IES OF CoON

defned
function ff defined
ONTINUOUS FUNCTIONS
UNCTIONS
The by f{z)= 330
8.1(c)) 1/z ifz
y 0. (See Fig.
discontinuit at
=
4 0,
f0) =
khas an infinite
infinite
AY

(0.k)

8.1(¢) Infinite discontinuity at

Fig. 8.1(a)
Fig =0. Neither f(0+0) nor
exists. f(0-0)

The function f defined

(iv)The function
det by
hatever )sin/2)ifz
may be the value #0, f(0) khas a
neither f{0 +0) discontinuity
=
of k. In this
st case
eists.(SeeFig. 8.1(d). nor f{0-0)

8.4 lementary Properties of Continuous

Functions
rn8.4.1 Let f ond g be tuo
functions of z defined in some
f f and g are both continuous atc. Then neighbourhood of
0) f+9%f-9 and 1.9 are all continuous at c.

6i) f/g is also continuous at c, provided that g(c) # 0.

lgl is continuous at
(") If| or c.

Proof: All these results follow as immediate consequences of theorems on limits of

Chapter 7 (see Theorem 7.4.2).
For example, we wish to prove that f+g is continuous atc if f and g are both
continuous there.
Since f is continuous at e, we have, im f(o) =
f().
Also since g is continuous at c, have, lim g(r)
we =
g{¢).
By the theorem on limits,

limfa)+9tz)} limf(e)+ lim gte)

= =
flo) +9lc).
 DIFFERENTIAL CaT
340 AN INTRODUCTION TO
ANALYSIS:
CALCULUs
Hence, f+g is continuous atC
Similarly, we may discuss the
cases for f- 9 , f*9 and f/g (g(c) #0
Again, from the result: lim |f(r)| lim f() we may obtain part (i).

Composite Function
Let f(r) be defined on a set D with range R. Let g(u) be defined on a set Di.
contains R(D' 2 R). If z E D, then f(c) E R and consequently f(z)=.uc
ED
which
that 9{f(r)} is defined. The resulting function F, dehned by F()g{f(a)},iscal
the Composition of g with f. We say that the composite function F is defnedon D

Theorem 8.4.2 Letf and g satisfy the conditions of the previous paragraph. ff a
g are both continuous at a certain point c E D, then the composite function F =al
is also continuous at c.

Proof: Let uo = f(c), uo ¬ D'. Let e be any given arbitrary positive number.
Since g is continuous at u = uo, corresponding to the given e, 3 a positive number
61 such that
g(u)-g(uo)<6, Yu in lu-uol<61.

Again, since f is continuous at a =, corresponding to that determined positive

number 01, 3 a positive number d2 s.t.

If(r) -

f(c)| < 61 Vz¬ |r - cl <

ðz
ie., luuol<81 Vae |T-e <ô2.
Givene> 0, 3 61(c) whence find 62(61,e) such that lg(u)
we can
-g(uo

i.e., lo{f()} -9{f(©)} <e, Vz e

lr -c< &a:
or, F()- F(c)| < e,
i.e., F is continuous at o = cED.
The theorem is thus proved.

REMEMBER: A continuous function is also a

continuous function. of a continuous functio
pon,

Corollary 8.4.1 f a junctaon f has its values

a
a t

0, and if f is
uous

t = is also
c, then Vj continuous at c. contnu
 .4.
ELEMENTARY
PERTIES OF
CONTINUOUS FUNCTIONS
341
eghbourho0d Property
. a 1ff
rheorem 8.4.3 Iff continuous at certain. point and if
is a
neiohbourhood of such that V in that
c flc) T
c
# 0, then there
sign
as that of f(e). neighbourhood f(r) has the

let cibe an interior point

w (i) First, of [a, b], where f
continuous at c
corresponding to e
is defined. Since it is
=
If(c).
i e at h a t

mberè such that There exists a

positive
fla)-flc< lfe), Vce|r -d <6.
ie,
f)-f©N < fa) <
ie)+ 5lfl)], Yzelz-d <6.
(8.4.1)
E fc)>0, (8.4.1) gives f¢) < fla) < /fl), Vze -cl <6.
Since fc) and zflc) are both positive, we conclude
that

VrETd<6, every f(z) is positive.

D)LE fIc) <0, (8.4.1) gives if() < f(e)< ~fl), Vz ¬
|r-d <6.
Since f(c) and f(¢) are now both negative, we conclude that
YE Tcl <6, f(«) is
every negative.
i) Next let c coincide with the left-end
a of
a)0. We know that f(r) is continuous at a [a, b| has fa
i.e., is continuous at a and
f(c) right-continuity at a.
rom the definition of
s right-continuity it will follow, as before, f(r) has the same
that of f(a) in a certain nbd like
(a,a + 0).
u1)f c coincides with b i.e., if f is continuous at b and f(6) #0 then 3 a nbd
6-6,6) where every fc) has the same sign as that of f{b).

9plications of the previous theorems

ExA MPLE 8.441v
*Aconstant function is continuous at any point in its domain.

)=c Vre R. At any point To E R.

hence ete.
-flzo) =
|e-ed =0< any given e(> 0);
EXAMPLE
point
8.4.2 Identity. entity function: Let f(r) = #, Vae R. 1hen f is continuous
zo in its
domain.
fle)Jel= lz -ol<e, Vse |z-cl <6 uhere d=e; hence etc.
 342 AN INTRoDUCTION TO ANALYSIS:
DIPFERENTIAL CALC
CALCULUs
EXAMPLE 8.4.3 f(z) = r", (where n is a ger) isis
positive intege; continuor
contin
Vaues
of ¬ R. For any arbitrary cE R, twe have for a
lim f(c) =
lim z" =lim z lim 2 lim T* *m z =c C.C C=
=
.
fe).

Hence n is continuous at c and c being any arbitrary point of R, t is eo

for all z e R.

Polynomials
EXAMPLE 8.4.4 Let P(r) = ag#" +a1"*+**tan-14+n be a polynOt
Xof degree n. Then P(z) is continuous for all values of z E R. ial in
SOLUTION: This statement is true because for any real c, we have

lim P(r) = lim " 4+a1 lim " + t a n - 1 limz+lim a

=
agc+a1c"*i+...+an-1C+an =
P(c); hence etc.

Rational Functions

ExAMPLE 8.4.5 f(r) = «", where n is a

negative integer, is continuous for all
ezcept forz 0. More generally, a rational function
=

point of R uhich is not

is P(c)/Q(7) continuous at eren
a zero
of Q() (i.e., uhere Q(z) #0).
See that r" = *"" =
uhere n= -m(m being a positive integer
Trigonometric Functions
EXAMPLE 8.4.6 sinz and cosa are continuous at
hencewe say they are continuous for all real any arbitrary point 7 -"
values of ,
See that sinz sin c -
=
2 cosT sin
S2-1- < la -

e< any e(> 0)

Nak vhenever |z -cd <6, choosing
DEXAMPLE 8.4.7 tan r and
I =
natT/2; cot z and coseczsecare continuous
(n being 0 or any positive
are
continuous for all values of T
ercept uhet

negative integer). for all values of z

or when
erCep
Transcendental Functions
EXAMPLE 8.4.8 The
1erse erponential function
function log, T is contiruous e* is
andits i

values of continuous everyuher

for all Us everywhere
See Ex. 8.5.11 for detailed 0.
discussions.
 D 6

1uous Functions
blluou1s
anctions
now
discuss (with proofs) the
importa part following more deep
s h a l .

de
We
which form an
actions
of Real Analysis: properties
roperties ofof continuous
continuous
termediate Value Property.
2. Boundedness,Property.

s. Uniform Continuity Property.

Compactness and Continuity.
5. Continuity and Existence of Inverse functions.

6. Continuity and Monotoncity.

1.
Intermediate
Value Property
Theorem8.5.1e (Bolzano's Theorem on
in a closed interval
Continuity). Let f be a real-valued con-
function [a, b]. Suppose that f(a) and f(b) are of opposite
ie.
signs (i.e., let fla) f(b) 0).
let < Then there erists at least one
pointc where a <c<b
such that flc) =0. e

The truth of this theorem seems obvious and indeed it is evident geometrically: If the
ruph of a continuous function lies above the X-acis at one end of an interval la, b} and below
the X-aris at another end, then it must cross (at least once) the X-aris someuhere in-betuen.
However, the point is to give a proof based on the arithmetical properties of real numbers
rather than on a geometrical picture. Remember that there are continuous functions for which
Bo geometrical picture can be drawn].
rooj: For the sake of definiteness, assume f(a) > 0 and f(6) <0.
Dunce f(a) >0, by neighbourhood property 3 some right-neighbourhood ofa where,
or
every z, f(r) > 0. Let
A = {z:ze la,b) and f(c) > 0}.
bounded. Therefore,
Clearly, A is non-em
mpty subset of [a, b] (. a ¬ A); A is obviously
has aa supremum = c (say).
Aas
A

Since 3 a right-nbd of a, where f(r) > 0, c#a.

Since 3 aleft-nbd of b, where f(«) < 0, c#6.
.'. a <c <b.

We shall
establish that f(c) = 0.

e0, then either f(c)>0 flc)

or
property, 3 8 > 0 such

Let, if possible, J(c)

f(c) > 0. > Then, by
the
neighbourhood

.>0 for all Ee [c -6,c++6. that c=

supA.
Hence f(c)
the fact
Por some c> contradicts

Anot be positive. J ( T ) > 0.

This
POsiti
 () Next suppose f(c) <0. Then 36>0 such that
flr) <0 when z e
le -6,e.
c=
Since c sup A, by definition of Supremum it follows that
3ne A such
such#.
that
c-6 n S c .

Now that 7 ¬ A f(n) >>0 and n ¬

we see >
(c-6,c > f(n) <0.
Thus, we arrive at a contradiction and as such f(c) is not negative.
From (i) and (ii) we arrive at the conclusion that f(c) = 0.

Alternative Probof

Proof of Bolzano's Theorem by the method of Bisection

Take a =
(a+b), the mid-point of la,b].
In case, we find f(a) = 0 the proof is complete.
I, however, f(a) £0 then either f(a), fla) or fla), f(6) muse haveopposiesi
Of the two intervals (a, a] and a, b the one at the ends of which f(7) has opa
signs, we re-name as a1, b1. Hence in either case, we get
aa1 bi < b;
 fa.) is m.i., bounded above and hence convergent and {bn} is m.d.,
From (8.5
ence
below and hend convergent.
bounded
Fom (8.5.2)

lim( n a n ) = lim
n+o0 noo (6- a) =0.
lim an = lim bn = l (say).
noo n+oo

either be an interior poit of la, b] or it may be an end-point of

l may
The point
poi prove
that f() = 0.
shall
We
let f() # 0.
f possible,
may be f () 0 or
erior point of la, b]. Since f() # 0 it >
I. Let l
be an
such that for
neighbourhood property there exists an same |l-8,1+8]
Case interval
By sign.
this interval f(z) and f(l) have
the
point of
7
every l from above, there exists a positive integer
Ca an Ifrom below and 6n
lies within |-6, + and accordingly f(am), f(bn) have the
ch that lam, bm conclusion is
arrive at a contradiction of (8.5.3). Hence the only
me sigT, Thus, we

f)=0.

coincide with a. In this case an a, V n. =

Case II. Let l interval

such that for every point r of this
Since f(a)#0, 3 an interval a, a +6 m such that
have the s a m e sign. Since bn > l = a , 3 a positive integer
flc and f(a)
a bm <a +0.
that we again have a
and f(bm) have the same sign so
Accordingly, f(an) f(a)
Outradiction. Hence l cannot coincide with a.

cannot coincide with

b.
mlarly, it may be shown thatl
0. Hence the
of la,b} where f{)
=

us, we have proved that 3 an interior point

theorem.
the following:
unnediate consequence of Bolzano's Theorem is
continuous
Let f be a real-valued
Theorex .5.2 Intermediate
ned
Value Theorem).
7 & closed interval la,b]. Suppose that f(a) # f(6)

at least once.
Then f assumes ry value
every
between f(a) and f(b)

Proof: Let j be a number between f(a) and f(6).

We onstruct afunction o such that do(r) = f(«) -L.
is obvic since f is so
and alsoo
y continuous in la, bl
f(6)
-

f(a)
-

4, o(6) =

de of o(a) =

oppositesigns. 0.
Hence 3 at lleast ô(c) =
that
where a < c<b such
One point c,
But ofe*0>f(c) = u. Hence the theore