equation 3x2 - 3: 0 to get
x: 1, - l. An examination of the function f'\x) :3xz - 3: 3(x .. 1)(x + 1)
shows that /'(x) > 0 for x < - 1; also we see'that f '(x) < 0 for - 1 < x < 1
and /'(x) > 0 for x > 1. We conclude that

/(x) is increasing for x < - 1,

/(x) is decreasing for - I < x < l,

/(x) is increasing for x > 1'
+ z
A table of values gives the following points.

y:f (x)

The graph is sketched in Fig. 15. n

GraPh of 1,x) : x' - 3x + 2 The fact that the curve reached a low point at exactly (1,0) and a high
flGUIf 15 point at exactly (- 1, 4) was discovered from studying the derivative. The
location of high and low points could not easily be obtained merely from
plotting points and sketching the graph. A low point such as (1,0) is a
minimum point only if we restrict our attention to a small interval about this
point. Such a point is called a relative minimum. It is clearly not the minimum
point on the entire curve, since the curve falls below the x axis as we go to the
 65
sEcTtoN 2.4 Gromrrnic lNrERptETATtoN oF THE DEnrvATtvE

left beyond x: -2. Similarly, the point (- 1,4) is called a relative r-naximum.

then /'(x): Q.

Notation. We saw earlier that thy6ymbol Ax can be used instead of lr in

calculating the derivative of a nction /. Writing y:f (x), we define the
symbol Ay by the formula 7
y:f (x + Ax)
Av, -/(x).
Then Step 2 in the three-srep rule states that
Ay_/(x+Ax)-/(x)
Ax Ax
The derivative consists of taking the limit of the above expressitin as Ax -* 0.
The symbol we have been using is /'(x). However, it is helpful to use a symbol

related to the quantity It turns out to be most useful to write

#
r$):! ax
where, at this time, dy and ilx will not be defined separately. We simply use the
.dv
ratro as an alternate expression for
1; /'(x). That is, we write

4 PROBTEMS

In Problems I through I 2, find the equation of the tangent line )

and the normal line at the point corresponding to the given ll y:2x + 1, xo: I
x
value of xo.
12 y:xa-2x*5, xo: -l
l y:*'*x-1, xo:1
2 y:x2+3x, xo: -l In.Problems 13 through 20, find the intervals in which is /
increasing and those in which it is decreasing. Plot the graph of
3 y=x'-2x2+3, xo:2 y:f (x).Note relative maxima and minima.
'4 y:xt-x+2, xo:-2
13 /(x):x2-4x+5
5 y: -r'*3x* 1, xo:2
M flx):l+2x-x2
6- y: x*l ^, xo:3
x- z 15 f (x):-1x3 - !x2 -Zx
t y:,/*, xo:2 16 [(x):*x3- x2 -3x+2
E y:.tr-, xo:-l 17 f(x):l+3x2-x3

9
I
---7, l8 /(x):x3+3x-2
y -- xo:6
j*-2 19 f(x)':3r4 - 8x3 6x2 + 24x + 2
-
l0 y:2x+31G, xs-4 20 f(x) * 3x5 - 25x3 + 60x + 10
 cHAPrEt 2 UMns. THE ornivrrm i{p?uclnoNs

;In Problems 2l through 23, a function y:f(x) is given. In Problems 24 through 30, fnd dyldx and evaluate the
Compute Ay : f (x + L,x) -/(x) and Ly I L,x for ttre given ratoes derivative at the given Value ofx.
of x and Ax.
A y:xa+2x-1, 4:-2 x2-l
2l y=f(xl=x2-3x+4, x:1, Ax:0.1 25 t: ?T1' x= 1

zri y: tx, x: 8 I
Y:7+ l'
H, x:2, Ax: -0.2
?t2 y=f'(x)= 27 x=2
_t
8 y=Jx2+x+2, x:l 29
x3
Y=;Tr, x:2
23 y=f(x)=j-#, x=0, Ax=o.ol
30 y=x5+3. x=-1

lNsrnNrnNEous vEtocrry AND spEED: AccEr.ERATroN

If the needle of the speedometer in a car is pointing to 6e we say

that the car
is traveling at 60 kilometers per hour*. An analysis of this
statement requirEs
a definition of the phrase "60 km per hour." As a start, we.ighiruy
that if rhc
car were to continue in exactly the same manner, then in oie
hour it would
have traveled exactly 60 kilometers. But such a statement
avoids the problem
completely, sincq the needle pointing to 60 does so without
any knowledge o[
the way the car behaved some minuies ago, or how it
intends tl behave in tnc
future.
Before any further discussiorr of this question we introduce
two simprify.-
ing assumptions.
l) Motion will be assumed
i9 take prace arong a $raight line, arthougfi
the object in motion may go in either diriction. oni airectlin will arbitraritr-r
be selected as positive, and the orher direction will then
be negative.
2) The object in motion is idearized to be a poini oi p"ii.r". S,*clr
e
simplification is necessary, since a mathematicai treatment
of th" .rr"t
motion of every portion of a complicated object such as an
automobile wourd
defy analysis
Suppose that the path of a particre is along a horizontal
distance to the righ.t designated as positive unain"r torl"
rine ! with
(Fig' 16). we let , denote time in ro*e
l&'u, negatirc
unit such as seconds or
Negative Positive minutes, and we measure the distance s of"ooueoient fr; il;;;;t
the object o in .ni*
o
, such as centimeters, meters, or kilometers. If thi particle
io *otilo is at thc
point s, at time t, aLd at the point s, at
HCUff 15
time rr, th", it took r, _ r, units of
time to traver the distance sz-sr. we definJ'tn" riro"ity of the
particle over this time interval to be "oog"
sz-sr
t,-tr'
If s is measured in meters and f in seconds, the units
of this average verocity
would be denoted as meters per second (m/sec). If a
car travels 40 kilometers

*The metric system

will be used throughout.
 67
sEcTtoN 2.5 INSTAMANEoUS vBocITY AND SPIED: Acczuunox

in one hour, the average velocity of the car is 40 kilometers per hcur (km/hr).
But we still know very little about how fast the car is traveling at various
times during the hour. The needle of a speedometer pointing to 60 is an
instantaneous event, and we need a notion of velocity at a given instant of
tirne along the path. The development of this notion proceeds in the following
way.
We mark off two points along a straight stretch of road, accurately
measure the distance between them, and provide a timing device to indicate
the exact moment the car passEs each location. The average velocity is then
computed. We believe that if the distance between the two points is small, the
car's motion cannot change too radically as it travels from one point to the
other; we therefore obtain a good indication of the velocity of the car at any
time during this interval. For example, if the points are 100 meters apart ind
the average velocity is 60 km/hr, the variation in the motion along such a
short stretch of road is probably not too great. If we desire greater accuracy,
we can shorten the interval to ten meters, improve the measurement of the
distance, and increase the precision of the timing device. Now we again argue
that, in such a short distance, the motion of the car varies so slightly that the
average velocity closely approximates the velocity indicated on the speed-
ometer. On the other hand, some variation may still be possible, and so we
cut the interval again, this time to one meter, say. The qontinuation of this
process shows that instantaneous velocity is the result of computing the
average velocity over a sequence of smaller and smaller intcrvals.
To make the above discussion precise, we first note that the distance s
traveled along the straight line is a function of the time ,; that is, s:,f(f).
Then the ayerage velocity between the times r, and r, is simply expressed as

f (tr) -f (tr) _ distance

tz- tr time

DEfIMTION The instantaneous velocity at time ti is defined to be

we see that the instantaneous velocity is the limit of the average velocities as
the time oVer which these averages are taken tends to zero. since t2 must be
different from fr, we can write rr:t, * At, where Ar may be plositiue or
negatiae. The above expression for instantaneous velocity then becomes

lim I$, + Al\ -f (t,)

,t+Ll)tr tr+Lt-tl

we note that this is precisely the derivative of f(t) evaluated at r,. Inother
words,
 CHAPTER 2 Llnrs. THr orniva;rvr AppilcATtoNs

we also use rhe symbol to describe this derivative. The quantity dsldt may
-
f
be positive or negative, according to whether the particle is moving
along the
iine in the positive or negative direction. The speed of the particle iidefined to
beldsldtl==if'Qil rhe speed is merely the magnitude oithe velocity and is
always positive or zero. It may seem strange to introduce a special term
specifrcallv for the absolute value of the velocity. However, thi notion of
speed wiil be especially useful later when we study motion along curved
paths. T'he speed teils us how iast the particle is moving but
lives no
inlbrmation about its direction.

EXAMPLE 1 A projectile shot directly upward with a speed of 100 meters/sec moves
according to the law

s: t00r - 512,

where s is the height in meters above the starting point, and r is the
time in
seconds after it is thrown. Find the velocity of the piojectile
after 2 seconds. Is
it still rising or is it faliing? For.how many seconds does it continue to risel
Ilow high does the projectile go?

Solution we assume that the rnotion ofthe projectile (particle) is in a straight vertical
line' If we let /(r) = r00r 5t2, the verocity is given' by dsldt. Aiprying the
three-step ruie, we obtain
-
ds
dt
= 100- l0r.
Let p denote the velocity at any time r; then
+
ds
= 100- l0t.
I
i D--

tt f 500 rueters , u:0, t = 10 dt

Il t:2, then u : 100 - l0(2) : 80 misec. Since I is positive, the projeailc
I

is still rising. The velocity is zero when r00- l0r:0, or r: r0. when is
I

ti larger than 10, the velocity is negative and the motion isdownw&rd,
r
i.e.. the
tf
il
projectile is falling. When r: 10, we see that,f(lO): 100(10)_ 5(tOj2:566
meters, and this is the high6st point the projectile reaches. when
ll particlehasreturnedtoitsstartingpoini.(SeeFig.17.)
r: 20 tbe
rl
,=20 il, -f startinc point
In motion along a straight line, whenever a particle chifues direction
velocity goes from positive to negative or from negative to politive.
t!
Therefire
FtcuRt 17 at the instant the particle reuerses direction the uirocity must be zero.

EXAMPTE 2 A particlc moves along a horizontal rine (positive to the right) according
to
the law

s: /3 -- 3t2 - gt l- 5.

During which intervals of time is the particre moving to the right,

and during
which is it moving to the left?

Solution
I
It is m_oving to the right whenever the velocity is positive and to the left whee
the velocity is negative. 'Ihe velocity u is just ihe ierivative of
11rf : ;, _ ;gt
- 9, + 5. Applying the three-step rule, we have
 69
SECTION 2.5 lNsrANrANtous vEtoctry ANI) ipEED: Acctunertor.r

u:!:3t2
dI - 6t- 9: t(r + t)(r - 3).

If r : - l, 3, thc velocity is zero. By exarnining the signs of the factors r + I

and r - 3. we find that:
If r < - l, u is positive and the motion is to the right.
If - I < r < 3. u is negative and the motion is to the left.
7f t>3, u is positive and the motion is to rhe right.
Motion to the left and motion to the right are separatcd by points of zero
velocity, i.e., aft:
- 1,3 (Fig. l8). Although the morion is along the line L. its
schematic behavior is showd by the path above this line. Zcro vclocitv occurs
at points P, Q. The point p corresponds ro r - -- l. ' s :, 10. ; =, 0; 0
corresponds to r: 3, s: -22, u:0. I

_25 _20 _ 15. t() _,5 0 5 10 15

Showing the direction of n:otion i;f a particle moving
according to the law .s : t.l _ lr 2 _ 9r + 5
FIGURE 18

Acceleration is a measure o{ the chonge in uelocity. If a particlc nrovcs

along a straight line with constant velocity, the accereration is zero.
In an
autr:mobile race, the cars pass the starting position travering iit .
unirorm
velocity-say 30 km/hr. within ten seconds one of the cars is traverin_e
at
120 km/hr. The average acceteration of this car is

r20-30
' :9 (kmihr)isec'
-m
The units appear rather strange, since velocity is in kilometers per
hour and
the time in the denominator is in seconds. we can convert i20 krrilhr
to
3-3j.m/sec if we multipty 120 by r000i3600. similarly, 30 km/hr: g*
m/sec.
With this change the average acceleration is expressed as
11.1 / ul
-)-ra f ol
-10 - : J.5 (m7sec),/sec.

DEFlNlrloN If the velocity u of the particte is varying according to the law tt F (t),*t.."1
=
is the time, then the instantaneous acceleration c, or simply the
acceleiation, is
defined as the limit of the average acceleration:

This is in complete anaiogy vrith the definition ,,rf instantaneous

velocity as
the limit of the averagr": velociiy.
If a particle is mcving (arong;r straight iiriei so that the distance
s is given
by s:/(r), then the velocity u is found Uy taiing the derivative
i,O.fn"
 cilAtTEn

derivative /' is the law which describes the velocity. The derivative of this
derivative is the acccleration. We call it tho.second derivetive and use the
notation f"(t). That is, if s:/(r), then

we could continue this process and take third derivatives, fourth

derivatives, etc., and indeed we sometimes do. The notation for third
derivative is f"'(t) or /(3)1r). [Beyond the third derivative, primes become
unwieldy, and a numeral in parentheses is used. For the seienth
"*r,,pi.,
derivative is denoted /(7)(r).1 we should mention that frorn the point of view of
physics, the first and second derivatives are the most importani-.rpecially
in,
the study of the motion of objects. Derivatives higher ihan the second are
of
occasional interest only.

EXAMPTE 3 A particle moves along a to the iaw

":Ti, "::rl'"g
s- 132 + ]::
108, - l6t2 +3t3,
sbeing the distance in meters and r the time in seconds. Find the velocity
and
acceleration at any time t, what is the velocity when r 2? what is the
=
acceleration when t: I and when r: 3?

Solution The velocity is found by taking the.derivative, which we do according

to the
three-step rule. Then
u= l0g -32t+9t2.
The acceleration is obtained by taking the derivative of this function,
which
we also do by the three-step rule. WJhave

a: -32+ l8,t.
when r = 2, we substitute in the equation for u to get u: lOE @ + 36
80 m/sec. Wlgn, :. 1, we see that a :
- =
32* I g : _ t+ lnoTsecysec, and when
t : 3, a : - 3?* 54 : 22 (m/sec)/sec.
-
since the acceleration is negative when , = 1, we conclude that the
particle is slowing down, while a is positive when t 3 and thereforc
= the
particle is speeding up.
u

5 PnoBtrrrns

In Problems I through 8, a particle is moving along a 5 s=t3-4t2*2t-1, t=-2

horizontal lioe (positive to thc righg, according to ihe stated
las. Determinc whether the particle is moving to the right or to 6 s=la+2t-1, !:!
IhE lcft at thc given time.

I s:t2 -2t+7, !=3

7
': #2,
,
t=2 * ,= r= 3

2 s=tz +t-5, t=l =+,

In Problems 9 through 16, calcutate the vclocity and acoeler-
ation of the given laws of motion. Determine thi times, if any,
3 s:t3 -2tz+7t-6, != -l when thc velocity is zero.
I s:t3 -5t1+Zt+4. t=3 9 s:r2:4t+2 !0 s:6-2t-l6tz
 71
sEcrroN 2.5 lnsrrrrrnrous vEoclry AND s?E> ACca.RAnoN

ll s:*r3 -2t2+3t-5 12 s= tr+2t2-t-l n s: bt1 + ct + d (b, c, d constants)

13 s= bt2 +ct + d (b,c,d constants) 28 s: bt4 + ctt + ilt + e (b, c, d, e constants) .

l{ s= btt + ctz + dt + e (b,c,il,e constants) 29 A stone is thrown upward from the top of a building 50
t2-1 t meters high with a velocity of 15 meters/sec. The height s at
15 s = ---------=
t+ 15 s=._
t'+4 :
any time .r is given by the formula s 50 + l S, 5r2 Find
2
-
the maximum height the stone will reach and how long it
In Problems l7 through 26, motion is along a horizontal line, takes to reach this height. When will the stone hit the
positive to the right. When is thc particle moving to the right
ground? _
and when is it moving to the left? Discuss the aceleration.
t7 s: +3t-l
t2
30 A projectile hurled straight up from the earth travels
according to the law s = xo * stol - 5t2, where xo is the
It s:t2-2t+3 height.above the ground at the start of the motion;-uois the
19 s:8-4t+tz initial velocity, r is the time traveled, and s is the hcight at
Z) s:r3+3r2 -9t+4 time t. Distance is measured in meters, time,in seconds.
A man on a tower throws a ball upward. He obscrvcs
2l s:2t!-3t2-l2r+8 that it passes him on the way down exactly 3 seconds later
22 s:tt -3tz +3t and that ii hits the ground in one more s€cond. How high is
. t
23 5=l+r':
l+r the tower? What is the initial velocity?
24 s:4+t2
3l A projectile hurled upward from the surface of the moon
f+l t2+4 follows the law s: oot - tgt2 where uo is the initial velocity
25 s:rJ 26 s: , . and g is the gravitdtional constant for the moon. A stone
thrown up with an initial velocity of 20 meters/sec is
In Problerns 27 and 28, find expressions for the velocity and observed hitting the surface 25 seconds hter. What is the
acceleration. value ofg, the gravitational constant?

DEFINITION OF ttMIT (OPTIONAD*

In Section I we introduced the notion of limit in an informal way. we spoke

of intervals as being "small," numbers as being "close," quantities ..approach-
ing" zero, and so forth. However, these nonmathematical words vary widely
in meaning from person to person and cannot be the basis for a mathematical
structure. Therefore we give the following statement as a precise definition of
limit:

DEHNffiON Given afunction f and numbers a and L, we say that !(x) terds to L es a limit
as x teds to a if{or each positioe number e there is a posititse number 6 such that
/(x) is ilefined antl

In abbreoiated notation, we write

f(x)+L 45 X+A
for this definition of limit.

'This scction may be omitted without loSs of continuity. Thc problems at the end of
this scclion are.abovc avcrag€ in difficulty.
 CHAPTT[ 2 Lruns. Txr DtRtvATrvE. Appucanoxs

we now elaborate on the meaning of the above definition. First

of all, the
definition imphes that there can be et most one limit L" (we ornit
the proof of
this fact.) Next we recall that

lx-al<6
is the same as the two inequalities

a-6<x<a16.
This double inequality states that x must lie in an interval of length
2d having
____l*+ c as its center (Fig. r9). In the deflnition, the part of the inequalitiwhich
states
a-, e q.+6 that 0 < lx - almerely means that x is not allowed to be equal to oitsetf.
HCUnt 19
This is
dcne for convenience. The inequality

l{(x)-Ll<t
is equivalent to

L-e<f(x)<L+e,
which asserts that the function / lies above the line | : L _e and below the
liney:L+e(FiS.20).
i The delinition :itself may be interpreted as a test. If I am givcn
any
-Rf--l
positive number whatsoever (cail it e), the test consists "a
-l of finding number
d such that /(x) lies between rhe values L- e and L+
i e if x is in ihe interval
ll'it)-rl<e (a-6, a+d) and x*o.r{ such a d can be found for euer,v positive number
t,^tl".n *..-t-u_y that /(x) has the limit L as x approaches a. Note
flcuRE 20 ,fr", ,i" ,"ir.
of delta will be differenr for different epsilons. Also the
test *rrt t. i.riorn .a
fo.r_euery positive epsilon, which means in generar
that it is an extremely
difficult thing to check.
v
t . Ih. geometric explanation, one simirar to that given in Section I, states
that if an e is given, a d can be found such that the graph
of the function / lies
in the rectangle bounded by the lines x : a * 6, x J a 6, y : y _
n, y : L + e

,-:r-N:
j
(Fig.21). Nothing at all is said about the value of whan
/
+
x is a.
. It is good to get some practice in finding the d which corresponds to a
--i ,
lri--;=-;-;+,-, i given e; the 6 can actually-be fo,nd in very simple
cases. To consider an easy
case, we let /(.x) : 3x - 2 and take c: 5. We know intuitively from our
earlier work that
l.x - al < d and l/(x) - Ll < e
frcuRt 21 lrm /(x):3(5) - 2:13.
We wish to show that, given an €, we can find a d such that

l3x-2 * l3l<c whenever 0<lx_51<6.

But l3x - 151 : l3(x * 5)1. If someone gives us an r, we simply
take 6:43.
Then, if lx - 5l < 6 : el3, we find
1by multiplying thiough Uy :i tfrritt, _ St
<e, which is the same as l3x- i5l <e, as is required.

EXAMPI-T 1 Draw a graph of the function

t
f (x):
fr, x* -1.
Find a d such that l/(x) - +l < 0.01 if 0< lx _ 1l < d.
sEcTtoN 2.6 DEFN!noN or L|MIT (omoNAD

Solution Part of the graph is sketched in Fig. 22.lnthedefinition of limit we have L: l

and a:1. We must find an interval of the x axis about x:1 such that thi
graph lies in the proper rectangle. The function decreases steadily as we go to
:r/(t*t) the right and therefore, when we erect vertical lines where the lines y:0.51,
! :0.49 intersect the curve, the largest possible interval on the x axis is
obtained. When we solve the following equation for x,

-J- : g.51.
x+l
we get * : t?, and similarly, solving 1l$ + l):0.49 gives x : f|. In
HGtRf zl
Fig.22 these values are shown with units greatly exaggerated. Fortunately,
once we find a 6, then any smaller 6 will also-be valid; for if the function lies in
a rectangle, it certainly lies in a similar rectangle which is of the same height
but narrower. So we may take 6 :0.03, since ff <0.97 and oi$ > 1.03. n

EXITMPTE 2 Draw a graph of

2(x
Lt 4)
at
J {X\: -+-+,
^ -
r x>0, x*4,
\/x
and find a 6 such that

lfft)-8l<0.01 whenever 0<lx-4l<r.

solurion The function / is not defined at x : 4, but for x * 4 we can multiply
rjumerator and denominator by .Ji *Z to obtain
f(x):2(x-4ir.vG+2)
(x-4)
:2t/i+q forx*4.
The graph of this function is shown in Fig. 23. We construct'the lines y :7.g9
and y: 8.01, shown with greatly enlarged units in the figure. The intersec-
,l--7.W --z tions of these lines with the function are found by solving the equations

Zr/i + 4:7.99 and ZJi +4 : 8.01.

3.S80025 3.985 4 4.015 4.020025 we get x: 3.980025 and x : 4.020025. since the function steadily increases
m.rf 23
to the right, an adequate selection for 6 is 0.015. In other words, it is true that

l/(x) - 8l < 0.01 whenever 0 < lx _ 4l < 0.015. t_l

Remark. To establish the existence of a limit we must find a 6 for each
positive e. It is not necessary that we find the largest possible 6 and, if work
can be avoided by selecting a 6 smaller than the largest possible one, we
usually do so.

EXAMru 3 Show directly from the definition that

fim r :1
,-ix+l 2'
and draw the graph of the function /(x) : x/(x + l).
74
CHAPTER 2 LIMrs. THE DEf,tvAnvE APPilcAnoNs

We have 1: ] and a : l. We must show that for every r > 0 we can find a
6>0suchthat
I x ll
whenoer 0<l.x-llcJ.
l- ,-=l2l <c
lx+l
In order to get an idea of the appearance of the function we sketch the graph
(Jig.2q. Frcm the graph we see that the function is a steadily increasing one.
We verify this fact by writing the identity
'* I

x*1 -l- x+l'

FIGURE 2,r and noting that as x gets larger, l/(x + 1) gets smaller and. as a result,
I- [/(x + l)]
increasesl
Let us first suppose that s <j. Then L+ e< I and L- e > 0, since L:j.
Next, to see where the lines y-t* e and y:)+e intersect the curve, we
solve the equations

xl and
I
r* I 2 x*l -+e.
2

The first equation gives

1^
,,:(1-e)(x+ 1) (*+e)x:+-e € *:?-t =*,-
*+e
Similarly, the second equation yields

i+t
x:T___:=xz.

We select as d the smaller of the distances between I and x, and between I

and xr. The reader can check the fact that 1-x, is smaller than xz- l.
There[ore

d:l-xr:1--:z-1-L +:.y-
1-re l+2e'
ir
Remark. We restricted e to a value smaller than j and then found a 6 for
every positiv e e < +.While it is true that the basic definition states that a 6 has
to be found for e.uery e, in actuality this is not so: Once we have found a d for a
specific 6, we can use the same d for alllarger e. Geometrically, this means that
once the function is known to lie in a particular rectangle, clearly it lies in every
rectangle which has the same sides and contains the particular rectangle.

6 PROBTEMS

In Problems 1 through 17, the numbers a, L, and e are given. 3 f|x):(x2 -9)1ft+3t, a: -3,L: -6,e:0.005
Determine a number 6 so that l,f(x) - Il < e for all x such that
< ai< a f G) : ,,fr, a: l, L:1, e : 0.01
0 i-x - 6. Draw a graph.
I f(x) :2x :- 3, a: l, L:5,e=0.001 5 f{xl:li. a:1,L:l.e :0.01
7 IG):r-2x. a: -t,L:3.e:0.01 e f G) : iG, a:0, L:0, e :0.1
3tcTtoN 2.6 DEMnoNoru fr(omoN/u) 7t

7 f{x) = r/2x, a:2, L=2"e :0.02 In Problems 18 through 25, show that
! f(x) = t1x, a:2, L:i,e=o.oo2
, f(x) : 21.,/i, a= 4, L:t,e
hm f(x): L
=0.1
lfl(x) : 3l$ +2|, a= l, L:1, e =0.001 directly by finding the d corresponding to every positive
e. (Usc
the method of Example 3.)
ll{(x)=lla, a: -l,L: *l,e:0.01 18/(x) : a, a: 5, L:5
lTftrl =(x- !)/(x+ l), a=0,L: _l,e=0.01 t9l(x) : J, a:.1, L:5
20f(x):)y, a:nl2,L=n
l3-f{x) =(6
- \lg - t), a: t, L:*,e=0.01 2t f(x) : (x? 4)lg 2), a 2, L: 4,
14 f(x) : (J2x
- - -
- 2)l@ - 2), a : 2, L=*,e : 0.01 2271x1:rG, a:2,L:UD
15-fix) = x2, a: l, L: l, 6 : 0.01
15{(xl:x3-6, a: 1,L:_5,s:q.l Bl$)=fi, a-3 L=$
'14 f (x): l/(x + t), a:2, L: *
'17 {lx):x3+3x, a: - l,I,:
-4,e:0.5 B_f(x):t/(x+2), a: -3,L= _l

THEOREMS ON TIMITS

--
-In Sections r through 5 we
discussed limits without any
or exact mathematicar statements. In the process attempt at rigorous
of doing thir, iL p".ro.r.a
all sorts of algebraic manipulationr. m"ri.piical
reader realizes tliat each of
these needs justification, €ven though
on th; surface many appear obvious.
The first step in such a justificatioi r.quii",
the precise definition of rimit
which we have just given in Section o.
and the proofs of the theorems which
rr,...-t,;dffi;d b"'inrli"r.rn"n,,
allow us to manipurate timits. However,
we shall restrict ourserves to statements
the proofs are beyond the scope or u
of the ih";;;;;;r,
.ir", ,"r,
statements themselves will help the reader
*uir" ",
in elementary calculus. The
understand the kind of work that
has to be done.
The first theorem we state asserts that
a function cannot approach
than one limit. While trrilra3!rpry1f more
requires the rigorous definition
-of fairry oiviou, th";;;;f,;;iJi*.
rrip,
tirnit *, gu"" in Section 6.

THTOREM 1 (Uniqueness of Limits) Suppose that f (x) + Lr as x -+ a, and f (x) -- iJ r

-, a. Then Lr: Lz.

As an illustration of Theorem l, suppose that

.f(x): x'+
-+ l and a:2.
Then we already know from our intuitive discussion
that

l,T/(S =*:1
 CHAPTER 2 tlMlrs. THE DERtvATtvE. ApprrclnoNs

Theorem I tells us that t is the only number that /(x) canapproach as x tends
to 2.

THEOREM 2 (Limit of a Constant) If c is a constant and f(x) = c fu all ualues o{ x, then

for any number a
lim
x1a
/(x) :6.

Theorem 2 is estabrished by apprying the definition of rimit

particular function f (x):c. Geometiicalty, tfre function
to the

a line parallel to the x axis and c units from it. For

f (x): i iepresents
example , if c = _ 4, then
Theorem 2 states that

lim(-4)=.-4,
and this limit holds for euery number a. We have

lim (-a) - -4, lim (-a) - _4,

x-3 r{8
and so on.

THEOREM 3 (obvious Limit) If a is a rear number ora J1r7 -- 4or arJ x, then
limf(x\=a.
t1d

Th.gl.*-l says that for the special function f (x):x, the limit is always
.
obtained by ilirect substitution ofthe varue afoi x. in otiri
*;r;;;-
' limx:2,
r+2
lim x=-3, limx=5,
x+-3 r+5
and so on.

THEORE\'|4 (Limit of Equal Functiors) suppase there is

a number h>0 such that
/(x) : g{x) for all x for which 0 < lx _ al < h. Suppose also that
lim g(x) : l.
x)a
Then

lim/(x) =1.

This theorem is useful whenever the rimit of

/(x) cannot be found by
direct substitution but where therc is a "simflided iunction,,
g o,lri"n i,
identical with;f except at the point a and is such that the.r;it
of g'"", * o i,
found easily. For example, suppose

,r(x):ry#, g(x) =x.*2.

We see that-/(x) and f(x,f are identical except at x 2,where
=
and g has the value 4. Then Theorern 4 says that since
/ is not defiod

,1
t
sEcrtoN 2.T 7'
Trcornrrsor.ruurn

limg(x) -4, then lim /(x) = d.

x:2 x-2
Another illustration is given in Example 2 of section l. The
are
functions / and g

By rationalizing the denominator we see that these functions are identical

except when x:4. Therefore
lim lim s(x):f.
"f(r:
THtORf^,f S (The Limit of a Sum Is the Sum of *re ,
with

lim /(x) : L, and : Lz,

!r11s(x)
r+a
then

lim (l (+) + g(x)) : L, + L2.

The hypothesis states that r/(x) Lrl can be made ..small,,

-.-
1'close -
to" a; the same is true auout';g1x;
if x is
-'ir1. rn. conclusion asserts that
t/(r)^+ ctD - L, - Lzl can be madei.small" ii, i, .tio;;;,;;.*"
once Theorem 5 is established, we can use it over
and over to add the
limits of any number of functions. For example, if
lim /(x) = L,, lim g(x) = Ir, and lim h(x):
X+O X+d
7r,
x10
then

lim (/(x) + c(x) + tr(x)) = Lr * Lz + Lr.

To demonstrate this, we first apply Theorem 5 to

f and g, designating
"f(x) + g(x) = F(x). Then *" Theorem 5 again to F(x) and lr(x).
This technique of combining and"ppiyusing the same
theorem over and ovcr
occurs frequently in the study of limits. In fact, we
may use Theorem 5 any
number of times to get the followeg general result.

COROttARy Thelimitofthesumof any (fnite) * -

each of the functions.

THEORETIi 6 (The Limir of a product Is rhe produ

functions, with "t
lim /(x) : L, and lim g(x) : Lr,
then

lim [/(x) .g(x)] : Lr. Lz.

CHAPTT[ 2
trrurs. Txr DERtvATIvE nppucenoni

following coroilary.

coRoLLARY
[,rPrr,r*u
of rhe produc, of ony fi

EXAMPIE 1 Given
Irm f (x): yr, lim g(x) = Lr, and lim 1(x) : L.,
:+d
find the value of
lim [/(x) .sU) + h(x)].

solution we define F(x):f (x)'g(x).-From Theorem

6 we know that
lim/(x).g(x): Lr. Lz.
x+a
That is,

lim F(x) : Lr. Lz.

r+a
pplying Theorem 5 to F(x) * h(x), we now find that
lim (f'(x) + n(x)) : L1. L2 * L3. tr

THEOREM z (The Limit of a euotient I. ,h

functions, with
lim
X-A
/(x):lr, limg(x) - tr, and Lz*0,
X-a

then

_ L, (x)
,r^f
**,9(x) L2'
It is necessary to assume that Lr*
meaning.
0 if the expression LrlL, isto have a

EXAMPU 2 Given F(x) : x2, show that

: a2'
l* "ra
solution It is possible to obtain this result
by appearing
of limit. to the definition
However, we can.arso prove the,sta,,ir"",il
ri-piy'ty-rorr..t apprication of
the theorerns on limits. Using Theorem
3, we have

limx=c.
. x-o
Ixt !(x): x ard g(x) : x. Then
Iim
.trA
/(x) = a and lim g(x) : a.
X)4
 79
sEcTto\ 2-7 THEoR[Ms oN uMtrs

Wc gan apply Theorem 6, with Lt: a, Lz: a, to Eet

lim /(r.g(x): e.a= a2,

which says that

'
lim F(x) : a2. D
rro

EXAMPIE 3 Given F(x): x2lex - 2), show that

a

g,.=:;"- ira*i.
Solution From Example 2 we know that

lim xz : a2,

and from Theorems Z Z,*. ino* ,nu,

ana

, - r, Iim (- 2): -2, r: r.

l,Ij l,I
From Theorem 6 with f lxl :3 and gt J : x, we see that

lim i.r, --. 3a.

An application of Theorems 5 and 2 yields

f1t:' - 2) :l,y f:, + ( -2)l : 3:a * (-2) : 3a - 2.

Finally, using Theorem 7 with f (x) = x2, g(x): 3x _ 2, Lr: az, and
Lz:3a * 2, we conclude that

a2
llm-_= t-l
'-, 3.x - 2 3a-2'
THEOREM 8 (The Sandwich Theorem) Suppo.se that
"f \x),gtxt, h(x) haue the pr:oprrry rt*,
' l(x) <s(.x)<h(_x) lbr 0<lx-cl<6.
Suppose also that

lim
\+a
/(.x): L, lim ft(.x): L.
x-a
Then

Iim g(x) : 1.
h\r)
(c) The situation is shown crearry in Fig. 25. The function g(x)
is squeezed
between f and i and so must tend to L as x tends
l@) to c] we provide an
application of Theorem 8 in Example 6 below.

iI
- In working the problems at the end of this section the reader shourd give
the reason for each step. An abbreviation for the name.
ratherl itran the
oitu number of the theorem being used, helps in the deveropm.;;
oi a crear
I
understanding of the processes. we illustrate this procedure
Hctnt 25
in the next three
examples.
80
CHAPTEN 2 UMrrs. THE DERtvATIvE AmLcIrIoNs

ilAMPtE c Find the value of

-'v3+1
lim ,' -
,-2 2x, + 5
and justify.each step,

Solution Step lim 2:2,

1: x-2 lim 3 : 3, lim 5:5 (lim. const.)
x-2 x+2

step 2: l,I, = 2 (obv.lim.)

Step 3: lim 2x2 : lim (2 'x ' x) : 2'2'2:8.

x-2 x-2
limx3:lim(x.x.x):g (lim. prod.)
xr2 x-2

Step4: lim (x3+ 3):8+ 3:11,

x-2
lim (2x2 + 5):8 + 5: l3 (lim. sum)
x-2

.. x3+3 11
5: l,i (lim. quot.)
Step
Xr=: 6
I
THEOREM g (Limit of a Composite Function) Suppose f and g are functions, a and b are
numbers, and

lim /(x) :f (b) and lim g(x) = !.

r+t

Then

lim /[g(x)] =f (b).

x40

THEOREM 10 If n is a positiue integer and a>0, then

lim
x+4
jG: fi.
The next result can be obtained by combining Theorems 9 and 10.

THEOREM 11 If n is, a positiue integer, L> 0, and

lim f (x): l-, then tim {f g) = tL.
xr4 x+o

EXAMPLE 5 Evaluate

lim
17-ro
l-
'
x-2 v x3_8
 81
stcTroN 2.7 THEo*EMS ON IIAIIIS

Solution Straight suostltution shows that the function is undefined whea x: 2. If we

denote the quantity under the radical by,f(x), we find by factoring that for
x *2:
(x-2)(x3+2x2+4x+8)
f(x): (x-2)(x2 *2x+4)
Setting
_,..\ xt +2xz* *
8(r):-7Trx+4
4x g

'

we observe that g(.x) is defined for all values of x and f (*) :g(x) except when
x:2.
Step l: lim 2 : 2, lim 4: 4, lim 8:8 (lim. const.)
. x- 2 .x-2 x'.2

Step lim x:2

2: t.- (obv.lim.)
2

Step 3; ,t '- t, hq 2x2 : 8, lim 4.x: 8,

]13
lim x2 :4, lim 2.x:4 (lim. prod.)
x-2 x-l

Step 4: .r*2
lirn (xr + 2x2 + 4x + 8):32,

lim (x2 .+ 2x + 4) : 12 (lim. sum)

x-2

x3 +2xz*4x* 8 32
Step 5: lim
Y{, x'+2x+4 12
(lim. quot.)

Step lirn g(x) :

6: x-2 lim /(x) : 328 (lim. equal fcts.)
x-2 123

7: itm
-:- /8
(lim.,//(x))
.
SteP
x-2 ;r_ 8 v3
n
EXA .IPLE 6 Let

g('x) =
lr'-:l
and show that
lim g(x) :0.
x+0

Solution Straight subsiitution shows (hat s(x) is undefined when x:0. Since
g(x) > 0 always, and since

l,i" ll r
lxl =
 CTIAPTEI

* 0, if we define the functions /(x) and ft(x) as f (x) :0

for all x for all x and
h(x): lxl for all x, we have
,f(x) < S(rc) < h(x) for all x * 0.
We then proceed:

Step l: lim X:0 (obv. iim.)

xro

Step 2: lim ft(x) : lim lxl = 0 (lim. comp. function)

r+0 r-O

Step 3: lim /(x) : S (lim. const.)

x-0

Step 4: lim g(x) : Q (Sandwich Theorem)

r-O
n
It is important to know the eleven theorems of this section and have a
clear understanding ofhow they are used. However, as a practical approach
to the evaluation of limits of algebraic expressions, stiaight substitution
always works except when there is a zero in- the deno,minatir.In such cases
either there is no limit or, if the numerator is also zero, Theorems
4 anc g can
sometimes be used. In chapter 10 we will study additional techniques
for
evaluating limits.

7 PROBTEMS

In Problems I through 2e evaluate the limits by following the

methods given in Examples 4 and 5. Give the reason for each
step as in those examples.
,'m*f,x>o
l lim(x2+4x-3)
,-2
2 lim (x3 +2x2 -7x + 4) rc ri- 1/-l- l\
r-oi\x+tr - x) ,,mi(ft-,)
4 llm-=-+6x-5
x2
,-t xr *2x.1 7 -5-tl
lg I.m(l+rr)3/2-l 19 lim^ "
i-o h x-z x-2
6 li* "*8
,--2 x*2 -
*6 ?'ro

,2
--. _a r-3 x*3
8 lim
,--z Xr * 8
In each of Problems 2l anel22, use the Sandwich Theorem to
evaluate the limit.
l0 lim l--7
.) l +I
2l lim
,-o -/xa +
22 lim ,'luin lY
II l2 3x2 +6 r-o \ x/
*23 Prove Theorem I by employing the definition of limit.
*24 Prove Theorem 2 by employing
l3 the definition of llmit.
r25 Prove Theorem 3 by ernploying the definition of limit.
 83
sEcTloN 2.7 THEoRt ,ts oN ui trs

'2.5 Show that

,.,n //(x) -g(x) + ft(x)p(x)\2"
",' *-o \ g(x) + r,(x) /
limxsinl:0 _-t, +_t=rto\2t3
r-O X _ ( r, , if L2 + Lt # o.
by using the definition of limit in Section 6. \ Lr+L. /
?? Given 2E Use the definition of limit to show that if a is any number
then
lim /(x): /-,, iim g(.x) : l.r,
h*':o''
iim &(x) : L., lim p(x):1o,
29 Let P("r) : a,xn + an tX' ' + ... + arx + ao be any poly-
:tate the theorems which justify the following statements: nonrial. Show that

. lim P(x) : P(b).

a, /(x) +8(x)
Iim : '' l t, , if Lt # o.
rtb
x+a htx) L\ 30 a) Suppose that lim*-, f (x\: L. Show that lim_-. l/(x)l
-= li-l.
O, ,,* "f(x)g(x) -
ft(x) _ LrL,- L3
if L2+ L4*0. b) Suppose that lim,*o lJ'(x)l: ll,l. Show that it is not
xro g(x) +p(x) L2+ L4 neci:ssarily true that lim,-,.f(x) : L. Give an example.

COUNruUITY. ONE.SIDED LIMITS

In Section 6 we analyzed the meaning o[

l,* t k):
L.

In doing so we made a point of ignoring the actual value of the function

when.r: a. In fact, for many expressions the function was not even defined at
/
x:a. Suppose we have a function / which is defined at x: a,and suppose
limit L, which l'approaches when .x tends to c, is the value of / when x is
the
a-just the quantity we call l'(a). When this happenr, *. ,r.y that the
function is continuous at .x: rz.

DEFINITIONS "The / is continuous at the number a' means that

Most, but not all, of the furrctions we have been studying are continuous
everywhere. For example, the functions

f (x).: 3x - 2, g(x) : 5- 2x * 4x2, and l(x): Il]

2+x2
 CHAPTTR 2 bMrrs. THE DInIVATIVE AmIcATloNs

are defined for all values of x and are continuous for all values of x. on the
other.hand, we shall show that the function

F(x) : -:
{;:::;, :;: :
is discontinuous at x: l. The graph of part oi F is shown in Fig. 26. we
define the functions

G(x) : 2x + l, H(x) : i*, - 3,

in each case the domain being all of Rl; that is, G and H are given by the
above formulas for all values of x. Now we apply Theorem 4 on the limit of
rtctfrE2f equal functions. we observe that F and G are identical for all values of x in
the interval - co < x < 1; therefore the limit as x tends to I from the left (that
is, as x tends to I through values less than l) of the functions F and G must be
the same. Since G is continuous everywhere, we see that F( l) :2, I + 1 : 3 is
equal to the limit of G(x) as x tends to I from the left. we employ a similar
argument for F and rI; these functions are identical for I < x < @. Also, .tI is
continuous for all values of x. Therefore the limit as x tends to I from the
right (that is, as x tends to 1 through values iarger than l) of the functions F
and H must be the same. We have H(1) :
- f and so F(x) tends to - j as x
tends to I through values larger than l. These statements are intuitivelfclear
in the graph shown in Fig.26. The definition of continuity for F is not
satisfied at x: 1; this assertion is justified in Theorem 12 on page g5.
In the definition of limit, when we state that
f (x) --+ l, as x --+ a,

we must consider values of x larger than a and values smaller than a as x

tends to c. However, in the above discussion on the discontinuity of F, we saw
that it is important to consider the limit of F as the number 1 is approached
through values entirely on one side of l. This leads to the following symbol
for "one-sided limit." when f (x) tends to L as x tends to a through oarues
larger than a, we write

f(x)--L as x-*a*.
The plus sign after the letter a indicates that x tends to a through values
larger than c. we also say that L is the one-sided rimit from the riint or
1.
The definition of one-sided limit from the left is entirely analogous to that
above. with the change that x is smaller than c instead oi lurg.ithun c. we
use the notation

f(x)-L as x)a,
the minus sign after the letter a indicating that x tends to c through values
smaller than a. Additional notations for one-sided limits are

,'1T.
[Q): L, ]yn tt'l: t.
All the theorems on limits stated in section 7 carry over to one-sided
limits with the exception of Theorem 9 on composite functions. ln that
theorem wc must assumc that the limit of / exists, ilthough one-sidcd limits
of g are allowed. In additioq we have the following extension of Thcorem lo.
SECTION

THEOREM l0' If n is a positiue iiteger, then

[m ,f:g.
x*O+

we note that the restriction ic one-sided limits is necessary

in Theorem 10,
since .r"& is not defined for negativ.
,utu", oi, when n is an even integer. An
extended theorem similar to Theorem I I holds for

,-on
lim {f (i
" "
' if .f(.x) - o* as x - a* .

The next theorem is an immediate consequence

of the definitions of one-
sided iimits.

THEOREM 12 If f is afunction and a and L are numbers, then

lim f (.x):7
if and oniy if
lim f{x):l
x-a+
and
x
lim
+a'
/(x): l.
Theorem l2 asserts that the limit of exists
/
exist and the1. haue the same ualue. For
if and only if bothone-sided limits
ordinary limit is the two_sided limit.
this reason, *;,;;;;;;;;;;
;ffi;
Using Theorem 12, we can justify the conclusion
that the function F
given on page g4 is not continuous at :
x t. Both one-rio"a'rrnirr;;;';;lr;,
but they have different values. Therefore tt e
t*o-siaeJhmit (ordinary limit)
does not exist and the function cannot U"
.".tirrrr*
ExAMptE 1 Find the value of rhe one-sided Iimit

and give a reason for each step.

''1?-#
Solution We factor the expression, getting

-.x2 _ (3 + x)(3 _ x)
9

\// r + zx - x. .r/(l + x)(3 _ x)

(3+x).Al
:_--+- for-l<x<3.
Jl+x
From the theorems on limit of a constant
and obvious limit, applied,to
sided limits, we have one-

lim
r$3 -
l:1, lim 3:3, lim x=3.
x+3 - r+3 -
Therefore, from the limit of a surn (one-sided case), we find
lim-(3*x):6, lim (t*x)=4,
-'r (,
r-3- r-3_ ,,iT_ - x) = 0.
CHAPTER 2 LIMITS. THT DBIVATIVE AMICANONS

Also,

hm .uA -- x :0, lim 1E + x : 2, tlim. {f G) I

,,,o Q:{1p : T : o, (lim. prod.; lim. quot.)
x*3- Jl +x
9-xz (lim. equal func.)
lim
,-r-.r/3 +2x-xz
--:0. tr

EXAMPIE 2 For what values is the function

x2 -9
"f(x): x+3
continuous?

Solution This functiort is defined for all numbers except x : - 3. Straight substitution
- / /
at 3 yields $ and is undefined. Therefore is not continuous at 3. For -
all other numbers in the domain of !,the function has the same value as the
: -
function g(x) x 3' and so is continuous. In fact,
lim /(x): -6. u
' :--3
Remark. The important distinction between the functron / in the above
example and the function g(x) : x - 3 is found in the ilomain of thesc
r, I
functi,ons. The domain of g is all of R while the domain of / is all of R with
the exception of -3. The functions coincide wherever they are both defined,
but the functions are not identical. These fine points occur frequently and they
cannot be overlooked.

EXAMPTE 3 Find the values of x where the function

h(x): 1{2*', -l<x<l

l<x<2
t3-x,
^

is continuous.

Solution The domain of h is the half-open interval l-1,2\. Clearly h is continuous

except possibly at - I and at l. Taking one-sided limits, we find that
lim 2x2:2 and lim
:- I +
(3 - x):2.
ra I -

Since both one-sided limits are equal and since h(1) = 2' we conclude that ft is
ct.utinuous at x:
l. (See Fig.2l.)
At x :"-
1, the function h has the value 2. Also, the limit as x tends to 1
-
from the right exists and the limiting value is 2. Under these circumstances,
we say that h is continuous on the right at i: - 1. tr
F'GIfi,T
Remark. If a function / is defined on a closed interval [a, b], then continuity
of may be defined at every interior point. The function / is continuoE on thc
/
right at the endpoint a if and only if
 sccttoN

lim /(x) :f(a).

t{a +

Similarly, / is continuous on ate hft et b if and only if

Iim /(x):f(b).
r+t -

DEFINITION A function / rs continuou, on

interiar point, continuous on th"|tgnl oi a, and
"^r-":t::i::ll:r"ry
on the left at b.
rinrinioi

To say that a function has a derivative at the

varue 4 means that a
tangentline to the curvecan bedrawnat thepoint(a
the derivative, defined as
,[email protected]"i, tt
"r"ruror
(a + h)
,r*f -f (a)
i-o h
is simply the slope of this tangent line. As
we have seen in Exampre 3, a
continuous function may havc corners, and
it is hard tt-irnugn" how a
tangent to a curve can be constructed at
a corner; in grn..ur,-ii cannot. Not
every continuous function possesses a tangent
line at e;;;iri. A, example is
z the function

,,-r-12*',
r\^'-f21,-z;', -l<x<l
t<x<2
As Fig' 28 exhibits, F(x) is continuous
at x :-r, but does not possess a tangent
A continuous function with no line at (!, 2). The two dotted.lines in the grrpf,.no*
derivative at x: I the tangent line to a curve is unique m"ust
th"t;;;.;;r"ment that
Ue violated.
nc{ft 28 lf a function has a derivative, however, it is continuous,
as we prove next.

THEoREM13Ifthefunctionfpossessesaderiuatioe,iM.

Proof We recall that two things must be shown:

i) f (a) is defined and ii) lim f (x):f (a).

r+4
lf a function has a derivative at a, then
,r^{(a+b.l-f@) :f,(a)-
r \v'
l
[-o h
exists' This statement wourd have no
'having meaning it f (a\ did not; therelore
a derivative at c means thar/(a) must
' ha-ve-a';;;rl;g. As for (ii), wc
see that it has the same meaning
as

lT /(' + h\:f (a).

We can write

f(a+h)-fat:(eire),"
CHAPIEN 2 UMTIS, THI IxnvAIIvE Amlcmoxs

On the right, the term in parentheses approaches f'(a\, a finite numbcr, as lr

tends to zero. The second part, namely lr, tends to zero. Therefore, as fi tends
to zero, the right side-according to the theorem on the limit of a product-
tends to zero. We conclude that

o
and (ii) is esrabrished ifi::"i[l;:^,:
'-

8 PROBTEMS

In each of Problems I through 12, evaluate the one-sided limit, function is continuous. Sketch the graph.
giving a reason for each step.
I
x-l 15l(x):,-+: for -4<x<7
I lim--4
,-1. Jx2_l
2
,'lT.
('+ G=)
x-t 3

3lim:
4-x2 *
4 lim: -3x+2 x- - l_Y - l)
,-z- J2*x-x2 x-2, Jr, -d
f(-3):2
x4-16 lxn - to
5lim 5 lim l_ -4 x2
0<x<5,.r*2
,-z-.Jx-2 ,-z'{ x - 2 11 f(xl: ir_Ar-+f Z,
-
t ,r^ iF(4 8 lim- lx-ll+lxl f (2):1
xr I
'-o- ./x(3 - x) x*2
l8/(xt for-l<x<3
ttr, x--zx-x
9 lim
r-O +
*,x-I l0 x
x-o Vl1+-.
lim
.Y- . x3-l
19 f(x) = --------^, 0<x <2,x*l
x'+x--l
,,,,lT-1x-JG];F1 ^,,-4_rI
_.:_____
12 lim
,-z-,t/ 6-5.x+x2 f(tj : 1

13 Evaluate 2o f(xt:['^ o' -t<x<2

2<x<5
[r'-6.
1i* l, - tl and
,l]1 l, -
t|.
(*'-6, -co<.,r<-!
I
Does (the two-sided limit) lim,-, 2t[tx):{-5, -l<x<10
lx - 1l exist? Draw the
(x-15, l0<x<co
graph.
14 Evdluate (x2 -
l" -l -l<x<2, x*l
22f(xl:{xa-l'
lim a and lim a. lx2 +3x-2, 2<x<5
'-or lxl ,-o- lxl
f (1):,
Does
23f(x)=lx+31 for all values of x
24 f (x): l2.x - 5l for all values of x
lim A
,-o lxl -_1
for all x except .x:2
exist? Give a reason for your answer.
lx-21
f (2J:0
In each of Problems 15 through 30, a function is defined in a
certain domain. Determine those values of x at which the
26 f (x): ' i f ,r'' for atl vatues of x
 89
SECTION 2.8 CoNnNUrY. Oxr-soco rnrurs

-2-*-{^;
" for all values of x except x: 3l Given
27flx): ^ :x-J
.- 3
{x-2n for 2n<x<2n+1, n:0,1,2,...
:Izn-x
/(3): s "l(xl for 2n-l<x<2n, n=1,2,...,
'2S/tr):.in1 for -l <x< l. x#0 where is / continuous? Sketch the graph for x in the
interval [0.6).
,r(0) :0 32 a) Given /(x):lx-ll+t2-xl, show that has theI
constant value 1 on the interval I < x < 2. For what values
'?B.f(x):rrin1 for -1 <x< 1, x*0 ofx is y' continuous?
.r(0) :0 /
b) Write a formula for a continuous function which has
the constant value 2 on the interval 0 < x < 3 and is not
30./(*):lxlall -xl for -1 <x<3 constant outside this interval.

9
LrMtTs AT tNFtNtw; tNFtNrTt uMtTs (opfloNAD*

The graph of the function /(x) :

x2l(l + x2) is shown in Fig. 29. The values
of this function are always less than 1; in particular, we note the following:

Ilt :r, I
/(ro) :
l#, .r(roo) :
frffi /(rooo) - #,
As x gets largbr and larger, the graph of / gets closer and closer to the line
I: 1, as shown in Fig. 29. We use the term "x tends to infinity" when refer-
ring to-values which increase or decrease without bound. If x is becoming
c larger through positive values only, we write x + * coi if through negative
values only, we write x --) - oo. The symbol x --+ .o means that lxl increases
EELtf, 2!t without bound.t In the above example, we would write

/(x)--+1 as x+co.
More generally, we formulate the following definition.

DEFINITION Wesay that/(x) -+casx--+ o, iffpr eache>0ther:eis anumber A>0,such

that lJ$) - cl < e for all x for which lxl > A.

If we want to define
. f{x)--c as x++co
/(x) tends to c as x tends to infinity through positive values only), the
(i.e.,
only change in the definition would be to write x > .,{ instead of lxl >,,{.

*This section may be omitted without loss of

continuity.
tsome texts use the symbols o, - co, + @ corresponding to our use in this scction of
+ @, - co, o, respoctivcly,
CHAPTER 2 TIM]TS. THE DEnTATM. AmIcATIoNs

Further, the definition of

as x+-96
"f(x)*c
would require x <
- Ainstead of lxl >,{.
The theorems on uniqueness of limit and on lirnit of a constant, equal
functions, sum, product, and quotient, and the sandwiih Theorem remain
unchanged.In place of the theorem on limits of functions of
the forrn f (x) : x,
we have the following theorem on obvious limits.

THEOREM 14 ,. I
Itm _:0, -.
itm -I :0, -.
Ilm I
x-m X x- + r, X r--oI --0.

EXAMPI-E 1 Evaluate

u- 3'- ?
,-**5x*4'
giving the reason for each step.

sotution we employ a standard trick in working with infinite limits by dividing

each
term in the numerator and denominator by x:
3x-2 3-(2,,xt A
5x+4:5+(4A)' x#0, -5
For the numerator, we find that
/ t\ z .\
Iim
,-+m\ {:-:):x/ lim 3+ lim (-1).
xr+@ ,-*o,\ X)
since the limit of a sum is the sum of the limits. we now reason that
(using limit of a constant)
,IT- ':'

t,^ (- ?') 2 lim - : 0

1

..** \ x / = - x-*o
(using limit of a product and Thcorem
X 14)

Therefore

.IT- ('-:) :'

In exactly the same way,

.lT- ('* 1) :'

Now we can say that

lim (3 -(2lx))
3
, *. 5 + (4lx) - (2/.x) rr+a

,- . lim (5 +(4lx))'
r- + @
 5ECTtoN 2.e
91
lmns nr rNFrNrw; rNFrNrrf r.r,r,rrs toEioillil

since the limit of a quotient is the quotient of the limits.

: we conclude that
(limit of equal functions)

,. _:3x-2 _3
llm I
,-+- 5x*4 5'
EXAMPTE 2 Evaluate

,,^F
re+m 2x+l '
giving a reason for each step.

Solution We first divide numerator and denominator by

x, getting

JT:OM
2.+t :-2+{rd-
Since

I 1l
x2xx'
we have

. I
lim .. II . lim l:0,
lim I
, x-*otr -: x-*oX x_*oX
using the limit of a product theorem and ,Iheorem 14.
Then

,IT- (t - *) :,liT- t -,1T. { : ' (using limit or a sum).

Further, by using Theorem 11, we find that

I :
lim t.
r+ + @ x'
1

We observe that
v
I :
y : x/(r- l)
.IT- (, * i) r,
as in Example r, and when we apply the quotient rule
and the theorem on the
limit of equal functions, we have

,,, fE:1
r-+o 2x*l 2' t-]

:
i_l
The graph of the functio^n _ l)
/(x) x/(x is shown in Fig. 30. We note
the following values of this function:

"f(2):2, f(*)=3, /(t.t):11, /(1.01): tOt, /(1.00,): t,00t, ...;

"f(0) =0, f(il- -1, /(0.9) : -9, /(0.99) =
rlfx -99, f(o'gg) _ _ssg,....
This type of behavior impries that the function
grows without bound as x gets
 92
CHAPTTR 2
UMtrs. THE DERtviTtvr. Applcrrlorus

closer and closer to 1. We use the expression

.f(x)-.o: as x-1,
which we define in the following way.

DEFINITION We say that f (x) becomes infinite

,
each number A > 0 there is a 6 > 0
0<lx-af<6.
such'that lf(x)i> ), nr" ii| jo, *hirh

This definition does not state whether

thc function goes off to infinity
"upward" or "downward." In
worJs, *" .uy have
(x) - cases of-other
/(x) _-) + oo or
"f - co ?s special /(r) ; ;."For exampie, tf f (x)_+ + oo as
x-)a,we say that ,,/ becomes positively infinite as xiends io.ol,, Th. orly
change required in the above a"nnitior"i,
ii._."ptu.".nent of l/(x)l > A by
3 I shows the function g ('i)-:
FtcunE t t *, ,wr,icr, u"co,iies
31
{H],r" f; ?ire [ositivery
Similarly, a function m.ay become negatively
infinite as x tends to a. In this
case,rile replace l"f(x)l> A by f (x). _]l in ,r,. above definition.
we also consider functions which become infinite
approached from one side. In Fig.30 as a varue a is
it is clear that f {x)--+ *oo as x-_ l+
and f(x) -,
-oo 8s r+ I-. We-now e-'- gi* u- p*"ire definition of one-sided
limit from the right. I

DEFtNtTtON We say that f (x) becomes p*iti*ty

i, c a > 6 ,,,i ,n,i it i,,
{!i"::,!,y{u:,1,!:,,,
0< lx- al<6 and x is larger than a. fo, for ail x for which
ti"ir')7r'rrilt;;r;;,'i:r:"r:;:
notations

"f(x)
--, + oo as x --+ a+ qrd lim /(x) : + €.
x-q+

Analogous definitions and notations

are used for functions which
become negati-very infinite and for
o*-Jira-rmits from the reft.
Figure 3l shows a function which becomes
positiveiy infinite as x tends
to zero both from the reft and from the ..two-
right. There is
sided limit" theorem for functions
which #;e ^r;";i;;;;;e
infinite. w. oun""on.iude, for
example, that g(x): l/x2.becom.r poriiir"tyinfinite
as x _+ 0..
functions have infinite limits from"one
side only. for example, the
,*n.il3. _

h1x;: -J-.
J, -2
becomes positively infinite as x-* 2n, b ut /r is not defined
Therefore the limit from the left cannoi for x less than 2.
exist and consequently there can be
no (two-sided) lirnit as x tends to 2.
The rules for operating with limits are
somewhat trieky when one of the
quantities becomes infinite. we must
..*L*u". trrat co is.not a-number and
cannot be treated as such. Theie are
some simpie facts aboui infinite lirnits
which the'reader will easily recogni"r.
F;; **u*pt., t; ;;;;;;;';,
 93
SECTtON 2.9 UMtTs AT tNFtNtry; tNHNtrE ttMlrs (opTloNAt_)

f(x)-'+m as x-+a'
g(x)*c as x--+a,
where c is any number, then

/(x)+g(x)--)+o6 as x)a.
If r' * 0, we have

,f (*) 'gi.x) -'+ + oo when c > 0

and

,f (x) 'g(x) + - oo wheq c < 0.

On the other hand, if c:0, further inuestigation is necessary. To see this, we
let /(x):ll?-3)2 and g(x): (x-3)r. Then /(x)--+ +co as x-*3 and
g(x) - 0 as x --+ 3. However,

/(x)'s(x): l,
and so, f .s- l asx + 3.If wereplace,f(x)bythefunction AIG-3)2,where
A is any number, the product /. g tends to ,4. The reader can easily construct
examples in which f 'g- *m and f .g- 0as x*3.(See Problems 25and
26 at the end of this section.)
- For cases in which f (x) -* * oo and g(x) *
-.o as x --+ a, nothingcan be
said about the behavior of

f(x) + g(x)
without a closer examination of the particular functions.

9 PRosLrA s

in *-ft of Problems I through 15, evaluate the limit or show x'f I

fu.ilr :"m fuocrion tends to oo, + co, or - co. Give a reason for 15 lim
'ru lxl
rufi sco.
lr-4 In each of Problems 16 through 19, determine those values of x
iY 2 li^ *'=* 2* * 5 at which the function is continuous. Sketch the graph.
-
r
,-o 2x'- 6x * I
.r-x2r7x+5 -/57 +T
(l
-{, frm 4 lrm l/(x): x+r- for-7<x<-1, x*-3
=-l tt -:Zx*l r++@ x*3 16 {
_;r+4 U(-3):2
t'-
r--r x*2 6 lim (Jxz + a2 - x) (- x*3
r-+@
- -6<x<6, x+-3,3.
'l* s -i--: ,, {r,r,:"4,
t/( -3) :2, f (31: t
"= fim
'-t I -x' ,-24-x'
(2x
g h*tI*' lo lim l_ , for 0<x<2
r-oI' r{6 "+l
x rE/(x):l;,_; for 2<xs5
-z -i
nl Em\-x-2 Lr*u for 5<x<7
x-2-
12
,IL (,trF+2x-n
= for all x except r:2
,, {r,', H
Lf Q)=t
94
CHAPTER 2 LrilTs. THE DERtvATlvE. AmlcArloNs

In each of Problems 20 through 23, sketch the graph of the -@ as y+qand /(x)+g(x)++@ as xia. Find
given function. Then decide what limit, if any, is approached. functions fandg suchthat f(x\- +co,g(x)+ -o63sx
- c and /(x) + g(x) + - co. If .4 is any given number, find
*20 lim /(x) where /(x) : x sin 1 functions f and gsuch that f(x) - +co, g(x) + -co as
x-0 x + a, and /(x) +g(x) + A as x - a.
+

*21 lim /(x) where /(x): i t* i *27 Let P(x) : anf * a,- 1x'' r + ... + a1x * as and Q(x) :
t
bnxn * br- tx^- + '.' + b 1x + bs be two polynomials
with a, # 0 and b- + 0. Find the value of
*X2 lim f(x) where /(x) : ,l t* i
+ x-O P(')
ti,n
*23 lim
/(x) where /(x) - sin ,-n- 9(x)
I m: (iii)
by considering three cases: (i) m > n, (ii) n, nr < n.
*24 Given ihe function / defined by the conditions *28 Let P andQbe defined as in Problem 27. Find the value of
0x if x is a rational number,
It') : P(x)
to if x is an irrational number.
,IT- o(r'
Find lim /(x). *29 Suppose that P(x) and 0(x) in Problem 27 are written in
x-0 +

25 Findfunctions/andgsuchthat/(x) r +cx) andg(x) +Q the form P(x) : a"1r - cr)(x - c)(x - ca) "' (x - c,),
as x J d and /(x)'g(x) - +co as x - a' Also find func- Q$l: b.(x - d1)(x * dr) "'(x - d-)' Describe the vari-
tions / and g such that /(x) + * oo, g(x) - 0 as x * c and ous possible results in evaluating the
,f(x)'S(x) - 0 as x --+ a.'Is it possible that /(x) + * @,
g(x) -+ 0, and /(x)'8(x) - r{ as x + a, where ,{ is rititorffi as.x-d,.
negative? Justify the answer.
26 Find functions f and g such that f(x) - + 6 and 8(x) t wherc i is an integer between I and m.

CHAPTER 2
REVIEW PROBTEMS

In Problems I through 4, find the limit (if any) approached by 1l /(x):x-2 +2x 12
"f(x) as x tends to a.
x2 + x-2 t4 /(x) :
t /(x)=ffi, a=l 13 f(x): xn + x2 - 2 x5

In Problems 15 through 18, find the equation ofthe line tangent

2 r$):tffi,
-2x-3 x2
a:3 to the given curve at the given value of a.
15 y=2x2+3x-1, a:-l
3 /(x)= '[ij2=-t,
x-- lx
a:7 ,,_1

-6_.1
' x*4'
af!):^r'.',
x'+ x
a:-r 11 y=J*+5, a=2
Yz
In Problems 5 througb 14, use the three-step rule to find the l8 y: ^- a- -4
derivative.
x+z=,
5 f(x\:2x2-x+l 6 f (x): -xz + 3x -2 In Problems 19 through 24, a particlc is moving along a straight
7 f(x\:x3-2x2+l 8 f(x)=2x3 +x2 -3x+ I line according to the given law. Find thc times, if any, when the
3x velocity and the aceleration ate zoro.
e-f(x):= l0 f(*): r+l 19 s: i2 +2t-3 2I) s: 2t2 --3t + 5
CHAPTET

2l s=tt+2t2-t+1. 22 s=2tt+t2-t+4
,, -. I
23 s:t+:.
tL)
2A r: ,,1T.
Q+ G +2x) 32 llfit-
*-z- xt *2
t-l '"-t+l " ,, (lxl+12-xl) y lim (fxl+12-xl)
25 A man in a stationary balloon 500 meters high hurls a stone ,,g. x-2-
straight down with an initial v.eloc[ty of 2 meters per In Problems 35 through 39, determine the values of x for which
seqond. How Iong does it take for thl stone to strike the the given function is continuous.
earth?

ln Problems 26 through 29, evaluate each of the limits. citing the 35,r(x)-[x+2, -l<x<2
2 <x< 8
[6 -r,
rppropriate theorem to justify each step in the evaluationl

f r*3 $ f(xt_[x'z+2, -l<x<t

[m(x2-2x+5) 2i u* 1'-'--2 [4-r. l<x<2
'--txz*4x+3
't_ [- J7 f(x)_{lxl-lx+ ll, -tsxsl
n [m-4 v^
r-a 29 lim Jx3-2x+2 (-12-xl, l<x<5
-.X
I Suppose that lim f(x\: L and L is negarive. 3t f(x)-{txt, -5<x<2
tlx-41, 2<x<5
Show that
lg.f6: {i t n is an odd integer. 39 f'(x):5# :
for all x except x -2,3; I(-2)=t,
tn Problems 3I through 34, evaluate the one-sided limit. f(31=2.