You are on page 1of 429

THE BOOK WAS

DRENCHED

DIFFERENTIAL

CALCULUS

SIIANTI

NARAYAN

DIFFERENTIAL CALCULUS

(Review published in Mathematical


Gazette,
London, December, 1953

THE

book has reached

can be assumed that

it

Its

meets

5th edition
all

in

demands.

9 years and
Is

it

the revie-

wer's fancy to discern the influence of G. H. Hardy

opening chapter on
clearly dealt with

Or

the

in

numbers, which are well and

real

it

this only to

is

be expected from

an author of the race which taught the rest of the world

how

to count

.-

"i

The course followed


and there

is

comprehensive and thorough,


v
good chapter on curve tracing. The author
is

has a talent for clear


and
exposition,

is

sympathetic to the

difficulties of the beginner.

Answers to examples, of which there are good and ample


selections, are given.

Certaianly
excellent

Mr.

Narayan's

Our own young

command
scientific

of
or

English

is

mathematical

grumbling over French or German or Latin as


additions to their studies, would do well to consider their

specialist,

Indian

confreres,

with

English

technical education can begin.

to

master

before

their

DIFFERENTIAL CALCULUS
FOR
B. A.

&

B. Sc.

STUDENTS

By

SHANTI NARAYAN
Principal,

and Head of

Hans Raj

the

Department of Mathematics

College, Delhi University

TENTH REVISED EDITION


1962

S.

CHAND &

CO.
NEW DELHI BOMBAY
JULLUND tTR LUC KNOW

DELHI

Books by the same Author


Integral Calculus

Rs.

6'25

Modern Pure Geometry

Rs.

5'00

Analytical Solid

Rs.

5-50

Geometry

A Course of Mathematical Analysis

Rs. 15*00

Theory of Functions of a Complex Variable


A Text Book of Matrices

Rs. 12-50

7'50

Rs.

A Text Book of Vector Algebra


A Text Book of Vector Calculus
A Text Book of Cartesian Tensors
A Text Book of Modern Abstract Algebra
A Text Book of General Topology (Under preparation)

Rs.

6 50

Rs.

7'50

Rs.

6 50

Rs. 14'00

Published by

CHAND &

S.

for

Shyam

Lai Charitable Trust,

16B/4, Asaf

AH

Ram

Road,

New

Delhi.

book are spent on

(All profits from this

S.

CO.

CHAND

charities.)

CO

&
NEW

Nagar

Fountain

DELHI
DELHI

BOMBAY

Lamington Road
Mai Hiran Gate

JULLUNDUR

LUCKNOW

Lai Bagh

First Edition

1942

Tenth Edition 1962


Price

Published by

G S

Shartna. for S.

Rs. 7-00

Chand

Printed at Rajendra Printers.

&

Co.,

Ram

Ram

Nagar,

Nagar,

New

New

Delhi-1*

Delhi and

Preface to the Tenth Edition

The book has been

revised.

few more exercises drawn from

the recent university papers have been given.

SHANTI NARAYAN

30th April, 1962.

PREFACE
This book is meant for students preparing for the B.A. and
Some topics of the Honours
B.Sc. examinations of our universities.
standard have also been included. They are given in the form of
appendices to the relevant chapters. The treatment of the subject
is rigorous but no attempt has been made to state and prove the
theorems in generalised forms and under less restrictive conditions as
It has also been a
is the case with the Modern Theory of Function.
constant endeavour of the author to see that the subject is not preThis is to see that the student
sented just as a body of formulae.
does not form an unfortunate impression that the study of Calculus
consists only in acquiring a skill to manipulate some formulae

through 'constant

drilling'.

The book opens with a brief 'outline of the development of


Real numbers, their expression as infinite decimals and their representation by points 'along a line. This is followed by a discussion
of the graphs of the elementary functions x"> log x, e x sin x, sin- 1 *,
Some of the difficulties attendant upon the notion of inverse
etc,
functions have also been illustrated by precise formulation of
Inverse trigonometrical functions. It is suggested that the teacher
in the class need refer to only a few salient points of this part of the
book. The student would, on his part, go through the same in
complete details to acquire a sound grasp of the basis of the subject.
This part is so presented that a student would have no difficulty in
an independent study of the same.
,

The first part of the book is analytical in character while the


later part deals with the geometrical applications of the subject.
But this order of the subject is by no means suggested to be rigidly
different order may usefully be adopted at
followed in the class.

tho discretion of the teacher.

An

analysis of the 'Layman's' concepts has frequently been


to serve as a basis for the precise formulation of the corresponding 'Scientist's' concepts. This specially relates to the two
concepts of Continuity and Curvature.

made

Geometrical interpretation of results analytically obtained have


been given to bring them home to the students. A chapter on 'Some
Important Curves' has been given before dealing with geometrical
applications. This will enable the student to get familiar with the
names and shapes of some of the important curves. It is felt that a
student would have better understanding of the properties of a
curve if he knows how the curve looks like.
This chapter will also
serve as a useful introduction to the subject of Double points of a
curve.

Asymptote of a curve has been defined as a line such that the


distance of any point on the curve from this line tends to zero as
the point tends to infinity along the curve. It is believed that, of all
the definitions of an asymptote, this is the one which is most natural.
It embodies the idea to which the concept of asymptotes owes its
importance. Moreover, the definition gives rise to a simple method
for determining the asymptotes.

The various principles and methods have been profusely


by means of a large number of solved examples.

illus-

trated

am indebted to

Prof. Sita

Ram

Gupta, M.A., P.E.S., formerly

Government College, Lahore who very kindly went through the


manuscript and made a number of suggestions. My thanks are also
due to my old pupils and friends Professors Jagan Nath M.A.,
Vidya Sagar M A., and Om Parkash M A., for the help which they

of the

rendered

me

in preparing this book.

Suggestions for improvement will be thankfully acknowledged.


January, 1942

SHANTI NARAYAN

CHAPTER

Real Numbers, Variables, Functions

ARTICLE
ri.
1'2.
1*3.

1*4.

1'5.
1 6.

1*7.

PAGE
Introduction
Rational Numbers

...

...

Real Numbers
Irrational Numbers.
Decimal representation of real numbers
The modulus of a real number
Variables, Functions
Some important types of domains of variation

Some Important

13
14

Classes of Functions and their Graphs

Graphs of y=x*
Monotoiiic Functions. Inverse Functions
n
Graph of y x*i
x
Graph of y=a
Graph of >>^log a x
Graphs of sin x, cos x, tan x, cot x, sec x,

26.

11

...

II

2*2.

2-5.

...

...

2-1.

2-4.

...

...

Graphical representation of functions

CHAPTER

2-3.

...

2
5
6
9

...

18

...

20
22
24
25

...
...

...

...26

cosec x
2-7.

Graphs of shr^x, cos^x, tan^x, cot~*x see^x,

2*8.

cosec^x
Function of a function

...

2-9.

Classification of functions

...

CHAPTER

...

30
34
35

III

Continuity and Limit


3-1.

32.
3*3.

3-4.

35.
361.
3'62.
3-63.

Continuity of a function
Limit

...
...

..
Theorems on limits
Continuity of sum, difference, product and
quotient of two continuous functions. Con...
tinuity of elementary functions
Some important properties of continuous functions
...
Limit of x n as -> oo
...
Limit of x n jn as n -> oo
...
Limit of (1 1//I)" as n ->oo
...
Limit of [(a*~l)/Jt] as x -0
X
...
Limit of [(x
-^)/(x-o)] as x->a
...
as
x->0
Limit of (sin .v/x)
...
Hyperbolic Functions and their graphs
7;.
Inverse Hyperbolic Functions
I

3-64.
3-65.
3-66.
3-7.

3-8.

37
41
52

53
54
57
58
59
63

64
64
66
68

VI

CHAPTER IV
Differentiation
4-1.

4*11.
4-12.
4-14.
4-15.
4*16.

Indroduction.
Derivability.

Rate of change
Derivative

Derived Function
An Important theorem
Geometrical Interpretation of a derivative
Expressions for velocity and acceleration

4 22.

Derivative of x

4-3.

Spme

4-34.
4*35.
4-36.
4*4.

4*5.
4-61.
4-62.
4-71.
4-8.

4-91.
4*92.
4-93.

72
72
74
74
76
78
TO

...
...

...
...

...

...

general theorems on differentiation


Derivative of a function of a function

Differentiation of inverse functions


Differentiation of functions defined

81
86
88

...

...
...

by means
88
89
93
96
97
97
99
100

of a parameter
...
...
Derivatives of Trigonometrical Functions
Derivative of Inverse Trigonometrical Functions
...
Deri vati ve of loga x
...
Derivative of (f
...
Derivatives of Hyperbolic Functions
...
Derivatives of Inverse Hyperbolic Functions

Logarithmic differentiation
Transformation before differentiation

...

Differentiation 'ab

...

initio*

Appendix

...

102
104
108

...

113

...

116

...

118

...

CHAPTER V
Successive Differentiation
5-1.
5*2.

Notation
Calculation of nth derivative.

Some standard

results
5*3.

Determination of the

th derivative of rational

functions
5 -4.
5*5.

5 -6.

Determination of the nth derivative of a product of the powers of sines and cosines
Leibnitz's theorem
Value of the nth derivative for

x0

CHAPTER
General Theorems.
tvl.

6*2.
6*3.

6*4.

...

120

...

121

...

123

VI

Mean Value Theorems

Introduction

...

Rolled theorem
Lagrange's mean value theorem
Some deductions from mean value theorem
Cauchy's mean value theorem

...
...
...

...

130
130
133
136
137

vn

development of a function in a finite


form
Lagrange's and Cauchy's forms of remainders
Appendix

6*6,6-7. Taylor's
;

CHAPTER
Maxima and Minima,
7-1.

Greatest and Least values

Definitions

...

...

CHAPTER
Evaluation of

...
...

VIII

8-6.
8'7.

Limit of /(x)

8-2.

8-4.
8-5.

F'(l

\Y
'

"

148
149
151
157

Indeterminate forms

limits.

Introduction
...
Limit of [/(x)/ir (x)] when/(x)->0 and F(x)->0 ...
Limit of [/(x)/F(x)] when/(;c)->oo and F(xj -*oo ...
Limit of [/(x).F(x)] when/(x)->0 and F(x)->oo ...
Limit of [/(*) F(x)J when/(x)->oo and F(x)->oc

8-1.

140
145

VII

necessary condition for extreme values


7-3,7-4. Criteria for extreme values
7*6.
Application to problems
7*2.

...
...

under various conditions

...

165
166
170
173
174
175

CHAPTER IX
Taylor's Infinite Series
9*

Definition of convergence

and of the sum of an

...

179
180
181
185

Rigorous proofs of the expansions of


m
cos
...

188

infinite series

...

9-2,9*3. Taylor's and Maclaurin's infinite series


Formal expansions of functions
9-4.
9-5.

Use of

infinite series for evaluating limits

Appendix.
ex

sin x,

x, log (l-fx),

...
...

(l+x)

CHAPTER X
FUNCTION OF
VARIABLES

WO

Partial Differentiation
4

10 1.
10 '2.
10*4.

10*5.
10'6.

10*7.

10*71.

...
Introduction
Functions of two variables and their domains
...
of definition
...
variables
a
two
of
of
function
Continuity
...
Limit of a function of two variables
...
Partial Derivatives
Geometrical representation of a function of
...
two variables
Geometrical interpretation of partial derivatives
...
of first order

193
193
194
195
196

197

198

10*81.
10*82.

Eider's theorem on

10*91.

Theorem on

VIII

...

199

...

204

Calculation
...
Differentiation of Composite Functions
...
...
Differentiation of Implicit Functions.
Appendix. Equality of Repeated Derivatives.
Extreme values of functions of two variables.

206
210
213

Homogeneous functions
A new Nota-

Choice of independent variables.


tion

10*93*
10*94.

total Differentials.

Approximate

Lagrange's Multipliers

...

Miscellaneous Exercises I

...

230
232-237

...

238

...

239

CHAPTER XI
Some Important Curves
IT!.
11*3.
11*4.

Catenary
Explicit Cartesian Equations.
Parametric Cartesian Equations. Cycloid,
Hypocycloid. Epicycloid
Branches of a
Implicit Cartesian Equations.
Cissoid, Strophoid, semi-Cubical parabola
Polar Co-ordinates
Polar Equations. Cardioide, Leinniscate.

curve.
11*5.
11-6.

Curves r m =a m cos w0, Spirals,


Curves r~a sin 3$ and r=a sin U0

CHAPTER

...

243
247

...

248

...

XII

Tangents and Normals

and Paramedic Cartesian

12-1;

Explicit, Implicit

12*2.
12*3.
12*4.
12*5.
12-6.
12-7.
12-8.

equations
Angle of intersection of two curves
Cartesian sub-tangent and sub-normal
Pedal Equations. Cartesian equations
Angle between radius vector and tangent
Perpendicular from the pole to a tangent
Polar sub-tangent and sub-normal
Pedal Equations. Polar equations

CHAPTER

..

..
..
..

..
..

..
..

254
262
264
266
267
270
271
272

XIII

Derivative of Arcs

13-3.
13-4.

An axiom
the meaning of lengths of arcs.
a
function
as
an
arc
of
Length
To determine ds/dx for the curve y=f(x)
To determine dsjdt for the curve

13-5.

To determine dsjde

13'1.
13>2.

On

*=/(0, ?=/(')
for the curve

r=/(0)

...

.,.
...

...

274
275
275

276
277

CHAPTER XIV
Concavity, Convexity, Inflexion
14-).
14*2.

1 -t-3.

14-4.

281

Definitions
Investigation of the conditions for a curve to be
concave upwards or downwards or to have inflexion at a point
Another criterion for point of inflexion

...

Concavity and convexity

...

282
284
285

...

290

...

291

...

291

w.r. to line

...
...

CHAPTER XV
Curvature, Evolutes
15-1.
15-2.
15vJ.
15-4.

lo'-l'O.

15-47.
15*48.
15-51.
15*5").

Introduction, Definition of Curvature


Curvature of a circle
Radius of Curvature
Radius of Curvature for Cartesian curves.
Explicit Equations.
Implicit Equations.
Parametric
Equations. Newton's formulae for radius of curvature. Generalised
Newtonian Formula.
Radius of Curvature for polar curves
Radius of Curvature for pedal curves
Radius of Curvature for tangential polar curves
Centre Curvature. Evolute. Involute. Circle
of curvature. Chord of curvature

Two

properties of evolutes

...

...

...
...

...

...

292
298
299
300

304
310

CHAPTER XVI
Asymptotes
16-1.

Definition

16'2.

...
Determination of Asj^mptotes
Determination of Asymptotes parallel to co...
ordinate axes
,.:_<
of
rational
Asymptotes
general
Alge^raic'Equations

IO31.
lfi-32.

...

"

1(5-4.
lfi-5.

](>(>.

I(r7.
1C-8.

Asymptotes by inspection
Intersection of a curve and
Asymptotes by expansion

...

its

asymptotes

Position of a curve relative to its asymptotes


Asymptotes in polar co-ordinates

...
...

...
...

313
313

315
317
324
325
328
329
331

CHAPTER XVII
Singular Points, Multiple Points
17*1.
17-2.
17-31.
17*4,

Introduction

...

Definitions
Tangents at the origin
Conditions for multiple points,

...

...

,..

335
336
336
339

17-5.

Types of cusps

...

17-6.

Radii of curvature at multiple points

...

CHAPTER

342
345

XVIII

Curve Tracing
18*2.
18-3.
18-4.

18-6.
18-6.

Procedure for tracing curves


2
Equations of the form }> =/(x)
Equations of the form y*+yf(x)+F(x)-^0
Polar Curves
Parametric Equations

...

348
349
357
359
364

...

369

...
...

...
...

CHAPTER XIX
Envelopes

19

l.

One parameter family of curves


Definitions.
Envelopes ofy=MX-t-ajm

19*2, 19'3.
19*4.
19*5.
19*6.

...
obtained 'ab initio*
...
Determination of Envelope
Evolute of curve as the Envelope of its normals ...
Geometrical relations between a family of curves
...
and its envelope

Miscellaneous Exercises II

...

Answers

,,.

369
270
37:2

372
378-82
383-408

CHAPTER

REAL NUMBERS
FUNCTIONS
Introduction.
The subject of Differential Calculus takes its
stand upon the aggregate of numbers and it is with numbers and
with the various operations with them that it primarily concerns
itself.
It specially introduces and deals with what is called Limiting
operation in addition to being concerned with the Algebraic operations of Addition and Multiplication and their inverses, Subtraction
and Division, and is a development of the important notion of Instantaneous rate of change which is itself a limited idea and, as such, it
finds application to all those branches of human knowledge which
deal with the same. Thus it is applied to Geometry, Mechanics and
other branches of Theoretical Physics and also to Social Sciences such

as Economics
It

may

and Psychology.
be noted here that this application

is

essentially based

on the notion of measurement, whereby we employe-numbers to


measure the particular quantity or magnitude which is the object of
investigation in any department of knowledge. In Mechanics, for
instance, we are concerned with the notion of time and, therefore, in
the application of Calculus to Mechanics, the first step is td~ correlate
the two notions of Time and Number, i.e., to measure time in terms
of numbers. Similar is the case with other notions such as Heat,
Intensity of Light, Force, Demand, Intelligence, etc. The formula^
tion of an entity in terms of numbers, i.e., measurement, must, of
course, take note of the properties which we intuitively associate with
the same. This remark will later on be illustrated with reference to
the concepts of Velocity, Acceleration Gwry^fr&e, etc.

The importance of numlTe^l^/Bfe-sradj^bf i^subject in hand


being thus clear, we will in some* .of the; f#ltowig. articles, see how we
were first introduced to the notion of number and how, in course of
time, this notion came to* be subjected to a series of generalisations.
:

It is, however, not intended to give here any logically connected amount of the development of tl>e system of real numbers, also
known as Arithmetic Continuum and only a very brief reference to
some well known salient facts will suffice for our purpose. An
excellent account of the. -Development of numbers is given in
'Fundamentals of Analysis' by Landau.

It may also be mentioned here .that even though it satisfies a


deep philosophical need to base the theory part of Calculus on the

notion *>f number ialohe, to the entire exclusion of every physical


basis, but a rigid insistence on the same is not within the scope of

2
this
will

DIFFERENTIAL CALCULUS

book and intuitive geometrical notion of Point, Distance, etc.,


sometimes be appealed to for securing simplicity.
Rational numbers and their representation by points along a
1-1.

straight line.

Positive Integers.

I'll.

It

was to the numbers,

1, 2, 3, 4, etc. r

we were first introduced through the process of counting certain


The totality of these numbers is known as the aggregate of
objects.
that

natural numbers, whole numbers or positive integers.

While the operations of addition and multiplications are^,unrestrictedly possible in relation to the aggregate of positive integers,
this is not the case in respect of the inverse operations of subtraction
and 'division. Thus, for example, the symbols
are meaningless in respect of the aggregate of positive integers.
Fractional numbers.
1*12.
At a later stage, another class of
numbers like p\q (e.g., |, |) where p and q are natural numbers, was
added to the former class. This is known as the class of fractions
and it obviously includes natural numbers as a sub- class q being
;

equal to

in this case.

The introduction of Fractional numbers is motivated, from an


abstract point of view, to render Division unrestrictedly possible and,
from concrete point of view, to render numbers serviceable for
measurement also in addition to counting.
1-13.

Rational numbers.

Still later,

the class of numbers was

enlarged by incorporating in it the class of negative fractions including negative integers and zero. The entire aggregate of these numbers
is known as the aggregate of rational numbers.
Every rational
number is expressible as p\q, where p and q are any two integers,
positive and negative and q is not zero.

The introduction of Negative numbers is motivated, from an


abstract point of view/ to render Subtraction always possible and,
from concrete point of view^, to facilitate a unified treatment oi
oppositely directed pairs of entities such as, gain and loss, rise and
fall, etc.

Fundamental operations on rational numbers. An impor1-14.


tant property of the aggregate of rational numbers is that the
operations of addition, multiplication, subtraction and division can
be performed upon any two such numbers, (with one exception which
is considered below in
T15) and the number obtained as the result
of these operations is again a rational number.
This property
rational

numbers

is

expressed by saying tli^t the ag^egate of


closed with respect to the four fundamental

is

operations.
1-15.
Meaningless operation of division by zero. It is important
to note that the only exception to the above property is 'Division

REAL NUMBERS
by

zero'

follows

which

a meaningless operation.

is

This

may

be seen as

To

divide a

by b amounts to determining a number c such that


bc=a,
will be intelligible only, if and only if, the determi-

and the division


nation of c is uniquely possible.

Now, there is no number which when multiplied by zero produces a number other than zero so that a JO is no number when 0^0.
Also any number when multiplied by zero produces zero so that 0/0
may be any number.

On account of this impossibility in one case and indefiniteness in


the other, the operation of division by zero must be always avoided.
A disregard of this exception often leads to absurd results as is
illustrated below in
(i)

(/).

Let

jc-

6.

Then
;c

-36:=.x-6,

or

8.
(x--6)(jc-f.6)=Jt
Dividing both sides by jc 6, we get

jc+6=l.

6+6=1,
which

is

12

= 1.

clearly absurd.

Division by jc
absurd conclusion.
(ii)

We may

6,

^lf =

which

remark

also

is

zero here,

is

responsible for this

in this connection that


6)

*~^~

=x+6,

only when *j66.

...

(1)

hand expression, (je2 36)/(x 6), is meaningwhereas the right hand expression, x-f 6, is equal to 12 so that
For *=6, the

less

i.e.,

left

the equality ceases to hold for Jt=6.

The equality (1) above is proved by dividing the numerator


and denominator of the fraction (x 2 36)/(x 6) by (.x 6) and this
operation of division is possible only when the divisor (jc -6) ^0, i.e*
9

when x^6.

This explains the restricted character of the equality (1).


1.
Ex.
Show that the aggregate of natural numbers is not closed with
respect to the operations of subtraction and division. Also show that the
aggregate of positive fractions is not closed with respect to the operations of
subtraction.

Ex.

2.

Show

that

every rational number

is

expressible as

a terminating

or a recurring decimal.

To decimalise plq, we have first to divide/? by q and then each remainder,


after multiplication with 10, is to be divided by q to obtain the successive
figures in the decimal expression of p/q. The decimal expression will be
terminating if, at some stage, the remainder vanishes. Otherwise, the process
In the latter case, the remainder will always be one of the
will be unending.
finite set of numbers 1,2 .......
tf1 and so must repeat itself at some stage.

DIFFERENTIAL CALCULUS

From this stage onward, the quotients will also repeat themselves and the
decimal expression will, therefore, be recurring.
The student
some

will understand the argument better if he actually expresses


fractional numbers, say 3/7, 3/13, 31/123, in decimal notation.

Ex.

For what values of x are the following equalities not valid

3.

W --*+.

-!.

(0

tan -

1*16.
Representation of rational numbers by points along a line
or by segments of a line.
The mode of representing rational numbers
by points along a line or by segments of a line, which may be known
as the number-axis, will now be explained.

We

marking an arbitrary point


The
the
origin or zero point.
calling
be represented by the point O.
start with

axis

and

Any

one of these

it

The point

"ft

Fi

and the

On
it

may

"

tive

divides the

hand

axis into

zero will

two parts or

sides.

be called positive and the other, then negative.


is
number-axis
Usually, the

left

number

on the number-

number

side of

the positive side,

drawn

parallel to the printed


of the page and the right
hand side of O is termed posilines

negative.

we take an

arbitrary length

OA, and

call

the unit length.

We

say, then, that the

number

1 is

represented

by the

point A.

After having fixed an origin, positive sense and a unit length on


the number axis in the manner indicated above, we are in a position
to determine a point representing any given rational number as

explained below

Positive integers.

m.

We take

ger,

m.

Firstly,

we

consider

any

positive integer,

a point on the positive side of the line such that its


distance from O is m times the unit length OA. This point will be
reached by measuring successively m steps each equal to OA starting
from O. This point, then, is said to represent the positive inte-.

m, w<
Negative integers. To represent a negative integer,
take a point on the negative side of O such that its distance from
is
times the unit length OA.

This point represents the negative integer,


m.
Fractions. Finally, let p\q be any fraction
q being a positive
OB being one
Let OA be divided into q equal parts
integer.
of them. We take $ point on the positive or negative side of O
according as p is positive or negative such that its distance from O is
p times (or, p times jf p is negative) the distance OB.
;

The point so obtained represents the

fraction, p\q*

REAL NUMBERS

If a point P represents a rational number pjq, then the measure


of the length OP is clearly p\q or
p\q according as the number is
positive or negative.

Sometimes we say that the number, p/q,


segment OP.

is

represented

by the

We

Real numbers.
have seen in the
number
can
be
rational
every
represented by a
the line with
see
can
cover
of
a
it
to
that
is
line.
we
Also,
easy,
point
such points as closely as we like. The natural question now arises,
"Is the converse true ?" Is it possible to assign a rational number
to every point of the number-axis ?
simple consideration, as de1-2.

Irrational numbers.

last article that

tailed below, will clearly

show that

it is

not

so.

Construct a square each of whose sides is equal to the unit length


P on the nutnber-axis such that OP is equal in
the length to the diagonal of the square.
It will
now be shown +hat the point P
cannot correspond to a rational number
of OP cannot have a
i.e., the length
7i r
rational number as its measure.

OA

and take a point

Fig. 2.

measure be a rational number pjq so that,


by Pythagoras's theorem, we have
If possible, let

its

= 2,

i.e.,

p*

(0

2q*.

We may suppose that p and


factors, if any,

Firstly

q have no common factor,


can be cancelled to begin with.

we

for,

such

notice that

so that the square of

an even number

even and that of an odd

is

number is odd.

From thHtion (/), we


Therefore, p itlHpinst be even.

that

equal to 2n where n

Let, the

Thus, #

see,

is

p2

an

is

an even number.

integer.

is alsc

Hence p and

#mfcrommo*factor 2 and this conclusion conno common factor. Thus the


the hypothesis
thaPthej^fce
measure -y/2 of OP is not a rationaffiumber. There exists, therefore^
a point on the number-axis not corresponding to any rational number,
traflicts

is

Again, we take a point L on the


any rational multiple say, p/a, of OP.

line

such that the length

QL

DIFFERENTIAL CALCULUS

The length OL cannot have a


m\n be the measure of OL.

rational measure.

For, if possi-

ble, let

m
p
mq
/0
or v
-\/2=-\/ 2
v
n
q
p
which states that <\/2 is a rational number,
being equal to mqjnp.
This is a contradiction. Hence L cannot correspond to a
rational number.

>
'

Thus we see that there exist an unlimited number of points on


the number-axis which do not
correspond to rational numbers.
If we now require that our aggregate of numbers should be such
that after the choice of unit length on the line,
every point of the
line should have a number
to
it
(or that every length
corresponding
should be capable of measurement), we are forced to extend our system of numbers further by the introduction of what are called irrational

numbers.

We will thus associate an irrational number to every point of


the line which does not correspond to a rational number.
A

method of representing

notation

is

irrational

given in the next article

Def. Real number.


real number.

numbers in the decimal

1-3.

number, rational or

irrational, is called

Theaggregate of rational and irrational number


aggregate of real numbers.

is,

thus, the

Each real number is represented by some point of the numberaxis and each point of the number-axis has some real number,
rational or irrational, corresponding to it.
Or,

we might

OP

length

say, that each real number is the measure of some


and that the aggregate of real numbers is enough to

measure every length.


1-21.
Number and Point. If any number, say x, is represented by a point P, then we usually say that the point P is x.
Thus the terms, number and point, are generally used in an
indistinguishable manner.
1-22.
Closed and open intervals, set a, b be two given numbers
such that #<6. Then the set of numbers x such thata^x^fe is
called a closed interval denoted by the symbol [a, b].

Also the set

of numbers x such that

a<x<b

open interval denoted by the symbol (a,b).


The number b a is referred to as the length of

<.

is

[a, b]

called

an

as also of

*).

Decimal representation of real numbers. Let P be any


1-3.
given point of the number-axis. We now seek to obtain the decimal
representation of the number associated with the point P.

REAL NUMBERS

To start with,
side of O.

we suppose that the

point

on the positive

lies

Let the points corresponding to integers be marked on the


number-axis so that the whole axis is divided into intervals of length
one each.

Now, if P coincides with a point of division, it corresponds to an


In case P falls between two
integer and we need proceed no further.
l, we sub-divide the interval (a, a+l)
points of division, say a, a
The points
into 10 equal parts so that the length of each part is T1T

of division, now, are,


#, tf-f TIP

coincides with any of these points of division, then it corresIn the alternative case, it falls between
a
rational number.
ponds
two points of division, say

If

to

i.e.,

o.a v

where, a l9

is

We

tf.K+1),
one
of
the
any
integers 0,

1, 2, 3,

9.

again sub-divide the interval

a,

+ io'

a+
,

L*

<Ji+i
10

n
J

into 10 equal parts so that the length of each part

is

1/10

2
.

The points of division, now, are


l

fll
fl

io

_!_"

+io

+ io'
4_

The point P

a>

//-u' !

coincide with one of the above points


corresponds to a rational number) or will
between two points of division say

of division
lie

(in

will

cither

which case

it

10*
i.e.,

where a 2

We

is

one of the integers

0, 1, 2, ...... , 9.

again sub-divide this last interval and continue to repeat


the process. After a number of steps, say n, the point P will either
be found to coincide with some point of division (in this case it
corresponds to a rational number) or lie between two points of the

form

DIFFERENTIAL CALCULUS

the distance between which


and smaller as n increases.

is

and which

1/10*

clearly gets smaller

The process can clearly be continued indefinitely.


The successive intervals in which P lies go on shrinking and
clearly close up to the point P.
This point P is then represented by the infinite decimal
Conversely, consider

and construct the


[a,

series

will

any infinite decimal

of intervals

0+1], [a.a& a.a^+l],

Each of these intervals lies within the preceding one


their
lengths go on diminishing and by taking n sufficiently large we can
make the length as near to zero as we like. We thus see that these in;

tervals shrink to a point.


This fact is related to the
perceived aspect of the continuity of a straight line.

intuitively

Thus there is one and only one point common to this


and this is the point represented by the decimal

series

of

intervals

Combining the

results of this article with that of Ex. 2, !!,


see that every decimal, finite or infinite, denotes a number which
rational if the decimal is terminating or recurring and irrational in the

p. 3,
is

we

contrary case.

Let, now,
representing it is

lie

on the negative

side of O.

Then the number

#-#i#2**#i ......

where

a.a^ ...... a n ......


the number representing the point P' on the positive side oft? such
that PP' is bisected at O.

is

Ex.

1.

Calculate the cube root of 2 to three decimal places.

We have

I=*l<2and23 =8>2.
l<3/2<2.

We consider the numbers


1,M,

1'2, ....... 1-9,2,

which divide the interval [1, 2] into 10 equal parts and find two successive nunv
bers such that the cube of the first is <2 and of the second is >2. We find that

Again consider the numbers


1-2, 1*21, 1-22 ......

which divide the interval

[1*2, 1-3] into

1-29, l'3 r

10 equal parts.

REAL NUMBERS

We find

that
(l'25)

8
=l-953125<2and(l-26) -2'000376>2.

l-25<3/2<l-26.
Again, the numbers
1-25, 1-251, 1-752. ....... ,1*259,1-26

divide the interval [1-25, 1-26] into 10 equal parts

We find

that

8
(1-259)=1-995616979<2 and (1'26) =2'000376>2
!-259<3/2<l-26.

Hence
Thus

to three decimal places,

we have

?/2=l-259.

Ex. 2.

Calculate the cube root of 5 to 2 decimal places.

Note. The method described above in Ex. 1 which is indeed very cumbersome, has only been given to illustrate the basic and elementary nature or
the problem. In actual practice, however, other methods involving infinite
series or other limiting processes are employed.

x,

1*4.

The modulus of a

Def.

the modulus of a real number, x, is meant


according as x is positive, negative, or zero.

x or

real number.

the

By

number

The smybol x is used to denote the modulus ofx.


Thus the modulus of a number means the same thing as ite
numerical or absolute value. For example, we have
Notation*

|3

=3;| -3
5-7

=-(-3)=3;

= 7-5
|

=0

=2.

The modulus of the difference between two numbers is the


measure of the distance between the corresponding points on the
number-axis.

We

Some results involving moduli.


now state
useful results involving the moduli of numbers.

a+b |<

1-41.
|

i.e.,

the modulus of the

a
|

b
|

sum of two numbers

sum of their moduli.


The result is almost self-evident.
its truth more clearly, we split it up

is less

than or equal to the

To enable the reader

1.

In this

Let
case,

a,

we
I

e.g.,

and

b have the same

sign.

clearly have

=
b
a +
=
7
+ 3
-7-3 = -7 + -3
a+6
7+3

to see

two cases giving exampfes-

into

of each.

Case

some simple and

,
.

DIFFERENTIAL CALCULUS
Case

Let

II.

In this

case,

we
I

4=

e.g.,

Thus

b have opposite

a,

clearly

have

a+b
7-3

<
<

+
+

7
I

=10.

3
|

a+b < \a\


\

ab

1-42.

b
|

we have

in either case,
I

signs.

\b\

e.g., the modulus of the product of two numbers


duct of their moduli,
e.g-,

4-3

(-4)(-3)

If x, a,

1-43.

=12=
=12=

/,

4
|

-4

-3

|.

equal to the pro-

is

be three numbers such that

<*>

(^)

then

i.e.,

lies

between a

and

a-\-l or that

belongs to the open

interval

The

inequality (A) implies thai the numerical difference between


less than /, so that the point x (which may He to the
right or to the left of a) can, at the most, be at a distance / from the
point a.

a and x must be

Now, from the

d+t

a,

see that this

Fig 3

may

It

if

lies

figure,
I

clearly

and only
and a+l.

possible,

between a

we

if

also be at once seen that

*-!<'

is

is

closed interval
equivalent to saying that, x, belongs to the
[a-l, a+l].
Ex.

1.

//

a-b

We have

b-c

a-c

<m, show

that

</+/w.

= a-b+b
1

<

a-b +

Give the equivalents of the following

2.

notation

a-c

</,

b-c

<l+m.

terms of the modulus

in

(0

-1<*<3. () 2<x<5.

(m) -3<;t<7.

(iv)

/~s<x</+s.

Give the equivalents of the following by doing away with the modu3.
lus notation :
(/)

4.

If

y=

x_2

x
\

<3.

+
l

\x+\\ <2. (7) 0<

(//)

jc

2x,

1
|

then show that

fora;<0
forO<A:<l.

x-\

<2.

11

FUNCTIONS

1*5.
Variables, Functions. We give below some examples to
enable the reader to understand and formulate the notion of a variable and a function.

Ex.

"where

Consider two numbers x and y connected by the relation,

1.

we take only

the positive value of the square root.

Before considering this relation, we observe that there is no


is negative and hence, so far as real numbers are concerned, the square root of a negative number does not
real

number whose square

exist.

Now,

jc

equal to 1. This
the relation

i.e.,

is

is

x 2 is less than or
any number satisfying

positive or zero so long as

the case

if

and only

if

is

when x belongs to the interval [1, I].


If, now, we assign any value to x belonging

1, 1],

to the

interval

then the given equation determines a unique corresponding

value of y.

The symbol x which, in the present case, can take up as its


value any number belonging to the interval [
1, 1], is called the
independent variable and the interval [1, 1] is called its domain o?
variation.

The symbol y which has a value corresponding to each value of


x in the interval [1, 1] is called the dependent variable or & function
1, 1].
of x defined in the interval
["

2.

Consider the two numbers x and y connected by the relation,

Here, the determination of y for x ~2 involves the meaningless


operation of division by zero and, therefore, the relation does not
But for every other
assign any value to y corresponding to jc=2.
value of x the relation does assign a value to y.
Here, x is the independent variable whose domain of variation
consists of the entire aggregate of real numbers excluding the number
2 and y is a function of x defined for this domain of variation of x.
3.
Consider the two
defined by the equations

numbers x and y with


y=^x* when x<0,
y=x when 0<x<l,

y=ljx when x>l.

their

relationship

...(/)

(")
...(iii)

These relations assign a definite value to y corresponding to


every volue of x, although the value of y is not determined by a single

DIFFEBENTIAL CALCULUS

12

formula as in Examples 1 and 2. In order to determine a value of


to a given value of x, we have to select one of the
three equations depending upon the value of x in question. For

y corresponding
instance,

forx=

2,

forx=,
for

y=(

2)

=4, [Equation

(/)

[Equation

(i7)

[Equation

(in)

j=J
^=|

*=3,

Here again, y

is

v 2<0
v 0<J<1

3>1.

a function of x defined for the entire aggregate


9

of real numbers.
This example illustrates an important point that it is not necesthat
sary
only one formula should be used to determine y as a function of x. What is required is simply the existence of a law or laws
which assign a value to y corresponding to each value of x in its
domain of variation.
Let

4.

Here y

is

a function of x defined for the aggregate of positive

integers only.

Let

5.

Here we have a function of x defined


real

for the entire aggregate of

numbers.

It
follows

may

be noticed that the same function can also be defined as

6.

ys=

x when

y=

x when x<0.

Let
yt=z\lq f

when x

y=0 when
t

is

is

a rational number plq

in its lowest

terms r

irrational.

Hence again y is a function of


of
real numbers.
gate

x defined

for the entire aggre-

1*51.
Independent variable and its domain of variation. The
above examples lead us to the following precise definitions of variable
and function. Ifxis a symbol which does not denote any fixed number
but is capable of assuming as its value any one of a set of numbers,,
then x is called a variable and this set of numbers is said to be its
domain of variation.

1*52. Function and its domain of definition. // to each value of


an independent variable x, belonging to its domain of variation, there
corresponds, by any law, or laws, whatsoever, a value of a symbol, y+

FUNCTIONS

13

then y is said to be a function of x defined in the domain of variation of


K, which is then called the domain of definition of the function.

Notation for a function.

1*53.

variable

is

read as the

is

The

fact that

y is a function of

expressed symbolically as

'/'

of x.

If x, be any particular value of x belonging to the domain of


variation of X then the corresponding value of the function is denoted by/(x). Thus, iff(x) be the function considered in Ex. 3,
1*5,
9

p. 11,

then

If functional symbols be required for two or more functions,


usual to replace the latter /in the symbol f(x) by other
letters such as F, G, etc.

then

it is

Some

important types of domains of variation.


Usually the domain of variation of a variable x is an interval
a, b], i.e., x can assume as its value any number greater than or
equal to a and less than or equal to 6, i.e.,
1*6.

Sometimes

it

becomes necessary to distinguish between closed

and open intervals.


If x can take up

as its value any number greater than a or less


than b but neither a nor b i.e., if a<x<b then we say that its
domain of variation is an open interval denoted by (a, b) to distinguish it from [a, b] which denotes a closed interval where x can take
up the values a and b also.
y

We may similarly
[a, b),

have semi-clos3(l or semi-opened intervals

a<x<6

(a, b],

a<x^b

as domains of variation.

We may
bound

also have domains of variation extending


in one or the other directions i.e., the intervals

(00,

b] or

x<6

[a, GO

or

x>a (00
;

oo

or

any

without
x.

Here it
00,00 are no
numbers in any sense whatsoever. Yet, in the following pages they
will be used in various ways (but, of .course never as numbers) and
in each case it will be explicitly mentioned as to what they stand for.
Here, for example, the symbol ( 00, b) denotes the domain of variation of a variable which can take up as its value any number less
than or equal to b.
should be noted that the symbols

Similar meanings have been assigned to the symbols


(0, GO

Constants.
called a constant.

),

(00

oo

).

symbol which denotes a certain fixed number

is

14

DIFFERENTIAL CALCULUS
It has

like a, h, c
like JC, y, z

Note.

become customary to use earlier letters of the alp}ifl,bet r


a, (J, y v as symbols for constants and the latter letters
w, v, w*as symbols for variables.
points about the definition of a function should be

The following

carefully noted

function need not be necessarily defined by a formula or formulae so


1.
that the value of the function corresponding to any given value of the independent variable is given by substitution. All that is necessary is that some rule or
set of rules be given which prescribe a value of the function for every value of
the independent variable which belongs to the domain of definition of the function. [Refer Ex. 6, page 12.]
It is

2.

not necessary that there should be a single formula or rule for the
[Refer Ex. 3, page 11.]

whole domain of definition of the function.

Ex.

is

Show

1*

that the

domain of definition o f the function

the open interval (I, 2).

For

and

x>2

x=l

and 2 the denominator becomes zero. Also for x<l


the expression (1 -Jt) (x 2) under the radical sign becomes

negative.
is not defined for *<; 1 and x^2.
For x>l
the expression under the radical sign is positive so that a
Hence the function is defined
value of the function is determinable.

Thus the function

and

<2

in^heopen

interval (1, 2).

Show

2.

that the

the closed interval

Show

3.

are

(0, oo)

and

that the

domain of definition of the "function

^[(1

x)

(A*

2)]

is

[1, 2].

oo

domain of

definition of the functions

0) respectively.

Obtain the domains of definition of the functions

4.

(0 ^(2x+l)
1-7.

(//)

1/U + cos

A-)

(iii)

V(l-f-2 sin x).

Graphical representation of functions.

Let us consider a function

y=f(x)
defined in an interval

...

[a, b].

(/)

represent it graphically, we take two straight lines X'OX


at right angles to each other as in Plane Analytical
fwid Y'OY
Geometry* These are the two co-ordinate axes.

To

We take O

as origin for both the axes and select unit intervals


Also as usual, OX,
lengths).
arq taken as positive directions on the two axes.

on OX,

OF (usually of the same

To any number x

corresponds a point

OY

M on

Jf-axis such that

FUNCTIONS

15

To the corresponding number y, avS determined from


on K-axis such that
corresponds a point

said

Completing the rectangle OMPN,


to correspond to the pair of

numbers x,
Thus

we obtain a
y*

there

which

is

y.

to every number x belonging to the interval [<?, b], there

corresponds a number y determined


by the functional equation yc=f(x)

and to

point

(/),

this pair of

corresponds a point
above.

y
o

numbers,

x, y
as obtained

Fi g% 4

The

totality of these points, obtained by giving different values,


to x, is said to be the graph of the function /(;c) and y=f(x) is said to"
be the equation of the graph.

1.

page 11,

Examples
The graph of the function considered

in

Ex.

3,

1'5,

is

Fig. 5
2.

The graph

of >>

excluding the point P(l,

1) is

(jc

the straight line y=x-\-l

2).

Fig. 6
3.

The graph of y=*x

consists of a discrete set of points(1, 1), (2, 2), (3, 6), (4, 24), etc.
\

DrFFEBENTIAL CALCULUS

The graph of the function

4.

x,

when

1,

when
x when

is

as given.

Fig. 7

Draw the graph of


root o/x 2

5.

square

the function

which denotes the positive

As

*v/x

graph (Fig.

x according

=jc or

8) of

V*2

as

is

positive or negative, the

the graph of the function /(x) where,

*s

f x, when

=H
x,

xi

when x<0.

Fig. 8

Fig. 9

The student should compare the graph


graph

(Fig. 8) of

2
\/* with the

(Fig. 9) of x.

The reader may


6.

Draw

see that

the graph of

>-

We

have

x+lx=l
x+lx=l

2x when x<0,
when 0<x
{ x+x ^l=2x 1 when
Thus we have the graph as
drawn
The graph consists of parts of
:

3 straight lines
^

,,

corresponding to the intervals

^-oo,0].
Fig. 10

[0, !],[!,

x).

FUNCTIONS

Dawn

7.

17

the graph of

[*],
where

[x] denotes the greatest integer not greater

We

have

y=4

for

1,

1^2,

The value of y

than x.

<*<!,
<x<2,

f 0, for
1

<ix<3 and

for 2

so on.

x can

for negative values of

also be similarly

given.

The right-hand end-point of each segment of the line


point of the graph.
Yk

is

not a

Fig. 11.

Exercises
1

Draw the

<0/(*)

graphs of the following functions


1,

when A-<CO

-l,when.v>0
x, when OjcCjc^i
2-*, when^v^i

2
f A
A- ,

\
2.
x.x

Draw

^v,

when
wften A<TO
x^u
when A>0;

<)/(*)

,^

,.^ f

(")/(*)

( v /)/(A-)

,.^

00 x

(/)

The positive value of the square root


3.

(0
4.

W
where

[*]

is

Draw the graphs of the functions


(//)
|*|.
|*| + |x + l|.
Draw the graphs of the functions
+ [*fl]
()

w
when
when
*

{!-*,
,

+
,

when x = 4
1
A, when
I/*. when ^<
0, when A =
.

1,

^
f

the graphs of the following functions

^* 2

^
^

-l/jc.whe

^(x-\)

to be taken in each case.

(///)

MV

denotes the greatest integer not greater than

x.

2|*-1

+3|

CHAPTER

II

SOME IMPORTANT CLASSES OF FUNCTIONS AND


THEIR GRAPHS
This chapter will deal with the graphs and some
Introduction.
simple properties of the elementary functions
x n a x loga x
,

tan x, cot x, sec x, cosec x


1
1
sin~ x, cc5s~ x, tan"~ 3 x, cot~ 1 x, sec^x, cosec- 3*.
The logarithmic function is inverse of the exponential just as the
sin x,

cos x,

inverse trigonometric functions are inverses of the corresponding


trigonometric functions. The trigonometric functions being periodic,
the inverse trigonometric are multiple-valued and special care has,
therefore, to be taken to define them so as to introduce them as singlevalued.

Graphical representation of the function

2-1.

y=x*

n being any integer, positive or negative.


We have, here, really to discuss a class of functions obtained by
giving different integral values to n.
It will be seen that, from the point of view of graphs, the whole
of this class of functions divides itself into four sub-classes a&
follows

(i)

integer

odd

when n is a
(Hi) when n

positive even integer


is

(//)

a negative even integer

when n is a positive odd


(iv) when n is a negative

integer.

The functions belonging to the same sub -class will be seen to


have graphs similar in general outlines and differing only in details.
Each of these four cases will now be taken up one by one.
Let n be a positive even integer.
2-11.

The following

are,

obviously, the properties of the graph


n
of
whatever positive even integral
n
value,
may have.

y=x

^=0, when x=0


y=l, when x=l
>>=!, when x=

(i)

The

graph,

1.

therefore,

passes

through the points,

(0,0),

(it,

a positive even integer)


Fig. 12.

(1,1), A'

(-1,

when x

1).

is posiy
(ii)
positive
tive or negative. Thus no point on the
graph lies in the third or the fourth

is

quadrant.
18

19

GRAPHS

We have

(wi)

same value of y corresponds to two equal and opposite


The graph is, therefore, symmatrical about the .y-axis.

so that the

values of x.

The variable y

(iv)

larger numerically.

gets larger

larger, as x gets larger and


made as large as we like by
The graph is, therefore, not

and

Moreover, y can be

taking x sufficiently large numerically.


closed.

for positive even integral value of n


in
as
Fig. 12. page. 18.
general outlines,
given

The graph of y=x*

The following

are the properties


integral value n may have.

when

1,

The graph,

of

y=xn

whatever positive

whenx^O;
x=l

^=0,

(i)

in

Let n be a positive odd integer.

2-12.

odd

is,

therefore, passes

through the points


0(0,0),

(//')

,4 (I, I),

is

according as x

,4'(-l,-l)

positive or negative
is positive or nega-

tive.
lies

Thus, no point on the graph


the second or the fourth

in

(n,

odd integer)

a positive

quadrant.

13.

Fig

The numerical value of y

(///)

increases with an

increase in the

numerical value of x.
Also, the numerical value of y can be

by taking x
2-13.

made

we

as large as

like

sufficiently large numerically.

Let n be a negative even integer, say,

m> (w>0).

Here,

xn

n,

=x

(n,

a negative even integer)


Fig. 14

~~

xm

The following are the


x n whatever
properties of y
negative even integral value,

\U,D

1
_______

may have
(i)

x=Q

Determination

of y

involves the meaningless operation of division by


and so y is not defined for
for

20

DIFFERENTIAL CALCULUS

y=l when x=l or 1, so that


A (I, l)andX'(-l, 1).

(fl)

the graph passes through

the two points


(itt)

Thus no
positive whether x be positive or negative.
lies in the third or in the fourth quadrant.

is

point on the graph


(IF)

We have

so that the same value of y corresponds to the two equal and


opposite values of x. The graph is, therefore, symmetrical about
jr-aaris.

(p)

decreases.

as

As

x, starting

from

Alsojy can be

we like) by

increases, x also increases so that y


as small as we like (i.e., as near zero

1,

made

taking x sufficiently large.

m
Again, as x, starting from 1, approaches zero, x decreases so
that y increases. Also y can be made as large as we like by taking x
sufficiently near 0.

The

variation of y for negative values of

down by symmetry.
2-14.

x may

be,

now, put

Let n be a negative odd integer.

The statement of the various


properties for this case

is left

The graph only

reader.

is

to the

shown

here. (Fig. 15)

VJ)

Before considering the


graphs of other functions, we allow
2*2.

some digression in this


to facilitate some of the

ourselves
article

later considerations.

2-21.

+. a negat^ odd

integer^

Monotonic

functions.

Increasing or decreasing functions.

Fig. 15.

Det L A function y=*f(x)

is called a monotonically increasing


that
such
interval if
throughout the interval a larger
function in
value of x gives a large value of y, i.e., an increase in the value of x
always causes an increase in the value of y.

an

Def. 2.

it

is

A function is

if His such that an

called a monotonically decreasing function,


increase in the value of x always causes a decrease

in the value of the function.


Thus, if *i and x a axe any two numbers in an interval such that

21

GBAPHS
then for a monotonically increasing function in the interval

and

for

a monotonically decreasing function in the interval

A
ing

is

function which

called a

Show

Ex.

is

mono tonic
that for a

either monotonically increasing or


function.

monotonic function /(x)

in

an interval

decreas-

(a, 6),

the

fraction

keeps the same sign for any pair of different numbers, x l9 x, of


Illustration.

xn

Looking

at

the graphs

in

(a, 6).

we

2*1,

notice that

is

(/) monotonically decreasing in the interval (


tonically increasing in the interval [0, oo] when n
integer

00, 0)
is

and mono-

a positive even

(//')

when n

monotonically increasing in the intervals


a positive odd. integer.

oo, 0], [0, oo ]

is

(in) monotonically increasing in the interval (


00,0) and monotonically decreasing in the interval (0, oo) when n is a negative even
integer.
(iv)

when n

.monotonically decreasing in the intervals


a negative odd integer.

(00,

0), (0, oo)

is

2-22.

Inverse functions.

Let

y=Ax)

...(1)

be a given function of x and suppose that we can solve this equation


for x in terms of y so that we may write x as a function of y, say

x=*(y).

Then

<p(y) is

.-.(2)

called the inverse of f(x).

This process of defining and determining the inverse of a given


function may be accompanied with difficulties and complications as
illustrated

below

be
(i) The functional equation (1) may not always
as a function of y as, for instance, the case

solvable for

(1) may not always determine a


(ii) The functional equation
unique value of y in terms of x as, for instance, in the case of the
functional equation

j = l+x

which, on solving for x, gives

,
,

a
,

...(3)

22

DIFFERENTIAL CALCULUS

so that to each value of y there correspond two values


situation is explained by the fact that in (3) two values
are equal in absolute value but opposite in sign give rise
value of y and so, conversely, to a given value of y there
ally correspond two values of x which gave rise to it.

of x. This
of x which
to the same
must natur-

We cannot, therefore, look upon, x, as a function of, y, in the


usual sense, for, according to our notion of a function there must
correspond just one value of, jc, to a value of, y.
In view of these difficulties, it is worthwhile to have a simple
which may enable us to determine whether a given function
y=f(x) admits of an inverse function or not.
test

If a function y=f(x) is such that it increases monotonically in an


interval [a, b] and takes up every value between its smallest value f(d)
and its greatest value f(b), then it is clear that to each value of y

between f(a) and

/(/>),

of x which gave rise to

there corresponds one and only one value


so that *is a function of y defined in the

it,

interval [f(d)> f(b)}.

Similar conclusion

is

reached

easily

if

y decreases monotoni-

cally.

Illustrations of this general rule will

appears in the following

articles.

2*3.

To draw

the graph of
JL

y=x*
n being any positive or negative

integer.

From an examination of the graph


even otherwise, we see that y=x n
(i) is

y~x n

drawn

monotonic and positive in the interval


may have

integral value n
(11)

of

takes

up every

in

(0, oo

21, or
whatever

positive value.

Thus, from 2*2 or otherwise, we see that to every positive


value of y there corresponds one and only one positive value of x
such that

y=x n

or

xy^

Thus

determines

jc

as a function of y in the interval

always positive.

(0, oo

x being

also

OBAPHS

23

The part of the graph of y~x n lying on the


,

also the

graph of

x=y

first

quadrant,

is

n
.

y*

o
when n

is

when n

a positive integer.

is

a negative

integer.

Fig. 17.

Fig. 16.
i

To draw the graph of y~x H when x is independent and y


dependent, we have to change the point (h, k) in
t
i

j,

to the point

in the line

(k, h) i.e., find

y=x.

This

is

the reflection of the curve

illustrated in the figure given below.

Cor. x p u means (.x^) 1 /? where the


root is to be taken positively and x is also
positive.
Note. It may be particularly noted that
in order that x lfn may have one and only one
value, we have to restrict x to positive values
n
only and x l l is then to be taken to mean the
positive nth root of x.
It

is

nth root of x

not enough to say that x lln is an


we must say the positive wth root
;

o/the positive number x.


For, if n be even and x be positive, then x
has two nth roots, one positive and the other
negative, so that the nih root is not unique and
if n be even and x be negative, then x has no nth
root so that x lln does not exist.
Thus x and x 1 /" are both positive.

X
when n

a positive integer.
Fig. 18.

is

The exponential function a*. The meaning of a* when a is


and
x is any rational number, is already known to the
positive
2-4.

2*3 above).
student. (It has also been considered in
To give the
rigorous meaning of a* when the index, x is irrational is beyond the
scope of this book. However, to obtain some idea of the meaning of
a? in this case, we proceed as follows
:

24

DIFFERENTIAL CALCULUS

We

find points on the graph of y=a* corresponding to the


rational values of x.
then join them so as to determine a continuous curve. The ordinate of the point of this curve corresponding
to any irrational value of x as its abscissa, is then taken to be the
value of a* for that particular value.

We

Note. The definition of a* when x is irrational, as given here, is incomplete in as much as it assumes that what we have done is actually possible and
that also uniquely. Moreover it gives no precise analytical way of representing
a* as a decimal expression.
z
of

y=a

Graph

In order to draw the graph of


y*=<f>

we note

its

following properties

2-41.

The function a*

(i)

Leta>l.
always positive whether x b3 positive or

is

negative.
(//)

large as
(ill)

(iv)

It increases monotonically as x increases and can be made as


we like by taking x sufficiently large.
tf=l so that the point (0, 1) lies on the graph.
Since a*~\\a~* we see that a* is positively very small when
x is negatively very large and can
be made as near zero as we like
provided we give to x a negative
',

value which

is sufficiently

large nu-

merically.

With the help of these facts,


we can draw the graph which is
as shown in (Fig. 19).
Fig. 19.

2-42.
(i)

LetQ<a<\.

The function a*

is

always positive whether x be positive

or negative.
as x increases and can be
(//) It decreases monotonically
as near zero as we like by taking x
sufficiently large.
(ill)

fl=l. Therefore
on the graph.

the

point

(0, 1) lies
(iv)

Since a x ~\\cr x

we

see that

positively very large when x is


negatively very large/ Also it can be

cf

is

made

as large as

we like if we

a negative value which

is

give to

sufficiently

large numerically.

Thus we have the graph


in (Fig. 20).

as

drawn

Fig. 20.

made

GRAPHS

25

Note 1. In the case of the function a* it is the exponent x which is a


variable and the base is a constant. This is the justification for the name
n the base is a variable
"Exponential function" given to it. For the function x
and the exponent is a constant.
,

Note

2.

Ex.

Draw

a* never vanishes and

1
2 /*

(i)

1
2 /*"

</<)

Graph of

2-5.

is

always positive

the graphs of the functions


1

(iiflZ-

WZ-

/*

/*'.

the logarithmic function

y-log a x
a,

x being any

We

positive numbers.
know that

y=a*
can be written as

We have seen in 2'4 that y~a* is monotonic and takes up


oo
every positive value as x increases taking 'all values from
to

-f-o>

to any positive value of y there corresponds one and only


one value of x.
In the figure of
2*4, we take y as independent variable and
Thus we see that
that
it
to GO
suppose
continuously varies from
x monotonically increases from
x to oo if
(/')
.x
x if
oo
to
from
decreases
(//')
monotonically
x=.0
for
Also,
y~l.

Thus

Considering the functional equation

y=log a x,

we

see that as the independent variable x varies from


oo to
dependent variable y monotonically increases from

and monotonically decreases from


Also, y=Q for x = l.

We

oo

to

oo

if

if

thus have the graphs as drawn.

Fig. 22,

Fig. 21.

Note.

to oo the
oo

The graph of>'=log 3 ;t

is

the reflection

tfy=a*

in the line

.y

26

DIFFERENTIAL CALCULUS

Some important properties of lo&x.

We may,

is

now, note the following important properties of

(i) log a x is defined for positive values of


also positive.
(ii)

If

a>l,

x only and the base a

log^x can be made as large as


can be made as small as

sufficiently large and


sufficiently near 0,

we
we

like

like

For #<1, we have similar statements with obvious


log a

(i/O

alterations.

1=0.

When fl>l, we
When 0<1, we
2*6.

by taking x
by taking x

have, log a x>0, if

log a x<0, if

x<l

have,

loga x>0, if

x<l.

x>l, and
log a x<0, if x>l, and

Trigonometric functions.

It will be

assumed that the student is already familiar with the


and the nature of variations of the fundamental
functions
sin x, cos x, tan x, cot x, sec x and cosec x.
trigonometrical
The most important property of those functions is their periodic
character, the period for sin x, cos x, sec x, cosec x being 2?r and that
for tan x, cot x being TT.
definitions, properties

We

only produce their graphs and restate some of their

will

important and
2-61.

well

y-sin

known

properties here.

x.

,1\

Fig. 23.
(i)

It is defined for

every value of x.

1 to 1 as x increases from
from
to
1 as x increases
1 to
from
it
decreases
7r/2
7T/2
monotonically
from Tr/2 to 3?r2, and so on.
(ii)

It increases monotonically
;

(Hi)

have.

<

sin

x <1,

i.e.,

sin
|

x
[

<1, whatever

value x

may

27

GRAPHS

Fig. 24.

It is defined for

(i)

It decreases

(11)

to

TT

and

2-63.

mono ton ically from

to

as

increases from

so on.

cos

(/)

every value of x.

y^tan

<1

whatever value x

may

have.

x.

Fig. 25.
(i)

It is defined for all values of

+ l)7r/2,

x excepting

where n is any integer, positive or negative. It fact


be recalled that for each of these angles the base of the rightand so the definition of
angled trangle which defines tan x becomes
tan jc, as being the ratio of height to base, involves division by
which is a meaningless operation.
(2
will

i.e.,

it

28

DIFFERENTIAL CALCULUS
(if)

from

It increases monotonically

7T/2 to 7T/2

increases from

can

(Hi) It

oo

to

bfe

made

positively as

x<7r/2 and

x>
yr=cot

7T/2

and

large as

oo

to

increases

+00 as

large as well like provided

sufficiently near to it

and can be made negatively as

+00 as x

from

it

we take

2-64.

from

increases monotonically
to
7T/2
37T/2, and so on.
;

we

provided we take

like

sufficiently near to

it.

x.

Fig. 26.

It

(i)

is

defined for
4?r,

all

STT,

values of x excepting
27r,

TT,

0,

TT, 2?r, STT, 47T,

i.e., nir, where w has any integral value, since for each of these angles
the height of the right-angled triangle which defines cot x is zero and
the definition of cot x being the ratio of base to height involves

division

by

zero.

It decreases monotonically from oo to


to
it decreases monotonically from oo to
TT
/rom
from TT to 2?r and so on.

()

<x>

as x increases
as x increases

GO

(iii)

It can

be made positively as large as we

like

provided

we take

x>0and

sufficiently near to it

and can be made negatively as large as we like provided we take


x<0 and sufficiently near to it.

GRAPHS

29

2-65.

Fig.

It is defined for all values of

A:

excepting

i.e.,
(2n
l) ?r/2, where n has any integral value, since for each of
these angles, the base of the right angled triangle which defines
sec x is zero and as such the definition of sec x involves division

by

zero.
(11)

sec
|

^1

whatever value x

may have.

monotonically from

1 to oo when x increases
to Tr/2
it increases
oo to
1 when x
monotonically from
increases from ?r/2 to TT, and so on.
(zv) It can be made positively as large as we like by taking
x<7r/2 and sufficiently near to it
(ill)

It increases

from

and can be made negatively


x>ir/2 and
2-66.

y=cosec

as large as

we

like

by taking

sufficiently near to

x.

Fig. 28.

it.

DIFFBKENTIAL CALCULUS

30

It is defined for all values of

(i)

"STr,

-27T,

TT,"

x excepting

0,

TT,

27T, OTT,

when n has any integral value.


cosec x
^ 1 whatever value x may

i.e., nit,
(ii)

(lii)

have.

oo
from -1 to
decreases monotonically from oo

It decreases monotonically

from

7T/2

creases

from

to

it

as~""x

to]*l

increases
as jc in-

to 7T/2.

(iv) It can

be made positively as large as we like by taking

x>0

and

sufficiently near to it

and can be made negatively as large as we like by taking


x<0 and sufficiently near to it.
Inverse trigonometrical functions.
2-7.
metrical functions

The inverse

trigono-

sin- 1 *, cos- 1 .*, tan- 1 *, cot~ l jc, sec- 1 *, cosec- 1 x

are generally defined as the inverse of the corresponding trigonometrical functions. For instance, sin" 1 * is defined as the angle whose
sine is x. The definition, as it stands, is incomplete and ambiguous
as will now be seen.

We

consider the functional equation

x== sin y

where y

is

the independent and x the dependent

...

(/)

variable.

Now, to each value of y there corresponds just one value of x


in (i). On the other hand, the same value of x corresponds to an
unlimited number of values of the angle so that to any given value
1 and I there corresponds an unlimited number of
of x between
values of the angle y whose sine is x.
Thus sin" 1 A*, as defined above,
is not unique.
t

The same remark

applies to the remaining

trigonometric

func-

tions also.

We now

proceed to modify the definitions of the inverse


functions
so as to remove the ambiguity referred to
trigonometrical
here.

271.

y-sin^x.

Consider the functional equation


;c=sin>>.

We know that as y increases from Tr/2 to w/2, then x increases


1 and 1, so that to
monotonically, taking up every value between
1 and 1 there
each value of x between
corresponds one and only one
Thus there is one and only
value of y lying between
?r/2 and Tr/2.
and
a given sine.
with
one angle lying between
7T/2
ir/2

GRAPHS

sin~ l x

between
5/ne

the angle lying

is

Tr/2

define sin- 1 * as follows

we

Accordingly

and

Tr/2,

w^ose

x.

To draw the graph of

y= sin- 1 *,
we note that

monotoni(i) y increases
to
from
?r/2 as x
cally
Tr/2
1 to 1.
increases from
(if)

sin- 1

(fff)

We,

__

0=0,

sin- 1 *

the interval

defined

is
1, 1]

in

only.

Fig. 29.

thus, have the graph as drawn.

We

know that if x^siny, then ovaries monotonically taking up


Note 1.
1 and 1 as
y increases from w/2 to 3w/2, 3^/2 to 5w/2, and
every value between
soon. Thus the definition could also have been equally suitably modified by
restricting sin-'a to any of the intervals (w/2, 3*/2), (3*/2, fiw/2) instead of
What we have done here is, however, more usual.
(
7T/2, tf/2).
Note
2*72.

2.

The graph of ^sin-

y=cos""

1
.*?

is

the reflection of .y

sin

x in the line

x.

Consider the functional equation


cos y.
increases from

A:

We know

to TT, then x decreases


that as y
1. Thus, there
1 and
value
between
monotonically taking up every
and TT, with a given cosine.
is one and only on? an^lc, lyiiva; botwean
Accordingly we define cos~ x as follows
and TT, whose cosine is x.
cos-*x is the angle lying between
!

To draw the graph of

we note that
(/)

from

TT

to

decreases
monotonically
as x increases from
1

(/I)

Fig. 30,

COS"

cos-Ml) = 0.

1
(

to

1.

1)=7T,

COS^O^TT^

1
1, 1] only.
defined in the interval [
(MI) cos- ^ is
varies
taking up every
mondtonically,
that
x^cosy
Note. We know
w to 2w, 2* to 3* and so .on. Thus
value between -1 and 1 as y increases from
modified ty restricting
the definition could also have been equally suitably
instead of (On).
1
etc.,
3*]
2w],
[2*,
x
[,
the
intervals,
of
to
cosany

DIFFERENTIAL CALCULUS

32
2-73.

We

y=tair*x.

consider the functional equation

x=tan

We know that

y.

y increases from
7T/2 to 7T/2, then x increases
00 to oo taking up every value.
Thus there is

as

monotonically from
one and only one angle between
7T/2 and 7T/2 with a given tangent.
1
Accordingly we have the following definition of tan- x
l
tari~ x is the angle, lying between
whose tangent
TT/ 2 and Tr/2,
:

isx.

Fig. 31

To draw the graph


note that
creases from

(Fig. 31)

of

increases monotouically
oo to oo and then tan*" 1

from

?r/2

to Tr/2 as

in-

0=0.

2-74.

y==cot- x.
Consider the functional equation

x=cot

y.

We know that

to TT, then x decreases


as y increases from
from
oo
to
4-a>
monotonically
taking up every real value. Thus,
and TT with given cotanthere is one and only one angle between
gent.
In view of this, we define cot" 1 x as follows
:

cot~\

is

the angle, lying between

o
Fig. 32

and

TT,

whose cotangent

is x.

GRAPHS
To draw the graph

we note
from

that
oo

(Fig. 32)

of

y=COt~ l

X,

y decreases monotonically from


and cot" 1 0=7r/2.

TT

to

as x increases

to oo

2-75.

y^sec^x.

Consider the functional equation

X=9eC y.

We know

to ?r/2, then x increases


from 7T/2 to TT, then
increases
monotonically from 1 to oo also as y
X increases monotonically from
oo to
1.

that as y increases from


;

is one and only one value of the angle, lying between


1
whose secant is any given number, not lying between
l
x
of
sec
The following is, thus, the precise definition

Thus, there

and

and

1.

TT,

and

x is the angle, lying between


To draw the graph of

sec- 1

y=secr*

we note that y increases from


y increases from Tr/2 to TT as x
sec- 1

TT,

whose'$eant

is x.

x,

to ?r/2 as x increases from


1
oo to
increases from

1=0, and

sec- 1

to oo

and

Also

l)=7r.

Ylk

Fig. 33

2-76.

y=cosec-

x.

Consider the functional equation

x=cosec

We know

y.

that asj; increases from

7T/2

to 0, then

x decreases

and as y increases from to ?r/2, it


monotonically from
Thus there is one and only one
decreases monotonically from oo to 1
value of the angle lying between
?r/2 and Tr/2 having a given
1

to

oo

cosecant.

We thus
1

cosec"
cosecant is x.

To draw

is

say that

the

angle,

the graph of

lying

between

?r/2

and

*r/2,

whose

DIFFERENTIAL CALCULUS

34

we note that y decreases from


any y decreases from ?r/2 to
J

to
?r/2 as x increases from
as x increases from 1 to oo.

oo

to

Fig. 34
1
1
1
Ex. What other definitions of tan- *, cot- *, sec- * and cosec- 1* would
have been equally suitable.
1
cot- 1 *, sec- 1 * and
Note. The graphs of j^s in- 1*, cos- *, tan- 1*,
1
sosec- * are the reflections of y
*, cos *, tan *, cot *, sec * and cosec *

fespectively in the line

yx.

Function of a Function. The notion of & function offunction


will be introduced by means of examples.
2-8.

Ex.

1,

Consider the two equations

>>=sinw.

...(2)

To any given value of *, corresponds a value of u as determined from (1) and to this, value of w, corresponds a value ofy as
determined from (2) so that we see that y is a function of u which isagain a function of x, i.e., y is a function of a function of x.
From a slightly different point of view, we notice that since the
two equations (I) and (2) associate a value of y to any given value
3fx, they determine y as a function of x defined for the entire aggregate of real numbers.

Combining the two equations, we get


2
j>=8in x
which defines y directly as a function of x and not through the intermediate variable w.
Ex.

2.

Consider the two equations

w=sin

x,

.(!)

y=\ogu.

...(2)

The equation

(1) determines a value of u corresponding to every


x.
value
of
The
given
equation (2) then associates a value of y to
But we know that u is positive
this value of u in case it is positive.
if and only if x lies in the open intervals
............... (~47T,

~37T), (-27T, -7T),

(0,

W)

(2*, 3*}

..,(3)

35

FUNCTION OF A FUNCTION

Thus these two equations


domain of variation of x is the

define y as a function of
set of intervals (3).

x where

'

the.

General consideration.

Let

and
be two functional equations such that/(x)
[a, b]

and

<f>

Let, further, each value of f(x)

defined in the interval

is

(u) in [c, d].


lie

in

[c, d\.

Clearly, then, these two equations determine


defined for the interval [a, b}.
Classification of functions.
is either

2-9.

Any

y as a function of

Algebraic and Transcendental.

given function

Algebraic or

(i)

A function

(ii)

transcendental.

said to be algebraic if it arises by performing upon the


constants a finite number of operations
of addition, subtraction, multiplication, division and root extraction.
Thus,
variable

is

x and any number of

y=x+7jc+3, y=(7x 2 +2)/(3;c4 + 5;c 2

y^V(*+ 3 \/* 8 + Vx)

),

are algebraic functions.

A function

is

said to be transcendental if

it is

not algebraic.

The exponential, logarithmic, trigonometric and inverse trigonometric functions are all transcendental functions.

Two

particular cases of algebraic functions. There are


ally important types of algebraic functions, namely,

(ii)

Polynomials
Rational functions.

function of the type

(/)

where #
is

two

speci-

#,, .........

a m are constants and

is

a positive integer,

called a polynomial in x.

function which appears as a quotient of two polynomials

such as

a Qx m +al x m -*+ ...+a m ^x +a


is

called a rational function of x.

A rational function of x is defined for every value of x excludfor which the denominator vanishes.
those
ing

DIFFERENTIAL CALCULUS

36

Exercises

Write down the values of

1.

2.

For what domains of values of the independent variable are the following,

tan-i(-l),

(11)

[Sol.

sin-Nx

(/)

this will

O'O

is

log sin

(iv)

defined only for those values of x for which

and only

if,

open

if,

defined in the interval

x is defined only
sin

in the

2,

log a sinx.
is

be the case

50 that sin-Nx

(///)

(//)

I.*.,

cos- 1 (~i).

sin-H,

functions defined

and

sec~i

(/)

[0, 1].

for the values of

for

which

x>0

intervals
(0, *), (2n, 3*), ......

or generally in (2nn,
(ill) log sin-i*.
(v)

log

2/i-f-l *)

(x-H)/(x-l)

(v/0 (sin x)*.


2

(ix)

log [x+>J(x

1)

n being any integer.

].

Find out the intervals in


3.
cally increasing or decreasing

+*)/(! -x)

(iv)

log

(v/)

log (tan-'x

(1

(viii)

(Iog 6 x)^.

(x)

x tan-ix.

which the following functions are monotoni-

(02*.
4.

(//)

(i)*-

(ill)

1 ''*'

(iv)

Distinguish between the two functions


1

x,

G)

4 l/x2

'

(v)

2sinx.

CHAPTER

III

CONTINUITY AND LIMIT


Introduction. The statement that a function of x is defined in a certain
interval means that to each value of x belonging to the interval there corresvalue
ponds a value of the function. The value of the function for any particular
value of x
of x may be quite independent of the value of the function for another
and no relationship in the various values of a function is implied in the defini-

tion of a function as such.

so that
Thus, if *!, x a are any two values of the independent variable
then
the
function
of
)-/(*i)
2
values
/(*
fix),
/(*i) /(*) are the corresponding
may be large even though x a --x,. i? small. It is because the value /(x,)
value /(x^ of the
assigned to the function for x=x 2 is quite independent of the
I

x=x
We now propose

function for

change

t.

Xg-Xi

relative to the
/(x? )-/(x,)
study the change
of
and
of
discontinuity
notion
the
continuity
by introducing

to

a function.
In the next article, we first analyse the intuitive notion of continuity that
in the form of
possess and then state its precise analytical meaning

we already

a definition.

31. Continuity of a function at a point. Intuitively, continuous


variation of a variable implies absence of sudden changes while it
varies so that in order to arrive at a suitable definition of continuity,
we have to examine the precise meaning of this implication in its
relation to a function /(x) which is defined in any interval [a, b]. Let,
be any point of this interval.
The function f(x) will vary continuously at x=c if, as x changes
from c to either side of it, the change in the value of the function is
not sudden, i.e., the change in the value of the function is small if
only the change in the value of x is also small.

r,

We
Here,

h,

consider a value c-f-/* of x belonging to the interval [tf, b].


is the change n the independent variable x, may be

which

positive or negative.
f(c), is

Then,/(c+A)

the corresponding change in the depenmay be positive or negative.

dent variable /(x), which, again,

we

For continuity,
numerically small,

can
c _|_/j)_y( c )
small.
sufficiently
|yjr

h
be

if

require

is

c+h)f(c)

that

small.

numerically
as small as

made

we

This
like

be

should

means that

by taking

h
|

The precise analytical definition of continuity should not involve the US6
and
of the word small, whose meaning is definite, as there exists no definite
absolute standard of smallness. Such a definition would now be given.

To be more

precise,

c,

finally say that


c
continuous at

x=

corresponding to any
number S
arbitrarily assigned, there exists a positive

function /(x)

positive number,
such that

we

is

\f(c+h)-f(c)
37

if,

DIFFERENTIAL CALCULUS

38
for

all values of, h, such that

This means that/(c+A)


values of h lying between

Alternatively, replacing
is

f(x)

continuous at

around c such

that,

for

lies

and

between /(c)

c+h by

x=c,

all values

and /(c)+c

for all

8.

we can say that

x,

if there exists an interval (c


ofx in this interval, we have

8,

c-f 8)

being any positive number arbitrarily assigned.

Meaning of continuity explained graphically. Draw the graph of


the function /(x) and consider the point P[c,/(c)] on it.

Draw

the lines

which

lie

the point

on

different sides of
are parallel to

P and

x-axis.

Here, e is any arbitrarily


assigned positive number and
measures the degree of closeness of the lines (1) from each
other.

The continuity of /(x)

x=c,

then,

should be

for

requires that we
able to draw two

lines

g. 35.

x=c

8,

x=c+S,

...(2)

which are parallel to >>-axis and which lie around the line x=c, such
that every point of the graph between the two lines (2) lies also
between the two lines (1).
311. Continuity of a function in an interval. In the last
articles,

at

we

dealt with the definition of continuity of a function /(x)


now extend this definition
its interval of definition.

We

a point of

to continuity in an interval and say that

A function /(x)

is

continuous in an interval

[a, &], if it is

continuous

at every point thereof.

Discontinuity. A function /(x) which


said to be discontinuous for

is

not continuous for

x=c

is

xc.

The notion of continuity and discontinuity


ted by means of some simple examples.
Examples
1.

is

Show

continuous at

that

x=l.

will

now be

illustra-

39

CONTINUITY AND LIMIT


Here,
-and

We

will

now attempt

to

make the numerical

value of this

difference smaller than


(/)

any preassigned positive number, say '001.


Let x>\, so that 3(# -1) is positive and is, therefore, the

numerical value off(x)

/(I).

Now
|/(*)-/(l)| =3(x-l)<-001,
if

x-KOOl/3
ie., if

x<l+-001/3=*l-0003.
Let JC<1, so that 3(x
value of/(jc)-/3;i) is 3(L-x).

1) is

(11)

-..(0

negative and the numerical

Now

/(*)-/(!)

=3(l-*)<-001,

if

l-x<

001/3,

'

i.e., if

l~-001/3<x,
or

Combining

(/)

and
|

(//'),

we

/(*)-/(!)

seo that
|

<-001,

for all values of x such that


1

<

-0003

x <1 +-0003.

test of continuity for x~l is thus satisfied for the particular value -001 of e.
may similarly show that the test is true for
the other particular values of e also.

The

We

The complete argument


Let

however, as follows
be any positive number. We have
is,

Now
3

|x-

if

|*i.e

if

l-e/3

<

<

l+e/3.

Thus, there exists an interval (1 e/3, l e/3) around 1 such


that for every value of x in this interval, the numerical value of the
difference between f(x) and /(I) is less than a preassigned positive

number e. Here S=e/3,


Hence /(x) is continuous
Note*

lit

may be shown

for

x=l.

that, 3;c-M, is continuous for every value of x.

DIFFERENTIAL CALCULUS
2.

Show

Let

f/Ktf

/(x)=3x

+2x

1 is

We have

be any given positive number.

= 3x*+2x-l-15
= x-2
3x+8
|

We

suppose that x

between 1 and 3, 3x + 8
Thus when

lies

is

between

positive

and

x = 2.

continuous for

less

.
|

and 3. For values of x


than (3*3+8)= 17.

we have

Now
17
if
|

Thus,

we

x-2

<e/17.

see that there exists a positive

number e/17 such that

when

x-2

<e/17.
Therefore /(x) is continuous for x=2.
Note. It may be shown that 3x*-f2x 1,
|

of

continuous for every value

is

x.

Prove that

3.

say

sin

continuous for every value of x.

is

shown that

It will be

sin

is

continuous for any given value of

c.

Let

be any arbitrarily assigned positive number.


|

/(x)

/(c)

=
=

sin

x+c
-~ sin x

X + C!

xc

2 cos

have

We

sin c

Now

cos

x+c
;

for

every value of x and

xc^lx

Also

sm
c.

(From Elementary Trigonometry)

Thus we have

sm x sm

cos

xc

x+c

XC
Hence

sin
|

sin c

<e when x
|

c
|

<

e.

CONTINUITY AND LIMIT


Thus, there exists an interval
every value of x in this interval

sin

(c

e,

sin c

every value of x
4.

is

continuous for

c being

c such that

for

<s.

xc

Hence, sin x

c+e) around

and, therefore,

also for

any number.

sm* x

is continuous for every value ofx.


be any given positive number.

S7z0w

Let

We

have

f/?tff

\f(x)-f(c)

sin 2 x-sin 2 c
|

sin
|

<

(x+c)

sin (x

c)

x-c

sin (x

c)

when
1

Hence, sin x
every value of x
5.

Show

atx^.

is

x-c

<e.
continuous for

c being

x=c

and, therefore, also for

any number.

that the function f(x), as defined below,

x,

[x,
The argument

will

discontinuous

is

when
when

be better grasped by considering the graph*

of this function.
consists of the point P (,
the
Our intuition
excluding
point A.
there
that
is a disimmediately suggests
continuity at A there being a gap in the
graph at A.

The graph

1)

and the

*
*

lines

OA, AC:

*p ^ *
,

we can see that


round about x=,
/(x) differs from /() which is equal to 1,
by a number greater than so that there
is no question of making the difference
between /(x) and/ () less than any posiAlso, analytically,

for

every value of x,

tive

number
3-2.

for

Fi

X.

arbitrarily assigned.

Therefore /(x)
will

is

discontinuous at

x=:

Limit. The important concept of the limit of a function


introduced.

now be

For the continuity of /(x) for x = c, the values of the function


values of x near c lie near/(c). In general, the two things, v/z.
The value /(c) of the function for x= c, and
(i)
;

(ii)

the values of /(x) for values of

x near

c,

42

DIFFBBENTIAL CALCULUS

It may, in fact, sometimes happen that the values


are, independent.
of the function for values of x near c lie near a number / which is not

equal to f(c) or that the values do not lie near any number at all.
Thus, for example, we have seen in Ex. 5, above, that the values of
the function for values of x near
lie near
which is different from
the value, 1, of the function for
In fact this inequality itself
was the cause of discontinuity of this function for x= J,
The above remarks lead us to introduce the notion of limit as

x=.

follows

Mm

Def.

f(x;=l.

x -> c

A function f(x)

tends to c, if, corany positive number e, arbitrarily assigned, there exists


a positive number, 8, such that for every value of x in the interval
S, c+S), other than c,f(x) differs from I numerically by a number
(c
is

said to tend to a limit,

as

I,

responding to

which

is

less

than

e, i.e.,

\A*)-l\ -<e,
for every value

ofx

other than

such that

c,
\

Right handed and

left

handed

lim

xc

/(*)=/.

x->(c+0)

A function f(x)

is

<8.

limits.

said to tend to

lim /(*)=/
x->(c-0)
a limit, I, as x tends

above, if, corresponding to any positive number,


there exists a positive number, 8, such that

I/W-M

e,

to,

c,

from

arbitrarily assigned,

<*

whenever

c<x<c+8.
In this case, we write
lim fix)

and say

1,

the right handed limit of f(x).


A function f(x) is said to tend to a limit, I, as x tends to c from
below, if corresponding to any positive number, e, arbitrarily assigned,
their exists a positive number, S, such that
that,

/ is

whenever

c8<x<c.
In this case, we write
lim
/(*)=/
x -> (c-0)
and say that, /, is the left handed limit of/(x).
From above it at once follows that

lim

*)

/,

CONTINUITY AND LIMIT


jf,

and only

if,

Jim
It is

/fx)=/=

lim

important to remember that a limit

f(x)

may not

always exist.

(Refer. Ex. 3, p. 46)

Remarks

In order that a function may tend to a limit it is necessary


and sufficient that corresponding to any positive number, e, a choice of 8 is
possible. The function will not either approach a limit or at any rate will not
have the limit /, if for some e, a corresponding 8 does not exist.
2.

1.

The question of the

x approaches 'c* does not take


The function may not even be defined for

limit of f(x) as

any note of the value of /(or) for x^c.


x=c.

3-21.
Another form of the definition of continuity. Comparing
the definitions of continuity and limit as given in
3*1, 3*2, we see
that

x=c

f(x; is continuous for

if,

and only

if,

lim f(x)=f(c).
X ->C

Thus the limit of a continuous function f(x)


equal to the value f(c) of the function for x=c.

as

x tends

to

c,

is

>

We
in

any

continuous for X = c

see that a function /(x) can fail to be


one of the following three ways :

(/)

(11)

is

f(x)

not defined for

x=c

f(x) does not tend to a limit as x tends to


lim f(x) does not exist.

r, i.e.,

x ->c

Hm

(Hi)

f(x) exists and/(c)

is

defined but

lim f(x) ^f(c).

Examples
1.

Examine

<w x tends to

the limit of the function

The function

is

y=

defined for every value of

x2

=*+!, when x^l.

Case 1. Firstly consider the behaviour of the values of


values of x greater than L
Clearly, the variable

than

1.

is

greater than 2

when x

is

y for

greater

^4

DIFFERENTIAL CALCULUS

If x, while remaining greater than


difference from 1 constantly diminishes,

In

takes

up values whose-

then y, while remaining


takes up values whose difference from 2 constantly

greater than 2,
diminishes also.

like

1,

fact, difference

by taking x

between y and 2 can be made as small as we-

sufficiently

near

1.

For instance, consider the number

"001.

Then

if

x<l-001.

Thus, for every value of x which is greater than 1 and less thai*
1*001, the absolute value of the difference between y and 2 is less than*
the number -001 which we had arbitrarily selected.
Instead of the particular number -001,

number
Then

we now

consider

any

e.

positive

if

x<l +

e.

Thus, there exists an interval (/, 1 -f e), such that the value ofy,
for any value of x in this interval, differs from 2 numerically, by a
number which is smaller than the positive number e, selected arbitrarily.

/, is

Thus the limit of y as x approaches


2 and we have

Case

II.

We now

less

than

When x

2,

consider the behaviour of the values of

y for

1.

than

1, }> is less

than

2.

while remaining less than 1, takes up values whose


1 constantly diminishes, then y, while remaining less
takes up values whose difference from 2 constantly diminishes

If,

difference

than

is less

through values greater than

=2.

lim

values of

x,

from

also.

Let, now, e be

any

arbitrarily assigned positive

number, however

small.

We

then have
|

if

y-2

=2->>=2-(x+l)=l-x<
1

e,

<x,

so that for every value of x less than 1 but >1


value of the difference between y and 2 is less than the

s,

the absolute

number

e.

CONTINUITY AND LIMIT

Thus there
Jor any x in the
is

such that the value of y,


s, 1)
2 numerically by a number which
selected positive number e.

exists an interval (I

interval, differs from

smaller than the arbitrarily

Thus the
1, is

45

ofy, as x approaches

limit

1,

values

through

less

than

2 and we write
lim

y=2.

jc~(i-0)
Combining the conclusions arrived at in the last two
any arbitrarily assigned positive
e, 14**) around 1, such that
for every value of x in this interval, other than 1, y differs from 2
numerically by a number which is less than e, i.e., we have
Case HI.

cases, we
number e,

see that corresponding to


there exists an interval (1

y-2

or

any

x, other

than

1,

<

such that

|x-l|<e.
Thus
lim

y=2,

or,

y ->

2 as

x ->

1.

x->l
Examine

2.

of the function

the limit

y=x sin x,
as x approaches 0.

Case

Let x

I.

>

that y

0, so

<

is

also

> 0,

if

we suppose

7T/2.

For 0<x<7T/2, we have


sin
BO

r<x,

that

x
Let

sin

x<x 2

if

0<x<7r/2.

be any arbitrarily assigned positive number.

For values of x which are positive and

less

than

y'e,

we have

X <e

we have
x sin x<e.

so that for such values of x,

e ) such that for every value


exists an interval (0,
of the difference between
value
numerical
the
interval
of x in this
is less than the arbitrarily assigned positive number
x sin x and
1
so that we have a situation similar to that in case I of the Ex.

Thus there

e,

above.

Thus
lira

Case

II.

Let

x<0

so that

sin

^>0,

x=0.
if

we suppose

x>

Tr/2.

The values of the function for two values of a which are equal
Hence, as in case I,
in magnitude but opposite in signs are equal.

DIFFERENTIAL CALCULUS

46

we easily see that for any value of x in the interval


the numerical value of the difference between x sin x and
than

\/e, 0)
is^ less

e.

lim
x sin x=0.
x -> (0-0)
Case HI. Combining the conclusions arrived at in the last two
cases, we see that corresponding to any positive number e arbitrarily
\/s, \/ e ) around 0, such that
assigned, there exists an interval (
for any x belonging to this interval, the numerical value of the

Hence

difference

between x sin x and

i.e.,

Hence

<e,
x-~Q

is

sin

lira

sin

<e.

x=0.

x->0
It will be seen that the inequality

Note.

sin

x
|

< e is

even for x=0. But it should be carefully noted that no


difference would arise as to the conclusion
lim xsin x=0,

satisfied

even

if

x=0

were an exception.

we

Again,

see that for

#=0,
sin

which

is

for

is

it is

limit
conti-

l)/(jtl), as considered in Ex. 1, possesses


but does not have any value for x=l, go

discontinuous for

Examine

3.

Thus in this case, the


Thus this function is

x=0.

The function (x2


a limit as x approaches
that

is

0=0

x approaches 0.
the same as its value.

also its limit as

of the function

nuous

the value of the function

the limit

x = l.

of sin

(1/jc)

as x approaches 0.

Let
so that

is

a function defined for every value of x, other than


to understand

The graph of this function will enable the student


the argument better.
To draw the graph, we note the following points
(/)

from
1

toO

As x

7T/2

to

increases monotonically from 2/7T to oo, 1/x decreases


and therefore sin (I/*) decreases monotonically from

as x increases monotonically from 2/3-7T, to 2/7T, Ifx de(//')


greases monotonically from 3?r/2 to Tr/2 and therefore sin 1/x increases
1 to 1
monotonically from
;

increases from 2/5?r to 2/37T, l-/jc decreases monotoni(i/i)


to 3?r/2 and therefore sin l]x decreases from I to
from
5?r/2
cally

as

-i;
and so on.

CONTINUITY AND LIMIT

Thus the positive values of

number of

jc

47

can be divided in an

infinite-

intervals

r j-

-Li

r j!_

I*'

* J'
LlT' 5;: J' L57T> 3lTj'
such that the function decreases from 1 to in the first interval on
the right and then oscillates from
1
1 to 1 and from 1 to
on
interval
second
in
the
the
others
from
alternatively
beginning
the right.
It may similarly be seen that the negative values of
divide themselves in an infinite set of intervals.

A
[-

x also

_? 1

37T'

57TJ

such that the function decreases from


1 in the first interval
to
1
on the left and then oscillates from
1 to 1 and from 1 to
alternately in the others beginning from the second interval on the^
left.

Hence, we have the graph (Fig. 37) as drawn.

more and more


1
1 and
The function oscillates between
rapidly as x approaches nearer and nearer zero from either side. If
we take any interval enclosing x=0, however small it may be then
for an infinite number of points of this interval the function assumes
the values

and

Fig. 37.

There can, therefore, exist no number which differs from the


function by a number less than an arbitrarily assigned positive number
for values of x near 0.

Hence
lira

(sin 1/x)

does not exist.

->0

This example illustrates an important fact that a limit


always

may

not

exist.

Note.

The function considered above

the function defined for


jc=0. Neither
exist when x tends to zero.
is

is

x=0

not continuous for


nor does its limit

48

DIFFERENTIAL CALCULUS

Examine

4.

the limit

of

sin

x
as x approaches

0.

Obviously the function


zero values of

is

not defined for

x=0.

Now,

for

non-

*rini-

]*|

sin~L]<|*|

so that

xsm

sin

j\

when
Thus, we see that

be any positive number, then there


0, such that for every value of x in
vthis interval, with the sole exception of 0, x sin (I/*) differs from
than e.
Jt>y a number less

an interval

exists

if,

e, s)

e,

around

Hence

In

fact, as

may

be easily seen, the graph of

y=x
oscillates

between

y=x
as

sin

x tends to

and>>=

zero.

This function

is

not continuous for x=0.


and variables tending to

Infinite limits

3*22.

infinity.

Meanings

of
(i)

lim/(x)=oo

lim f(x)=

(//)

oo

A function f(x) is said to tend to oo as x tends to c, if,


Def.
corresponding to any positive number G, however large, there exists a
positive number 8 such that for all values ofx other than c lying in the
,

interval

[c-8, c+S],
/(*)

>

G.

as x tends to c, if,
oo
Also, a function f(x) is said to tend to
Corresponding to any positive number G, however large, there exists a
positive number S such that for all values of x in [c8, c+8] exclud,

ing

c,

AX)
Right handed and

left

<

-G.

handed

limits can be defined as in

p. 42.

We shall now consider

some examples.

3*2,

INFINITE LIMITS

Examples

Show

I.

that

lim

^cA*
The function

is

52

not defined for

]=OQ

'

x=0.

We

write

y=l/x

2
.

Case I. Lef ;t>0. If x, while continuing to remain positive,


diminishses, then 1/x 2 increases.
Also, 1/x* can be made as large as we like, if only we take x
sufficiently

and greater than

near

it.

Consider any positive number, however large, say 10 6

Then
l/x

>10 6

so that for all positive values of

if

X<1/10*,
3
x<l/10 , we have
,

Instead of the particular number


positive number
Then

10 6

we may

consider

any

G.

Thus, there exists an interval [0, lj\/G] such that for every
x belonging to it, y is greater than the arbitrarily assigned
t

value of
positive

number G.

Hence
lim

l*2 = oo.

Here also y is positive. If, x while contiII. Let x<0.


2
the
remain
nuing
negative, increases towards zero, then 1/x also
Case

increases.

Also

if,

be any* arbitrarily assigned positive number, then,


x in the interval ( l/y^* 0)> we have

for every value of

so that 1/x 2 tends to oo as


and we write in smybols

x tends to

2
lim
(l/JC )
a->(0-0)

through values

less

than

it

= oo.

Case III. Combining the conclusions arrived at in the last two


cases we see that corresponding to any arbitrarily assigned positive number G, there exists an interval [
II\/G, lj\/G], around 0,
such that for every value of x (other than 0) belonging to this interval,

we have
Thus
lim

lx*z=oo

or (l/;c*)-*oo as

50

WFFBRENTIAL CALCULUS
2.

Show

that

lim

=POO

--

lim

The function

is

1.1 =

,.

lim

00,

cfoes wo/ exist.

We

not defined for x=0.

write

y=I/x.

Case I. Let x>0 so that j is positive. If x, while


continuing
to remain positive, diminishes, then
Ijx increases.

Also, if

be any positive number taken arbitrarily, then

l/x>G,

if

x<l/G.

Hence
lim (l/x)

Case

Let

= oo

whenx->(0+0).

#<0

so that y=ljx is negative.


If x, while con^
tinuing to remain negative, increases towards zero, then y=ljx
decreases and numerically increases.
II.

Also, if

be any positive number taken arbitrarily, then

l/x<-G,

if

x>-l/G.

Hence
lim (l/x)=

Case

when x->(0

0).

Clearly when x->0, lim (1/x) does not exist.


We will now give a number of definitions whose mean-

III.

3-23.
ings the reader
(i)

oo

may

easily follow.

function f(x)

is

said to tend to a limit

/,

as

tends to

QO ), if, corresponding to
(or
any arbitrarily assigned positive
e, there exists a positive number G, such that

oo

number

'

!/(*)-/
for every value

ofx>G, (x<

(11)
i/,

A function f(x)

corresponding to

a positive number

is

oo
as x
number G, however
,

oo

].

tends to oo (or

oo )

large, there exists

such that

we
r

Symbolically,

[lim/(x)=/ when x->

said to tend to

any positive

<>

G).

we write
= / when x->oo

Symbolically,
lim /(*)

write

lim/(x)=oo when x~oo [lim/(x)=<x> when x-> oo )]


The reader may now similarly define
oo when x-*oo
lim /(*) =
oo when x-*
lim/(x) =
,

Note

1.

Instead of

oo,

we may

write +00 to avoid confusion with

oo
oo

Note
Rigorous, solution of any .problem on limits requires that the
problem should be examined strictly according to the definitions given above in
terms, of e'$, g'j, G's etc. Bat in practice this proves very difficult except in some
2.

THEOREMS ON LIMIfS

We may

very elementary cases.

not,

J5J

therefore, always insist

on a rigorous

solution of such probelms.

Exercises

Show

1.

lim

(2*+3)=5.

lim

(**+3)=4.

(0

(///)

that the following limits exist


(//)

2.

(3*-4)=2.

o;~>3

lim

Show

(l/aO=i

(vO

lim

that

(0

5x+4

is

continuous for z=2.

(11)

z a +2

is

continuous for *=3.

Also show that two functions are continuous


3.

lim

(iv)

a?->l
(v)

and have the values as given


lim

Examine the continuity

for every value of x.

for oj=0, of the following functions

(0
.

/()=/

(,7)

6,

sin

Prove that cos x

is

~,
whena?=0,

0,

j^

4.

when #=3,

continuous for every value of x.

Show that cos 2 x and 2+x+x* are continuous for every value of x.
6.
Draw the graph of the function V(x) which is equal to
when x
when 0<a;^l and to 2 when s>l and show that it has two points of

5.

to

discontinuity.
7.

function

Investigate the points of continuity and discontinuity of the following


:

f(x)=(x*la)-a for x<a, /(a)=0,/(aO=a-(aV#), for


8.

Show

that

when

/(z)=cos
is

discontinuous for
9.

Show

g=0.

that

lim

tan a?=oo

[Refer

lim

tan x does not exist.

lim

but

10.

2-63, p. 27]

Show

that

lim

logx=

oo.

*->(<HO)
[Refer
11.

x^O and/(0)=l.

2-5, p. 25].

Examine
lim

5^

tan a?=

oo

DIFFERENTIAL OALOTTLtTS
[If x tends toO through positive values, then I/* tends to ooand, therefore,
21 /* tends ooo.
i

If x tends to
through negative values, then I/* tends to
l/a
tents toO.
fore, 2

oo

and there-

>

Thus
JL_

-JL,

lim

2* =

lim

2*~=oo,

s-KO+0)
but

2*

lim

does not

exist.]

Show

12.

that

lim

Show

13.

where

that if

denotes the greatest integer not greater than rr, then


lim
lim
lim f(x) does not
/(*) = !,,
/(aO=0,

[a?]

*-Hi -0)
What

a?->(l

the value of the function for

is

exist.

<c->J

+0)

x=\

[Refer Fig. 11, p. 17]

Explain giving suitable examples, the distinction between the value of


the function for a?=a and the limit of the same as x tends to #.
14.

15.

but

is,

Give an example of a function which has a definite value at the origin


nevertheless, discountinuous there.

Theorems on

3*3.

Let/(x) and
lim

<p(x)

limits.

be two functions of x such that

/(*) = /,

lim

?(x)=w.

Then
lim
a?-0

(0
i.e.,

the limit of the

limits

lim [f(x)

the limit

of their

limits

(Hi)
i.e.,

of the

equal to the

sum of

their

(iv)

difference

/(x)lim

of two functions

y>(.x)

is

lim [f(x)-9(x)]=limf(x) lim *'(*)


x-*a
x~>a
x >a
.

m,

equal to the difference

of the product of two functions

is

= /m,
equal to the product of

lim

of the quotient of two functions is equal to the quotient of


provided the limit of the divisor is not zero.

^Ae limit

their limits

?(x)]=lim

ih* limit

their limits

i.e.,

is

(ii)

i.e.,

x-*a

x >a

sum of two functions

FEOPERT1ES OF CONTINUOUS ITUNCTIONS

63

These results are of fundamental importance but their formal


proofs are beyond the scope of this book.
These results can easily be extended to the case of any finite

number of

functions.

Note. Inversion of operations.


beginner who notices the above statements for the first time may fail to see any meaning in them and he may not be
able to appreciate the significance of the same. In order, therefore, to help
him dp so, some observations seem to be necessary here.
In mathematics we deal with different types of operations and sometimes the mathematical expressions are subjected to two or more operations
and in such a case the order in which the operations are made must be taken
into account. It cannot just be assumed that the order can be changed at will
without altering the final outcome, i.e., the final result may not be independent
of the order of the operations.

The fact is that in each case the question as to whether the inversion of
the order of operations is or is not valid has to be separately examined in
relation to the nature of the operations.
We now examine the following statements
:

(i)

logOnfn) ;*logw4-log

n.

+ B) ^sia A+ sin B.
=0x6 + axc,
(Hi) ax(b+c)
In (/) we find that the two operations
(it)

sin (A

involved,

viz.,

that of addition

and taking of log arc not invertible.


Similar is the case in (//) where we have the operations of adding and
taking of sine.
In (//'/), however,
plication are invertible.

we

two operations of addition and multi-

see that the

The theorems in this article amount to asserting the validity of the inversion of the operations of taking the limits with each of the four operations of
addition, subtraction, multiplication and division.
The reader is advised to think of other similar cases also.
3-4.

Continuity of the sum, difference, product and

quotient of

two continuous functions.

Let /(;c),

x=0,

<p(x)

be any two functions which are continuous for

so that

Prom

lim/(x) ^(a),
x -> a
the theorems in 3*3,
lim
x ->

lim 9(x)=?(0).
x -> a

we

see that

a.

=value of
so that/(x)+9(.x)

is

[/(*) +?(*)], for

The continuity

of/(x)

y(x)

and

f(x)<p(x)

proved.

For the quotient, we have


lim

*=0

continuous at x = (t.

/M -?^M

,
'

*here lim

may

similarly

be

D1FFB&BKTUL CALCULUS

54

JM

^value of t&.

so that/(x)-r?(x)

for

x-a,

?(*)'

?(a)

also continuous at x=fl, provided

is

when x ->

lim <p(x)=<p(a)^iQ,

a.

Thus, f&e sum, the difference, the product and the quotient of two
continuous functions are also continuous (with one obvious exception
in the case of a quotient).
3*41.
Continuity of elementary functions.

We have seen in Ex. 3, p. 40 and Ex.


GOB x are continuous for all values of x.
Hence from

sin

tan

x= cos

3*4, above,

x
x

cot

x
sm x

x=- --

51, that sin

x and

see that

cos

we

4, p.

sec

x=-

cos

,
'

cosec

x=- ------sin

all those values of x for which they are defined


points of discontinuity of these four functions arise when the denominators cos X sin x become zero and for such values of x, these functions themselves cease to be defined.

are also continuous for

Also it may be easily seen that, x, and a constant are continuous functions of x. Thus by repeated applications of the results of
3/4 above, we see that every polynomial

a continuous function for every value of x and that every rational


function

is

<*

x+a x-l +...+a n


l

a continuous function of x for every value of x except those


which the denominator becomes zero.

is

for

1
1
1
1
1
Finally the functions, sin- x, cos- x, tan"" x, cot- x, sec- x,
1
a?
are, continuous for all those values of x for which
cosec- x, log a x,
they are defined. This is geometrically obvious from the graphs
drawn in Chap. II. Analytical proofs are beyond the scope of this

book.
3*5.

Some

important properties of continuous functions.

Letf(x) be continuous for x=c and f(c)^0. Then there


an interval [c S, c+S] around c such that /(x) has the sign of

3-51.
exists

f(c)for all values

ofx

in this interval.

Its truth is obvious if we remember that a continuous function


does not undergo sudden changes so that if /(x) is positive for any
value c of x and also continuous thereat, it cannot suddenly become
negative and must, therefore, remain positive for values of x in a
certain neighbourhood of c.

PROPERTIES OF CONTINUOTJS FUNCTIONS


In informal, precise manner,

it

may

be proved as follows

To

every positive number s, however small


correspond a positive number S such that

for

65

it

may

be, there

every value of x such that


c

Letf(c)>Q. In this case we take e any positive number less


than/(c) so that/(c)-~ e and/(c)+e both are positive, Thus, /(*) lies
between two positive numbers and
x lies between
and c+S.

c8

is,

therefore, itself positive

when

Hence /(c)e is already negative and /(c) +s will


Ltf(c)<0.
be negative if e< /(c). Thus in this case if we take e any positive
number which is smaller than the positive number /(c), then we see
that /(x) lies between the two negative numbers /(c) e and/(c)+s
and is, therefore, itself negative when x lies in the interval
~
[C-8, C+8].
:_
*

Hence the theorem.


Note. This simple but important property of continuous functions will
be used in the chapters on Maxima, Minima and Points of inflexion.
3*52.
Let f(x) be continuous in a closed interval [a, b] and let
/(a), f(b) have opposite signs ^ Thenf(x) is zero for at least one value of
*> lying between a and b.
Its truth is intuitively obvious, for, a continuous curve
y=f(x)
going from a point on one side of x-axis to. a point lying on the other
cannot do so without crossing it.
Its formal,

analytical proof

however, beyond the scope of

is,

this book.

3-53.
Letf(x) be continuous in a closed interval [a, b]. Then
there exists points c and d in the interval [a, b], where f(x) assumes its
andm, i.e.,
greatest and least values

The proof

is

beyond the scope of

this book.

continuous func_ .The. theorem states that there is a value of a


tion greater than every other of its values and as also a value smaller
than every other value.
.

A
as

discontinuous function

we now

may

not possess -greatest or least values

illustrate.

Consider the function defined as follows-:


I

x,~fbr

\,~-

0<;

ibrJfc=X

DIFFERENTIAL CALCULUS
point C(0 *>' and the line AB excu u point 5. The function
.
the
possesses no greatest value; 1 not being
a value of the function. If we consider
>

any^ value less than 1, however near 1


may be seen that there is a value
of the function greater than that value.
;

it

(0'i)C

This

explained by the fact that


the function is not continuous in the
is

interval (0, 1)
Fig. 38.

#=0

being a point of

discontinuity.

Note. This property will be required to prove Rolle's Theorem in Chap.

Example

Show

when, n

is

that

any

integer.

We cannot

write

,.

hm

jc*~ a*
------

yr-S

--- ~>

r
for

,.

hm(x

In the present case the limit of the numerator

Case

Firstly suppose that n

I.

is

a)=0.
is

also zero.

a positive integer.

actual

By

division,

the equality being valid for every value of x other than a.


limit does not depend upon the value for x=a, we can write
-

lim

As the

~~
x-*a x a

x -> a

Being a polynomial, the function

is

continuous for every value of x and, as such,


Thus
to its value for x~a.

its limit

when x -> a

must be equal

x
n
-*a+ ...... +d-i=na*-*
-^-=a*-i+a
~
x ->a x a
lim

Case
where

is

Now suppose that n is a negative integer, say


a positive integer. We have

II.

m,

67

iMPOKTAisri LiMi'fs

-- =
x*

,.

lira

a*

,.

lim

1
x m -a m
m
a xm
x a
x m a m ,.
1

hm

m m
employing Case I and the fact that l/a x
In the Case II, we must suppose that

'

is

continuous for x=a.

Exercises
1.

Show

2.

If

that

^-2 and/(-2)*t
find

A so that/(x) may be continuous for

Show

3.

*=

2.

that

lim sin(*-f /t)=sin x.


/t~>0

Some important and

3*6.

Limiting value of x" when n tends to infinity through posi


x being any given real number.

3*61.

live integral values

Let

(i)

useful limits.

x>\.

x= 1 +h

We

write

By

the Binomial theorem,

where each term

is

so that h

is

positive.

we have

positive.

Hence
n
x=(l+h) >l+nh.

As

(\-\-nh) tends to infinity

with

n,

therefore x n also tends to

infinity.

To be more

rigorous,

we

consider any pre-assigned positive number G.

Then
If

be any^positive integer greater than (G l)/t, then


x n >G for H> A so that x*-*oo when w~>oo
.

58

DIFFERENTIAL CALCULUS
(ii)

Let

x~l. Here

x*

=l

and therefore

for all values of n

in this case.

(m) Let

We

0<x<l.

write

Here, also, x"


so that A

x=l/(l +h)

is

positive.

a positive number.

is

As before
<z __ __
^
_

1+nh'

Now

ll(l+nh)-*Q as n tends to

Therefore x"-*Q as

infinity.

oo.

To be more

rigorous,
,

If

m be any

we consider any

number

positive

t.

Then

if

integer which

>

is

--

then

J[A,

;c*<c for ri^m*

Thus we

n^m

so that for

see that x w , which

is

lim x w
(iv)

Let

(v)

Le/

lies

always positive,

between

and e

0<jc<l

x=0.

Here

when

^=0 for all n so that x->0.


Here x

l<x<0.

>co

negative so that x* is positive


We write .*= a

is

for even values of n and negative for odd values.


so that a is a positive number less than 1.

Absolute value of x n

is

x"
|

But a*->0.

i.e.,

=an

Hence

1 and 1, therefore
Since x n is alternately
1.
Let x=
i n this case.
neither tends to any finite limit nor to Jt

(vi)

x<

it

Here, again, x is alternatively negative and


in this case it takes values numerically greater than
But
positive.
n
any assigned number. Hence x does not tend to a limit.
{vii)

Let

/.

Thus, we see that lim x*, when n ->

oo,

exists finitely

if,

and only

if,

Abo ttisOif-l<x<I
9

i.e.,

if\x\ <1 and

Iforx^l.

is

3'6i. Limiting value of x jn !, when n tends to infinity


positive integral values and x is any given real number.

Let x be positive and


integers between..which x

lies

let

m>

ao that

m+l

we haye

through

be the two consecutive


.

.,

IMPOETAN'T LlMlfs

We

write
s

^L

1*2

nl

^L

-JL

'

'

"m

IT'

*
'

m+1

p=^.JL
12

Let
so that

'

'

'

'

^n+2'

'7T*

m..

a positive constant.

/? is

^.
m+3

-^~,

Also, each of

m+2

x*

\*~

,
'

is

'

"

<-~ ^

8ay

where

A:

is

a positive constant independent of

Since, x/(w

+ l)

is

n.

and <1, therefore when

positive

m being a constant.
Hence
xn
lim -^^=^0,'

when w-^oo

let

Again,
tive.

We

x be any negative numbed, say

a, so

have

that a

is

posi-

x*

Ij

Now,

since

j-->0,

Hence

we

whatever value x

may

see that

lim

also-0.

x*
n-y*=

have.

Here, of course, n tends to infinity through positive integral


values only.

3 63.

Limiting value of

when n tends

to infinity through positive integral values.

60

DIFFERENTIAL CALCTJLXJS
Step

1.

we have

By

the Binomial Theorem for a positive integral index,

The expression on the right


Changing n to fl + 1, we get

is

sum

of (n +1) positive terms.

__L_ViA_^
A n+i
)

......

(w-i-1

Here, the right-hand side consists of the


Also

sum

fi

of

(+2)

n+l

\
J

positive

terms.

'

1
i_l_<i__2_
< n + l' !_!
n

so that each term in the expansion of


41

is

greater than the corresponding term in the expansion of

C'+r)'Thus, we conclude that


l

\*

K'+T)
for all positive integral values of w,
cally as n increases.

Step

II.

We

have

+ l/M)"

i.e.,

(l

increases monotoni-

SOME IMPORTANT LIMITS

<l+l+-2-+

2"+. ..+2.2.2. ..(-!)

-g.

=1 +

61

factors

<3) foralln.

...(//)

Thus we

see that

N*

-1

-j

steadily increases

successively the series of positive integral values


less than 3 for all values of n.

as

/i

takes up

1, 2, 3, etc.,

and

re-

mains

H
(1

-----

Note

1,

From

(i)

approaches a finite

and

(11),

we

for every value of n so that the limit

limit as

n ->

oo

see that

is

some number which

lies

between 2 and

3.

The mere

existence of the limit of (1 -f l/w)*has beea shown here and at


this stage nothing more can be said about the actual value of the limit which is
denoted by e, except that it lies between 2 and 3.

The reader will be enabled to appreciate the argument better and also
begin to feel a little more acquainted with, e, if he calculates the value of

1+

for various successive positive integral values of n.


<?-,
tfj-2, a 2 =2 25, a 3 -2-37, a4 -2 441,

In Chapter

IX

it

will

be shown how the value of e, correct to any number

of places, can be determined without

Note

much

inconvenience.

The proof for

the existence of the limit has been based on an


intuitively obvious fact that a monotonically increasing function which remains less
than a fixed number tends to a limit. The formal proof of this fact is beyond the
2.

scope of this book.

Cor.
real

I.

Lim

numbers as

its

(1

4-l/x)*=e, when x tends to

infinity taking all the

values.

If x be any positive real number, then there exists a positive


integer n such that

JL > JL
x

or

n+i'

i.j__^],4_ -L>i_|__J_,
x
'

n+1

82

DIFFERENTIAL CALCTTLtT&
/

or

O+T)
number to a

(In case the base is greater than 1, raising a greater


Qf.itie inequality.) __

greater

power, does not alter the. direction

We thus

have

Let x ~>
values.

oo

We know

Then n and (w+1) ->

00

through positive integral

that

lim

lim

H-|

= lr=:lim

x*

(l-|

M+ ""Xf

/I

\ n4

=^ = lim (l+^~TTj

Therefore

Urn

x-+<
COT. 2.

lim

l-.

-C^-ccV
Let

x=

so that >>->

oo

'

lim (l+z)

3.

oo

-4-

We have

lim fl-f

=e

--j-Y

^y-V

Cor.

as

when z ->

-+

Let z=l/x so that jc


positive or negative values.

T ^(l+~-

lim
ooV

0.

according as z ->

or

Now
lim

(1+z)

z->(0+0)

a=s

lim
*->

[1+

s=f

and

lim
(1+2)
-> (0-0)

lim

1+ ~)

=^.

through

63

SOME IMPORTANT LIMITS

Cor.

4.

Hl+
^
-+0\

*.y'*-Iim
a)
X ^.Q

Note, tn higher Mathematics the number


logarithms which are then called Natural logarithms.
not mentioned so that log x means log e x.

To show

3-64.

that

a*
lim

Let

taken as the base of

is

The base

i
------ =log, a.

a?l=y so that y ->

as x ->().

We have
or

a=log

log

(1 +>>),

i.e.,

=x

log (l+>0/loga.

Hence
a*

y
1

-.

--..

&v

'logo

--- _
1

=loga.

-;

log

hm
..

...

^Q

log (1+7)

(1+7)

a*-\
--=
x

--------lus M .

hm
'.

r.

^Q]_
a

logo

lim

To

Iy

Hm

-^0

3-65.

-----

log

(1+7)

T7/:K

1
J

is

generally

DIFFEBENTIAL CALCULUS
Let x=a(l+y) BO that y ->

*X -<* X

= aX

...

Again,

[(1+JO

0, as

-1]

a.

x~ 1
fl

we put
(1-BO

l=z

=i +2

so that z -+ 0, as

y ->

0.

or

-l)

_ a (X-D
a

(X-D

Fog

_
(1+z)

log

(1+z j

_z

'

z_

>
*

'

log (1+z)

~~~y
.
'

log (1+z)

Hence
..

Urn -

..

~
->0 x a

3*66.

lim

hm
z-*0

=?va

---------

-^0 x

=1,

--------------

log

as proved in books

Examples

Show

that
ll

is

continuous for

when

x=0.

Now
lim/(x)=lim f

#->0
aothat

lim

a?->OL

f[x)=*

->0
Hence the

result.

hm
,.

'

metry.

1.

,,

on Elementary Trigono-

EXAMPLES
2.

Show

65

that
i

/(x)=(e*--l)/(e* +1)

whenx^Q,

/(0)=0
if

x=0.

discontinuous at

When x

through positive values, then

tends to

therefore e* -

oo.

tT

..

Thus

_1
-

When x

through negative values, then

tends to
JL

And therefore e * ->0.

Thus

hm
Hence we see that
lim

Hm

eo that

does not

lim

f(x)

f(x)

exist.

Hence the
3.

f(x)^

Show

as

that

continuous for

Now,

result.

x=l.

may

easily be seen
_

lim

;r
^ - 1 =B

oo,

e*"1 =0.

lim
a?->(l

0)

l/JC->

<x>

DIFFERENTIAL CALCULUS

Thus
lim

Km

/(x)=0,

*-Hl+0)

x=0=

lim

Hence the
4.

annum.

/(x)-0,

*->(!- 0)

result.

A sum of P rupees is given on interest at the rate of r% per


What mil be the amount after t years, when the interest is

being continuously added ?

Supposing that the interest


year, the

amount

after

is

added after every nth part of a

years will be
r

The required amount

We

is its

\tn

limit as n-> c.

have
"

fr/100

as n

which -> Pe

--*>

oo

and

is

thus the required

amoun

Exercises
1.

Prove that
sin

ax

-- COS
l

..

taf^

(j/i)

2.

lim

Examine whether or not the following functions are continuous

0,
:r

(iO

1,

2r
"5

sin

A*)-}"
^

tan 2*
|>

when
when

a;

i^O.

rc=0.

when #-0,
when a; =0.
wbencc^O.

when a;

0.

whena;=0.

HYPERBOLIC FUNCTIONS

e2 ,
(

when #=0.

-I/*

*)=)
C

when #=0.

1,

e-

f
(v/0 /(*)=

/*

whcn x

1+~^T~'

<(

f
vi/0

Show

whens^O.

-1

3.

I/a'

"777
*
e 11

/()--(

when # = ().

U,

67

when #=0.

1,

that
sin

lim

a?
sin
------

(a?-|-/0
-

=cos

A-->0

Carefully state the results you employ.

Show

4.

and

that

(//) (

when w->oo through

Draw

5.

positive integral values.

the graphs of the functions

3-7. Note on Hyperbolic Functions. In analogy with Trigonometric functions, the Hyperbolic functions are defined in the following

manner

QX _ p X

03?

sinh

x=

^
sinh

tanh X=s

sech
3-71.

=e^

e-^

Graph of

The following

C0th

cosech

sinh

is

sinh

x=

properties of sinh

will

-- ass ~

enable us to

continuous for every value of #.

0=-

-^

sinh x.

V>__gO'

(tf)

X= cosh x =

graphs:
(i)

12

x=

cosh x==

=0.,

draw the

68

DIFFERENTIAL CALCULUS

-_-L

e x -er*

e*

and as x increases from

0,

2
monotonically increases and
Thus as x
monotonically decreases.
increases from 0, sinh x monotonically
e*

increases.
(iv)

lim (sinh
x->oo

x)=oo

e~*- e*
(v)

sinh
-a

(-*)=
..

sinh x.

Thus we have the graph as


shown, in general outlines. (Fig. 39).
Fig. 39

3'72.
(/)

Graph of cosh

cosh

x.

continuous for every value of x.

is

(ii)

cosh

0=

(iii)

cosh

x=

and as seen above ( ex e-x)l2 monotonically


increases, as x increases from 0. Thus cosh x
increases
as x
increases
monotonically
from

0.

(iv)

(v)

x=

lim cosh
X->co
cosh

oo.

x)=cosh

x.

Hence the graph as shown.


3*73.
(i)

Graph of tanh

x.

tanh x is continuous for every value of x, the denominator


not vanishing for anv value of x.

() tanh 0=0.
(tit)

tanh x

e*+e~*

1-

69

HYPERBOLIC
BO that tanh

increases as

increases from

onward.

vi

y^tanh x
Fig. 41

lim tanh

(iv)

(v)

tanh

x)

We may note

\-e~**

e*-fi-

lim

lim

e*+e~*

1+e

tanhx.

that

tanh x

for every value of x.

Note. The graphs of coth x, sech x and cosech x have not been given.
The reader may, ifhe so desires, draw the same himself.
Some fundamental relations. The following fundamental
3-74.
relations can be at once deduced from the definitions
:

<

a)

cosh x

sinh

tanh x
2

=1

=sech2 x.

(2)

(3)

coth2 x-l

(*)

cosh 2x+sinh 2x

2 sinh x cosh x

=sinh 2x.

=cosech2 x.
cosh 2x.

Inverse hyperbolic functions. In this section we obtain the


3*8.
logarithmic expressions for the inverse Hyperbolic Functions
:

sinh- x, cosh- 1

X tanh- 1
9

1
1
1
x, coth- x, sech"" x, cosech"" x,

which are defined as the inverses of the corresponding hyperbolic


functions. This will also necessitate a slight modification of the definitions so as to

A.

make them

single-valued.

Let
1
y=sinh"" x,

<

so that

is

the

number whose

sinh

or j

is

x.

DIFFERENTIAL CALCULUS
It is easy to see that x+
V(* 2 1) is positive and x <\/(x*+l)
negative for every value of x, positive or negative. Also we know
that the logarithm of a
negative number has no meaning in the
2
field of real numbers so that
[x
<\/(*
1)] has to be rejected. Hence

is

we have
B.

Let

We

will see that there are

a given number.

always two values of y whose cosh

is

Now

y= J
*=xi/(x*
x=cosh

(<# -f-<r")
l)

e 2 *-2,x.^+l=0.

i.e.,

or >>=log [x

v^C*

!)]

22
Here we see that both jt +
1 ) are
v/(* -l), an d x-Vf*
positive and real when x>l so that in this case y has two values.
For x>l, we have
,

and

BO that
log

[*4V(* 2 --1)]>0, and

To avoid this ambiguity,


cosh"" x a little and
say that
cosh is x so that we have
cosh-1

log

[x-

usual to modify the definition of


cosh- 1 x is the positive number y whose
it is

x=log

[x

+ vXx*

1)].

Note. The ambiguity referred to here is a consequence of the fact that


in xcosh y two values of .y
give rise to the same value pf x so that to a given
value of x correspond two values of y which
gave rise to it.
(Refer 3*72).

C.

Let

7=tanh~1 x, where
.-.

x=

Thus
l

A "T" A

x== ___ fog

It is easy to see that


^0,

if

HYPB11BOLIC FUNCTIONS

We may
D.

similarly

show that

eoth-ix=^^f

log

[<*+!)/(*- 1)]>0,
E.

71

sech-*x=log

X
A ^]
~

if
|

M.

>1.

jc|

1+ ^fl-Tlg2)-.

[0

<x< 1].
'

F.

where the sign of the radical

is

positive or negative according as

is

positive or negative.

Ex.
oo

or

oo

oo

Show

that

sinh* tends

to oo

or

oo

according as x tends to

3.

Show that cosh x tends to oo whether x tends to oo or to oo.


1 according
Show that tanh x -> 1 or ->
as x tends to

4.

Show

2.

,lo

1.

that

sinh (;c+^)==sinh

x cosh .y+cosh x sinh y

cosh (x-f v)=cosh x cosh y i sinh x sinh y.

oo

or

CHAPTER

IV

DIFFERENTIATION
Rate of Change. The subject of Differenits origin mainly in the geometrical problem of the determination of a tangent at a point of curve, has
rendered possible the precise formulation of a large, number of physical concepts such as Velocity at a point, Acceleration at a point,
Curvature at a point, Density at a point, Specific heat at any temperature, etc. each of which appears as a Local or instantaneous Rate
of change as against the Average Rate of Change which pertains
to a finite interval of space or time and not to an instant of time
Introduction.

4-1.

tial Calculus

which had

and space.
The fundamental idea of Local or instantaneous Rate of Change
all

pervading

these concepts underlies the analytical definition of

differential co-efficient.

4-11.

defined in

We consider a function f(x)


Let c be any number of this interval

Derivability. Derivative.

any interval

(a, b).

the corresponding value of the function. We take


of this interval which lies to the right or left
of c (i.e., c+/*> or <c) according as h is positive or negative. The
value of the function corresponding to it is f(c -{-/?).

o that /(c)

is

c+h any other number


Now,

h, is

the change in the independent variable x, and

the corresponding change in the dependent variable /(c).


The expression [f(c+h)f(c)]jh which is the ratio of these two
changes, is a function of h and is not defined for A=0 c being a fixed
is

number.
It is possible that the ratio tends to a limit as h tends to 0.
This limit, if it exists, is called the derivative of f(x) for x==c and
the function, then, is said to be derivable for this value.

Def.

f(x)

is

said to be derivable at

and

the limit

is

the function f(x) for

if

^,

/r->0

exists

xc

called the Derivative or Differential co-efficient

of

x=c.

The limit must be the same whether h tends to zero thcpugh


positive or through negative values. The function will not b&'derivable if these limits are different.

The function /(x)

is

said to be finitely derivable if its derivative

isftflffe.

72

DIFFERENTIATION
Ex.

Show that x2

1.

derivable forx=l

is

for *=7.

73r

and obtain

its

derivative

Let
2
/(jc)=x so that/(l)

To find the derivative for


that the function changes from
Change

in the

=l

Ex.

is

derivable and

that
\

is
\

to

1-f-ft

&

to

functional +h)*-l=2h+h

Show

2.

=l.

x=l, we change x from

which approaches 2 as h approaches

Hence /(x)

2
.

0.

its

derivative

is

not derivable for

2 for x

1.

x=0.

Let

f(x)=

x
\

s>othat/(0)=0.

shown that the


when h tends to 0.

It will be
exist,

limit of [/(04-/J)

f(0)]//J

does not

Now

f(0+h)-f(0)
h
according as h

is

=
-=-f(h)
/r lor-l

positive or negative.

Hence, [f(Q+h) f(0)]/h -~> 1 as h ->


through positive values
1 as h ->
through negative values.

and ->

Thus [f(0+h)f(0)]/h approaches different limits when h


approaches
through positive or negative values so that it does not
*
tend to a limit as h tends to 0.

Hence
Ex.

3.

show thatf(x)
for

is
\

If

is

not derivable for

f(x)=xsin

continuous for

x=0

x=0.

We

x=0.

but has no differential co-efficient

have

=x sin

-0=xsin
X

DIFFERENTIAL OALOTTLUS

x
|

Thus

if e

*-

Bin

<

be any pre-assigned number, we have


!/(*)-/(0)

<e when x-0


[

<e.

Hence
lim /(x)=0

x-0
so that f(x)

is

continuous for

Again

x=0.

==
x

jc

8n

and, as seen in Ex. 3, p. 47 lim sin (I/*) does not exist when x ->
Thus/(x) has no differential co-efficient for JC=0.
>

Ex.

0.

Find the derivatives of

4.

(/)

2x 8 -f 3x-4 for .v=5/2.

(11)

1/jc

for

x-5.

have defined the


Derived function. In
4-11, Ave
of
a
for
derivative
a function /(x)
particular value, c, of the independent variable. Instead of considering a particular number c, we, now,
consider any number x and determine
4*12.

>/v

lim

~ -yv

when h -> A
.

where x is kept constant in the process of taking the limit. We


suppose here that the limit exists whatever value x may have provi-

ded

it

belongs to the interval of definition of the function.

^his

limit which

a function of x

is

of Derivative off(x) and

is

is

called the Derived function

denoted by/'(jc).

Derivative of function

also

is

called its

Differential co-effi-

cient.

The symbol /'(c) then denotes the derivative of f(x)


Examples
1.

Find the derivative of x 2

Let
/(*)=**.

/'<*)-lim

&^^

when h ->

for x=*c.

DIFFERENTIATION

when h ->

^^I^

]i ra

ft

=
Thus 2x

is

lim

the derivative or the derived function of

x2

Puting x a=l, we get


which is the derivative of
Ex. 1, 4-11, p. 73.
2.

Find the

.x

for

x=l

and agrees with the

differential co-efficient

result of

of \/x.

Let

f(x)=\(m

We start

afresh to find derivative at

when'/* ->
through positive values
tive values of h.
Ex.

Find the

--r

,
'

x=0.

when X

We have

being not defined for nega-

<\Jh

of
derivations
(///)

(i/y 1/^Jx.

x8

</v)

ax*+bx+c.

Another notation for the Derivative. In this notation the


variables x and y are denoted by the composite symthe
changes in
bols Sx and Sy respectively, so that
4-13.

The

derivative

t.e. 9

lim

(Sj>/Sx),

as 8x ~+ 0,

is

then denoted by

another composite symbol dyjdx.

Thus
<

dx

lim

when 3x

^ 0.

ox

of y=f(x) for any partiAgain, the value f'(c) of the derivative


cular value,

c,

of x

is

denoted by

DIFFERENTIAL CALCULUS

76

dyldx is a composite symbol denoting lim SylSx and is not to be


regarded as the quotient of dy by dx which have not so far been defined as separate symbols.

Note

1.

Note

2.

The changes gx and Sy

1.

Find

Ex.

2.

If .K=M(x*-i-l), find (dy/dfc)

An

(-! ) x=Q

y are

*- when y

and

Ex.

4-14.
ts

in x,

*f-

also

known as

increments.

=-,

when *=-l-

important theorem.

Every

finitely derivable function

continuous.

x=c

Let/(x) be derivable for

so that the expression

[f(c+h)-f(c)]lh

We

finite limit as h tends to 0.

tends to a

write

lim

J^L^'J^J- x

lim

"

ii

A--0

Hence
lim fa+h)=f(c),

Therefore /(x)
Cor.
nition,

then

Note.

is

i.e.,

lim /(x)=
x ~> c

continuous at x=c.

If f(x) is derivable for every point of


it is continuous in that interval.

The converse of this theorem

is

its

not necessarily true

interval of defi-

i.e.,

a function

be continuous for a value of x without being derivable for that value.

may
For

example, the function

yis

1*1

continuous but not derivable for x=0.

[Ex. 2, 3

Ex. 1. Show that the function x + x


of x but is not derivable for x=0 and xl.
|

is

Ex. 2. Construct a function which


not derivable for three values of x.

Ex. 3. Show that/(x)=x*


and derivable for x=0.
[For

mor examples and

sin (1/x)

is

is

page 73)

continuous for every value

continuous for every value of x but

when x-^0, and/(0)-0,

is

continuous

exercises refer to the appendix to this chapter.)

DIFFERENTIATION

To show

Geometrical interpretation of a Derivative.

4*15.
/'(c),

77
that

t.e. 9

is the tangent of the angle which the tangent line to the curve
at the point P[c,f(c)] makes with x-axis.

y=f(x)

We take two points P\c, f(c)] and Q[(c+h), f(c+h)] on the


curve yf(x).
and
Draw the ordinates PL,
draw

PN

JL

MQ. We

QM

have

yi

and

NQ

Here,

_XRQ

is

Fig. 42.

the angle which the chord

makes with the A'-axis.


As h approaches

PQ

of the curve

0, the point Q moving


along the curve
the
the chord PQ approaches the tangent line
P,
point
approaches
TP as its limiting position, and XRQ approaches /. XTP which we

denote by $.

On

taking limits, the equation

(1) gives

tan</r=f'(c).

Thus f'(c)

is

the slope of the tangent to the curve


y=f(x) at
P[cJ(c)}.

Cor.
The equation of the tangent at any point P[c, f(c)] of the
curve y =/(*) is
y-f(c)=:f'(c)(x-c).
Note. The student should note that it is not necessary for every curve to
have a tangent line at every paint thereof. The existence of the tangent demands
the existence of the derivative and we have seen in Ex. 2, and 3, 3 4'11, p. 73
that every function

is

not derivable for every value of jc.

For example, we know that x is not derivable at JC=0. The


curve y= x Cannot therefore possess tangent at (0, 0).
This fact may be seen directly from the graph Fig. 8,
16 also.
]

1.

Find the slope of the tangent to the parabola

p.

Ex.
point

thS

(2, 4).

Derivative of x* is 2x and its value for


required slope of the tangent is 4.
angle

y=x 2 at

x=2

is

4.

Ex. 2
Show that the tangent to the hyperbola y^ljx at
with x-axis.

3*r/4

Hence the
(1, 1)

makes an

DIFFEBBNTIAL CALCULUS

78
Ex. 3.
points (-1,

Find the equations of the tangents to the parabola y=x* at the


and (2, 4).

1)

4*16*
Expressions for
in a straight line.

and acceleration of a particle

velocity

Every thinking person is aware of


and Acceleration of a moving point. The

moving

the concepts of Velocity


In practice,
difficulty arises in assigning precise measures to them.
velocity at any instant is calculated by measuring the distance
travelled in some short interval of time subsequent to the instant
under consideration. Ihis manner of calculating the velocity cannot
clearly be precise, for different measuring agents may employ different
In fact this is only an approximate value
intervals for the purpose.
of the actual velocity and some approximate value is all that we need
in practice. The smaller the interval, the better is the approximation
to the actual velocity.

In books not employing the method of Differential

Calculus,,

velocity
any instant is defined as the distance travelled in an
Now there exists
infinitesimal interval subsequent to the instant.
no such thing as an infinitesimal interval of time. We can take
intervals of time as small as we like and in fact interval with duration smaller than any other is conceivable. The definition as it stands
A meaning can, however, be attached to the
is thus meaningless.
above definition by supposing tluat the words 'velocity' and 'infinitesimal' in it really stand for approximation to the velocity' and 'some
short interval of time', respectively.
at

The

meaning to the velocity of a moving particle at any


by employing the notion of Derivative.
Expression for velocity. The motion of a particle

precise

instant can only be given


4-161.

along a straight line


equation

is

analytically represented

by a functional

where, 5, represents the distance of the particle measured from some


fixed point O on the line at time /.
Let P be the position of the particle at any given time /. Let,
some short interval 8t, and let PQ=8s.
again, Q be its position after
ratio 8s/S/ is the average velocity over this interval and
know intuitively
is an approximation to the actual velocity at P.
smaller
be
obtained
will
better
considering
by
that
approximations

The

We

values of 8t .

We

are thus led to define the measure of the velocity at


J>eing equal to
f

lim

Hence,

if v

8s

T
6t

ds

denotes the velocity,

i.e.,

fr
dt

we have

P as-

DIFFERENTIATION

19

Expression for acceleration. Let v be the velocity at


t, and lot v-\-Sv be the velocity after some short
interval of time St 8v is the change in velocity during time 8t.
4*162.

any given time

8t

The ratio SvjSt is the average acceleration during this interval


and is an approximation to the actual acceleration at time t. The

smaller values of St will correspond to better approximations for the


acceleration at time t
We are thus led to define the measure of
acceleration as
.

7
Jim

Bv

dv

--,

i.e.,

dtf

Ex.

Find the velocity and acceleration


each of the following cases

1.

(//) initially,

in

(a)

j=i*-f2f+3.

Ex,

2.

function of/

Ex.

A particle
;

3.

(/')

at

the end of 3 seconds,

(b)

(c)

moves along a

prove that
Ff ,y-/ 3

s= \l(t -{-!).

its

straight line such that


acceleration remains constant.

5=

is

a quadratic'

-2/ a -h3f-4, give the position, velocity and acceleration

the particle at the end of 0,

1,

of

2 seconds.

4-2.
The remaining part of this chapter is devoted to determining the derivatives of functions. Home general theorems on differentiation which are required for the purpose will- also be obtained"
To provide for illustrations of these general theorems,
4*3.
in
we obtain, in this section, derivative of x a where a is any real'

number.
4-21.

Derivative of constant.

Let

V=r,
where,

c, is

a constant.

To every value of x corresponds the same value of y, so that theincrement Sv, corresponding to any increment Sx, is zero.
*y

8x

dy

..? -o
8.x
..

By

...

dc

Note. Looking at derivative as the rate of change, this result appearsalmost intuitive, as the rate of change of anything which does not change i
necessarily zero.

The result may also be geometrically inferred from the fact that the sloper
cf the tangent at any point of the curve y=c, which is a straight line parallel to*
x-axis, is 0.

4-22,

Let

Derivative of

xa where a

is

any real constant number.


"*

90

DIFFERENTIAL CALCULUS
Let 8y be the increment in y corresponding to the increment 8x

in x.

y+8y=(x+8x)*

"

Sx

8x

(x+8x)-x

'

lim
(

3-65, p. 63)

Hence
d(x

=a x^
Vdx

where a

is

any

real

number, rational or

1,*

irrational.

Another method. The derivative of xa for rational values of a


can also be obtained without employing the general limit theorem of
3-65,

Case

I.

Let a be any positive rational number, say, p\q*

Here

We write
Then

z+8z=(x+8x)
8y

8x

Let 8x ->

so that Sz also -> 0.

l lq

or

(z+Sz^^x+Sx.

SOME GENERAL THEOREMS ON DIFFERENTIATION

81

= axoc-1
Case II. Let a be any negative rational number, say
We have
being both positive.
Sv
oy

8x
Writing

Sx

Let
dy

Jx

p,

v~*Pl4
x

fYJL&v-\~~PlQ
(x-^-bx)

~~

p\q

Sx

z=x llq

and z+Sz=(x+Sx) ^, we obtain

Sx ->

so that Sz ->

""

z*.z p

p'

^1

As

also.

'

a-2^"-

before,

get

9
1

Ex.1.

(/)

()

dx

ar

4*3.

Some

general theorems on differentiation.

Derivative of the
4-31.
of x.
functions
able
derivable

Denoting their sum by

y,

sum

we

or difference.

Let

w, v

be two

write
...(i)

Sw, Sv, 8y be the respective increments in u, v, y, corresan increment Sx in x so that x, u, v, y become x-\-8x,
to
ponding

Let

u-\-8u,

r+Sv,

y+Sy

respectively.

We

have

y+Sy^U+Sll+V+Sv.
Subtracting

(i)

from

(H),

JSj^ ==s

Sx

and dividing by Sx we get


SM

Sv
'

"SjT"^"8jc

...(//)

82

DIFFERENTIAL CALCULUS
Let Sx ->

0.

8y

to

,.
=lim

/Sw

8u\

(to+to)du

dy

We may

"

Sv

+lim S*

>

dv

dx^dx +

Sw

,.
1

ta-

similarly prove that

d(u

v)

dx

du

dv

dx

dx

Generalisation.
By a repeated
obtained above, it can be proved that
number of derivable functions and

"

application
if

u l}

t/

2,

of the results
u n be buy finite

then

dy

dx

We

du
= ~dx

duz

du 3

du n

~dx

~dx

dx

'

thus have the theorem

Algebraic

derivable functions is itself derivable


equal to the sum of their derivatives.

and

sum of any finite number of


sum is

the derivative of the

Ex.1.

_ 11

_
= 1 4.
-TV
/

(0

i '\

-i

(io

1
3

x -i- 1 _

Find the derivatives of

i*.
>|X

'

dx

dx

Ex. 2.

""""

(l-/o

__

DIFFERENTIATION
4-32.

Derivative of a product.

Let

y~uv
where

are

u, v f

Let

two derivable functions of

Sw, 8v, 8y,

be the increments in

ponding to the increment 8x in

We

x.

x.

y respectively corres-

u, v,

have

8y=u.8v+v.8u+8u.8v,
or

Let 8x ->
function,

is

Then 8u

0.

Sw

8v

8y

also

for

8v

w,

which

is

a derivable

continuous.

W
-limf
L
,.

*V

'

fix

Sv \

/
.

=hm(

ljc

H-to.
f
Sx

v.

when S* -,

'

,.

Su \

dx

..

)+hm( v.---)+hm3w.l,m
dv

..

,.

du

dv

dx

Thus we have the theorem


functions is itself derivable and its

The product of two derivable


is the sum of the twa
with
each
obtained
derivative of the
products
by multiplying
function
:

derivative

other.

The

first function
derivative of the product of two functions
derivative of the second -f second function x derivative of the first.

The

result

(i)

may

.-

y dx
Note.

It

also be re-written as

u dx

may be noted

dx

,
'

that the operations of differentiation and multi-

plication are not invertible.


l

'

d(u

e"

__
"^

v)

"dx

du
dx

dv

dx'

'

Cor.

1.

Generalisation.

number of derivable functions.

We first take

Directive of the product

of any finite

84

DIFFERENTIAL CALCULUS
dy

du,
Hi Un

"

du

On

by

dividing
I

du l

dy

dx

dx^U!

~if

By

y=u u2 u^ we

obtain
1

du z

u%

dx

repeated application of this result,


l^

dki

dy

27

,-a

-r~

du

dc

'3x~~

dx

dx

du

du z

u3

dx

we obtain

du^t

J_

^^

where y==w1t/ 2 w3 ...t/ n

Cor. 2. Derivative of cu, where c


vable function of x. Let

dy

is

TT

wr

a constant and u any deri-

t/w

Hence
du

d(cu)

2.

dx

t/x

Find the derivatives of


(0 (x+2)(3+x).

4-33.

(//)

(x-l-2) (2x:-3).

Derivative of a quotient.

Let
J>=n/v,
are two derivable functions of
for the value of x under consideration.
w, v

x such that

v is not zero

85

DIFtffiBENTIATION

V.SM--M.8V

T(v+Sv)"'

By

Sx

Now,

~
v(v

+Sv)

being a derivable function of x,

v,

is

continuous

Hence
as Sx -> 0.

Sv ->

Thus
dv

*du

<.:* =

d(

"~~

dx

v2

dx

so that derivative of the quotient of two functions

= [Derivate of Numer.

(Denomr.)

(Numer.) (Derivative of Denomr .)]

,/

d(
...

Ex.

^ dx
(1+^) ax
7

--

+x--)/

\l

Square of Denominator.

dx
(lJr x).lx.l

L 1+X*J
-[--i]

Obtain the derivatives of

2.

n\

'

^x -f~5

ft:\
v-*

r *)

x*

"f

Being a derivable function, v is continuous. Also, we have


value of x under consideration. There is, therefore,
an interval around x such that v^O for any point of the interval. (3*51, p. 54.)

Note

supposed

tf

at

1.

v=o Tor the

86

DIFFERENTIAL CALCULUS
Thus

if

ding value of

v,

we suppose x+ Sx
/.<?.,

to lie wittvn this interval then, the corresponjustifies division by v+gv in step (/)

This fact

v-j-gv^O.

Note 2. The importance of the results obtained above in 4 3 lies in


the fact that the derivative of any function which is an algebraic combination
[/.*., built up through the operations, of addition subtraction,
multiplication md
division of several others is itself expressible as an algebraic combination of the
derivatives of the latter.

34.

Derivative of a function of a function.

11= <l>(x), soy

is

y=f(u) and

//

a function of x,"*then

dy

dy
du

dx
i4r

'and<t> being derivable functions

du
dx

'

ofu and x

respectively.

Let Sx be any increment in x and Su the corresponding increment in u as determined from u = (f>(x). Again, corresponding to the
increment Su, in u let 8y be the increment in y as determined from
9

y=f(u).

We

write

Sy

Let Sx ~>

Su

Sy
8u

8x~~~

Sx

so that Su -> 0.

n
nm
Sx ->

( _ S ^y \\

/ Sy
-rL,

nm

lirv,

\ Sx J

sx

->

lim

Su

8>L

->

6//

Su
.
'

Sx
lim

Sw

5x

8u
.

S-^

>

Hence
du

jdy ~~ dy

dx

The

functions so that

dy

dx
Ex.

(i)

dx

immediate generalization.

result is capable of

be three derivable

'

du

>>

a function of x,

is

du

dy
~ du

'

dv
'

dx

dv

Find the derivatives of

We

"

Hence

write

du

dy

= l +x 2

j=w2.
1

i
-

Thus

if

we have

87

DIFFERENTIATION
or,

without introducing

u,

_
~

dx

(U)

Let

we have
'

</(l+x

dx

u=-~, y=u

du

</x~

(I-*)

-1)

(1-x)

__

J2

'

~~(1

x)*

Hence
dy

~~

C/A:

dy
du

du

~~

'

dx

/l4-x\

or, directly

l+x

Ex.
(i)

Find the derivatives of

(ax+b)

n
.

(M
{tv)

Differentiation of inverse functions.


4-35.
function derivable in its interval of definition.
admits of an inverse function

Let y=f(x) be any


suppose that it

We

as explained in 2*22, p. 21.

We

have to find a relation Jbetween/'(x) and

f '(y).

88

DIFFERENTIAL CALCULUS

Let 8y be the increment in y corresponding to the increment


8x in x, as determined from y~-f(x). The increment bx in x corresponds to the increment 8y in y as determined from

We have
,
1

Sx

Sy

Let 8x -

^y

by

d[v/rfx

dy

and dx/dy are reciprocal to each


y=x* when x = 2.

Differentiation of functions defined

4*36.

We consider

by means of a para-

two derivable functions

x =ft) y=9(t)
9

of

other.

Verify the theorem for

Ex.

meter.

0.

dx

Thus

8x

" A*

t>x

'*'.

Assuming that x=f(t) admits an inverse .function t=$(x) (2-22)

we

obtain

so that

By
(

4*34),

is

a function of x.

the rule for the differentiation of a function of a function

we have
dy

dx

dt

dy
*

dt

'

dx

dt

AI
Also

dx*=
Note,

Ex.

We

have

'f* is

1.

dt)

'

~dt~T\t)

called a parameter.

Finddyjdx, when

x=a/ 2 y=
,

89

DIFFESENTIATION
2.

Find dyjdx, whea


...

(0

I/'
,.
*- flt
,-6

2t

....

*=

*=

(ii)

,.

3at
8,

7-

2a/
.,

44. derivatives of Trigonometrical functions. The symbols


Bx and 8y stand for the increments in x and y and will always be
used in this sense without any frequent mention ^of their meanings.
4 41. Derivative of sin x.
Let
>>=sin x.
sin (x+8x)
8y
""

8x

sin

_x

S^

_2
""

cos

%(2x+8x)
Bx

sin |8x

sin iSx
gy-

=lim

cos

(x+^Sx)

UJ\

As

cos

AT

Also

is

lim

wo

a CDntinuous function,
lim cos

o
when Sx
->

r,

0,

sin iSx
- .2

hm

-^$ O^t

when Bx ->

have,

when Bx

0.

-> 0,

= 1.
,

dy -=
,-

cos.v.

Thus
d

(sin
x)y
v

dx

cos x.

Derivative of cos x.

4-42.

Let*

By

_cos (,V+SA:)~ cos x

'Bx^

SAT
2 sin |(2x
-

_
=
As

sin

A: is

dy

sin

-sin

sin iS
.

$A

a continuous function, \ve have,


lim sin (x+^Sx) = sin x.

= hm
,.

i* v, iC- sm x +iSx)] hm
.

a?->0

+Sx)

s^

sin

xl

sin x.

1-

when Bx ~>

90

DIFFBBBNTIAB CALCULUS
*->(Cos x/)

Tk
Ihus
Ex,

Let

We

smx.

Find the derivatives of

1.

sin 2x.

(/)
(i)

j;

= sin

cos 3 x.

(ii)

(Hi)

^(sin

2x.

write

= 2x,

dy

so that >>:=sin w.

du

dy

or briefly

~~=

d(sm 2x)
(ii)

Let

We

write

2x\
d(sm
~
}

d(2x)

- cos

2x

2==2 cos

.y

t/=cos

.y=w 3

so that

du

rfv

rfy

A:

3 cos 2 JC

)=

sin x.

or, briefly

__ </(cos x)
~
dx

dx

=3

Let^^V

(Hi)

We

write

(cos x)

rf(cos x)

x(

rf(cos x)
*

~d(Qo* x)
sin x)

dx
3 cos 2 x

(sin

v=sin

u=<\/x=x^

u,

j= v/v=v2.

so that

dy

dx

~~

dv

\v
""

du

dv

dy
'

du
*.

dx

cos w

x~~^

COS <\/X

VC^ V*)
11

V^*

or, briefly

d\/(sm \/x) ~ d\/(sm \/x)


dx
d sin \/x

= | (sin
sss

JL

v^*)""""

cos
*

dsin y/x

cos

V^- I*

sin

91

DIFFERENTIATION
Find dy\dx for /=7r/2, when

2.

x=2

cos

y=2

cos 2/,

sin

sin 2t.

We have
sin

^=--2
-^ =2

cos

(cos 2/)2

dy dx
dy
dx ~~dTj ~dt
,

Putting

= 7T/2,

z=r

=2

~2(sin

sin 2/,

2 cos 2/>

cos 2/)
sin 2/)

2(cos

cos

/+2

2 sin

/-(-sin 2/)(2)=

we obtain

Find the derivatives of

3.

(0 sin x,

(//)

sin x

(1/1)

m
(/>)

cos 8

--*,

(v)

cos mx,

(vi)

.f

cos

(v/0

sin^x. cos x.

<v//7)

4.

Find dy\dx when


y

(/)

(i/y

x=a (cos /-f sin /), y=a


x = 3 cos r~ 2 cos 3 /,r=3
x-a cos 3 /, ^- sin /.
/

sin

(sin
/

cos

2 sin

/).
t.

(///)

4-43.

(D.U. 1955)

Derivative of tan x.

Let

>=tan x

(x

tan x

Sx
sin

+ Sx)

cos

(x+Sx) sin x
(x+Sxj"~cos x
Sx

cos x
~ sin (x+Sx)
Sx cos

cos

(A:

(x 4 S^j

sin (x-4-S.x
x)
8x. cos (x+ox). cos
*

cbs~fx

x
sin

+8x)

cos :f

+8x)
x

cos

Sx

sin

DIFFERENTIAL CALCULUS

= cos
J_._L.l
=8ec*x
x cos x

dx

Thus
Or,

we

x.
^>=sec*
dx

write

x= sin

}>=tan

cos

cos x.

x
x

,,

(sin x)^

sin x.

-,

dx

dy
dx~~

so that
(cos x)'

j
dx

cos lc

cos x. cos

x+sin

x. sin

cos 2
1

"cos 2
/

4-44.

proof

Ex

1.

(i)

tan x-f cot x.

(iv)

cosec 2 x.

Find the derivatives of

tan x

cot

sin x. tan 2x.

(/i)

;c

(v^

tanx-t-cotjc

4-45.

x.

to the reader.

Its

is left

=sec 2

tanx\

A //I
\
VJ

x tan x cot 2x.

(Ill)

i.

\l + tan x/

.,

(v/)

A
A

//"I
/ i 1i

y V

Derivative of sec x.

Let
>>=sec jc=-

COS

1
" Sy

cos_(x+8x)
~
cos

cos

cos

"Sx. cos (Jt-f6x). cos

2 sin | (2x+$x) sin


bx. cos (x+S.x) cos
1

Sx

x
sin i

OX)
1

COS

COS

-=tan x

sec x.

8x

D CFFEK ENTI ATIO N


d(sec
~ x)

Thus
Or,

dx

we

-=tan x

dy
-/-

dx

cos

sm x)
--=tan x

s-2

is left

so that

x 01
---= cos
cos

dfcosec x)
'=
-^-,

Its proof

x cosec

cot

sec x.

x.

to the reader.

Find the derivatives of

Ex.
(/)

sec x.

write

v=

A A*
4-46.

93

cosec8 3 x.

(//)

(Hi) secVfc-f&x).

(/v)

>l[sec

(ax+b)].

sec (cosec x).

Darivatives of inverse trigonometrical functions. The precise


4*5.
definitions of inverse trigonometrical functions as given in
2*7, p.
30 will have to be kept in mind to obtain their derivatives.
4-51.

Derivative of

sin"" 1

x.

Let
1
^=sin"" x so that x=sin y.

dx

4y
j^

__
^

where tha sign of the radical

By

sin"" 1 x,

the def. of

is

we have
1

A:<'7r/2,

i.e.,

positive.

Hence
4-52.

it

the same as that of cos y.

is

7r/2<sinso that cos y

_
,,

fi

Derivative of cos-

^/(TT^p
x.

Let

y^cos- x
1

dx

so that x^=cos y.

7r/2<><7r/2

DIFFEBENTIAL CALCULUS

94

dy

f\f

_.__

_.

where the sign of the radical


def. of cos"" 1 x, we have

<
Also

if

lies

4-53.

.._

and

dfcos- 1 x)/

same as that of sin

is .the

cos" 1 * <TT

between

Hence

,
I

then sin y

TT,

=-

---,-

<

i.e.,

is

<

By

y.

the

TT.

necessarily positive.

Derivative of tan- 1 x.

Let
"1

dx

so that

x=tan

y.

or

4.54
454

d(cot-

dx

x)_
~

J^
1+x*

Its proof is left to the reader.


4-55.

Derivative of see" 1 x.

Let

y=sec~ 1 x so that #=sec


dx
j =sec v tan v

y.

dy

dy
dx

or

sec

_
We take,

y tan y

--^-

c*f\f* i^ //afvr>2
2
sec
^ V(8e

s ig n before

+>

S"

dx

x)

1 \

1)

-_

// 2 x-^ VCx
1)
>

-*>2

'

the radical and write


(Refer note below)

x7(^)

Thus
d(sec~

J'

1
'

95

DIFFERENTIATION

Note. This note is intended to show precisely what sign should be


chosen before the radical. Now it is clear that the sign before the radical is
the same as that of tan y. By the definition of sec~ T x. [Refer 2*75, p. 33],
we have

When x
ween

is

positive so that it lies in the interval


is positive ;

[1,

oo].

then y

oo,

1],

lies

bet-

then y

lies

and n j2 and so tan y

When x is negative so that it lies in the


between w/2 and n and so tan y is negative.
Thus the sign of the
Hence

radical

is

interval,

positive or negative according as

is

posi-

tive or negative.

:r"=
2
^
dx
x^(x*l)

if

*>

and

-------,/-

~ir

if

<-

jH(x -l)

so that

dy

_._

=-:

ctx

x
I

Thus

*\

[x~

for every admissible value of x.


1 )

__~

dx

_
2

|x| V(x -!)'

Derivative of cosec- 1 x.

4-56.

Let

dx
-j-

or

dx

cot

x=cosec

so that

y.

y cosec

cosec

cot

__

-1

cosec 2

We take,

+,

-1

yl)~-^

x\7(x

-1)

sign before the radical and write

"

dfcosec" 1 x)
-

Thua
Note.

By

dx"
The

and

nr/2

when x
ween

7T/2

and

xV(x-iy

sign before the radical

the definition of cosec

When x
ween

*,

is

the

same

as that of cot y.

we have

is positive so that it lies


and so cot y is positive

in

the interval

[1, oo],

then y

lies bet-

then y

lies bet-

is

negative so that it lies in the interval


and so cot y is negative.

[oo

1],

DIFFERENTIAL CALCULUS

96
fckw

Thus the

sign of the radical

is

positive or negative according as

=^br

and
so that

we

dKcosec^x)^^

Thus
Ex.

(v;

'

posi-

if

x<0

>

for every admissibl e value of x.

1_

Find the derivatives of

1.

sin-H*.

(i)

is

can write
l-

(//;

tan-H(l +*)/(! -x)].

(i//)

Hence we have

tive or negative.

sec
>t

tan-Xcos Vx).

(/v)

/ X-

Jt!~~l\

(v/iFcos-

l4x
,

Ex.

2.

4-61.

Find

when

rfy/d*

Devivative of

(, x are both

log,, x.

positive)

Let
J>=log x.
tf

Now

Cor.

lim (

--^

Let

a=e

1+

X
* )
^

=e.

so that

^=log^ x=log
f

_
-.,
rrs"^

dx

iut&
6e

3-63, cor. 3)

D1FFBBENTIATION

JL
*&*_
x
dx

Thu8
4-62.

Derivative of ax

Let

a*.

y
~x -f ox

a
=
&x~
by

dy

2~*

x
ax

8x

,.

8x

&x

-,

1
*

8x

-l

ii? """&T~

=a*

(3-64, p. 63)

log^a.

Thus

-f-J=a-

Cor.

Let a^=e so that ^=6^.

Thus

-fdx

Find the derivatives of

Ex.

97

(/)

log sin x.

(//)

(/v)

log[sin (log x)].

(vi)

log tan (Jx -hj").

cos (log

x).

(Ill)

(v)
(vii)

sln *

logV(#

-I y.

(V//0 -,-~r'
log X
(x) logioCsin"

(x/v)

('*>

^1 ).

w *+e- w;r

(jci")

).

eV^7

(jci//)

(jrv)

v \
V(

fl

log[x+

)/

V(jc

a** sin 1*.

a+b *
x
los-a b tan x

Derivatives of hyperbolic functions.


4*71.

-t-;t4-l).

log(sec x-ftan x)

0^X

(xi7) log(e

Derivative of sinh x.

Let

e*-e~*
*

98

DIFFBBENTIAL CALCULUS
m,

d(sinh x)

Thus
4-72.

===coshx *

d
Derivative of cosh x.

Let

x=

e -e._.__

dy

d <co h
*)=sinh
;
dx

Thus
4-73.

!t

x.

Derivative of tanh x.

Let
,

>>=tanh

x= sinh
-,

cosh

^(sinh-- x)

cosh x

-,

d(cosh
'
- x)

~smh x
.

t/x

T
Jx

""

cosh 2 x
cosh x.cosh
"

cosh 2

sinh 2 x

_ ----------

cosh 2
_.

Thus
A

*A

4-74.

xsinh

.^ ---

dx

-=

^-^dx

cosch 2 x.
ft

Its proof is left to the reader.

4*75.

Derivative of sech x.

Let

v=sech

dx

^
""

cosh

1
x=~cosh
vx

xO

sinh

l.sinh

cosh a x

^^ _- S ech10

cosh 2 x

.
d(tanh
x)/
~
=sech29 x.

d(coth x)
/

x.sinh

cosh 2 x

x.

DIFFERENTIATION
d <S

Thus

-*>=-tanhxsechx.

dx

An*
4-70.

d(cosech
x
x)<
-_

Its proof

is left

..

coth x cosech x.

to the reader.

Derivatives of inverse hyperbolic functions.


4-81.

Derivative of sinh- 1 x.

Let
3^= sinh- 1 x so that x=sinh}>.

dx
dy
-A:

or

where the sign of the radical is the same as that of cosh y which we
know, is always positive, ( 3'72, page 68).

^
Hence
4-82.

dfsinhv x)
-'

...

dx
Derivative of cosh- 1 x.

Let

x=cosh

1
j^cosh""" x so that

y.

dx
dy'

^_J_
Gin
n
Ol
1111

//~Y*
t*^V

v
X

"

A /^rr^ftri*
tV ^^ A
1
I v/v/Oll
\/
?

^
i

H /iI

^^

I**
*Hs

^^

where the sign of the radical is the same as that of sinh y.


Now, cosh- 1 x i.e., y is always positive so that sinh y
(3-71, page 68).

'

4-83.

ll00

^^^

Hence

Derivative of tanh- 1 x.

x
\

<1]

Let

j^tanh- 1 x

so that

x=tanh

y.

dx

l^l

Thus

T~- =sech* "


V,
dy
1
1
dy ^.^
dx ~sech 2 >~ 1-tanh2
_r_ _

._

--=.- -----

dx

= *1

Its proof is left to the reader.

x^a

>>

~~l-x*

1
*

is

\I

positive

100

DIFFERENTIAL CALCULUS
Derivative of sech- 1 x.

4*85.

Let
1
>>=sech~ x so that x=sech

sech y

-j

sech y tanh y
--1

where the sign of the radical

is

is

-1

the same as that of tanh y.

But we know that sech- 1


tanh y

y.

dy
dx

or

tanh

y.

x,

i.e.,

always positive, so that

is

always positive.
dteechr 1 x)
--- J _.

_,

Hence

._!.

dx

xyXl

Derivative of cosech-

4*86.

_
1

x.

Let
1
>?=cosech~ x so that x=coseoh y.

dx
T

coseoh y

or

cosech y

sign of the radical

Now,
as

is

y,

coth

____

^cosech y

when the

,.

coth y.

is

________
2

=,

^(cosech y~+I)

x<\/(x*+l)'

the same as that of coth y.

and therefore coth y

is

positive or negative according

positive or negative.

*2?!*^x) -

TH
rhus

dx

ifx<0.

=1
VO

for all values of x.

Find the derivatives of

Ex.
(i)

log (cosh x),

4-91.

(//)

necessary to take
is

slnh2

*,

Logarithmic differentiation.

v
a function of the form u

which

known

its

(///)

In order

tan

to

tanh x.

differentiate

where w, V are both variables, it is


logarithm and then differentiate. This process
,

as logarithmic differentiation

is

also useful

when the

101

DIFFERENTIATION
function to be differentiated

The following examples


Ex.1.

Differentiate

log

dy

dx

,-

x
sln x

^ = 16g

we

Differentiating,

x sin

=x*

Let

(x

Hence

=:COS

+ Sin X.

,
.

log
&

JC

y=x *
l

log

+ (sin

Differentiate [x

write

log x.

get

tan *

Ex. 2.

*+(sin

sin x\
x -----

-\

cos x

x)

].

cos

*.

x)

tan

Let

t/=;c

and

v=(sin x)

so that

..

*,

(1)

cos

dy _ du

From

(1),

we

*,

dv-

obtain, taking logarithm,

w=tan x

log
1

du

dx

log

we

tana;

/
f

^ec

log

v=cos x

</v
,

--==

dx

sin

dx =(sm x)

3.

(3)

and

log

x+

tanx

J-

log sin x.

1
.

cos^/

dv

Adding

obtain, taking logarithm

*-e.,

rfx^^
(2),

;c.

21logx+tan x

=sec 2 x

rfw

j.^.,

From

...(2)

dx^dx

rfx

Ex.

)=sin x

dy = sin * /
~
x
(cos x

TT

We

the product of a nuniber of factors.

is

illustrate this process.

(4),

jc+cosx.

^-sin

we obtain

Differentiate

log
a sin

log sin

sin

COBJC

/ov
...(3)

DIFFERENTIAL CALCULUS

102

Putting

it

equal to y, and taking logarithms,

logx+| log (l-2x)-f


Differentiating, we obtain

log }>=

dy

log

-2

(2-3*)- J

l-2x~~

log (3-4x).

-Z'

we obtain

2-"3x

3-4x

_ __

+5(3-.4x)
9

2JC

Ex.

(/)

4.

Find the

(cos x)

(1

(3-j?)
sin

16

"I

+ 5(3-4x)" J'

(in

tanA:

-f(cotx)

8/4

x
.

(4-^ 4 ) 4
log x

'

*(v)

(log

;c)4(sin-

xx

/C

-*)V (2-x)/

(vi/j)

differential co-efficients of

cot

"

log

(/vfaanx)
v

^3(l-2;c)

+ 4(2 -

Preliminary transformation. In some cases, a preliminary


[transformation of the function to be differentiated facilitates the pfocess of differentiation a good deal, as is illustrated by the following
4*92.

examples.

Ex.

1.

Differentiate
Sin

!+-*
Putting

x=tan

we have

6,

Yr
1

-|-t/an

dy

~dx

-!+*'

*'

Ex.

2.

=Bin-i

(sin

1
2^)=2^=2 tan-

2_

Differentiate

Putting x=cos

0,

we have
2

^'

=V2
'

cos

x.

DIFFERENTIATION

103

=
e
COS x

f\

cos

'2 sin

l\

+sin-

rt

l-tan
l+tan-2-

7T

7T

J.

Ex.

F/wJ /A^ differential

3.

tan~ l

,
JL

of

to sin" 1

2x

x=tan

.*
J.

_
Putting

r^c/

wft/i

*~2
X

co- efficient

0,

we

(P.C7. 1954,

1956)

2y

_ sin-i

see that

2
-

J*

(tan 20)

2= sin~

z=sin- 1

( sin

= 2^=2

tan-1 x.

tan~i x.

26)^20=2

dy^

dy ,dx

dz~dx dz~
Also otherwise, we have
y***>

dy

j- ==1.

so that

Ex.

4.

//

V( !*)+ V(l -y*)=a(x-y), prove

Putting x=sin 6 and y=ain


sin

that

(D.U. Hons. 1949


<f>,

we have

+ *>(lBm*

)=a(sin B

sin

P.U. 1952)

DIFFERENTIAL CALCULUS
or

cos

0+cos ^=0(sin

sin

<f>)

2 cos |(04-0) cos \


~
2 cos i(0+<) sin~ (V

or

<

^(0~<)= cot- 1 a

0^=2 cot"

"**

sin^

Differentiating,

sin"

we

get

y=2

cot" #.

or
<fa

Ex.

5.

.(/)

.....
%

(ltt)

Find

tan- 1

tan-

the differential co-efficients

(>.t/. 1952)

j^.
Vx

of

(it)

..

sin~ l

Ov) tan

/\l*vos x\l/2

[Show that
Ex.

6.

this is equal to sin""

sin

Express in their simplest forms the differential co-efficients with

respect to x of

4 93. Differentiation "ab initio". To differentiate "ab initio"


from first principles, means that the process of differentiation is
to be performed without making any use of the theorems on the

or,

differentiation of sums, products, functions of functions etc., nor is


any use to be made of the differential co-efficients of standard forms.

We have already had numerous illustrations of it.


Ex.

1.

Differentiate sin-*x <ab

Let

^sssin-

initio'.

,*.

.y+Sy^sin- (x+Sx).

We have
and
so that

x=sin y
x + &t s= sin

(y

8>>)

Sx=sin (y+8y)~8my.

(D.U. 1955)

DIFFERENTIATION

105

Thus
Sx

sin

(y+Sy^&'my

Let Sx

(y+ %Sy)

~"
2.

F//irf,

cos

3;

/rom

>0.

~~~

cos

$8y

Vsin ^ Sy )

so that 8y also

-->

</x

Ex.

/
'

cos

^(F

~x*y*

first principles, the differential co-efficient

of

~sin*:~

Let

y~ Vsin

x.

\/sin x

'

Sx
sin

sin

(x+Sx)

= COS (X + iSx)
.

Let Sx->

\/sin

sin A Sx
TO

'

---- ------

0,

^V

=cos x

COS

.^

~~

Vsinlc
Ex.

Find, from

3.

sin

<'>

first

principles, the differential coefficients of

()

sin

x.

(i/O

Vx.

Exercises
Find, from

* 8)

*(

4-

tanx 8

first

principles, the differential coefficients of


2.

5.

>/(tanx).

sec^x.

3.

:-

(D.U. 7955)

106

J>^FFBBBNTIAlj CALCULUS

Find the

differential co-efficients of

6XT^iwo 3T

__

-.*.

7.

X*

b tan- 1

9.

tt tan
*
11.

*
i

tan- 1

~\

** --*

tan-*

12.

1+sin-jc

-I

..

O-fcOjH^*).

14

16.

8
a

.sin- 1^.

17. [l-x")"

22. 10

lo 8 sin

Mj

tan-11

26.

/7
V

23.

Y*
A-

-T-J

b + a cos x

IJX*
.

'

21. sin-

x
L./
f
t/ f*O^
V^^Jo

18. log [tanh (fcc)].

\/(f^H).

20-

(l-4x)
~

j^

x log x

log(log x).

f <^Q.
v^-V/O

^V

25. (sin
v
x)y

sin (x log x).

27.

^29.

&*
/ eaX
_"~_?_

sin^ 1

sin- 1

-;^^-^

31.

*^ v-v/k

X.

33. 9x4 sin (3 jc- 7) log (1-5.)

34. e aj?cos (^ tan- 1 x).

35. log [l-**i


x
37. xa sinh *.

cot x coth x.

^^

Find the

+)

differential coefficient of

^~ w sin
/

107

DIFFERENTIATION
42.

Differentiate
'

(0

,.,,

00

,2x+l

+ ^T tan

^3-

Find dyjdx when

43.
. .,

(i)

(/f)
(///)

sin 8 /

x =

cos 8

'a"-*/ 6

-^(cos2/)

'

x=sin H(cos 2/), y=cos /V(cos 20x=a(cos /-flog tan i/), .y=a sin /.

44.

If x* **<-*, prove that

JI5.

Differentiate sin*

46.

Differentiate

47.

Differentiate tan- 1

dy/dx^log

jc/(l

+log

x)

(P.U. 7955, 56)

(Differentiate logarithmically)

x with respect

sm x

(D.U. 1950)

to (log x)*.

with respect to (sin

(P.U. 1959)

x)*.

1
2
with respect to tan[{V(l -hx )-l)}/x]

(P.U. 1956)

when x=e

48.

Find

49.

Differentiate

with respect to cos- 1 x 2


50.

(D.U. Hons. 1949)

Differentiate (log x)

tan*

with regard to sin (m cos- 1

x.)

(P.

51.

Differentiate the determinant.

F(x)=

From

first

/,(*)

<fr,(x)

principles

we have

F(x+A)-F(x)

;-j
;

/<*)

AW

fi(x+h)-A(x)

*)

f&c+h)

(*+*)

U".

1955, 56)

108

DIFFERENTIAL CALCTJLUS

<P,W

Dividing by h and making h tend to zero,

we

get

<Pi
!

^*

^.

The

rule can be easily extended to the case of determinants of any order.

APPENDIX

EXAMPLES
1.

Show

that

/(x)=x

8 sin
(//x)

when

x^O

/(0)=0
is

derivable for every value

but the derivative

ofx

not continuous for

is

x=0.

(D.U. Hons. 1954)

For xr0,

sin

f'(x)=2x

coa(

o-sin
=2x

cos
.V

^ )(_

---

For x=0, we have


1

x'sin

_!1
fw-m^
x
x

=x

* _

sin

>

as

/'(0)=0.

x and

Thus the function possesses a derivative /'(x)


is given by
f'(x)=2x

We

have to show that


cos

=2*

sin

cos

for every value

when

f'(x) is not continuous for

sin

(2x
\

of

sin
*

x=0.

cos
-*

We write
...(1)

DIFFERENTIATION

109

Here
lim ( 2x sin

J=0.

y ^Q

'

In case

lim (2xsin --- cos


x
^

i.e.,

had

x-0
existed,

would follow from

it

'

that lim (cos 1/x) would also

(1)

;c->0

But

exist.

this

is

Henca lim

x-0
for

THUS

/'(*) doas not exist.

f'(x)

is

not continuous

x=0.
Examine

2.
(

not the case.

oo

oo )

the continuity
the
following function
/or
)

=l

and

derivability

in

the

interval

oo<x<0,

in

n x

in

-|7r)

O<X<|TT,
2

in

JTT<X<

oo.

(Mysore)

The function /(x) is derivable for every value of x except


parhaps for x*=0 and x = 7T/2. Thus we shall now consider x=0 and
X=7T/2.
Firstly

we consider x=0.

Now
lim
/(x)
-0)

and

lim
/(*)=
x->(0+0)

lim

Hence /(x)

is

/x=l=:

continuous for

x=0.

for

x<0,

Again,

so that

Also for

x>0
"""

1+sinx-^.
x

lim

(l+sinx) =

DIFFERENTIAL CALCULUS
80 that

lim
*->(OH-0)

M*~

x-KO+0) *

Thus
*

Hence the function

Now we

not derivable for

is

x-KO~0)

x=0.

consider x=irj2.

We have

lim

is

X ^(-|7T

0)

lim

Hence /(jc)

lim

/(x)=

lim

/(x)=

continuous for

Again, for

Putting

x=/ we

^TT

see that

sin

x~"

==

Isin
"" (JTT

cos

/)

so that

lim

|TT

lim
iii-u
(}n^

exists

and

is

equal to

0,

2 sin 2 \~t

_ sm
.

sin \

"""t

DIFFBBBNTIATION

111

Exercises
Discuss the existence of/'(x) and/"(x) at the origin for the function

1.

/(x)=x

sin -

when

(B.U. 1953

Examine the

2.

D.U. Hons. 1952)

differentiability of the function

m
/(x)= x

sin

when

x^r.0,

w>0

when x=0
at the point

x=0.

Determine

m when/'(x)

Determine whether /(x)

3.

is

is

continuous at the origin.


(D.U. Hons. 1952)

continuous and has a derivative at the origin

where

Show

4.

is

that

continuous but not derivable for


5.

x=0

and x=l.

Examine the function

as regards its continuity

and the existence of its derivative at the

origin.

(D.U. Hons. 1951)


6.

Discuss the continuity and

the differentiability of the function f(x)

where
o,

when x

is

W ^ en

irrational or zero

JC==

a fraction in

'

its

lowest terms.

(D.U. 1953)
1.

Find from

and/(0)=0

first

at a point

principles the derivative of f(x)

x=0 and show

that the derivative

=~
is

when x-^0

continuous at x=0.
(B.U. 1953)

8.

Show

that the function


9

/(x)=x{l-fisin(logx )} for x^0,/(0)=0


but has no differential co-efficient at x=0.
continuous
everywhere
;

is

(B.U. 1952)
9.

If /(*)=*

tan- 1

ous but not derivable for

when x^O and /(0)=0/show

x0.

that /(x)

is

continu-

112

DIFFERENTIAL CALCULUS
10.

Is the function

f(x)**(x-a) sin

continuous and differentiate at

x=a

_-- for

Give your answer with reasons.


(P. U.)

11.

f(x)

is

Discuss the continuity of f(x) in the neighbourhooi of the origin

defined as follows
(/)

f(x)=*x log sin x for x?*Q, and /(0)=0.

(ii)f(x)=e
12.

5x

when

function f(x)

4whenO<x^l,

11 *

is

to 4x a

when x^O and/(0)=0.

defined as being equal

3x when

l<x<^2

(D.U. 1955)
2

Jo

discuss the continuity of f(x) and the existence" of f'(x) for

x
when ^0, to
3x+4 when x=2
,

and to

x=0,

and

2.

(D.U. Hons. 1957)

CHAPTER V
SUCCESSIVE DIFFERENTIATION
5*1.
Notation. The derivative f'(x) of a derivable function
f(x) is itself a function of x.
suppose that it also possesses a
derivative, which we denote by/"(;c) and call the second derivative of
The third derivative off(x) which is the derivative of f"(x) is
f(x).
denoted by/"'(jc) and so on.

We

Thus, the successive derivatives of f(x) are represented by the


symbols,
/'(*),/"(*), ........ ,/"(*), ......

where each term

is

the derivative of the preceding one.

dny\dx*

Alternatively, if y =/(*), then


derivative of y. Sometimes

also

denotes the

rtth

are used to denote the successive derivatives of y.

The symbols

denote the value of the nth derivative of y=f(x) for x=fl.

Examples
1,

Ifx=a

0+0

(cos

sin 0),

)>=0

(sin

00 cos 0),find

d*yldx*,

We have
-.

c/v

-V--

M0
rfy

~a (sin

0-j-sin 0-J-0 cos

=a (cos
rfy

~dx=~d~e
rf*v

Q)=a

0+0 sin 0)=0

cos

cos

0sin

0,

0.

_dy \dx ~ tan ^


'dt~dijd7

v-Y=sec
^~SeC 3 va

t/(9

^0

-v-

[Note this step]


sec s

1
-

--

"~~

"^~ rsz

cos d

113

ad

fl

114

DIFFERENTIAL CALCULUS
2.

Ify=sin

(sin x),

prove that

Tx

COS* JC==0

+y

(/>

'

1953)

We have
dy

Making

,
=cos (sm

x) cos x,

sin (sin x) cos

sin (sin x) cos 2

substitution,

we

cos

dy

Change

cos (sin x) sin

cos (sin x) sin x.

see that

^ x j- +}>
-j-^+tan
3.

cos*

#=0.

the independent variable to Q in the equation

d 2y

2x

dy

by means of the transformation

x=tan

We

6.

have

_dy
dt ~de

dd__ dyjdx _dy


dx ~~~de/de '"del

dy^

= -2 cos
rt

dd
Oane..
.

=-

2 cos 6 sin

2 sin 6 cos 3

cos 2

dy

'

+">*

dy

^T+ cos2

2>in
or

cos8

^--|-cos

dy
do

dO

d*y

n
'

'

'

j~

cos 2

dv
0.

^r

Substituting the values of x, dy\dx and


dififerential equation, we see that it becomes

or

U. 1932)

(P.

0^4-2

d ly\dx*

in the given

sin 0. cos 3

^j-|-co8

SUCCESSIVE DIFFERENTIATION

Ifax*+2hxy+by*+2gx+2fy+c=Q, show

4.

that

d*y

dx z
Solving the given equation as a quadratic in y, we get

x [2(h*-ab)x+2(hf-bg)]

Differentiating again,

for

Substituting

we get

dy

+h

,-

from

(1),

we

get the "required

result.

Exercises

;>~log (sin x) show that y9

r.

Show

^3.

that

y=x f tan x

/^/AO
--

V
.

satisfies

the differential equation

/;
If >~[(fl+te)/(c + <fc)] t then

If

5.

x2

cos /-cos It and

^=2

sin

/-sin

2/, find

the value otd*y!dx*

when =i*.
y\

If
(f.

'

#=a

sin 2^ (l-fcos 20), >>=a cos 29 (I

If ^-(I/A:)*,

show

that

^ (1) -0.
%

'4s
,

\If p

^^ 1 cos 2 9+6 a sin'fl,

prove that

-cos

29),

prove that

DIFFERENTIAL CALCULUS

116
9.

.J*

1 j.

lfy**x log Kaxy-i+fl- 1 ], prove that

If

*=sin

/,

y=sin />/, prove that

Change the independent variable to

^-4 cot x

~J~

z in the equation

+ 4>> cosec*x=-0

by m^ans of the transformation

z=log tan
12.

of x and

If y is a function

jc<=l/z,

i*.

show

that

dz*'
.

and

Show

that

dx

__

~dy

~dy$x

find the value ofd*xfdy* in

'

""
tf^^^

dy*

terms of

4vA/Jc, d*yldx*> d*y[dx*.

Also show that

dx*

Calculation of the nth derivative.

5*2.

Let

5-21.

dy*

Some

standard results.

y=(

so that in general

jV=w(m
In case,

is

a positive integer, yn can be written as

7--

(m

w)

m
60 that, the /wth derivative of (^x+fr) m is a constant viz.,
\ a
and the (m-fl)th derivative along with the other higher successive;
derivatives are all zero.

SUCCESSIVE DIFFERENTIATION
Cor.

1.

m=

Putting,

1,

we

,117

get

JV(-lX-2) ...... (-n)a"(

dx"
Cor.

Let y=\og(ax+b).

2.

a
yi== a

dx*
5-22.

(ax+b)*

Let

y l --=ma mx

log a,

so that, in general

Cor.

Putting *e for

a,

we

~~

dx
5-23.

get

7*

'

Let

jsin (ax+b).
y =a cos (ax-\-b) = a
l

J 3 =a

cos

sin (tfx-f 6-f |TT),

(ax+b+i7i)=a*

sin

(ax+6+|7r),

o that, in general
d*

5-24.

sin(ax+b)

Similarly

r
d*cos (ax+b)
<=a wn cos
j\
dx*
L
.

5-25.

mr"|

Let

y=ea * sin

(6^:+c).

ax
y^=ae sin

*=eax

\a sin

(6jc+c)+e

a*

6 cos

ax+b+
,

DTTI
^~
2 J'

118

DIFFERENTIAL CALCULUS

In order to put y in a form which will enable us to make the required generalisation, we detarmine two constant numbers r and
f such that

a=r cos

b=r

V,

sin ?

r^i/
Hence, we have
ax

Thus y l

arises

increasing the angle

Thus

from

by

>>

y l =re sin
on multiplying by the constant

the constant

and

<f>.

similarly

y 2 ==r*e

ax sin

Hence, in general
d w 1[e * sin (bx-f c)]l
a*
-----^V___r_y ==r e sm

,.

fl

where
5-26.

Similarly

ffi!.fll)].

m.

+ W)4^

cos

tan-^

(bx+c-fn

1.).

Determination of nth derivative of Algebraic rational funcIn order to determine the nth derivative of
Partial Fractions.
rational
function, we have to decompose it into partial
any algebraic
fractions.
5-3.

tion.

Sometimes it also becomes necessary to


theorem which states that
(cos

where n

is

0/ sin 0)

any

=cos

n0i

sin

n()

integer, positive or negative,

apply

Demoivre'a

andi=\/(

l).

Examples
1.

Find the nth derivative of

Throwing

it

dr

into partial fractions,

_ __

(x+2)(2x+3)~ 2 L
xa
"I

we obtain
"
1

x+2 "2x+3j.

''

2(2*
8

SUCCESSIVE DIFFERENTIATION
Find the differential

2.

119

of

co-efficient

We

have
x2 + a2

(jc

-~~
2

form,

+ <r/j(x

a.

-+

f
L*

J
-

^x + a/

To render the result free from T and express the same


we determine two numbers r and such that

x=r

-r-rr?
n+1

and

^--

cos

9,

a=r

sin 9.

'

nl

"i i v 1

j=

.(cos
v

0-\-i

(cos

= ^-ii

sin 0)y

sin

t cos

Hence

x
~"

n
(-l) n!

cos(K-fl)fl

rfx"

where

r=\/(x

+
1.

Find the

-f-0

),

0=tan- 1

(ajx).

Exercises

/ith differential co-efficients

of
^.^

2.

in real

Find the tenth and the wth differential co-efficients of

x+1

DIFFERENTIAL CALCULUS

120

Find the nth differential co-efficients of

3.

(//)

X
Show

4.

that the wth differential co-efficient of


1

is

K-l)w ("

!)

sin

<l

+ 0[sin (n 4- 1)9 -cos


l

(w

+ (sia

Prove that the value of the nth differential co-efficient of


if /i is even* and
(
!) if n is odd and greater that 1.

5.

for

0-fcos 0)-*- 1 ]. where fl^

x=0 is zero,
Show

6.

that the wth derivative of .y^tan- 1

(-l)n-i (w -i)

is

w (Jn-^) sin' (iw+^).

s in

(P-U.)

(P.U. 1935)

Find the wth differential co-efficients of

7.

_..

If ^=tan-i x,

8.

ton -,

show

If

y=jc(x+

provided that

log

cos a

(jc-f I)

cos

3
,

[n,y

+(-!)

cos"^.

prove that

/*

If >;=A:

10.

1)

Jf

that

=(n-l)
9.

....

A:

log

-~, prove

that

X~j~ I

The nth derivative of the product of the powers of sines ai*4


In order to find out the nib derivative of such a product,

5-4.

cosines.

we have to express it as the sum of the sines and cosines of multiples


of the independent variable. Example 1 below, which has been
solved, will illustrate the process.
Ex.
(i)

i)

Find the nth differential co-efficients of

() #*

cos* x.

We know that
1

+cos 2x

cos 2

sin x,

(/>./. 1951)

SUCCESSIVE DIFFERENTIATION
cos *

121

2x

2x +1(1+008

1+4

-.
)

cos

cos

4x

*+- +4

cos

x=4(l + cos

cos 2* sin

2x+J

cos

2x) sin x

= J sin x+J.2 sin X.cos 2x


x+J (sin 3x
= J sin x+J sin 3x.
==| sin

sin

Hence

sn x

sin

-(

x)+J

sin
sn

e
(e

sn
sin
2.

Find the nih differential co-efficients of


sin 8Ar.

(/)
(//)

sin

(v) e

2ay

(i/)

x cos
cos

A:

x.

cos x cos 2x cos 3x.

(/v)

e* sin 4 x.

sin 2 2x.

55. Leibnitz's Theorem. The nth derivative of the "product of


two functions. If w, v be two functions possessing derivatives of the
nth order, then

This theorem
Step

I.

By

will

be proved by Mathematical induction.

direct differentiation,

and

we have

(wv),j=w a
t/

Thus the theorem

is

true for

C 1t/1 v 1

n=l,

2,

Step II. We assume that the theorem


value of n, say m, so that we have

is

true

for

a particular

mC w
a w-a vJ + ......
t

......

+ m C muv m

122

DIFFERENTIAL CALCULUS
Differentiating both sides,
+i

v+u m Vj+^C,

um

B,

we

get

^+"0, M W _,

va

V^rC^C.,)

V2

!/_!

U^-OK^ V,+

...+

OT

+
CW

But we know that

from which we see that if the theorem is true for any value
then it is also true for the next higher value m + l ofn.
Conclusion.

for==2.

In step

Therefore

=3 + l,

i.e., 4,

it

and

I,

we have seen that the theorem


for /f=ii + l, i.e., 3 and

must be true

of

is

so

tt,

true
for

so for every value of n.

Examples
1.

Find the nth derivative of x* ex cos


2

To

find the ttth derivative of ;c e


as the first factor and x 2 as the second.
.-.

2
(x e* cos x) n ^(e

x.

(P-U-)

cos x,

we

look upon e cos x

cos x) n x*+ n C l (e* cos x) n .,.2x


-t-"C 2 (e

n x
2k e cos
.

(x+n tann

l).x

~ 1)

" 1

cos

2.

cos

x+n^l -J-) +w(w-l)

(x+^"2 tan-U).2

cos

//

.V=a cos

show

(x+n-lj.tan- l)2x

~2

.2*

+2?.nx cos f

cos x) w _ 2 .2

(/og

x)+& ^m

(fog x),

that

(D.C7. 1952)

12$

SUCCESSIVE DIFFERENTIATION

we

Differentiating,

get

or

xj^ 1

x)

x)+b

a sin (log

Differentiating again,

xy*+yi=*
or

b ??_C!.o 8^t
-

a sin (log
-----

__
Vl

we

cos (log x}.

get

----

^(xya+J'iH

a cos (log
x)
-

1> cos (log

x)
- b sin(log
*-

x)+6

sin (log *)];=

* j>3+*yi+:y=o.

or

Differentiating n times

by Leibnitz's theorem, we obtain

* aJ^2+"Ci-2x^,fi+^
or

x'y n
Exercises

Find the nth derivative of

1.

(i)

jc

e*.

(i/i)

2nax+n(n + l)].

4* [a*x*

(//)

x 3 cos

x.

(iv)

e" log

(P.U.1954,56)>
2.

If y=x 2 sin x, prove that

sn
3.

If/(x)=tan

x+-x

cos

x, prove that

[P.J7. 5w/?/>. 7936).

[Write /(x) cos x

= sin x and

apply Leibnitz theorem.]

Differentiate the differential equation

4.

dx*

times with respect to x.

Find the nth derivative of the

5.

(D.U. 1950)1
differential equation

Determination of the value of the nth derivative of a funcit is possible to obtain the value of the nth
derivative of a function for x =0 directly without finding the general
expression for the nth derivative which cannot, in general, be
obtained in a convenient form. Examples 1 and 2 below which have
5*6.

tion for

x=o. Sometimes

124

DIFFBRBNTIAL QALCIJTJS

been solved will make the procedure cl *ar. As will be seen in Ch. IX,
the values of the derivatives for x=0 are required, to expand a
function by Maclaurin's theorem.

Examples
1.

Find the value of the nth derivative of e m 8in

or

"w=*r

Differentiating,

get

(1-x 2 2y,y 8 2xy1 -2ifi f


Dividing by 2y t) we obtaain
(l-jffo-xyi^m'y.
)

Differentiating n times

Putting

From

x=0, we

(1), (2)

...

(1)

(2)

(3)

ox

x*)y*=nfy*.

(1

we

x0,

msiiTl x

*-

for

"
sin Ax

y=e m

Let

"1 *

by

Leibnitz's

theorem, we get

get

and

WO)HHm%

(3),

B (0).

we obtain

XOHl^OHm^COHin".
Putting n=l,

2, 3, 4, etc. in (4),

we

get

In general

'~

fm2 (2 a +/n 2 )(4 2 +/M 2 )...[(n-2) 2 +m 2], when

(w(lHJ

2
)

(3

+m )...[(n-2) +m
3

2
],

is

when n

is

even,
odd.

(P.U. 1955)
2.

If >>= (sin- x)
(1
x

2
,

prove f/wf

x 2 ')^-x^-2:=0
ax*
dx

...

(1)

Differentiate the above equation n times with respect to x,

Also find the value of all the derivatives of y for

x=0.
(P.U, 1955)

Differentiating

(sin-

x)

2 sin" x

we

get

*=

v (!-*).

9
,^>

SUCCESSIVE DIFFERENTIATION
.

(1

- x*) y* = (2

125

= 4y.

sin- 1 x)*

we get

Differentiating again,

2(l-x ^2-2x^=4^.
Dividing by 2y l9 we get
2
(l-* )>>2-*yi-2=0

...(3)

which

...(4)

is (1).

Differentiating this n times

by Leibnitz's theorem, we obtain

(1-^+2+^+1

n(

(-2x)+

~^y(-2)-xy ni

-ny n

1=0.

l-*

or

Putting

x=0, we

obtain

JWOHa^CO).
From
From

(2),

>>i(0)^0.

(4),

y,(0)

Putting H=l,

In general,

if

= 2.

(7)

3, 5, 7 successively in

w=2,

Again, putting

is

>> n (0)

4,

even,

= 2.2

,.,(6)

6 ...... in

(6),

we

(6),

we

see that

see that

we obtain

'4 ? .6 2

..

(~-2)

2
,

when w

Note. The result (5), obtained on dividing (4) by 2y t is not a legitimateconclusion when ^t=0, which is the case when x=0. Thus, it is not valid to
derive any conclusion from (5) and (6) for x=0.
,

But these results may be obtained by proceeding to the limit as x->0 instead of putting x=0. This may be shown as follows
:

We can

easily convince ourselves that the derivative of every order of y


x 2 ) in its denominator
as calculated from (2) will contain some power of (1
1 , so
that lim ^t
and will therefore be always continuous except for
and lim ^==^(0), as x-0.

x=

Exercises
1.

If

u~ tan~

x, prove that

and hence determine the values of the derivatives of u when


2.

x=0

(M.T,

If
1
j>=sin (m sin" x),

show

that

(1 -x')*, + ,=(2n f IkVn-n+V-'"')*,

and

find

yn

3.

(P.U.

(0).

Find y n

(0)

when

j>=*log

[x+

4(\

{ x')].

/P5>

126

DIFFERENTIAL CALCULUS
4.

It y=lx+4(I +**)]", find

5.

If

yn (Q).

ys~JM

*how

COS"" 1*

that
(1

*nd find yn (0).


6.

-x^+i-Vn + ^xyn^-W+m^y^O
(/)./. Hons. 1953)

If ^(sinh- 1 *)

Hence

x=0

find at

prove that

the value of

dnyfdx n

(B.U.1952)

Exercises
1.

Show

that if
l

- 36^-0

then

2.

n" 1

If >>n 'denotes
(bja),

the

th

differential co-efficient of e ax sin

bx and

prove that
.

y n =(a sec

n aa! sin
^) ^

Also show that

y=*(ABx)

cos

;c+(C4-/>;c) sin AJC, prove that

3.

If

4.

Find the third differential

:5.

Prove 'that

6.

Show

that

if

co-efficient of

=sin wx+cos nx then


y

w r =^[l-f(~lKsin2x] i

when

Wr denotes tht rth differential co-elficient of u with respect to x.

(Lucknow)
7.

Show

that

__
23

......

__
n
(P.U.)

j '

8.

Prove that

where P and

and
respectively.

Q stand

for

nx^-Mi-lXfl

f|

2)jt

-t

4-

EXERCISES
9.

127

Find the value of nth derivative of

jx*^x
2
a
-4)

U
for

x0.

10.

Prove that

dn

if

;t=cot

fi

^=(-l)
where n

is

11.

Trinity College

(0<0 <n), then

0,

n- 1

(-l)

sin nO sin*0

any positive integer.


If

Prove that

IifIn-i+(-l)

and hence show that


(D.U. Hons. 1949)

Rewriting this relation as

Jn_
/j

and replacing

by

If

y=

12.

1,

d n

In-i

~(/i-i)

____

-l)

n,

3, 2,

show

JL

we

get the required result.

that

(x*-l)y n +t+2xy n + l -n(n+ l)^ n

Hence show

13.

If

that

.y

U n denotes

satisfies

0.

the Legendre's equation

the nth derivative of (Lx-t-M)l(x*2Bx+C)., prove

that

Wn+t +
14.

? (-*r*Ji

If y=x*e x , then

z,

15.

P./., D.(/., 7955)

If

prove that
(D.I/. //ow5.,

and Pass, 1949 ;


P.U. 1958 Sept.)

DIFFERENTIAL CALCULUS

128
If

16.

(tan"

;'

*)*,

then

Deduce

that

(Birmingham}
If

17.

x=^ tan (log y),

prove that
(1

+*)+

(2n X

-l)% +(-!)

m +y __m _
=2x, prove

!>=(>.

(B.U. }

__
18.

that

(D.U. Hons it 1947

where > w denotes the wth derivative of y.

If

19.

cos x,

y=*e*

show

P.U. 7959)

that

>W (0) -4n>>2n(0) + (2- l)2n ^ n - (0) =0.


2

y =(l+ x rf

If

20.

gin (m tan" 1^),

^ n(0)=Oand^ 2n+1(0)-(-l)

show

w(m-l)'w~2).

.(m-2n).

y=sin (m cos" Vx) then

If

22.

If x4-y==l,

prove that

(D.U.Hons.l950\M.U.)

21.

23.

that

By forming

in

two

different

8
ways the nth derivative of* ", prove

that
1

+ "!-+

is

723

i*

2 2 73 2

^"f

an
with the
[Equate the nth derivative of x
of x n and x n and put x=l].

24.

Prove that
sin

"Of!?
th derivative of the product

129

EXERCISES
25.

If

jf=s~.__ prove that

9=tan ^

Where

26.

show

Ifw=

that

().

(7,

CHAPTER VI
GENERAL THEOREMS
MEAN VALUE THEOREMS
Introduction. By now, the student must have leanrt to distinguish
between theorems applicable to a class of functions and those concerning some
particular functions like sin x, log x etc. The theorems applicable to a class of

known

functions are

as general theorems.

Some

general theorems which will play a very important part in the


following chapters will be obtained in this chapter. Of these Rollers theorem is
the most fundamental.

M &1.

Merval
'c'

ofx

Rollers

(a, b],

Theorem.

// a

is

function f(x)

and also f(a) ~f(b), then there

exists at

derivable
least

in

an

one value

lying within [a, b] such thatf'(c)=Q.

The function f(x) being derivable

in the interval [a, b] is con4*14, p. 76). By virtue of continuity, it has a greatest


value
and a least value in in the interval, (3-53, p. 55) so that
there are two numbers c and d such that

tinuous,

Now

M~w,

either

... (i)

M^m.

or

..

.(//)

When
case

(/),

the greatest value coincides with the least value as in


the function reduces to a constant so that the derivative
for every value of x and therefore the theorem is
equal to

f'(x) is
true in this case.

When

M and m are unequal, as in case

(//), at least one of them


from the equal values /(a), f(b). Let
/(c) be
different from them. The number 'c' bsing different from a and 6,

must be

different

lies within

the interval

[a, b].

The function /(x) which is derivable


x~c, so that

in the interval [a, b]

is,

in

particular, derivable for

Hra
exists

and

is

As/(c)

the sani3
is

/(c+/t)-/(c)

when h ->

wh0nA

_|.

through positive or negative values/

the greatest value of the function,

we have

f(c+h)<f(c)
whatever positive or negative value h has.

Thus
...(I,

130

GENERAL THEOREMS

131

and
f(c+h)-f(c)
Let h -*

>0for h<0.
From

..-(2)

through positive values.

(1),

we

get
...(3)

Let h -+

through negative values.


f'(c)

The

relations (3)

and

From

(2), .we

get

>0.

(4) will

...(4)

both be true

if,

and only

if

/'(c)=0.

The same conclusion would be


Beast value

m which

differs

Hence the theorem

is

similarly reached

if

it

is

the

from /(a) and/(6).


proved.

Geometrical Statement of the theorem.


If a curve has a tangent at every point thereof, and the ordinates
extremities
A, B are equal, then there exists at least one point
of
curve
than A, B, the tangent at which is parallel to x-axis.
the
other
of

its

In this geometrical form the theorem is quite self-evident, as


the student may himself realise by drawing some curves satisfying
the conditions of the statement.

"X

Fig. 43.

Fig. 44.

This intuitive consideration


proof of the theorem.

is

also

sometimes regarded to be

"a

Note 1. It will be seen that the point


where the derivative has been
to vanish lies strictly within the interval [a, 6], so that it coincides
neither with a nor with b. In view of this fact, the theorem is generally stated

shown

as follows

If a function f(x)
\

is

such that

It is

continuous in the closed interval

It is

derivable in the open interval (a, b),

[a, b],

2.
3.

/()=/<.

then, there exists at least one point

V of the

open interval

(a, b),

such thatf'(c)~Q

132

DIFFERENTIAL CALCULUS

Note 2. The conclusion of Rolle's theorem may not hold good for a
function which does not satisfy any of its conditions.
To illustrate this remark, we consider
the function >>=/(*)= x
the interval
|

[-1,

lt

/(

1)

is

continuous in

are both equal to


Its

1,

1]

and

/(I),

1.

derivative/ (x)

1^

for

1,

li-

< 0, and

is 1

for

0<z<l

and

does not exist for

jc=0, so that/'(x) nowhere vanishes.


(Ex. 2. p. 73>

~%r

The

failure is explained by the fact that


x is not derivable in [ 1, 1], in as much as
the derivative does not exist for jc=0, which is
a point of the interval.
|

Fig. 45.

Ex.

1.

Verify Rolle's theorem for

x2

(i)

in

[-7,

1].

'

(ii)

[-3,

0].

(0 Let
2
/(;c)=;c so that /(!)=! =/(~l).

Also,

f'(x)

x1

is

derivable in

1, 1].

The conditions of the theorem being satisfied, the derivative


must vanish for at least one value of x lying within [1,1].
Also directly we see that the derivative 2x vanishes for
lies within [1,
Hence the .verification.
1].

which value

x=O

Let

(ii)

We have
/(- 3) =0 =
*nd/(x)

is

derivable in the interval

[3,

0]*

We

have

*
x

Putting/ (A:)=0, we get


This equation
satisfied by x==
Of these two values of
2, 3.
Jis
is zero,
2, belongs to the interval [ '3, 0] under

x, for which /'(x)

consideration.

Hence the
Ex.

verification.

Verify Rolle's theorem in the interval [a, b] for the functions


(0 log[(x+*&)/(<i+&)K].
() (x 0)*(x 6)* m, n being positive integer*.
2.

GENERAL THEOREMS
Ex.

Verify Rolle's theorem for the functions

3.

sin x/e

(/)

Ex.
jc

log

133

in [0, *]

(//)

(sin

x-cos

By considering the function (x2)

4.

x-2 x

is satisfied

x) in [w/4, 5w/4].

log x,

show that the equation


1 and 2.

by at least one value of x lying between

6*2. Lagrange's mean value theorem.


If a function f(x) is derivable in the interval [a, b], then there exists at least one value c of x
lying within [a, b] such that
9

f(b)-f(a)
"

b-a

_
-

(C) '

To prove the theorem, we define a new function <p(x) involving


and
f(x)
designed so as to satisfy the conditions of Rolle's theorem.
Let

V(x)=f(x)+Ax,
where

is

a constant to be determined such that

Thus

-~
Now, /(x)
constant.

is

derivable in
<?(x), is

Therefore,

Thus,

<p(x)t satisfies all

b-a

[a, b]. Also x is derivable and, A, is a


derivable in [a, b] and its derivative is

the conditions of Rolle's theorem. There


'c' of x, lying within [a, b] such that

therefore, at least one value

is,

f'(c)=0.

Q = <e>(c)=f'(c)+A,

JWz>

=f (c)

i.e.,

A = -f(c),

...

(0

Another form of the statement of Lagrange's mean value theorem.


is derivable in an interval [a, a +h], there exists at
and 1 such that
least one number '#' lying between
If a function f(x)

f(a

We
Ja, b]
-a

write 6

which

+ h)=f(a)+hf(a+0h).

a = h so that h denotes the length

may now be

written as

a+h],
between a and

The number, c, which lies


a-\-h
by some fraction of A, so that we may write %

where
becomes

or

is

of the interval

[a,

some number between

and

f(a-fh)=f(a)+hf'(a+0h).

1.

is

greater than

Thus the equation

(i)

134

DIFFERENTIAL CALCULUS

Geometrical statement of the theorem. If a curve has a tangent


at each of its points, then there exists at least one point P on the
curve such that the tangent at P is parallel to the chord AB joining its
extremities.

01
Fig. 46.

Fig. 47.

We

will now see the equivalence of the analytical


cal statements.

and geometri-

Iff(x) is derivable in [a, b], then the curve y=f(x) has ar


tangent at each point of the curve lying between the extremities

The

slope
r

of the chord
.

Let

Pbe

AB= f(b)-f(a)
L
b a
V

the point (c,f(c)) on the curve

c,

being such that

...(01

^>-M=/'(c).
The

slope of the tangent at P~f'(c).

From
chord

(i7),

we

see that the

AB are equal.
Thus there

is parallel to

exists a point
the chord AB.

slopes of the tangent at

P on

P and

thex

the curve the tangent at which

Note 1. Now [f(b)-f(a)] is the change in the function /(x) as x changes


from a to b so that [f(b)f(a)]l(ab) is the average rate of change of the

function f(x) over the interval [a, b]. Also/'(c) is the actual rate of change of
the function for x=c- Thus the theorem states that the average rate of change of
a function over an interval is also the actual rate of change of the function at
some point of the interval. In particular, for instance, the average velocity
over any interval of time is equal to the actual velocity at some instant belonging to the interval ; velocity being rate of change of distance w.r. to time. This
interpretation of the theorem justifies the name 'Mean Value' for the theorem.

we

Note 2. If we draw some curves satisfying the conditions of the


will realise that the theorem, as stated in the geometrical form,

theorem*
almost

is

GENERAL THEOREMS
Ex.l:

135

//

f(x)=(x-l)(x-2)(x-3)
find the value

of

a=0, 6=4,

c.

D.U. Hons. 1954)

We have
/(6)=/(4)=3.2.1=6,
f(a)=f(Q)

f(b)-f(a)

-7,~a

12

= -6,

~6

'

Also

Consider

now

the equation

i.e.,

3=3c 2 -12c+ll.

or

3c 2 -12c+8=0

Taking \/3 = 1*732..., we

may

see that both these values of

c'

belong to the interval [0, 4],


Ex.
(i)
(ii7)

Ex.

2.

Verify the

log x in

Find

'c'

value theorem for

[1, e].

lx*+mx+nin
3.

mean

(//)

x3

in [a, 6].

[a, b].

of the

mean

value theorem,

/\x)=x(x- l)(x-2)

if

a=0, &=*.
(D.U. Hons. 1951)

Ex.
cases

4.

Find

'c'

so

that

the following

(c)=[/(/))~/(fl)]/(6~a) in

6=3.

fl=2,
(x-4)
/(*)=*; a-0, 6=1.
;

Ex.
tions log

5.

x and

Applying Lagrange's mean value theorem, in turn to the funcdeteimine the corresponding values of in terms of a and h.

ex ,

Deduce Jhat
(0 0<[log

Ex.

6.

(1

+*)]-*-*-'<!

(//)

< 1 log

~e

"

Explain the failure of the theorem in the interval

- <1.

[1,

1}

when

DIFFERENTIAL CALCULUS

136

important deductions from the mean Value theorem.


consider a function /(x)
the sign of Derivative.
derivable in an interval [a, &]. Let xv x2 be any two points belonging to the interval such that x^>xv Applying the mean value
theorem to the interval [xly xj, we see that there exists a number

Some

6-3.

We

Meaning of

between xl and X2 such that


'

f^-ftoHfv-Xi)*'^

Letf(x)0

6-31.

From

(/)

...

throughout the interval [a, b].

we get

where xl9 x 2 are any two values of x. Thus we see that every two
We
values of the function are equal. Hence f(x) is a constant.
thus prove :
"If the derivative of a function vanishes for
interval, then the function must be a constant.
This

all values

of x

in

an

the converse of the theorem, "Derivative of a constant

is

is

zero/'

Cor. If two functions f(x) and F(x) have the same derivative for
every value ofx in [a, b] then they differ only by a constant.

We

write

Hence, ^(x),

f(x)F(x)

i.e.,

6-32.

Letf(x)>0for
From (i), we get

for,

*!*!

is

a constant.

every value

ofx

in [a, b].

and/' () are both positive.


f(x) is an increasing function of

Hence
proved

x.

We

have thus

"A function
interval is

whose derivative is positive for every value ofx in an


a monotonkally increasing function ofx in that interval."

6-33.

From
JOT,

X2

Xi

Letf'(x)
(/),

is

Hence
proved

we

every value

ofx

in [a, b].

get

positive

f(x)

<Qfor

is

and/'()

negative.

a decreasing function of x.

We

have thus

whose derivative is negative for every value of x in an


a monotonically decreasing function of x in that interval."
Note. The above conclusions remain valid even if/'(x) vanishes at the
*n4 points a, * of the interval, for the V of the mean value theorem never

"A function

interval is

GENERAL THEOREMS

137

6'4.
Cauchy's mean value theorem. If two functions f(x) and
F(x) are derivable in an interval [a, b] and F'(x)-^Qfor any value ofx
in [a, b], then there exists at least one value 'c'
ofx lying within [a, b],

such that

f(b)-f(a)

f'(c)

F(a)^

F'(c)

we note that

Firstly,
[F(b)
F(a)]^0 for if it were 0, then
would
the
of
conditions
the Rolle's theorem and its deriF(x)
satisfy
vative would therefore vanish for at least one value of x and the
would be contradicted.
hypothesis that F (x) is never
:

Now, we

define a new function <p(x) involving f(x) and


to satisfy the conditions of Rolle's theorem.

F(x)

and designed so as
Let
where

a constant to be determined such that

is

f (*

Thus
f(a)

Now,
Therefore,

f(x)

+ AF(a)=f(b)AF[b).

and F(x) are derivable in [a, b]. Also A is


derivable in [a, b] and its derivative is

a constant.

<f(x) is

f(x)+AF'(x).
the conditions of Rolle's theorem. There
therefore, at least one value, c, of :c lying within [a, b] such that

Thus,

-f(x) satisfies

is,

f(c)=0.
or

f'(c)=-AF'(c),
Dividing by F'(c ) which^f-O, we get

_
F'(c)

~F(b)-F(a)

Hence the theorem.


Another form of the statement of Cauchy's mean value theorem.
If two functions f(x) and F(x) are derivable. in an interval [a, a+h] and
and 1,
F'(x)^0, then there exists at least one number 6 between
such that

The equivalence of the two statements can be easily seen as in


the case of Lagrange's mean value theorem.

DIFFERENTIAL CALCULUS

138

Note. Taking F(*)=jc, we


only a particular case of Cauchy's.

may

theorem

easily see that Lagrange's

is

Ex. 1. Verify the theorem for the functions x* and x* in the interval
b being positive.
Ex. 2. If, in the Cauchy's mean value theorem, we write for/Yx), F(x)
x
is the arithe- x show that in each case
(i) **, x
(ii) sin *, cos x
(///) e
metic mean between a and b.

[a, b]

a,

Ex.

3.

If,

in the Cauchy's

V* and 1/Vx respectively then,

and

we

write l/x 2 and

Ijjc

then,

6*5.

value theorem, we write for f(x), and


the geometric mean between a and b>
the harmonic mean between a and

mean

F(x),
if

c, is

c, is

Examples

1.

monotonically increasing

in

every interval.

Let

Thus f'(x) >0 for every value of x except 1 where


Hence f(x) is monotonically increasing in every interval.
2.

is

Separate the intervals

it

vanishes.

which the polynomial

in

increasing or decreasing.

Also draw a graph of the function.

Let

so that
for

x<2

/'(*)<0for2<x<3

x>3
=
x=2
for
/'(*)

/'(*)>0for
()

is

and

3.

oo , 2)
positive in the interval (
negative in the interval (2, 3).

and

(3,

oo

and

monotonically increasing in
2] [3, oo) and monotoni(
cally decreasing in the interval [2, 3].
To draw the graph of the function, we
note the following additional points

Hence /(x)

the intervals

is

oo

(iii)

(iv)

(v)

Fig. 48.

/(0)=2,
/(#)->

<

as

oo

GENERAL THEOREMS
Show

3.

that

*/(!+*)

We

<

<

log

(1+x)

>

and

xfor x

>

0.

write

Thus

/'(x)

Hence f(x)

>

for

for

x=0.

monotonically increasing in the interval

is

= 0,
/(*) > /(O) =0

Also

=0

/(0)

Hence /(x)

for

>

0.

that
positive for every positive value of x, so
0.
x/(l+x) for x
log (1+*)

is

>

>

Again, we write
BO that

F'(jc)==l

Thus,
oo],

>

F(x)

Therefore F(x)
[0,

[0, QO ].

is

for

>

and

monotonically

for

is

increasing

#=0.
in

the

interval

AlsoF(0)=0.

>

F(x)

Hence

F(x)

is

>

F(0)=0forz

0.

positive for positive values of x, so that

>

log (1+Jt) for

>

to

x=n/2.

0.

Exercises
1.
(i)

(ii)
(//i)

Show

that

x/sin x increases steadily from

x=0

x/tan x decreases monotonically from x=0 to jc=n/2.


the equation tan x x=6 has one and only one root in

(P*U.%
(

Jw, Jn).

(.{/. 7952)
O'v)

2.

*>-2 ^

tan-

<

tan-"

<

that x~sin x is an increasing function throughout any


Determine for what values of a axsin x is a steadily

Show

of values of x.

ing function.
3.

interval
increas-

(M.J7.)

Determine the
(x

intervals in which the function


4

+ 6jcH 17x a + 32x+ 32)*-*

is

increasing or decreasing.

is

Separate the intervals in which the function


(*+*+!) /(*-- * + l)
increasing or decreasing.
4.

5.

Determine the intervals in which the function (4-x 8 ) 8


Also draw its graph.

or decreasing.

is

increasing:

140

DIFFEBENTIAL CALCULUS
6.

Find the greatest and least values of the function

x 8 -9x 2 -f24xin[0,

x-

7.

Show

that

$.

Show

that if

x-

(//)

Show

>

6].

decreases as x increases from

(i -f x)

<

log

< -log(l-x) <

that e~ x lies

Prove

Show

If

*.

d+x) < x-

(1

-x)-

*
l\

(B.U. 1953)

that sin

<

<

1,

lies

<

<

1.

between
,

x5

show that

taking * =2/1
+

Hence

for

between

x9

12.

0,

1-xand 1-x + Jx 2
11.

to

that

x
10.

log

,-*-< lo, (!+)< -

(/)

9.

W
1'

>

rdcduce that

5.

13.

Show

that

tan^c

sm

x
14.

Show

ifo<x<

..

jc

that

x-1 > logx>


jc~l
x

and
for
15.

7957)

If /(x) is

where

(jc-l)x-

1
,

> 2xlogx > 4 (x-l)-2 log*


>

(M.T.)

1.

derivable in the interval [a-h, a+h], prove that

<^<

(a+h)-2f(a)+f(a-h
where < t < 1.
fl

The derivative of a function /(x) is positive for every value of x in an


16
show that /(c)
interval [c-A, c], and negative for every value of x in [c, c+h] ;
b the greatest value of the function in the interval [c-h, c+h].
*

of
Higher mean value theorem or Taylor's development
derivatives
the
function in a finite form.
// a function f(x) possesses
*
to a certain or^er n f r every value
f'(x)>f"( x )'f'"( x )> ..... /"(*)> UP
x in the interval, [a, a+h], then there exists at least one number 0,
between and 7, such that
6-6.

TAYLOB'S THEOREMS
f(a +

141

h)=f(a)+hf'(a)+*f f" (a)+

We define a new function 9(x)


x
f'( )>f"(x), ...... fn (x) designed so
,

Rollers theorem.

involving /(x) and its drivatives


as to satisfy the conditions for

Let
......

,i

where A

is

a constant to be determined such that

Thus we get

from the equation


-

Now,

it is

in the interval

given that/(x), f'(x),f"(x) .......

[a,

fn ~\x) are derivable

a+h].

2
Also, a+h-x, (a+A-x) /2 !, ........ , (a+h-x)"ln I are derivable in [a, a+h]. Therefore <t(x) is derivable in [a, a-f^]- Also

f'(x)=f(x)-f'(x)+(a+h-x)f*(x)-(a+h-x)f'(x)

ether terms cancelling in pairs.

Thus

<p(x)

satisfies all

exists, therefore, at least

the conditions for Rolle's theorem. There


and 1 such that

one number 6 between

DIFFERENTIAL CALCULUS

142

f"(a+6h)=A

for

Substituting this value of

(1-

in

(i),

we

get

f(a+h)-f(a)+hf\a)+~f''(a)+ ...... +

The (+l)th term

Lagrange's form of remainder after n terms in the Taylor's


expansion of/(0-f7z) in ascending integral powers of //.
Note. Taking /?=! we see that Lagrange's mean value theorem is only
a particular case of the Taylor's development obtained here.
Cor. Maclaurin's development. Instead of considering the interval [a, a+h], we now consider the interval [0, x] so that we change
a to and h to x in (//). We get
is called

'-

'

'

-.

..(//I)

which holds when the function f(x) possesses the derivatives


f'(x),f"(x), ......... ,/(*)

Mervai

in the

[0, x].

f
,

The formula
[0, x] in

the

finite*

(HI)

is

known

as Maclaurin's development of /(x) in

form with Lagrange's form of remainder.-

67. Taylor's development of a function with Caiichy's form of


If a function f(x) possesses the derivatives /'(*)> f"( x ) ......
fn (x) up to ft certain order n in the interval [a, a +/*], then there exists
between
and 1 such that
at least one number
remainder.

;7 2

We define

/)-i

+_

a ftew function f(x) as follows

f(x)=/(x)-f (a+h-x)f(x)

where

is

a constant to be determined such that

1)

/-i

(a)

TAYLOR'S THEOREMS
Thus, we get

143

from the equation


h n ~i
f

It is easy to see that v(x)

Hence

<p(x) satisfies

there exists at least one

derivable in

is

[a,

a+h].

the conditions for Rolle's theorem so that


between and 1 such that

number

Now,

y'(x)~

other terms cancelling in pairs.

o=?'(a+0/o=
hn

Substituting this value of


.

h*

ri-i)'i

f n ( a + 6h )- A

'1

in

(/),

we

get

The (+l)th term

is called Cauchy's form of remainder after n terms in the expansion


of f(a+h) in ascending integral powers of h.

Maclaurin's development.

Cor.
(//),

we

Changing a to

and h

to

in

get

is known as Maclauriris development off(x) in the interval


with
Cauchy's form of remainder after n terms. It holds when
x]
n

which
[0.

jc

f(x) possesses
Ex.

1.

^derivatives /'(*), /"(*), ...... f


,

Show

that

(x) in [0, x].

DIFFERENTIAL CALCULUS

144

Substituting these values in (Hi)

Ex.

2.

Show
JC

3.

JC

we

obtain the result,

x
~

X*** 1

"2~!

Ex.

6*7, p. 142),

that, for every value of

JC

Show

X^
4!

that

Ex. 4. Find, by Maclaurin's theorem, the first four terms and the remainder after n terms of the expression of e ax cos bx in terms of the ascending
powers of x.

APPENDIX
Examples

Show that the number 6 which occurs in the Taylor's theorem


1.
with Lagrange's form of remainder after n terms approaches the limit
n+1
(x) is continuous and
ll(n-\-J)ash approaches zero provided that f
different from zero at x=a.
(D.U. Hons. 1950)
Applying Taylor's Theorem with remainders after n terms and
terms successively, we obtain

(rt-f-1)

f(a+h)=f(a)+hf'(a) + ...... +

n-l

fn
\

These give
hn

hn

/(+*)=-

f n ()+

,-

fn (a+eh)-f (a)= n

or

(n+l)

,/"(+'*),

+l f<*\a+e'h).

Applying Lagrange's Mean Value Theorem to the left-hand


we have

Let h ->

+i

'/" +

0.

lim
2.

Show

0=^.

that

x*>(l+x)

We
.-.

put

side,

f(x)

= x* -

/'(x)=2x-l.

[log

+ x)]*for x>0.
[log 1 + x)].

[log (l

+x)
(l+x)]-(l+x).
(

=2x-[log (l+x)] -2 log (l+x).


The form of /'(*) is such that we cannot immediately
2

as to its sign.
vative.

We,

therefore, proceed to determine the second

145

decide
deri-

DIFFERENTIAL CALCULUS

146

/"(x)=2-2

which

is

>0

for

'

/'(*)
Also

log (l+x).l/(l

+ x)-2/(l+x)

x>Q.

i"3

(See Ex. 3, p. 139)

monotonically increasing in the interval [0, oo

).

/'(0)=0.
Therefore

f'(x)>Qforx>0.
,

Hence f(x)

is

monotonically increasing in the interval

[0, oo).

Also

/(0)=0
Therefore

f(x)>0

for

x>0.

Hence

x>(l+x)[log (1+*)]
3.

for

x>0.

If <p"(x)>Q for every value of x, then


f [*(*!+*)]< tt*(*i)

for 'every pair of values of x1


Suppose that xa >xt

+ *(*.)],

and x z

We write
Applying Lagrange's mean value theorem to the function
X 2 ] we see that there
for the intervals [x l9 (x l
2 )/2] and [(x 1
2 )/2,
exist numbers 19 % belonging to the two intervals respectively, such

+x

+x

that

and

= [K*i+^) -*i]^ (i)=l(*i-^'(&)


f

[f K*i+*i)-^(*i)]

Thus from

(1), (2), (3),

-(3)

we obtain

Applying the mean value theorem to the function 9' (x) for the
interval [^, | 2 ], we see that there exists a numer y such that

From

(4)

9'(&)-9'(4i)=(!t-Si)9"(*)and (6), we obtain

..(5)

APPENDIX
Since Jta xl9 | 2
required result.

147

an(* ? "(?) are

ft ll

we obtain the

positive,

Exercises

Show

1.

that

*0M which

approaches the limit i as

'A'

mean

occurs in the Lagrange's

approaches

0,

value theorem)

'

provided that/ (a)


(P.U. 1949

is

[It

2.

-where

should also be assumed that f"(x)

Show that
/(a+/0=/(aH

lies

between

and

V W+4* /
/

//

is

not zero.
Hons. 1949)

).(/.

not continuous.]

(fl+ 0/0,

and prove that


lim 0=J,

H-+Q

specifying the necessary conditions.


3.

where

(C.U. Hons. 1953)

Assuming f"(x) continuous

in [a, b] 9

show

both lie in [a, 6].


and
(Take <v(x)^f(x)+Ax+ Bx* and determine

that

Now apply Rollers Theorem


4.

^^*^^

/4,

to the intervals

such that

[a, c]

and

[c, 6].]

The second

derivative /"(*) of a function f(x) is continuous for


anc^ at eac^ P^ nt x tlle s *8ns of /(x) and /"(jc) are the same. Prove that

if f(x) vanishes at points c

between c and

and

where a<^c<d<^b, then

d,

it

vanishes everywhere
(B.U. 1952)

d.

/(*) 9(*) an(* ^U) ar e three functions derivable in an interval


such that
that there exists a point
5.

*how

f(b)

Deduce Lagrange's and Cauchy's mean value theorems.


6.

Prove that
lim

if /

7/

(jc)

exists.

7.

Discuss the applicability of Rolle's


(/)

/U)tan x and a=0,

Theorem when

b^n.

(m)
(iv)

/W=U-c) 3 -c 3

when a=0 and 6=2c

(a,

b)

CHAPTER

VII

MAXIMA AND MINIMA


GREATEST AND LEAST VALUES
In this chapter we shall be concerned with the application
7'1.
of Calculus for determining the values of a function which are greatest
or least in their immediate neighbourhood technically known as
Maximum and Minimum values. A knowledge of these values of a
function is of great help in drawing its graph and in determining its
greatest and least values in any given finite interval.

assumed that/(x) possesses continuous derivatives of


come in question.
that
order
every
It will be

Maximum value of a function. Let, c, be any interior point


of the interval of definition of a function /(x). Then we say that/(c)
is a maximum value of /(*), if it is the greatest of all its values for
values of x lying in some neighbourhood of c. To be more definite
and to avoid the vague words 'Some neighbourhood', we say that
f(c) is a maximum value of the function, if there exists some interval
(c8, c+S) around c such that
for

all values

ofx, other than

c,

lying in this interval.

This, again,

f(C)

c+8

is

f(c)

for values ofh lying between


small in numerical value,

equivalent to saying that


value of /(;c), if

is

maximum

> f(c+h),

and

i.e.,

S, i.e., for

f(c+h)-f(c)

<

values of h sufficiently

Minimum

value of a function, /(c) is said to be a minimum value


the least of all its values for values of x lying in some
neighbourhood of c.

of /(#),

if it is

This

is

equivalent to saying that f(c)

is

a minimum value off(x)

if there exists a positive 8 such that

f(c)

<f(c+h), i.e.J(c+h)-f(c)

for values ofh lying between


small in numerical value.

and

8, i.e.,

>

for values of h sufficiently

Note 1. The term extreme value is used both for a maximum as well as
for a minimum value, so that /(c) is an extreme value if f(c+h) f(c) keeps an
invariable sign for values ofh sufficiently small numerically.

148

MAXIMA AND MINIMA


Note

149

While ascertaining whether any value /(c)

is an extreme value or
the values of the function for values of x in any
immediate neighbourhood of c, so that the
values of the function outside the neighbourhood do not come into question at all.
Thus, a maximum value may not be the
greatest and a minimum value may not be the
least of all the values of the function in any
finite interval.
In fact a function can have
several maximum and minimum values and a
minimum value can even be greater than a

not,

2.

we compare /(c) with

maximum

value.

A glance

at the adjoining

graph of f(x)

shows that the ordinates of points P 19 P3 P5


are the maximum and the ordinates of the points
P2 P4 are the minimum values of f(x) and that
Fig. 49.
the ordinates of P4 which is minimum is greater than the ordinate of P x whi:h
a maximum.
,

is

To prove that a
necessary condition for extreme values.
is that
value
extreme
an
condition
be
to
off(x)
for f(c)
necessary
7-2.

Let f(c) be a
c-\-h

maximum

value of f(x)<

There exists an interval (c S, c+8), around c, such that,


any number, belonging to this interval, we have
Here, h

may

/(c+/0</(c),
be positive or negative.

Thus

'^<0iffc>0,

If h tends to

...

through positive values, we obtain from

If A tends to

through negative values,

.-(Hi)

we obtain from (),


...

/'(c)>0.

The

relations (in)

and

(/v)

(0

(/),

/'(c)<0.

only

if,

is

will simultaneously be true,

if

(iv)

and

if

/'(c)=o.
It

can similarly be shown that /'(c)=0,

if /(c)

is

minimum

value of f(x).
Cor. Greatest and least values of a function in any interval. The
either f(a)
greatest and least values off(x) in any interval [a, b] are
x
andf(b), or are given by the values of for w/i/c/i/'(x)=0.

The greatest and least value of a function are also its extreme
values in case they are attained at a point strictly within the interval
so that the derivative must be zero at the corresponding point.
The theorem now

easily follows.

DIFFERENTIAL CALCULUS

150

Note
for

1.

where the ordinate


point
x-axis."

When

the necessary condition


to a curve at a

Geometrically interpreted,

extreme values obtained above states


a

is

stated in this

maximum

"Tangent

or

minimum

is

parallel

to

geometrical form, the theorem appears

[Refer Fig, 48 p. 149.]


The vanishing off (c) is only a necessary but not sufficient condition for f(x) to be an extreme value. To see this, we consider

almost self-evident.

Note

2.

the function /(Jc)==x 3 for

x=Q.

For value of x greater than 0,/(x) is positive and is, therefore,


and for values of x less than 0, f(x) ia
greater than/(0) which is
less
negative and is, therefore,
than/(0).
Thus/(0) is not an extreme value even though /'(0)=0.
Note 3. Stationary value. A function f(x) is said to be stationary
;

also
for x^cif the derivative f (x) vanishes for x=c, i e. if /'(c) =
then f(c) is said to be a stationary or a turning value of f(x). The
term stationary arises from the fact that the rate of change /'(#)
of the function f(x) with respect to x is zero for a value of x for which
;

f(x)

is

stationary.

noted that a maximum or a minimum value is also a


stationary value but a stationary value may neither be a maximum
nor a minimum value.
It will be

Ex.

Find the greatest and

1.

least values

of

in the interval [0, 2J.

Let

Thus

/'(*)=0forx=l, -1,
The value

x=

J.

does not belong to the interval


to
be
considered.
not, therefore,
1

[0, 2]

and

Now
Also

Thus the least value


Ex.

2.

is 1

and the greatest value

Find the greatest and

least values

is

21.

of

in the interval [-2, 5].

Change of Sign. A function is said to change sign from


to
negative as x passes through a number c, if there exists
positive
some left-handed neighbourhood (c ft, c) of c for every point of
which the function is positive, and also there exists some rightfhanded neighbourhood (c, c+h) of c for every point of which the

Def

unction

is

negative.

151

MAXIMA AND MINIMA

A similar
the statement

meaning with obvious alterations can be assigned to


function changes sign from negative to positive, as

"A

passes through c".


It is clear that if a continuous function /(x) changes sign as
passes through c then we must have/(c)=0.
Ex. 1. Show that the function

= (x + 2)(x-m2x- l)Cx~3)
<?(x)
changes sign from positive to negative as x passes through 4 and from negative
to positive as x passes through
2 or 3 ; also show that it does not change sign
as x passes through 1.
Ex. 2. Show that the function
changes sign from positive to negative as x passes through
.* and 2.
negative to positive as x passes through

4 and

and from

72. Sufficient criteria for extreme values. To prove that f(c) is


an extreme value off(x) if and only iff'(x) changes sign as x passes
through c, and to show that f(c) is a maximum value if the sign
changes from positive to negative and a minimum value in the contrary
case.

Case

Let f(x) change sign frcm positive to negative as x

I.

passes through

c.

In some left-handed neighbourhood of c,f'(x) is positive and so


/(x)is monotonically increasing in his neighbourhood. (6*32, p. 136),
Therefore /(c)

is

the greatest of

all

the values of /(x) in this

handed neighbourhood.
In some right-handed neighbourhood of
so

f(x)

p. 136).

in

is

monotonically increasing
Therefore /(c) is the greatest of

left-

negative and
neighbourhood ( 6*33,
the values off(x) in this
c,

f'(x)

is

this
all

right-handed neighbourhood.
the greatest of all the values of /(x) in a certain
value
complete neighbourhood of c and so, by def., f(c) is a maximum

Hence

f(c)

is

of/(x).

Case

Let f'(x) change sign from negative to positive as

II.

passes through

c.

the least of all


the values of /(x), in a certain complete neighbourhood of c and so,
by def., /(c) is a minimum value of /(x).
It can similarly be

shown that

in this case/(c)

is

Case III. If /'(x) does not change sign, i.e., has the same sign
in a certain complete neighbourhood of c> then /(x) is either monothis
tonically increasing or monotonically decreasing throughout
value
extreme
so
of/(x).
neighbourhood
that/(c) is not an
Note. Geometrically interpreted, the theorem states that the tangent to
a curve at every point in a certain left handed neighbourhood of the point P
whose ordinate is a maximum (minimum) makes an acute angle (obtuse angle)
and the tangent at any point in a certain right-handed neighbourhood of P makreJJ
an obtuse angle (acute angle) with x-axis. In case, the tangent on either suit of r
makes an acute angle (or obtuse angle), the ordinate of P is neither a maximum
nor a minimum.

DIFFERENTIAL CALCULUS

152
Ex.

I.

Examine

the polynomial

10x6
for

24x 5 +15x 4 ~4Qx 8 +108

maximum and minimum

values.

Let

2)

=60x*(x +l)(x-2)
Thus, /'(;t)=0 for x-=0 and x=2
values of f(x) for #=0 and 2 only.

so that

we expect extreme

Now,
forx<0,

/'(*)<<>;

forO<x<2

/'(x)<0;

forx>2

/'(*)> 0.

Here,/'(*) does not change sign as x passes through


/(O) isjieither a maximum nor a minimum value,

so that

since f'(x) changes sign from negative to positive as


100 is a minimum value.
passes through 2, therefore /(2)
Also,

Ex.

2.

Find the extreme values of

and distinguish between them.


Let

24

so that

/(x)=0

for

x=l,

2, 3.

Therefore, /(x) can have extreme values


for *==-!, 2, 3 only.

Now, for* <1,

1<*<2,
for2<x<3,

for

/ (x)<0;

/'W>0

forx>3,
pig, 50

therefore /(l)--8

Since /'(x) changes sign from negative


to positive as x passes through 1 and 3,
are the two minimum values.

and/(3)=-8

from positive to negative as


Again since, f'(x) changes sign
is a maximum value.
-7
therefore, .fl2)
passes through 2,

MAXIMA AND MINIMA


Ex.

153

Find the extreme values of

3.

5*+18;c 5 +15;c*-10.
Ex.

are

Show

4.

that the

maximum and minimum

values of

-9 and - 1/9 respectively.


Ex.

Show

5.

that

9x5 +30;c4 +35;c 8 -fl5;c 2 +l


is

maximum when x=
Also, find

[-2,

2/3

its

and minimum when x=0.

greatest

Ex.

6.

Show

that

value

-5x+5;c3 -l
when x = l, a minimum
jc

has a

and least values in the intervals [2/3,

0],

and

2].

maximum

value

when x=3 and

neither

*=0.

when

(D.U. 1948)

Use of derivatives of second and higher orders. The derivatives


order only have so far been employed for determining and
As shown
distinguishing between the extreme values of a function.
in the present article, the same thing can sometimes be done more
conveniently by employing derivatives of the second and higher
7-4.

of the

first

orders.
All along this discussion it will be assumed that f(x) possesses
continuous derivatives of every order, that come in question, in the
neighbourhood of the point c.
7*41.

Theorem

1.

f(c)

is

a minimum value off(x) if


y

f'(c)=0andf"(c)>0.
Applying Taylor's theorem with remainder after two terms, we
get

f(c+h)^f(c)+hf'(c)+

r(c+0ji)

is positive for X
c, there exists an
for every point of which the second derivative is
3-51, p. 54)
(

As the value f"(c) of f(x)


interval around,

c,

positive.

Let
point of
ft

/2

is

c+h

be any point of this interval.

th''s

interval

Then,

and accordingly /"(C +M)

is

c+6 2 h,

a
Also

is also

positive.

positive.

see that there exists an interval around c for every


of
which, f(c+h)f(c) is positive, i.e.,f(c)<f(c+h).
point/ +A,

.Thus

we

Hence /(c)

is

minimum

value of/(x).

154

DIEFERENTIAL CALCULUS

Theorem

7-42.

As

in theorem

f(c

i'.*.,

2.

1,

f(c) is a

maximum

we have by

value off(x) 9 if

Taylor's theorem,

+h)-f(c) =

Asf"(c) is negative, there exists an interval around c for every


Thus as in the
point of which the second derivative is negative.
c
for every point,
preceding case, there exists an interval around
c+h of which /(c+/0/(c), is negative, i.e.,f(c)>f(c+h)
;

Hence /(c)

is

maximum

General Criteria.

7-43.

value of/(x).

Let

Thenf(c)is
a minimum value off(x), iffn (c)>Q and n

(i)

is

even

() a maximum value of(x), iff (c)<0 and n is even


neither a maximum nor a minimum value ifn is odd.
(iii)

Applying Taylor's theorem with remainder after n terms, we

so that because of the given conditions,


n h).

As/n (c) ?^0,

there exists an interval around c for every point


the nth derivative fn (x) has the same sign, viz., that of

x of which
Thus

for every point, c-f-A, of this interval,

fn (c + 9 n h)

has the

sign of f

(c).

is positive, whether, h, be positive


Also when n is even, h n \n
or negative and when nis odd, h n \n !, changes sign with the change
in the sign of h.
!

Hence, as in the preceding cases we have the criteria as stated.


Note. The result proved in 7*41 and 7*42 is only a particular case of
the general criteria established

in^

7*43 above.

MAXIMA AND MINIMA

155

Examples

maximum and minimum

Find the

1.

Let

values of the polynomial

4
2
5
/(x)-=8x --15x -l-10x

3
4
/'(x) = 40x -60x +20x

=20x(2x 3 ~-3x* + l)
2

Hence

-20x(x-l) (2x+l).
/'(*)=0 for x=0, 1,- J.

Again

/"(x)=-160x

- 180x 2 +20
9x 2 +l).

Now, /"(

maximum

45 which

)=

negative so that /(

is

I)

T-S-

is

value.

Again, /"(0)=: 20 which

is

positive so that/(0)

is

minimum

value.

As/"(l)-0, we have

Now

to

examine /

(1).

2
/'"(*)= 480x -360x.

= l20

f'"(l)

Hence /(I)

is

neither a

which

maximum

Investigate for 'maximum

2.

//;

sin JC+cos

is

nor a

not zero.

minimum

and minimum values

value.
the function

2jc.

Let

x+cos 2x.
x 2 sin 2x
=cos
x 4 sin x cos
%
we
dyldx=Q,
get
>>=sin

...(i)

dy fdx = cos
Putting

cos

We
function

x=0

or sin

consider values of
is

x= J.

x between

and

cos

x~ Ogives

x=
x = siii"

and
J lying

sin

x=|

gives

and

between

Now

d*y/dx

^
J

sin

For x=7r/2, d y/dx*=:3 which


For x=37r/2, d 2y/dx 2 = 5 which
For x=sin~ 1 J and TT sin* 1 J,

df^/rf^^-sin
negative.

and

TT

sin* 1 J,

Tr/2.
2

is

only, for the given

and

which

2ir

periodic with period 2?r.

Now
sin*" 1

x.

x-4 (1-2

4 cos 2x.
is

positive

is

positive

;.

sin 2 x)

= -15/4

DIFFERENTIAL CALCULUS

156

Therefore y is a maximum for x


minimum for x=7r/2, 37T/2.

= sin~"

J4

sin" 1 J

TT

and

is

Putting these values of x in (i), we see that |, | are the two


2 are the two minimum values.
values and 0,

maximum

Exercises

maximum and minimum

Investigate the

1.

(/)

(it)

values of

2x 3 -15x 2 + 36x+10.

(P.U.I945)

3x*-4x 3 +5.

Find the values of x for which x e -6ax 3 -f 90 2 x4 -f


(a>0).

'

2.

values.

Determine the values of x

3.

minimum

has

which the function

for

12x 5 -45x<+40x 3 -f6


attains (1) a

maximum value, (2) a minimum value.


maximum value of (x-l)(x-2)(x-3).

(P-U- 1941)

4.

Find the

5.

Find the maxima and minima as well as the greatest and the

values of the function

Find

6.

minimum

for

y=x*

12x 2 +45x in the interval

(P>V. 1939)
least

[0, 7].

what values of x the following expression

is

maximum

or

2x 3 -21x 2 4- 36x-20.
7.

Find the extreme values of x 3 /(x 4 -f

8.

Show

that (x + !)*/(*+ 3) s has a

9.

Show

that x x

10.

Show

1).

value 2/27 and a

maximum

minimum

value 0.
is

that the

minimum

maximum

maximum

x=e~ l

for

value of

(1

l
/*)* is e

l*.

<x<oo

11.

Find the

12.

Find the extreme value of a 3*

13.

Find the maximum and minimum values of x+sin 2x

14.

Find the values of x for which

value of (log x)lx in


1

~as -x.
sin

(a>

(P.U. 1955)

1).

x cos x

is

in

<S x

maximum

2*.

or a

minimum.
15.

Find the

maximum and minimum


x

in(

*^jc5^0;
Show
16.
17.

that sin

Discuss the
(/)

values of

sin 2x-f J sin 3x,

x (1-fcos x)

is

maximum when x=$n.

maxima and minima

sinx-fisin 2x+ j

in the

interval

[0, n]

of the sums

sin 3x.

cos K^\ cos 2x + $ cos 3x.


Also, obtain their greatest and the least values in the given interval.
(//)

18.
(i)

(Hi)
19.

for

x=0.

Find the minimum and

sinxcos^x.

(i"0

sinxcos 2x.

Show

that (3

maximum

(/v)

x)e**

4x^-x

values of

flsecx-f&cosecx,
ex cos (x

has no

(0<a<6).

a).

maximum

or

minimum

value

EXAMPLES
Find the maxima and minima of the radii vectors of the curve

20.

ra

21.

Find the

P
~ _a?__ +
"cos
sin*0
l

(Delhi, Aligarh)

"fl

maximum and minimum

values of x*+y* where

ax*+2hxy+by*=l.
[Taking x=rcos
treme value of r 2 where

0,

y = rs'm

0, the

question reduces to finding the ex-

2
-T=fl cos 0-f 2/i cos $ sin

sin 2 0]

In the following, we shall apply the theory of

maxima

and minima to solve problems involving the use of the same.

It /will

7*5.

be seen that, in general, we shall not need to find the second derivative and complete decision would be made at the stage of the first
derivative only when we have obtained the stationary values. In this
connection, it will be found useful to determine the limits between
which the independent variable lies. Suppose that these limits are
a, b.

If

is

for

x=a

xb and positive otherwise and has only

and

owe stationary value, then


maximum and the greatest.

the stationary value

is

necessarily the

If y -> QO as x -> a and as x -> b and has only one stationary


value, then the stationary value is necessarily the minimum and the
least.

In connection with the problems concerning spheres, cones


cylinders the following results would be often needed :
1

Sphere of Radius

r.

Volume = ?Ttr*.
Cylinder of height,

2.

Surface

Cone of height,

h,

47ir.

and radius of circular base, r.


Curved surface =2Trrh.

h,

Volume =7rr 2 /?.


The area of each plane
3.

and

face

Trr

2
.

and radius of circular base,

r.

tan" 1 (/*//*),
Semi-vertical angle
Slant height- V( r2 +/* 2 ),

Volume = |TT r 2 h,

Curved 8urface=-7Tr\/(r 3 +/j 2).


Here cones and cylinders are always supposed to be right
circular.

Application to Problems.

7-6.

Examples

Show

1.

greatest volume

Let
surface

that the height of an open cylinder


equal to the radius of its base.

of given surface and

is

be the radius of the circular base

and K, the volume of the open

h,

cylinder.

the height
Therefore,

the

158

DIFFERENTIAL CALCULUS

Here, as given, S is a constant and


are variables. Substituting the value of
(')> we get

V in

which gives

terms of one variable

is a variable.
Also, h, r
as obtained from (i), in

h,

r.

dV __S

'

dr~~~

so waat ur /u/=v only when r =\/(Sl37i)


negative value of
inadmissible. Thus V has only one stationary value.

As

V must be

positive,

Srnr >

we have

Thus

r varies in

Now F=0
for every other

r=V

being

0, i.e.,

>

Sr

the interval

for the points

Trr

(0,

or r

>

y"

(S/Tr).

(5/7:)).

r=0 and

admissible value of x.

^/ (S/n)

Hence K

and
is

is

positive
greatest for

(S/37T).

Substituting this value of r in

(i),

we

get

c
h

3?r
2?rr

_2S
~
3

Hence

27T

hr for a cylinder of greatest volume

and given

surface.

Show that the radius of the right circular cylinder of greatest


2.
curved surface which can be inscribed in a given cone is half that of the
cone.

Let

be the radius

OA

of the base and

h,

OV of the

the height

given cone.

We inscribe in it a cylinder, the radius


of whose base is OP=x as shown in the
figure. We note that x may take up any value
y

between

and

r.

To determine the height PL,


cylinder, we have

PL

Fis. 51.

PA
= ~'

PL

of this

r-o:
'

159

EXAMPLES
If

S be

the curved surface of the cylinder,

we have

= -?5^?L ^^

S=27T. OP. PL

^J^
dx

(r-2x)

=0

(rx-x

for

Thus S has only one stationary value.


Now S is for x=0 and x=r and is
and r.
between
lying
S
is greatest for x=r/2.
Therefore

2
),

x=r/2.

positive

for values of

circular cylinder of greatest


surface of the right
a
in
sphere of radius r.
surface which can be inscribed
is the
construct a cylinder as shown in the figure.
radius of the base and CB is the height of this cylinder.
3.

Fi/irf

//*

OA

We

Let Z_AOB==0 so that


between and 7T/2.
t

lies

OA

=008

0.

= OB cos
AB
=sm
OB
.-.

cos

6.

0=rsin

6.

0,

AB=OBsm

IfS be the

0=r

surface,

we have

5=27r. OA* +2v.OA.BC.


=27rr 2 (cos2 0+sin

dS
=27r/-

= 27rr

2
(

20). ..(1)

2 cos

(2 cos

sin

20

Fig. 52.

0+2 cos

20)

sin 20).

=0

gives
2 cos 20

sin

20=0,

i.e.,

tan 20=2.

...(2)

Let, 0j, be a root of tan 20 = 2.


sin 20 1 = 2/\/5 and cos 20 1 =l/'V/5.
.-.
As tan 20X =2,

From (1) ,we see that


when 0=0, S=27rr 2 when
;

when 0=0,, 5=27rr2

which

is

2
greater than Sirr

Hence

0=7T/2,

(1+^^+

5=0,
sin

20,)

wr
is

the required greatest

sjurface. (Cor.

p,149),

160

DIFFERENTIAL CALCULUS

Prove that the least perimeter of an isosceles


triangle in which
of radius r can be inscribed is Qr^/3.
(P. u. 1934)
We take one vertex A of the triangle at a distance x from the
centre O. Let AO meet the circle at P.
The two tangents from A
.
and the tangent at P determine an isosceles
4.

circle

ABC

triangle

We

Also

circumscribing the given

circle.

have OL~r.

BP=AP tan

denote the perimeter of the triangle,

If* P>

Fig. 53.

/_BAP

we have

p=AB+A C+BG
=2AB+2BP
=2(AL+LB)2BP
=2AL+4BP,

(for,

BL=BP)

-rr

"dx
=2 2{x+r)(x*-r

for

so that dpjdx=0
admissible.

)-x(x+r}

2r

negative value,

Now, x may take up any value >r


the interval

From

(r,

(i),

oo

only,

r,

of x being in-

so that

p-

Hence p

we

see that

also as X->QO
is

varies in

).

->

oo

as x ->

r.

Again, dividing the numerator and denominator by

so that

it

least for

x2 we get
,

x=2r.

see that the least value of p is

Putting this value of x in

(/),

we

EXAMPLES

161

circumscribed to a sphere of radius r ; show that


is a minimum, its altitude is 4r and its
semi-vertical angle sin~i f
(Madras 1953 ; P.U. 1930)
5.

cone

is

when the volume of the cone

We take the vertex A of the cone at a


O of he sphere. .(See Fig. 53. p. 160).

distance

x from the

centre

By drawing tangent lines from A, as shown in the figure, >ve


construct the cone circumscribing the sphere.
Let the semi- vertical angle BAP of the cone be 0.
Now,

if v

be the volume of the cone, we have

V=i7T .BP*. AP,


which

now expressed

be

will

Since sin

=7=r-,= x
OA

A
Again, since

Thus

.'.

BP

tan

have

0=

,
-=tan
AL

(r+x)

mi

We

in terms of x.

0,

r2

v=ir\,- 92----

9
2

^
(x+r)=
,

--

Trr

'

r)

Thus dvjdx is for x=3r.


Here x can take up any value ^r and v -* cc when x
Thus t; is minimum and least for x=3r.
when x -> QO

>> r

and

Hence, for minimum volume, the altitude of the cone

and the semi -vertical angle

= sinNormal

6.

is

drawn

at

~sin- 1 r

= sin~

3r

a variable point

P of an

i.
^

ellipse

maximum

distance of the normal from the centre of the ellipse.


(P.U. 1935)
take any point P(a cos 6, b sin 0) on the ellipse
being
the eccentric angle of the point.
find the

We

The equation of the tangent at P


at
Therefore, slope
r of the tangent
&

of the normal at
Hence, slope
*
1

-p

is

P~ ----a sin
.

P=tb ---cos

-r

=1.

DIFFERENTIAL CALCULUS
Therefore equation of the normal at
,

"

0x

>P> be

,.

sin

y-b sm

0=^~

0by cos

0=(a

/
<5r

a sin

is

(x-a

2
)

cos 0.

sin

tance from the centre


perpendicular distance

its

(0, 0),

we

obtain

jm

(9

cosj?

*~6+b*^o&
-

,--

/
(

CL

^ 2 cos4

ka\\
u

---------

(a

Putting

------

sin

6-}

<
-

ft

cos

dpld6=Q, we get

Because of the symmetry of the ellipse about the two co-ordinate axes, it is enough to consider only those values of 6 which lie
between and 7r/2 so that we reject the negative value of tan 0.

Now, />=0 when 0=0


between

and

or

7T/2

and p

is

positive

when

lies

Therefore p is maximum when tan 0=v'(^0Substituting this value in (/), we see that the maximum value of

is

7T/2.

b.

7.
Assuming that the petrol burnt (per hour) in driving a motor
boat varies as the cube of its velocity, show that the most economical
speed when going against a current ofc miles per hour is | c miles per

hour.

Let v miles per hour be the velocity of the boat so that (v -c)
miles per hour is its velocity relative to water when
going against
the current.
Therefore the time required to cover a distance of d miles

u
hours.

3=

The

8
petrol burnt per hour=A:v where k
the total amount, y> of petrol burnt is given by
,

y=k
'

v*d

vc

Thus

Of these r;=0

is in-

t8

7
=fo/

VC
.

dv

Putting dy/dv=0

a constant.

is

(v

we

get

c)

0=0 and |#

admissible.

Also y _>

oc

when

t?

Thus v=^c gives the

-> c and when


least value of y.

_> oo

183

EXAMPLES
Exercises

1.
Divide a number 15 into two parts such that the square of one
multiplied with the cube of the other is a maximum.

Show

2.

that of all rectangles of given area, the square has the smallest

perimeter.
3.
Find the rectangle of greatest perimeter which can be inscribed in a
of radius a.
4.
If 40 square feet of sheet metal are to be used in the construction of
an open tank with a square base, find the dimensions so that the capacity is

circle

(P-U.)
a semi-circle with a rectangle on its diameter.
Given that perimeter of the figure, is 20 feet, find its dimensions in order that
its area may be a maximum.
(Patna, Allahabad)

greatest possible.

A figure consists of

5.

6.
A, B are fijced points with co-ordinates (0, a) and (0, b) and P is a
variable point (x t 0) referred to rectangular axes ; prove that x*o6 when the
( p -# 1935 )
angle APB is a maximum.
7.
given quantity of metal is to be cast into a half-cylinder, /.*., with
a rectangular base and semi-circular ends. Show that in order the total surface
area may be minimum the ratio of the length of the cylinder to the diameter
of its circular ends is */(*+2).
(Aligarh 1949)

8.
The sum of the surfaces of a cube and a sphere is given show that
when the sum of their volumes is least, the diameter of the sphere is equal to
;

the edge of the cube.


9.
The strength of a beam varies as the product of its breadth and the
square of its depth. Find the dimensions of the strongest beam that can be
cut from a circular log of wood of radius a units.
(D.U. 1953)

10.
The amount of fuel consumed per hour by a certain steamer varies as
the cube of its speed. When the speed is 15 miles per hour, the fuel consumed
The other expenses total Rs. 100
is 4 J tons of coal per hour at Rs. 4 per ton.
per hour. Find the most economical speed and the cost of a voyage of 1980

(P.V. 1949)

miles.
11.

Show

maximum volume

12.

Show

that the right circular cylinder of the given surface and maxisuch that its height is equal to the diameter of its -base.

that the semi-vertical angle of the cone of


and of given slant height is tan~N2.

mum volume is
13.

surface

is

that the height of a closed cylinder of given


equal to its diameter.

Show

(D.V. 1952)

volume and

least

14.
Given the total surface of the right circular cone, show that when the
l
volume of the cone is maximum, then the semi-vertical angle will be sin
15.
Show that the right cone of least curved surface and given volume
has an altitude equal to 42 times the radius of its base.
.

16. Show that the curved surface of a right circular cylinder of greatest curved surface which can be inscribed in a sphere is one-half of that of the

sphere.
17.

well as

its

A cone is inscribed in a sphere of radius r


curved surface

is

greatest

when

its

altitude

prove that

its

volume as

is 4r/3.

Find the volume of the greatest cylinder that can be inscribed in a


18.
(D.U. 1955)
cone of height ft and semi-vertical angle a.
the length
n
one
times
have
to
is
box
edge
19. A thin closed rectangular
of another edge and the vojume of the box is given to bev. Prove that the
least surface s is given by

DIFFERENTIAL CALOVLUS

164

JO. Prove that the area of the triangle formed by the tangent at anj
9
point of the ellipse x*la*+y*lb **l and its aves is aminimuu for the point

21. Find the area 9f the greatest isosceles triangle that can be inscribed
in a given ellipse, the triangle having its vertex coincident with one extremity

ef the major axis.

(Allahabad 1939)

A perpendicular is let fall

from the centre to a tangent to an ellipse.


Find the greatest value of the intercept between the point of contact and the
22.

foot of the perpendicular.

show
23.
;
tangent to an ellipse meets the axes in P and
least value of PQ is equal to the sum of the semi-axes of the ellipse
that PQ is divided at the point of contact in the ratio of its semi-axes.

that the

and also

24.
is the foot of the perpendicular drawn from the centre O on to the
2
2
tangent at a variable point P on the ellipse x*la +y lb*=l (a>b). Prove that
2
2
the maximum area of the triangle OPN is (a
i )/4.

25.
One corner of the rectangular sheet of the paper, width one foot, is
folded over so as to just reach the opposite edge of the sheet ; find the minimum length of the crease.

a^x^b,

26. If f'(x) exists throughout an interval


prove that the greatleast value of/(x) in the interval are either /(a) and/(fc) or are given by
the values of x for which /'(*)=0.

est

and

grocer requires cylindrical vessels of thin metal with lids, each to contain exactly a given volume V. Show that if he wishes to be as economical as
8
possible in metal, the radius r of the base is given by 2wr =V.
If, for other reasons, it is impracticable to use vessels in which the diameter exceeds three-fourths of the height, what should be the radius of the base

of each vessel

(P.U.)

/, feet
long is in the shape of a frustum of a cone the
radii of its ends being a and b feet (a>b). It is required to cut from it a beam
of uniform square section. Prove that the beam of the greatest volume that
can be cut is alll(a-b) feet long.
(Agra ; P.U.)

27.

tree trunk,

28. Find the volume of the greatest right circular cone that can be described by the revolution about a side of a right-angled triangle of hypotenuse
1 foot.
(P.U. 1940)

29.
rectangular sheet of metal has four equal square portions reat the corners, and the sides are then turned up so as to form an open
rectangular box. Show that when volume contained in the box is a maximum,

moved

the depth will be

where

a,

b are the sides of the original rectangle.

(Banaras 1953)

The parcel post regulations restrict parcels to be such that the length
plus the girth must not exceed 6 feet and the length must not exceed 3 feet
Determine the parcels of greatest volume that can be sent up by post if the
from of the parcel be a right circular cylinder. Will the result be affected if the
30.

greatest length permitted were only If feet.


31.

Show

that the

maximum

(Patna)

rectangle inscribable in a circle


(P.C7.

is

a square.

Suppl 1944)

CHAPTER VIII
EVALUATION OF LIMITTS
INDETERMINATE FORMS

We know that x ->

8'1.

a,

hmF(x)
F(x)
so that this theorem on limits fails to give any information regarding
the limit of a fraction whose denominator tends to zero as its limit.
Now, suppose, that the denominator F(x) -> as x -> a.
The numerator f(x) may or may not tend to zero. If it does
not tend to zoro, then f(x)/F(x) cannot tend to any finite limit. For,
if possible, let-it

tend to

finite limit,

say

/.

We

write

we have

so that, in this case,

lim/(x)=lim

[fe

F(x)

lim

lim

F(x)=/.0=

Thus we have a contradiction.


Three types of behaviour are possible in this case. The fraction
oo or it 3 limit may not exist. For example,
to + oo or
tend
may
when JC -> 0, (so that the limit of the denominator in each of the
,

following cases is 0),


8
(i) lim (I/* )

we

see that

=+oo

(Hi)

zero

lim

(I I x)

(//)

lim(l/-Jt )=:-ao

does not exist.

The case where the limits of the numerator of a fraction is ale


more important and interesting. A general method of deter-

is

mining the limit of such a fraction will be given in this chapter. For
the sake of brevity, we say that a fraction whose numerator and
denominator both tend to zero as x tends to a, assumes the indeterminate form 0/0 for x=a.

may be

of interest to notice that the determination of the


dyjdx is itsolf equivalent to finding the limit
of a fraction 8yjbx which assumes an indeterminate form 0/0.
It

differential co efficient

Other cases of limits which are reducible to this


be considered in this chapter.
<(.x)

form

will

also

In what follows it will always be assumed that /(x), F,x) and


possess continuous derivatives of every order that come in

question in a certain interval enclosing x^=a.

165

186

DIFFABBNTUL OALOVLV8
8-2.

The Indeterminate form


lim

-^

To determine

x-*a
when
lim

/(x)=0

Urn F(x).

As/(x), F(x) are assumed to be continuous for

f(a)= lim

By

yfr-0 ;*

x=0, we have

Fx=Q.

lim

Taylor's theorem, with remainder after one term,

we have

Hence

\f

y._
f,\

ttli

This argument fails if F'(a)=0.


~
has already been discussed in

The

J \ )
Tflf/**\i
Jj \CL\

case

:f

pt

when F'(a)=Q but

8-1.

Now, lQtf'(a)=F'(a)=0.
Again, by Taylor's theorem with remainder

after

two terms,

we get

Hence
lim

-^-li

lim

h^(
The case of failure which
as before*

In general,

arises

when -F"(a)=0 can be examined

let

f(a)=*f'(a)=*f"(a)=*

F
Mid

INDETERMINATE FORM 3

By

167

Taylor's theorem, with remainder after n terras, we get


An
/iti-i
...+
(a )JL. f( a +6n h)

^Lfn-i

-l

h
F^(a)+ ~- F(a+0' nh)

F(a+6' nh)

Hence

Ex.

Determine lim

1.

-.
sin

where x ->0.

.,'

/(0)=0,
F(0)=0.

-.

F(x)=x

sin

jc

-.

f'(x)=e*+e-*- r

~,

{ F'(x)=x cos x+sin x.

/'(0)=0,

'.

F'(0)=0.

...

/" (0 )= 2>

Again

)=

_hm ----

sin

JC+2 cos

F"(0)=2.

x,

_.>

The process may be conveniently exhibited as

follows

:-

*- e-_21og(l + x)

x cos

^o
Ex.

2.

e'-e-a +2/(l+x) 8

-x sin x+2

cos

x~

_
~

Find the values of a and b in order that


x(l
a cos
f
x) - b sin x
^
fim
nm
3
*
x~>0

ay be equal

to

'-

'

U. 1944, 1959)

DIFFERENTIAL CALCULUS

The function

of the form (0/0) for

is

all

values of a and

6.

-frsinx

->-i- ----hm x(l+acosx)


x-3'
,.

x ->

l+a

,.

cos

x -> .

finite

ax sin x
3x 2
JX

b cos x
'

The denominator being for x=0, the fraction will tend to a


limit if and only if the numerator is also zero for x=0. This

requires

Again supposing

l+a

we have

this relation satisfied,

cos

xax sin x

b cos x

3x2
6X

2# sin
------

= hm
..

xax cos x+& sin x


-

,.

bx

x-*a

30 cos x+tf* sin x+fe cos x

*}/

As given,

From

Ex.

and

(1)

=1,

(2),

i.e.,

3a = 6.

we have

Determine the limits of the following

3.

....

sm x x

x..v

lv

cos x

cosh

A:

cos

( V)7

(x->0).
v

log cos x

,- M -,,-v

(vi)

Jog(14-6x)

A:

-.

(D.C/. //b/i5. 7952)

Ex.

Evaluate the following

4.

Jim

*"
.

^^

Af

.(D.U.1952)

(//)

lim

(D. U. Hons. 1951, P.U. 1957)


i

"

'

1'

0*1*

.*

*v

^v

r\ T

(D.U. 1955)

INDETERMINATE FORMS
Ex.

If the limit

5.

169

of

2x+a

sin

sin

X9
as

x tends to zero, be

the value of a and the limit.

finite, find

(P.U.)

Preliminary transformation. Sometimes a preliminary


transformation involving the use of known results on limits, such as
8*3.

..

hm

sin

x ->
simplifies the process

tan x

..

lim
* =1, x->

a good deal.

=1

These limits

may

also be used

to

shorten the process at an intermediate stage.

Ex.

hm 1+sin

*
1.

i?

|.

ITFind

cos

(1x)
x+log
J

x tan 25x

~> 0).
(x
v
'

The inconvenience of continuously differentiating the denomiwhich involves tan 2 x as a factor, may be partially avoided as

nator,

We

follows.
1

write

+ sin

cos

x+log

x)

(1

x tan^x
1

~~
1

,.

101

+ sin xcos x+log


x

+ sin x

x)

\tan x/
(1

x)

xcos x+log (1

x)

Ax 3

1+sin
-

\2

"

cos x+log
x tan 2 x
1-j-sin

(1

x-*

xcos~ x+log
-

(1

x)
~

x >0
1

lim
jc^O
tor

+ sin

cos x+log(l
-3
x

x)

To evaluate the limit on the B.H.S., we notice that the numeraand denominator both become for x=0.
l+sin x cos x + log (I x)
..

hm

x3

cosx fsui-v-

_x- T

-cos x^sin

------

DIFFERENTIAL CALCULUS
Ex,

Evaluate the following


cosh # cos x
..

2.

x un x

We

have

xcos x = cosh x cos


^g
x
^

cosh

x
*

----- cos x =

cosh x

,.

.....

lira

xsmx

cos x
-----x*

lim

cosh

CQ sh

cos

cos

..
.

hm

and denominator are both zero

Since the numerator


therefore
rvi

cosh x

,.

I*
i

sin

sinh
__ ,.
Aim
^

for

x=0,

*+sin~~x
OY

cosh
x
lim ____ - x+cos
iim
J
x->0
,.

=1.
Ex.

3.

Determine the

(i)

limits of the following functions

te -

-"J~.

dV) log (1

0).

-) cot

x.

->

0).

*..(- o,

8-4.

The Indetermine form

~.

Tc?

determine

00

ton

lim

*Let f'(x)IF'(x) -+
also ->

/(x)=
I

oo

asx-+a.

//m F(x).
It

mil be shown

that f(x)/F(x)

/.

Suppose, x>a. As f\x)IF'(x) -> /, when x -> a, we make it


arbitrarily near / by taking x sufficiently near a, say between a, and

a+8.
I* Another

proof is given io the following sub-section.]

INDETERMINATE FORMS

171

We now

take any two numbers c and x which lie between a


apply Cauchy's mean value theorem f(x) and F (x) for
the interval (c, x). We thus have

and

a+8 and

where

We

F(x)-F(c)
between c and x and, therefore, between a and a

lies

re-write

(/)

as

7"T~T~I
x

'/'(I)

F(x)

l-FWlFWJ'F'fc)

a+S

c fixed, we make x -> a.


Tlien/(c) and F(c) are fixed
to infinity.
tend
and
our
F(x)
and, by
hypothesis, f(x)

Keeping

Therefore,

F(c)
1

as *~* 00

made arbitrarily near / by taking x


as x -> a through values greater than
Similarly,
values less than

'

Thus

^)/^W

Y-f(c{ f(x)

it

sufficiently near

can be

a so that

it

-W

a.

may be shown that/(x)/F(x)

-*

as

x -> a through

a.

Hence

^ hm

hm

,.

,.

/'(*)

when
lim

/x=

oo

=s lim

Note. The above investigation rests on the hypothesis that /'(x)/F'(*)


tends to a limit as x -+ a. This part of the hypothesis necessarily implies that
x near a so that f'(x), F'U) both exist
f'(x)IF'(x) has a meaning for values of
and F'OOs^O for such values. This justifies the use of Cauchy's mean value
theorem in the above investigation.

8*41.

A proof of the

above result

lim /(*):=

oo

is

also often given as follows

and lim F(x)=

A*)
F'(x)

3BES
f'(x)

oo

172

DIFFERENTIAL CALCULUS

Let
lim [f(x)IF(x)]
.-,

from

we

(1),

where /^O and

get

Supposing now 1=0.

/+.-U-

.-.

Applying the above

+ l-

result,

F'(x)
/

Finally

By

let

/=*>

= m

)
-

...(8)

so that

the preceding result,


.

..

0=hm

F(x)
.
-

..

=lnn

F'(x)
r' -<
-

- (4)

i)*'
Hence we

see that always

when lim/JC
,

= 30

^- ^
"m

<>

lim JPx.

Note 1. The above second proof is incomplete in the sense that it


assumes that Urn lf(x)IF(x)] exists. The first proof did not assume this
existence.

Note

2.

While evaluating
Urn

-j~r*

when

it is

of the form

we must try to change over to the form 0/0 as soon as it may be conveniently
possible, for, otherwise we may go on indefinitely without ever arriving at the
end of the process.

INDETERMINATE FORMS
Ex.

1.

Determine lim

y-p^

173

L, as

Here, the numerator and the denominator both tend to


tends to a.

oo

as

log(x-a)

.-.

lim T --~-JT --L

..

Inn

lim

a
=*

-JT-

e*

Ex.

2.

Determine the

w,.

Hint,
(v/)

8-6.

limits of the following functions

log

tan x

tan 2x =

log (l~.v) cot (xw/2),

U ->

The indeterminate form

log tan 2x
log tan *

1).

0*oo

7b determine

Urn [/(x).
x->a

when
lim /(x)=0,

To determine

this limit,

we

write

so that these new forms are of the type 0/0 and oo /oo respectively
8-4.
limit can, therefore, be obtained by 8*2 or by

and the

In this case, we say that /(*). f(x) assumes the indeterminate

form

O'oo

Ex.

at
1.

x=a.
Determine lim (x logx), as x->0.

We write
x

log

x-

_.

f/i

lim (x log

x)-

lim

lim

^
r=

(*-)

DIFFERENTIAL

The reader may

see that writing

xlogx=

(I/log x)

which is of the form


(0/0) and employing the corresponding result of
8-2 would not be of
any avail.
'

Note

w e know that

haye

does nottend to a limit as x -*

lx

lim
=00,
*-(0+0) *

Ag ~ in

0.

In fact

*-(0-0)

x is defined for positive values of x only so that there


* *
question of making x ->
through negative values while determining
log

'

we

is

no

lim (xlogx).

x->0
Thus, here x <+
Ex.

(/)
(///)

8-6.

really

means x -> (0+ 0) so

Determine the

2.

x log tan

x,

that

/*

does tend to a limit.

limits of the following functions

(x -*

0).

->

The Indeterminate for


lim

x -> a

x tan (w/2-x).

(//)

(a~-x) tan (nxfta), (x

(x

-*

0).

0).

oo

<*>

7b determine

[f(x)-F(x)l

when
lim /(x)=<x>

X->fl

We

write

so that the numerator and denominator both tend to


as x tends to
a.
The limit may now be determined with the help of 8*2.

In this case, we say that [/(x)


nate form oo
oo for x=a.

F(x)] assumes the indetermi-

Note. In order to evaluate the limit of a function which assume! the


form, oo -oo , it is necessary to express the same as a function which assumeg
the form 0/0 or oo/oo.

Ex.

1.

Determine

We write

___

_1

x-2
and

see that the

log

new form
*.

tog

(X- !)-(*'

(x-1)
is

(x-2)

(x-1)
of the type 0/0 when x -*

_ __L__1

log

2.

iog(y~i)-(y~2

175

FORMS

for x=2. On
The numerator and the denominator are both
the
that
we
show
of
the
method
required limit is
8-2,
may
using

Ex.

limits of the following functions

Determine the

2.

'

The Indeterminate forms,

8'7.

as

'

oo

To determine

when
(i)

(fl)

limf(x)=:Q

F(x)=0.

limf(x)=0; lim F(x)=<x>

(Hi) //w/(jc)

We

lim

= oo

lim F(x)=0.

write

so that

In each of the three cases, ^ e see that the right hand side
assumes the indeterminate form 0. oo and its limit may, therefore, be
determined by the method given in 8*5.
Let

lim

x->

[^(*),log /(*)]==/.

;\

limlogj;=/,

or

log lim

Hence

lim

or brevity,

forms

we say

1,

oo

y=l or

that

j/(x)

lim y=e*.

^1

respectively for

assumes the

x=a.

indetermi-

DIFFERENTIAL CALOVLVS

176

Ex.

Determine

I.

x~a

Urn (x

a)

os x ->

a.

form).

(0

Let

lim

log>>= lim

{} ^~^
I

~~
.

Hence

log lim }>=0,

Thus

i.e.,

lim

lim y~e<>

x~a

lim (x

(xtf)=0.

=I when

a)

Note. Here it is understood that x tends to a through values greater


x ~ a would
for otherwise the base (x-a) would be negative and (x a)
have no meaning.
than

a,

Ex.

2.

Determine
lim (cos x)

1/x 2
tfs

1/x

Let

y=(cos
IT

\f\Ct

iv'fti

,.

,.

hm logy= hm

=
log (lim y)

..

lim

i or

x
"~

log cos
---^

tan x
~
s
^JT

lim (cos x)

JL-

y=e

lim

x~>0

Hence

0.

x)

cos
log
^^

x ->

'

'

*.

x->0
Ex.

3.

Determine the

->

(/)

x*, (x

ii )

(cot x)

sin

limits of the following functions

0).

(//)

^
,

(x

->> 0).

(l

~' x

)~\ (x ->
tan x

(/v)

(sin x)

(x

(P.C7. 7923)

1).

->

1.

12).

INDETERMINATE FORMS

177

(v)

(vii)

x*

(viii)

(D.C7. 7949)

(x-->0).

(x ->1).

(P.C/.

7957)

Exercises

Determine the limits of the following functions


e_e- x
12.
o
->0).
v
^ 2 sin A: (x

3.

1+^cos A:~cosh x

(2jc

tan x

^3

,
'

(A:
v

-> <>.
'

log(l-f x)

7.

log
& -AC

>

>-

n sec x),

(A:

->

n/2).

j_
8

log x
.

(cot *)

u -> o).

a*

10.

13.

(cos ax)

(x -> 0).

14.
15.

tan
16.

(2-O

17.

sin

-=-

2a

Ct

\w

',

(*->).

(B.V.1953)

sec 2

20.

cot x
(sec x)

(x -> n/2).

21.

(2-x)

tan

ff

(x^-

1).

'

DIFFIRENTIAL CALCULUS

178

a t5>*', (,+,).
'*,

..

26.

(fl.tf.

[For solutions of Ex. 25 and Ex. 26 by Infinite series refer

log

cos

Hons. 1959)

9-5, p. 185

sect*_
sec

ijc

28.

!+-

,(*->

co)

31.

Evaluate

sinxi
lim

32.

If

/(0)=0,

show

that tho derivative of every order of


/(x) vanishes for

/
33,

x=0,

/.<?.,

(0)=0,foralln.

Discuss the continuity of f(x) at the origin when

/(*)=* log

sin

x for

x^O and/(0) =0.


(D.U. Hons. 1955)

CHAPTER IX
TAYLOR'S INFINITE SERIES
EXPANSIONS OF FUNCTIONS
9-1.

Infinite Series.

Its

convergence and stun.

Let

be an

infinite set of numbers

given according to some law.

Then a

symbol of the form

an

Here each term is followed by another so


no last term.
If we add the first two terms of this infinite series, and then
add the sum so obtained to the third, and thus go on adding each
term to the sum of the previous term, we see that, as there is no
last term of the series, we will never arrive at the end of this process.
In the case of a finite series, however, this process of addition will be
completed at some stage, howsoever large a number of terms the series
is

called

that there

may

infinite series.

is

consist of.

Thus, in the ordinary sense, the expression 'Sum of an infinite


has no meaning. A meaning is assigned to this expression by
employing the notion of limit in the manner we now describe.
series',

Let Sn denote the sum of the first n terms of the series so that
a function of the positive integral variable n. If Sn tends to a
finite limit S, as n tends to infinity, then the series is said to be comergent and S is said to be its sum.

Sn is

In case, Sn does not tend to a


does not converge.

finite limit,

we say that the

The question of the sum of a non-convergent

series

series

does not

arise.

We may find an approximate value of the sum


infinite series

by adding a

sufficiently large

Consider the infinite geometric series

Illustrations.

l+r+r*+r* + ......

We know

of a convergent

number of its terms.

that

#n ,=-

1
JL

r*
*f

Sn =n

when
179

DlFFEBENTIAL CALCULUS

180

We have now to
[Refer

examine lim Sn when n ->

oo

3-61, p. 57].

r=l, S n =n which tends

For
For

<

1,

lim r"=0,

r>

1,

lim

r
k

For
For

r^

1,

lim

<

only if

1,

and

the

Sn

does not

exist.

geometric series converges if and

infinite

sum of the

infinite

2."

Sn =l/(l-r).
therefore, lim S n =oo

and, therefore, lim

Hence, we see that the

oo

so that lim

=oo and,

to

infinite series then,

is

We suppose that a

given function

series.

Taylor's
f(x) possesses derivatives of every order in

an interval

Then, however large a positive integer n


and 1, such that
0, lying between

may

[a,

1/(1

r).

a+h].

be, there

exists

number

f(a+h)=f(a)+hf'(a)+

=^

where

*^
n

(Taylor's development with Lagrange's form of remainder.)

We

write

so that

Suppose that R n ->

0,

lim
so that

we

as n ->

ao

It

is

then clear that

Sn =f(a+h),

see that the series

f(a)+hf'(a)+
converges and that

*",

its

sum is equal

to /(

Thus we have proved that


(0
[a,

(/*/(*)

possesses derivatives

a+H] and
(ii)

the remainder

tends to

as n tends to

infinity,

then

of every order

in

the

interval

EXPANSIONS
This

known

is

as Taylor's theorem

j(a+h) in an infinite
power series in h.

The

as Taylor's series.

infinite series.

series,

for the

we

for a

Putting

(ii)

for

in

in

the interval [0, x]

the remainder

tends to

as n tends to

This

known as

is

expansion of/v x) in
i.e.,

and x

see that if

(0 /(*) possesses derivatives of every order

and

x,

known

Maclaurin's

the Taylor's infinite

development of
ascending integral powers of h, i.e.,

series of

series (1) is

9'3.

181

power

The

an

infinity,

then

Maclaurin's theorem for the development or


infinite series of ascending integral powers of

series in x.

series (2) is

known

as Maclaurin's series.

Note. It may be seen that instead of considering Lagrange's form of


remainder, we may as well consider Cauchy's form.

Formal expansion of functions. We have seen that in order


any given function can be expanded as an infinite
Taylor and Maclaurin series it is necessary to examine the behaviour
9-4.

to find out if

of R n as n tends to infinity. To put down R n we require to obtain


the general expression for the nth derivative of the function, so that
we fail to apply Taylor's or Maclaurin's theorem to expand in a
power series a function for which a general expression for the nth
derivative cannot be determined
Other more powerful methods
have, accordingly, been discovered to obtain such expansions whenever they are possible. But to deal with these methods is not within
the scope of this book.
,

Formal expansion of a function as a power series may, however,


be obtained by assuming that it can be so expanded, i.e., R n does
as n tends to infinity.
Thus we have the result
tend to
:

then
Iff(x) can be expanded as an infinite Maclaurin's series,

/(*)=/(0)+*/'(0)+

**,

/*(<))

+ ......

...(1)

Such an investigation will not give any information as to the


range of values of x for which the expansion is valid.

To obtain the expansion of a function, on the assumption that


values of its derivatives
possible, we have only to calculate the
x=0 and substitute them in (1).

it is

for

182

DIFFERENTIAL CALCULUS
In the Appendix, we shall obtain the expansions of

e, sin

x, cos x, log

(1+x). (l+x)

without assuming the possibility of expansion by actually examining


the behaviour of R n for the functions,

In the following, however, we obtain these and other expansions


the possibility of expansion.

by assuming

Expansion of

9-41.

e*.

Let

Thus
Substituting in the Maclaurin's series,

we obtain
-v

-v-3

which

is

known

Cor.

1.

as Exponential Series.

Changing x into x log a in


* log *
,

=e

This result

(1),

we

get

= 1+(x log a) +

may

also

be obtained directly by employing Maclau-

rin's series.

Cor

2.

Putting,

for x,

we obtain

.LI
+ 2! +
L

from which we

\
~3!

L.4.
+ '"
+ ~4!
JL

obtain the values

may

of,

e,

upto any number of

decimal places.
9-42.

Expansion of sin

x.

Let
/(*):=: sin

x
mr

Thus
/-(0)-sin
1,

so that

we

-*

AOJ^O,/^)^-!,

/-'(0)-0,

etc.,

see that the values of / (0) for different values of n

a successively appearing periodic cycle of four values


1,0,

-1,0,

form

183

EXPANSIONS

Making these substitutions


ths sine

in the Maclaurin's series,

we obtain

series.

sin

*=*-__

We may

9-43.

similarly obtain the cosine series

x2

cos

Expansion of log

9-44,

x 2n

xl

*=:!-_ +_-_ ...... +


(1

(-!)

+x).

Let

Thus

.\/'(0)

= l,/"(0)=-l

Making substitutions
Logarithmic Series
log

/'''(0)==2

in the Maclaurin's seriep,

wo

so on.

obtain the

(i+x)=.x-^+-^+

9 45.

!,/""(0)= -3 land

Expansion of

(1

+x)

......

+ (-i)"-i.

..... .

OT
.

Let

Thus

Making

substitutions,

2f
In case

is

we

obtain^the Binomial Seriesj


......

a positive integer,

n\

we obtain

a finite series on the

right.

Note 1. If we examine the behaviour of R n as is done in the Appendix,


can show that (i) the Exponential, sine and cosine series are valid for every
value x, (ii) the Logarithmic Series is valid for
Kx<J, and (in) the Binomial
Series is valid for
Note 2. In the following we shall consider some more cases of expansions of functions, in each case assuming the possibility of expansion. It will
be seen that in some cases we may also obtain the expansion by vsing any of
the series obtained above.
,

we

184

DIFFERENTIAL CALCULUS
Examples
1.

Assuming
as the term inx*.

the possibility

of expansion, expand, tan

as far

x,

Lot

/(x)=tan
/.

/'(x)=sec

x.

x=l+tan 2 x.

/"(x)=2 tan x

=2

sec 2 x

tan x(l+tan 2 x)

= 2 tan x+2 tan


/'"(x)=2

sec

x+6

x.

tan x sec 2 *
2

=2 (l+tanx)+6 tan 2 x (l+tan2 x)


= 2+8tan 2x+6 tan x.
/ '"(x) = 16 tan x sec 2 x+24 tan 3 x sec2 *
4

= 16 tan x +tan x) -f 24 tan x(


= 16 tan x -f40 tan x+24 tan x.
= 16 sec x + 120 tan x sec x + 120
(

/ v (jc)

+tan 2 x)

tan 4 x sec 2 x.

Thus
Substituting these values in the Maciaurin's series
obtain

anx-x+

9*4,

+
e

Assuming the possibility of expansion, expand /(x)


ascending integral powers of x.
In Ex. 1, 5-6, p. 124 we proved that
2.

in

(w 2 (2 2 +w 2 )(4 2 +w 2
2
2
2
2
(m (! +m )(3 +m

[(w-2)

-fw 2

+m

2)

[(

Substituting these values in the Maciaurin's series,

Use of known
3.
find the expansion of

2i

by

the term in

w(l -fw
o

2
)

(2

m sm x

2
]

we

+m

By Maciaurin's theorem

series.

sin

upto and including

we

n even
w odd.

get

2
)

or otherwise

(e*-l),

x4

First four derivatives are neaded to expand the given function


Maciaurin's theorem upto the term required.

The required expansion can also be obtained by employing the


Exponential and the sine series and thus avoiding the process of calculating the derivatives which is often very inconvenient.

EXPANSIONS

=z,

Now,

sin (e

say.

l)=sinz
z
O

z5

v*2

"v4

"vS

+
2T

9*5.

185

3~"T

+ 4l +

Use of infinite series for evaluating the limits of indeterThe following examples will illustrate the procedure.

minate form?.

Examples
1.

find Urn

Using the

?*?

*-*-**
(D.U.1953)

infinite series for e

e" sin

x) 1

.x

x
,

sin

x and

log (I

.v),

we obtain

= lim
=lim

= lim
2.

-i-i

Eh
(0

(//)

lira

x->0

'

-_e

ie v
:

(D.U. Hons. 1954)


(Cf. Ex. 25,

Ex. 26,

p. 178)

186

DIFFERENTIAL CALCULUS
Let

y=(i+ X
log

7=

l lx
)

log(l+x)

2
L

(T>

or

'6

as

1^

O
3

Q
2

-e.e.

24

IX

JL

./V

+-STV-

'

2
/

""

24

llx
+ "24

'

Exercises
(/w f/w following ,
1.

//re

possibility

of expansion may also be assumed)

Prove that

187

EXPANSIONS
2.

Prove that
e a * sin

3.

b -~-

bx=bx+abx*+-

*+....

Obtain the wth term in the expansion of tan" 1 x

in ascending

of*.

powers

(P.U.1951)
4.

Show

5.

Prove that

6.

If >>=log

x=*l-x 2 +$x*-^\x

that cos 2

e x sin

*= *

-f * -f

Jx

......

____

2
[x+ >!(!+ x )], prove that

Differentiate this n times and deduce the expansion of y in ascending


in the form

powers of x

yss *~

x3

T*

x5

+ T-' 4

x7

'

'

'

~5~~ ~2~' 4

'+
,

'

'

'

(P.C/.

7.

If >>=sin log

2
(;c

7P57)

-}-2x-M), prove that

Hence or otherwipe expand


8.

'

.v

ascending powers of x as

in

far as

Prove that
.
,

9.

Show

that
2)

Obtain the following expressions

10.

(/)

log tan
(l

(i"fx)=Zx+*x

log

(v)

tan
^

__.1

e.^!-

(vi7)

A:

(m

8)

-JxH ----

log sec

T~ 2T~

=-^ +

xM ""
A'

47

"90"

'

2"+fi

30"

'

'

x4

^c

sm-^^xl,1
i

(v/)

x2

1
4

...

-f

+ sin
1+

(iv)

2!

^8

'

""

30
l

X*

x5

"5

'

--f-2^-

*=2> +

12

+ *"-

4"!

^+45

x%

-+

+"-

(D.V.1953}

188

DIFFERENTIAL CALCULUS
log sin (x+/i)=log sin

(v//7)

x+ h

cot

/i

+JY
-

tan-Ms+AJ-tan-^+Asinz

(/AT)

2
cosec x

*--yT

-~?

-(A

1
sin z) --

2*

--

v.sin3z
8

.,

x- ----

2 cot x cosec

4-(Asmz)

...

when zcofr^x.
Obtain the expansion of e

11.

sm X
in

powers of x as

far as

x4

(D.I/.
12.

Obtain

the

first

three terms of the expansion of log

Powersoft
13.

App]y Maclaurin's theorem

as the term in x 8
14.

to find the expansion

Expand log

sin

in

powers of U-3).

2
Expand 3x3 -2x 4-x-4

16.

Show

that

(/)

Uin

in

x
in

lim

(//)

and A by

x~3

in

sin

"-.
x-smx

=1.

*3

Obtain by Maclaurin's

powers of x-2.

*-55=;
^

x->0
X

(D.U. 1955)

15.

sion of e

in

(P. U. 1955)
of e x l(e x 1) as far

[Write log sin x=log sin (3+ x-3) and replace x by


the expansion of log sin \x+h) in powers of h.]

17.

(1-Manx)

16

theorem, the

first

four terms of the expan-

ascending powers of x. Hence or otherwise show that

xcosx

_
~-A-

lim

x-0

--

=3.

smx

Appendix

We shall now
e*,

obtain the expansions of

sin x, cos x, log

(1+x),

(l

+ x) m

without making any assumption as to the possibility of expansion,


r
Expansion of e

x
n
Let /(*)=*.
Therefore, f (x) = e so that/(x) possesses
vatives of every order for every value of x.
Taking Lagrange's form of remainder, we have
,

..

deri-

'

V!

We know

that

when

->

oo

x//i

whatever value x

may

have.

->

(See

3*62, p. 58)

EXPANSIONS

Rn

->

as n ->oo

189
.

Conditions for Maclaurin's infinite expansion are thus satisfied.

/(0) = e=l

Also

Making

v-2

--l+*+
which

is

we

substitutions in the Maclaurin's series,

2!

get

v-n

y3

3,

+-;,-!+"

valid for every value of x.

Expansion of sin
Let f(x)= sin x.

x.
.-.

f n (x) = sm

(x

+ ^nn),

so that/(x) possesses derivatives of every order for every value

We

of

jc.

have

Since
*

x"
n\

we

see that

Rn

->

as n -> oo for every value^of x.

Thus the conditions

Maclaurin's infinite

for

Now

satisfied.

expansion are

n*

/"(0)=sin

Making these

substitutions in the Maclaurin's series,

sinx=x
f

which

is

-j-

,-

we

get

....

valid for every value of x.

Expansion of cos

x.

cos x
for

As above, it may
2
x4
x6
_i x
2

j-

+-4

easily

be shown that

every value of x.

Expansion of log (1+x).

Let

We know that log (1+*)


(l+x)>0,

i.e.,

Moreover

for

x>~l.

possesses derivatives of every order for

DIFFERENTIAL CALCULUS

190
If

Rn

denotes Lagrange's form of remainder,

<

Let

we have

<;i, so that x/(l+0.x) and, therefore,


whatever value n may
[xl(l
(tx)]
positive and <1,
Since, also, 1/n -^ 0, as n -> oo we see that R n -> 0.
(i)

is

also,

have.

Thus
/? n

Let

(ii)

as H ->

->

1<JC<0.

In this case x/(l+fix)


so thafc

we

when 0^x<J 1.

oo

to

fail

may

draw any

not be numerically less than unity


from Lagrange's form

definite conclusion

Taking Cauchy's form of remainder, we have

Here

(1

0)/( !

+ #*).

is

positive
v

and

less

than

1,

and

i_

Also

Hence,

x n ->

as n -+

co

Rn

as n ->

oo

->

3'61, p. 57)

the conditions for Maclaurin's theorem are satisfied for

Also

in the Maclaurin's

Making these substitutions

x*

series,

we get

(l+x)=x- -2+-$-- ~T~ + ---.^

log
for

v3
A:

<v2

A.

./>

!<*<!.
m
Expansion of (l+x)

(m

is

any

real number).

Let

/w=(i +*r-

Whenm

is

any

real

number, (l+x)

vatives of every order only

when

+ x>0,

possesses continuous deri/.^.,

when

x>

1.

Now,
m
f(x)=m(m-l)(m-2) ...... (m-n+l)(l+x) ~.

We notice that

if

m is any positive integer, then the derivatives

EXPANSIONS

191

otf(x) of order higher than mth vanish identically and thus, for
m
n
w, R n identically vanishes, so that (l+x) is expanded in a
finite series consisting of (m+l) terms.

>

Ifwbe

not a positive integer, then no derivative vanishes

we have

identically so that

If

Rn

to examine this case

still

denotes Cauchy's form of remainder, we

further.

get

Let
1

<

<

1, i.e.,

<

1.

Also

0<0<
< 1<(

or

<

1.

l+Ox,

<

;-}

1,

0<
Let (m

1) be positive.

Now

< l+9x <

ml be

Let, now,
Now, since

negative.

x
I

or

X )-!.
(l + fl*)-l^(l_
we know that
m(m
l)...(w
w+1)
^
_J
""*
when
\

*Also

--

Thus,

Hn

~>

The conditions
fied.

"ZTiri

as n -> oo

when

<

<

1.

for Maclaurin's expansion are, therefore,

Now
/(0)=w(w-l) ...... (w-n
Aefer to the foot note on the next page.

satis-

DIFFERENTIAL CALCULUS

192

Making substitutions, we get


m(m--l)
,+
xm t
L__/xa
,

ft

(l+x)*=l+mxH

when

<

<

A:

JL

l)(m2) x
-v
+ m(m
3
,

'>

,
3

-f ...

1.

Ex.

1.

Justify the Maclaurin's expansions of


1
e ax sin bx e ax cos bx, tan" x.

Ex,

2.

Show

3.

Show

that log

x and cot x cannot be expanded

as

Maclaurin's

series.

Ex.
f(x)

that the Maclaurin's

infinite

expansion

is

not valid for

where

e~

when x^O and /(0)=0.

[Refer Ex. 32. p. 178.]

*To prove

that

when

<

1) ____

Changing n

to w-f

1,

we

1,

(w

"^

get

or

As

p such

<

we can

1,

<

that

Wn

>

k, for

1,

and a positive integer

<

we get

Now

up

//c^ is

a constant and k* ->

and so also Un ->


'

<

Thus

/>.

Multiplying

number k

find a positive

as n >oo.

as w

->

oo.

as n

->

oo.

CHAPTER X
FUNCTIONS OF

TWO

VARIABLES

PARTIAL DIFFERENTIATION
The notions of
two

10*1.

continuity, limit and differentiation in


variables, will be briefly explained in
few theorems of elementary character will also be

relation to functions of
this chapter.

proved.

The sub ject of functions of two variables is capable of extension to functions of n variables, but the treatment of the subject in
this generalised form is not within the scope of this book.
Only a
few examples dealing with functions of three or more variables may
be given.

102. Functions of two variables. As in the case of functions


of a single variable, we introduce the notion of functions of two
variables by considering some examples.

The

(/)

between

relation

x, y, z,

of numbers x,

determines a value of z corresponding to every pair


which are such that x 2 + y 2 ^l.

y,

Denoting a pair of numbers x, y, geometrically by a point on


a plane as explained in
1-7, p. 14, we see that the points (#, y) for
which X 2 +J> 2 <1 lie on or within the circle whose centre is at the
origin and radius is 1.

The region determined by the point (.x, y) is called the domain


of the point (x, y).
Now, we say that the relation (1) determines z as a function of
two variables, x 9 y defined for the domain bounded by the unit circle

>=!.
(/)

Consider the relation

a<b c<d.
Now (a x)(x

where
is

non-negative

if

b)

is

non-negative

if

a^x^b

and

(cy)(yd)

ct^y^d.

The points (x, y) for which a^x^b, c^yt^d determine a


rectangular domain bounded by the lines

x=a x=b y=c,y=d.


y

The

relation (2) determines a value of z corresponding to every


Thus z is a function of x
point (x, y) of this rectangular domain.
and y defined for the domain.

193

194

DIFFERENTIAL CALCULUS

The

(ill)

relation

r2
V2
z=e x y

determines z as a function of

x,

defined for the whole plane.

In general, we say that z is a function of two variables defined


for a certain domain, if it has a value corresponding to every point (x y)
of the domain.
y

The
etc., as in

relation of functionality is expressed by the symbols /, ?


the case of functions of a single variable, so that we may

write z=f(x,y),
Ex.

</(*, y), etc.

Determine the domains of

definition of

Neighbourhood of a point
(x, y) such that

10-3.

Let 8 be any positive

(a, b).

The points

number.

determine a square bounded by the lines

x=a
Its centre

is

S,

at the point

hood of the point

(a, b).

x = a+8

y=^b~8, y

This square
(a, b).
For every value of,

is

called

S,

we

a neighboura neigh-

will get

bourhood.
10*4.

Let

Continuity of a function of two variables.

(a, b)

be any point of the domain of definition of

As in tho case of functions of one variable, we say that f(x, y)


continuous at (a, b), if for points (x, y) near (a, b), the value
f( x y) of the function is near f(a, b) i.e., f(x, y) can be made as
near f(a> b) as we like by taking the points (x, y) sufficiently near
is

>

(a. b).

Formally, we say that /(A: y) is continuous at (a, b), if, corresponding to any pre-assigned positive number e, there exists a positive
,

number

such that

1
P(a,b)

for

y)-f( a b )

f(*>

>

all points (x, y) in the

<e

square

Thus for continuity at (a, b), there


exists a square bounded by the lines

x=a

8,

such that,
square,

x=fl+S; y^b

for

any point
between

8,

y=b+8

(x, y)

of this

/(x, y) lies

f(a, fc)-e,
p.

j.

where
small.

e is

any positive number, however

PARTIAL DIFFERENTIATION

Continuity in a domain. A j"unction f(x, y) is said to be


if it is continuous at every point of the same.

10 41.
continuous

195

a domain

in

10-42.
Let /(.x, y) be continuous at (a, b) and
Special Case.
be any positive number, however small. Then there exists a
square bounded by the lines
let e

x-=a8, x = a-\-8

;>'=

S, >'

such that for points (x, y) of the square, the numerical


difference bet ween f(x, y) and f(a, b) is less than e.

value of the

if we consider points of this square lying on the


see that for values of x lying between a
8 and a
8,
the numerical value of the difference between f(x b) and f(a, b) is less
than e. Also, such a choice of S is possible for every positive number
This is equivalent to saying that f(X, b) is a continuous function
.
of a single variable x for x-a.

In particular,

line

y=b, we

It

of y

for

may

y = b.

be similarly shown that f(a, y)

a continuous function

is

Thus we have shown that a continuous function of two


is also

variables

a continuous function of each variable separately.


10

5.

Limit of a function of two variables.


lim/(x, y)

l,

as (x, y) -> (a b).


y

A function f(x, y) said to tend to the limit /, as x tends to a


andy tends to b, i.e , as (x, y) tends to (a, ft), //, corresponding to any
pre-assigned positive number z, there exists a positive number 8 such
that
is

\f( X ,y)-l\

for

<

all points (x, y), other than (a, 6), lying within the square,

This means that corresponding to every positive number e,


there exists a neighbourhood such that for every point (x, y) of this
neighbourhood other than f(a, b),f(x, y) lies between / e and /+e.
,

Limit of a continuous function. Comparing the


10-51.
104, and
10'5,
tions of limit and continuity as given in
y) is continuous at (a, 6), if and only if

defini-

we

see

that/(x,

limf(x, y)=f(a, b) as

(x, y)

->

(a, b),

ie.,

the limit of the function^ actual value

The same thing may


nuity at (a, 6), we have
as

(0,0).

also be expressed

of the function.

by saying that

for conti-

196

DIFFEBENTIAL CALCULUS
Partial derivatives.

10-6.

Let
*=/(*>

J>).

Then

if

it

exists, is

said to be the partial derivatives of f(x, y) w.r.t. x at

and is denoted by

(a, b)

3z

f_
V

or

8x /(a, b)

x at

fA
.(a,

b).
'

It will be seen that to find the partial derivative of/(x, y) w.r.t.


we put j equal to b and consider the change in th&

(a, b),

function as

x changes from a

to a

-\-h

so that

/(*, b)
is the

x=a.

ordinary derivative off(x, b) w.r. to xfor

Again,

if

it

and

15

exists,
is

is

called the partial derivative off(x, y) w.r. to

y at

(a, b),

denoted by

the ordinary derivative off(a, y) w.r. to

y for y=b.

If f(x, y) possesses a partial derivative w.r. to x at every point


of its domain of definition, then the values of these partial derivatives
themselves define a function of two variables for the same domain.
This function is called the partial derivative of the function w.r. to x
and is denoted by gz/gx or fx (x, y) or simply fx
.

Thus
fix v]-\im
lim
")
Jx\x>

<*

f(X+h
-------'

y)

-f(X

' y)
-----------

as/z-^0
AI -> u,
as

f,

We

where y is kept constant in the process of taking the limit.


can
similarly define the partial derivative of/, w.r. toy which is denoted

by

gz/8^, f*(x, y) or fy .

10 61. Partial derivatives of higher orders. We can form partial


derivatives of 3z/gx and dzjfty just as we formed those of z.

PARTIAL DIFFERENTIATION

197

Thus we have

which are called the second order


partial derivatives of z and are
denoted by

oZ

Y>

or

v,

lvx

respectively.

Similarly the second order partial derivatives

aj \dy
-are

respectively denoted by

Thus, there are four second order partial derivatives of z at


*tny point (a, b).
2

Ths

by
y.
is

partial d3rivatives 3 z/g>> 3* and 3 z/dx $y are distinguished


the ordzr in which z is successively differantiated w.r. to x and

But, it will be sesn that,


given in the Appendix.

in

general, they are

The proof

equal.

Geometrical representation or function of two variables.


and OY. Through the
perpendicular lines
plane and
point 0, we draw a line Z'OZ perpendicular to the

10

7.

OX

We take a pair of

XY

oall it Z-axis.

Any one
sense.

of the two sides of Z-axis may be assigned a positive


z, will be measured parallel to this axis.

Lengths,

The three co-ordinate axes, taken in pairs


determine three planes, viz
XY, YZ and ZX
which are taken as the co-ordinate planes.
Let z --/(#, y) be a function defined in any
,

domain lying

in the

XY plane,

To each point (x, y) of this domain, there


corresponds a value of z. Through this point,
we draw the line perpendicular to XY plane
qual in length to z, so that we arrive at another
point P denoted as (x, y, z), lying on one or the
other side of the plane according as z in positive
or negative.

Fi g 55
.

Thus to each point of the domain in the XY plane there corresponds a point P. The aggregate of the points P determines a surface
which is said to represent the function geometrically.

198

first

DIFFERENTIAL CALCULUS
10 71.
order.

Geometrical interpretation of partial derivatives of the

Let

z=f(x

We have seen

y).

...(/>

that the functional equation (/) represents a surface


We now seek the geometrical interpretation of the
geometrically
partial derivatives
.

The point P[a,

b.

f(a, b)]

corresponds to the values

dent variables
If a

#,

on the

surface*

b of the indepen-

x, y.

variable

from

starting

point,

position on the surface such that

changes its
remains constantly equal 1o b

then

it

is

clear

that the locus of the point is the curve of intersection of the surface and the plane y=b.
Fig. 56

On

this curve

x and

z vary according to the relation

*=/(*,*).
T

--

the ordinary derivative of/(x, b) w.r. to x for

is

)
3* Aa>

(C5

&)

x=a.
Hence, we see that

--

is

\OX

J(a>

the tangent

of the angle which

6)

the tangent to the curve, in which the plane y = b parallel to the


plane cuts the surface at P[a, bj(a, b)], makes with X-axis.
/

Similarly,

it

may

be seen that

ZX

)T V

is

^ oy /( a

>

the

tangent of the-

ft)

angle which the tangent to the curve of intersection of the surface and
the plane x=a makes with Y-axis.

Homogeneous Functions.

10-8.

a homogeneous function of order

n. if

Thus

in

x and y

is

a homogeneous function of order

is

equal to

n.

Ordinarily, /(x, y) is said to be


the degree of each of its terms

n.

This definition of homogeneity applies to polynomial functions


To enlarge the concept of homogeneity so as to bring even
only.
transcendental functions within its scope, we say that z is a homogeneous function of order or degree n, if it is expressible as
xf(ylx).

199

PARTIAL DIFFERENTIATION

The polynomial function

is

(/)

which can be written as

a homogeneous function of order n according to the


The functions

new

definition

also.

are homogeneous according to the second definition only.


n
degree of x sin (yjx) is n. Also

'

L
so that

it is

of degree

Here the

1+

* J

J.

10 81. Euler's theorem on Homogeneous Functions.


7/z be a homogeneous function of.x, y of order

n, then

We

have

Thus

Cor.

// z

a homogeneous function of x, y of degree n,

Differentiating
/

^"r

^T
?-!

-i,

...(i)

200

DIFFERENTIAL CALCULUS

x and y

partially with respect to

separately,

we obtain

d z

32

'" (

9*'

'

8z

Multiplying

(2), (3)

by

y respectively and adding, we obtain

x,

where we have employed


and

(1)

and assumed the equality of


Examples

I-

//
-1

^
v

2 f

-1

*
yv

^v

'

prove that
2

_9

"

3x3^

Wa

l~ y
x 2 +>' \2

/955; P.f/.)7
(D.t/.
v

have

rtAtan-1l
---- 2v

-1

dxdy

= 12.

//

PARTIAL DIFFERENTIATION

show

that
3

-i^

We

201

"-+

_|_

a"^

(D.U. 1952

P.U. 1949)

have

Similarly or

by symmetry

2
3 w

---=:

Adding, we obtain the result as given.


3.

//

prove that

=sm
.

d*

Here u

is

not a homogeneous function.

z=tan
so that z

is

ft

2u.

dy

u= -

xy

We, however,

l(ylx)

write

'

a homogeneous function of x, y of order

2.

%7

>?

...(i)

dy

But

--*-

'

Substituting in

(1),

we obtain

sec 2 w

)=2z=2 tan

d* J

3w

+y
ax
,

3w

-<

W~

2 sin w
cos

cos

M=

w,

DIFFERENTIAL CALCULUS

202

Exercises
Find the

1.

first

order partial derivatives of


e ax

(/)

(Hi)

tanr (x+y).

logU

(//)

sin by.

).

-j->>

Find the second order partial derivatives of

2.

r^

Verify that

3.

//

a*a>

when

is

(/)

(Hi)
4.

sin- 1

Xy

/;;x

y
log (y sin

x+x sin

Find the value of


1

a2

when

5.

Verify Euler's theorem for

liii)

(v)

z^sin- 1

xy

+ taxri- x
-

(iv)

z=jc

log

~.
*

2=

rj7l'
6.

If u=f(yjx),

show

that

*!"_
1.

If

(D.[/. 7950

P.U. 1954)

dz

If z=-tan

dZ

-j-^-r.

(y+ax)+(y-ax)'*

J!^
>.

z=xyf(x/y) show that


x~~^

8.

+J 8_ =0

^J!?.

=2z.

p rj
^.L/./
i

find the value of

(P.U. 1945)

1
If r=tapr (y/x), verify that

1-^-4-1^ -0,

(AC/.)

205

EXERCISES
r
10.

If

z(x+y)=x

show

V
J

*L
3y
11.

11
12.

that

lfz=3xy-ya -\-(y -2x) i

Tr
If w =log

(
\*

dx

If

( fl )

(A) If

w-sin- 1

w-sia- 1

show

,
'

^^"^

(AlhLatcd)

show

=1.

that
*

3w

y ^
9x^9^

A:

>>

-f

\
J

-^, prove that ^ ^- + 7

9Z-

dy

verify that

13.

az-

=tan

w.

(P.t/.

1954)

that

^x+->!y

/
14.

^^

3 A:

If

z=f(x

a
a*
15.

V9^
2

If w=/(cix -f

_9 z\
2

9y9^y

2/7^+^

If

9=t

9^

e~ r2/4/
i

18.

If

-=/(r)

^^^
\9y
*

v-^(^

),

V,
x
V .Jr.V

9^
17.

'

ltz = (x-{-y) + (x+y)<9(ylx), prove that

/9 2 z
16.

(P.U. 7955)

prove that

ay),

ay)-\-<f>(x

-I

.=
9y

9*

-f-2/?xy-h^

which

8r

9^

where r= VU 2 -f j/2 ), prove that

&+C;-^m
19.

If

=log

UHy-hz
f a
2

),

If

K=r m where

21 .

If

w = tan- 1

[Refer Ex.

3, p.

- y-

201]

=x2 -Fy

find

dxdy'

9zax

3>az
20.

prove that

prove that
(M.r.)

.dY^ao^^ao
/
V
9r

),

\
-f
9^ x

f
V

find the value of n

+ z 2 show that
,

will

make

(M

r<)

204

DIFFERENTIAL CALCULUS
Differentiating

(2), w.r.

to

x and

y,

we

=2
<>

2
*
sec w tan w

As

z is a

w
9*

*
-=2

w
9;>

2 sec w tan

get

/0-~
(

2
^sec w
,

Y +sec
,

}
\9^ /

8*

;^.
2
a^

homogeneous function of jc, y of order

2,

we

have, by Cor.

10*81

P. 199.

Making substitution, we obtain

Using

(3) p.

201,

we

obtain

/ ^ ]=2 tan

22.

-2 sec*

tan

sin

2u.

If

w~sin~

show

that

10 82.
introduce the
that

notation,

we

x=r cos
a<jid

we

new Notation. To
consider a particular case.
Suppose

Choice of independent variables.

new

6,

y=r

sin 6

(0

are required to find 9x/9r.

Before beginning partial differentiation, we have to ask ourselves a question.


"What are the independent variables ?" Hitherto
the notation has always been such as to suggest readily what the

independent variables are and no ambiguity was possible.

connected
In the present case we have four variables x, y, r,
relations so that any one of them may be expressed in terms

by two

PARTIAL DIFFERENTIATION
Thus, x

of two of the remaining three.


(a) r, e

or (b)

r,

may

or

205

be expressed in terms of

(c) 0, y.

In case (c), 3.x/9r has no meaning. In cases (a) and (&), where
9x/gr has a meaning there is no reason to suppose that the two values
of (9x/3r) as determined from them, where we regard
and>> conSome modification of the notation is
stants respectively, are equal.
therefore necessary to distinguish between these two values.

For the sake of distinction, these two values are respectively


denoted as

Thus

means

L
r,

0?

Jv

-i.

rax

J0,

Ldr

the partial derivative

of x w.

r.

to r

when

are the independent variables.

From x

To

cos

find

Ldrjy

From

(/),

we have

.we have

1.

If

*!
>'
fl

in

terms of

and

y.

we obtain
r2

Ex.

to express

rcos
r

=x +y
2

so that

5, >'=' sin 9, find

*!

L ar Je'

r^-i
L'w

x= \?(r

-y*).

the values of

9x
a<-

i
Jfl

r^
L'w
(P.t/.

Ex. 2. v is the volume, s the curved surface, h, the height and


radius of the circular base of a right circular cylinder, show that

Ex.
are

its

the

Explain the meanings of the partial differential co-efficients


x, y are the rectangular cartesian co-ordinates of a point and
polar co-ordinates. Prove that

3.

and ar/Jx where


r,

1938)

DIFFERENTIAL CALCULUS

206
Ex.

4.

Prove that
)

to

-&r

- ~

0*(logr)
8

--

-F cos

>

\vhere

x=rco5

0,

yr sin

0.

For the following developments,

10-9.

it

f( x y) possesses continuous partial derivative


domain of definition of the function.
>

10 91.

Theorem on Total

Differentials.

will

w.

r.

We

be assumed that
x and y in the

to

consider a

func-

tion

z=f(*, y)..(')
be
two
so
that
Sx 8y are
any
(x, y), (x-\-8x. y+8y)
points
the changes in tlie independent vnriablcs x B,ruly. Lst Sz be the
consequent change in z.

Let

We

have

From

(/)

and

z+Zz=f(x \-Sx,y+8y).
we get
(//'),

...(//)

t-ix,y)~f(x
so that

y)]

...(Hi)

we have subtracted and added

Here the change Sz has been cxpresse 1 as the sum of two


which we shall apply Lag range's mean v.ilue

differences, to eaoh of

theorem.

We regard f(x-\-8x, y) as a function of y only


supposed constant, so that by the mean valu? theorem,

x \-x being

We write

...(/v)
fy(x+&x, y+OW-f^x, ^} e a
so that e 2 depends on $x, Sy, aid because of the assumed continuity
offy (x y) tends to zero as x and Sy both tend to 0.
t

Again we regard f(x, y) as a function of x only, y feeing sup*posed constant, so that by the mean value theorem, we have

f(x+8x, y)-f(x, y)=Sxft (x \-Ojx,

We

y).

write

Ux + OJx.yl-Ux.y),**!

...(v)

s, depends
upon 8x and, because of the assumed ontinuity
to
tends
as 8x tends to 0.
off
y),
tso x (x

that
t

EXAMPLES

Prom

(i/7), (iv), (v),

we

get
,

207

y)+Byfv

(x,

----A----

---y---

in r consists of two points as marked.


the differential of z and is denoted by dz.

Thus the change Sz


these the

first is called

Of

Thus
9

dz=^~8x +

-S>>.

dx

...(v/)
v

dy

Let
z

=x

so that

dx=dz=l 8x^8x.
.

by taking z=y, we show that


8y=dy.

Similarly,

Thus

(v/)

takes the form

dz=-^dx n+ a/
ax

dy.

ay

It should be carefully noted that the differentials dx and dy of


the independent variables x and y are the actual changes 8x and &y, but
the differential dz of the dependent variable z is not the same as the
change Sz it being the principal part of the increment Sz.
;

Approximate Calculations. From above we see that the


approximate change dz in z corresponding to the small changes 8x and
&y in x, y is
Cor.

dz

8r

dx ^
+
dx
2y
which has been denoted by dz.

</>>,

Examples
in the area of an ellipse when an
per cent, is made in measuring the major and minor axes.
If a, b, A denote semi-major axis, semi-minor axis and area of
an ellipse respectively, we have the relation
1.

error of

Find the percentage error

-._
=7rV

TT

Here

^ =7m

dATrbda+Tradb.
Since

we

are given that


j

JL

208

DIFFERENTIAL CALCULI! S
-Too

dA

+ ioo ""loo

'

A = 2.

Therefore the percentage of error in

2,
The sides of an acute-angled triangle are
that the increment in
due to small increments in a,

measured. Prove
b c is given by the
y

equation

be

sin

S/l

=a(cos

CS&-f cos J3&c$a).

Supposing that the limits of error in the length of any side are
M per cent, where p is small, prove that the limits of error in A are
approximately
1

5(fjLa

jbc sin

From elementary Trigonometry, we know

A~

2 cos

Here

A is

.*.

9 sin
-2

(M. T. >

degrees.

that

be

a function of three variables,

a, b, c.

A A 2b*c-c(b*+c*-a*)
--->--------- ' db
AA dA=*~
'

~ ----------

a*)'
--------

2a

^c

b'c*

or

-2O

sin A

JA
<M=

2ab cos

.-.

Tho

be sin

A dA~

a(cos

(7

&/i/100.
is

2 c# cos

c/6+cos

B dcdd).
.

triangle

c<*/100,

acute-angled,

therefore cos
dA are

JB,

cos

are

Therefore the limits of the error

C+c

cos ff-f a)
0(6 cos
6c sin"^

a(

100

180
'

100
1

15

'

Cc cosj8a)

Bc"s5tt

or

6cos

'TOO'

sin

or

limits of the errors d&, Jc, da are

As the
positive.

222

IT

At

.--A
t/c

-,

sin A.

degrees approximately.

M
'

100

PABTIAL DIFFERENTIATION

209

Exercises
Find the percentage error in calculating the area of a rectangle when
error of 2 per cent is made in measuring its sides.
1.

an

Show

that the error in calculating the time period of a pendulum at


zero if an error of +/* per cent be made in measuring its length
gravity at the place.
2.

anyplace

and

is

In a triangle ABC, measurements are taken of the side c and angles


is calculated from these measurements.
If Ac, A/4, A/*
are the small errors in these measurements, show that the error Atf in a is
given by
3.

A,

and length a

is an acute-angled triangle with fixed base EC.


If Sb, Sc, 8A
small increments in 6, c, A and 5 respectively, the vertex A is given
small displacement S* parallel to EC, prove that

4.

ABC.

and SB are

(i)

c8b+b8c+bc

cot

/I

S/l-0.

(//)

cSB-hsin BSjt=0.

(M.T.)

a triangle whose sides are a, b, c is A. Prove that the


error corresponding to errors 8a, 8b, 8c in the sides is approximately given by
5.

The area of

2 A S A = S 8p - s8q - abcSs,
2

2s=a+b + c,

Avhere
6.

The work

distance s in time

is

2/?=0

that

-f

+c

2
,

3?

must be done

to propel a ship of displacement

for

proportional to
rZ>f/ 2

Find approximately the percentage increase of work necessary when the displacement is increased by 1%, the time diminished by 1%, and the distance
diminished by 3%.
7. The height // and the semi-vertical angle a of a cone are measured, and
from them A, the total area of the cone including the base, is calculated. If h
and a are in error by small quantities S/z and 8* respectively, find the corresponding error in the area. Show further that if a=n/6, an error of -fl per cent
0'33 degree in a.
in h will be approximately compensated by an error of
(M.T.)

10*92.

Composite

functions.

Let

-=A*.
and

y)

0")

let

so that x, y are themselves functions of a third variable t.


The functional equations, (/), (//), (///) are said to define z as a
composite function of t.

Again, let

x=^(ii

y^(u

...(h)

t>,),

v)

,..(*;)

so that x y are functions of the variables u, v. Here the functional


zas a function of w, v, which is called
equations (i), (/v), (v) define
a composite function of u and v.
t

210

DIFFERENTIAL CALCULUS
Differentiation of composite functions.

10*93.

Let
***f(x>y)*
possess continuous partial derivatives and

let

possess continuous derivatives.

Then

Jz =
dF
'Let
in x, y, z

f-fSf be

f,

dx

gz

dt

ax~

dy

gz

+ 8y~

'

any two values.

Let

consequent to the change St in

'

'

dt

be the change*

Sx, Sy, Sz

t.

We have

y+8y)-f(x, y)

+ [f(x,y+*y)-f(x, y )1
As

we apply Lagrange's mean value theorem

in

10'91, p. 207,
to the two differences on the right,

and obtain

Let 8f->0 so that ?x and 8y-*Q.


Because of the continuity of partial derivatives, we have
Urn
(

fJx+Ol

lim

Hence, in the

fi
limit, (/)

dz

becomes

_
"~

dt

Cor

Bx y+Sy)=fj(x,y)=-&--,

Sx, 8y)->(0, 0)

1.

dx

9z
'

dt

9x

8z

dy

9y

dt

'

Let
z=/(^,

possess continuous

first

y),

order partial derivatives w.r. to x, y.

Let

X=>(,
y=$(u,
possess continuous

first

,),

v),

order partial derivatives.

'

'

*' '

211

To obtain 3z/gw, we regard v as a constant so that x and y may


be supposed to be functions of u only. Then, by the above theorem,
we have
az

It

may

similarly be

a*

az

a*

ay

shown that

_az __ az
~av ~~ax

ax

a_z^

av

ay

j?y
*

av"'

Examples
1.

Find dzjdt when

z=xy 2 +x*y, x=at 2 y


,

Verify by

direct substitution.

Now

-.

Substituting these values in

(//),

10*93, p. 210,

*L = (yZ+2xy) 2at+(2xy+x*)2a

Again

~
Hence the
2.

verification.

a function ofx andy.

Prove that

if

then

We look

upon z as a composite function of u,


.-

'

8j>

"

3V

9"

v.

we

get

212

DIFFERENTIAL CALCULUS

Subtracting,

we

du ~"av

get

^a

---.

9*
3.

9>>

JfHf^yz, zx, xy)

prove that,

^c

Let

u=y

vzx, w=xy,

z,

so that

H=f(u,

We

have expressed

v,

w).

H as a composite

function of x,

>>,

2.

'

'

9*

Similarly

__
'

'

'dy

9M

'

3"

Adding, we get the result.


4.

th(it

is

a homogeneous function of x, y, z of ordern

prove

[This

is

Euler's theorem for a homogeneous

independent variables.]

H=xnf

We
,

function of three

have

|j=x/\w,

v)

where yjx-u, z 'x=v.


(

IMPLICIT FUNCTIONS

213

But
8w

=
~~ - JL

a*

;c*~

?
>

^ _

z
'

~~;c a

a*

Hence

= njt-'/(,
/V

0*

v)'

'

'

~ x* -

+z -

( v
V '9W

0V /

Again

dH^xnfdf
a^

8", ?/

+ 8v~
8.y

'

\8

?/,

for

aw

'

v >

aj/

8B

a>>

A;

-3J!..o.

ay

Similarly

3-

two

10*94.
Let /(jc, _y)
Implicit functions.
variables.
Ordinarily, we say that, since
f(x,

we can
an

be any function of
on solving the equation

y)=o

...(0

obtain y as a function of x, the equation

(/)

defines

y as

implicit function of x.

difficulty here. Without investigation


that corresponding to each value of A', the equation (/)
must determine one and only one value of y so that the equation (/)
always determines, y, as a function of x and in fact, this is not the
The investigation of the conditions under which
case, in general.
the equation (/) does define y as a function of x is not, however,
within the scope of this book.

There arises a theoretical

we cannot say

Assuming that the conditions under which the equation (/) dey as a derivable function of x are satisfied, we shall now obtain
the values ofdy/dxandd 2yldx 2 in terms of the partial derivatives 3//3X,
2
2
2
3 //ay, 3 //8*
d fld*dy, d*f/dy* of / w.r. to x and y.
fines

'

Now, f(x y) is a function of two variables x, y and y is again


a function of x so that we may regard f(x, y) as a composite function
of x. Its derivative with respect to x is
9

dx

3/
'

3x

~dx

df
+ dy~

a/

dy
'

dx

*'

ax

dy

df

+9*

'

dx

'

Also/(x, y) considered as a function of x alone,


equal to 0.

Therefore

derivative w.r. to

its

is 0.

is

identically

214

DIFFERENTIAL CALCULUS

__

,_.

/v

dfldy

Differentiating again w.r. tox, regarding 9//9X


composite functions of x, we get

'

~dx

and 3//8y as

'

dx\dxdydy*

>dy

dy

dy

a.y

2V

dx

8*A

>3
3

Hence

-__
~
dx

ftf

'

and

_._

.^__

Without making use of


follows

10'93,

we may obtain dyjdx

...

also

as

Now

f(x>y)=0Let Sx be the increment in x and Sy the consequent increment

in y, so that

f(x+Sx,

.or

or
,

fy (x,

Sx

Let 8x ->

0.

Example
Prove

/fart

i//

3flrx

+x3 =0, then

y+Sy)

2J6

EXERCISES

f(x, >>)=>>

-3ax2 +Jts =0.

Substituting these values in

on

(Hi),

p. 214,

we

get

_
dx*

27>>

- a)(3ajc2 (x

y
Thus
j

<7r, directly, differentiating the

given relation w.r. to x,

(2a2x)y*2y(2axx 2 )(2ax-x 2 )ty

as before.
Exercises
1

1.

Ifii=Jt- y ,JC=2r-35 + 4,

2.

If

3.

z=(cos y)/x and

i-

If

-,

cosv"

y=-r+85-5,
x=w 2 - v, y.=^ v find

sin*

v=

smy

If

5.

Find dy/dx in the following cases


(/)

sin (x-^)-(x-|-y)=0.

(j/0 (cos x)f -(sin >^)*=0.


(v)

(tan x)* +y

eoi

z)=0

6.

If F(x, y,

"7.

If 2=xyf(vjx)

find v
ar/
/

x=tan (2r-5 2 ), y-eot(r 2j),

4.

Ji=(x+ y)l(l-*y)

find

(//)
(i v)

x.

(P.U. 1955)

find 0z/ax, 3z/ay.


is

=sin

x* ^y*.

x =a.

and z

find

a constant, show that

216

DIFFERENTIAL CALCULUS
8.

y)=Q,

If/(jc,

<p(y,

z)=0, show that


9

?/
If

9.

JH(l-yH^(l-*

If u

and

1*

?/..

show

)=fl,

(/>./.)>

that

y?=2
dX
10.

4L=

?,.

(P.U. 1935\

_^

,j

-J

are functions of x and

xu + e~

z'

sin w,

>>

defined by

v
yv-\-e~ cos

w,

prove that

d^ = dv
If

11.

,4,

B,

(P.U. 1936)

dx

dy

are the angles of a triangle such that


sin

/1

+ sin B]

sin

C=constant,

prove that

dA^

C-lan B

tan

Ifax 2 -} 2hxy + by* -\-2gx-\-2fytc-Q, prove that

12.

......

"dx*

Show

13.

where

//*-!-

that at the point of the surface

x=yz,
r

ff~z

"" 1
"I"

X log

tfX

14.
(/)

(Hi)

15.

Find dy/dx 2

in the following cases

x *+y*=laxy.
Jc

+^ 5 =5fl 8

If AX,

(P.U.)

Jc.y.

=_

d 2x

/y
'

dy

fx

(//)

x 4 -i y*=

(iv)

jc

5
-!

/(x,

>>)=.xM y

dy

^_S^fv}*-2fa

2
.

(/

Given that

dx 2
17.

>^)=0 and/.T ^0, prove that

dx
16.

If

is

3ax>>:=0,

show

that

'

dy

a homogeneous function of the nth degree in

and if

(x, y> z)

where X, Y, Z are the first differential co-efficients of, w, with respect to x, y,


respectively, prove that

1950)

217

PARTIAL DIFFERENTIATION

APPENDIX
EQUALITY OF REPEATED DERIVATIVES
It has been seen that the two1.
Equality of fxy and fyx
are generally equal. They
derivatives
second
order
repeated
partial
ate not, however, always equal as is shown below by considering twoexamples. It is easy to see & priori also whyfyx (a, b) may bo diffe-

A.

rent

from/^

We

(a, b).

have

-Iim
A ->

and

t,

/y (0, &)=

i-

Inn
k -*

/K
yv

f(a,b)

b-\-k)'
--

Jim,
li

>0

A-

r
Iim
/7

It

_>

similarly bo

may

r
Inn
A ->

<

say.

/7/c

shown that

->

Iim
->

//fc

/z

Thus wo sec that/ (a, ft) and/^ffl, /;) are repeated limits of the
same expression taken in different orders. Also the two repeated
limits may not be equal, as, for example
rj!/

r
lirn

v
Jim

fc
.

hk

and
A

Iim
Iim -.----_> o/j-^0 n

+K

=
=

v
Iim

''

A->0

ll

1,

_k
Iim
A:->0

7
^

Examples
1

x,

Prove that fxy ^ fyx at the origin for the funct ion

y are not simultaneously zero andf(Q, 0)=0. (D.U. Hons. 1954)

We

have/_(0, 0)= Iim

A->0

'--

^
.

'-J v^

L
*

-'

...(1)

218

DIFFERENTIAL CALCULUS
Also

/,(0,

and

/,(

Thus from

Again

0)=

lim

lim

MH

/(0.

0+k)-fM

--

... (2 )

= m
li

=0..

lim

and

(1), (2)

(3)

/((>,

OH

lim

/,,(0 (

0)-

lim

fl( 0, 0)=

lim

t- =

l.

...,4)

Also

**
and

/fin
x (0,

From

(4), (5)

to

/c)=

and

/.>,
Thus

/JO,

==

"

h^-0

OH *->o
I

"

=0,

.(5)

lim
lim

(6),

lim

A->0

*=
K

0)=!^- 1 =/^0, 0),

2.

2
1 -/(x, .y)=* tan- x

and

We

is

have

zero elsewhere.

2
>^

tan- 1

^/
(B.U. 1953)

219

PARTIAL DIFFERENTIATION

= hm
..

r
\

fc-oL

=A

/ta,n- l klh\
'

h 1

k tan- 1

,,

klh

0=/J, for [tan-

/c

*/*]

->

as

J
-* 0.

A:->0

We may

similarly

Hence the

show that

result.

A. 2. Equality of fxy and f^. Theorem. Ifz=f(x, y) possesses


continuous second order partial derivatives d^ld^dy and d*?ldydx, then

Consider the expression

For the sake of brevity, we write

t(x)=f(x,y+k)-f(x,y),

...(1)

so that
<?(h,

By

Lagrange's

k)=$(x+h)-t(x).

...(2)

mean value theorem,

Also

...(5)

ji, y)].

Again applying the mean value theorem to the right side of

(5),

we obtain
)-

0<0 t <l.

Thus

Again considering
proceeding as before,

/r(^)=/(A:+/j, y)

we may prove that

f(x, y) instead of

</t(x)

and

DIFFERENTIAL CALCULUS

220

Let h and k -> 0. Then, because of the assumed continuity


of the Partial Derivatives, we obtain

fvx (x,y)=fz V (x,y).


A.

3.

Taylor's theorem for a function of two variables.

V /(*

y) possesses continuous partial derivatives of the nth


any neighbourhood of a point (a, V) and if (a+h, b+k) be
any point of this neighbourhood, then there exists a positive number 9
which is less than J, such that

order

in

f(a+h, b+k)=f(a,

b)

+k A/(<7,

b)

Lemma.

We

write

-=/(*, y)

and
so that z

is

a function of

t.

Now, we have, by

10'93, p. 210,

dz

8z

dx_

di'^'Wx

~dt

dz

Now, we agree

dz

dy
'

"di

dy

dz

form of the operator


hL

operating on the operand

+k
,

ax

z.

This equality shows that the operators


d
are equivalent.

8z

to write

dz
in the

dz

TWO VARIABLES

TAYLOR'S THEOREM FOR

221

Employing the equivalence of these two operators, we obtain


d / dz \
d*z__
dt^ dt \dt )
d
8
d \
'

8y)
8

lx +k
,

If,

now, we agree to write

we have
d*z

Continuing in this manner, we see that


dn z (
8
,
8 \w
*

'

where the operator

implies the repeated application of the operator

/i

Thus we have arrived at the following


Jfz=f(x, y) and x=a+ht, y = b-\-kt, where

times.

result
a, b,

//,

k, are cons-

tants, then

\n

'

By)
Proof of Taylor's Theorem.

We

write

f(x,y)=f(a+ht b+kt)=g(t)<
and apply Maclaurin's theorem to the function
9

variable

of the single

g(t)

t.

There exists a positive number 6 between

For

= !,

this

becomes

and

(0

<

such that

<

1)

.(/)

DIFFERENTIAL CALCULUS

222

Now
)=/te+A, b+k).
f(0)-/te

b).

Also since

Substituting these values in


as stated.

(i),

we have the Taylor's theorem's

Exercises
1.

Show
>

2.

that

(*

By mathematical

indication or otherwise,

3 7
/ h. 9 +
+/t _3_\" -A" V
a>-/
3^
3^"

show

that

"

8 z^^ A-^ a^-^r


+
l

......

on,
n K
-"
4-"c.h
f..t
C,n -rkr
n-r
4-

3.

Show
(i)

that

sin

sin

^^ary

T[(**+ 3x^

2
)

cos 0a sin
sin

(W)

e*

sin

ex cos

0,y]

<

< K

6^A>+afrx,y
2
u
+|. e [(a*x* -3ab*xy ) sin

where u~a9x, v~b$y.

v-|- (3fl

^ ^-6^) cos
a

],

MAXIMA AND MINIMA FOB TWO VARIABLES

MAXIMA AND MINIMA


A.

Maxima and Minima of a function of two variables.


Maximum value. f(a, b) is a maximum value of the function

4.

Def.

fix, y)> if there exists some neighbourhood of the point (a, b) such
for every point (a-\>h, 6+fc) of this neighbourhood,

Minimum

value,

f(a,b)>f(a+h,b+k).
/(a, b) is minimum value of the function

that

f(x, y) r

some neighbourhood of the point (a, b) such that for


every point (a+A, b+ k) of this neighbourhood
f(a, b)<f(a+h, b+k).
Extreme valuej /(a, b) is said to be an extreme value of f(x, y),
if there

if

it is

exists

maximum

A. 41.

or a

minimum

value.

The necessary conditions for f(a, b) to be an extreme

value offix, y) are that

fx (a b)=Q,fv (a,b)=0.
9

If f(a b) is an extreme value of the function /(;c, y) of two


variables x and y then, clearly, it is also an extreme value of the
and as such its derivative
function/(jc, b) of one variable x for
t

x=a

x (a, b) for

a must necessarily be zero.

Similarly

we may show

that
/,(*,

Note

As

1.

*)=0.

in the case of single variable, the conditions obtained above


not sufficient. For example, if f(x,y)Q when jc=0 or

are necessary and


and/(x, v) = l elsewhere, then

but/(0, 0)

is

Note

2.

jc=<i,

y=b

/*(0,0)-0,/V(0,0)=:0,
not an extreme value.

Stationary value. A function f(x, y) is said to be stationary for


or f(a, b) is said to be a stationary value of /(jc, y) if

Thus every extreme value


be

y=0

is

a stationary value but the converse may not

true.

A. 42.

Sufficient conditions.

To show

that

/.(M)=O //a,&HO,
f

and
f*(a.

b)=AJxy (a,

b)=B,tf(a,

is

a max. value

ifACB*>0 and A<0,

a min. value

ifAC-B*>0

b)=C

then
(i)
(ii)
(iii)

f(a b)
t

f(a.

b) is

f(a> b) is not

an extreme value if

AC-B*<O
(iv) the case

It

may

is

and A>0,

doubtful and needs further

be noticed that

A^O

if

consideration, if

224

DIFFERENTIAL CALCULUS
Taylor's Theorem with remainder after three terms, we.

By
obtain

fta+h. b+k)=f(a,

or

&)+(

~ +k

f(a+h, b+k)-f(a, b)=hfx


(, b)+2hkfacv(a,

(a,

b)+kf,

b)+k%*

(a, b)

(a, b)]

where u=a+0h. o=b+6k.

Now
fx (a,
Also,

where

p is

We

6)=0,/,(a, 6)=0.

we have written

of the third degree in h and

fc.

assume that for sufficiently small values

of

h,

k,

the

sign of

is

the same as that of

Case

1.

Let

ACB >0.
2

In this case neither

nor

C can

be zero.

We

write

=-r
Since

is

ACB*

is

positive,

we

[(Ah+Bk)*+(AC-B*)k*].

see that

always positive except when

Ah+Bk=Q,
i.e.,

when

/*=0, fc=0,

Thus we
which

is

when

it is

fc=0.
zero.

see that Ah?-\-2Bhk-\-Ck 2 always retains the

same sign

that of A.

Thus f(a>b)

maximum

or

is

minimum

an extreme value
according as

in tbja case
a^J ,will be
or
negative
positive.
t

is

EXAMPLES
Case

225

II.

Let

AC-B*<0.

Firstly,

We

we suppose that A^O.

write

ACB

2
Since
is negative, \ve see that this expression lakes
values with different signs when fc=0 and when Ah-\-Bk= Q;

up

Thus in this case/(#, b) is not an extreme


The proof is similar when Cr^O,
In case

A=Q

as well as

value.

C=0, we have

so that the expression does assume values with different signs


accordingly f(a, b) is not an extreme value.

and

Case HI.
Let

AC-B*=0.

Suppose that

We

A^Q.

have

Here the expression becomes

zero,

when

so that the nature of the sign of

f(a+h,b+k)-f( a,b)
depends upon
If,

have

now,

the consideration of

A = Q then,

p.

The

case

is,

because of the condition

therefore, doubtful.

AC=B

2
,

we must

5=0.

so that the expression


case is again doubtful.

is

zero

when fc=0 whatever h may

Examples
1.

Find the extreme values of

We

write

xy(axy).
/(*, y)

We now

=xy(a-x-y)=axy-x*y-.\y*.

solve the equations fx

~Q and/y=0.

be!

The

DIFFERENTIAL CALCULUS

226

Thus we have
2xy
x2

ay

ax

=0,

.y

These are equivalent to


so that

we have to consider the four

pairs of equations, viz.,

Solving these, we obtain the following pairs of values of x and


the function stationary

y which make

J"

(0, 0), (0, a), (a, 0), (J*. Jfl).

(0, 0),

^1=0,

5= a, C=0
is

Thus/(0, 0)

For

(0, a),

so that AC B* is negative.
not an extreme value of /(*, y).

A2a, B=

Thus/\0,

is

^r)

We may

a,

C=0 so that AC U2 is

negative.

also not

an extreme value of /(x,

>>).

show that

similarly

f(a, 0)

is

also not

an extreme

value of the function.

For

(\a, la),

A\<t, B-

\a,

C=

\a so that

AC

J3* is

positive.

Thus /(|0, i#) is an extreme value and will be a maximum or


a minimum according as, -4, is negative or positive, i.e., according as,
a, is positive or negative.

The extreme value /(^a, Ja)=^T^ 3


2.

JVmf

f/ie

2(x-^)

We

write

/(x,

fy*(x,

We now

extreme value of
a

-x -^4
4

(/>.[/.

Hons.)

j;)

y)=4-

solve the equations fx -=zQ

fy=Q which
y

are
...(1)
...(2)

Adding these, we obtain

-4(x3 +}")=()
or

(x+y)(x*- xy+y*) =0,

EXAMPLES

227.

which shows that

x -\-y~Q

either

or

We

xy+y*=0.

have, thus, to consider the two pairs of equations,

--

and

viz.,

>

The equations

(3)

give the following pairs of solutions

(0, 0), (v/2,

The equations

(4)

-V2), (-V2, V2).

give only

as the real solution.

(0, 0)

For (0,0),

A =4,
and

5= -4, C=4 so that ACB*^Q

accordingly this case needs further examination.

For(V2,

X=

-V 2
20,

.. /(\/2,

value

as,

A,

5= -4, C=-20 so
is

\/2)
negative.

is

We may

that

4C B

an extreme value and

similarly see that/(

is,

is positive.

in fact, a

maximum

also a

maximum

as well as ,y=0 can

be disposed

y2,

\/2)

is

value.
Note.

The

case which arises

of by an elementary consideration as

when x=0,
follows:

Now/(0,0)=0.
For points, (x

0) along x-axis,

where .y=0, the value of the function

=2x-;c=;t'(2--;c

which

is

which

is

positive for points in the

8
)

neighbourhood of the

origin.

for points along the line, y=*x the value of the function
negative.

Again

2x 4

Thus in every neighbourhood of the point (0, 0) there are points where
function assumes positive values i.e., >/(o, 0) and there are points where the
function assumes negative values i.e., </(0, 0).

Hence /(O,

0) is not

an extreme value.

3.

Find the minimum value of

We

write

when

so that we have to find the minimum


variables x, y z which are connected
t

We

value of a function of three*

by a

single relation, viz.,

re-write this relation in the form

DIFFERENTIAL CALCULUS

228

so that z has been expressed

We now

a function of x and y.

as

obtain

where,
ables

w,

has been expressed as a function of two

x and

We

independent vari-

y.

have
(*>

;>)=?*-

(p- a * - hy)

"*

>

<>b

u(x.

y)=2y-~c t[(p- ax-by).

Equating to zero these two

order

first

partial derivatives,

we

obt&jn

Again, we have
2

2^2

2
C'

2ab

a
-

Since

minimum

4C

jB

positive and,

is

for the values of

x and y

A is also positive, therefore, w, is


in question.
The minimum value
}

of, w, therefore, is

Exercises
1.

Examine the following functions


(/) ^+4x^+3**+ x.
//)

for extreme values:


(11 )

x*+xy+y*+ax+by.

(vi)

4).
(v/i)

(ix)

3x 2 ~^ 2 4-x a
2 sin (jc+2y)+3 cos (2x-y).
.

(iv)

(vi/i)

2x*y+x*~y*+2y.
z
-x-y).

(x) x*y (\

(B.U. 1952)
2.

Find

all

the stationary points of the function

examining whether they are maxima or minima.

(D.U. Hens. 1952)

STATIONARY VALUES
3.

A.

229

Find the shortest distance between the lines

Stationary values under subsidiary conditions.

5.

To find

the

stationary values of

u=f(x,y, z,w).
where the four variables x, y,

r,

...(i)

are subjected to

the

two subsidiary

conditions
...(2)
...(3)

ing

Now, we can look upon the equations


any two of the four variables x, y, z, w

We may

two.

(2)

and

(3)

as determin-

in terms of the remaining


suppose, for the sake of definiteness, that (2) and (3)

determine z and w as functions of x, y. Thus u is a function of x,


V, z, w where z and w are functions of A: and y so that u is essentially
a function of two independent variables x and y. Therefore for
stationary values of, w, the two partial derivatives of, w, with respect
to x and y obtained after z and w have bo3n replaced by their values
in terms of x and y, are respective^ zero.
Equating to zero the partial derivatives of, w, with respect to
>

x and

y\

we obtain

.-0.

.-(4)

-0.

...(5)

+/.
Again, differentiating
v,

(2)

and

(3) partially

with respect to

and

.v

we obtain

"-=0.

...(7)

-<>.

...(8)

^-0.

'...(9)

9z
--

If, now, 3Z/3X, 3>v/gx, 9z/3j^, 3w/9>> be eliminated out of the six
equations (4) (9), we shall obtain two eliminants, say,

F,(x y z w)=0,
9

...(10)

...(11)

^t(^, y, z, >v)=0.

Then the four equations (2), (3), (10) and


values of the unknowns x, y, z and w for which u

(11) determine the


is

stationary.

230

DIFFERENTIAL CALCULUS

Note. This method for determining stationary "values of a function


under subsidiary conditions outlined above is unsymmetrical in character
regard to its treatment of the variables involved. This defect has been remedied
by Lagrange who has given a symmetrical method based on the introduction of

certain undetermined multipliers.

This method

is

explained in the following

section.

Lagrange's method of undetermined multipliers. We


the
multiply
equations (6) and (8) of the preceding section by Aj and
A 2 respectively and add to (4), so that we obtain

A. 51.

*)

Again,
section

by

Xl

we multiply the equations (7) and


and A, and add to (5), so that we

(9)

-1-AalM

We, now, suppose

~=0.

...(12)

ox

of the preceding

~ =0.

...(13)

make

that A t and A 2 are determined so as to

...(14)

...(15)

With

this choice of the values of A x

and

A2

the equations (12)

and

(13) give
...(16)

The equations (2) and (3) of the preceding section along with
the equations (14), (15), (16), (17) determine values of A x A 2 and of
x, y, z and w, which render u stationary.
,

If,

now, we write

we

see that the equations (14) to (17) are obtained by equating to zero
the four partial derivatives of g, with respect to x, y> z and w so that
In practice,
all the four variables are being treated on a uniform basis.

therefore, the necessary equations are to be put


the auxiliary function g.

down by

setting

up

Note. The method outlined above is applicable to a function of any


number of variables which are subjected to any set of subsidiary conditions.

Examples

Find the lengths of the axes of the section of the


*+z*lc*=l, by the plane lx+my+nz^Q.
Let (x, y> z) be any point of the section. We write
1.

ellipsoid

EXAMPLES

231

We

have to find the stationary values of


subjected to the two subsidiary conditions
jfrVa'

We
g(x, y,

x,

>>

2
,

where

/x=0.

1,

z are

x, y,

...

()

write

zM

Equating to zero the


and z we obtain

partial

derivatives of g(x, y, z) w.r. to

The equations
AI, A lf x, y,

z.

and

(/v)

(11), (///),

These values of x,

y, z,
r2 .

determine the stationary values of


eliminating

^^

x,

y and

from

(v)

will

when

determine the values of


substituted in

(/),

will

These operations amount to

(/), (11), (///), (iv)

and

(v).

that,

Wo multiply (///), (iv) and (v) by x, y, z respectively and add so


on making use of (/) and (//), we obtain

With

this value of 7^,

the equations

(//'/),

(iv)

and

(v)

can be re-

written as

2x

r^~
!-_-'.
1
c2

Multiplying these by
/

/,

2z
A"

Ai

m, n respectively and adding, we obtain

a quadratic in
8
stationary values of r

which

is

r*

and (Jetermines, as

its

roots,

the two

232

DIFFERENTIAL CALCULUS

The geometrical considerations, now show that these two


2
stationary values of r are the squares of the semi-axes of the section,
in question.
,

2.

If

where

x
show

'

that the stationary value ofu,

is

given by

We write
g(x, y,

z)=flx'+&V+c'2 a +>

( I

Equating to zero the partial derivatives of


x,

y and

z,

-1
)-

g(x, y, z)

r,

to

we obtain

y-

=0,

2A

=0.

These give
or

ax=by~cz.
The equation
=1, determine

(/)

x,

.,,

(/)

along with the given subsidiary condition

y and

z.

Exercises
2

1.

Find the minimum value of

2.

xyz=a*.
Find the extreme value of xy when

3.

Find the greatest value of axby when

,v

-h^

fz 2 when

(w)
(MI)

(D.U. Hons. 1953)

x*+xy+y*^3k*.
Find the perpendicular distance of the point

4.

(a> b, c)

from the plane

b^the Lagrange's method of undetermined multipliers.


Which point of the sphere 2x*=l is at the maximum distance from
5.
tbepohU{2, 1,3)?
6.

Find the lengths of the axes of the conic


z
ax*+2hxy+by ~l.

7.

In a plane triangle, find the maximum value of


ens A COS B COS C.
'

8.

Find

the?

extttme Va lues of

MISCELLANEOUS EXERCISES

233

subject to

Show that the maximum and minimum values of, r 2 where


2
r =a*x*+b 2y 2 + c 2 z 2 jc'+yH z ? =l and /x-f iny+wz=0

9.

are given by the equation


-.2

72

>7_^ + i,*,v + f .l>.-0.


10.

show
will

that

(D.C/./9J5)

two variables x and y are connected by the relation


ax*+by*=ab,
the maximum and the minimum values of the function
If

be the values of

given by the equation

4(0-a)(0 -b)~ah.

(D.I/,

//ww. 7957)

Miscellaneous Exercises
Find the points of continuity and discontinuity of the following

1.

functions

2.

Determine the points of continuity and discontinuity of the function

f(x) defined by
fin

where n

is

3.

l-/)-o, if r = i
(/-I, ifi</<l,
(B.U. 1)52)

any integer.

Determine the points of discontinuity of


tan

4.

and

Draw

f"

i
L

the graph of

find the points of discontinuity.


5.

Determine the points of discontinuity of


(0 [*] + [-*?.
/ 2
lim
(///)
(
n
/I

6.

Show

~>

oo

(')

tan^/

that the function fPU)

to J when * = ; to
three points of discontinuity which

whenO<x<J
7.

which

is

equal to

when 4<x<l
%xare
to
you

required

when x=-0

and to

find.

to

whenx--=l. has
(Patna)

Examine the continuity of the function

/(0)=0.

(D.U. Hons. 1947)

234

DIFFERENTIAL CALCULUS

d0W4

Find

8.
a

where

Care

,4, J0,

relation

S 5/it

J/H

where,

/r,

C sin A+sin A

sin

B=h,

constant.

is

9.

C-f sin

the angles of a triangle and satisfy the

Find

V + ^0'(~r
\

in terms of

r,

/)'

when
r a ~a*cos 20.

10.

Itax*+2hxy+by***l, prove that

x
11.

Find
W

^ V i+
12.

show

If

cos fl+a

j,

that

*
13.

If ^

when iw=6,

If >>=cos

(w sin" 1 ^:), prove that

can be expanded in a series of ascending powers of

.v,

prove that
(M.T.)

14.

If

prove that

Assuming

that

,y

flo

can be expanded as a series


2
+ QiX+ a z x -\- ----

prove that

00=0, #!=! and for


2-4 6.... (2m)
-

=(^iivm
/

^ cosa

if e
cos (jc sin a) can be expanded
that the co-efficient of x n is cos wa/w !

15.

show

16.

(a)

Give the

first

(b)

powers of or,

(B.U.)
three non-vanishing terms in the expansion of
sin"" 1

Show

in ascending

J sin x).

that the expansion upto

x4 of x/sinh x

is

MISCELLANEOUS EXERCISES
Use Taylor's theorem

17.

g=
is

satisfying
prove that the only f unction /(x)

to

-/(*) for

/(O)-

all x,

/'(0)= 0,

given by

/W=1 ~^!+4T ......

adinf

'

(D.U. Hons. 1952)

Prove that whatever the functions /and

18.

z**xf(x+y)+yg(x

(I)

g,

\ry)

satisfies the relation

z=xf(ylx)+g(ylx),

(//)

satisfies the relation

Given that

19.

a function of u and

z is

= jc

t?,

/-2xy,

while

= v,

prove that the equation

<*>) g+f*-y)j;-ois

equivalent to 9z/av^0.
20.

21.

If M

If

= (l

ii^sin' 1

,2

22.

t2
2xy F3 r^, prove that

J_7w
;

then

tv

4cos 2 w

Prove that

'

if

dxdy

is

dy*

homogeneous function of the wth degree

in x, y,

then

x .*?+ y

9w

3x+

ay

+,-

. nM

8?l

a^

Prove also that


/*

^2

^2

^2

(bF+a7

9* w
9 "
9 2u
i /
"+ 3 J*' ( 3x + 3? + 'a?-

^)(*>-

where

13.

show

that
3'/(r)

9 V(r)

3 V(r)

3?

236

DIFFERENTIAL CALCULUS
24

^^ =

If

9*

dV

Ir

97

'

has a solution of the form

Ae~ a
prove

sin (\vt~bx).

that

H=Io

25.

(x

x*y-xy*), prove that

+y*

'

Show

26.

mean
1

to

that

value theorem
as,

/*,

27.

increases

=0

if

is

and /U)=log

(l

+ x)then

a continuous function of,

from

Show that
(/)
*<log

[1 /(I

to oo

A,

'0'

of the Lagrange's

which decreases steadily from

-*)!<*/( I

-.Y)

where

0<x<l.

,where-l<x<0.
28.

slight error

x is

made

measuring the semi-vertical angle of a

in

cone which circumscribes a sphere of radius R. Find the approximate error in


the calculated volume of the cone and show that, for a given Sac, the least value
of the error is
647TR3

V5 -* a

75-

29.

'

If the three sides a, b, c of a triangle are

angle A, due to given small error in the sides,


A
c/A=
,

da
--cot C

sin
r~ir-; ~.
sin B
sin

measured, the error

in

the

is
~,

db

cot

dc
c

-.

when
(i)

31.

a=3, 6=~5,

Findlim

=4

(ii)

a=3, 6=~4, c=2

~~-

32.

33.
</)

34.

Obtain
lim

Find

.7.,

lim tan(sin^-sin(tan^)
z
sin x "" x cos ^^ sm x

x >o
35.

.(//)lim

(///)

(D.U. Hans. 1955)

Exarr i le where the following functions are continuous for *=0.

MISCELLANEOUS EXERCISES

237

cot2 x
(/) f(x) = (cos *)

36.

Having

(*?*0),/(0)-

sin v

n
+jc +BJC 3 )/x
(tanx-A:e
non-zero limit as x tends to zero, prove that n must be equal
'

maintains a
i

finite

37.

38.

from

-3;c-f 2)/(x

+ 3x+ 2).

Sketch the curve

*0 to

sin

indicating correctly the positions of


any, of the function.
39.

Indian Pol ice 1932)

Find the maximum values of the function

if

1/>U?.

given that the fraction

A:

==2rr,

Show

maxima and minima

that the function

4(jc-l)-

has a single maximum and a single minimum value and that the function does
not lie between its maximum and minimum. Draw the graph of the function.
40.
Find the maximum and the minimum values of (1 x)"e x
.

Show
to

and that

that e x
it

*= + <.

(1

steadily decreases as x increases from


only one minimum for values of x between x=l

+ x)j(l~x)

has one and

Prove that |(35 sin 4 * 40 sin 2*-}


and has also | as a maximum value.

41.

and
.J.

42.
Show that #(*)
the interval
<jc<n/2.

43.

sin

x tan

A:

(P. U. 1938)
in
value
between
ranges
8)
unity

log sec

is

positive and increasing in

(M.T.)

Find the maxima and the minima

of,

2 cos

x+

>', where
1 and (adbc)y*Q.

(B.U.)

x*axy* oPy = Q prove that y is maximum where 3xy+ 4o =0 and


minimum where * = 0.
(B.U.)
45.
Draw the curve r=a(2 cos 04-cos 39).
Show that the extreme
44.

=i

oo

and

If

values of the radius vector are 3a and

/3>/3.

The sum of the surfaces of a cube and a sphere is given.


when the sum of their volumes is least, the diameter of the sphere
46.

Show
is

that

equal to

the edge of the cube.


47.

surface

Given the volume of a

is least

right cone

required

its

dimensions when the

possible.

48. Prove that the volume of aright circular cylinder of greatest volume
which can be inscribed in a sphere, is \3/3 times that of the sphere.

49. An ellipse is inscribed in an isosceles triangle of height h and base


2k and having one axis lying along the perpendicular from the vertex of the
triangle to the base.

Show

that the

maximum

area of the ellipse

is

>|3nM:/9.

sector has to be cut from a circular sheet of metal so that the


remainder can be formed into a conical shaped vessel of maximum capacity.
Find the angle of the sector.
(M.U.)
50.

51. Find the greatest rectangle which can be described so as to have two
of its corners on the latus rectum and the other two on the portion of the curve
cut off by the latus rectum of the parabola.
52.

Find the greatest and

least values of (sin x)

sm *

(B.U.)

CHAPTER XI
SOME IMPORTANT CURVES
The following chapters will be devoted to a discussion of
11-1.
such types of properties of curves as are best studied with the help
of Differential Calculus. It will, therefore, be useful if we acquaint
ourselves at this stage with some of the important curves which will
frequently occur.
11-2. Explicit Cartesian equations of Curves.
A few curves
whose equations are of this form have already been traced in Chapter
We now trace another very important curve
II.
,

which

known

is

as Catenary.

The following
trace

it

particulars about the

curve will enable us to

Since

(i)

cosh

on changing x to

x,

=s

we

"o-l

see that

y=c cosh

=c cosh (.
\

\
c J'

so that the two values of x which are equal in magnitude but


opposite in sign give rise to the same value of y.

Hence
(ii)
(111)

Thus, y

is

the curve

When x=0,
~j- =sinh

which

is

positive

monotonically increasing in

when x

[0, oo

is

positive.

).

=0, i.e., the slope of the tangent


-Jx \j
a
so that the tangent is parallel to x-axis at A (0, c).
Also,

x=0

symmetrical about y-axis.


y=c so that A (0, c) is a point on the curve.

is

"

is,

0, for

-^^

Hence if the point P(x, y) on the


curve starts moving from the
position
A (0, c) such that its abscissa increases,
then its ordinate also increases. Also
since y -> GO when x -^ oo the curve is
not closed.
,

The curve being symmetrical about


>>.axis, we have its shape as shown in the
adjoining figure.
Fig. 57

238

SOME IMPORTANT CURVES

239

Note. In books on Statics it is shown that the curve in which a uniformally heavy fine chain hangs freely under gravity is a Catenary.

113.

We

Parametric Cartesian Equations of Curves.


two functions of t defined for some interval.

Let/(/), F(t) be

write

To each value of /, there corresponds a pair of numbers x, y as


determined from the equations (/), (i7). To this pair of numbers x, y
there corresponds a point in a plane on which a pair of rectangular
co-ordinate axes has been marked. Thus the two equations associate
The two equations (i) and
to each value of, t, a point in the plane.
(ii) then constitute the parametric equations of the curve determined
by the points (x, y) which arise for different values of the parameter t.
The point P on the curve corresponding to any particular value,
of the parameter is denoted as the point P (t) or simply 'f.
We assume that the reader is familiar with the standard parametric equations of a parabola and the Ellipse so that we may only
restate them here. Afterwards the parametric equations of a Cycloid,
Epicycloid and Hypocyoloid will be obtained from their geometrical

f,

definitions.

11*31.

Parabola.

The parametric equations


x=at a , y=2at

represent the parabola having its axis along x-axis and the tangent
at the vertex along >>-axis, and latus rectum equal to 4a.
The point P(t) describes the part ABO of
y,
the parabola in the direction of the arrow-head
as the parameter, f, continuously increases from
oo to 0.

For the vertex 0, f=0. Again, the point


of the parabola in the
describes the part
direction of the arrow-head as the parameter t
to oo
continuously increases from

OCD

11-32.

The

Ellipse.

The parametric equations

x=a

cos 0,

y=b

Fig. 58

sin

represent the ellipse whose axes lies


along x-axis and y-axis and are of
lengths 2#, 26, respectively.

The point P(0) describes the


parts AB, BA', A'B, B'A of the ellipse
in the direction of the arrow-head as
the parameter, 6, continuously increases in the intervals (0,
TT), (I?r, TT)
(TT,

59.

ITT), (ftf, 2ir) respectively.

DIFFERENTIAL CALCULUS

240

The Cycloid is the curve traced out by a point


11*33.
Cycloid
the circumference of a circle as it rolls without sliding along

marked on

a fixed straight

line.

Let the rolling circle start from the position in which the
line X'X.
generating point P coincides with some point O of the fixed

as origin and
the fixed line as *-axis. When the

Take the point

has rolled on to the


position shown in the figure, the
generating point has moved from O

rolling

Fig.

circle

-*X to P(x, y) so that

O/=arc PL

60

Let, 0, be the angle between


Let, a be the radius of the circle.
and C/so that it is the angle through which the radius drawn to
the general point has rotated while the circle rolls from the initial
have
Thus arc PI=a0.
to its present position.

CP

We

x=OL=OI-LI--=OI-PM=aO-a

sin

6)

= <j(l-cos 0).
Thus

x=a(0-sin
y=a(l cos

6))
6))

are the parametric equations of the Cycloid

being the parameter.

Note. While the circle makes one complete revolution, the point
describes one complete arch OVO of the Cycloid, so that
increases from
to
2n as the point P moves from O to O'.

The position K of the generating point which is reached after the circle
revolved through two right angles is shown as the verttx of the Cycloid.

The Cycloid evidently consists of an endless succession of exactly congruent portions each of which represents one complete revolution of the rolling
circle.

11-34.
The curve traced out by
Epicycloid and Hypo cycloid.
a point marked on the circumference of a circle as it rolls without sliding along a fixed circle, is called an Epicycloid or Hypocycloid accord-

jpg

as the rolling circle

We

is

outside or inside the fixed circle.

shall first consider the case of

an Epicycloid.

Let O be the centre and, a, the radius of the fixed circle. Let
the rolling circle start from the position in which the generating
point P coincides with some point O of the fixed circle. We take the
puint O as origin and OA as A^-a is.

SOME IMPORTANT CURVES

The generating point has moved on from


the

241

(x, y)

Fig.

61.

to

when

has rolled on to the


position shown in the figure so that
rolling circle

arc

Let
a0

J/~arc PL

Z_40C=0,
arc ^47

i.e.,

/_ICP=<f>

arc

PIb<t>

(/>=aejb.

Here, 0, is the angle through


which the line joining the centres of
the two circles rotates while the rolling
circle rolls from its initial to its
present
position.

A PP= /. OCP- /_ OPiV


T

/_

Therefore

x=OL

= OP cos + OP sin

|TT)

cos 6

b cos

cos

b cos

sin 6

b cos

sin 6

b sin (0+<)

y -LP

sin

Thus

ft

sin

(0+^
fl+fc

x=(a+b)

cos

b cos

y=(a+b)

sin

b sin

are the parametric equations of the


meter.

Epicycloid

a+b

were within the fixed


obtained by changing

btob

a+b
'

0,

being tbe para-

b) cos

0+b

if

the rolling

equation can easily be


in the equations of the Epicycloid.
Its

circle.

Thus

x=(a

0,

The tracing point would describe an Hypocycloid,


circle

Jir)

cos

= (a-b) sin 0+b sin b

0,

are the parametric equations of the Hypocycloid.

242

DIFFERENTIAL CALCULUS
In making one complete revolution the rolling

cribe

p\q

2bno

circle will

the length of the circumference of the fixed

Let the ratio a/b of the radii of the


terms so that

des-

circle.

be a rational number

circles

in its lowest

a
b

= -p

i.e.,

aq~bp

or 2na.q.~27Tb.p.

This relation shows that in making, /?, complete revolutions,


the rolling circle doscrib3S the circumference of the fixed circle q,
times and then the generating point returns to its original position.
Therefore the path consists of the repetition of the same, p, identical
portions.

In case a\b

is

its original position

irrational, the tracing point will never return to


and so the path will consist of an endless series

of exactly congruent portions.

Some

Particular cases of Epicycloid and Hypocycloid.

For a
epicycloid become

the

b,

(i)

x=2a

cos

y2a sin
Fig.

ing

The shape of

a sin 20.
will

original position after the

roll-

circle

the curve

0a cos 20,

In this case the generating point


return to
62.

equations of the

is

its

has made one complete revolution.

shown

in the adjoined figure.

Four cusped hypocycloid. If a~4b, the equations of hypobecome


cycloid
jt=| a cos + tf cos 30,
(//)

^=|

a sin

\a sin 30.

Now,
cos

30=4

cos 8

x=a

3 cos 0, sin 30

cos

0,

y=a

=3

sin

are the parametric equations of a curve


hypocycloid or astroid.
Here 0//?=4 so that the path consists
of the repetitions of four portions. The
thickly drawn curve ABA'B' gives the
path of the point.

For the points A, B, A' B' the values


are 0, Tr/2, TT, 3?r/2 respectively.
9

of

between the parametric


Eliminating
this
of
we obtain
curve,
equations

sin

04 sin

0.

0,

known

four

cusped

SOME IMPORTANT CURVES


h

is

cycloid

243

form in which the equation of a four-cusped hyposometimes given.

also the

is

Show

Ex.

becomes a

that hypocycloid

straight line for

114.

Implicit cartesian equations of curves.


function of the two variables x and y, then

a=2b.
If f(x y) be a
f

fa, y)=0.
the implicit equation of the curve determined by the
points whose
co-ordinates satisfy it.
is

many

Curves whose equations are of the form f(x, y)~Q possess


points of interest not offered by curves with explicit equations

of the form y=F(x).


In the following, we consider only rational algebraic functions
f( x y) so that the form of the equation, when arranged according
to ascending powers of x and y, is
>

The same equation may,

in a concise form, be written as

where, u k represents the general homogeneous polynomial of the


degree in x and y.
,

The degree of any term means the sum of the indices of x and
term and the degree of the curve means the highest of the

in that

degrees of each term.

Branches of a curve.

of degree w, we
an equation in,

in the rational

If,

algebraic

equation

x by any fixed value, then there will result


whose degree will bo less than or equal to n. On

replace
y,

many values of, y, as is its degree.


as abscissa there will be as many
real roots of the equation.
points on the curve as are the difTeront
As, x, goes on taking different values, each of these points will separately describo what is known as a branch of the curve.
solving, this equation will give

Thus with the given value

as

of, x.

We

now trace a few important curves. In each case we have


to examine how, y, will vary as x starting from some fixed value,
increases or decreases.
}

DIFFERENTIAL CALCULUS

244
11-41.

We

so that,

(x

+y2)x-ay 2 =0.

>

(a

0).

Cissoid of Diodes.

write the equation as

we

see, that

to a value of

x correspond two

values" of

which are equal in magnitude but opposite in sign. The two values
of y determine the two branches of the Cissoid which are symmetrically situated about x-axis.

We consider

one branch,

y=>
and the form of the other branch can be seen by symmetry.
We have

dx

The following
trace

it

=~

about the curve

particulars

will

enable us to

xj(a-x) under the radical is negative when


(/) The expression
x is negative or when x is greater than a and is positive when x lies
and a. Thus for y to be real, x must lie between
between
and a.

Hence the curve


and x=a.
(ii)

is

entirely

When x=0, >>=0

so

situatedjjsetween the lines A'=0

that the ^branch passes through the

origin.
.(ill)

x=0 or %a. The values


of the admissible values of x.

dy/dx=Q, when

the interval

[0, a]

a of x

is

outside

Thus the slope of the tangent at the origin to the branch


so that the branch touches the x-axis at the origin.
(iv)

For values of x between

and

a,

the value

positive so that the ordinate


increases.

Also as x

a,

is

of dy\dx

is

y monotonically

Hence we have]" the shape of the


curve as shown, the* two symmetrical
branches lying in the first and the fourth
quadrants.
Note. It is important to notice that origin
a point common to the two branches and the
two branches have a common tangent there.

is

Fig 64.

SOME IMPORTANT CURVES


11-42.

We

(x

+y 2 )x~a(x 2 -y 2 )=0,

245
Strophoid.

write the given equation as

i.e.,

so that we see that to a value of x there correspond two values of


v (giving rise to two branches) which are equal in magnitude but
opposite in sign. The two branches of the curve are, therefore,
symmetrically situated about X-

We have
dy

__

dx

Some

enable us to trace the curve will

particulars which

now

be obtained.
(/) The expression
x)l(a-i-x) under the radical is positive
(a
a and a. Thus the curve entirely lies
and only if .v lies between
between the lines x~ a and x = fi.

if

(ii)

points

When .x=0

(0, 0)

and

When

(a,

0)

both the values of y are


on both the branches.

or 0,
lie

so

that the

I so
that the slopes of the two tandy/dx
(///)
gents at the origin to the two branches are 4-1. Hence y=x and
x are two distinct tangents to the two brandies at the origin.
v

at

.x=0,

For both the branches dyjdx tends to


is
(a, 0) the tangent to either branch

infinity as

x ~> a so that

parallel to j'-axis.
(/v)

dyjdx0

for values of

.v

given

by
a8
i.e.,

a.x

.x^O,

x-=-a(liV 5 )/--

for

The value

0(1
belong to the interval [

-f-

\/5)/2 does not


of the ad-

a, a]

missible values of x.

Thus

for

both the branches,

for

only.
(v)

When x->-a,

v->-foo

for

one branch and

oo

for the

other.

Hence we have the shape of the curve


Note.

as shown.

Both the branches of the curve pass through the origin and have

distinct tangents there.

246

DIFFERENTIAL CALCULUS

ay2- x (x-f a)^0,

11-43.

(a

>

0).

Clearly,

are the

two branches of

this curve.

Also

dx
(/) The point (a, 0) lies on both the branches.
negative value of x corresponds a real value of y.

The value of dyjdx


(-*,

is

To no other
not real

for

0).

()

on both the branches.

lies

(0, 0)

Also, dy/dx tends to infinity for either branch


as x ~> 0.
Thus .y-axis is a tangent to the
two branches at the origin.
(Hi)

As x takes up

one value of dy/dx

is

positive values only


always positive and the

Thus the ordinate y


one branch monotonically increases and

other always negative.


for

for the other monotonically decreases,


Fig. 66.

y ->

oo

(iv)

and ->

for one branch

Also when

x ->

oo

dy

= -^
-t
rf*

~T~

When x
oo

oo

for the other.

P3V*'

y'tfL

\2

a
4-i

j
1 tends
j.

-V* J

c -x
to inanity.
*

This shows that as we proceed to infinity along the curve, the


tangent tends to become parallel to j^-axis. This is possible if and
only if at some point, the curve changes its concavity from down-

wards to upwards.
We have the shape of the curve as in Fig. 66.
Note. The peculiar nature of the point ( a, 0) on

the curve may be


This point lies on either branch, but no point in its immediate
neighbourhood lies on the curve.
carefully noted.

11-44.

x3

ay

=0,

semi-cubical para-

The

discussion of this equation which


very simple is left to the student.
Its

shape

is

shown

in

the

is

adjoined

Pig. 67.

11 45.

x 3 +y 3 ~3axy=:0. Folium of DesFig. 67.

cartes.

It will be traced in

Chapter XVIII.

247

SOME IMPORTANT CURVES


Important Note
we obtain

Putting

y=tx

in

2
the equation y (a-x)=x* of

the

Cissoid,

at 3

at*

which are

its

parametric equations

We may

similarly

show

being the parameter.

that

*~

3af
'

1+Y

__

3af

2
'

1+t 3

are the parametric equations of the Strophoid, Semi-cubical Parabola and

Folium

respectively.

This method of determining parametric equations


we have here considered.

not general and

is

applies only to such curves as

115. Polar co-ordinates. Besides the cartesian, there are other


systems also for representing points and curves analytically. Polar
system, which is one of them, will be described here.
line

In this system we start with a fixed line


fixed point on it, called the pole.

OX,

called the initial

and a

P be any given point, the distance


and /_XOP~0, the vectorial angle.

If

vector

OP=r

is

called the radius


re-

The two together are

ferred to as the polar co-ordinates of P.

11 51.

we were

If

Unrestricted variation of polar co-ordinates.

concerned with assigning polar co-ordinates to only individual points


in the plane, then it would clearly be enough to consider the radius
vector to have positive values only and the vectorial angle 9 to lie
between and 2ir. But, while considering points whose co-ordinates
satisfy a given relation between r and 6, it becomes necessary to remove this restriction and consider both r and 6 to be capable of varyx oo ). The necessary conventions for this will
ing in the interval (
be introduced now.
,

The angle, 0, will be regarded as the measure of rotation of a


which starting from OX revolves round it the measure being
positive or negative according as the rotation is counter-clock-wise
line

or clock-wise.

To find the point (r,


value, we proceed as follows

0)

where

r is

negative and,

6,

has any

OX

revolve though 0. We
Let the revolving line starting from
produce this final position of the revolving line backwards through
r
is the
O. The point P on this produced line such that OP
\

required point

The
(

1,

(r,

positions of the points

97T/4)

6).
(1, 97T/4),

(1,

9ir/4),

have been marked in the diagrams below.

(1,

248

DIFFERENTIAL CALCULUS

'X
Fig. 68.

Fig. 69.

v\

Fig. 70.

Fig. 71.

It will be seen that


according to the conventions introduced
here a point can be
For
represented in an infinite number of ways.
example, the points (-1, ir/4), (I, oir/4), (I, 2/17T 57T/4), (n is any
integer), are identical.

11 52.

Transformation of co-ordinates. Take the initial line


as the positive direction of Jf-axis and the
The positive direction of
pole O as origin for the Cartesian system
K-axis is to be such that the l\neOX after
revolving through ?r/2 in
counter-clock-wise direction comes to coincide with it.

OX of the polar system

Let

(x, y)

and

(r,

be the cartesian and polar co-ordinates respectively of any point P in the plane.

9)

From

the

OMP, we

OJf/OP=cos o,
AfP/0P = sin 0,
The equations

get

'X x=r cos


i.e.,

(/')

y=r

and

(//)

sin d

..

(i)
(//')

determine

cartesian co-ordinates (x, y) of the point


terms of its polar co-ordinates (r, 6) and
Fig. 72

..

the

in

vice

versa

6.
Polar Equations of Curves. Any explicit or implicit rebetween, r and 9 will give a curve determined by the points
whose co-ordinates satisfy that relation.

11

Thus the equations


determine curves,

The co-ordinates of two points symmetrically situated about the


or of the form (r, 9) and (r,
9) so that their vectorial

fnitial line

angles differ in sign only.

SOME IMPORTANT CURVES


Hence a curve
changing f) to
curve /=#(! +cos

249

symmetrical about the initial line if, on


equation does not change. For instance, the
symmetrical about the initial line, for,

will be
its

0,

0)

/-=tf(l

+ cos

0)

= a[l+cos(

0)].

be noted that

may
r=a represents

It

a circle with

its

centre at the pole and radius

a and
;

9=b
the

represents a lino through the pole obtained


through the angle b.

by revolving

initial line

A
curves,

To trace polar
will now be treated.
varies.
generally consider the variation in r as

few important curves

we

cos

11-61.

r=a(l
The curve

(/)
(//)
(/'//)

is

Cardioide.

0).

symmetrical about the

When 0-0, r = 0.
When increases

initial line.

from

to 7r/2, cos 9 decreases from 1 to


and, therefore, r increases contito a. When
nuously from
7r/2,
r
a.

When 9 increases from


to
cos 9 decreases from
1 and, therefore, r increases from
to 2a.
When
IT, r='2a.
(iv)

Tr/2 to

TT,

-~..^

Fig. 73

The variation of 9 from TT to 2ir ne^d not bo considered because


of symmetry about the initial line.
Hence the curve is as shown.
Ex.

Trace the curve

11*62.

^ a-

--

cos 20.

--0(1-1-

cos

0).

Lemniscate of Bernoulli!.

It is symmetrical about the initial line and so


the variation in r as 9 varies from
to TT only.
of
r
values
We consider positive
only.
(/)
(//)

When 0=0, r=a.


When increase from

so that cos
a to 0.

2'y

decreases from

to

we need

consider

to ir/2
to ir/4, 20 increases from
and, therefore, r decreases from

When = ir/4, r = 0.
When increases from
(in)

y4
'

.*''
I

^v

Tr/4

to

to 3-7T/4, cos 20 remains


7T/2 and from ir/2
Thus no
r is not real.
so
and
negative
to these
curve
the
on
corresponds
point

values of 0.

When = 3ir/4,r=0.
When increases from
(iv)
fl

NX

Pig. 74

TT,

20 increases from 3ir/2

to

2-rr

3ir/4 to
to that

250

DIFFERENTIAL CALCULUS

cos 20 increases from

Hence we

to

and, therefore,
have the curve as shown.

increases

from

to a.

It is easy to see that the point


will describe exactly the
if we take, r, to be negative.
The curve consists of two
situated between the lines

same

curve even

loops

To obtain the cartesian equation


the polar equation as
r2

=a

r4

=a

re-write

sin 2 0)

we

of the lemniscate,

(cos

or

Therefore,

which

(r

r 2 sin 2 0).

cos 2

we obtain

a well-known form of the equation of the lemniscate and

is

of the fourth degree.

11 63.

rw

=a m

(1)

r=a

or

1T^

'

/.*.,

For

(2)

is

m=

J.

cos

=0rcos

circle

(Fig.

and radius

(0/2, 0)

75

2,

x 2 +^ 2 =^x,

which
Fig.

cos m0,

m = \, 1, 2,
For w 1, we have

for

is

75)

with

its

centre at

(a/2).

we have

r^sstf" 1 cos

0),

# = r cos

or

a~-x
which

is

straight

line

(Fig. 76) to the initial line


a,

from

perpendicular
at a distance,

it.

(3)

For

w~2, we
r*=a

which

and

is

have

cos 20,

lemniscate.

Fig. 76.

(4)

/w m=
r-

or

>X

we have

=a- cos(-20)

a 2 =r 2 cos 20

=r
or

2,
2

a 2 =x

which is known to be a rectangular


hyperbola (Fig. 77). To trace this
curve, write its equation in the form
20

Fig. 77

sin*0)

(cos

SOME IMPORTANT CURVES


and note the following points about
(0
(//)

When 0=0, r=a.

creases from a to oo
(iii)
(i'v)

When
When

initial line.

When

increases from

to

7T/4, r in-

varies from Tr/4 to 37T/4, r remains imaginary.


varies from 3?r/4, to

m=

F0r

(5)

it

symmetrical about the

It is

251

TT,

decreases from

to a.

oo

we have

J,

=a cos 2

or
or

2r=a(l+cos

which

a Cardioide.
For
(6)

0),

is

|,

we have

i.e.,

=r

or

cos 2

=r(l+co"s

0)J2

= l+cos0,

or
v/hich

is

To

known
trace

Fig. 78.

to be a parabola (Pig. 78).


this curve (Fig. 78), we re- write its equation in

form

2a

r ~~

1+cos
and note the following points about
(/)

(//)

It

is

symmetrical about the

When 0=0, r~a.

decreases from 2 to
11-64!
(/)
(//)

it

r--a0,

initial line.

When
r

and, therefore,

to
increases from
increase from a to oo

TT,

1+cos

Spiral of Archimedes.
so that the curve goes through the pole.

(a>0).

When 0=0, r=0,


When increases,

r increases.

Also,

when
when

~>
->

oo

oo

oc

Thus the curve

starting fron*

the pole goes round it both ways an


infinite number of times. The continuously draw n line corresponds to
and the dotted
positive values of
r

one to negative values of


Fig. 79.

0.

252

DIFFERENTIAL CALCULUS
1165.
Here

Hyperbolical Spiral.

r0:=a,(a>0).

r=flf/0.
(i)

(iV)

r is positive or

When

6
|

negative according as

increases,

is

positive or negative.

decreases.

r
\

Also

when |0
when

0,|r|->x>.
oo

(ill)

->

0.

Now

or
r sin

or

0=(fl sin 0)/0

>>=(a sin 0)/0 which

-;>

a as

-> 0.

The ordinate of every point


on

curve

the

approaches
as drawn.

approaches

as

We thus have the curve

0.

The continuously drawn

line

corresponds to positive values of

r=aeJ>0

1166.

(ii)

Also when

(i/i)

(a,

b>0). (Equiangular Spiral)

When 0=0, r=a.


When increases,

(/)

r is

GO

also increases.

> GO

when

->

/*

->

always positive,

We thus
The

->

0,

while the dotted one corresponds to


negative value of 0.

Fig. 80.

have the curve as drawn.

justification

of the

adjective

'Equiangular' will appear in Chapter XII,

Ex,

1. p.

269.
Fig. 81

11-67.

r=asin30,

(a >$}. Three leaved rose.

(D.U. 1949)

SOME IMPORTANT CURVES

When 0=0, r=0. As

(i)

from

increases from

to 7T/2 and, therefore, r increases

from

to
to a.

253
7T/6,

30 increases

(ii)

As

increases from 7T/6 to Tr/3, r decreases from a to 0.

(///)

As

increases from 7T/3 to


to a.

remains negative and

7T/2,

numerically increases from


(iv)

As

increases from ?r/2 to

27T/3,

remains negative and

numerically decreases from a to 0.


It

57T/6

may

increases from 27T/3 to


similarly be shown that as
P
the
to
describes
the second loop
TT,
57T/6
point
(r, 0)

and from

above the
If

ted and

initial line.

increases

we do not

11-68.

r=a

beyond TT, the same loops of the curve are repeaget any new point.
sin 20.

Four leaved

The discussion of this equation


shown in figure 83.
is
only

rose.

is left

Fig. 83.

to the reader.

Its

shape

CHAPTER

XII

TANGENTS AND NORMALS


[Introduction. This and the following chapters of the book will be devoted to the applications of Differential Calculus to Geometry. The part of
Geometry thus treated is known as Differential Geometry of plane curves.]

Section I

Cartesian Co-ordinates

12

EQUATIONS OF TANGENT AND NORMAL.

1.

It was shown in Ch.


Explicit Cartesian Equations.
be
the
if
which
the
77
that
^
IV 4-15, p.
angle
tangent at any point
makes
with
then
curve
the
on
x-axis,
y=f(x)
(x, y)

12 11.

tan

*=

=/'(*).

-j

Therefore, the equation of the tangent at

curve y=f(x)

&ny point

(x, y)

on the

is

Y-y-f'()(X-x),
where X,

are

the

...(i)

current co-ordinates of

on the

any point

tangent.

The normal to the curve y=f(x) at any point (x, y) is the straight
which
line
passes through that point and is perpendicular to the
tangent to the curve at that point so that its slopa is, -!//'( jt).
Hence the equation of the normal at (x, y) to the curve y=f(x) is

(X-x)+f(x)(Y-y)=0.
Implicit Cartesian Equations.

12-12.

the curve /(*,

y)=Q

where dfldy*

0,

For any point

(x, y)

on

we have

'

dy

dx

Hence the equations of the tangent and


are
(f,y) on the curve /(x, y)=0,
rt

the

=0.n d(X-x,

normal

at

any point

_(V-n

=0.

respectively,

symmetrical form of the equation of the tangent to a


rational Algebraic curve. Let f(x, y) be a rational algebraic function
of x, y of degree n. We make f(x, y) a homogeneous function of
Cor.

254

TANGENTS AND NORMALS

255

three variables x, y, z by multiplying each of its terms with a suitable power of z. Then by Euler's theorem, ( 10'81, p. 199) we have

i.e.,

"^

'

dz

d'y~~'~

so that the equation


/

V_ Y\ df

yr__

8/

^Q

of the tangent takes the form

or

x.j
9.x
or

8f

+Y

cx

8y

+Z?8z

=0,

where, for the sake of symmetry, the co-efficient r of 8//8- has been
replaced by Z.

The symbols

and

both put equal to unity after

r are to be

differentiation.

This elegant form of the equation of the tangent to a rational


algebraic curve proves very convenient in practice.
Parametric Cartesian Equations. At any point *f of
12-13.
the curve x=f(t) y=F(t), where/' (0^0, we have
y

dy _
~ dy

dx

Hence
9

*t

dt

dt
'

dx

_F'(t)

^f

(t)

and

the equations of the tangent


y=F(t) are

'

the

normal

at

any point

of the curve *=/(/),

[X -f(t)]F'(t) - [Y- F(t)]f '(t) =0,

and
[X-f(t)]f'(t)+[Y-F(t)]F(t)=0,
respectively.

Examples
1.

(x, y)

Find the equations of the tangent and the normal at any point

of the curves
(/)

y=c

cosh (xjc).

(Catenary)

(ii)

x m la m -\-y m lb m = l.

256

DIFFERENTIAL CALCULUS
(i)

For

y~c

we have

cosh (xjc),

dy =

sinh

ax

x-

Hence the equations of the tangent and the normal


point

(x, y), i.e., (x, c

respectively
(//)

X,

any

c cosh

=(A

(x

and

at

cosh xjc) are


x) sinh

sinh
c J

cosh

^X
+

x^O,

being the current co-ordinates.

Let

/(x,

y)^*

df

~~

dx

1=0.

tt

my

m -*

Hence

Therefore, the equation of the tangent at (x, y)

or

on the given curve.


Also, the equation of the normal at (x,y)

for (x, y) lies

is

or

x-x

The given equation

Another method.
it

homogeneous, we

y-v
is

of degree m.

get

M **)%. +>-*".
jW

so that

yw

Making

TANGENTS AND NORMALS


Hence the equation of the tangent

^
Putting

Z.mz-

\-~/ti---

Z=z = l, we

257

is

^.

12*12, p. 254)

(cor.

obtain

am

bm

'

'

as the required equation of the tangent.


2.

Find the equations of the tangent and normal

at

6=n/2

to the

Cycloid

x^a(9sin

6),

y=a(lcos

9).

Wejiave
-r-

a(

a0

dy

cos

= 2a

sin 2

=a

sin

rt

2a sin

civ

dy
dx

~i

cos

dy
d9

dd

"

dx

dx

dy
;

rftf

2i

2i

rf^

__
~" C

9
'

Thus

Also, for

7T/2,

we have

jc=0(7T/;2

equation of the tangent at 0=7T/2,

-a =

i.e.,

-^Tr-

and the equation of the normal

1)

and j

a.

Hence the

at the point [0(7T/2

or

-=aTr-a

is

y-a=-l[*--<i(i7r--l)] or -Y+r=Jair.
3.

may

Prove that the equation of the normal to the Aitroid

be written

in

the form

x
Differentiating

we

get

sin

<f)y cos

(f>

+a

cos

1), a] is

DIFFERENTIAL CALCULUS

268

Therefore the slope of the normal at any point

(x,

y)

=**/A
But the

We

slope of the given line

= tan

<f>.

write
...()

Equations

They

(/)

and

have now

(11)

to be solved to find x and y.

give

or
cos 2 ^,

y$=a*
Substituting this value of y in

we

(//),

x*=flrsin

i.e.,

}>=0cos

<.

get

<f>,

x~a

i.e.,

sin 3 <.

the slope of the normal at (a sin 3 ^, a cos 3 <).


equation of the normal at the point is

Thus tan

<f>

is

a cos 3 <i=-^-^(x
T
cos

The

a sin 3 <i),
Y/

<f>

or

cos <f>acos*<f>~x sin

a sin 4 ^,

<f>

i.e.,

sin

2
y cos ^+fl(cos

<

sin 2 ^)(cos 2 ^+sin 2 ^)=0 ?

</>

i.e.,

x
which

is

4.

sin

y cos

</)

<^-f

a cos

2^=0,

the given equation.

Find

the condition for the line

x cos

0+y

sin

6=p,

...(/)

x/a+ylb m =l.

...(/I)

to touch the curve

was shown that the equation of the

In Ex.

1, (//) page 255, it


tangent at any point (x, y) of the curve

(11) is

where X, Fare the current co-ordinates.

We re-write

the equation

XGOB
taking X,

Y as

(/)

in the

0+Y

sin

form

0~p=0,

current co-ordinates instead of x,

The equations

(in)

and

(/v)

>>.

represent the same line.

TANGENTS AND NORMALS


a m cos 6

b m sin

'

or

(a* cos

~\

The point

\l/(m

>

p~

(x, y) lies
1

cos S

1)

/6 W sin 0\l/(m

y ~(

on the given curve.

\w/(w

1)

1)

Therefore

/6* sin

-r

or
'

(0 cos 0)

which

is

(&sin0)

==P

the required condition.

Show that the length of the perpendicular from the foot of the
5.
ordinate on any tangent to the Catenary

y=c
is

cosh (x/c),

constant.

Equation of the tangent at any point

sinh x/c

The

foot

F+(rcosh

xjcx

sinh

_c
is

6.

free of x,

cosh x/r __
~~
cosh x/r

y and

is,

is

1. p.

the point

is

255).
(x, 0).

0) to the tangent

x/c

x sinh x/c)

therefore, constant.

SViou' that the length

of the portion of the tangent

intercepted between the co-ordinate axes

is

Differentiating the given equation,

,''-*+
dy

(x, y)

(x,

+ (r cosh

__x sinh x/c

of the Catenary

x/c)=0 (See Ex.

of the ordinate of the point

The length of the perpendicular from

which

(x, y)

to the astroid

constant.

we

get

!'-'y*

Therefore the equation of the tangent at any point

yt

Y-y=- y r (X- X

),

(x, y)

is.

DIFFERENTIAL CALCULUS

260
or

=x*
The tangent
with

y*

where y=0. Hence

(/)cuts JV-axis

(0

fl*.

its

intersection

its

intersection

A'-axis is

A(x^
on putting ^=0 in

Similarly,

with F-axis

0).
(/),

we

find

that

is

5(0, yl($).

AB=<x$a

Therefore

which

is

x and y and

free of

is,

therefore, a constant.

Exercises
I-

Find the tangent and normal to each of the following curves


(//) x*l<P+y*lb*=l at (*', /).
20).
(/) y*=4ax at (0,
:

xy=c f

at (cp, c/p).
(iv)
2
2
<v) (x +^ )A:~0(jc ->' )=0 for jc=-3a/5.
2
?
<v/) x (jc-^)-f0 (x+>>)=0 at (0, 0).

<iii)

(v/i)

x = 20cos^
2

(x 4-y

2.

+/)*-0>>

=0

at

x-0/2.

0cos 2^,^=2a

)=^V

a sin 2^ at (9=^/2.

sin ^

at (c/cos 0, c/sin ^).

Find the equation of the tangent to the carve c*U 8 -fy t )=JcV

form x cos 3
3.

Show

in

the

= c.

3
0-f>> sin

that the tangent to the curve

curveagainat (16/5,

3xy*-2x-y=l

meets the

at (1,1)

1/20).

4.
Prove that ths equation of the tangent at any point (4m 2 8m 3 ) of the
2
3
semicubical parabola x 3 =y is
and show that it meets the curve
m 1 ), where it is a normal if 9m a =2.
(D.C/. Hons., 1957)
again at (m\
,

y~3mx4m

Find where the tangent


5.
F-axis for the following curves

is

parallel to yf-axis

and where

it is

parallel

to

(/)

6.

x*+y*=a*.

x 3 4-^ 3 = 30xy.

(Folium)

Find the equations of the tangent and the normal to the curve

at the point where


7.

(//)

Show

it

cuts JT-axis.

(Delhi 1948)

that the line

x cos O+y

sin

/?=0,

touch the curve

if

sin n ^.
(P.I/. 7957)

261

TANGENTS AND NORMALS


8.

Tangent

ordinate axes in
co-ordinates is

,4

at

and

any point of the curve Oc/a)l +(ylb$ =1 meets the coShow that the locus of the point with (OA, OB) as
2

Show

9.

li

y m =ax m

makes

x*/a +y /b
that the tangent at any point x, y)

-l

-\-x

on the curve

in
t

intercepts

ax

ay

and

\)a+mx

(m

m(a-\-x)

on the co-ordinate axes.


10.

Prove that the sum of the intercepts on the co-ordinate axes of any

tangent to
is

constant.
11.

Prove that the portion of the tangent to the curve


a

intercepted between the point of contact and A'-axes


12.

Show that the normal at any


x=acos Q+aQ sin 0,

constant.

point of the curve


sin
cos

ya

is

is

QaQ

at a constant distance from the origin.


Show that the length of the portion of the normal to the curve
13.
3
x=a(4 cos

3 cos 0),

y=a

3
(4 sin

09 sin 0)

intercepted between the co-ordinate axes is constant.


Show that the tangent and the normal at every point of the curve
14.

x=ae u

fl

(sin

are equidistant from the


15.

Show

0-cos0), y=ae

fl
u

+ cos

(sin

0)

origin.

that the distance from the origin of the normal at any point

of

the curve

is

x=ae^ (sin i0 1-2 cos


twice the distance of the tangent,
16.

Show

makes angle i

0),

y=ae^

(cos

i0- 2

sin 10)

that the tangent at any point of the curve


x = a(/-f sin t cos /), >>=c(l-f-sin /) 2 ,
(*

+ 2/)

with

A'-axis.

origin to the curve v=sin x.


Tangents are drawn from the
=x 2 -/.
their points of contact lie on

Prove that
(D.U 7955>

17.

xY

18.

If

p~ x

cos

0+y

sin 0,

touch the curve

prove that

p n = (tfCOS
(P.U. 1941

D.U. 7955)

262

DIFFERENTIAL CALCULUS

The tangent at any point on the curve x3 +y*=2a*


7tm die co-ordinate axis show that

cuts off lengths

p and

The tangent

20.

at

any point P(x v

x3

y=x*
nusefcs it

again at
chat its locus is

of the curve

y^)

Find the co-ordinates of the mid-point of

PQ

and show

y=l-9x -\-2Zx* ~28x*.

(B.U.)

Prove that the distance of the point of contact of any tangent to


2a mxy m ~~i =zy2in a 2m

21.

from the

origin as

the segment of the

-Ac-axis

between the origin and the

tangent.

Show

22.

at

Show

23.

at

tangent at the two points where

3) is

(0,

12-2.

it

meets the curve again.

that the tangent to the curve

also a normal at

1, 1) is

Def.

that the normal to the curve

two points of the curve.

Angle of intersection of two curves.

The angle of intersection of two curves


to

intersection is the angle between the tangents

the two

a point of
curves at that

at

point.

Examples
Find the angle of intersection of the parabolas

1.

/>

and
at the point other than the origin.

We

have

or

x=0

Substituting these values of

and

in (#),

we

>=4a^
Therefore
intersection.

(0, 0)

and (4a*6 f

4a*b*)

get

for

x=

are the two

points of

TANGENTS AND NORMALS


Differentiating

(i),

we get
or dy/dx=2a/y.

2ydyjdx=4a
Therefore, for the curve

(//),

we

n
^L
2x=46

(/),

4(0M, 40M) =^12$.

(dyfdx) at

Differentiating

get

^V
-~-

Therefore, for the curve

Thus,

if

/,

wi

(//),

(4a%$,

(dy/dx)Q,i

= !-/
dx 26
t/y

or

t/x

263

4a^)

=2a*/A*.

be the slopes of the tangents to the two curves,

we have

The required angle

ft^

2.

JF/w/

//?e

condition that the curves

should intersect orthogonally.

Let (x

r
,

y')

be a point of intersection so that we have

or

Differentiating the

first

equation,

we

get

dy

=_

dx](x', y')~

by"

ax

264

DIFFERENTIAL CALCULUS
Similarly for the second curve

__.

*L1

For orthogonal intersection, we have


ax'
by'

0ijc'__
~~~~

'

hy'

Substituting the values of


*-.

_r

L.

l'

'

aa ^ x

x', y' in (/),

**._..

!'

we obtain

r=0

or

/.*.,

bl

a^

as the required condition.

Exercises
1.

2.

Find the angle between


2
2
2
(/) x -/=a and x +/=a^2.
(//) ^=ax and ^-f-^^

(P.C7.

7955)

Prove that the curves

interesect at tan" 1 (88/73) at {3a, -2a).


3.

Find the condition for

y=mx

to cut at right angles the conic

ax*+2hxy+by*=l.

Hence

find the directions of the axes of the conic.

12*3.
Lengths of the tangent, normal, sub-tangent and subnormal at any point of a curve.

Let the tangent and the normal at any point (x, y) of the curve
meet the A'-axis at Tand G respectively. Draw the ordinate PM.

Then
sub- tangent

the lines

TM,

MG are

called the

and sub-normal respectively.

The lengths PT, PG are sometimes


referred to as the lengths of the tangent
and the normal respectively.
Clearly

Also

dy

TANGEENTS AND NORMALS

From the

/i

we have

Tangent ^TTWMPcosec

(i)

figure,

Sub-tangent

^=

= TM = A/P cot ^ = y

Normal =GP==MP

(/'//)

265

-=?

sec /r=

-V
dv

(iv)

Exercises
1.

Prove that the sub-norrral at any point of the parabola

y*=4ax
is

constant.
2.

Show

that the sub-normal at any point of the curve


2

y*x =a*(x*-a

2
)

varies inversely as the cube of its abscissa.


3.
Find the length of the tangent, length of the normal, sub-tangent
and sub-normal at the point 0, on the four cusped hypocycloid
x=a cos 3 0, y-=a sin 3 0.

Show

4.

that the sub-tangent at any point of the curve


^m _. Q rn+ n
yti

varies as the abscissa.


5.

Show

(Banaras}

that for the curve

x=a+b log

+ V( &'->>)] -V(6 a ->' 2 ),

sum of the sub-normal and sub-tangent


6.

Show

constant.

is

(P.C/.

1938}

that for the curve

the product of the abscissa and the sub-tangent is constant.


Show that the sub-tangent at any point of the exponential curve
7.

is

constant.
8.

For the catenary

y-c cosh U/c),


2
(P-U. 1941}
prove that the length of the normal is v /c.
Prove that the sum of the tangent and sub-tangent at any point of
9.
the curve
*

/<

= *-<!.

varies as the product of the corresponding co-ordinates.


10.

Show

that in the curve

y=a

log

-0

a
),

the sum of the tangent and the sub-tangent varies as


ordinates of the point. [Compare Ex. 9 above].

the product of the co(D.U. 1952}

268

DIFFERENTIAL CALCULUS
Pedal equations or p, r equations.

12*4.

Del A relation between the distance, r, of any point on the


wrvefrom the origin (or pole), and the length of the perpendicular* p,
from the origin (or pole) to the tangent at the point is called pedal equa*
lion of the curve.

To determine the pedal equation of a curve whose Car-

12-41.

tesian equation

is

given.

Let the equation of the curve be

Equation of the tangent at any


point

(X, y) is

Y-y=f'(x)(X-x)

y>

or
If,

cular

p be the length of the perpendi}

from

(0, 0)

to this

tangent,

we

have
Fig. 85.

y-xf'(x)
Also

r*=x*+y 2
Eliminating x

y b3twe3n

...(Hi)

(/),

required pedal equation of the curve

(il)

and (m), we obtain the

(/).

Example
Find the pedal equation of the parabola

Tangent at

(x, y) is

Y-y^-.(X-X)
or

The
tangent

length,

(/')

is

given

of the perpendicular from the origin to the

by

<

ISO

TANGENTS AND NORMALS

From
which

is

(//)

and

(///),

267

we obtain

the required pedal equation.


Exercises

1.

Show

that the pedal equation of ellipse

;
2.

Show

j 4 1 ;;

IP.*. 1955)

that the pedal equation of the astroid

x=a

cos 3 0,

y=a

sin 8 0,

is

3.

Show

that the pedal equation of the curve


A'

-a

a cos 29, ;y=2a sin

2a cos Q

sin 20,

43
2

9(/-

4.

Show

-^)=8//-.

that the pedal equation of the curve


x^--ae

+ cos

cos 0),,)'=^ (sin

(.sin

0),

is

r=
5.

Show

V2p.

that the pedal equation of the curve

x=a

(3

0-cos 3 0), y=a

cos

(3 sin

0-sin 8 0),

as
s

3/> (7fl

6.

Show

-^)

=(10fl

-r)

a
.

that the pedal equation of the curve


<*(*'

f >')=**>'.

is
s

l/P*+3/r'=J/r

Section II
Po/^fr Co-ordinates

12-5.

Angle between radius vector and tangent.

Let P(r,

d)

be any given point on the curve.

Take any other point Q(r+8r, 6+80) on the curve.

OP to L.

Produce

268

CALCULUS
Let

PT

be the tangent at

Let

and

let

^LPT=

Z_PQ=a so that

limit of a as

Q ->

P.

We

=<>

^_

is

the

have

Also

the

Applying

"\ we

sine

formula

to

the

get

OQ
OP

Fig. 86

sin
=

sin

or
sin

sin

a)

(TT

a
*

sin (a

8r

sin (a

80)

80)

sin a

"""

sin (a

8^)

sin a

sin (a

sin (a

=2

89)

cos

sin

se
2 sin

or
5r

Let

80

P so

-*

sin

~~ cpsJa-4S0)
sin (a -80)

t
'

that 80 -> 0, 8r

^
'

-->

0,

and a

Therefore
1

dr __ cos
~~

dti

(^
"

sin

<

or
.

tan cj=r
^
r
dr

Cor. Angle of intersection of two curves. If <^,


be angles
2
between the common radius vector and the tangents to the two curves
it a point of intersection, then their angle of intersection is
<

Note.

#1-*,

Precise meaning of

<p.

For a point P of the curve,

q> is precisely defined to be the angle through which the


of the radius vector (/.<?., the direction of the radius vector
rotate to coincide with the direction of the tangent in which
direction of the tangent in which, 0, increases is taken as the
of the tangent.

positive direction
produced) has to
9

increases.

The

positive direction

TANGENT3 AND NORMALS

269

Examples

Show

1.

that the radius vector

inclined, at

is

a constant angle to

the tangent at any point on the equiangular spiral

Differentiating (i) w.r. to


dr
,
hfi

or

which

<=
<i

dO

,,'

do
tan

we get

6,

dr

tan" 1

a constant.

This property of equiangular spiral justifies the


il'66, p. 252]
[Refer
Find the angle of intersection of the Cardioides

is

adjective 'Equiangular'.
2.

r=fl(l+eos
Let

P (r

r=6(l

0),

cos

6).

OJ be a point of intersection. Let <^ 1


which OP makes with the tangents to the two curves.
For the curve r=a(l +cos 0) we have
t

<

dr

--

be the angles

or

dO

__

2Ujos* 0/2
____
"~
2 sin 6]"2.GOs 0/2

_a(i+cos0) ~~
a sin

dr

=tan

tan

,1T
-

/)

**

*^

or

Hence

^i=*f + 2
r=6 (I cos

For the curve

dr

_b

'd9~~

0),

we have

8m

'

or

tan cA=
Y

Hence

t/0

t
dr

fe(l
=--.

cos0)

b sin

--tan
,

DIFFERENTIAL CALCULUS

270
Therefore,

^^

and hence the curves cut each other

-Tr/2

afc

right angles.

Exercises
Find for the curves

1.
(i)

cos

r=0(l

0)-

(Cardioide).

m =a rn cos m 6m
m (cos m$ + sin wifl).
r -a

(11)

cos 0. (Parabola)
(/v)
that the two curves
2
a
r =a cos 20 and r=a(l-f cos 0)

(Hi) 2alr=*\ -f

Show

2.

intersect at an angle 3 sin- (3/4)*.


m
m
3.
Show that the curves r =a cos mfl,

w- w

cut each other

sin

orthogonally.
4.
(i)

Find the angles between the curves


r=aO, rS=a

(Hi) Y

=a

r=ae

(vi)

Show

npp' be drawn
meet in

T, then
6.

cosec (0/2),

r=6

sec (0/2)

re

+ 0),

r=-0((l-f

(v) r

sin

20-4,

- 16

sin 20

=ft.

that in the case of the curve

P and

P'

r=a

(sec 0-J-tan 0), if a radius


and if the* tangents at P, P'
(Af-C/.>

that the tangents to the cardioide

r^=(l+cos 0)
whose vector ial angles are rr/3 and 2^/3 are respectively

perpendicular

cutting the curve in


TP= IT'.

Show

at the points

19

r=flfl/(l

r=alog0, r-0/log0

(iv)

5.

(//)

parallel

and

to the initial line.

on the same side of the initial


7. The tangents at two points P, Q lying
show that
0), are perpendicular to each other
r=a(H-cos
cardioide
the
of
line
at the pole.
an
n/3
angle
subtends
the line PQ
chord of the
B Show that the tangents drawn at the extremities of any
the pole are perpendicular to
which
cos
through
passes
0)
cardipide r=a(l-f
;

each other.

r-=<i(i + cos 0) are parallel, show that


an angle 2"/3 at the pole.
subtends
of
contact
the line joining their points
that
10. Show

9.

If

two tangents to the cardioide

(Agra 1952}

from pole to the tangent.


From the pole O draw OY perpendicular to the tangent at any point
P (r e) on the curve r=/(0) Y being its foot.
Let OY^p
126*.

Length of the perpendicular


;

From the &OPY, we


p
r
-

=sin

get
.

<f>

or

p=r
which

is

a very useful expression for p.

We now
ordinates
9.

sin <,

r,

We have

express, p, in terms of the co~

and the derivative

of, r, w.r.

to

271

TANGENTS AND NORMALS

tan

O
j
cfc=?r -~j-

-*7\*

which we can write as


1

__r

+(dr/d0)_

11

/dr_\

Yet, another form of p will be obtained,

if

we

write

r=l/,
so that

dr

~"~

Substituting these values in

w2

(//),

'

</6i

we

get

12 7. Lengths of polar sub-tangent and polar sub-normal. Let


the line through the pole 0, drawn perpendicular to the radius
vector OP, meet the tangent and normal at P in points T and G
respectively.

Then, OT, OG are respectively called


polar sub-normal at P.

From

the

the polar sub-tangent

&OPT, we get

OT
OP =tan
or

Hence

polar sub-tangent

=r

d^
,-

dr

d#
where
j
du
.

From the A^PG, we

w=

get

Fig.

88.

= cot ^,
or

OG=OP cot

Hence, polar sub-normals

<A

du

~A* where
dtf

w=

and

272

DIFFERENTIAL CALCULUS

Note. The lengths PT and PGara sometimes called the lengths of the
polar tangent and polar normal at P. We have
>

r=OPseccp=rV[l-ftan q>]=r/\/rH-r
8

PG=OPcosec<p=rV[l+cot <p]=<\/r
12-8.

is

('

r -f

-fa

1.

To obtain the pedal equation of a curve

Pedal equations.

whose polar equation

* VI

given.

Let
r~f(6) be the given curve.

We

...(/)

have
1

Eliminating 6 between
equation of the curve.

and

(i)

fdr

we obtain the required pedal

(//),

The pedal equation is sometimes more conveniently obtained


between (/) and
6 and

by eliminating

<f>

dQ

tan <=/
r

dr

p=r

'

sin <.

Exercises

Show

1.

that for the curve

the length of polar tangent

is

constant.

Prove that for the curve

2.

the polar sub -tangent

is

constant.
a9, the polar sub-nDrmal

3.

Show

4.

Find the polar sub-tangent of


(/)

that for the spiral

r-a(H-cos0).

(Cirdioide)

(i7)

2^/r^=l

+e

(Hi)

r~a0 1(0-1).

Show

5.

that for the curve

*
rae L/)2

pol^r sub-normal varies as 9 2


'
polar sub- tangent
.

Find the polar sub-normal of


(/)

1.

r=aje

Find p for the curve

(//)

=a+b cos

Q.

is

C3nstant.

cos Q.

(Conic)

KXERCISE8
Find the pedal equation ofr m =am cos m0.

8.

Logarithmically differentiating,

dQ

=r
or

f^

Thus

p-r

-.

sin

mm

is

we get

-=~-m

which

273

tan/n 9

-=-cot w0 = ta

p=r
P

(sin Jfi-}-m0)=r cos

m9.

or

the required pedal equation.

9.

Find the pedal equations of

(iV)
(v)

//r-

1 -j

<?

cos 0-

r-ci(Mcos0).

(v) r(l-sinj0) 2
(ix)

m =u m

a.

(iv)

r= o0.

(v/)

r--asin/ii0.

(viii)

sin mO~\ b m cos

r^-a + bcos$

iwfl.

10. Prove that the locus of the extremity of the


polar sub-normal of the
curve r=/(0) is the curve

r=-/'lfl-l").

Hence show

that tlie locus of the extremity of the polar

the equiangular spiral


11.

a?

w^

is

sub-normal of

another equiangular spiral.

(P.U. 1940)

Prove that the locus of the extremity of the polar sub-tangent of the

curve
!
-

4/(0)--0

it

Henoe show

/'(R4

that the locus of

ttie

0).

extremity of the polar sub-tangent of

the curve
r

is

(H

tan 10)/(m f-wtan 10)

a cardioide.
12.

form

Show

that the pedal equation of the spiral

Allahabad 1943)

r=^a sech

is

of the

CHAPTER

XIII

DERIVATIVE OF ARCS
On the meaning of the lengths of arcs. The intuitive
13*1.
notion of the length of an arc of a curve is based upon the assumption
that it is possible for an inextensible fine string to take the form of
the given curve so that we may then stretch it along the number
axis and find out the number which measures its length.
This assumption, which is not analytical, cannot be the basis
of analytical treatment of the subject of lengths of arcs of curves.
Also to define lengths of arcs analytically is not within the scope of
this book.

Hence we have to base our treatment on an axiom which concerns the numerical measure of lengths of arcs and we now proceed
to give it.
Axiom. If P, Q, be any two points on
a curve such that the arc PQ is throughout its
length concave to the chord PQ, then
Chord

PL

where

PQ<arc

and

QL

PQ<PL+QL

are any two lines enclosing

the curve.
Fig. 89

An

important deduction from the axiom.

If P,

be any two

points on a curve, then

PQ
= l.
rfxs
PQ
near P that the

arc

,.

lim

Q-+P
we take Q

chord

so
To prove it.
concave to the chord PQ.
Let PL be the tangent at P.

arc

PQ

is

Draw

Ai/)*

QLPL.
Let /_LPQ=a.
chord

PQ

<arc

we have

PQ<PL+LQ,

^ PQ

chord
or

chord

PQ

everywhere

cc+sm

a.

274

Fig.

90

275

DERIVATIVE OF ARCS
Let

-*

P so

limiting position

From

(/),

that chord

and a ->

we

PQ

tends to the tangent

PL

as its

0.

get

arcPQ
_^ ^ chord

^^*

~~

PQ

'

for
(cos

a+sin

a)

>

as a -> 0.

13-2.
Length of arc as a function. Let y =f(x) be the equation
of a curve on which we take a fixed point A.
To any given value of x corresponds a value of y, viz. 9 f(x)
to this pair of number x and/(x) corresponds a point P on the curve,
and this point P has some arcual lengths 's* from A. Thus 's' is a
function of x for the curve y=f(x).
Similarly, we can see that 's* is a function of the parameter '*'
for the curve
;

(Parametric Equation)

x=f(t) y=F(t),
a function of for the curve
9

and

is

r=f(6).
Cartesian Equations.

13*3.

(Polar Equation)

To prove that

for the curve

Let

V denote arcual distance of any

fixed point

We

A on

point P(x 9 y) from some

the curve.

take another point

AQ=s+Ss

Let an arc

Q(x+8x

yi-By) on the curve near P.

so that

PQ=Ss.

arc

From the

rt.-

&PQN,

angled

by

we

have

or

L
Fig. 91

or

fchord PQ-\*

L~a?cPe J
Let

->

P so

( 8s \

^J

that in the limit


'

or

'

()'='+()

MX

DIFFERENTIAL CALCULI^

276

We make a

convention that for the curve >> =/(*), 's


ed positively in the direction of, x, increasing so that,
with x. Hence ds\dx is positive.

is

measurincreases-

s,

Thus we have

taking positive sign before the radical.

Cor 1. If x=f(y) be the equation of the curve, thenfunction of y and, as above, it can be shown that

Cor.

From the

2.

cos

right-angled

Aror,
= &*
L NPQ=
pQ s/

's'

is

a>

PNQ, we h ave
arc

chor

is the angle which


-> ^ so that
-> //, where
j2
A'-axis.
with
at
jP'nmkes
direction
of
the
positive
tangent

NPQ

Let

cos r
d/=

dx

_
.

J^

1=

dx
ds

Hence
003

Similarly,

*=

dx
'

ds

we can show that


.

,
sin r
dt=

dy
,

ds

Parametric Equations.

13-4.

To prove

that

for the curve

Let *s' denote the actual distance of any point P(t) on


from a fixed point A of the same.

We
be

take another point

Q(t+St) on the curve.

its co-ordinates.

Let
B,TC.PQ

From

the

A PNQ>

= SS.

(Fig- 91, p. 275),

we get

Let

the-

277

DERIVATIVES OF ARCS

'

St

Bt

or
2

/chord PQ\*( &s


1.
arcPQ / ( St

Let

We
that dsjdt

Q -> /* so that

measure
ie

2
8y \
/

^Uf

in the limit

positively in the direction of,

/,

increasing so

Thus

positive.

ds
:

It

dt

Polar Equations.

13-5.

To prove

)'+(dt )*T
that

for the curve

denote th'3 arcivvl length of any point P(r, J) from some


fixed point A on the curve.
We take another point Q(r+Sr, fl-f 80) on the curve near P.

Lot

'5

Let arc

AQ=s+$s

so that arc

PQ^bs.

Fig. 92

Draw PN.OQ.

Now
JP7V/OP=sjn 85 so that

PJVW

sin S^.

Again

ONjOP^cos 89
Hence

so that OJV^==r cos 89.

NQ^OQ-ON
= r+8r~- r cos 80
=:rl

cos 80

+ 8r

DIFFERENTIAL CALCULUS

278

From the

&PNQ

Dividing by

(S0)

we

get

we

get

~~^
PQ^

/chord

_
~~

Let

We

measure ^' positively in the direction of Q increasing so

that dsjdff

-> P.

is

Therefore

positive.

-2

Thus

-V

Cor. 1. If 0=f(r) be the equation of the curve, then


function of r and, as above, it can be shown that

_
Cor.

2.

..,,,

sin

PN

s' is

r sin

= T sin 80

arc

80
"

~8lT

Us

P so that /_PQN ->

chord

PQ

^ where, <, is the angle between


of
the
the positive directions
tangent and the radius vector.
Let

->

sm ^=r.
,

/.

1.

-j~

or

Again,

cos

2r sin2

TQ

279

EXERCISES
2r sin 2

JSj+Sr

'

80
sin

arc

80^
S7~

^
'

chord

PQ
PQ

Sr \ 80

80

arc

8T

Let

->

chord

P
f r\
cos r
<i=(V r. 0.
i

^r N do .
.1

1+
,

dQ ) as

or

dr

Exercises
Find

1.

(i)

for the curves

t/5/d[x

cosh

y=c

jc/c.

^==a log

(//)

3V=^ (~ x).
5

(///)

(v)
(v/i)

8ay-x
4^-f

ind

2.

3.

Show

2
(a -jc

c/5/c/y

that ydsjdy

(v/)

).

logx=x

for the

2
.

curve

is

constant for the curve

j^jhi^jz^

_i
~ g

tf^^-rr)

>

fl

Find dsldQ for the following curves,

4.

(/)

(//)
(111)

d'v)

x=^a cos

^,

y^b

x=a cos 3

^,

^=sin 3 ^. (Astroid)

x=a(9

sin ^),

x=ae

sin 0,

(v) jc==fl(cos

Find dsjdB

5.

(/)

r=>ae

(iv)

r =00.

(v/i)

(v/ii)

6.

y=a(l-cos

0-f0 sin

0),

cos

0).

(Cycloid)

cos

y=a(sin

(Cardioide).

0).

(//)

=fl cos 20. (Lemniscate)

(Equiangular spiral)
8

(v)

r=a(0 -l).

w =aw cos nt0. (Cosine spiral).


m m (cos w0+sin w0).
r =a

Show

0.

for the fo'lowing curves

cot a

being the parameter

sin ^ (Ellipse)

y=ae"

r=a(l+cos0).

(///)

x 3 =ay*.

(iv)

that for the hyperbolical spiral

rfr

r0=a,

DIFFERENTIAL CALCULUS

280
7.

Show

that r ds/dr

8.

Show

that

is

^r

constant for the curve

.is

constant for the curve

dr

If /?i and /? 2 be the p^rp^ndicular^ from tlv^ origin on the tangent and
9.
normal respectively at the point (x, y\, and if tan ^ -dy/dx, prove that

p^x sin ty~y cos ^,

and

/i a =.=a:

hence prove that

10.

Show

that for any pedal curve


ds

dr

r
2

V(r -/?

"

2
)

pf(r],

cos ipi-y sin

^f

CHAPTER XIV
CONCAVITY CONVEXITY
;

POINTS OF INFLEXION
14-1.

Definitions.

Draw

P[c,f(c)] thereon.

We suppose
f'(c) is

some

Now
sider

finite

Consider a curve y=f(x) and any point


the tangent at P.

that this tangent

is

not parallel

X-axis so that

number.

there are three

mutually exclusive possibilities to con-

V"

+X
Fig. 93
(/)

may
of

be,

Fig. 94

Fig. 95

portion of the curve on both sides of P, however small it


lies above the tangent P (i.e., towards the positive direction

X-axis).

In tliis case \vo say that the curve


convex downwards at P. (See Fig 93).
(//)

it

may

A
be,

is

concave upwards or

portion of the curve on both sides of P, however small


lies below tho tangent at P (i.e., towards the
negative

direction of X-axis).

In this ease we say that the curve


convex upwards at P. (See Fig. 94).
(Hi)

is

concave downwards or

The two portions of the curve on the two

different sides of the


at P. In this case we

tangent at
say that P

P,
is

sides of

(See Fig. 95).


Thus, the curve in the adjoined
figure 06 is concave upwards at every
point between P and P2 and concave

downwards
and Po.

at every point between

P2 and P 3 are
flexion

lie

on

the curve crosses the tangent


a point of inflexion on the curve.
/ e.,

Pa

the points of in-

on the curve.
Fig. 96

281

DIFFERENTIAL CALCULUS

282^

From these

figures, it is easy to see that the curve changes the


of its bending from concavity to convexity or vice versa as a
point, moving along the curve, passing through a point of inflexion.
This property is also sometimes adopted as the definition of a point
of inflexion.
Note. The property of concavity or convexity of a curve at any point

direction

is

not an inherent property of the curve independent of the position of axes.

directions, as also positive and negative directions of


by convention and have no absolute meaning attached to
them. But this is not the case for a point of inflexion. The point where a
curve crosses the tangent is an inflexional point so that its existence in no way
depends upon the choice of axes.

Upward and downward


y-axis, are fixed

14 2. Criteria for concavity, convexity and inflexion. To determine whether a curve y~f(x) is concave upwards, concave downwards,
or has a point of inflexion at P [c,f(c)].

We take a point
the point P[c,/(c)].

Q[c+h, f(c+h)] on the curve y=f(x)

lying near

Q lies to the right or left of the point P according as


or
negative.
positive
Draw
\_X axis meeting the tangent at P in R.
The point

is

QM

The equation of the tangent at P is


y-Ac)=f'(c) (x-c),
and therefore the ordinate
abscissa c+h is

MR of this

f(c)+hf'(c).

tangent corresponding to the

CONCAVITY, CONVEXITY, INFLEXION

MQ

Also, the ordinate

283*'-

of the curve for the abscissa

c+h

is-

= MQ~MR=f(c+h)-f(c)-hf'(c).
For concavity upwards

MQ >

MR,

at P, (Fig. 97).

i.e.,

RQ,

For concavity downwards

MQ <
For

i.e.,

positive

when Q

lies

on

either side of P.

lies

on

either side of P.

at P, (Fig. 98).

RQ is

when Q

negative

inflexion at P, (Fig. 99).

RQ
Q

MR,

is

is

positive

when Q

on the other side of

lies

lies

on one side of P and negative when?

P.

Thus, we have to examine the behaviour of RQ,

i.e.,

f(c+h)-f(c)-hf'(c)

which are

for values of, h,

theorem, with remainder after two terms,

By Taylor's

Case

sufficiently small numerically.

Let

I.

/"(c)

As the value f"(c) of/"(x)

is

>

we have

0.

for

positive

x=c,

there exists an-

every point x of which the second derivativef(x) ia positive. (3*51, p. 54). Let c+h be any point of thisThen c+yji is also a point of this interval and accordingly
interval.
interval around

/"(-f

2 A) is

for

positive.

Also
/*

/2

!,

is

positive.

RQ >0,
for sufficiently small positive

Hence
Case

and negative values of

the curve is concave

II.

Let

upwards at

f"(c)

<

P iff"(c) >

the curve is concave

Case HI.

By

Let

downwards at

/"(O^O

0.

0.

In this case we may how, as above, that


small positive and negative values of h.

Hence

h.

RQ <

P iff"

for sufficiently

(c)

<0.

fy///'"(c):0.

Taylor's theorem, with remainder after three terms, we*

DIFFERENTIAL CALCULUS

:284

have
h*

so that

RQ=

3*

f'"( c

+e

h )> f r /"(c)=0.

There exists an interval around c frr every point X of which


the sign of/"'(c). Let c+li be any point on this interval.

f'"(x) has

Then

c-\-Q.Ji is also a point of this interval and accordingly f'"(c-{-Pji)


has the sign of f"'(c)

But A 3 /3

changes sign with the change in the sign of h.


changes sign with the change in the sign h.

Thus RQ
Jhe curve has inflexion
Case IV.

at

Hence

P iff"(c)^0 andf"

Generalisation

Let

Taylor's theorem, with remainder after n terms,

By

f(c+h)=f(c)+hf'(c) +

**

f"(c) +

^7!

/n

we have
~
1(C)

so that

There exists an internal around,


sign

off

n
(c)

c,

such that / n (c-f-0 n /i) has tho

or every point c4-h of this interval.

Also h n fn changes its sign or keeps the same sign while the
sign of, /i, changes, according as n is odd or even.
I

Thus
/i

is

RQ

changes sign ifrcis odd and keeps tho sign of/

n
(c),

if

even.

Hence if n is odd th*. curve hxs inflexion atP\ifn is even,


n
concave upwards or downwards according as f (c) > 0, or <0.

it

is

Another Criterion for points of inflexion. We know that


inflexion at P if it changes from concavity upwards to
concavity downwards, or vice versa, as a point moving along the curve
P such
passes through P, i.e., if there i a complete neighhourhood of
that at every point on one sid^ofP, lying in this neighbourhood, the
curve is concave upwards and at every point on the other side of P
14-3.

a curve has

CONCAVITY, CONVEXITY, INFLEXION


the curve

IR

concave downwards. We thus see that the curve, has


iff(x) changes sign as x passes through c.

in-

flexion at P[c,f(c)],

'n

Note. We have already remarked that the position of the point of inflexion on a curve is independent of the choice of axes so that, in particular, the
positions of x and y axes may he interchanged witnout affecting the positions or
the points of inflexion on the curve. Thus the points of inflexion may also be
For pomts \vhere
determined by examining d z xldy 2 just as we examine d zy/dx 2
the tangent is parallel to X-axis, i e., where dy/dx is infinite, it becomes neces.

ary to examine d*x/dy instead

14-4.
Concavity and convexity with respect to a line. Let P be
a given point on a curve and, /, a straight line not passing through
Then the curve is said to be concave or convex at P with respect
P.
to /, according as a sufficiently small arc containing P and extending
to both 'sides of it lies entirely within or without the acute angleformed by, /, and the tangent to the curve at P.

Fig. 100

Thus the curve

Fig. 101
is

convex to

at

in Fig. 100

and concave

inr

Fig. 101.

In the noxt section, we deduce a test for concavity and convexiwith


ty
respect to the X-axis.

14-41.

test

of concavity and convexity with respect to the X-axis.

Fig. 103

Fig. 102

103,

From the examination of


we deduce the following

the figures 97, 98 and the figures 102 r

(i)

positive)

A
is

curve lying above, the axis of. x (so that the ordinate y is
convex or concave wit h respect to the axis of x according

DIFFERENTIAL CALCULUS

286
as

it

concave upwards or downwards,

is

i.e.,

according a,ad*y/dx*

is

positive or negative.

A curve lying

(ii)

below the axis of x (so that the ordinate y is


convex or concave with the respect to the axis of x accoris concave downwards or upwards i.e., according as d 2yl<tx*

is

negative)
ding as it
is

negative or positive.

Thus, in either case, we see that a curve


.P with respect to the axis of x according as

convex or concave at

is

d*y
dx*

y
ds positive or negative at P.

Examples
I.

.is

Find if the curve

concave or convex upwards throughout.

(P.U. 1932)

Now
dy
dx

~~~

d*y

which
i.e.,

is

1
~

always negative.

Hence the curve is concave downwards,


convex upwards throughout.
Note.

We know that y=Iog x is negative or posi0<x<l or x >1. Thus for 0<x<l,

tive according as
2

yd*yldx is positive and for x>\,yd*y(dx* negative, so that


*k e curve is convex w.r. to x-axis if 0<x<l and concave

Fie
104
rig. ju^
2.

iy.r.

5Aow

//ra/

to a?-axis if

j=x

4 is

#>1.

concave upwards at the origin.

We have
=

dx

~dx*

24,

that

but positive at
Therefore the curve
3.

is

concave upwards at the origin.

Find the ranges of values of x for which

as concave upwards or downwards*


Also, determine

(0, 0).

its points

of itfft&xion.

CONCAVITY, CONVEXITY, INFLEXION

287

We have

Now,

-V~>0

forx<l,

forx = l

"'=0;
dz v

for

1<*<2,

<0; forx=2,

for

*>2,

>0.

=0,

Thus the curve is concave upwards in the intervals


and is concave downwards in the interval [1. 2].

oo, 1),

<(2, GO ]

It has inflexions for

Therefore
the curve.
4.

Show

(1, 19)

x=l

and

(2,

that the curve

and x = 2.
33) are the
2

(a

two points of

+x 2 )^=a 2^

inflexion

has three points of

on
in-

flexion.
'

Here

_
~

dy
dx

~~ a 2

dx*
"

*=Q

for

x= ^3^,

0,

2
2
It is easy to see that d y[dx changes sign as, x, passes
the
Hence
curve has inflexions at the
ach of these values.

through
corres-

ponding points.

Thus
(V3a, V3a/4),

(0, 0),

(-

the three points of inflexion of the curve.


5.

Find the points of inflexion on the curve

Here, y,

is

the independent and, x, the deper\dei|t variable,

dx

--3

xl

(logy)-,

1
.

DIFFERENTIAL CALC0LUS

288

d'x

.y=0 or

log

>

=0

if log

y~2,

i.e., if

=e=
e*.

Thus we expoct points of


Again
6

~~

or

inflexion oti the curve for

121oj; """ Qlocry

^~

=l

and".

Therefore

and e 2 give points of


are the two points of inflexion.

Hence
9
(8, e )

y~

inflexion,

so that

(0,1)

and'

Exercises

y~e x

everywhere concave upwards.

1.

Show

2.

Examine the curve .y^sin x for concavity and

that

is

convexity in the-

interval (0, 2*).


3.

is

concave upwards or downwards.


4.

is

Also,

which the curve

fin

the points of inflexion.

Also

find its points

of inflexion.

Find the intervals in which the curve


y= (cos* i sin xie x

concave upwards or downwards


6.

in

Find the ranges of the values of x in which the curve

concave upwards or downwards.


5.

i*

Find the ranges of values of x

x varying

in the interval (0, 2n).

Find the points of inflexion the curves


2

(i)

(Hi)
* V'

y--ax*-\-bx +<QX
4

A*=^-3.y

y ~~a 2

+x

3
4.v -f 5.

[xi)

y*^x(x\-\)*.

ny

(n)
(iv)

y=(x*

(yla).

=jc f (fl 2 -x).

(x)

*)/(3.x -f !)

x-~(y~\)(y2)(y3)"
a--j-x

xy^a* log

(xii)

Vl ' ;

'

(ix)

\-d.

2
y- x log

(D.U. Rons. 1947)


(D.U. Hons. 7955)

'

289

EXERCISES
Also, obtain the equations of the inflexional tangents

(7) and

jjto

the curves

(it),

(xi).

Show

1.

that in the curve

the abscissae of the points of inflexion are

Show

8.

that the line joining the

h.

two points of inflexion of the curvs

y*(x-ay x*(x+a)
subtends an angle w/3 at the origin.

Find the values

9.

of, x, for

which the curve

54,y=U+ 5)^-10),
has an inflexion and draw a rough sketch of the curve for

-6<x<3, making

Show

10.

the

(MT.)j

inflexions.

that the curve

ay*=x(xa)(xb)
has two and only two points of inflexion.

Find the points of inflexion on the curves

11.

(/)

(ii)

/,

y~a

sin

cos

x2a cot Q, y=^2a sin

f.

0-

of the points of inflex on on

Show

that the abscissae


satisfy the equation

12.

y*=f(x)

(Hi)

jc=0(2<9-sin 9),;y=fl(2-cos 0).


x--=a tan

[f'(x)?=2f(x)f''(x).
13.

Show

the curve

(PV.)

that the points of inflexion of the curve


2

y =(jf-fl) (.v-6)
lie

on the

line
4/>.

(Luckjow)

CHAPTER XV
CURVATURE
EVOLUTES
Introduction, Definition of Curvature.
15'1.
In our everyday
language, we make statements which involve the comparison of bendFor instance, at
ing or curvature of a road at two of its points.
times, we say, "The bend of the road is sharper at this place than at
that." Here we depend upon intuitive means of comparing the
curvature at two points provided the difference is fairly marked. But
we are far from intuitively assigning any definite numerical measure
to the curvature at any given point of a curve. In order to make
'Curvature' a subject of Mathematical investigation, we have to
assign, by some suitable definition, a numerical measure to it and
this has to be done in a way which
may be in agreement with our
intuitive notion of curvature.
This we proceed to do.

We

take a point Q on the curve


P.
near
lying
Let A be any fixed point on the

Let

curve.

arc

AP=s,

arc

AQ=s-\- 8s,

arc

PQ=Ss,

so that

Let

i/s

i/'+St/',

be the angles which

the positive directions of the tangents


at P and Q make with some fixed line.

The symbol, 8^ denotes the angle


through which the tangent turns as a
point moves along the curve from P to
a
8s.
distance
According to our intuitive feeling, 8$ will
Q through
be large or small, as compared with 8s, depending on the degree of
sharpness of the bend.
This suggests the following definitions
Fig. 105

(i)

the total bending or total curvature of the arc

be the angle 8$
(//)

the average curvature

ratio 8(pj8s

and

(ill)

PQ

is

defined to

of the arc

PQ

is

defined to be the

the curvature of the curve at

lim

Q->P

'"

Ph
ds

290

defined to be

CURVATURE
Thus, by def.,

A/r/ds is the curvature

15-2.
Curvature of
circle is constant.

291

of the curve at any point

To prove

circle.

Intuitively, we feel that the curvature


throughout its circumference and that the

that the

curvature oj

of a circle

is

uniform

larger the radius of the


It will now be shown that

the smaller will be its curvature.


these intuitive conclusions are
consequences of our formal definition
of curvature.
circle,

Consider any circle with radius, r, and centre O.


Q be any points on the

Let P,
circle

and

'

Also let
tangents PT,

We

let arc

be the point where the

QT at P and

meet.

have

From Elementary Trigonometry


r

Fig. 106

top

!_

""""

Us
Let

->

so that in the limit,

we have

~~

ds
its

Thus the curvature at awjy point of a


radius and is, thus, a constant.

Also,
r increases.

15-3

it is

circle is

clear that the curvature, 1/r,

Radius of curvature.

and

is

decreases as the radius

The

reciprocal of the curvature


curvis called its radius o
generally denoted by, p, so that we have

of a curve at any point, in case


ature at the point

the reciprocal of

it is

=0,

= dThus the radius of curvature of a

circle at

any point

is

equal

t<?

its radius.

only

15-31.
The expression ds}dfy for radius of curvature is suitable
for those curves whose equations are given by means of a

between s and (p. We must, therefore, transform it so that


be applicable to other types of equations such as Cartesian,

relation
it

may

Polar, Pedal, Tangential Polar, etc.


It may be remarked that in the case of Cartesian and Polar
equations, the fi^ed line will be taken as X-axis and initial line res-

For the curve y=f(x), the positive direction of the tanone


in 'which, x, increases, and for
the
gent
r=/(0) the positive*
is the one in which, 0, increases.
direction of the
pectivelyis

tang&j^

DIFFERENTIAL CALCULUS

292;

Ex.

Find the radius of curvature


(j)

any point of the following


(//)

(///)

(v)

at

j=c tan (Catenary).


s-4a sin ty (Cardioide).

(/v)

s=4a

s~c

sin

ty

log sec

(Cycloid).

(Tractrix).

sa log (tan 1^4-sec t//)fa tan ^ sec ^ (Parabola).


Radius of curvature for Cartesian Curves.

15*4.

15 41.

Explicit equations

y=f(x).

Now,
tan Y
\b =
Differentiating, w.r. to

s,

dy
/--

dx

we

get

dx

But, we have

(13-3, p. 275}*

where the positive sign

to be taken before the radical.

is

Hence

_~
~"d^

d2y

y2

dx 2

The radius of curvature,

Cor.

is positive or
P,
negative
or
d*yjdx*
positive,
negative ie., according as the
curve is concave upwards or downwards. Also, the equation (A)
shows that curvature is zsro at a point of inflexion.

according as

iSi

is

S'mss the valus of p is indspendent of the choice of X-axis


interchanging x and y we see that, p, is also gfcven by

Tfhis

formula

is

specially useful

when the tangent

is

perpendicular

K^axis in which case


L5;42.

Implicit equations

f(x,

and

y)=0.

^t points where dfjdy^fyjkQ-, we have, (JlO-94/ p. 213)

to?

CURVATURE

"

293

....... "

.....

<***

(/,)*

2
2
Substituting these values of dy\ dx and d yjdx in the formula
A) above, we get

the sign
Note.

"

is

or

according as fv

The form of the formula

Parametric equations

At points where

f'(t)^0

negative or positive,

(B) shows that the expression for p will


^0 but/^0, The exceptional case

etain the same format points where /


y
^hich arises for the points where fx and fv
onsidered in chapter XVII.

1543.

is

become simultaneously

be

x=f(t), y=F(t).

we have

*L_^/_>dx~f'(t)'
(l

will

"'

w
/

yf'(t)F"(t)-F'(t)f"(t)

2
2
Substituting these- values of dyjclx and d y/dx in the formula
above, we get

Lf(t)+F'(t)]*_
vhere the sign
Note.

ame form

is

or ~~ according as/'(0

The formula

for the points

is

'

>t<l

positive or negative.

(C) shows that the expression for, p, will retain the

where /'(0=0 but F r U)^0.

If a curve passes
Newtonian Method.
15-44.
at
is
the
x
origin, then
tangent
origin and the axis of
,

rives the radius

of curvature at the

as

through

the

x ->

origin.

Here we obtain the values of y l and y%

at the origin.

2
Now, x l'2y, assumes the indeterminate form 0/0

Iim -*-= Urn

= hm
..

^
1

as

x ->

0.

(--)
1

204

DirFBBBNTlAL CALCULUS
Also, from the formula (A) of

Thus at the

15 41,

we have

at the origin

origin where x-axis is a tangent,

<.=

Um
shown

It can similarly be

that

at the origin where

Y-axis

is

tangent,
=

lim

(g-).
These two formulae are due to Newton.
15-45.

Generalised Newtonian Formula.

If a curve

through the origin and X-axis


tangent at the origin, we have

X 2 -4-V 2
lim

SX^

=lim

'X

V \

+-~^
2 J
\2y

,~=lim[

2y

passes
is the

-----

2y

= p, at the orgin.

Fig. 107

=OP

2
2
2
is the square of the distance of any point
Here, X +J
n
the
curve
the origin O and, j, is the distance of the
from
y)
P
from
at O.
the
X-axis
point
tangent
,

P( x *

Interpreted in general terms this conclusion can be stated a


follows (see Fig. 108.)
:

of a curve, and
If OT be the tangent at any given point
PAf, the length of the perpendicular drawn from any point P to the
tangent at O, then the radius of curvature
at

lim

when the point

OP 2
2PM'

P tends

to

as

M
its limit.

Fig. 108

Note. It will be proved later on in 17-31, on p. 332 that the tangent


at the origin lying on a curve is obtained by equating to zero the lowest degree
terms in its equation.

Examples
1.

In the cycloid

x=a(/+sin

t),

j;=

cos

t),

CURVATURE

295

prove that

L
&

P=4acos

(P.U.1944)

We have

~a
dy

>

s t),

a sin

fc

a(

_ 2_8i ?

+0087)

1
g

QPO ,
8ec

2
1

*/

__

'~

,..__
'

-_

11

rfx'

<

/.

2"ooB772

^sec

sin

--

2a cos 2

_
4a

cos

=4a

__

C08

"

>

COS*

2.

5Aoif

f/iar

Folium x*-\-y*=Zaxy
Differentiating,

/Ae curvature
is

we

of the point (3a/2, 3a/2) on

8^2/30.
get

xl+^^^ + flx.

or

Again, differentiating

(/),

we

...(/)

get

+V [*J^.g..

Substituting 3a/2,

1 for x, y,

3fl/2,

Hence the curvature

dyjdx respectively,

"o )

8V2
f

we

get

DIFFERENTIAL CALCULUS

296

Find the radius of curvature at the origin of the curve

3.

It is easy to see that X-axis is the tangent at the origin.


15*45, p. 294]
[Refer note

Dividing by y

Let x -

we

get

so that lim

= 2p.

(x jy)

x->0
0.2p+5.2p-8=0,
or

p=4/5.
Exercises

Find the radius of curvature at any point on the curves


c cosh (xlc) (Catenary).
(/) y

1.

(//)

x-=a (cos

sin

t-l-t

(W)x% -Kyf =--fll


(iv) x^(a cos t)lt,
2.

/),

y^a(sin

cos

t).

(D.U. 1953)

(Astroid)

y^=(a sin

t)it.

Find the radius of curvature at the origin for


2

(//)

x*y-xy*-\ -2x y
2

(ill)

3.

is

2x*'\-ly*-\-4x

Show

y+xy-y

-\-2x- -0.

(P.U. Supp. 1939)

that the radius of curvature of any point of the astroid


x=a cos 8 0, y-^a sin 3

equal to three times the length of the perpendicular from the origin to the
(Andhra 1951)

tangent.

4.

Show

that for the curve


),

the radius of curvature


is

is, a,

Show

2c cosh 2
6.

(tic)

Show

sinh

(//c),

where

cosh

is

y^~-2c

cosh

t
.

the parameter.

that the radius of curvature at a point of the curve

x^ae

*
fi

fi

(sin 9

cos

0),

y=a?

(sin 0-fcos 0)

twice the distance of the tangent at the point from the oiigio.
7.

Prove that the radius of curvature at the point


(

8.

at

0),

that the radius of the curvature at any point of the curve

x=tc smh

is

cos

which the value of the parameter

TT/4.

5.

is,

(H

>>=flsin

at the point for

Show

2a, 2a)

on the curve

jc

f
l

y=flU

+y i )

is

"~2a

that the radius of curvature of the Lemniscate

the point where the tangent

is

parallel to x-axis is

CURVATURE
Prove that the radius of curvature

9.

is

297

point (2a, 0) on the curve

at the

(65)1 n/(536).

Find the radius of curvature for 4(xla)-4(ylb)=l at the points

10.

where

it

touches the co-ordinate axes.

Show

11.

that the ratio of the radii of curvature at points

on the two

curves

which have the same abscissa varies as the square root of the ratio of the ordinal es.

Show

12.

that the radius of curvature at each point of the curve


#=-a(cos /-flog tan J/), y=a sin f,

the
is inversely proportional to the
length of the normal intercepted between
point on the curve and the X-axis.
13. Show that 3V3/2 is the least value of
for j?=log x.
p
j

at which the curvature is maximum,


and sh^w that the tangent at this point forms with the axes of co-ordinates a
triangle whose sides are in the ratio 1
V2 ^3.
(Rajputana 1951)
14.

Find the point on the curve

y=e x
:

15.

is least for

16.

Show

that the radius of curvature of the curve given by

the point x^a and its value there


Prove that for the ellipse

is

9a/lO.

=
p, being the perpendicular

from the centre upon the tangent at any point

(x, y).

(P.U. 1944)
17.

Prove that for the ellipse


x*la*-'r y*lb*

where

CD
18.

is

-l,p-C>

/a&

(Madras 1953)

the semi-conjugate diameter to CP.

Employ

generalised

Newtonian Formula to show that the radius of

curvature at any point of the ellipse x 2 /a 2 -! j 2 //> 2 -- 1 is equal to


3
(normal)
__
2
(scmi-idtus rectum)
19. The tangents at two points P,
Q on the cycloid

___

show

cos 0)

sin0), y=--a(\

;c=--a(0

are at right angles

that if p lf p a be the radii of curvatures at these points,

then
a

Pi

20.

ture

is 7.i

+P,

lfa l

Find the points on the parabola y a =8x

at

which the radius of curva-

21.

(a)

Prove that

parabola y*=4ax and S be

if,

its

be the radius of curvature

at

2
3
focus, then p varies as (SP)

any point

P on

the

(P-C/- 1952)

a focal
(b) Pi Pi''are the radii of curvature at the extremities of
chord of a parabola whose semHatus rectum is / prove that
;

22.

Show

that in the curve

x=- A fl(sinh u cosh u


if the

normal at P(x,

equal to 3 PG.

y)

fi/),

meets the axis of x

y=a

in G,

cosh 3 w,
the radius of curvature at

is

(B.U.1954)

298

DIFFERENTIAL CALCULUS
23.

If x,

y are given

** -

dylds

p~
also,

show

show that

as functions of the arc s 9

dxjd*
'

d*ylds*

that

jL/^iY

*2y

Find p for the catenary

x=c log
24.

Show

LS+ V(c*+s

)], y--

that for the curve s=f(x),

_ [V~ VA /
25.

f^\_

Prove that for the curve s=ce xlc >

Find p for the curve s*=8ay.

(Patna, 1952)

Radius of curvature for polar curves

15*46.

Here we are to express


with respect to

terms of

ds\dty in

r=f(0).

and

its

derivatives

0.

From

the figure,

we

see that

_>
ds

_
~

ds

dO
ds

di

+ de

dQ
'

ds

IT
Fig. 109

tan

Now,

^= -T

_T

'

860

>

i.

A.
or

-"

TT-

d<f>

r.

tt

ae

dff*

tie

_^V J
I

-T"

77a

r.

CII+-"

Also
...(3,

CURVATURE
where positive sign

From

(1), (2)

ds

is

to be taken before the radical.

and

we

(3),

obtain]

__

f
do

dp

f
J
dr

Therefore

where

and

r=

Cor. Since curvature is zero at a point of inflexion, thereforeat a point of inflexion on a polar curve,
ra

15-47.

+2r1 2 -rra =0.

Radius of curvature for pedal curves.

We know that

= 6 +<f>.

//

Differentiating w.r. to

r,

the relation

/?=r sin
we.

^,

have

=sm +r cos
A

Now

sin

^=r

-,

cos

^=

dr

13-5, Cor. p. 278)

~~ds

300

DIFFERENTIAL CALCULUS
AT

or

Note. The relation $^0+$ is true whatever be the position of the point
on a curve and the tangent at it relative to the initial line. We can satisfy
ourselves on this point, it we examine a few different curves, keeping in view

the following conventions

6 is the angle through which the positive direction of the initial line has
to rotate to coincide with the positive direction ot the radius vector ;
is the
angle through which the positive direction of the radius vector has to rotate to
coincide with the positive direction of the tangent which is that of, 0, increasing;
$ is the angle through which the positive direction of the initial line has to
rotate to coincide with the positive direction of the tangent.

or

For

Fig. 111.

or

A
15 48. Radius of curvature for tangential polar equations.
called
of
is
a
for
an'd
between
curve,
relation
holding
every point
p
i/>,

Tangential polar equation.

The perpendicular drawn to the


tangent at any point (x, y) from the origin makes angle,
?r/2, with #-axia.
i//

Also p,

is

the length of this per-

pendicular.

Therefore the equation of the tan-

gent at
Fig. 112

is

p=sX sini/'

y cos

&

CURVATURE

,301

X, Y being the current co-ordinates of any point on this tangent.


Since the point P(x, y) lies on this tangent, we have

p=x
which

is

Differentiating

(/),

ipy

sin

a relation between

;;,

i//,

to

r.

cos

we

i//,

(p

...(/>>

any point on the curve.

for

x,

get

sn
XT
Now

dx
,

a^

dx
-

=-->rfs-

^-f^ sin

-^

= x cos i^+>
(//) H'.r.

i//-f

sn

Differentiating

dy

to

ds

dy

-TTP
d*p

ds
d>,-= --}

(p,

dip

=x cos

ds
,-,-=? cos

p cos

t//

sin

/>
j//

sin T
<//

sin ^ cos

^.

i//,

we

get

dx
COS

=
=

x sin
x

p=x

i//

COS

cos

-f>'

sin i/^+j cos

sin

y cos

ip

i//

or

For the

First

Example
m
m
curve r =a cos md, prove

Method.

By

logarithmic differentiation,

dr

~~~~

'

dO

r,1

that

,.

dd

m9
m
cos
6
sin

tan md.

dV*

rtn sec 2

*s}

rm*'Qd(fi

md
f

tan

dr
rn9.

m0+r tah 2 mtf.

we obtain

DIFFBBENTIAL CALCULUS

Hence

gec 2

_______
m 0_ tan m g
2

r2

15-46, p. 298>

(See
r a sec 8
_

mQ

(cos mtt

Second
is
273
tpage

Method.

Its pedal

equation, as obtained in Ex.

8,

*~ r~~dn

Hence

r.a m

Exercises

.r.

1.
Find the radius of curvature of the curve r=a cos H0 as a function of
2
(P.C/.)
Also, show that at a point where r=a its values is a/( 1 + /i ).

Find the radius of curvature

2.

ing curves

at

the point

(r, 0)

on each of the follow-

Kl+cos
3.

Find the radius of curvature of the curve r=a sin n$

4.

Prove that for the cardioide r=a(l +cos

at the origin.

(Allahabad)
2

0), p /r is

constant.

(P.U. 1954
5. Find the
following curves

radius of curvature at

the

point (p,

r)

/>.(/.

1955)

on each of the

6.

Find the radius of curvature for the ellipse

7.

Find the

radii of curvature of the curves

(/)

p=a sin # cos

(//)

pwA'm+1)

^.

-a

/(+)

sin

8. Show that at the points in which the Archimedean spiral r=a$ interthe hyperbolical spiral rQa, their curvatures are in the ratio 3 : 1

CURVATURE
9.

If Pi> P2

be the

303

of curvature at the extremities of any chord of

radii

the cardioide

r=a(l-fcos

0),

which passes through the pole, then


10.
Find the radius of curvature to the curve
point where the tangent is parallel to the initial line.

r=a

(1

+cos

0) at the

line is drawn
through the origin meeting the cardioid
a*id the normals atP,
cos 0) in the points P,
meet in O; show
are proportional to PC and QC.
that the radii of curvature at P and
11.

r==fl(l

Find the points of

12.

inflexion

(0 r-aJVW'-l)-

Show

13.

(11)

on the curves
r

0=a2

(See Cor.

that (a, 0), in polar co-ordinates,

is

15'46, p. 299)

a point of inflexion on

thecurver=0e0/(l-{-0).

Show

14.
lies

r=aQ n has

that the curve

and

between

fi

15.

Show

16.

Prove that

points of inflexion if and only

if,

and they are given by 0=^[/i(/i-f 1)].

that the curve re


for

=a (1+0)

has no point of inflexion.

any curve.
r/p=sin 9[l+<fr/</0],

where

p is

the radius of curvature and tan <p~rd$ldr.

(PU.1951\
If

17.

the equation to a curve


e be given in polars
pola r=/(0) and

prove that the curvature

is

Deduce or otherwise prove

The curve r=ae"

Pn

_i_

mn

cuts
If Pn

log

is

given by

any radius vector in the consecutive


denotes the radius of curvature at P n ,

j*.
pn

constant for all integral values of m and


19.

u*=

sin ' ? -

that the curvature

points PI, Pi, .............. ,


prove that

is

if

given by

($F+ U )

18.

Delhi 1948)

n.

(Delhi Hens. 1947)

Prove that in the curve


r

-a f sin

20.

the tangent turns three times as fast as the radius vector and that the curvature
varies as the radius vector.

(Delhi, 1949)

Centre of curvature for any point P of a curve


on the positive direction of the normal at P and
which
lies
point
15-5.

distance,

p,

The

from

is
is

the
at

it.

distance,

p,

must be taken with a proper sign so that the


pn the positive or negative direction of the

normal of /curvature

lies

centre according as,

p, is

positive or negative.

DIFFERENTIAL CALCULUS

304

thepositive direction of the normal is obtained by rotating


direction of the tangent
of
the
direction
(the
positive
tangant
positive
to y~f(x) is the one in which x increases) through ir/2 in the anti-

The

clockwise direction.

Fig. 114

Fig. 113

Fig. 116

Fig. 115

From an examination of the figures above, we see that thecentre of curvature at any point of a curve lies on the side towards which*
the concavity is turned.
15-51.

To find the

co-ordinates of the centre of curvature for any

point P(x, y) of the curve y =f(x).

Let the positive direction of the tangent make angle, $ 9 with


X-axis so that the positive direction of the normal makes angle
-j-^/2 with X-axis.
\jf

The equation of the normal

X-x

Y-y

_
""

cos ($ +7T/2)

is

'sin

(4r

+7T/2)

or

X-x
sin

\p

COS

wheje X, Y are the current co-ordinates of any potrit 6nthe


mal and, r is the variable distance, of the variable pbftit (X, Y>

fir

CURVATURE

305

Thus the co-ordinates (X, Y) of the point on the


distance, r from (x, y) are
(x

normal at a

r sin ^, y-\-r cos


^),

For the centre of curvature,

if

Hence,

(X, Y) be the centre of curvature,

=y+p
But,

we know

sin ^

we hare

cos w.

that

X=x

cos

^~h ~

Another method.

Y=v4-

be the centre of curvature for P,

If

we

have
'

nrn

'

/ft

* TD/1f

Also,

=x

p sin

^LP+POcos^
=}>+P

cos

Fig, 117

</r.

Substituting the value of sin ^, cos ^ and


and Y.
values of

curve

p,

we can

obtain the

Evolute.
The locus of the centres of curvature of a
15*52.
is called its evolute and a curve is said to be an involute of its

evolute.

15-53.
circle

The

whose centre

of curvature of any point


of a curve is the
at the centre of curvature G and whose radiu

circle
is

The circle of curvatura Will Clearly touch the curve at


Curvature will be the same #3 thdfr of the curve.

Pand

its

15 54. Chord of curvature drawn in a given direction for any


a curve is the chord 6f fche circle of curvature through tu ~
point of
point drawn in the gaid direction. ^
.

306

DIFFERENTIAL CALCULUS

We will now

determine expressions fpr the lengths of someimportant particular cases of the chords of curvature.

Let

(i)

PT

be the

tangent

and C

thfr

centre of curvature for

We draw
PC

any point -Pof a curve.


the circle with C as centre and

as radius.

PI?
to A'-axis

and PQ are chords of this


and Y-axis respectively.

Clearly, /.

RSP= $

==--

/_

SPQ

Thus, the chord of curvature

=2p

PQ=PS cos

Thus

the chord

of curvature

||

\\

sin

t//.

X-axis

i/>.

cos

<Jr.

Y-axis =2/ cos

i/>.

the pole. PT is the chord


of curvature through the

is

(ft")

of the
pole O.

Z_S?# = 2p

sin

parallel

PR=PS sm_RPS=2p

Now,

Also,

circle

circle

PS is the chord of the circle of


curvature perpendicular to the radius
vector OP,
Clearly

TQP=<l>=:^SPQ.
sin /_TQP=:2f

PT=PQ

/.

Thus,

f/ie

the pole=:2p sin


Also,

sin

<f>.

chord of curvature through


Fig. 119

<f>.

PS=PQ cos /_SPQ2p cos

<?>

Thus, the chord of curvature J_ wrf/wj vec/or=2 p co^

i//.

Examples
1.

(*>

J')

JTwrf /te co-ordinates of the centres


of curvature at any point'
pcrabola y*=*ax. Hence obtain its evolute.

Differentiating,

we

get

d*y

Jf

(X Y) be the
y

2a

centre of curvature

CURVATURE

307

(0

4*2
14-

Y=

4a*

UOX}t

To

= ^^~.

-..(>

we have to eliminate x from

find the evolute,

(/)

and

(//).

Thus,
a

27^r2

or
is

the required evolute.


2.

.F/n J //i^ evolute

^_^ w

We

of the four cusped hypocycloid

^,

v,

,,.

^ +j;3=a?.

..^.,

can easily show that

dy

tan

d 2y

1
-.-=-

cosec

sec*

9.

-.

sec* g coseo

sin 2

cos g,

----- 4 -

=a sin3 0+3a
To

eliminate

0,

we

-----

sec 4

cosec

cos 2

sin

...

(/)

...

(ii)

0.

separately add and subtract

There-

(/), (//).

fore

X+ Y=a (cos 0+sin 0)


X-Y=a (cos 0-sin 0)
On

or

(X+ 7)*=a* (cos

or

(^-7)'=^*

(cos

^-f sin

0).

0-sin

0)

squaring and adding, we obtain

as the required evolute.


3.

y=a log sec


constant
length.
of

In the curve

to Y-axis is

(x/tf),

the chord

of cuivature parallel

308

DIFFERENTIAL CALCULUS
TT

d)>

Here

tan ^^-r-

AI

^0=

Also,9

dx*

x
=tan

sec*

.'.

/.

is

= a sec

x
a

chord of curvature parallel to F-axis

=2p
which

(14. tan 8 xla)*


'
sec x/a

p=a ------- -29

^=

X
cos

cos i/>=2a sec

= 2a,

constant*
Exercises

Show

tint the evolute of the ellipse


$9

y=b sin

18

2.

(a)

Show

that for the hyperbola

ad the equation of the evolute is


(axfi
(6)

-(byfi -(

Prove that the evolute of the hyperbola

3.

Shaw

that the evolute of the tractrix

xa (cos
it

-Hog tan

J/),

y=

sin

the catenary

>>=0cosh

(xla).

IProve that the centres of curvature at points of a cycloid lie on an


(P.U. Supp. 1944)
equal cycloid.
4.

Show that (210/16, 210 (16) is the centre of curvature for the point
5.
(3a/2, 3n/2) of the Folium x*+y*=laxy.
6. Show that the centre of curvature of the point P(a, a) of
xi +y*=2cPxy divides the line OP in the ratio 6:1; O being the

co-ordinates.
7.

Show

that the parabolas

j^-aHx+l. *=-;K2 -f.y-f


have the same circle of curvature at thQ point
8.

is

ftt

Show

that

the circle of curvature of the curve

the point (a/4, a/4).

(1, 1).

the curve
origin of

OTJKVATURB
9.

Find the

309

circle of curvature at the origin for the curve

x+ y=ax*+by*+c**.
10.

Show

(Delhi Hons. 1951)

that the circle of curvature at the origin of the parabola

is

mx ).

(D.U. 1955)

2
(a) Show that the circle of curvature, at the point (am 2am) of the
parabola y*=4ax has for its equation
x*+y*-6am2x-4ax+4am*y=3a*m*.
(D.U. Hons. 1957)

11.

(b)

the ellipse
12.

Find the equation of the


*
2

Show

+ ^=

the point (0, b) of

(P.U. Hons. 1959)

that the radii of curvature of the curve

x==ae^ (sin

and

circle of curvature at

cos^),^=a^

(sin

0+cos

0),

evolute at corresponding points are equal.


13.
Find the radius of curvature at any point P of y=c cosh (x]c) and
show that PC=PG where C is the centre of curvature at P and G the point of
intersection of the normal at P with x-axis.
(Allahabad)
its

14.

Show

spiral

spiral

that the chord of curvature through the pole of the equiangular

is 2r.
r=ae
15. Show that

r=ae
16.

bisected at the pole.

is

If cx

the chord of curvature through the pole of the equiangular

and cv be the chords of curvature


x a

point of the curve

y=ae

parallel to

prove that

17.

If Cx

and

cv

point of the catenary

18.
is

Show

the axes at any

{P

'

'

1948)

be the chords of curvature parallel to the axes at any


cosh (xlci, prove that

y=c

that the chord of curvature through the pole

r=0(l

Jr.

of the cardioide

cos 0).

and c * be the chords of curvature of the cardioide r=o(l 4-cos


through the pole and perpendicular to the radius vector, then
19.

Lf c r

20.

Show

0)

that the chord of curvature through the pole of the curve


r

m_ m C08 m Q
fl

is

2r/(ro4-l).

21.

Show

(Guj rat 1952)

that the chord of curvature through the pole for the curve

is

2/(r)//'(r).

(Lucknow)

br the chord of
curvature through the
22. Show that for the curve p~ae
,
pole is of constant length.
For the lemniscate r 2 =fl2 cos 20, show that the length of the tangent
23.
the
from
(B.U.)
origin to the circle of curvature at any point is r^3/ 3.

310

DIFFERENTIAL CALCULUS

24. If P is any point on the curve r 2 =a2 cos 29 and Q is the intersection
of the normal at P with the line through O at right angles to the radius vector
OP, prove that the centre of curvature corresponding to P is a point of trisection

ofPQ.

(L.U.)

Pis any point on the curve r=a (1-l-cos 9) and Q is the intersection
of the normal at P with the line through the pole O perpendicular to OP, prove
that the centre of curvature at P is a paint of trisection of PQ remote from P.
25. If

26. 1 he circle of curvature at any point


meets the radius vector OP at A, show that

OP: AP=l

P of the

Lemniscate

r2

=a2 cos

29

2;

O being the pole.


27.
p lf p, are the radii of curvature at the corresponding points of a
f
8
cycloid and its evolute ; prove that Pi +p t is a constant.
28. Show that the chord of curvature through the focus of a parabola is
four times the focal distance of the point and the chord of curvature parallel to
the axis has the same length.
(Rajputana 1952)

29. Prove that the distance between the pole and the centre of curvature
n
w
is
corresponding to any point on the curve r =a cos

15-55.

Two

important properties of the evolute.

If (X, Y) be the centre of curvature for


any point P(x, y) on
the given curve, we have

X=x
Differentiating w.r. to

dX

p sin0,
9

ds- dx

-=-

dif>

dY

cos

X we obtain

=1

dx

Y=y+p

dib
y--

ds dx

sin

rfp

-=-.
* dx

dy

^x
= dy

ds dy

dip,

cos
Jx-d* dsdx-+

dp
dx,

rfp

-COS ^
fc

From

(i)

and

...

(//)

(//),

Now, dY/dX is the slope of the tangent to the evolute at pf


cot </>, is the slope of the normal PP' in the
and,
original curve at
P. By (//) the slopes of two lines, which have a point P' in
common,
are equal, and therefore they coincide.

CURVATURE

311

Thus, the normals to a curve are the tangents

we square

Again,

one

(/)

and

(//)

and add.

Let, S, be the length of the arc of the


fixed point on it upto (X, Y) so that

Here, x

is

to its evolute.

evolute measured

from

a parameter for the evolute.


...(/*)

dx)

dS
dx

dx'~

or

where,

c, is

constant.

on the
,
2 be the two points
evolute corresponding to the points L 1?
L2 on the original curve. Let p lf P 2 be
the values of, p, for L l9 L 2 and S 19 S 2 be
Afa
the values of 8 for
19
Let Af

Thus

i.e.,

arc A/,

M = difference
2

the radii of curvatures at

L L9 L 2

between

Fig. 120

we have shown

that difference between the radii of


Thus,
curvatures at two points of a curve is equal to the length of the arc of the
evolute between the two corresponding points.
Note. We suppose that 5 is measured positively in the direction of x, increasing so that dSldx is positive. Also, dpfdx is positive or negative according
as, p, increases or decreases as x increases.

Thus

.-= T

ax
according as,

p,

dx

or

dx

increases or decreases for the values of, x, under consideration.

easy to see that the conclusion arrived at in this section remains the
same if we consider dSldx^dpfdx instead ofdS'dx^dpldx. It should, however,
be noted that the conclusion holds good only for that part of the curve for which
p , constantly increases or decreases so that d?/dx keeps the same sign.
Tt is

312

DIFFERENTIAL CALCULUS
Ex.

which

is

Find the length of the arc of the evolute of the parabola

intercepted between the parabola.

The evolute is
Let L,

27a^=4(x-2a)
be

(Ex.

1,

the points of intersection of the evolute

p.

30)

LAM

and the parabola.

To find the co-ordinates of the points


of intersection L, M, we solve the two
equations simultaneously.

We get
07/7

*i I C*

A
/ry
^IM^V

--

---- ft/r\*^
A.("Y
-*1
*\^^
AC*y y

a are the roots of


x=Sa is the
of
which
equation
on ly admissible value
a being negative.

Now,

8a,

fl,

this cubic
*

Fie 121
/.

(84','4\/2a), (8a,

4^20)

are the co-ordinates of L,

If(Xt Y) be the centre of curvature for any point


parabola,/we have

X^3x+2a,

4\/2a)

is

(x, y)

(Ex.

I4:a*.

the centre of curvature for


f
the centre of curvature for P(2a,

/Thus A(2a, 0)
L(8<t,

Y=y

is

M.

1, p.

arc

Hence the required length Af4L=4a(3 v'3


Ex.

2.

Show

Ex.

3.

Show

that the

that the

x~
it

120.

1).

whole length of the evolute of the

306)

O(0, 0) and

The radius of curvature at/? = 0.4=20.


The radius of curvature at P=PL=(
/;

on the

ellipse

whole length of the evolute of the astroid

CHAPTER XVI

ASYMPTOTES
16-1.

an

Definition.

straight line

said to be an Asymptote of
recedes to infinity along
from the straight line-

is

P
P

branch of a curve, if, as the point


the branch, the perpendicular distance of
infinite

tends to 0.
Illustration.

The

line

#=a

is

an asymptote of the Cissoid

(Seell-41>

y\a-x)=x*.

moves to

It is easy to see that as


(x, y)
line x--a tends to zero.

infinity,

its

distance

from the

Ex.

What

arc the asymptotes of the curves


j>=tan x ; ,y=cot x ; .y=sec x and >>=cosec x.

We know that the equation


16-2. Determination of Asymptotes.
of a line which is not parallel to X-axis is of the form

ymx+c.
The

...(/>

must tend to

abscissa, x,

infinity as the point P(x, y}

recedes to infinity along this line.

We shall now determine, m, and,


an asymptote of the given curve.

c,

so that the line

(i)

may

M
X
Fig. 123

Fig. 122

Ifp=MPbethe perpendicular distance of any point P(x,


on the infinite branch of a given curve from the line (/), we have
v mx c
p ->

as

x ->

lim(>>~ mx

oo

c)=0,

which means that, when x ->oo

313

be

314

DIFFEBENTIAL CALCULUS
Also,

lim (yl*

w)=lim

mx). lim l/x=c.

(y

0=0

or

lim (yly)~m,

-when

x ->

oo.

Hence

m=

lim (y/x),

X ->

c=

lim (y

X ->

oo

mx),

oo

We have thus the following method to determine asymptotes


which are not parallel to j>-axis :
Jet lim (y/x)~m.
(/) Find lim (yjx)
X - oo
x - oo
;

(11)

Find lim (ymx)


X - oo

Then

y=mx+c is an

(ymx)^c.

lim

let

X ->

oo

asymptote

The values of y will be different according to the different


branches along which -Precedes to infinity, and so we expect several
values of lim (yjx) corresponding to the several values of y and also
Thus a curve may have
several corresponding values of lim (y mx).
more than one asymptote.
This method will determine all the asymptotes except those
parallel to 7-axis. To determine such asymptotes, we
start with the equation x^my+d which can represent every straight
line not parallel to A'-axis and show, that when y -^ oo

whiah are

m^lim
The asymptotes not

(xjy)

and d~]im (xmy).

parallel to

any

axis can be obtained

way.
Ex.

Examine

1.

the Folium

for asymptotes.

The given equation is of the third degree.


To find lim (y/x), divide the equation (/) by X s

xx

x J
Let

Jc

oo.

We then get

l-[-m

=Oor

(m+l)(/n

The roots of w 2 -- m + 1 =0 are not

m+l)~Q.
real.

so that

either

ASYMPTOTES

To
put

mx) when

find lim (y
so that,

y+xp

Putting

/?, is

for

m=

315

l,

i.e.,

to find lim (y+x),

a variable which -> c when x ->

in the equation

(i),

we

oo

we

get

or

which

is

of the second degree in x.

2
Dividing by x

Let # -*

oo

we

get

We then have

3(c+a)=0

or

c=

a.

Hence

>>=
is

the only asymptote of the given curve.


If

is

a or .*+}> 4-0=0,

we

start

with x=my-\-d, we get no new asymptotes.

the only asymptote of the given curve.


Ex. 2. Find the asymptotes of the following curves

Thus

(i)
(ill)

16*3.

Working

rules

for

determining

asymptotes.

Shorter

In practice the rules obtained below for determining


-asymptotes are found more convenient than the method which involves direct determination of
Methods.

lim (yjx) and lim

(ymx).

Firstly we shall consider the case of asymptotes parallel to the


co-ordinate axes and then that of oblique asymptotes.
16-31.

Determination of the asymptotes parallel to the co-ordi-

nate axes.

Asymptotes parallel to Y-axis.

Let

x^k

..(t)

be an asymptote of the curve, so that we have to determine k.


Here, y, alone tends to infinity as a point P (x, y) recedes to
infinity along the curve.

816

DIFFERENTIAL CALCULUS

The
line

(t) is

PM of

distance
equal to x

any point P(x,

on the curve from

y)

the-

k.

fc)=0 whenj>

oo

lim #=fc when y -

oo

lim (x

or
,

which gives fc.


Thus to find the asymptotes parallel

we find

Y-axis,
fc

2,

etc., to

the

which

x=k2

Then x=fc,,

to*

definite value or values fcr


x tends as y tends to oo .
,

etc. are

the required

asymptotes.

We will now obtain a simple rule to


obtain the asymptotes of a rational Algebraic

Fig. 124

curve which are parallel to F-axis.

We
y, so that

arrange the equation of the curve in descending powers of


it takes the form

where
are polynomials in x.

Dividing the equation

Let y ->

oo

We

(i)

by y

so that,

fc,

Let

is

fe

lf

(ii)

get

write

lim

The equation

m we

x=k*

gives

a root of the equation <(x)=0.


fc

etc.,

be the roots of

<j)(x)

=0.

Then the asymptotes?

parallel to 7-axis are

x~k
From

l9

we know

x=fc 2

etc.

that (x

(x fc 2 ),
fc,),
algebra,
factors of ^(x) which is the co-efficient of the highest
in the given equation.

etc.,

are the

power y

m of
y

The asymptotes parallel to Y-axis are


Hence we have the rule
obtained by equating to zero the real linear factors in the co-efficient of
the highest power of, y, in the equation of the curve.
The curve will have no asymptote parallel to F-axis, if the
co-efficient of the highest power of, y, is a constant or if its linear
factors are all imaginary.
:

Asymptotes parallel to X-axis. As above it can be shown that


the asymptotes, which are parallel to X-axis, are obtained by equating
to zero the real linear factors in the co-efficient of the highest power of>
x, in the equation of the curve.

317

ASYMPTOTES
To determine

16-32.

the

asymptotes

the general rational

of

algebraic equation

Ur is a homogeneous expression
We writo Ur in the form

where,

where

<f>,(ylx)

is

a polynomial in y\x of degree,

So we write

x
"+-("x

of degree,

(i)

r,

r,

in x, y.

at most,

in the form

)+*""*-' ()+*"<*

(i )+

Dividing by x*, we get

On

taking limits, as x-^oo

we obtain the equation

which determines the slopes of the asymptotes.


Let

We

be one of the roots of this equation so that

write

Substituting this value of yjx in ('),

we

get

Expanding each term by Taylor's theorem, wo get

..(wJ+^-f ..,(111,)+.,

DIFFERENTIAL CALCULUS

318

Arranging terms according to descending powers of x, we get

1
Putting ^ n (m,)=0 and then dividing by x"- , we get

Let x -

oo

We write lira />!=!.

-K

=0

Therefore

or

Therefore

is

the asymptote corresponding to the slope

A,

if

^(

Similarly

are the asymptotes of the curve corresponding to the slopes


2 m#
which are the roots of ^ n (m)=0, if <f>' n (m 2 ), ^' n (w 3 ), etc., are,

etc.,

not

0.

Exceptional case.

Let

^(m,) =0.

If ^' n (m,)=0 but ^-1(^1)^0, then the equation (vi) does not
determine any value of c l and, therefore, tho/e is no asymptote corresponding to the slope m v

Now

suppose

#'n('i)=-0=^ _ 1 (m 1 ).
In this case, (vi) becomes an identity and we have to
examine the equation (v) which now becomes
fl

On
tion

taking limits, as x ->

oo

we

see that

is

a root of the equa-

'

which

dejf'ermines

two values of c,, say

c,', c,",

re-

provided that

ASYMPTOTES

3 1 fr*

Thus

y=m,x+ c/, yare t he two asymptotes corresponding to the slope

These

are-

clearly parallel.

This

is

known

as the case of parallel asymptotes.

The polynomial J n(m)

Important Note.

x=l andy=m

is

obtained by putting

xn n (ylx), and n -i(tn),


n
^ n - 1 (^/x), x ~^ n ^(yjx), etc., in a'

the highest degree terms


n^1

in

^n- 2 ( m ) etc -> are obtained from x

<f>

<f>

similar way.

Fxunples
Find the asymptotes parallel

1.

curves

(x +>>*)* --tfy*-=0.
(//)
(i) The co-efficient of the highest
the asymptote parallel to y axis is
(i)

of

jcy-d*(jt+y)=<).
power y of y is x a.

ther

0.

The co-efficient of the highest power x3 of x is 1 which


Hence there is no asymptote parallel to x -ax is.
(//)

The

the highest power y^ of y

co-efficient of

Hence

xa

stant.

to co-ordinate axes,

is

x2

a con-

is

a2

Also,

rfence

x
are the

two asymptotes

a=Q,

x+a~

parallel to }>-axis.

It may similarly be shown that>> #~0, y 4-^=0 are the twoasymptotes parallel to x-axis.
2. Find the asymptotes of the cubic curve
2x 3 x 2^ +2xy* +y* 4x 2 +8xy 4x + 1 =0.

Putting x=l,
separately,

we
<f>

The

y=m in

the third degree and second degree terras

get

2m 2 +m 3

(m)=2m

slopes of the asymptotes are given

^3 (w)=w

by

2m2

or

111=

Again,

cf

c. is

-1,

1,

given by

4i+3/ I )+(--4+8m)~0,

320

DIFFERENTIAL CALCULUS

m=

1, 1, 2, we get c=2,
Putting
Therefore the asymptotes are

3.

4 respectively.

2,

Find the asymptotes of

x2

4y*
l

m (l
2

m)

The slopes of the asymptotes, given by


To determine c, we have

<

m)

(w)=0, are

1.

1, 1,

i.e.,

c(~l-~2m+3m 2)+(2+2m-4w 2 )==0.


For w=s 1, this gives c=l and therefore, j>=

...(0

x+1

the

is

^corresponding asymptote.

For

the equation (/) becomes


In this exceptional case,

w=l,

cally true.

0.

c+0=0

c,

is

which is identidetermined from the

equation

i.e.,
2

(c

For

/w

/2)(-2+6m)+c(2-8m)+l(l+m)=0.

= l, this becomes
2c'~6c?+2=0,

i.e.,

c=(3v^5)/2.

Hence y~x+(3<\/5)l2 are the two


responding to the slope

We have

parallel

cor-

asymptotes

1.

thus obtained

all

the asymptotes of the curve.

Exercises
Find the asymptotes parallel to co-ordinate axes of the following curves
1.

ytx-a^x-a^O.

2.

x*y~-3x*-5xy+6y+2=Q.

4.
a2 /*2
y=xl(x*-l).
Find the asymptotes of the following curves

3.

+W=i.

5.

x(y-x) =x(y-x)+2.

6.

x\x-y)*-\-o?(x*-y*)=o?xy.

1.

(x~-y)*(x*+y*)-W(x-y)x*+l2y*+2x+y~0.
x 2y+xy*+xy+y*+3x=Q.

8.

(D.U. 1952)
(P.U.)

(P.U.)
'

10.

(x-y+l)(x-y-2)(x+y)=Sx-l.
y*-x*y+2xy*-y+l=Q.

11.

ylx-y) =x+y.

12.

^V^

9.

-/) 2

(P.U.)

(P.U.)
(B.U.)

ASYMPTOTES
2

321

~* 2 -l.

13.

Xj>-l)

14

xy***(x+y)*.

15.

(x+y)(x-y)(2x-y}-4x(x-2!*)+4x

16.

xy*-x*y-3x*-2xy

17.

2*(>'-3)

18.

(y~a) (x -a*)=x'-\

19.

U-f/r(x

20.

^SyV-x'^-SxM^-ZJOH- 3* 2 4>-f5-0.
2
3
3
(A:-j) (.v--2v)(.v-3r)---2a(A' -v )--2aV-2v)Ul>')^-0,

21.

(L.I7.)

(L

-3X*-l)
2
-t-jty -f-^

x~1y-\

U)

(L.U.)

0.

(L.U.)

y*

0.

(^ra)

(Lucknow}

a*.

)-o 2 JC 2

a*:y~x).

(DehiHons. 1950)
22.

yW

Show

/(fl

-* 2

(P.U. 1955 Supp.)

).

that the following curves have

no asymptotes

23.

jc^f^---^^-^).

24.

25.

aV 2 -.v\2a-JC).

26.

y ---x(xfl)

27.
Find the equation of the tangent to the curve
parallel to its asymptote.
28.

An asymptote

is

x3

2
.

2
3
^ --3ax which

is

sometimes defined as a straight line which cuts the


Criticise this definition and replace it by a

curve in two points at infinity.

(P.U. 1955 Supp.)

correct definition.

16 33.
(/)

Some

deductions from

16 32.

The number of asymptotes of an algebraic curve of the nth

degree cannot exceed

n.

The slopes of the asymptotes which are not parallel to F-axis


are given as the roots of the equation n (m)~ which is of degree n
at the most.
<f>

In casa the curve possesses one or more asymptotes parallel to


is easy to see
that the degree of ^ n (/w) Q will bo
smaller than n by at least the same number.
K-axis, then it

Hence the

result.

an algebraic curve are parallel to the lines


(//) 77?e asymptotes of
obtained by equating to zero the factors of the highest degree terms in its
equation.

Let, m, be a root of the equation


^1^=0 is parallel' to an asymptote.

^ n (^)--0,

so that

the line

n (yl*)
elementary algebra, (y/xrn^ is a factor of
n
converse!
t/
Also
a
x
is
of
factor
n
n
y,
i.e.,
hence, yr^x,
(ylx),
we see that if, y m x x, is a factor of U n then m v is a root of

By

<f>

<f>

^,(wt)

= 0.

factor, then a
consideration will show that the curve will possess asymptotes
parallel to x=0, i.e., to .y-axis.

In case the highest degree terms contain, x, as a

little

DIFFERENTIAL OALOQLUS

322
(iit)

Case of

In this

fore,

case,

satisfies

iy

the three equations

Since ^ n (m) and its derivative ^' n (m) vanish for


by elementary 2algebra, m l is a double root of

therefore,

m^

a root of

is

l)th degree terras

m=m

Un _ v

<f>n

_ 1 (m)=0,y

there-

lt

<^ n (m)~0

a factor of the highest degree terms

is

/w^)

(>>

Also, since
(AI

parallel asymptotes.

f/ n

and

a factor of the

is

Thus, we see that in the exceptional case of


16*32, a twice
repeated linear factor of U n is also a non-repeated factor of U n _ r

There

be no asymptote with slope m l9


butw>* of ^ n -i(fH)--0, i.e.,
and >> fl^x is not a factor of Un ~i-

will

if

^ n (m)=0, <' n (w)=0


factor of C/ n

w is a root of
(ym^) is a
1

if

The

results obtained in the paragraphs (11) and (///) above enable


the process of determining the asymptotes as shown in the
following examples.

Note.

us to shorten

The first step mil always consist in fact or i sing the expression
highest degree terms in the given equation.

formed of

the

Examples
1.

Find the asymptotes of the Folium

The curve has no asymptotes

parallel to co-ordinate axe*.

Factorizing the highest degree terms,

so that,
terms.

y+x,

is

the only real linear factor of the highest degree

Hence the curve has only one


whose slope is

to the line y-\~x~

We have, now
We have

In the limit,

real

we have

xa

asymptote which

is

parallel

1.

to find, lim (y+x),

y=
is

we get

i.e.,

tha only real asymptote of the folium.

when x ->

<x>

and yjx ->

1.

ASYMPTOTES

323

It is easy to see that we could have eliminated the step


simplified the process by saying that the required asymptote
v 4- x =-- Km
lim
y+X

and

(i),

is

IB

when x ->

oo

and yjx ->

1.

Find the asymptotes of

2.

Equating to zero the

we

flO.y

-^ f

co-efficient of the highest

-0.

power y* of

y,

see that

4xH-10=0,
is

+ 15xy

-f 5x

i.e.,

2x+5--=0,

one asymptote.
Factorising the highest degree terms,

we

get

Here 2y +x is a repeated linear factor of highest degree terms,


3rd
There will, therefore, be no asymptote parallel to
i.e.,
degree.
2y+JC~0 if (2y+x) is not a factor of the 2nd degree terms also.
But this is not the case. In fact, the equation is

+5(x +y)(x +2y) - 2y + 1 - 0.


curve has two asymptotes parallel to

x(2y
Therefore, the

-f-*)

We have now to find lim (y+\x) when x ->


Let lim (y -f Jx)=--c so that lim (2^+x) 2c.
Dividing by x, the equation becomes

x and

yjx ->

In the limit,

4cM-5.2c(l-i)-2(-i)+0=0
4c 2 +5c + l=-0

or
-

r
C

Hence
are the two
It

is

ified
simplified

y=

5,

...(0

>

i
1.

J^-~J and.y=

Jx

1,

more asymptotes.

easy to see that we could have eliminated the step


that the asymptotes are
the process by saying
sayi

(i)

and

(2y+xy+5(2y+x). lim (l+y/x)+ lim


i.e.,

(ty+x)*+5(2y+x).

+1-0

or 2(2>> fx) 2 +5(2>>+*)+2-0,

which gives

=0

and

4>>-f 2x

+ l=0.

324

DIFFERENTIAL CALCULUS
Find the asymptotes of

3.

The asymptotes parallel to the two imaginary lines


are imaginary. To obtain the two asymptotes parallel to the lines
x y~Q, we re-write the equation, on dividing it by (x 2 +^ 2 ), an
--

1+0

t+--

We take the limits when x ->


asymptotes are

i.e..

(x-,y)

and yjx

oo

--> 1

Therefore the

2=0, x

.v

i.e.,

/jr)

3>-3-=0.

/7m/ /Ae asymptotes of

4.

(x-y +2)(2jt- 3j +4)(4jc - 5y +6) +5* ~fi v -4-7=0.


The asymptote parallel to x--jH-2=0 is
-

.*

when

.v

->

x and
,

v/x ->
i

A'--v4-2fhm

5-6^/x-f7/x
---'

"1

-v

=0.

'

---,;

L^-^/^+^WC

4 "" 5

,,,,,.

^-!-^)

x~^+2-0,

or

as the limit

is

zero because of the factor 1/x.

we can show that

Similarly,

3j>+4=0, 4x- -5y -|-6-^0

2x

are also the asymptotes of the curve.


16-4.

Asymptotes by Inspection.

If the equation

of a curve of the

nth degree can be put in the form

Fn _ a is of degree (w
when equated to zero

2) a^ ///^ wort, //^ ev^r.y /me^r /ac/or of


will give an asymptote, provided that no
equating to zero any other linear factor of n is

where

Fn

straight line obtained by


parallel to it or coincident with

Let

where

ax+by+c=Q

Fn _ t

ax +by 4-0=0

is
is

of

it.

be a non-repeated factor of

degree

(w

1).

Fn

The asymptote

We

\*rite

parallel

to

32&

ASYMPTOTES

p
ax+by+c+lim /~^0,
fn-i
when x

->

and y/x ->

oo

a\b.

limit (/ rn-a/' rn-i)> w ^ divide the numerator as


well as the denominator by x n ~ l and see that 1/x appears as a factor

To determine the

so that jPn _ 2 /Fn ^


1

>

x ->

a*

Thus
is

on asymptote.
Exercises
Find the asymptotes of the following curves

2.

(x--i)(;c-2)(;c-Kx)+*

j-x+1 -0.

-*M .y2 + X + y- X +\^Q.

3.

4.

x(y ~lby+2b*)^y*-3bx*+b*.

5.

^ 3 f6.v 2 ^MlA:/--h6^4-3^4-12x^f 1 ly^-ZxfSy f-5-0.

6.

.x

2 2

+ ^-2x 2 )(2>'

7.

(y -HA7^2x

8.

x(y-3)*^.4y(x~\)*.

9.

(a-}-x)

3 ><

16

5.

(/?

5A-y

-}-(y

-}A'

)--xV

-jc)-7/

-l9xy-23x

-hx f 2/

3-^>.

2
.

-h8^^4jc -3/-f 9A7

-6A' 2

Intersection of a curve and

Any asymptote of a curve of the

its

i2^

2x f

-0.

asymptotes.

nth degree cuts the curve

in (n

2)

points.

Let y =niK-\-c be an asymptote of the curve

To

find the points of intersection,

we have

to

solve

the two

equations simultaneously.

The

abscissae of the points of intersection are the

roots

of the

equation

xn

<l>

n (m

>

(3>'^,v)MM3vfAr)(xHr)-h9^-|-6A7f9y^6x+9-0.

10.

(7\l/.

+clx)

+x

^n-i(m +c/x) +x*-*<f> n - 2 (m +cjx)

+ ...... -0.

by Taylor's theorem
to
according
descending powers of x, we get
Expanding each term

...(/)

and arranging

326

DIFFERENTIAL CALCULUS

Aay^mx+c

is

an asymptote, the

co-efficients

of x n and x*~ l

are both zero.

Thus the equation

(11)

reduces to that "of (n 2)th degree


Hence the result.

and,

therefore, determines (ttr-2) values of x.

Cor,
in

1.

n(n2)

The,

/i,

asymptotes of a

curv^e of the

nth degree cut

it

points.

If (he equation of a curve of the nth degree can be put in


where Fn _ 2 is of degree (n 2) at the most and
n consists of n, non-repeated linear factors, then the n(n2) points of
intersection of the curve and its asymptotes lie on the curve

Cor. 2.

the form

Fn +Fn _ 2 =0
9

result follows at once from the fact that Fn =0 is the joint


At the points of intersection of
of
the, n, asymptotes.
equation
Fn =0 and
the curve and its asymptotes, the two
equations
Fn n _ 2 ~Q hold simultaneously and therefore at such points we have

The

+F

*-=<>.
Particular cases
3, and therefore the asymptotes cut the
(i) For a cubic, n
curve in 3(3 2) =3 points which lie on a curve of degree
1,
i.e., on a straight line.

32 =

asymptotes cut the


(11) For a quart ic, n=4, and, therefore the
curve in 4(4 2)~8 points which lie on a curve of degree 4 2=2
i.e., on a conic.

Examples
1.

Find the asymptotes of the curve

x*yxjP+xy+yP+x-y=Q
and show

(P.U. 1955)

that they cut the curve again in three points

which

lie

on

the

line

x+y-^0.
The asymptotes of the given curve, as may be
are

The
i.e.,

joint equation of the asymptotes is

x 2y-xy*-\-xy+y*-2y=Q.

The equation of the curve can be written

as

Here

F^tfy
Henee the points of

intersection lie

on the

line

(P.U. 1940)
easily

shown

ASYMPTOTES

Show

2.

that the asymptotes

327

of the quartic
2

cut the curve in the eight points which

The asymptotes of the curve

lie

on a

=0,

circle.

are

x+2y=0,x-2 y+l=0, x-3>'=0,


i

so that their joint equations

is

.e.,

The equation of the curve can be written


X
X
5V 2^^
( 2_4^2)( 2_9^2)_|_
Hence the points of intersection

3,

lie

as

on the

circle

Find the equation of the cubic which has the same asymptotes

as the curve

and which passes through

the points (0, 0), (/, 0) and(Q,

I).

(Delhi Hons. 1948)

We

write

Fjssx

- Gx-y +

1 1

xv 2 - 6y 3

= (x-y)(x-2y)(x-3y).
Fjsx+y+1.
The equation of the curve can be written in the form /rs +jF1 =0
where F% has non -repeated linear factors. Thus 7^=0 is the joint
equation of the asymptotes of the cubic.
The general equation of the cubic is of the form
or

where ax -\~by-\-c

(0,

is

the general linear expression.

In order that it
we must have
1),

may

pass through the points

c=0,

6+6=0 or
Thus the required cubic

is

(0, 0), (1,

0)

and

DIFFERENTIAL CALCULUS

328

Exercises

Show

1.

that the asymptotes of the cubic


2
xy - 2xy -f 2x - y
points which lie on the line

x9

cut the curve again in

0,

3.v-->>---0.

If a right line is drawn through the point (a, 0) parallel to the asymptote of the cubic (Jc a) 3 x 2y=Q, prove that the portion of the line intercepted
2.

by the axes

is

bisected by the curve.

(C.C/.)

2
2
Through any point P on the hyperbola x / ~ 2ax, a straight line is
drawn parallel to the only asymptote of the curve x 3 -f-.y3 =30;t2 meeting the curve
in A and B
show that Pis the mid-point of AB.

3.

4.

Show

that y ^x + a

is

the only asymptote of the curve

A straight line parallel to the asymptote


that the mid-point of PQ lies on the hyperbola

meets the curve

in P,

show

x(x-y)+ay---0.
5.

Find the equation of the straight

line

on which

lie

three points of

intersection of the cubic


2
8
xV-jn' - 2r

and

its

4r 2

2xy

asymptotes.
6.

Find the asymptotes of the curve


4.x*

~13*V+V

a
3
f-32xV-42.v -20x

i-74>'

-56v-f

4,v f

16

0,

and show that they pass through the intersection of the curve with
(D.U. Hons. 1953)
7.

Find

ail

the asymptotes of the curve

3x 3 -f 2x*y -Ix

Show
on a straight
8.

that the asymptotes


line,

and

meet the curve again

in three

find the equation of this line.

points which lie


(D.U. Hons. 1952)

Find the equation of the cubic which has the same asymptotes as the

curve
x* -6x*y\

\\xy*-6y*i-x+y+\-

0,

and which touches the axis of v at the origin and passes through the point (3, 2).
(Delhi Hons. 1949, 1955)
9.
Find the asymptotes of the curve

4(*H y 1 ) - 17x

2 2
>'

- 4x(4y 2 ~ x*) 4- 2(;c

2)

-,

and show that they pass through the points of intersection of the curve with the
a
(Delhi Hons. 1951, 1959)
ellipse x + 4/-4.
10.

Find the asymptotes of the curve

9-0
and show that they intersect the curve again
straight line. Obtain the equation of the line.
16-6.

y=mx-\
is

To show

Asymptotes by expansion.

which He on a
(D.U. Hons. 1957)

in three points,

that

an asympotate of the curve


(I)

ASYMPTOTES

329

Dividing by x we have
9

y\x=m +C/X+A/X* -\-Blx*+C/x* +


so that

when x

*x>

...

]im(ylx)=m.
Also from (1)
(y

so that

- mx) = c +A/x + B/;t2 -f Cj 8 + .....


jc

when x->x

lini

From
is

(2)

..(2)

we have

and

(ymx)

r.

...(3)

we deduce that

(3),

an asymptote of the curve

(1).

Position of a curve with respect to an asymptote.


the position of the curve
16*7.

with respect to

its

To

find

asymptote

y=mx-\

c.

Let A^O. Let > t and >' 2 denote the ordinates of the curve and
the asymptote corresponding to the same abscissa x.
We have
f

By

taking

.v

+....)
we can rnako

sufficiently large,

We

as small as we like.
suppose that x
this expression is numerically less than A.
values of .r, the expression

A
lias

+ B/.v + C/A"

2
|-

D/.v

..(1)

so large numerically that


Thus, for sufficiently large

is

(2)

the sign of A.

Thus

be pjsitive, the expression (1) is positive for suffivalues


of A* so that, from (1) we deduce that when x is
ciently large
lies
positively sufficiently large, then J^
>' 8 is positive, i.e., the curve
above the asymptote and when x is negative but numerically large,
yiy.2 is negative, i.e., the curve lies below the asymptote.
if

Similarly, we may deduce that if A be negative, then the curve


below the asymptote when, jc, is positive but sufficiently large
and lies above the asymptote when, .v, is negative but sufficiently
lies

large numerically.

Let

A=Q, B^O we have


;

As above we can show that


values of x, the expression

has the sign of B.

for numerically sufficiently

large

330

DIFFERENTIAL CALCULUS

In this case, the curve lies on the same side of the asympotate
both for positive and negative values of x it will be above or below
the asymptote according as B is positive or negative.
If 5=0 and C^O, we will have a situation similar to that of
;

case

(1).

Ex.

Find the asympotates of the curve


'

y*=x(x~-a)(x-2a)l(x +3a),
and determine on which side of the asymptotes the curve
We have

a*

~2x
x ~8&

.....

Y
A

Y
2?-" A
'

a2

a
i

*~

Thus we have two values of y,

y=x3a+-

lies.

3fl
i

2^

+ 27aS
8.v
-u

viz.,

a 2 x ......

y=sx+3a*J-a*x ____
Therefore

y=x3a y=
9

are

the asymptote

we

x+3a

two asymptotes.
The difference between the ordinate of the curve and that of

y=x3a being

see that the curve lies

and below
It

it

when x

is

above the asymptote when x

is

positive

negative.

the second
similarly be seen that the curve lies below
is positive and above it when x is negative.
easy to see that

may

asymptote when x
It

is

*=:

3a

To find the position of


we suppose that
x= -3fl +A/y +Bjy* +C/>* + ......
Substituting this value of x in the equation of the

is also

an asymptote of the curve.

the curve

relative to this asymptote,

curve,

have

-5*+

we

ASYMPTOTES

Equating the

co-efficients

331

of like powers of y, we have

4: = 0.

60a 8

#:=

The

which

is

curve

lies

Ex.

between the abscissae of the curve and the


same value of y, being

difference

x~3tf,

asymptote

etc.

for the

negative whether y be positive or negative, we see that ths


towards the negative side of JC-axis.
Find

2.

following curves

the asymptotes and their position with regard to

(Hi)

16*8.

Asymptotes

Lamma.
any

the

x*(x-y)i

2
>>

-0.

in polar co-ordinates.

The Polar Equation of a

Line.

The polar equation of

line is

pr cos (6-

a),

where, p, is the length of the perpendicular from the pole to the line
angle which this perpendicular makes with the initial line.

and

a, is the

OY

Let
given line

We

be the perpendicular on the

Y being

its foot.

are given that

0y=p;
If

(r, 6)

Z_

be any point on the

line,

we

have

Now,

~p= C oS /_YOP.
Fig. 125.

/>/r=cos(0

which

is

a), i.e.,

p=r

cos (0

a),

the required equation of the line.

To determine

the asymptotes of the curve

..(0

r/(0),

we have

to obtain the constants,

p~r
is

/?,

and,

cos(0

the asymptote of the given curve.

a),

so that any line


..(it)

332

DIFPffiUBNCIAL CALCULUS

Let
(0.

P(r,

be any point on the curve

ff)

Draw OY
Draw PL

J_ the line (n).


and
J_

PM

OY

J_ the line

(ii).

Now.

OY-OL
Fig. 126.

Now
>0 t

=p--r

cos (0-a).

r->oo as the point recedes to infinity along the curve.

when

..(ill)

Let

r->- oo.

We have
Now when

oo,

=~
PM

cos (6

->>

a).

Oso that
l

>and
lim cos (0
or

0.

lim (0

a)=0

a)=7T/2,

or

This gives

a.

Again, p O Y is the polar sub-tangent of the point at which the


asymptote touches the curve, i.e., the point at inanity on the curve. This
may be seen as follows
:

Join the pole O to the point at infinity on the curve i.e., draw
O a line parallel t'> the asymptote. This line is the radius
vector of the point at oo.

through

Draw through O a line perpendicular to the asymptote meetis the polar sub-tangent of the
point
ing it at Y. Then, by def.
at infinity on the curve.
12-7, p. 271).
(

OY

Thus

>

Note

point at

Without employing the notion of the polar sub-tangent and the


value of p, may also be obtained as follows

infinity, the

From

(///)

we

have,

when

r->

oo,

/>=lim

[r

cos (0-a)]

lim

[r

cos (e-Oi 4-*/2)l

lim

which

is

of the form (0/0).

ASYMPTOTES

..

-> p

hm
Hence the asymptote

orfei
r*-j
</r

Oi is

the limit of o as r-+

Working
Change
as u ->

^i
,-

<fo

where ii=-

*~

=-

rcosf

r sin

oo re., as

rule for obtaining

r to

is

limf-

where

..

=iim

33$

6-Oi-f-

(Oi0).

->0.

asymptotes -to polar curves.

l\n in the given equation and find out

the limit

of 9

0.

Let

1?

be any one of the several possible limits of

Determine (- dOjdu) and


Let

this limit

its

limit as u -> 9

0.

and 0-+6i-

be p.

Then
p=r
is

sin (0i-0),

the corresponding asymptote.

To draw

the asymptote.

Through the polo O draw a line making angle


on this line take a point Y such that

initial line

(0 L

|TT)

with the

OY=\im(-d6ldu).
The

line

drawn through Y perpendicular

to

OY

is

the

asymptote.

Examples
I.

Find the asymptote of the hyperbolic spiral rd~a.

Here
a\r

an so that

~>

as u

> 0.

Here
w = 0/0,

Since

we have
dujde^lla

or

dOjdu^a.

Therefore

/.?.,

is

the asymptote.

-a=r sin (00)


sin d = a,

r sin 0,

required

334

DIFFERENTIAL CALCULUS
Find the asymptote of the curve
a
__

2.

'

~-cos0

Here

= -~(i-COS0).

"=
f

When

w ->

0,

so that cos

cos 0) ->

Now,
t/w

or

sin

-rr
d9~

du

=
-

r-

sin0

2a

de

TT

and

2a

d0

2a

TT

_---_r
=r

--

a-r

i.e.,

sin^-g-

and

TT

sin

\/3

J,

4tf

), i.e.,

sm

oos

= r(A/3 sin + 3 cos 0),

are the two asymptotes.


Exercises
Find the asymptotes of the following curves
2

'

3.

r=o

7.
9.

rn

11.

sec

sinno=? o

rologO8

13.

r(l-e )=o.

15.

17.

C()s 9

sin

0=o*e

=o2 (secl e+cosec 2

4.

rt

6.

2r 2 =tan 20.

8.

10.

sin 8 e-fccs 8 9'

""e-l"

Q+b tan 6- (P-I/0


r sin 20=a cos 3e.
rO cos 0=0 cos 20. (P.tf.)

5.

"

r=

8).

r sin /io=o.

r=o

tan

12.

rlogO==o.

14.

r(o

16.

r(Tr+0)^o^

(P.C/.)

0.

-f 8)-2oO.
6
.

Find the equation of the asymptotes of the curve given by (he

equation

ry-W+r'^/n-itoH

+/t(e)-0
(P.C/.

18.

Show

Hons., 1938)

that all the asymptotes of the curve


r tan n$a,

touch the circle


19.

ro/fi.
Find the asymptotes of the curve
r

cos

2o~o

sin 30.

(Delhi

Ho*s

194%)

CHAPTER XVII
SINGULAR POINTS
MULTIPLE POINTS, DOUBLE POINTS
17-1.
Introduction.
Cusps, Nodes and Conjugate points. The
cases of curves considered in
11-4, p. 243 show that curves with
implicit equations of tho form/(x, j>)=0 exhibit some peculiarities
which are not possessed by the curves with
explicit equations of the
form y=F(x). These peculiarities arise from tho fact that the
equation f(x, y)=0 may not
define, >', as a single valued function of*.
In fact, to each value of x
corresponds as many values of y as is tho
degree of the equation in y, and these different values of y give
rise to different branches of the curve.

We

recall to ourselves the

following three curves considered in

11*4.

Fig(i)

Origin

is

127.

a point

Fig. 128 .

common

to the

two branches of the Cisaoid

(Fig. 127)

y*(a-x)=x*,
and the two branches have a common tangent there.
Such a point on a curve is called a cusp.
(ii)

Origin

is

a point

common

to the

two branches of the

phoid (Pig. 128)

and the two branches have

different

Such a point on a curve

is

tangents there.

called a node.

335

n 41)
.

Stro-

DIFFERENTIAL CALCULUS

336
(Hi)

0,0)

is

a point

common

to the

two branches of the

curve (Fig. 129)

ay*-x(x+a)*=^Q

and the two branches have imaginary tangents


point in the immediate neighbourhood of the point
lies on the curve.
Here, a, is positive.

There

there.

Such a point on a curve

is

called an

isolated

a,

11-43)
is

no

0) which

or conjugate

point.

Fig.

17-2.

129.

Definitions.

Double points. Cusp. Node. Conjugate point. A point through


which there pass two branches of a curve is called a double point.

curve has two tangents at a double point,

one for each

The double point will be a node, a cusp or an


according as the two tangents are different and real,
imaginary.

isolated point
coincident or

branch.

A point through which there pass, r, branches


called a multiple point of the rth order so that a curve
tangents at a multiple pDint of the rth order.

Multiple point.

of a curve
has,

r,

is

Thus a double point is a multiple point of the second order.


point of the third order is also called a triple point. A
multiple
^A
multiple point is also, sometimes, called a singular point.

17*3.
simple rule for writing down the tangent or tangents
at the origin to rational algebraic curves is obtained in the following
article.

Tangents at the origin. The general equation *>f rational


curve
of the wth degree which passes through the origin 0,
algebraic
17*31.

SINGULAR POINTS

when

337

arranged according to ascending powers of x and y,

form

is

of the

...(/),

where the constant term

is

absent.

Let P(x, y) be any point on the curve. The slope of the chord
<OP is y\x. Limiting position of the chord OP, when P -+ O, is the
and y -> 0,
tangent at O so that when x ->
lim

()>/*)

=m,

is the slope of this tangent.

From

On

(/),

we have,

taking limits,

after dividing

when x ->

=0

0,

we

by

x,

get

so that

m=

bjb 2

if

Hence

ylx^-bjb*
d.e.,

b,x+b 2 y^Q
is the tangent at the origin.
t;o

zero the lowest degree

If& 2 =0but

6^62=0
(c A x

may be written down by equating


degree) terms in the equation (i),

(first

then, considering the slope of OP with


can be shown that the tangent retains the same

6^0,

reference to F-axis,
form.

Let

...(ii)

This

it

so that the equation takes the form

+c2 xy +w*) +(4* 3 +djc*y + ......

Dividing by

x*

and then taking limits as x ->


C1

+... =0.
0,

we

+c 2m+c 3m 2 =-0,

...(wi)

get
...(iv)

a quadratic equation in, m, and determines as its /wo roots


the slopes of the two tangents so that the origin is a double point in
^diich

is

this case.

The equation of

either tangent at the origin

is

y=mx,

Tvhen
obtain

is

a root of

(iv).

Eliminating

..(v)

m between

c^+CiXy+Csy^Q,

(iv)

and

(v),

we

...(vi)

as the joint equation of the two tangents at the origin. This can be
written down by equating to zero the lowest degree terms in (111).

The equation (vi) becomes an identity if c l =c2 =c3 =0. In this


case the second degree terms, also, do not appear in the equation of
the curve. It can now be similarly shown that the equation of the
tangents can still be written down by equating to zero the terms of
the lowest degree which is third in this case.

DIFFERENTIAL CALCULUS

338

In general, we see that the equation of the tangent or tangents at


the origin is obtained by equating to zero the terms of the lowest degree
in the equation of the curve.

The

on a curve whose equation


and
contain
the
the first degree terms*
constant
least,

origin will be a multiple point

does not, at

Illustrations.
(i)

The

origin is a

node on the curve

and x=0, y=0 are the two tangents thereat.


(ii) The origin is a cusp on the curve
and

y=Q
( ill)

and

is

the cuspidal tangent.

The

origin

axiby=Q are
(iv)

The

origin

and x=0, x=y,

is

an isolated point on the curve

the two imaginary tangents thereat.


is

a triple point on the curve

xy

are the three tangents thereat.


Exercises

Find the tangerts at the origin to the following curves


1.

.3.

5.

(xH;y

a 2

) ==4fl jty.

+y2 )(2a-x)=b2x.
2
2 2
2
2
8
(x +/) =a (x ->> )

(x

-x 2 )=x2 (b-x)^

2.

y*(a

4.

a*(x*-y*)=x*y*.

Example
Find the equation of the tangent at

and show

We

that this point is

2) to the curve

1,

cusp.

will shift the origin to the

point

1,

2).

To do

so

we

have to write

where X,
are the current co-ordinates of a point on the curve with
"reference to the new-axes.
The transformed eqation is
or

Equating to zero the lowest degree terms, we get

-y

=0,

i.e.,

(r-Jf)

==0,

339

SINGULAR POINTS

which are two coincident lines, and, therefore, the point is a cusp
and the cuspidal tangent, i.e., the tangent at the cusp with reference
to the

new axes

is

To find the equation of the cuspidal tangent with reference to


the given system of axes, we write
Hence the tangent at

(1,

2) is

Exercises

Find the equations of the tangents to the following curves

1.

rV+x

2.

(;c-2)

3.

4
x*-4ax*-2ay* + 40V+3aV--a ==0 a t (a, 0) and (20, a).
Show that the origin is a node a cusp or a conjugate point on the

4.

)=*V--* )at(a,

2
=X>>-l) at(2,

0).

1).

curve
2
2
y =ax +ax*

according

as, a, is positive,

Conditions for any point

17*4.

zero or negative.

(Delhi Hons. 1950)


(x, y) to

be a multiple point of

the curve

f(x,y)=0.
In

10*94, p.

213,

we have

seen that at a point

(x, y)

of the

curve
/(*.

the slope of the tangent, dy/dx

is

y)=o,
given

by the equation

At a multiple point of a curve, the curve has at least two


tangents and accordingly dy\dx must have at least two values at a
multiple point.

The equation (1), being of the first degree in


by more than one value of dyjdx, if, and only
/.0,/f =0.

satisfied

dyjdx,

can b^

if,

Thus we see that the necessary and sufficient conditions for any
point (x, y) on/(x, }>)=0 to be a multiple point are that

/.(^y)=0,/w (x y)=0.
f

To find

multiple points (x, y),

we have

therefore to find the values.

of(x, y) which simultaneously satisfy the three equations

340

DIFFERENTIAL CALCULUS
17*41.

To find

the slopes of the tangents at a double point.

Differentiating (1) w.r. to x,

we have

Ay
Jf*+r
* +J **
dx

d?

<( Jv
f * +J
+f**y
*
+1
dx~) dx
point, where /y =0,/ .=0,
\

so that at the multiple


are the roots of the quadratic equation

the values of dyfdx

not all zero andfx =Q=fv the point (x, y)


In
casef^tf^fy*, are
be a double point and will be a node, cusp or conjugate according
as the values ofdyjdxare real and distinct, equator imaginary i.e.,
,
according as
,

will

(/*,)

2
2
-/* /,, >0, -0, <0.

2
If/a =/art =y^ =0 the point
order higher than the second.
2

(x,

will

y)

be multiple point of

Example
Find the multiple points on the curve

Also, find the tangents at the multiple point.

Let

/(*, y)

=x4

f*(x>

J>)=0 gives *=0,

fv( x *

y)=

Hence the two

gives j;=0,

a.

partial derivatives vanish for the points

(0, 0) <0,-fl), (a, 0),

Of these

a.

a,

(fl,-n)

(-a,

0),

(-a,-a).

the only points on the curve are


(fl.0),(-fl,0), (0,-a).

Hence these are the only three multiple

To
follows

find the tangents at the

points,

multiple points,

On the curve.

we proceed

as

First method.

We have
Since at

(a,

0).

fJ
therefore,

by the equation

(2),

the values of dy/dx at

by
a

-f 8a

= 0,

(a, 0)

are given

341

SINGULAB POINTS

The two values being both

real,

the point

a node. The

(a, 0) is

tangents at (a, 0) are


It
(0,

may

similarly be

Second method.

we

shown that the tangents at

a, 0)

and

a) are

Differentiating the

given equation w.r. to x,

get,

4x 3

&a 2yy l

Qay^

which identically vanishes

for the multiple points.

Differentiating again,

we

12x 2 - 12ayy^

From

this

^=

,
(/) for (a, 0),
for
(ii)
(-fl, 0)^^=1,

(iii)

for(0, -a),

Knowing the

get

6ay*y 2
see that

we

4a*x =0,

^^f,

- Qa yy 2 - 4a
2

6a*y\

/.*.,

=0.

i>.,^
/.e.,

^-

slopes of the tangents,

we can now put down

their

equations.

Third method.
origin to this point.

To find the tangents at


The transformed equation

(a, 0),
is

we

shift

the

or

The tangents at the new

The tangents
It

origin are

at the multiple point

may similarly

y=

(a, 0),

therefore, are

\/(4/3)(*-).

be shown that

are the tangents at the multiple points

0, 0)

and

(0,

a) respec-

tively.

The three mutiple points on the curve are nodes.


Exercises

Find the position and nature of the multiple points on the following
curves

1.

x*(x-y)+y*=Q.

2.

y*=*x*+ax*.
x*+y 2x 8 +3>> 2 =0.

3.

(D.U.1951}
(D. U. \950\P. U.)
(P. U.)

DIFFERENTIAL CALCULUS
4.
5.

xy
2
>>

=(jc-l)(jc--2)

2
.

6.

ay*=(x-a)*(x-b)*.

7.

x*-4ax*+2ay*+4a*x*-3a*y*-a*=Q.

9.

10.

x*+y(y+4a)*+2x*(y-5a)*=5a*x*.

11.

X*+2x*+2xy-y*+5x-2y=Q.

12.

(2 y+x-fl)

13.

(B. U.)
(P. U.)

5
-4(l~x) =0.
2
8
(jc+>>) -^2>>-*+2) =0.

(P (70

<

14.

(/-a^-hx^-f

15.

xV 2 =(a+}0 2 (& 2

Find the
following curves

(P.V.)

2
30) ==0.
2
.y )

(P. U.)

distinguishing between the cases

equations of the

tangents at

b^a.

(D. U. Hons. 1953)


the multiple points of the

16.
17.
18.

19.

(y-2)*=x(x-\)*.
Show that each of the curves
(x cos

cc

sin

2
=c(;c sin <x.+y cos a) ,
cusp show also that all the cusps lie

a6)

for all different values of a, has a

on a

circle.

17 5. Types of cusps. We know that two branches of a


curve have a common tangent at a cusp. There are five different
ways in which the two branches stand in relation to the common
tangent and the common normal as illustrated by the following
figures

Fig. 130

Fig. 131

Single cusp of 1st species

Fig. 132

Double cusp of

Single cusp of 2nd species

Fig.
1st species

133

Double cusp of 2nd species

SINGULAR POINTS

343

Fig.
Point of oscu-inflexion

mon

In Fig. 130, the two branches lie on the same siue of the comnormal and on the different sides of the tangent.

In Fig. 131, the two branches

mal and on

the

same

lie

on the same

side

of the nor-

side of the tangent.

In Fig. 132, the two branches lie on the different sides of the
different sides of the tangent.

normal and on the

In Fig. 133, the two branches lie on the


normal and on the same side of the tangent.

different

sides

of the

In Fig. 134, the two branches lie on the different sides of the
lie on the same, and on the other on
opposite sides of the common tangent. One branch has inflexion at
the point.

normal but on one side they

It will thus be seen that the cusp is single or double according as


the two branches lie on the same or different sides
of the common nor-

mal. Also it is of the first or second species according as the branches


lie on the different or the same side
of the common tangent.

Examples
Find the nature of the cusps on the following curves

1.

(i)

y*=x\

(ii)

^-xi^o.

(0 y=0 is the cuspidal tangent. Since x cannot be negative,


the two branches lie only on the same side of the common normal
eo that the cusp is single. See
Fig. 130.
3

y=xf

so that to each positive value of x


^gain,
correspond
two values of y which are of opposite
signs and hence the two
branches lie on different sides of the common
tangent and the cusp is

of first

species.

(11)

Two

branches of >>*

ndj>+x a =:0 which

y=0

and

the origin

* 4 =0,

are the

two parabolas y

on

x*=0

different sides of the common tangent


extend to both sides of the common normal x=0. Thus
is

lie

a double cusp of

first species.

See Fig. 132.

344

D1FFBBENTIAL OALOTTLTJS
(7) Here

where the two signs correspond to the two>


x

"^"^
Fig. 135

y~Q

the cuspidal tangent.


Since x cannot take up negative values,
the two branches lie only on the sime sideof the common normal and the cusp, there-

branches

is

fore, is single.

Now, one value of .y is always positive and, therefore, the corresponding branch lies above X-axis. Again
JT

QX

^> X

II

Thus, for the values of x lying between and 4^, the second
value of y is also positive and, therefore, the corresponding branch
lies above Y-axis in the vicinity of the origin.
Thus the cusp is of
the second species.

Show

2.

that the curve

has a single cusp of the first species at the origin.

(Delhi ffons. 1949}

Equating to zero the lowest degree terms,


origin is a cusp and y=0 is the cuspidal tangent.

We re-write the

we

see

that the

given equation in the form, quadratic in y y

y*-y
and solve

it for y,

so that

_.

For positive values of #, we have


*4 (2+x) 2 +4;c 3 >;c 1 (2+;c) 2

or

so that to positive values


Thus the

opposite signs.

when x
>

of x correspond two values of y with


two branches lie on opposite sides of x-axis

is positive.

Again,

we have

For values of x which are


is

sufficiently small in

positive, for the same -> 4 when x -> 0.


Thus for negative values of x which are

numerical value,

numerical value,

sufficiently

small in

SING o LAB POINTS

345

is
negative so that the values of y are imaginary.
take up negative values.

Hence the curve has a

Thus x cannot

single cusp of first species at the origin.

Exercises
Find the nature of the cusps on the following curves
14
3.
5.

7.
8.

x\x-y)+y*=0.
x *+y*-2ay2 =Q.
2
(y~-x) +x'=0.
x*ax*ya*x*y+a*y 2 =Q.

2.

4.

a*

6.

x'

Examine the curve

for singularities.

(Delhi Hons. 194SY


Prove that the curve jc s +y 8 =ax2 has a cusp of the first species at the
origin and a point of inflexion where x=a.
(Lucknow Hons. 1950}
10.
Show that the curve
9.

has a single cusp at

(a, 0).

(D.U. 1955)

Radii of curvature at multiple points. The formula for


the radius of curvature at any
point (x y) on the curve f(x, y)~Q,
as obtained in
15-42, page 292, becomes meaningless at a multiple
point where/a. y^~0. At a multiple point we expect as many values
Of course, these values of p may not be all
of, P, as its order.
17-6.

distinct.

Tho following examples will illustrate the


ing the values of p at such points.

method of determin-

Examples
1.

Find the radii of curvature at the origin 0* the branches of the

curve

Here, 2*y*=*
at the origin so that

To

find,

find Iim(.y 2 /2.x).

p,

i.e.,

it is

for

To do

x^Q, )>=(1/V 2 )*
a

the
this,

branch

we

which

touches

jc=0,

wa

write
p l9

and

are the three tangents

triple point.

.e.,

X=

substitute this value of x in the given equation.


Lim
is
the radius of curvature of the
at
branch
the
origin.
corresponding
We get

or

p^P

346

D1FFEEENTIAL OALOTLTTS
Let y -*
Thus,

To

we have

so that

branch=

p, for this

find, p, for

+a/p=0.

a.

we proceed

the other branches

Suppose that the equation of either branch


J>=/(0)+x/'(0)

YTe have,

+^1/"(0)+

as follows.

is

given by

....

ere

/(0)=0.
Also

we

write /'(0)=p,/"(0)

Making substitution

Equating

Thus we have

<7.

in the given equation,

co-efficients of x*

and

x*,

we

get

we get

2ap*=a, p*+2apq=l.
These give

=<

=- 3V2

'

for the two branches.


2.

Show

that the pole

is

r=a(2

and

that the radii

cos

point on the curve

8+ cos

30),

of curvature of the three branches are

V3a/2,

The radius

triple

vector,

a/2,

V30/2.

vanishes for the values

r,

0+ cos

2 cos

of, 0,

given by

30=0,

!>.,

2 cos

0+4 cos 8

03 cos 0=0,

tpr

cos

2
(4 cos

1)=0,

or
cos

Thus,
point.

r=0 when

0=0,

cos

0=7r/3,

0= J,

?r/2,

cos

0=

J.

27T/3 so that the pole

is

a triple

347

SINGULAR POINTS

We now proceed
/

=a(2

to find

p.

0-3

sin

We have
2 cos

sin 30), r2 =0(

09 cos 30).

For 0=7r/3,

r^

y^Sa,

r2

=8a

r2

=0

r2

for 0=7T/2,
r 1==a

or 0=27r/3,

r^
Also,

r=0

y3a,

8a.

for each of these branches.

Putting these values in the formula


(r'+r,*)*

we get the required

result.

Exercises
1.

Show

that the radius of curvature at the origin for both the branches

of the curve

is

2.

Find the radius of curvature at the point (1,2) for the curve

3.
Find the radii of curvature at the origin of the two branches of the
curve given by the equations

y=t-t\x=\-t 2
(For the origin /==
values of /).

1,

Find the radii of curvature


4.

x*+y*=*3axy.

5.

-3^y5
jc +ox

6.

7.

JC

at the origin of the following curves

two

(Folium).

Show

that (a, 0), in polar co-ordinates,

r=a(l4-2sin
and

(P.U.)

so that the two branches correspond to these

find the radii of curvature at the point.

Jo),

is

a triple point on the curve

CHAPTER XVIII

CURVE TRACING
We have already traced some curves
The general problem of curve tracing,
will be taken up in this chapter.

18*1.

and XII.
aspects,

in

Chapters II

in its elementary

It will be seen that the equations of curves which we shall trace


are generally solvable for y, x or r.
Some equations which are not
solvable for y or x may be rendered solvable for r, on transformation
from Cartesian to Polar system.

18

Procedure for tracing Cartesian Equations.


Find out if the curve is symmetrical about any line. In this
connection, the following rules, whose truth is evident are helpful
2.

I.

(/)

curve

which occur in
(ii)

its

curve

which occur in

its

is symmetrical about A'-axis


equation are all even

if

the powers of

the powers of

symmetrical about
equation are all even
is

7-axis

if

(MI) A curve is symmetrical about the line y=x if, on interchanging x and y, its equation does not change.
II.
Find out if the origin lies on the curve. If it does, write
down the tangent or tangents thereat. In case the origin is a multiple point, find out its nature.

Find out the points common to the curve and the co-ordinate
be any. Also obtain the tangents at such points.
there
if

III.

axes

Find out dyjdx and the points where the tangent is parallel
co-ordinate axes.
At such point, the ordinate or abscissa
generally changes its character from increasing to decreasing or vice
IV.

to

the

versa.

V.

Find out such points on the curve whose presence can be easily

detected.

VI.

Find out points of inflexion.

(It

may

not always be neces-

sary).

Find out multiple points, if any, and their nature.


VIII. Find out the asymyplates and the points in which each
asymptote meets the curve.
IX. Find out if there is any region o?the plane such that no part
VII.

of the curve

lies in

it.

Such a region

is

generally obtained on solving the equation

348

for

349

CURVE TRACING

one variable in terms of the other, and find out the set of values of
one variable which make the other imaginary.

X. If possible, solve the equation for one variable in terms of


the other and find out, by examining the sign of the derivative, or
otherwise, the ranges of the values of the independent variable for
which the dependent variable monotonically increases.
Important Note. The student must remember that a mere knowledge of
symmetry, asymptotes, double points, etc., will not enable him to trace a curve,
this knowledge being only piece-meal; asymptotes indicate the nature of the
curve in parts of the plane sufficiently far removed from the origin and the
tangent at any point gives an idea of the curve in the neighbourhood of the
point. To draw a curve completely, it may be necessary to solve the given
-equation for one variable, say, y, in terms of another and then to examine how
y varies as x varies continuously. For this purpose th& examination of dyfdx

proves very helpful.

Examples
Equations of the form

18-3.

1.

Trace the curve

We

note the following particulars about this curve

(i)

(11)

It is

symmetrical about both the axes.


through the origin and

It passes

gents thereat.
(in) It

Thus the origin

meets A^axis at

dy
r

ax

y=x are the

two tan-

a node.

(a, 0), (0, 0)

The tangents at

axis at (0, 0) only.


jt=- tf respectively.

/M
(V)

is

and (
and

(0, 0)

a, 0)
fl,

and meets Yx=a and

0) are

fl

-A(

which becomes

fl*-2a x

Thus the only

(v)
(vi)

when

zero,

It has

-x4 =0,

i.e.,

real values of

when x2 =(-l

for

v'2)fl

s
.

which dyjdx vanishes ar

no asymptotes.

Solving the equation for y,

We see that,

we

re-write

it

as

2
x 2 must be non-negative and
y to be real, a
a
and a. Hence the entire curve lies
therefore, x must lies between
between the lines x=a and x= a.

for

350

DIFFERENTIAL CALCULUS

Conclusion. In order to trace the curve we have to connect


the above isolated facts. We proceed to do this now.
The curve

being symmetrical about both the axes,


we firstly consider the part of the curve
lying in first the quadrant where x, y
are positive and

/P

'-* A

(a.o)

VL

When x=0,

then ^=0.

When

increases, starting from 0, then y >


which is positive, also increases and

goes on increasing

x=a

till

x=v

/
(

l+y'2)^

i.e., where the tangent


Fig. 136.
is parallel to Z-axis. Since ,y=0 when
we see that when x increases from \/(l + \/2) a to fl, y

where dyjdx=0

decreases.

/A
<*

Fig. 137

The variable x cannot take up values greater than a, for, then,


y is not real. Thus the part of the curve in the first quadrant is as
shown in Fig. 136. The curve being symmetrical about the two axes,
its complete shape is as shown in Fig. 137.
2.

(/)

Trace the curve


It

is

symmetrical about A'-axis only.

It passes through the origin


are the two imaginary tangents thereat.
lated point.

()

*axcP
(IV)

ivhich vanishes

Since

Q,

i.e.,

origin

is

an

iso-

meets J-axis at ( a, 0) and (0, 0) and >-axis ,at the


a is the tangent at (a, 0).

(Hi) It

origin only

and yz +xz
Thus the

when

oxa =0,
2

i.e.,

whenx=J

CURVE TRACING
For x=(l
See

(vi)

(v)

(vi)

\S5)aj2,

which

lies

351
a and

between

a,

is

not reaL

below],

y=(x+a)

and

see that, for

be

> a and <

three asymptotes.

y= x*^* =x **^^>

Since

we

x=a are its

to be real,

jc

2-

a 2 must be non-negative,

i.e.,

must

a.

Hence no part of the curve lies between


a.
and x==

xa

Fig. 138

Conclusion. The curve being symmetrical about the A" axis, we


consider the part of it lying in the first two quadrants where y is
positive so that we have

where we consider

s ig n

or

according as x

is

positive or

nega-

tive.

When x=
responding point

Let
is

y=0.

Also jc=

a, is

the tangent at the cor-

a, 0).

vary continuously from

a to

ao

Then, since

tfy/flbc

any of these values of x, and y= (x+a) is an asympwe have the part of the curve in the second quadrant as shown

notO

tote,

JC

a,

for

in Fig. 138.

a and fl, y is not real.


For values of x lying between
Since jc=0 is an asymptote and dyjdx^Q for JC=(l+\/*>)

fl

/2,

DIFFEBBNTIAL CALCULUS

352

we see that y, starting from oo goes on decreasing as x increases


'from a to (1 4-^5)0/2. As x increases from (l+<\/6)0/2 to oo
y
'

again goes on increasing and, y=x fa being another asymptote,


have the part of the curve in the first quadrant as shown.

The curve being symmetrical about the


shape is as shown in Fig. 138.

We

A"-

axis,

its

we

complete

thus have the curve as drawn.

Note. The curve appears to have two points of inflexion, apparently


necessitated by the fact that the curve could not be asymptotic to the lines
y= (x+a) unless it changes the direction of its bending after leaving the point
(a, 0) where it touches x= a. In fact, by actual calculation, we see that
- 2a/>|3) are its two points of inflexion.
<-2fl, 2a/>l3) and ( 2a,
3.

Trace the curve

'(/)

It is symmetrical

about both the axes.

does not pass through the origin.

(n) It

a, 0) but it does not meet


(111) It meets A'-axis at (a, 0) and (
'yaxis at any point x=a and ;c = a are the tangents at (a, 0) and
'.{0, 0) respectively.
;

'

_ ~~
dx

JC

v (x2

^__
2

}'

which never becomes zero.


(0)

J>=1

Prom
ibut

it

will

(vi)

we

are the

two asymptotes.

x=0 is also an asymptote,


cannot be an asymptote.

16-31. p. 319, it appear*? that

be seen below that

it

Writing the equation in the forms

see that for

x and y

to be real,

we have

<a

a2

>

0, i.e

1-y*

>

0, i.*.,--l

jc

or

>

and

mo

<y

<1.

As no part of the curve lies between the lines x= fl, and


branch of the curve can possibly be asymptotic to x~0.
Conclusion.

Consider

>which corresponds *to the part of the curve in the

first

quadrant.

CURVE TJUGINQ
and a^ y is
values of x lying between
y=Q so that (a, 0) is a point "on thfc curve
seen above, is the tangent there.
As x increases from a onwards, y which is positive
.

If

x=a,

must

stantly increase

con-

dyjdx being never zero.

y=l is an asymptote, the part of the curve in


quadrant can now be easily shown.
The curve being symmetrical about both the axes, its complete

Since the line

the

first

shape

is

as

shown

in the Fig. 139.

Y*

Fig.
4.

139

Trace the curve

y *=(x-a)(x-b)(x-c).

We

suppose that a, b c are positive and in ascending order of


magnitude so that we have to consider the following four cases
y

a<b<c. (//) a = b<c.


Case I. a<b<c.

(in)

(/)

symmetrical about

(/)

It is

(//)

It does not pass

(Hi) It

meets

meet 7-axis

x=--a

a<b = c.

a^b^c.

.Y-axis.

through the origin.

A'-axis at (a y 0), (b, 0)


t

(/?)

x=b, x=c

nnd

(c, 0),

are the tangents

but

at- (a,

it

docs not

0), (6, 0), (c, 0),

respectively.
(iv)

(v)

It has

no asymptotes.

When x<a,
y*<0,
whena<x<&, .y 2 >0,
when b<x<c, y*<Q
2
whenx>c,
^ <0.

e.,

is

i.e.,y is

Hence, no part of the curve


between the lines x=6, x = c.
for

lies to

not real

not real

the

2
(vi) As y vanishes for x=a and x=fe,
some value of x between a and.&.

left

it

of the line x = a and

must have a maximum

354

DIFFERENTIAL CALCULUS
(viii)

As x

increases

beyond

c,

for

y* also constantly increases,

each of the three factors then increases.

We

Fig. 140

have

S~-2(a+b+c)(\lx)+(ab+bc+ca)(\lx*)
which ->

oo

&s

x ->

ao

Thus the curve tends to become parallel to 7-axis as x --> oo


Since the curve, in departing from (c, 0), where the tangent is
parallel to F-axis, must again tend to become parallel to F-axis, we
see that the curve must change its direction of bending at some
and as such must have a point of
point between (c, 0) and oo
.

inflexion.

Hence, we have the curve as shown in Fig. 140.

an oval and an infinite branch.


a=b<c. The equation, now, is

It consists of

Case

II.

(a.oi

Fig. 141

355

CURVE TRACING
It is easy to show that (a, 0) is an isolated point, as
of case I shrinks to the point (a, 0).

if

the

oval

The curve consists of an isolated point and an infinite branch


and can be easily drawn. (Fig. 141 on the preceding page).
As

in case I,

it

can be shown that the curve has two points of

inflexion.

Case

III.

a<b=c. The

equation, w>w,

is

Fig. 142

It is easily

shown that

(b, 0) is

a node and

are the nodal tangents.

The curve may now be


Case IV.

easily

a=b=c. The

drawn.

equation, now,

Fig. 143

(Fig. 142).
is

DIFFERENTIAL CALCULUS

356

It is easy to see that (a, 0)

is

a cusp and

"j^O

is

the cuspidal

tangent.

The curve may b3

drawn.

easily

(Fig. 143).

Trace the cissoid

5.

y*(a-x)=x*.
This curve has already bsen considered in

11-41, p.

244

also.

(See Pig. 64, p. 244).

We note the

symmetrical about x-axis.

It is

(i)

following particulars about the curve

through the origin and y =0, i.e., J>=0, jj=0


are the two coincident tangents at the origin.
Thus the origin is a
It passes

(ti)

cusp.

meets the co-ordinate axes at the origin only.

(Hi) It
(iv)

x^a

(v)

Since

is

the only asymptote of the curve.

y*--x*/(ax),

we

see that

and

is

positive,

i.e.,

is

real

only

when x

lies

between

a.

which vanishes when

x=f#

or 0.

fais outside the range of admissible values of x. Thus


vanishes
at no admissible value of x except x=0.
dyjdx
Combining all these facts, we may easily see that the shape of
the curve is as shown in Pig. 64, p. 244.

But X

Note.
1

1*4?,

The student would do well

to learn to trace

the curves given in

by the methods of this chapter.

1-43 also

Trace the curve

6.

It is neither

(/)

symmetrical about the coordinate axes nor

about the
(if)

Origin

is

a cusp, and

x=0

is

the cuspidal tangent.

(HI) It meets A"-axis at .(0, 0) and (3a, 0) bjut meets


he origin only x=3a is the tangent at (3a, 0).

F-axis at

(10)

y+x=aia

asymptote at

its

only asymptote and the curve meets the

'

(a/3, 20/3).
(t;)

(vi)

x and y cannot both be negative.


Now, y*dyldx=x(%dx) and, therefore,
for x=2a.
y

OtiBVE TRACING

Solving for y,

357

we obtain

Fig. 144.

If

*=0, then y =0 and

7-axis

is

the tangent there.

<JC<3a, then >>>0, and if x increases from 0, y also inwill go on increasing with x till x =2a where dyfdx=Q.
When x increases beyond 2a, y will constantly be decreasing
y=0
for x=3a and is negative for x>3a.

If
creases ;

We now consider the

negative values of JC.

If x is negative, y is positive and constantly goes on increasing


as x increases numerically, i.e., as x varies from
to
oo.
Also,

y+x=a is the only asymptote

of the curve.

Taking all the above facts into consideration, we aee that the
complete curve is as shown in Pig. 144.
1
18-4.
Equation of the form y +yf(x) + F(x) = o.
7,

Trace the curve

(i)

It

about the
(ii)
(///)

(tv)

curve at
(v)

it.

is

line

symmetrical neither about the co-ordinate axes, nor

y=x.

a cusp and y =0 is the cuspidal tangent.


It meets the co-ordinate axes at the origin only.
Origin

is

y=x-\-l

(-,
We

We have

is

the only asymptote of the curve.

re- write

the

meets the

equation as a quadratic, in y and solve

yx*+x*~ 0,
or

It

i).

358

DIFFERENTIAL CALCULUS

so that

and

is

imaginary

if x s (x4.) is

negative

i.e.,

if

lies

between

4.

Fig. 145.

Thus, there

is

no part of the curve lying between the

line

x=0

and x=4.

x=4

for both the branches, and ;c=4 is the


y=8 when
the
at
tangent
point.
The following additional remarks about the two branches of
curve will facilitate the process of tracing the curve.
(vi)

y=

and
are the
or 4.

two branches. They meet where x 4

Thus the two branches meet


Consider the branch

When

x, starting

4x 8 =0,

at the points (0, 0)

i.e* 9

and

where

#=0

(4, 8).

(i).

from

4, increases,

then y

is

positive

and

also

constantly increases.
*

Also, when x starting from


decreases,
values whose numerical value increases, then

stantly increases.
Now, consider the branch

When x>4,
therefore,

x4

is positive.

4;c

and takes up negative


y is positive and con-

(//).

positive and <\/(x*4cX*)<\/x*=x* and,


Also, when x, starting from 4, increases, y
is

decreases in the beginning.

When x < 0, x 4

4*3

is

positive

and

\/( x

4x*)

> -y/x

=* x 1

so

CUBVB TRACING

359

that y is negative. Also, when x, starting from 0, decreases, then y


numerically increases. Thus, we have the shape as shown in Fig. 145.

Polar carves.

18-5.
8.

Trace the curve

Fig. 146.

We first

consider positive values


only.
dr
2a0

is

have

_
=

^0

which

We

(f-jL02)2>

always positive so that

r,

constantly increases as,

0, in-

creases.

Also,

r=0 when 0=0;


Again we have
-

Thus

s=

_2

which -> a as

-> oo.

starting from its initial value 0, constantly increases as


-> oo so that the point (r, 0)
and approaches, a, as
and
nearer
nearer
the
circle
whose centre is at the pole
approaches
and radius equal to a.
The circle is shown dotted in the Fig. 146.
The figure shows the part of the curve corresponding to the positive values of
only, and the part of the curve for negative values
r,

increases

is its

reflection in the initial line.


9.

Trace the curve

r~a(sec d+cos

0).

Here
1

+
,

\cos
(i)

(ii)

The curve
r

cos

x~

\
cos 9 ]=^a
/

+ cos

--

cos

symmetrical about the initial


0=a, i.e., x=rtis its asymptote.
is

line,

DIFFERENTIAL CALCULUS

360

~
and

so .that

4- is

positive

when

lies

between

ir/2.

Fig, 147.

is

When 0=0, dr/dOQ so that <=Tr/2 and, therefore, the tangent


perpendicular to the initial line at the point (2a, 0)
.

Also, r=2a, when 0=0


Hence we see that when increases from
cally increases that 2a to oo and the point P(r,
of the curve drawn in the first quadrant.

and

-> oo as

->

Tr/2.

to TT, or monotoni0) describes the part

When 6 increases from TT to TT, r, remains negative and decreases in numerical value from oo to 2a and so the point P(r 6)
describes the part of the curve as shown in the fourth quadrant.
,

As the curve
point

will

is symmetrical about the


be obtained when 6 varies from TT to

initial line,

no new

2ir.

In the case of some curves

it is found convenient to
polar as well as the Cartesian form of their equaare obtained from the Cartesian and the others
from the Polar form.

18*6.

make use of the


tions. Some facts

Curves whose cartesians equations are not solvable for x and


but whose polar equations are solvable for r, are generally dealt
with in this manner.

y,

10.

Trace the curve

(/)

It is symmetrical

about both the axes.

CUKVE TRACING
Origin

(//)*

is

a node on

361

the curve and y*

are the nodal

tangents.

and

meets Jf-axis at (0, 0) (a, 0) and ( a, 0), but meets


x=a andx= flare the tangents at (a, 0)
(0, 0) only

It

(Hi)

F-axis at

a, 0).

(iv)

On

no asymptotes.
changing to polar co-ordinates, the equation becomes
It has

0* cos 2

"cos 4 0+sin 4 0'


(v)

We see

that

cto~~

___
0)~

4
(cos ~0-f sin"*

'

so that drfdQ remains negative as


varies from
to ir/4
fore r decreases from a to
as
increases from Q to ?r/4.

and

there-

2
changes from ir/4 to ir/2, r remains negative and
therefore no point on the curve lies between the lines 0=7r/4 and
0-7T/2.

As

(vi)

Fig. 148.

As the curve is symmetrical about both the axes, we have


(Fig. 148).
shape as shown.
Trace the curve

11.

y*~-x*+xy=Q.
It is

(i)

about the

line

(P.U.)

neither symmetrical about the

It passes through the origin


JC=0,
a
so
that
the
is
thereat
node.
gents
origin

(iv)

On

co-ordinate axes,

nor

y=x.
:

(ii)

(1/1)

its

y~Q

are the two

tan-

It cuts the co-ordinate axes at the origin only.

y=x,

y=*x are its

asymptotes.
transforming to polar co-ordinates, we get
ra
tan 20.

(v)

ir/2,

and,

When

increases from

a
therefore, r

to w/4, 20 increases from


from to oo.
increases
monotonically

to

362

DIFFERENTIAL CALCULUS

When increases from -ir/4 to ir/2, tan 26 and therefore also


remains negative and, thus, there is no part of the curve lying
between the lines 0=ir/4 and =ir/2.

rf

Y4

Fig. 149.

When

$ increases

from

ir/2

to

2
3-7T/4, r

When
there

increases

from

to oo.

6 increases from 3?r ,/4 to TT, r remains negative and so


= 3Tr/4, and
no part of the curve lying between the lines

is

0=TT.

We
from

TT

can similarly consider the variations of

as

increases

to 2?r.

Hence we have the curve


12.

as drawn. (Fig. 149)

Trace the Folium of Descartes

y=x

It is symmetrical about the line


point (3a/2, 30/2).
(i)

and meets

it

in

the

x0,

)>=0 are the tangents


(11) It passes through the origin and
there so that the origin is a node on the curve.
(//i)

meets the co-ordinate axes at the origin only.

x+y+a=Q is its

only asymptote.
cannot
both
be
00 x, y
negative so that no part of the curve
on the third quadrant.
(iv)

lies

It

On transforming
r

Now, r=0

dr^ __

46

for

to polar co-ordinates,

_
~

30 sin

we

get

cos
'

cosM~+sin
0=0 and 0=7r/2.

3a(cos
'"

sin

0)(l+sin
81

"(cos ^

+sin

cos
3

0)

0+sin

cos2 0)

363

CURVE TRACING
which vanishes only when
cos

sin

0=0,

For 0=7T/4,

i.e.,

r:-=3a/

tan 0=1.

V2

i.e.,

0=7r/4 or

57T/4.

Yt

Fig. 150

Thus
from

/*

to

to 3a/\/2, as Q increases
monotonically increases from
and monotonically decreases from 30/ y'2 to 0, as 9

7T/4,

increases from Tr/4 to

Tr/2.

Again as Q increases from ?r/2 to 37T/4, r remains negative and


to oo so that the point (r, 0) describes
numerically increases from
the part of the curve shown in the fourth quadrant. (Fig. 150).

As increases from 37T/4 to TT, r remains positive and decreases


from oo to so that the point describes the part of the curve shown
in the second quadrant.
is easy to see that we do not .get
increases
from TT to 2?r.
when, 0,

It

Find the double point of the curve

13.

and

any new point on the curve

trace

(P.U.I 95 5)

it.

Let

Putting

/JC--0

fv (x,

y)

and

/y =0, we

have

or

(a,
origin

Thus we see that fx and / vanish at the points (0, 0), (a, a)
Of these only (0, 0) is a point of the curve so that the
a).
is

the only double point of the curve.

364

DIFFERENTIAL OALODLUS

To

trace the curve


It is

(i)

symmetrical about the line y=*x and meets

Fig.
(it)

It

following particulars aboht

we note the

it

it

at

151

passes through the origin and

0,

y0,

are the

two

tangents there.
(Hi)
(tv)

meets the co-ordinate axes at the origin only.


It has no asymptotes.
It

x and y cannot have values with opposite signs, for such


(v)
Thus
values make 4fflxy negative whereas x*+y* is always positive.
the curve lies in the first and fourth quadrants only.
(vi)

On

transforming to polar co-ordinates, we have


r2

dr
r

so that

dr/dO^O

'

if

cos 0/(sin 4 0+cos 4 0).

sin

^ 2a

1'

4a

and only
tan 0=1,

i.e.,

0=7T/4 and

18-6.
Parametric Equations.
trate the process.

x=a(Q+sin
(

co* 2 0)

57T/4.

(Fig. 151).

The

following examples will illus-

Trace the curve

as* 0, varies in the interval

if

Thus we have the curve as drawn

14.

(cos 0-~ sin ~ 0)(cos 0+sift 0-f 4" sin "

0),

y=a(\+cos

TT, TT).

Yt

Fig.

152

0),

CURVE TRACING

We

have
dx

dy
~0(l+eos0), -~-=
U0

de

^
dx

365

asin0.

ten-?^

~ VjtlJL

positive for every value of 0, we see that as, 0,


always increase. Also, dy\dQ is positive when, 0, lies
and negative when, 0, lies in the interval
TT, 0)
(
increases from
TT to
(0, ir).
Therefore}* constantly increases as
and constantly decreases as, 0, increases from to TT,

Since dx/</0

is

increases, .v will
in the interval

The following
arid

dy/dx

table gives the corresponding values of

0,

x,

y,

Intermediate

TT

an

increases

',

Intermediate

7T

an

increases

increases

y
dy/dx

2a

decreases

oo

CO

and dy/dxco, we see


Since for
air, y
TT, we have x=
that the point A( an, 0) lies on the curve and the tangent thereat is
increases from --TT to 0, x and y are both
parallel to F-axis when
increasing so that the point P(x, y) moves to the right and upwards
till it reaches the position K(0, 2a) corresponding to
where, dyjdx
being 0, the tangent is parallel to A'-axis.
;

When
to TT, x increases but y decreases so
increases from
that the point moves to the right and downwards till it reaches the
position A' (an, 0) corresponding to 0~7r where, dyfdx being infinite,
the tangent is parallel to F-axis.
Thus the point F(x,
from

TT

to

y) describes the curve

AV A'

as

increases

TT.

It is easy to see that the point P(x, y) describes congruent portions to the right and to the left of the portion AVA' as
varies in
the intervals
...... (-57T,

-STT), (-37T, -TT),

(TT, STT),

(3?r, 5?r) ......

Trace the curve

15.

x=^a

sin 20(1

+cos

20),

y=a

cos 2o(l

We have
:

=40

cos 30 cos

d0

dx
This shows that

-4a cos 30 sin

0.

cos 20).

DIFFERENTIAL CALCULUS

366

The

following table gives corresponding values of

0,

x,

y and

dyjdx.
7T/2

00
7T

With the help of


shown.

this

of values, we have the curve as

table

(Fig. 153).

Since x, y are periodic functions of 0, with TT as their period, the


values of x, y will repeat themselves as
varies in the intervals
(TT,

so that the

same curve

27r), (2;r, 3?r), etc.,

will

be traced.

6
Fig. 153

The curve has 3 cupps

viz.,

A,

B and C.

EXERCISES

367

Exercises
Trace the following curves
1.

3.

2.

3ay'=x*(x-a).

ay*=x(a
2

*=x 2

4.

).

5.

8.

+x2)=a2x.
(B.U.)
2
2
2
2
X (y +x )=a (x*-y 2

9.

13.

y(a

6.

(4-x*).

10.

;c=0 8 (a-jc).

15.

yV+x )=0
y x=a(x

19.

/x =x -K

20.

=x

22.

21.

a y

23.

25.

-a2

;c

).

(D.U. 1955)

(2a-x)(x-a).

36;>

^(jc

-l) (7-;t

Trace the following curves

26.

x-(y-l)(y-2)(.v-3).

28.

30.

r log

r=a(0 -sin

29.

31.

r=alog0.

32.

33.

r=a^

34.

sin 30.

sin 0.

Trace the following curves


35.

x*+y* = 5ax y

37.

x*=ay*-axy*.

39.

x(yx)=(y+x)(y

cos 8
2

cos

0=a

0=

cos 20.
2

(L.U.)

sin 30.

0=a.
e).

36.

38.
2

+x*).

Trace the following curves


2
a:

(/*{/.)

r-a-t-6cos0.

0=a

^5)

24.
2

27.

cos

(#

18.

).

2
2

flV-x^a 2 -* 1 )-

16.

(2;c-l)=x(;t-l).

+y2 )=a(x*-y 2

14.

17.

>>

).

;v

x(x

41.

40.

.v

42.

<flx?-&lf**\.

43.

44.

axy=x*-a*.

45.

(jt-}-a).

=Q

2
has a cusp at the origin and a
46. Show that the curve x*(x+y)y
rectilinear asymptote x+y=*\ and no part of the curve lies between the lines
4 and that it consists of two infinite branches, one in the second
and
quadrant and the other in the first and fourth. Give a sketch of the curve.

x0

x=

Trace the following curves


47.

**-(* +J0 a

49.

x a (y+l)-y2 (x-4).

51.

2xy=x*+y*.

(L.U.)
(P.U.)

48.

y(x-y)*=(x+y).

50.

y-x(*

52.

x4

-m2>fy

(B.U.)
(B.C/.

DIFFERENTIAL CALCULUS

368

lies

53.

Xl~*2 )(2-~*) 2 ==*(l--*).

55.

Show

between

54.

r(l

that no part of the curve

x=

x=~l

4 and

x0 and x=3

or between

give a sketch of the

curve.

(Af.T.)

56.

Show

that the cubic


tl

xy*-x'*y-lx*-2xy+y +x-2y+\=-Q,
has double point at (0,

1)

'

asymptotes and sketch the curve.

find its

Trace the following curves

57-

\x=a(0~sin0),>'=a(l"-cos0).

58.

60.

0), y^a (1-cos 0).


x=a cos 30, y=b sin 8 0.
x=a cos 8 0, y=a (sin 30+9 sin 0).

61.

x=0(3 cos0

62.

63.

0+0
x/=log (1+cos

64.

x=a

59.

65.

x=a(0+sin

cos 3 0),
sin

x=fl(co$

log

x/a=2

sin

0),

y=a

^^=a(sin

0)-~cos

0+tan

(sec

0-log

0)

(sec

0,
1

(3 sin

0-sin

00

cos

y/a=sin

y=a log

0+tan

0)

0).
0).

cosec 0+cot

66.

x=a cos

67.

x=0(sin

68.

*==0(0+sin

.y=a(2+sin

+i

C03

0/(3+cos

0-log

0).

j=a (cos 0-J cos 30).


6-in 0), y=a(cos 20-cos 0).

sin 30),

cos

0) sin

.
|

ylb=2
2

0)

(cosec

0+cot

0)

CHAPTER XIX
ENVELOPES
One parameter family of Curves.
19-1.
function of three variables, then the equation

If f(x y, a) be.any
y

f(x,y,*)=o,
determines a curve corresponding to each particular value of

The

a.

of these curves, obtained by assigning different


said to be a one-parameter family of curves.

totality

values to a,

The

is

variable, a,

which

is

different for different curves

is

said to

be the parameter for the family.


Illustration.
(i)

The equation

determines a family of curves which are circles with their centres


on A'-axis and which pass through the origin. Here, a, is the
parameter.
(//)

The equation

y=mx 2am- am 3

determines a family of straight lines which are normals to


parabola

Here the parameter

is

m.

the

We now

introduce the concept of the envelope of a oneparameter family of curves by means of an example CDnsidered in the

next

article.
'

Consider the family of straight lines

19*2.

y^mx+afm,
where, w,

is

the parameter and,

a, is

...(i)

some constant.

The two members of this family corresponding, to, the param l and Wj-f 8m of the parameter, m, are

metric values

~
m

369

..()

370

DIFFERENTIAL CALCULUS

We
4ine

fixed and regard. 8m as a variable which


l
so that the line (///') tends to coincide with the

shall

keep

tends .towards
(ii).

The two

lines

As 8w ->

0, this

tion on the

line

(//)

(//), (///)

intersect at (#, y)

which

lies

on

where

point of intersection goes on changing


and, in the limit, tends to the point

1
N

its posi-

(ii).

This point

is

the limiting position of the point of intersection


latter tends

of the line (ii) with another lino of the family when the
to coincide with the former.

lie a point, similarly, obtained, on every line of the


locus of such points is called the envelope of the given

There will
family.

The

family of lines.

To

the
(//), we notice that
a
on
such
the
line
are
of
*m
y)
point lying
given by

find this locus for the family of lines

oo-ordinates

(x,

x=

2a

m*

y= m

Eliminating m, we obtain

as the envelope of the given family of

lines.

The envelope of a one-parameter offamily of


Definition.
19-3.
is the locus of the limiting position of the points of intersection
of any two curves of the family when one of them tends to coincide with
curves

the other which


19*4.

is

kept fixed.

Determination of Envelope.

Let

/(x,y,a)-0,
j^be

any given family of curves.


Consider any two members

and
f(x,y,

a+Sa)=0,

^corresponding to the parametric values a and oc+Sa.

..(/)

ENVELOPES

371

The points common to the two curves

(/),

satisfy the

(11)

equation

f(x, y, a +8a)
_ -/(*,
g

Let Sa ->

common

to

(/)

0.

and

JF,

a)

Therefore the limiting positions of the points,


satisfy the equation which is the limit of (ill)

(11)

viz.,

/a(*,J>,)=0.

..(/v)

Thus the co-ordinates of the points on the envelope


equations (i) and (/v).
Let the elimination of a between

(i)

and

(/v)

lead

satisfy the

to an equa-

tion

This

is,

then, the required envelope.

To

Rule.

obtain the envelope of the family of curves

/(*,J>,a)=0,
eliminate, a, between
f(x, y,

where fax,

is

y, a)

and

are the

then (w)
(v/)
being the parameter.
Illustration.

Tve eliminate, a,

is

To

is

a)=0,

(/)

and

(/v) for x, y,

we obtain

*=#(),

.-(v)

y=*(a),

..(vi)

parametric equations of the envelope

find the envelope of the family of lines

y <x.x fl/a-=0,
between (v/7) and

obtained by differentiating

The eliminant
which

y,

the partial derivative off(x, y, a) w.r. to a.

on solving the equations

If,

which

a)=0, andfa(x>

(vii) w.r.

-(w'O

to a.

is

the envelope of the given family of lines.

This conclusion agrees with the one already arrived at 'ab


9

initio

in

19-2.

372

DIFFEKENTIAL CALCULUS

The

Theorem.

19-5.

a curve

evolute of

is

the

env3lope of its

normals.
are the normals and PT,
the tangents at two
of a curve. L is the point of intersection of the tan-

QT

QR

PR,

points P,

gents.

LPRQ = LTLT'=W,
arc

Applying

PRQ, we

PQ=Ss.

sine-formula to

the

A.

get
sin

/_RQP

P:=sin /_RQP.
^ -~
sin

or

chord

Let

^ P, so that Z./?gP ->

Lim P/?=sin--

Q-+P

L RPT^ir/2.
~~ 1 = P.
1
.

Thus the limiting position of R which is the intersection of the


P and Q is the centre of curvature at P.

normals at

Hence the theorem.


19-6.

Geometrical relation between a family of curves and its

envelope.

To prove

19-61.

family

is

a point on

We have to

its

that

any singular point of any curve of a given

envelope.

show that

if for

any point on any member of the

family

we have

/,=/w=0,
then, for such a point,

we

will also

have

/a=0.

From

(i),

we

get

Putting

dfldx=dfldy=Q, we get
9//9a=0.

Hence the

result.

ENVELOPES

Thus the
is

373

locus of the singular points of the curves of a


family

a part of its envelope.


1962. To prove that,

curves touches each

in general,

the envelope

of a family of

member of the family.

Let
/(*,?, a)=0,

...(/)

be a given family of curves.


Its envelops is obtained

by eliminating a between

(/)

and

Let

=$

(a)

be the parametric equation of the envelope obtained by solving


and (//) for x and y in terms of a.

The equation
every value of a.

We

(Hi) satisfy

differentiate

(/)

the equation

w.r. to a,

regarding

(i)

jc,

identically,

as

i.e.,

(/)

for

functions of a,

we obtain

o that

3/
dx
which, with the help of

dx_

df

'

(//),

dy
'

</<x

^dy

'da

+ 9/_
a<x

becomes
...(IV)

^f()+^*'(a)=0.
The slope of the tangent at an ordinary point
"V of the family is

(x, y)

of a curve

Also the slope of the tangent at the same point to the envelope
<///)iB*'(a)#'(a).

We see from (iv) that these two slopes are the same. Thus
the slopes of the tangents to the curve and the envelope at the common point are equal. This means that the curve and the envelope
have the same tangent at the common point so that they touch.
N ote. If for any point on a curve, 3 ffox and 3 ffoy are both zero, then
the above argument will break down, so that the envelope may not touch a
curve at points which are the singular points on the curve.
7

We

know that a straight lino and a conic cannot have


Cor. 1.
singular point. Hence we can say that the envelope of a family of
straight lines or of conies touches each member of the family at all their
common points without exception.
a

DIUFBBBNTIAL CALCULUS

374

Examples
Find the envelope of the family of semi-cubical parabola*

I.

Differentiating

w.r. to a,

(/)

*Y

we

Eliminating a between

we

(//),

(i)

and

get

y=o,
which
-

the required envelope.


We already know that
is

the locus of singular points


O f i and also it touches each

y=0

i.e.,

-.

155

iff

cusps,

member

of the family.

Find the envelope of the family

2.

where a

is

a constant and

we

which

is

a parameter.

to m we

Differentiating w.r.

Eliminating m,

is

get

get

the envelope.

Thus the envelope

consists

of two

ines

x0 and x=a.
If

we

trace the given curves


Fig. 156

x+a
we

will find that j-axis

x=a is tangent

(x=0)

is

the locus of

its

singular points and

to each curve.

Considering the evolute of a curve as the envelope of its


2
2
2
normals, find the evolute of the ellipse x /a +y /6-=l.
3.

The equation of the normal


the ellipse

at any point (a cos

0,

b sin

0)

on

is

ax

^Jby^^tf^ip

cos

Thus, (i)
the parameter.

is

//v

"~~sin 0~~

the equation of the family of normals, where

Differentiating

(/)

ax

partially w.r. to 0,

sin

cos

<T +

by GOB
sin

__

is

we have
,..v

"*

l)

375

ENVELOPES

To obtain the envelope, we have


From (ii), we get

(//).

tanvan v = ---

to eliminate 6 between

(i)

and

()*

Substituting these values in


[(<*x)%

(/),

we

get

+(by)$] [(**)* +(*)>)*]* =**-&*,

or
or

which

is

the required evolute.

Find the envelope of the family of ellipses

4.

/z^

two parameter

a, 6, flre

connected by the relation

a+b=c
fomg

constant.

(B.U. 1954)*
eliminate one parameter and express the equation of
have
given family in terms of the other.
c,

tz

We will

&

=c

We
a,

so that

ia

the equation of the family involving on parameter


Differentiating (i) partially n\r. to a we have
y

~0

or

ca
-

which gives

---- ^

/(

Substituting these values in

or

or

which

is

the required envelope

(/),

we

get

a.

OALOTTIJJ8
f'W>

ewehpt

$) ,:$w4';&?

off he family" of

lines
,(0,

a and b ore connected by the relation

the parameters

Here, if, as in the example abo^e, we eliminate one parameter,


the process of determining the envelope will become rather tedious.
This tedio^sn^ss may be avoided in the following manner.

We

pohsider, b, As a 'functibq. of, a, as determined

Differentiating

(/)

and

(11)

and

(iv),

a,

we

(//).

get

b
,

(Hi)

fr6m

da

'

From

to the parameter,

w.r.

we

The equation of the envelope will, now, be obtained by


nating a and b from (/), (11) and (v). Now (v) gives

---xfa+y/b
r =--

y\b

=:xc w or a

^n

'

eliminate db\da and get

.x

x'/a
-^

or

L-

[From
L

(/)
v

and

elimi-

(//)]
/J
v

^^

Substituting these values in

(//),

we

get

or

xn/(n+l)
as the required envelope.
Show that the envelope of a circle whose centre lies on
6.
parabola y*=4ax and which passes through its vertex is the cissoid

Now,

(a/

the vertex (0, 0)

2at) is

y*(2a+x)+x*=0.
any point on the parabola.

the

(B.U. 7953)
Its distance

from

is

Thus, the equation of the given family of circles


2
2
(x at*)+(y-2at)*=^a t*+4a*t

is

or
Differentiating

(i),

Substituting

x*+y*2at*x-4:aty=Q.
t, we get
4a/x 4oy=0, or /= y\x.
this value of t in (/), we get the required

...(vi)

w.r. to

envelope.

377

ENVELOPES
7.

Find the envelope of straight


of the Cordioide.

risht angles, to the

drawn at

lines

radii vectors

through their extremities.


Let P be any point on the curve.
then its radius vector OP=^a(l +cos a).

The equation of the


radius vector

line

P at

vectorial

angle,

right angles to the

a)=tf(l

rsin (0
eliminate

(a)

(r

+cos

cos

from

we

Qa)

tan

cL

get

sin a.

a)^

(1)

sin 9 cos a

(4) gives

and

(2),

we

= r sin

Substituting these values in


2
(r cos 0-tf)
2

+a

Qa).

0/(r cos

(3),

+r

as
.

.(3)

..(4)

0a

r cos

a __

C S

y(r

..(2)

them
a=a.
a =0,

re-write

cos a-\-r sin 6 sin


a) sin
(r COR

r sin

__

..(I)-

a).

different for different straight lines.

is

Differentiating (1) w.r. to a,

'

its

OP is

The angle a

Now,

be

drawn through

r cos (6

To

If a

'

cos

0'

we obtain

sin 2

0_

"~~

2ar cos 0)

'

or
r 2^02_oar cos

which

is

0a

2
,

i.^.,

r=2a

cos

the required envelope.


Exercises

Find the envelope of the following families of lines


2
2
(i) y^wx+^cFm ^ 6 ), the parameter being m
3
3
the parameter being o
x
cos
sin
0=a,
Q-\-y
(//)
(iii) x sin
y cos G~flO. the parameter being o
n
n
(/v) xcos Oi->>sin 0^=0, the parameter being 6
p
(v) y=mx + am , the parameter being m
^cot o~c.
(v/) xcosec o
2.
Find the envelope of the family of straight lines xla+ylb=-\ where
4, 6 are connected b> the relation
1
2
2
2
(m) a^-c
(/) a+ 6=c.
(//) a +6 -c
c is a const ant.
3.
Find the envelope of the ellipses, having the axes of co-ordinates as
principal axes and the semi-axes a, b connected by tne relation
:

1.

',

4.

Show

that the envelope of the family of the parabolas,

under the condition


(/)

(11)

a6=c 2

is

a-f 6==c

a hyperbola having

is

an astroid.

its

asymptotes coinciding with axes.

378

DIFFERENTIAL CALCULUS

5*
ag the evolute of a curve as the envelope of
*~A*I>
9on5idcri
find the evolute
of the parabola y*=4ax.

its

normals,,

Prove that the equation of the normal to the curve

6.

may

be written in the form x sin


f-y cos #+ cos 20=0 and find the envelope
of the equation of the normals.
(P.U. 1944}
7. Find the equation of the normal at
any point of the curve
*=0(3 cos
cos8 0, y=a(l sin t-2 sin 8 f)
and also find the equation of its evolute.
[P.U. (Supp.) 19361
8
Fromac y point of the ellipse x2 /a2 -f >2 /6 2 =l, perpendiculars are
A
drawn to the axes and their feet are
joined; show that the straight line thu*
formed always touches the curve

t2

UAO* +(ylb)% =1
Find the envelope of the curves

9.

x*loP cos Q+b*ly* sin

(B.U. 1952)

6=1.

10.

Circles are described on the double ordinates of the parabola


as diameters
prove that the envelope is the parabola y*=4a(x+a).

2
y =4ax

[P.U. (Supp.) J935]


that the envelope of the circles whose centres lie on the
rectangular hyperbola x*-y*=a* and which pass through the origin is the

Show

11.

lemmscate

r2

13.

=4a2

cos

(B.U. 1955)

20.

Find the envelope of the circles described upon the radii vectors of
'*V+y a /6*= las diameter.
Find the envelope of a family of parabolas of given latus rectum and

parallel axes,

when the

locus of their foci

is

a given straight line

y=px+q.

(P.U. 1931)
that the envelope ol the straight line joining the extremities of
2
a pair of
semi-conjugate diameters of the ellipse * 2 /a 2 +rV6 =l is the ellipse
14,

Shnw

15.
Find the envelope of straight lines drawn at right angles to the radii
vectors of the following curves
through their extremities
n
n cos n.
(i) r=*a+b cos 6.
(//) r =a
:

Gcota

(///)r=<*
(B.U. 1953)
16.
Find the envelope of the circles described on the radii vectors of the
following curves as diameter
.

///= i + e cos 9.
Find the envelope of the curves
(/)

17.

(//)

n-. n cos
a

jfQ,

where the parameters a and 6 are connected by the relation

a+

18.

Show

b =c .
(P.U. 1955)
that the family of circles (x-0) 8 +V 2 =fl 2 has no envelope.
(P.U. 1942)

Miscellaneous Exercises II
1.

Show

that all the curves represented by the equation


*~a~

+ ~~b

VflHhfr/

'

for different values of n. touch each other at the


point x^-y^abl(a
the radius of curvature at this point is

f b) and

that

(DV.

1956)

MISCELLANEOUS EXERCISES

its

37 9*

2.
If the polar equation of a curve be r**a see* ie, find an expression for
radius of curvature at any point.
3.

Show

that

06

cos
x=a(cos 0+0 sin 8), y=0(sin
is the involute of a circle.
4.
Find the radii of curvature of the curve
nV-a 1* 4 -**,
for the points *=0 and x~a.
5.
If a curve be given by the equations

--'

0>,

2=f 2--2,
2 -

-2

of curvature in terms of
to the axis
a curve passes through the origin at an inclination,
of x, show that the diameter of curvature at the origin is the limit of
(x*+y*)l(x sin <t-y cos a)
when x -* 0. Hence show that the radius of curvature at the origin of the curve
y*+ lay 2ax 0,
find the radius
6.

is

t.

If

-242a.

7.
If P,, p f be the radii of curvature at the extremities of two conjugate
semi-diameters of an ellipse semi-axes a, b t prove that

1948
that the projections on the X-axis of radii of curvature at the
of >>=log cos x and its evolute are equal.
points
corresponding
m m cos
and Q is the intersec9.
If P is any point on the curve r =a
/up
tion of the normal at P with the .line through O at right angles to the radius
vector OP, prove ihat the centre of curvature corresponding to P divides PQ
the ratio / m.
cot
Prove that if
10.
The equation of the equiangular spiral is r=ae
O be the pole and P any point on the curve, then a straight line drawn through
Oat right angles to OP. intersects the normal at P in C such that PC is.this
(M.u.y
radius of curvature at P.
8.

Show

11.

Show

that the

normal to the Lemniscate


r

^?

cos 28

at the point whose vectorial angle is */6 is perpendicular to the initial line and
that the centre of curvature at the point at a distance 420/12 below the initial
line.

12.

Show

that a point

*=3

P on

the curve
2

cos 0-cos*0, >'=3 sin 0-sin


and that the centre of
where 0=7r/4, the normal* passes through the origin
(B.U.)
curvature at P divides OP internally in the ratio 4:1.
13. Show that the pole Hes within the circle of curvature at every point
of the cardioide
a(l-f cos 0), and that its power with respect to it is rN3.
14.
Find the co-ordinates of the point in which the circle of curvature of
the parabola y* = 4x at the poini (f*, 2f) meets the curve again.
The circles of curvature of a fixed parabola at the extremities of a focal.
passes through a
chord meet the parabola again at/fandtf; prove that
(Indian Police 1935)
fixed point.

HK

and its evolute at corres15.


p, p' are the radii of curvature of an ellipse
at P make*
ponding points P, P' A is the area of the triangh which the tangent
with the axes ; show that p/p' varies as A*
16.
Prove that the maximum length of the perpendicular from the pole
(B.U.)
on the normal to the curve r=fl(i cos 0) is 4<V3a/9.
17.
For what points on the curve xy*~a*(a-x) is the square of sub-tangent maximum.

380

DIFFERENTIAL CALOTTLTT3
18.

Show

19.

Find points

that the tangents at the points of inflexion of the curve

5xH3y-4a=0,

(B.U.)

on

of* inflexion

x(xy-a

(M.T.)

)*=b'.

Find the co-ordinates of two real points of inflexion on the curve


2
y =(x-2Y(x-5)
and show that they subtend a right angle at the double point.
21.
If O.is a fixed and Q, a variable point on a given circle whose centre
is C and if QQ is produced to P so that
QP=2. OC, prove that the radius of
20.

curvature of the locus of P at the point

P is

22.

If

is

the extremity of the polar sub-tangent at a point

of the

curve

prove that the locus of

r=atan(0/2).
is.

(B.U.)

r=0(l-fsinO).
23.

Find polar sub-tangent and polar sub-normal at any point of the

-curve

r=*-

-show that the polar sub-tangent constantly increases as


increases and the
polar sub-normal attains its maximum value 3>j3a/8 at the point (a/4, 1/^3).
24.

Show

that the circles of curvature of the parabola

y*=4ax
for the ends of the latus

and

rectum have

for their equations

that they cut the curve again in the point (9a, qp 6n).
25. Show that the centre of curvature at every point of the curve

Avhere

it

r=0(0
meets the positive side .of initial

26.

Show

that

sin 0)
line is the pole.

the co-ordinates of the centre of curvature

in any of the following ways

(P-U.)

may be

given

dx 1
rf

_1

Show

27-

that the area of the triangle


2
j/

its

asymptote and
28.

formed "by a tangent to the cissoid

(2a-x)=A,

A'-axis is greatest for the point given


x 3a/2,

by

Find the equations of the tangents to the curve

parallel to the line y=x.


29. Trace the curve

r^=a(l+cos

e)

^ind prove that the greatest distance of the tangent to the curve
point of the axis is >l2a.
30.

Show

of dinate to
is constant.

that the length of the perpendicular


normal at any point of the curve
jt=0(sinh 26-20), y=4a cosh 0,

from the middle


CP U-)

from the foot of the

MISCELLANEOUS EXERCISES
Show

31.

that for the curve

0)-cos
,

381

II

y/a=sin

0,

0,

the portion of the A'-axis intercepted between the tangent and the norrral at
any point is constant.
32. Find the nature of the double poiut on the curve

and show that

it

has two real points of inflexion.

33.

Show

that the pedal equation of the curve

34,

Show

that the tangential pedal equation of the curve

+(;/*)

<*/*)

is
1

la

sin2

cos 2

$ -f 1 Ib-

on the tangent at each point of curve, a constant length be


measured from the point of contact, prove that the normal to the locus of the
point so formed passes through the corresponding centre of curvature of the
35.

If

given curve.

Show

36.

that there

as the parameter

is

a curve for which the circles


cosh a+l=0,

U-a) 2 + y 2 ~2y

varies, are the circles of curvature.

Prove that the distance from a fixed point P (x y e ) to a variable


on the curve f(x,y)^V is stationary when PP is normal to the curve

37.

point

at P.

The tangents

38.

at any point
x=--a(0

'0'

of

+ sin 0),.v=0(Hcos

0)

normal to the curve at the point where it meets the next span on the right "
cot (0/2)^=-- n/2.
prove that

is

Show

39.

that the locus of the intersection of perpendicular

the Astroid

x=^a cos 3 0,
is

y=a

the curve
2

2(jc*+/)=fl (*

- y2
i

2
)

and the locus of the intersection of perpendicular normals

to>

is

(D.V. Hans., 1959}


the curve

For the curve

40.
find the

tangents

sin 3

x^=a(2 cos r-f cos 2f), y=a(2 sin /-sin 2f),


equation of the tangent and normal at the point whose parameter

is t

and show that

(0 the tangent at
are -i/ and?r-i/,

P meets

the curve in the points Q,

R whose

parameters

QR is constant,
ihe
tangents at Q and R intersect at right angles on a circle,
(Hi)
(j'v) the normals at P, Q and R are concurrent and intersect on a concentric
(11)

circle.
2
P, Q are any two points on ay =x* such that PQ always passes
9
8
a
or
fixed
on
the
curve
)
show that the locus of the
(at
through
point
lying
point of intersection of the tangents at P, Q is the parabola

41.

42,

again at

where

Tangent

OX is

show

of the
at any point
that
cot 2 LXOP+ tan

Jf-axis.

Folium x*+y*=3axy meets the curve

DIFFBBENTIAL CALCULUS
43.

Show

that

x=0/4

is

a bitangcnt of the cardioidc

r =a( I

414.

The envelope of the


x cos

-cos

(Birmingham)

0).

straight line

#+y sin

4s the curve whose polar equation

?$=0(cos

is

//(I-*) ~*nl(l-n)
45.

Show

that the radius of curvature of the envelope of the line

XCOS a+ysin

a=/(<x),

that the centre of curvature

is the point
*=-/'(<*) sin a /"(a) cos a
y=/'(a) cos a -/"(a) sin a
Discuss the nature of singularity at the origin of the curves,
a
2
5
2
(i) j>a=*(jc -l).
(//) x -0(x --a>0 ==0.
Show that the curve
|

(M.U.)

>46.

-47.

G0.J7.)

bytx* sm2 (xja)

has a cusp at the origin and an infinite series of nodes lying at equal distances
from each other.
Trace the following curves
48.
:

(i)
'(///)

49.

r-fl(2 cos e-f cos 36).


x=0[cos f 4-log taa Jf]-

(11)

y=a sin

>^-2cV+a2^ 2 -0.

/.

The tangent

the parabola

is

to the evolute of a parabola at the point where it meets


also a normal to the evolute at the point where it meets the

evolute again.
50. If p be the radius of curvature of a parabola at a point whose distance
measured along the curve from a fixed point is, s, prove that

51. If p and p' be the radii of curvature at corresponding points of a cusp


evolute, and p, q, r first, second and third differential co-efficients of>>
with respect to x, prove that

.and

its

P
9
2
[Hint. P '=</V<ty -]
52. Show that the radius of curvature at a point of the evolute

-curve

rn

-corresponding to the point

53.

(a)

lias a single

~an cos n$,

(r, e) is
/

of the

l~l\*

'sec/iO tan n$.

Show that the curve


x*-2x 2 y-xy*-2x 2 -2Ky+y 2 -x+2y+l=*0

1).
cusp of the second kind at the point (0,
Show that the radius of curvature of the curve /(r, 8)=0

(b)

is

rcosecy

the chord of curvature perpendicular to the radius vector


denotes the radius of curvature.

is

2p cos ?

where

(D.U. 1956)

ANSWERS
467.]

JPages

Page

4.

Ex.

Page

3.

(//)

a,

(iv) nit

i) 1,

n being any integer.

9.

Ex.2.

Page

0.

(/)

1-710.

10.

Ex.2.
Ex.3.

Page

x-1 <2.
(ii)\x-l\ <$.
x-l <e.
x-2 <5.
(iv)
(in)
(i)-l<x<5. (//) -3<x<l. (in) -l<x<3,
(/)

14.

Ex.4.

(J) (-1,00)
The entire aggregate of real numbers excluding the numbers
is any integer.
(2 + l)ir, where
(/') The set of intervals

(')

when n

is

any
36.

Page
1.

(I) 7T/6.

2.

(in)
<v/i)

(*)

[!.

(iv)

(/)

Page

1].

0]

27T/3.

(IV)

[-co

(v)

+ lir),
J

>

-1],

[1, oo

].

(vi) [0, oo ].

(viii)

[1, oo

].

(ix) [1, oo

]-

(^"0

The

numbers (H+i)^.

(xiii)

[1,

Increasing in
in

and decreasing
!)TT,

oo

[00,00].
,

and

0]

in [0,

oc ].

in
(v)

(ii)

Decreasing in

[00

continuous.

6.

7.

Continuous for every value of

11)

x=0

and x = l.
x.

57.

-f

6667.]
(/)

continuous.

(//)

discontinuous,

333

oo

].

[0, oo].
i*v)
Increasing
Increasing in the set of inter-

discontinuous.

(i)

Discontinuous for

2.

(2n+^)fr] and decreasing in [(2n+|)7r, (2fl+f)7r].

3.

[Pages

1].

in

51.

2.

].

entire aggregate

[0, QO],

Decreasing

vals [(2n

Page

[-1,

(Ill) 7T/3.

set of intervals [2mr,

3.

(//)

The

set of intervals (2mr 2n


~" 2
L (^O [~- >
]^ [

<xiv)[-oo,0],

(Hi)

-7T/4.

numbers excluding the

real

[00

(/I)

(i) j(0, 1).

[0, 1).

The

of

integer.

(ill)

continuous.

384

DIFFERENTIAL 'tEBptHTMf*
(iv)
(vii)

Page

Page 75,

4.

(vi)

discontinuous.

discontinuous, (vw) continuous.

(/)

13.

() -,V-

4-12.

Ex.1.

Page 76.
Ex.1.

(/)

2,

-2x/(x+3)
2

2
.

2(l-x )/(l+x

(//)

-1/2A

2 2
)

(n/)3x.

Ex.2.

(/v)

-1/V2.

78.

Ex.

Page

continuous.

(v)

74.

Ex.

Page

discontinuous,

[Page 74-89

3.

2x+>>+l=:0,

4Ar=^-j-4.

(0)8, 2; 2,

(ft)

79.

Ex.1.

2.

-1/16, 1/32
(c) 1/4, -1/32
1/2, -1/4.
-4, 3, -4; -2, 2, 2; 2, 7,8.

-1,

2.

Ex.3.

Page

83.

Ex.2,

(i)

(Hi)

Page

(x-l)/2A
(3x*+x-I)l2x%.

(//)

(2x~*+5x~l'*)/l2.

(iv)

(1+7^/6".

85.

Ex.

2.

(i)

2x+5.

(ii)

2(x+2)(3x+6).

(Hi)

Page

86.

Ex.2.
(Hi)

Page

(/)

-(4x+6)-.

(tf)

-x(x+2)-.

[(x'-

87.

Ex.2.

(iv)

(v)

Page

(2x+3) (10x-3)/2x^.

89.

(i)an(ax+b)-

1
.

2 2
() -2x/(l +x )

----_. --r

,..

(w)

- ----

[Pages 91-97

ANSWERS

91.

Page

3.

(m)

mx m ~i

cos

(v)

(x cos

(v/ii)

Ex.

Page

(ii)
...

(in)

{V)

sin-^*
xm

cos x.

sin x)/x 2

(ii)

(iv)

sin 2x.

(vi)

sin x.
2

(ax+b)(ax*+2bx+c)~^ sin V(^


2
(m cos x w sin 2 x) sin- 1 * cos""" 1 *.
4.

(/)

tan

r.

(//)

wx.

sin

cot

+2feo;-f

r.

c).

tan

(ii'i)

r.
'

.4 cosec 2x cot 2x.


(/)
x tan 2x+2 sin x sec 2 2x.
sin 4x
8x sin 2 x
r
2
4 cos x sin 2x
1.

cos

,.

-;

(iv)'
v

sm

V[(cos 2x)](cos x+sin

(V/)

* 8e

2x.
2

x
'

x)'

93.
(i)
(i/i)

9 cosec 3 3x cot 3x.


\b sec ^(a-\-bx) tan

(//)

^a^/[seo (ax+b)} tan (ax +6).

a
N/(

+^)/v/ (+*x

tan (cosec x) cot x coaec

sec-(cosec x)

(/v)

Page

(/)

92.

Ex.

Page

Ex.

(vii)

)-

x.

96.

Ex.

1.

(i)

(/V)

/X)

Ex.

Page

385

Vx

'

1+cosVx2

...

__ 2

8l

'

2.

1.

97.

n
Ex.
B

(/)

,->

cot x.

(ii)

,...,

ain

(in) e

cos x.

cot (log x)
(v >

(wi).

sin (log
v
" x);
----

secx.

....

(viii)

---ji-f-r,

e 2 ".2x log

x-e-*

,.

(w)

x
.

lo

DIFFERENTIAL CALCULUS

386

Page

97.

a 2 cos 2 x

sin 2x).

-!

sin 2

(x/v)

100.

Ex.

tanhx.
() sinh 2x
sec 2 x tanh x 4 tan x sech 2 x.

(i)
(///)

Page

ft

/V(**)

2
log a

Page

Pages 97-105]

102.

Ex.

4.

[(log cos x)/x

(i)

(cosx)

(/v)

(tan x)

eo

tan

x. log x\.

cosoc 2 x log (e cot x)

tau x
(cot x)

(v)

+ lo
-^)-

(logx)*(log>gx
sin

x/

i;a
,x)

(co

-x2

(2

1-1
sm

x)

3
'

-1

log

f"

sec 2 x log (e tan x).

'

.v

1
'

9x 2

4.v_

2(1

"3(2~Tx

-r)

4
)

+4(3- j
_

(vm)

Page

+ 15jc* + 36)/3(JC

(2.x*

(v/i)

sin

.v

;r
.

.r

log

+3)

'

X* fcot .vf

f 4).

(.v

log .v)-'

\-

log

<?-.r|.

104.

Ex.5,

(i)

(i/)

(///)
:2

'

Ex.6.

Page

CO

1.

(//)

105.

Ex.

3.

(/)

2x cos x 2

m x cos x.

(ii)

sin 2x.

(///)

l/2y'.v.

'

(iv)

(v)

eV/Jf /2 v/x.

(v/)

]/(l

+x

2
)

ANSWERS

[Pages 105-106

Page

387

105.
-.

*2

2.

sec 2 x\/(cot x)
~

5.

- --1

1.

2x e

4.

2jcsec 2 x 2

3.

'.

(jc

.,

cos x

sin x).

l
2

Page 106.
COS- J JC

Xy/( I

'

(l-X
^,^

o
/

/'/I

"

-^9v

9.
"

10.

(tanx-

12.

X2 )

"

2 3/ 2

log

13.

(1

~\'"*i

(i

.'

.(5x)

15.

A'

16.

"
l

\"

f 7.Y)-/

7
(

6.7
'"1

}-7.x
"

V2 3(

-1.

14

"

11.

hotl(f.Y

i.

y
A

*'j_/

7"'""'

-"-Vx)"

'.)

v)~4(l

8.v"":V(r

!-3.r)

"f>(l

2
).

"

17.

-x log
.v-

log ex\x

.Y

(xt)

l
.

cosooli

18.

sin-'.v.

-.Y-).

.'5.Yv(l

'

.v.

I
'

.x

v/(l

2a

21
"'

^/

(l

log (^

x)
(a-

26.

(sin.v)

.v

log

-j

-b

los log

.v.

sin

lopRinX

r/

.cot.x.lo^lO.

xf

log sin

cot

.v

^cos-

^.

axl
[cos (x log x)

log

^v-2ax

sin (x log x)].


2fl

27.

win .v)v/(cos 2x)'

.x

cos* jc)-j-4ft cos


1

.cos""

25.

10

.v

^)

loo;

_____

24

22.

v
23.

(cos

sech ax.

28.

29.

[Pages 106-111

DIFFERENTIAL CALCULUS

388

30
W

*2 ~1

~'

31

32

'

2x cos

Page 106.
9x 4

33.

sin

(3*- 7)

log (1

-5x) x
cot S

eax [a(l

34.

+x2

cos (6 tan- 1 *) ~fc sin

cosec

cot x,e

w/sin
7

oosec 2

37.

a*

sinh x+xfl* sinh x

38.

V3/(sec

cfcth

^(sec

x)

tan^x)]/^ +x

(fe

2
).

_/2/sin
'

/r1

cosech 2

36.

*~

w/sinx"

cot x.

log a fxa* cosh

x.

3x).

39.

40

'

(1-x 3 )"^.

41.

Page 107.
42.

(i)

43.

(I)

45.

l/(x*+l).

(H)

x/(x^l).

0.

(ii)

-tan

x cos x/log

sin

sin

_1

-,

^o,

~-

47.

48.

1,
(sin

x+x cos
-__--

(x cos x-f-sin

(sin x)

_/(l-x2

(1/1)

tan

t.

x.

3/.

x log
.

x)

log sin x)

49.
-1.
x(2+2tanlogx+secMogx).
tan * [sec2 xl g (1 S ^)+tan x(xlogx)^]

(logx)

cog

cos""1 x)

Page 111.
1.

/'(O) exists

/x

but/

2.

Differentiable at
the origin for

3.

/(x)

is

5.

/(x)

is

6.

x=0

m > 3.

does not.
for

m>

2.

/'(x)

is

continuous at

continuous but does not have derivative at the origin.


continuous but not derivable at the origin.

Continuous when x

for other values of x.


7.

(0)

is irrational

Differentiable for

or zero
value.

no

and discontinuous

ANSWERS

Pages 112-120]

Page

112.
10.

Continuous but not differentiate for x=a.

11.

(i)

12.

Discontinuous at

derivable at

Continuous.

(//)
;

Discontinuous.

derivable at

continuous but non

2.

Page 115.
5.

Page

116.

Page

119.

-3/2.

^v'r
4

..?.

L (x2)*+

10

L(X~I)"

Page

(x+2)+

J'

f
r

(-- 1)"

^ ( X __ jp,

(v

+ i-

^])

(jr+o)"* J'

120.
3.

(-!)

!f

w+1

fl)

(X + tf)"*

2 sin [(n

+ l)

cot~ 1 (

-]
'

(+l)^, where

sin

7.

(j)

l)-i( w

(ii)

l)

n -J

(M

1)

1)

sin" 6 sin nd, where

cosec" a sinn

d=cot~ 1 x.

sin n&,

where
[(x

coBa)/sin

390

[Pages 121-126

DIFFERENTIAL CALCULUS

Page 121.

+ \n it) - (3/J) sin (3x -f


() J[2 n cos(2x + IWTT) -f/$ cos(4x + JTT) +6"

2.

(i)

<7'0
(iv)

-|

sin.

(x

J/ITT).

cos(6x

+ |/wr)]

3 n cos (3.v+|/i7r)--5n co6(5jc+^w)l.

TV t 2 cos (x+imr)

[3e*-4-5^"cos (x+w tan" 1 2)


-f

(v) 4

(4x+n tan

17^'cos

4)]/8.
i

(3x+n tan

[25*" cos (jr+n tan-i i)- 13-" cos

|)

Page 123.
1.

e'f ^

(/)

(iii)

.v

cos

(.v

-v

|TT)

-f-

-I

/l2

*
-i-

)2
.v

^"-.?

f-3x 2

cos [Ar-j-|(n -l)7r|

3(W- 1)^008

+ J(-2)7T]4

[X

n(/i- ])(/i- 2) cos


(///) 0"+'-.

(iv)

*.

fi

e"[log

x-pfjX

V3

'

jr

f-T 3 2

!.

5.

Page

([
2

x-)y n

x ;vh2

x~*~
1

I)""

!-....-}-(
4.

3)ir|.

(i~l)

"<'

-v

"|.

+(2tt

125.

23-

Page 126.
4.

^ f n(0)=0;^ ta+1 (0)=/n(L*-./n )(3


V
yan (0) -0 v, -+1 (0) = - 1 )" I

W-

2
.

^(0) = w-(m= -

2 2 )(m2

-4

a
)

(2n

- (2/i - 2)]

.... [w-

.... [(2

4.

>Wi-=0, ^ J n=(-l)"- 1

22

42

62.

.[/

.[(2n~2)

126.

56.

6.

-1)*- m*J.

[(2

-3*)(/i|S-5).

Page

[x+K/i

X2

2
)

-(2n-

+/].

4-/n

.(2-2)

2
|.

2
.

I)*].

ANSWERS

Pages 127-156]

Page

391

127.
If n

9.

Page 135.
Ex.3.
Ex.4.

is

even, y n (Q)

-0

if

odd, y n (0)--~n

is

(3/f- 5)

= (6-V2l)/ti.

(n)V5.

(/)|.

(//7)log(e

-1).

139.

Page

For a ^1, the function

2.

the function
3.

is

is

for

steadily increasing;

d<[

1,

00

2]

steadily decreasing.

Increasing in

2,

1]

and

[0, 1]

decreasing in

1-1,0], [l,oo).
5.

and

Increasing in

4.

Decreasing

[2, oo

i, 1

in (--oo

decreasing in

and

2]

fO, 2]

(00
;

1],

and

increasing in

[1, oo
[

).

2, 0]

).

Page 140.
6.

o.

:it>,

Page 144.
2

4.

ar

A
cos />.V

-& 2

--~Jt 2
-^
2t

-f

fl(l*-36)^A' 3,
V

..+^j

fl

24ft)

-!

....

J/l

^cos^ftflx+wtan--

)-

150.

Page

Ex.

2.

i:m.

Ex.

3.

max.

Ex.

5.

-s.

Page 153.

Page

55/27,

8
1

min.

1109,

21).

-10,

-27.

156.
1.

2.

3.

min. 37.
(/) max. 38
Max. for x -2a and minimum
Max. forx--l, min. for x-^2.
;

(11)

for

min.

4.

2^3/9.

5.

max. 64, min. 50 least 0, greatest 70.


max. for x--l, min. for x--=6.
max. for x -=\/3 min. for x-= */3.

6.
7.

= 3</.

e-

12.

rain,

13.

max. values

value

miii. values

integer,

.v

11.

14.

4.

and

log [(ae

e) log tfj/log a.
'

(27T+3v/3)/6

(87r+3y/3)/6.

(47T-3V3))6 (10:: 3\/3)/6.


max. for JC=/ITT where n is an odd positive or even negative
and minimum for x~mr where n is an even positive or odd

negative integer.

392

DIFFERENTIAL CALCULUS

[PagOS 156-169

Page 156.
min. values :-J(5ir+3 x/3+2),

15.

max. value
VT
17.

~4

mm. value V3
.

,.v
(i)

least value

least value
18.

(ii)

min.
min.

any

is

0,

2/3-v/S

-A'-

max. values

1-

-V

-V -.

max. 1,2/3^6.

max.

greatest value

min. -1, -2/3v/6


(a* +b%

greatest value

max. values

(/)

(in)

where n

min. values

(if)

(27r--3 v/3+2).

(-77+3^3-2), ^

max.

(a$

+b%

0,

2/3V3.

fi

integer.

Page 157.
max. c*l(a+b).
The required values are the roots of the quadratic equation

20.

21.

(a~-r-*)(b

r~*)=h*, in

r.

Page 163.
1.

9,6.

4.

V(40/3),

5.

9.

Sq. whose each Ride =/2ff.

3.

vW3), h/(>/3).
Diameter of the semi-circle = 40/(Tr +4)
Height of the rectangle =-20/(7r+ 4)
Breadth y^ depth \/%a.

15(^

10.

miles per hour, 19800(|-)^ rupees.

18.

Page 164.
21.

3V30&M-

22.

26.

(3v/8ir)*.

28.

Length 2
and girth 4J ft.
30.

ft.,

girth 4

a-ft.

25.

3^3/4.

2V37T/27.
ft.

yes, length should

now be

If

Page 168.
Ex.3.

(0-f

(H)**l-2e.

(Hi)

(/v)

()f

(tfi)

-|.

(iv)

i.

(vi) 2a/6.

Ex.4.

(/)|.

Page. 169.

Ex.

5.

limit is

1.

(v) 2.

ft.

Page 170.
Ex.3,

T\.

(i)

1.

(iii)

cofc itx

Change
(v) |.

(0 i

to cot \Ttx.

(v) 4.

173.

Ex.

Page

-J.

(//)

2/7T.

(iv)

Page

393

ANSWERS

Pages 170-188]

2.

(i)

0.

(ii)

3.

(iii)

1.

2.

(i)

0.

(ii)

1.

(iii)

20/7T.

2.

(i)

(ii)

0.

(iii)

3.

(i)

(ii)

e' 1

(iv)

1.

(iv)

2
7r /6.

(iv)

1.

1.

(v)

0.

(vi)

174.

Ex.

Page 175.
Ex.

Page 176.
Ex.

1.

1.

(iii)

Page 177. [8-7].


i

(v)

Page

e.

e.

(vi)

(vii)

e*.

(viii)

J.

6.

e.

177.
1.

e~\

2.0.

-5.

3.

9.e~*
2A

13.

e-"

17.

e~~

-iV

10.

4.

5.

11.

\4, e

1 2

15.
18.

*.

e"

16.

".

alogrt.

22.

'

<>

A /37T/6.

-2.

7.

e~^

12.

2.

2 /7r
21.

-1.

1
""

01 -.

a^ b

4.

*-'*
20.

19.

2.

23.

tog (e/*) log be.

1.

Page 178.

...aj\

25.

-<?/:>.

26.

29.

aV.

30.

24.

(a,a,<i 3

28.

-iflV.

33.

Continuous at the origin.

lie/24.

27.

f"(a)l'2f'(a).

16.

31.

Page 187.

7.

2.v-

*-

g-jc+

*4

^5

x *2

Page 188.
Y4

Y2

11.

12.

i+x-Hg*

12
-!+ A

14.

log sin 3

+ (x- 3)

13.

cot 2

^^

J-+J

cosec 2 3

+
C ot

3 cosec 2

3.

TB

394

DIFFERENTIAL CALCULUS
J

15.

Page

14-f29(x---2)+16(;c-2) 4-3(*-2)

3
.

188.

<ReacU* ->*

Page

LP ft ges 188-215

cos

cos x

in place of e-e*

).

202.

a
ae * sin

2.

fofl;r cos

&>>,

--,

eB -' -g*-",

(i)

ye^

(ff)

-.

&-*[yx+y-l].

log

'

(i-xy)
4.

3'

(1-xy)*'

(l

0.

8.

]/c*r.

Page 203.

n_3.

2-

Page 205.
Ex.

1.

sin

0,

cos

0, r

cos 0, sin Q,

cosec 0, cosec 0.

Page 209.
2.

4%.

6.

-3J.

Page 215.
1.

3.

4.

(M cot
(

(oosyxys'my)lx

2.

4x-\-2y.

x cos M

+*)(! +J

sin

y+z

sin v sin x)/(cos v sin y).

)(25 f>2)/( 1

-xy)*.

5.

()

[>;+x cos (x-y)-\l\x+x* cos (x-^)J.


v ~l
v
(cot x yx.
log y)ylx (y log x log y-4

..

sin

(i)

>'(>;

sin

cos jf(sin
(/v)

x_+cos x
log cos

log_sin

jc

X^-* log y,lx(x-y


2
1
j(tan
x)*- sec x
""

_
8x

~~dFidz

'

j')

cos

>>)

log x).

cosec*x log y(y) e

8JL

8Fl 8y

8y~~

8F/8z

'

"'

1).

395

ANS\VKKS

Pages 216-233]

Page 216.
14

Page 228.
1.

min. at

(i)

(5,

min. at[J(6

(ill)

(i7)

).

rain, at (0, 0) if

t/

>

20), J(0 --26)].

at (2, i).
max. at (2, 1).
(v) min.
at (
2, 0).
1, 0).
(v/i) max.
(vi) min. at (0, 0), (
extreme
value.
(v//7) No
1 + 4m +2/i)|,
(ix) max. at [-&n( I +4w -f!2/i),
min. at [-^-71(74-4^ + 12^), ^7r(l 4 -2m-f -2/?)],
where m and w are integers.
(x) max. at (|,
).
2.
max. value 112 at (4, 0) min. value 108 at (6, 0).
(iv)

Page 229.
3.

2(x

x 2 )(/;/

/2
1

MJh)l\ 2( n h ?L2

2
1

i)

Page 232.
1.

30'-.

(/)

2.

2
fl

3.

2k^(a*

3a*.

(//)
2

(Hi)

3w 2

and Ja (max.)

(uiiu.)

ab~\b*).

4.

......
'

'

v/14

/14

v'

Squares of the semi-axes are tho roots of the quadratic

6.

71
'

""8 O
'2

'

Miscellaneous Exercises

Page 233.
1.

Zero

2.

is

+ i,

the only point of discontinuity.


are the only points of disis any integer,

whore w

continuity.
3.

f2/;H-l)/(l- -2/w), in being

4.

1,

5.

(/)
(11)

(Hi)

any integer and

-1.
All integral values of ,v.
All integral values of x.

x-0.

x=

I.

DIFFERENTIAL CALCULUS

396

[Pages 233-260

Page 233.

Page

6.

jc=-0,

7.

Discontinuous at the origin.

1.

234.

(CA)+ sin

sin

(B- Cf+sin

sin

(cos
_.

,4(c~os

cos
C)
......

',

_ lf x _
Jv
JA -A-v3

16

3a 4
^

r6

(7--cos 5)'

(-l)n!
J.W.

where

tan

^=a

sin

a/(x+a

cos a).

Tfg

Page 236.
30.

(i)

31.

|.

33.

(i)

34.

-2.

35.

(i)

!;(//)

-1

-1.

;(ffi)

32.

()

1.

continuous,

i.

(Hi) 0.

2a/b.

(//)

discontinuous, (HI) continuous.

Page 237.

4-3v2

--

max. at 7T/3 and min. at


min.
4/^ max. and

38.

40.

(3db)l(a3c)

43.

min. (max.)
47.

if,

ad

5TT/3.

a max.

is

(min.)

and -(b+d)l(a

c) is

Height=2(3i;/7r)*,radius

= (l/V2)(3i;/7r)^

where

is

the

volume of the cone.


50.

27r{l

51.

The

-(V2/3)} radians.

vertices of the rectangle are

('

The greatest value

52.

2a \

2a\

2a \

C^'-VS)' (3-

3>
V3')'
ax being the equation of the parabola.

(3'
y* =

is I

least value

and the

is

(l/e)

Page 260.
.

(i)

(ii)

;c+.y+a=-0, x->> -30.


2
7 2
a t/(x-x )=fr 1 .x 0' -/).
v'^-a 6
/

6Vx+a

(Hi)
.

(v)
(vi)
(vii)

(viii)

(ix)

be, is positive (negative).

3lJC8>^+9a=0

8x^31^ + 42^-0.

x .y^O.
x+y=--3a x y+a=Q.
x cos 3 5+^ sin 3 0=c x
13x-~16);=2a

x+^^0;

sin 3

8
-v cos ^

+ 2c*

cot 20 --Q.

ANSWERS

Pages 260-279]

397

Page 260.
5.

(0

(0, a)

(a, 0).

(11) (

(^.T|->

>

/2a, y*a)

(V4a,*/2a)

(-1-. *

jc2Qy=7, 20* +7 = 140.

6.

Page 262.
"
'

Page 264.
1.

(i)

ir/4.

(/)

m h+m(a
2

3,

tan" 1

(ii)

/z0.

b)

two

tion in, m, are the slopes of the

V 16

The

roots of this quadratic equaaxes.

Page 265.
a sin 2

3.

0,

a tan

sin 2 0, a sin 2

cos 0, a sin 3

tan

0.

Page 270.
(00/2. () 7r/2-fw0.
1
(i) ir/2. (ii) tan- .?.

1.

4.

(v)

(m) Tr/2-0/2. (/v) w/4-fmfl.


1
2 (iv) tan- [2^/(^
(/) ir/2.

2ir/3.

(v/) 7T/2.

Page 272.

6.

(i)

7.

()

0.

0/02.

/>-<*>/(

a^ 3

2fl

cos s |0 cosec

(i) 2fl

4.

g--^ ^
ft

(//)

1+0M2+20

C")

tf

_2

sin 0.

2
).

Page 273.
2

9.

(i)

l/2J
1

(iv)

(v/)

(viii)

2
.

= l/r 2 +a 2 /r
2
8
=/J [o m +(l

l/p
4

=^l/r +l/a
e 2 -!
2

(//)

4
.

r 1 .-=(#s

yj^rsina.

-a2 +2ar)/>

(v)
J

)r

J.

(v/7)

3
.

(ix)

Page 279.
'

x
cosh

CO

'

,
<

2.

I)]

DIFFERENTIAL CALCULUS

398

[Pages 279-29(>

Page 279.
4.

2
2
2
2
v'(^ sin 0+6 cos 0).
2
2
cos
3 sin
v/(a cos'

(/)
(ii)

(iv)

v/2tf e

(/)

2acos

5.

sin 2 0).

sin J0.

2tf

(ill)

a m lr m

(v) 00.

0/2.

(//)

ay (sec

<V(l+

(iv)
(vii)

f/>

(v)

)-

a(0

l
.

+l).
m m~

v/2a

(v//7)

20).

cota

a cosec a e G

(111)

tf(02+ l)/(02 -1)2.

(v/)
l

/r

Page 288.
Concave downwards
Concave upwards in

2.
3.

and upwards in [ir, 2ir].


concave down2, 0] and [2, oo)
Inflexions at (-2, 198), (0, -20),

in [0,
[

TT]

wards in ( -oo, -2] and [0,2].


(2, -238).
Concave upwards in ( oo,
concave down4.
1] and [1, GO]
wards in [ 1, 1]. Inflexions at ( I, 2e) and (1, !(>/<?).
Concave upwards in [0, ir/l] and [5?r/4, 2ir] and concave
5.
-

downwards
6.

in [ir/4, 5ir/4J.

For x--^blZa.

(/)

(Hi)
(v)

j-

(v/)

(0, 0), (1, 0),

(ii)

(\V- I)a and <r/(2


For x
and itf \/3.
For x(5, 0),

("')

For x

(v/7)

-ttf/v'-.

For

(/.v)

x=

(-v/)

ae* }

(/'/')

are

x-\-y-.

(///)

are

x----

9x

Inflexional tangents to (x/) are

0,

5 and

(\ \>

"f

.x

y'*>

---2j'l--

0.

,'{(9.x+<>)>)

3 v /3v

{-

0.

Page 289.

...

For*-

2,

(-4

i-

v 'l8).

(.-..-.-l-T).

f].

(0,0).(///)

Psge 292.
Ex.

(i) c

(zv)

<

sec

'
1

tan

(/'/)

i//.

(v)

>{i.

4aoos^.
2fl

see 3 !/-.

Page 296.
at.

2.

(i)

-1/2.

(//)

1.

(HI)

3(a.vj)

(/v)

'

Y
/

(0, 0).

(.v/7)

9.

3\

(\ae

and

*
(1,3)-

Inflexional tangents to
Inflexional tangents to

()).

\ '3).

(v//7)

(1,

(0, 2).

-*(

Inl.

/:K

ANSWERS

Pages 297-321]

399

Page 297.
10.

W-la.

2tflb,

14.

[-(log2)/2, l/v/2].

20.

(8,

3).

Page 298.
9

23.

25.

(c -+s*)lc.

Page 302.

(r*+aW-n*r*f

,
*

'

r*-r*n*+~2a*/i*

/(+ l)r"-i.

(i/i)

3.

o/2.

(/v)

vfy-'-o").
3

5.

(i)

2p

/o

(")

(f'O r*//</>.

(m) a

6.

7-

(/)

Page 303
10.

--a.

12.

(3<i/2.

(/)

.U

(\/-.

(//)

1/3).

Page 309.
9.
11.

(a-\-b)(x*+y

2
)

O//)2 _
^

* 2 -fy 2

(0)

>o2
;

Page 315.
Ex.2.

Page 320.
1. * = 0,

y U/ = 0.

.r

(/)

y-v-

A*^ : ui,y^().

5.

.v=(),

7.

y=.Y

-2, y---~.v--3.

9.

y-

10.

x,

3,

y=0, y

y = x + ].

y ---x+2,

.Y-fy =

= a.

\-y

Lj-</.

3.

r.v--

(//)

13.

K&'-2fl-)=<>.

)'

/7

(Hi)

x-:2, jc=3, y=-3.

x=a. y-6.

6.

y^xia, x^=a.

8.

.x

I
i

=0, y -0,

().

y ---(),

.Yfy=-h v /2,

.v

i:l,y=

Page 321
13.

y = (>,

15.

x+y=2,

16.

yA'= 1.

14.

y==x-f -2, y=2jc

*=!.

4.

17.

3=0, JC + 1=0, y-jc


x=0, y=--0, 2,y--,?x =

19.

x-hya^0.

20.

4(y~x) + l=0, 2(y f Jf) = 3, 4(y + 3;c) +9=0.

fl.

4.

18.

/c.

x-\-ajX.

2.

x=l.

12.

4.

y = xV2.

11.

;c

y =-o

400

DIFFERENTIAL CALCULUS

[Pages 321-334

Page 321.
21.
22.
27.

x=y+
x=0.

Page 325.
1.

y^a, x=0, x+y+a=Q.

3.

3(>>-x) 4-2=0.

5.
6.

X4-.X 4-1=0,

7.

;>4-2

8.

10.

x=l, x=2, x 4-^ + 1=0.

2.

3y+

x=

^=2^ + 1, y=*2x+2.

jp=x t

Page 328.
5.

7.

6.
3y+*=l.
y
8(^-x)+7=0. 2(^-3x)+3 =0.

8.

x3

9.

y=2x8 4-l, y=

10.

x 4-^=0, 2x

106^-381x4-105=0.

3(2^+^)4-5=0,
3
x=0.

6x 2 .y4-llx7 3

6j
2*

1,

x+2j=0.

x^=2j,

1=0, 2x->3y4-3-=0, 4x- 6^ 4-9=0.

3^

Page 331.
Ex.

The curve lies above or below the asympx4-y


x is positive or negative.
The curve lies above the asymptote both for
x-f-j; 4-0=0.
2.

(f)

tote according as
(//')

positive as well as negative values of x.


(Hi)

The curve

y=zx\-\.

accordiug as x
Page 334.

above or below the asymptote

lies

positive or negative.

is

+ ^2r sin

1.

tf=rsin

2.

3.

rcos6b=a.

4.

r cos

5.

2ram0=a,2e=ir.

6.

0=7r/4.

7.

r sin

rsmf/ 6

5.

00,

/>

\
9.

10.
12.

15.

(1

0).

r sin

8a^=0.

cos 0)4-2a=0.

7r(r

-- )=
W TT

>v

7i0=/W7r where

n cos
is

rcos0=0.

AWTT

any

a=rsin(0-.l).

System of

(04-i7r)=0.

0a=;Q

parallel lines

where mis any

integer.

integer.
11.

0=0.

13.

y+a=0.

y~ae

W7r

where n

14.
is

any integer or

zero.
16.

^4-11=0.

17.

-^T\V

19,

a=2r

(ft\

==rr

(cos

sin

ei

6 )>

where

sin 0), a4-2r(oos

6 l is

any root of the equa-

0+sin 0)=0.

ANSWERS

Pages 338-347

401

Page 338.
I.

4.

x=0, >=0.
>>==*.

2.
5.

Page 339.
1.

*=a.

(y-l) =
2(*-2a) =

3.

x=0.

y=x.

2.

V3>'-v'2(*-*)

3.

bx~ay.
(x-2).

V3(y-*).

Page 341.
1.

Cusp at

(0, 0).

3.

Node

at

(0, 0).

Node
Node

at (a, 0).
at (2, 0).

2.

Cusp at

(0, 0).

Page 342.
4.
5.

(a, 0) is a node, cusp or an isolated point according as


than, equal to or greater than 1.
7.
Conjugate point at (a, 0).

6.

b\a

is less

Conjugate point at (2, 3).


9.
Conjugate point at (a,
10.
Cusp at (0, -4a).
llr Cusp at ( 1, -2).
12.
Cusp at (1, -1).
8.

b).

13.

No

14-

(0,

multiple point.
0) are triple point

(fa,

0) are

cusps

node at

(-a,0).
15.

(0,

a) is

point according as

a double point

being a node, cusp or conjugate

b>a, b=a or b<a.

-a

16.
17.

18.

Page 345.
Single cusp of
Single cusp of

1.

2.

first species.
first species.

Single cusp of first species.


Single cusp of first species.
Isolated point.

3.

4.
5.

7.

Double cusp of second species.


Oscu- inflexion.

8.

Oscu-inflexion.

6.

Page 347.
2.

2^2.

5.

-VV17,

7.

3.

V2.

2v/2 for each.

4.
6.

3fl/2 for each.

a,

-*/2,

-5*

402

DIFFERENTIAL CALCULUS

C/O*'

Page 367

Page 367

85'

AXSWEAS

403

404

DlFfffiBENTIAL CALOTTLtfS

Pages

Page 368

405

DIFFERENTIAL CALCULUS

406

[PagQS 377-382

Page 377.
b*=l.
(iii)
,.

x=a

c*(

(ii)

0+aO sin 0, y=a sin Q-~a


/v
2/(2-w)
/v
2/(2-)
2/(2-n)
/v
'

cos

cos

'

0.

'

(iv)

(v)

(pX2

(VI)

2.

=C

2
.

x^+^=ci

(i)

(ii)

x^+^ = ci

(ii)

xyc=0.

(iff)

Page 378.
3.

2xy=c*.

(i)

5.

6.

(jc^j+j^j;)

9.

12.

13.

The

straight line

= 2a.

7.

p*xpy+(a+ap*+pq)=().
cos

,...,

r sin

(///)

16.

(i)

Kae(a

r 2 (e 2

1)

-7T/2)

2/ er cos

cot a

Gr=

*)

cot a
CO

fl

e
2

0+2/ =0.

Page 379.

17

+ p)

Miscellaneous Exercises

II.

Page 379.
2a see 3

2.

4.

(0/2).

4:4/2

a.

Page 380.
2

14.

(9/

17.

(Jfl.

-6/).

" UlS
1Q

a).

/2256 5
5

448i* \
'

3375V

'

(b '

Page 381.

28.

>'=xi*.

1) is an isolated point
(1,
(5, 3) and (5,
5) ate the
two points of inflexion.
x~a tanh a, .y^sech a is the required curve.
36.
382.
Page
x sin \t-\-y cos fat ---a sin /, x cos \ty sin ^/..-3fl cos ?/.
40.

32.

47.

(/)

Isolated point.

(//)

Single cusp of second species.

INDEX
(Numbers refer

to

pages)

Discontinuity, 38
Double points, 336
Acceleration, 79

Angle of intersection,
of two curves, 262, 268
Approximate calculation, 207
Arcs, 274
, Derivative of, 274
Astroid,212,257
Asymptotes, 313
by expansion, 328
by inspection, 324
parallel to axes, 315

Cardioide, 249
Catenary, 238, 259
Cauchy's Theorem, 137
Cissoid, 244

Closed interval, 1*3


Composite functions, 209
Concavity, 281

Conjugate point, 336


Convexity, 281
Continuity of elementary functions, 54
Continuous functions, 36, 194
properties of, 54
Curvature, 290
Centre of, 303
Chord of, 305
Circle of, 305
,

Fnyelopes, 369
Epicycloid, 240
Equations,
,

Radius

of, 291

Curve Tracing, 348


Cusps, 336
Cycloid, 240, 257

254

Parametric, 255, 276


Pedal, 266, 272
Polar, 248, 277, 298
, Tangential polar, 300
Equiangular spiral, 252, 269
Euler's Theorem, 199
Evolutes, 305,310, 372
Expansions, 179
,
,

Extreme values, 148

Folium of Descartes, 246, 362


Functions, 11
Algebraic, 35
,

Composite, 209
Continuous, 37
Exponential, 23, 97
Homogeneous, 199
Hyperbolic, 67, 97
Implicit, 209, 292

,
,

Implicit, 243,
Intrinsic, 291

Inverse, 21, 88
-- Inverse Hyperbolic, 70, 99
Inverse Trigonometric, 30, 93
Logarithmic, 25, 96, 100
20
, Monotonic,
of a function, 34, 86
of two variables, 193
Transcendental, 35
Trigonometric, 26, 89
,

Derivative, 72
,

of arcs, 274
Partial, 196
Sign of, 136, 150

Determinant, 107
Differentials, 206
Differential Coefficient, 72
Differentiation, 72
, Logarithmic, 100

of function of a function, 86
of implicit functions, 209
, of inverse functions, 88
Partial, 196
Repeated, 217

Geometric Interpretation,

77, 131, 197

Graphical representation, 14
Greatest values, 149

H
Homogeneous

functions, 199

Hypocycloid, 240

Successive, 113

292
Implicit Functions, 209,
Indeterminate forms, 165, 185

FFEBENTIAL OALOITLUA

408
Infinite limits, 48
179
, series,
Inflexion, 281

Points, Stationary, 150, 223


Polar coordinates, 267, 277, 331, 359

Interval, Closed, 13, 55


, Open, 13

Inversion of operation, 53

Radius of curvature, 291, 345


ve.ctor,

Lagrange's Theorem, 133


multipliers. 230
Least values, 149

267

Remainder, 142
Repeated derivatives, 217
Rolle's Theorem, 130

Leibnitz's Theorem, 121


Lejnniscate, 249
Limits, 41, 195
Indeterminate, 165
,

Right handed, 42
Left handed, 42

,
,

M
Maximum value, 148,
Minimum value, 148,
Mean value theorem,

223
223
133, 137

Modulus, 9
Monotonic functions, 20
Multipliers, 230

N
Newtonian Method, 293
Node, 336
Normal, 259
Numbers, Irrational, 5
Rational, 2
Real, 6

,
,

Series, Binomial, 183


, Taylor's, 140, 220

Maclaurin's 142
Sign of derivative, 136, 150
Singular points, 335
Smallest values, 149
,

Spirals, 251, 252


Stationary values, 150,223
Strophpid, 245
Subsidiary conditions, 229
Sub-tangent, 264
Sub-normal, 264
Successive differentiation,

,
,

,
,

interval, 13

Euler's, 199
Lagrange's, 133
Leibnitz's 121
Maclaurin's 142
Mean value, 133, 137
Rolle's, 130
Taylor's, 140, 220

on

limits, 52
Total differentials, 200
,

Parametric equations, 255, 276, 293, 364


Partial differentiation, 196
Pedal equations, 266, 272
Values, Greatest, 149
Maximum, 148, 223
Points, Conjugate, 336
Minimum, 148, 223
Cusp, 336
149
Double, 336
, Smallest,
Stationary, 150, 223
Node, 336
, Multiple, 335
Variables, 11
281
, of inflexion,
Velocity, 78
,
,

Singular, 335

13

Tangent, 254
Tangential Polar equations, 300
Taylor's Theorem, 140, 220
Theorem, Cauchy's, 140

Open