UNIT 5 FIRST ORDER PARTIAL

DERIVATIVES AND . 1
DIFFERENTIABILITY
Structure
5.1 Introduction
Objectives
5.2 First Order Partial Derivatives
Definition and Examples
Geometric Interpretation
Continuity and Partial Derivatives
5.3 Differentiability of Functions from R2 to R
5.4
5.5 Summary
-
Differentiability of Functions from Rn R, n > 2

5.6 Solutions and Answers

5.1 INTRODUCTION

You are already familiar with the concept of a derivative of a real-valued function
of a real variable (Calculus, Unit 3). In this unit, we shall study this concept for
functions of several variables. You have seen that the notions of limit and con-
tinuity can be easily extended to these functions. But the definition of the -
derivative of a real-valued function of a real variable cannot be applied, as it is, to
a real-valued function f(x) of n variables (n > 1). This is because for any h E Rn,
f(x+ h) - f(x)
h # 0, the quotient does not make sense as the division by a .
h
vector in Rn is not possible.
However, if we examine the definition of derivative more closely, we realise that a
function of a single variable is differentiable at a point if and only if the two
directional derivatives, i.e., the right hand derivative and the left hand derivative,
exist and are equal at that point. You will see later, in Unit 7, that this concept of
directional desivative can be generalised to functions of several variables. The only
hitch is that in Rn, we have to deal with infinitely many directions. gowever, in
the beginning we shall confine ourselves to the special directions which are parallel
to the coordinate axes. This leads us to the notion of partial derivatives.
In this unit we'll discuss the concept of partial derivatives of a function of several
variables in detail. Note that this notion of partial derivatives does not fully
generalise the concept of derivative of a real-valued function of a real-variable.
Later in this unit we introduce the concept of differentiability for functions of
several variables, and discuss the relationship between differentiability, continuity
and the existence of partial derivatives.
Throughout this unit by the word 'function' we shall mean a real-valued function
of' several variabfes, i.e., a funcsion from D -R, where D is a subset of
Rn, n > 1. We shall give the definitions for general n, but host of the time dur-
ing our discussion we shall confine ourselves to the case n = 2, i.e., to real-valued
a
functions of two variables. We'll briefly discuss the case n = 3 also.

Objectives
After reading this unit, you should be able to
a define partial derivatives of the first order for a function of several variables,
a partially differentiate a given real-valued function of several real variables with
respect to any particular variable,
give the geometrical interpretation of first order partial derivatives of functions
of t\ko variables,
 decide whether a given function of two or more variables is differentiable or First Ordcr
not, Partial Derivatives
give examples to establish relationships between the continuity, differentiability and Differentiability
and existence of partial derivatives at a point for a function of several
variables.

5.2 FIRST ORDER PARTIAL DERIVATIVES

In this section we shall see what the partial derivative of a function at a point
means. We already know how to define the derivative of a function of a single
variable. We'll use this knowledge in defining the partial derivatives o f functions
of several variables.

- 5.2.1 Definition and Examples ---+I&---

Consider a function f : D -
R; where D E R". Let (xl, x2, .... x,) be an
interior point of D, i.e., there exists an open sphere with centre (xl, x2, ....., x,)
contained in D. Then for each i, 1 Ii In, we can construct a real valued func-
tion of a real variable from this function f in the following manner.

Choose a small number 6 > 0 such that the point (xl, x2, ..., xi + h, xi+
....., x,) € D for all h 6 ] - 6, 6 [. Such a 6 exists since (x,, x2. ......, x,) is an in- Fig. 1
terior point of D. We have shown this for n = 2 in Fig. 1. Now we can define
fi : ] - 6, 6 [ -
R such that
, xi-], xi + h, xi+ 1,
f(xl, ~ 2 ..., ..., x,) - f(xl, ~ 2 ....,
, x,)
fi (h) =
h
for all h c ] - 6, 6 [, h # 0.

Now if lim fi (h) exists, then we say that f has a first order partial derivative with
h-0
respect to the ith variable xi at the point (x,, x2, ...., x,) and the value of
lim fi (h) is called ith first order partial derivative of f at the point (xl, x2,
h-0
.... x,).'

You should note here that for lim fi (h) to exist, it is enoug': that f be defined in
h-0
a neighbourhood of the point under consideration.

I More formally, we have the following definition.

Definition 1: Let f : D -
R, where D E R n , and let (x,, x2, ...., x d be an
interior point of D. We say that the function f has ith partial derivative at the
point (xl, x2, ...., x,) if

lim
f(xl, ~ 2 ....,
, xi-], xi + h, xi+ 1,...., x,) - f(xl, ~ 2 ....,
, x,)
exists.
I- h-0 h Whenever we say that a limit
exists, we mean tvat a finlk
The value of this limit is called the ith partial derivative of f at the point limit exists.
( ~ 1 X2r
, xn).
a*..,

There are different symbols available in literature to denote the partial derivatives af is read as
of a given function. However, we shall use only the following symbols axi
"del f by del xiw
af
- fx, Di f,
axi

according to our convenience to denote the ith partial derivative of first order of f.
In case we want t o emphasise the point (xl, x2, ...., x,) at which the partial
derivative has been calculated, we write
 You have seen in the Calculus course that we write y = f(x) to express a
function of a real variable. For functions of two variables it is customary to write
z = f(x,y), and the two partial derivatives of f at the point (x,y) are then denoted

Remark 1 : i) Note that the notion of partial derivative of a function (like con-
tinuity) is local in character. That is, we check the existence of the partial
derivative of a function d 8 point. Thus, when we say that a function has partial
derivatives on a set A, we mean that the function has partial derivatives at each
point of A.
ii) It is obvious from the definition of a partial derivative at a point that the func-
tion must be defined in a neighbourhood of the point. Thus, we can talk about the
partial derivatives only at the interiof points of the domain D. For example, if D is
a disc in R2, then we cannot talk a b q t the partial derivative of a point on the cir-
cumference of this disc.
iii) If f has a partial derivative at (x,, x2, ...., x,,), then'its value depends only on
the values of f in an open sphere around the point. If the function is changed out-
side this sphere, it would not affect the value of the partial derivative.
We now give some examples to show how to obtain partial derivatives of given
functions at given points. If we do not mention a specific point at which partial
derivatives arc: to be calculated, wemean that they are to be calculated at a generic
point (x,y) or (x,y,z) according as n = 2 or n = 3.
-
Exrmpk 1: Let f : R2 R be a function given by f(x,y) = x2+xy+ y3. Let us
find fx (x,y) and fy (x,y).
By definition,

~ ~ + 2 x h + h ~ + x y + h- ~x + -'xy
~' - y3
= lim
h-0 h

= lim (2x
h-O
+ h + y)

Similarly,
- fy (x,y) = li!m f(x,y + k) - f(x,y)
k -0 k

When we are considering functions of two variables x and y, then for the
increment in x we normally use the letter h and for the inctement in y, the letter k.
Similarly, when we arc dealing with functions of three variables x, y an8 z, we use
the letters p, q and r for increments in x, y and z, respectively. This' is q~ilya
matter of &vention, not a rule.
In the next example we consider a function of three variables.
Example 2 :Let f : R' -- R be a functibndebned by f(x, y, z) = xy .+ yz + y.
- -
La us find the partial derivatives at the point (a, b, c). By definition

f, (a, b, c) = .lim
f(a+p, b. c) - f(a, b, c)
P-0 P

fy (a, b, C) = lim
f(a, b +q, c) -1 jf(a, b, c)
I
q-0
(1 -
= lirn
a@+& + (b4& c + c a - ab - bc - ca
q-0 1 9
I
= a+c

f, (a, b, c) = lirn
f(a, b,,c+r) - f (a, b, c)
r-0 r

= lirn ab+b(c+r) + (c+r) a - a b - b c -ca

r-0
.
,
/
- C

In the next example we shall calculate the partial derivatives of a function of n

variables.
-
Example 3 : Let f : Rn R be a function defined by
f (x,, X2,.... ,x") = ,x: + xf + + x;. ....
To find the partial derivative of f with respect to xi at the point (al, a2,.....a,,),
we write

= lirn
a:+ai+ ...+ abl+(q+h)2+af+1+...+af.-a: - a$ -...-a, 2
h-0 h

= lim
2a,h + h2 = 241,.
h-0 h
Try to solve these exercises now.

- R be the constant function defined by

- - - - - -

El) Let f : R~
f(x, y) = c for all (x, y). Show that f, (a, b) = 0 = fy (a, b)
for all points (a, b).
E2) L a f : R~ - R be defined by
1if (x, Y) if (0, 0)
f(x, Y) =
0 if (x, y) = (0, 0)
Show that fx (0, 0) as well as fy (0, 0) does not exist.
E3) Let f : R~ - R defined by
be

f(x, Y) = [
x if (x, Y) (0, 0) f

0 if (x, y) (0, 0) =
Show that fx (0, 0) = 1 and fy (0, 0) = 0.

From these examples and exercises you must have observed that f, (x, y) is
nothing but the derivative of f(x, y) considked as a function of a single variable x,
treating y as a constant. Similarly, fy(x, y) is nothing but the derivative of f(x, y)
considering it as a function of the single variable y, and treating x as a constant.
In general, we can say that f,, (x,, XI,....,x,) is nothing but the derivative of
Partial Derivatives f (x,, x2,....,x,) with respect to xi treating all the other variables as constants.
Thus, for calculating partial derivatives, we can use our knowledge of calculating
derivatives of functions of a single real variable and avoid the lengthy limiting
process.
Here is an example to illustrate this.
Example 4 : Let us find the partial derivatives of the following functions.

ii) z = x sin y + y cos x

iii) z = x eY + y ex.
In the Calculus course (Block 1) We first note that in all the three cases, the functions involved arc either
you have seen that polynomial, polynomials or trigonometric or exponential functions. This ensures that the partial
trigonometric and exponential derivatives exist. By direct differentiation, we get
functions of a single variable are
differentiable.

az az
ii) - - sin y - y sin x ; - = x cos y + cos x
ax ay

The calculation of partial derivatives is not always as simple as in these examples.

In some exceptional cases, we have to use the limiting erocess as in the case of one
variable. You will be able to recognise such cases with practice. Lef us consider
one such situation.
Example 5 : Suppose f : R2 - R is definkd by

Let us find the two partial derivatives at the points (0, 0), (a, 0). (0, b) and (a, b),
where a # 0, Ib # 0.
Now, by definition,
f (O+h, 0) - f (0, 0) 0-0
f, (0, 0) = lim = lim - = 0.
h-0 h h-0 h

fy (0. 0) = lim
f (0, O+k)- f(0, 0) -
. = lim 0 0 = 0.
-
k-0 k k-0 k
f (a+h, 0) - f (a, 0) 0-0
fx (aJ 0) = lirn ,= lirn - = 0.
h-0 h h-0 h

f ( a , O + k ) - f (a, 0) a4 + k4 I.
fy (a, 0) = lim = lim = -
k-o k r-o k a3

f (0+ h, b) - f (0, b) b4, + h4 - -1

fx (0, b) = lim , = lim
h-0 h h-0 h b

fy (0, b) = lim
f (0, b + k ) - f ( 0 , b) = lim - -
0 0 L 0.
k-0 k k-0 k
f (a+h, b ) - f (a, b)
f, (a, b) = lirn

(a + h14 + b4a4 + b4
= lim
h-0 h
(ab + hb) (a4+ b4) - (ab) (a4+ 4a3 h + 6a2h2+ 4ah3 + h4+ b4)
= lirn
h-o h (a4+ b4) [(a+ h14 + b4]
 b (a4+ b4) - (ab) (4a3+ 6a2h+ 4h2 a + h3) First Order
* = lim Partial Derivatives
h-o (a4+b4) [(a+ h)4+b4] and Differentiability

fy (a, b) = lim
f (a, b+k) - f (a, b)
k-0 k
a (b+k) - - ab
a4+(b+k14 a4 + b4
= lim
k-0 k

(ab + ak) (a4+ b4) - (ab) (a4+b4+ 4b3k+ 6b2k2+ 4bk3+ k4)
= lim
k-o k (a4+b4) [a4 + (b+ k)4]

a (a4+b4) - (ab) (4b3+6b2k+4bk2+k3)

= lim
k-o (a4+b4) [a4+(b+k14]

Here,, by direct differentiation, we could have obtained fx (a, b) and f, (a, b),
(a, b) # (0, O), correctly, but not fx (0. 0) or fy (0, 0).
Can you see why? Since f is defined as a quotient of two polynomial functions for
all (x, y) # (0, 0), we can use direct differentiation to calculate partial derivatives
at these points. But to calculate fx (0, 0) or fy (0, 0) we need to use f (0, 0), which
is not defined by the same quotient. Also note that after obtaining f, (a, b) and
fy (a, b), we could have substituted a = 0 or b = 0 to get fx (0, b), fy (0, b),
f, (a, 0) and fy (a, 0).

You must have come across functions from R--R which do not possess derivatives
at some points. For example, f(x) = 1x1 can not be differentiated at x = 0. Here
is an example of a function of two variables whose partial derivatives fail to exist
at some points.
Example 6 : If f : R2 - R is defined by

-x+ - , yY# O , x # O
, otherwise,
then fx (0, 1) and fy (1, 0) do not exist.
Let us prove this.
1
h + 5 - O
f (O+h, 1) - f (0, 1) - 1
Here = I + -
h h h

f (1, O+k) - f (1,O) - .-k1 + ' k - 0 1 +

and -- 1.
k k kZ
1
Since lim - = a,neither
h-0 hZ

lim
f (O+h, 1) - f (0, 1)
nor lim
f (1, O+k) - f (1,O)
h-0 h . k-0 k
exist. So fx and fy do not exist, respectively, at the points (0, 1) and (1, 0).
Why don't you try some exercises now?
 af af
E4) If f (x, y) = 2x2 - xy + 2yZ; find - and -at the point (1, 2).
ax ay
ES) Find all the fust order partial derivatives of the following functions.
a) sin (x2 - y)

c) y sin xz
d) xy
e) x ' ~+ exy2
E6) Show that the functions u = ex cos y, v = ex sin y satisfy the conditions
-au- - - aav
n d - = - -au 'av
ax ay a~ ax'
E7) Let f : R -
R and g : R -- R be two differentiable functions.
Let F(x, y) = f(x) + g(y) for all x and y. Show that
Fx(x, Y) = f'(x) and FAX, Y) = g'(y).
E8) Let f and g be two real-valued functions for which fx(a, b) and gx(a, b) exist.

Show that a ('+

ax
g, exists at (a, b) and is equal to fx(a, b) + gx(a, b).
Is the converse true?

Through these exercises you must have gained enough practice of calculating
partial derivatives. In the next sub-section we shall try to interpret these
geometrically.

5.2.2 Geometric Interpretation

In the case of a real-valued function f of one variable x, you know that the .
'
derivative f'(x) gives the slope of the tangent to the curve y = f(x) at a generic
point (x, y). Now we shall try to visualise the partial derivatives of a real-valued
function of two variables. Such a function, as you know, represents a surface in
R'.
So, let f(x, y) be a real-valued function of two variables and let S = ((x, y, z) ( z
= f(x, y)) be the surface represented by the function f(x, y) in R'. Suppose that
f(x, y) has both the partial derivatives at a point (a, b) and let c = f(a, b). Let PI
be the point (a, b, c) on the surface S. Now the plane y = b, which is parallel to
the XOZ plane and passes through P1will intersect the surface in a curve (see
Fig. 2(a)). We are giving an enlarged version of the curve C in Fig.. 2(b).
 Let P2 be a point with coo~dinates(a', b', c') on the curve C close to PI. Since Fire( Order
P2 lies on the surface S and the plane y = b, we have b' = b and c' = f(a', b). Partial Derlvallves
and Ditfekntlabi~lt~
Clearly PIP2is a line joining the points (a, c) and (a', c') in the plane y = b and
its slope is given by

f(a+h, b) - f(a, b) \
= , where we have put a ' = a + h.
h
As h approakhes zero, the secant PIP2approaches the tangent at the point PI to
the

curve C. Consequently, lim

f(a+h, b) - f(a, b) af .
, 1.e.. -(a, b) or
h-0 h ax
gives the slope of the tangent at the point (a, b, f(a, b)) to the curve C which is .
the intersection of the surface z = f(x, y) and the plane y = b. Similarly,

(3.. b)
is the slope of the tangent at the point (a, b, c) to the curve CI, which

is the intersection of the surface z = f(x, y) and the plane x = a (see Fig. 2(c)).
We use this fact in the following example.
Example 7 : Suppose we want to find the slopes of the tangents to the curves of
intersection of the planes x = 2 and y = 3 and the surface z . = xy 3x2 at the +
point (2, 3, 18).
We know that the slope of the tangent to the curve of intersection of the plane
x = 2 and the surface z = xy + 3x2 at the point (2, 3, 18) will be given by

Now, ($) (2. 3)

= 3) = 2.

Therefore, the slope of the tangent at the point (2, 3, 18) to the curve of intersec-
I
tion of the plane x = 2 and the surface z = xy + 3x2 is 2.

Similarlyr since +
= (y 6 x ) , 3) = 15, therefore, the-slope of. the tangent

I at the point (2, 3, 18) to the curve of intersection of the plane y = 3 and the sur-
1 face z = xy + 3x2 is 15.

I
You should be able to do this exercise now.

).
-
E9) Find the slope of the tangent at the point (1,. 2, 14) tc ~ n crw:..*r of
intersection of the plane y = 2 and the-surfct? z =: 5 : 3 <-' .

You know that if a function f from R -

R is differentiable at a point, it is 'also
continuous at that &t. In the next sub-section we shall see whcthe~any such bnk
exists between the cbntinuity and the uristena of partial dcrilrativts of functtons .
from R~ -- R ot not.

5,2.3 C o n d a n between Continpity a d PutW Derivatives

If the two partial derivatives of a function f(x, y) exist, then what can we wfer
from this? La's see. Suppose f(x, y) is a red-valued function having partial
derivatives at a point (a, b). Then for h # 0,
Partial Derivatives nd therefore,
f(a+ h, b) - f(a, b)
lim [f(a + h, b) - f(a, b)] = lim lim h
h-0 h-0 h h-0

= fx (a, b). 0

Therefore, we can say that f(a+ h, b) -- f(a, b) as h -- 0, i.e., f(x, y) -- f(a, b) as

(x, y) approaches (a, b) along a line parallel to the x-axis. Similarly, the existence
of the other partial derivative shows that f(x, y) -- f(a, b) as (x, y) approaches
(a, b) along a line parallel to the y-axis. The existence of fx(a, b) and f,(a, b) does
not give us any further information. So we do not know whether the limit of
f(x, y) exists or not if (x, y) -- (a, b) along any other path. But you have learnt in
Unit 4 that f(x, y) would be continuous at (a, b) if f(x, y) -- f(a, b) as (x, y)
(a, b) along any path (which need not even be a straight line). Thus, it is clear
-
from the above discussion that the mere existence of partial derivatives need not
ensure the continuity of the function at that point. This, in fact, is the case as
shown by the following example. Later on we shall see that if the partial
derivatives satisfy some additional requirements, theh their existence does imply
continuity.
Example 8 : Let the function f : R~ -- R be defined by

You will see that both the partial derivatives of f at (0, 0), that is, f, (0, 0) and
f, (0, 0) exist, but f is not continuous at (0, 0).
f(O+ h, 0) - f(0, 0) 0-0
Now, fx (0, 0) = iim = lim - = 0, and
h-0 h h-0 h
f(0, O+ k) - f(0, 0) 0-0
fy (0, 0) = lim = lirn - = 0.
k-0 k kk0 k

So, f possesses both the first order partial derivatives at (0, 0). However, this
function is not continuous at the point (0, 0) since lim f(x, y) does not exist.
(x,y)-(O.O)
(see E 3) a), Unit 4).
f(x) = 1x1 is continuous on R We know that a real-valued continuous function of a real variable need not be
but is not differentiable at x = 0. differentiable. The same is true for functions of several variables. This means that
a function of several variables which is continuous at a point need not have any of
the partial derivatives at the point. The following example illustrates this fact.
-
Example 9 : Let f : R3 8 be defined by

We'll show that f is continuous at (0, 0, 0) but does not possess any of the three
first order partial derivatives at (0, 0, 0).
Now at the point (0, 0, 0), we have
f(O+ h, 0, 0) - f(0, 0, 0) - -( h J
fl Ol) =
h h
lhl
Therefore, lim fl(h) = lirn - = lirn - = 1, h
h-O+ h-0 h h-0 h
h>O h>O

but lim fl(h) = lirn -

Ih 1 = lim -
-h = - 1.
h-0- h-0 h h-0 h
h<O h<O
Hence, lirn fl(h) does not exist.
h-0
You can similarly check that
lim f2(h) and Lim f3(h) also do not exist.
h-0 h-0
 Thus, f does not possess any of the first order partial derivatives at the point First Order

I (0, 0, 0). But this function is continuous at (0, 0, 0) as you have seen in E7) d),
*nit 4.
Partial Derivatives
and DiffemntinbUity

I If you have understood these examples, you should be able to solve the following
exercises.

I
E10) Let f(x, y) = for all (x, y) 6 R ~ Show
. that f is continuous at (0, O),
but f,(O, 0) as well as fy(O, 0) do not exist.

1 1
E l l ) Let f(x, y) = [;sin; + y s i n - ,Yx y $ 0
xy = 0.

Show that fx(O, 0) as well as fy(O, 0) exist. Also show that f is continuous at
(0, 0).
E12) Show that the function f : R~ - R defined by
'4,. itx4+y2f o
0 ,ifx=O=y
possesses first order partial derivatives everywhere including the origin but
I the function is discontinuous at the origin.

I- o
I

1
", ify z
E13)Letf(x,y) = lyl
0 ,ify = 0
I

a) Prove that f is continuous at (0, O), and both f, (0, 0) and f, (0, 0) exist.
b) Show that fx (1, 0) exists, but f, (1, 0) does not.
I
i In the above examples and exercises we have seen that the existgnce of partial
derivatives docs not imply continuity. However, if the partial derivatives satisfy
some more conditions, then we can ensure continuity. You will study this in
Lagrange's m.v. thwrim: La f
be a real-valued functioh con-
tinuous on [a, a + h] and>dif-

i
Theorem 3. In order to prove this theorem we need a simple result which follows ferentiable on ]a, a + h[. Then
easily from Lagrange's mean value theorem (Block 2, Calculus, also see margin 3 B e 10. I[. s.t.
remark). We first state this result. f(a + h) - f(a) = hfl(a +Oh).
it Theorem 1 (Mean valne tbeorem) : Let f be a real-valued function defined on a
I
neighbourhood N of (a, b). If f, exists at all points of N and fy exists at the point
(a, b), then
1

1' +
f(a h, b + k) = f(a, b) + hf, (a+Bh, b + k) + k (fy (a, b) + g)
for all real h, k such that (a+ h, b + k) belongs to N, where B depends on h and k
I

and 0 < 8 < 1. Moreover, q is a function of k, which tends to 0 as k 0. * -

You can see that this is an extension of Lagrange's mean value theorem to func-
tions from R~ to R. Interchanging x and y in the hypothesis of this theorem we get
another theorem. which we now state.
Theorem 2 : Let f be a real-valued function
, ~-. - defined on--- a
~- - nei~hbourhood
.-- - - -"-. N. of
.- -- -- - - - - - -

f (a, b). If f, exists at all points of N and f, exists at (a, b), then for real h and k
+
such that (a+ h, b k) t N , we have
f(a+h,b+k) = f(a,b) + kf,(a,b+Ofk) + h(f,(a,b) + q')
where 6' depends upon h and k, 0 < 8' < 1, g' is a function of h and tends tb
zero as h -
0.
Both Theorem 1 and Theorem 2 are easy consequences of Lagrange's mean value
theorem, but we shall not discuss their proofs here. We will now use these
theorems to prove Theorem 3.
Theorem 3 : Let f be a real-valued function of two vaiiables defined in a
neighbourhood N of a point (a, b) such that one of the fust order partial
derivatives exists at all points (x, y) c N atid is bounded on N, whereas the other '
partial derivative exists at the point (a, b). Then the function f is continuous at the
point (a, b).
Proof :Without loss of generality we can assume that fx exists for all (x, y) c N
and is bounded on N, whereas fy exists at the point (a, b). By Theorem 1, for all
real h, k such that (a + h, b + k) c N, we get
f (a+ h, ti +.k) = f (a, b) + hfx (a+ Oh, b + k) + k (fy (a, b) + q), ...(ly"
whercOc8< landq-Oas'k-0.
Since fx is bounded on N, it follows that

Consequently, from (1) we get,

1
lim f(a+h,b+k)=f(a,b)
Q, k) - (0.0)
Hence the function f is continuous at (a, b) and the proof is complete.
While proving Theorem 3 we had assumed that fx exists at all points of N and is
bounded on N, and fy exists at (a, b). If instead we had assumed that fy exists at
all points of N and is bounded on N and f, exists at (a, b), then we could have
proved the result using Theorem 2.
Now here is a result which follows easily from Theorem 3.
Corollary 1 : Let f be a real-valued function of two variables defined in a
neighbourhood N of a point (a, b), such that both the partial derivatives of f exist
at all ppints of N and one of them is bounded on N. Then the function f is con-
tinuds everywhere on N.
Note that the conditions given in Theorem 3 are only sufficient and are not
necessary. We have already seen in Example 9 that a function may be continuous
even when none of the partial derivatives exist.
In the next example we use Corollary 1 to prove the continuity of the given
function.
-
Example 10 : Is the function f : R2 R given by f (x, y) = yex continuous
everywhere? Let us find out.
We first note that fx (x, y) = yex.and fy (x, y) = ex.
Let (a, b) be any point in R2, and wnsider a neighbourhood
N = ( ( x , Y ) IJ G T T F W < 1 1
of (a, b). Now, fx (x, y) and f, (x, y) exist at dl points of N. Further, since.
IX -a1 s J(x-a12 + (Y-biz,
therefore, for all (x, y) c N, we have-ix - a1 c .1,

i.e., en-' r: ex < en+'.

So, f, is bounded on N. Therefore, in view of Corollary 1, f is continuous at the
point (a, b). Since (a, b) was any arbitrary point of R2, we can say ihat f is con-
tinuous everywhere on RZ. -
Try this exercise now.

E14) Use Corollary 1 to show that the following functions are continuous
everywhere on ltZ.
a) f (x, y) = xeY
b) f (x, Y) = 3xy
 In this section we have seen that the mere existence of partial derivatives docs not First Order
imply continuity. This shows that the concept of partial derivatives does not Partid Derlvatlves
generalise the concept of differentiation of functions from R -
R. In the next sec-
tion we'll introduce a concept which is a generausation of the concept of differen-
and Differentlability

tiation of real-valued functions of a single variable.

5.3 DIFFERENTIABILITY OF FUNCTIONS

FROM R~ TO R

What do we mean when we say that a function f from R to R is differentiable at a

point c?
Do you agree that f is differentiable at a point c if and only if there exists a cons-
tant A (depending on the function f and the point c) such that
f (c+h) - f (c) = Ah + hq(h)
where q(h) - 0 &h '- 0 ?
While checking this you will find that A is nothing but fl(c).
This definition of differentiability of a function of a single variable can be
generalised in a natural way for functions of several variables. In this section we
shall study the differentiability of real-valued functions of two variabld. Here is its
definition.
Definition 2 : Let f be a real-valued function defined in a neighbourhood N of a
point (a, b). We say that the function f is differentlabk at (a, b), if
f (a+h, b+k) - f (a, b) = Ah + Bk + hc$ (h, k) + k$ (h, k), where
h and k are real numbers such that (a + h, b + k) E N,
A and B are constants independent of h and k but dependent on the function
f and the point (a, b),
c$ and $ are two functions tending to zero as (h, k) -- (0, 0).
We would like to make an important remark here.
- Rcervlr 2 : (i) In the literature you may find another definition of differentiability
which we give below.
Let f be a real-valued function defined in a neighbourhood N of a point (a, b).
Then the function f is said to be differentiable at the point (a, b), if there exist
two constants A and B (depending on f and the point (a, b) only) such that
f (a+h, b+k) - f (a, b) = Ah + Bk + ~-c$(h, k) where 4(h, k)
is a real-valued function such that c$ (h, k) -
0 as (h, k) -
(0, 0).
The equivalence of these two definitions can be established easily by using the
<
identity

ii) For a function f from R to R, if f ' ( ~ exists,

) then we can approximate
f(x) - f(a) by the linear function (x - Q) fl(xo) near Q. Similarly, if g(x, y) is
.differentiable at (a, b), then g(x, y) - g(a, b) can be approximated by the linear
function (x - a) A + (y - b) B in a ~ncighbourhoodof the point (ai b).
We now illustrate the definition of differentiability with the help of a few
I examples.

1 Emmpk 11 : Let f(x, y) = x2+y2. Then we can show that f is differentiable at

any point (a, b).
For any two real numbers h and k, we have
i

f(a + h, b + k) - f(a, b) = (a + h)* + (b + k12 - (a2+ b2)

 - - - ---- --
If we set A = 2a, B = 2bi +(h, k) = h, $(h,k) = k, then
f(a+ h, b+ k) - f(a, b) = Ah + Bk + h4 (h, k) + k$ (h, k), where A and B

-
are constants independent of h and k, 4(h, k)
(h, k)
-
0 and $ (h, k)
(0, 0). Thus f is differentiable at the point (a, b).
0 as -
Exrunpk 12 : Let f(x, y) = -XY Then let us show that f is differentiable at all
points (a, b) in the domain of definition of the function.
Since f is not defined for y = 0, we take b # 0. Let h and k be two real numbers
such that (a+ h, b + k) is a point in a neighbourhood N of (a, b), which is
contained in the domain of f. Then b + k # 0, and
a+h a
f ( a + h , b + k ) - f (a, b) = -- -
b+k b

Then f (a+h, b+k) - f (a, b) = Ah + Bk + h tj&, k) + k$ (h, k), where A

and B are constants independent of h and k, 4 (h, k)
(h, k) - -
0 and $ (h, k)
(0, 0). Hence, f is differentiable at the point (a, b).
0 as -
Now here is an example of a function which is not differptiable.
Exrnrpk 13 : We will prove that the function given by f (x, y) = 1x1 +
lyl is not
differentiable at (0, 0). Suppose, if possible, that f is differentiable at (0, 0). Then
f(O+h, O+k) - f(0, 0) = Ah + Bk + h4 (h, k) + I($ (h, k)
where A and B are constants. 4 (h, k)
Therefore, I h 1 +
-
0 and $ (h, k) 0 as (h, k)
I kl = Ah + Bk + h4 (h, k) + k$ (h, k).
- - (0, 0).
Let h = 0 and k > 0. Then
k = Bk + k$ (0, k),
i.e., 1 = B + $ (0, k).
-
Taking limits on both sides as (h, k)
because (I (0, k) 0.
- (0, 0). we get B = 1,

Now let h = 0 and k < 0. Then

Taking limits on both sides as (3, k)- (0, 0), we get B = -1, because flQ, k)-0.

Thus the assumption that the given function ia differentiable at (0, 0) leads us to
the contradiction B = f = -1. Hence, 1x1 + 1 y 1 is not differentiable at (0, 0).
Now see if you can do these exercises.

ElS) We have listed some results about the differentiability of real-valued func- ,
tions of a real variable in the fmt column of the following table. Write
analogous statements for real-valued functions of two variables in the second
column, and check whether each of them is true.
 One variable Two variables First Order
Partial Derivatives
a) A constant function is differgntiable and Differentiability
everywhere

b) I f f is differentiable at a E R, then
cf (c t R) is also differentiable at a.

c) If f and g are differentiable at a t R,

then f +_ g is also differentiable at a.

d) If f, g are differentiable at a c R then

fg is also differentiable at a.

E16) Show that the function x2+y +xy is differentiable at (0, 0).

E17) Show that cos (x+ y) is differentiable at the point (T 1).

. 4' 4
E18) Show that the following function f is not differentiable at (0,O)

In the case of a real-valued function of a real variable, continuity does not imply
differentiability. The same is true for real-valued functions of two variables.
Consider the function f(x, y) = 1x1 + Iyl of Example 13. According to E7) d) of
Unit 4, it is continuous at (0, 0). In Example 13 we have seen that it is not
differentiable at (0, 0). So continuity at a point does not imply differentiability at
that point. However, every function which is differentiable at a point, is also
continuous at that point. This is proved in the theorem that follows.
Theorem 4 : Let f be a real-valued function defined in a neighbourhood N of a
point (a, b). If f is differentiable at (a, b), then f is continuous at (a, b).
Proof : Let h and k be two real numbers such that (a+ h, b + k) E N. Then
differentiability of f at (a, b) implies that there exist two constants A, B and two
functions t$ (h, k). JI (h, k) such that

--
where 4 (h, k) -- 0, $ (h, k) -- 0 as (h, k) (0, 0).
Now taking the limit on both sides of (2) as (h, k) -
(0, 0), we get
lim (f(a+h,b+k)-f(a,b))=O,or
a, k) - (0, 0)

r'
(h,k)
l
-im(0. 0) f ( a + h , b + k ) = f ( a , b )
This shows that the function f is continuous at (a, b). We can use this result to
establish the non-differentiability of a function at a given point. For instance, we
have seen in Sec. 4.4.2. that the function
t

is not continuous at (0, 0). Thus, in view of Theorem 4, we can conclude that this
/ function is not differentiable at (0, 0).

I You can now use Theorem 4 to solve this exercise.

E19) Show that the following functions are not differentiable at (0, 0) by showing
that they are discontinuous at (0, 0).
 You can check that A- f, (a, b) So far we have not said anything about the values of the constants occurring in the
and .B= f, (a, b) in Examples 11. ' definition of differentiability of a function at a point. We shall now show
12 and 13. (Theorem 5) that the constants A, B mentioned in Definition 2 are nothing but the
two partial derivatives of the function under consideration at a point. This, in
I
particular, would show that if a function is differentiable at a point, then it has
both the partial derivatives at that point. But is the converse true? That is, if a
function has both the partial derivatives at a point, can we conclude that it is
differentiable there? No. The existence of partial derivatives, as we have seen in
Example 8, does not guarantee even continuity. So, obviously, it cannot guarantee
differentiability.
Look at the following theorem now.
,,.1.
Theorem 5 : Let f be a real-valued function defined in a neighbourhood h.bf the
point (a, b). If f is differentiable at (a, b), then f possesses both the partial*"
derivatives at (a, b).
Proof :Let -h, k be real numbers such that (a + h, b + k) E N. Since f is
differentiable at the point (a, b), it follows that
f (a + h, b + k) - f(a, b) = Ah + Bk + h4 (h, k) + k$ (h, k).
where A and B are constants, 4 (h, k) - -
0, $ (h, k) 0 as (h, k) -
(0, 0).
You can see from Fig. 3, that if (a + h, b +k) € N, then (a + h, b) and (a, b + k)
also belong to N. So if we let k = 0 in the above equation, then we get
f (a+h, b) - f (a,.b) = Ah + h4 (h. 0)
i.e., (a+h* b ) - (a = A + 4 (h. 0) for h # 0.
h
Therefore, lim
f (:!+h, b) - f (a, b) = A,
h-0 h
i.e., fx (a, b) = A.
Similarly, by setting h = 0 and proceeding as above, we can prove that
fy (a, b)*= B.
This completes the proof of the theorem.
In view of Theorem 5, if f is differentiable at the point (a, b), then
f (a+ h, b + k) - f (a, b) = hfx (a, b) + kfy'(a, b) + h4 (h, k) + k$ (h, k)
where 4 (h, k) - 0 and $ (h, k) - 0 as (h, k) - (0, 0). I

From this we see that for small values of h and k we can approximate
-
f (a+ h, b + k) f (a, b) by the expression hf, (a, b) + kty @, b). 'his expression
is given a special name, as you will now see.
 Definition 3 : Let f (x, y) be a real-valued function defined in a neighbourhood of ~b4 Older
the point (a, b). If f (x, y) is differentiable at (a, b), then the linear function Paid ~ o t i v e s
T : R' -. R defined by and Wferedebility

T (h, k) = hf, (a, b) + kf, (a, b)

is called the differentid of f at (a, b). It will be denoted by df (a, b).
We will now give an example to show that a function may possess partial
derivatives and still not be differentiable.

Example 14 : I f f (x, 4') = I- x3 - y3

x 2 + y 2 , (x, Y) Z (0, 0)

then f possesses both the first arder pattial derivatives at the point (0, O), but is
not differentiable at (0, 0).

Now, lim = lirn

f (0, O+ k) - f (0, 0) -k - 0
and lim = lim - - -1

Therefore, both the first order partial derivatives exist at the point (0, 0) and
f, (0, 0) = 1, f, (0, 0) = -1.
Suppose, if possible, that f is differentiable at (0, 0). Then, by Remark 2 (i),
f (O+h, O+k) - f (0, 0) = hf, (0, 0) + kf, (0, 0) +J W . 4 (h, k)
where (h, k) -- 0 as (h, k) - (0, 0).
f (h, k ) - h + k
Now, 4 (h, k) =
JiZP
f(h, k)- h + k
This means that lim
ol. k) - .
(0. 0) Jh2+kZ
= 0

Now, if h '= r cos 8, k = r sin 8, then

f (h, k ) - h + k
= cos3 8 - sin3 8 - cos 8 + sin 8
Jm
f (h, k ) - h + k
Therefore, 0 = lim
7-J
= lim (cos3 8 - sin3 8 - cos 8 + . sin 0,)
&. k)-(0. 0) 1-0
.- 4 3 )
- - +
Now since the expression cos3 8 sin3 8 cos 8 sin 0 is independent of r, (3) .
implies that
cos3 9 - sin3 8 - cos'8 +
sin 8 = 0 for all 8. But this is not true. So we have ar-
rived at a contradiction, proving that the given function is not differentiable at
d
(0, 0).
We now present the results of this section in the following chart.

1
',

I
I
But the arrows in this chart cannot be reversed. We shall now give (without proof)'
a sufficient set of conditions, which would ensure the differentiability of the func-
tion under consideration.
T b t o r m 6 : If f is a real-valued function defined in a neighbqurhaod of (a, b)
such that
i) f, is continuous at (a, b),
I
and ii) fy exists at (a, b),
then f is differentiable at the point (a, b).
Similarly, the statement that f is differentiable at (a, b) if fx exists at (a, b) and f y
is continuous at (a, b) is true. Thus, the continuity of one of the partial derivative
and the existence of the other guarantees the differentiability of the function under
consideration.
Now a function, both of whose partial derivatives are continuous, is given a
.special name. Here is the precise definition.
befinition 4 : A real-valued function f of two variables is said to be continuously
differentiable at a point (a, b) if both the first order partial derivatives exist in a
neighbourhood of (a, b) and are continuous at the point (a, b).
Note that the above definition requires that a neighbourhood of (a, b) should be
contained in the domain D of the given function.
J

An immediate consequence of Definition 4 and Theorem 6 is the following.

Theorem 7 : A function, which is cont(nuously differentiable at a point is differen-
tiable at that point. -
Now we shall illustrate the above discussion with some examples.
Example 15 : Let f : R~ - R be a function defined by
C0 ifx = o.= y
/
We'll show that f is differmJAableat (0, 0). ..
Wc shall prove the result ,&ing Theorem 6.
'.
q h , 0) - f (0, 0)
Now, fx (0, Q) = lim
h-0 // h
=o /

Similarly, fy (0, 0) = 0 and for (x, y) # (0, O),

Using polar coordinates, x = r cos 8, y. = r sin 8, we g!et

1 fx (x, y)( = r I(cos4 6 sin 8 + 4 cos2 0 sin3 8 - sir? 8))
s 6J-, since sin8 s 1 andcos8 s 1.
.
This can be made less than a given E if we take (x( c -E and (y( < E
475 rn -a

This means that lim fx (x, y) = fx (0, 0).

(x, Y) - (0. 0)

Thus, we can say that fx is continuous at (0, 0). Consequently, f is,differentiable at

(0, 0) in view of Theorem 6.
Example 16 : We will now show that f (x, y) = sin x + x2 + 2xy is
differentiable everywhere.
Since fx (x, y) = sin x + cos x + 2x + 2y alnd
fy (x, y) = sin x + 2x
are continuous everywhere, it follows that f is differentiable everywhere in view of
Theorem 6.
The following example shows that the conditions statfed in Theorem 6 are sufficient
but not necessary. That is, a function can be differentiable at a.goint even when
none of the partial derivatives is continuous at-that point.
Example 17 : Cansider the function f : R2 - R given by

x2 sin -
1
+ y2-sin -
1
, if xy # 0
Y

# Oandy = 0

1 y2 sin ;
1
, if x = o and y # 0

( 0 ,ifx=O= y
,
We'll prbve that f is differeitiable at (0, 0). but neither f, nor f, is continuous at
(0, 0).
\

1' 1
2 x sin - r cos - if x # 0
Here f, (x, y) = x
ifx= 0
1 1
2 y sin - - cos - if y # 0 Even though the limits of two
and fy (x. y) = Y Y .given functions do not exist, the
O
ify = O limit of their difference may
hist. heref fore it is necessary tc
1 1
Since lim cos - does not exist, and lim t sin - = 0, it follows that
check that lim t sin -I exists in
r;-0 t
t -0 t t-0 t addition to npting that Lim cos
lim fx (x, y) and lim fy (x, y) do not exist. t
-I does not exist. 1-0

(x, y)-(O, 0 )
yx. O) y)-(Op
This means that f, and f, are discontinuous at (0, 0).

However, f (h, k) - f (0, 0) = 0.h + 0.k + h

!>
1
where <]imh sin - = 0 = lim ' k sin -.1
(h, k)-(0, 0) h (h, k)-(0. 0) k.
Therefore, f is differentiable at (0, 0).
See if you can solve these exercises now.

E20) Show that the function f : R 2 -R defined by

1
+ y sin - , xy # 0
Y

f (x, Y) =
-
xsin - ,
1
y = 0, x.# 0
X
1
ysin-, x=O, y#O
Y

is continuous but not differentiable at the origin.

E21) Check-the following functions for continuity and differentiability ht the
origin.

1
y sin - if x # 0
b) f (x, Y) = x
ifx=O 39
E22) Show that the following functions are differentiable everywhere.
a) f (x. y) = eX+Y
b) f (x, y) = 2 sinhx + 3 coshy

In the next section we shall discuss the differentiability of functions of three or

more variables.

5.4 DIFFERENTIABILITY OF FUNCTIONS

FROM R" -- R, n > 2

In the last section we defined and studied differentiability of real-valued functions

of two variables. Most of the results established in the last section can be extended
to real-valued functions of three or more variables. Let us start by defining dif-
ferentiability of real-valued functions of three or more variables..
Definition 5 : Let f be a real-valued function defined in a neighbourhood of the
point a = (al, a2,...,a,,). The function f is said to be differmthbk at the point a,
if there exist constants A,, A2,...,An (depending on the function and the point a)
such that
f ( a l + h l,...,a,,+h,,) - f (al,...,a,,)
n n

where each & -

0 as (hl, h2,...,h,) (0, 0,,...0). -
As in the case of two'variables we have the following results.
af
i) Ai = - at (a,, ...,a,,)
hi
ii) If f is differentiable at a, then f is continuous at a.
iii) f is differentiable at a if and only if
n
f (a+h) - f(a) = C hiAi + Ihl ml.h2....,h d
i-1

where 4(hl, h2,...,h,) -- 0 as h - 0,

h = (hl, h2,...,h,,) and (hi = /, C h:

iv) If f is differentiable at a, then f has all the partial derivatives at a.

V) If the partial derivatives8f f are continuous at a, then f is differentiable
at a. 'r
Don't worry, we don't expect ~ o u t prove
b them.
As for functions of two variables,$we &y that the linear function T defined by
n

is the differenthl of f at 8 and denote it by df (8).

Here are a few examples which will help you in understanding Definition 5.
Exmupk 18 : Let f (x, y, z) = x2+ y2 + z2. We'll prove that f is differentiable
everywhere.
r ( x + h , y + k , z+l) - f (x, y, z) - hfx (x, y, z) - kfy (x, y, z) - If, (x, Y, z)
= (x+h12 + (y+k)2 + ( z + 1 l 2 - x 2 - y 2 - z 2 - 2 h x - 2 k y - 2 1 2
= h2 + k2 + 12
= h 41(h, k, 1) + k& (h, k, 1) + 14, (h, k, 1)
where 6,(h, k, 1) = h, h (h, k. 1) = k, 63 (h, k, 1) = 1, First Order
''artid Dcdvatives
and each one of them tends to 0 a8 (h, k, 1) -- (0, 0, 0). and r3ifferentiabllity
Thus the given function is differentiable at every point.
Frnrrrpk 19 : Let f (x, y, z) = Ix+ y +zJ. Then we'll show that f is differentiable
+
at all those points (a, b, c) for which a b + c # 0. We'll also show that f (x, y, z)
is not differentiable at' those points (a, b, c) for which a + b + c = 0.
+
Suppose that the point (a, b, c) is such that a + b c = 0. Then, by definition,

fx (a, b, c) = lim
, C) - f
( a + ~ b, b* , Bit
the limit e ~ t s .
P-0 P
f (a+p, b, C) - f (a, b, C) = Ia+p+b+cI - Ia+b+cI = (PI, sincea+b+c = 0.
Therefore,.

lim
f (a+p, b, c ) - f (a, b. C) = lim ,- IPI
P-0 .P P-P P '

-
Now we know that lim lpl d g s not exist. This peans that fx (a, b, c) does not
P-0 P
exist if a + b + c = 0. Hence, f is not differentiable at such a point.
Now, let's take the case when a + b + c < 0. Then there exists a neighbourhd N
of (a, b, C)such that (x, y, Z) E N implies that x+ y + z < 0 because the function
(x, y, z) -x +y + z is continuous everywhere (see Theorem 6 of Unit 4). Thus for
+
those (p, q, r) for which (a p, b + q, c + r) E N, we have
. .
f (a+p, b+q, c+r) - f (a, b, c) = -p -q -r
= ~ f ~ ( a , b , c ) + q f ~ ( a , b , c ) + r f ~ ( a , b , c )r,i + b + c < o a n d a + p + b + c
< 0. Therefore,
( a + p + b + c J- ( a + b + c J (a+ b + c ( = -a-b-c and
because f, (a, b, c) = lim
P-0 P l a + p + b + c ( = -a-p-b-c.

Similarly,
fy (a, b, C) = -1 = f, (a, b, c).
Thus, f is clearly differentiable at (a, b,' c) if a + b + c < 0.
The remaining case i.e., when a + b + c > 0 is similar.

Now here are some exercises which you can try to solve.

E23) a) Show that every constant function is differentiable.

b) LA f : R' -R and g : R' -- R.be two functions such that each is
differentiable at a point (a, b, c). Show that f + g is also
differentiable.
c) Let f : R' -- R be a differentiable funetion:Prove that the function
Af is differentiable, where A is a constant.
E24) Show'that the following functions from R'
everywhere.
- R are differentiable
a) f (x, y, z) = x+2y+4z
b) f (x, y, z) = xy+yz+zx'
.
E25) Prove that the function f : R3 -- R defined by
ex+ eY
f(x,.y,z)=- is not differentiable at (0: 0, 0). , ,
Z

Let us briefly recall what we have covered in this unit.

5.5 SUMMARY
In this unit we have extended the concept of derivatives to functions of several
variables. In the process, we have
1) Defined first order partial.derivatives of a function at a point. If f : Rn -- R
then

lirn
f (xi, ~ 2...A
, - f (xi, x2,...,U
+. h, ~i+l,...,U
h-0 h
if it exists, is called the ith partial derivative of f at (xl, x2,...,x,,).
2) Discussed the methods to prove the existence of first order partial derivatives
at a point and to evaluate them.
3) Given emmples to establish that the existence of first order partial derivatives
at a point need not imply the continuity of the function at that point.
4) Defined a differentiable function of several variables: f is differentiable at
a .= (al, a2,...,a,,) if there exist constants Al, A2,..;,A,, s.t.

f (al+hl, ...,% + h J - f (al,...,an) = zn

i= l
hi4 + zn

i= 1
hi 44 (hl....,h,,)

5) Brought out the connection between the existence of partial derivatives,

differentiability and continuity of a function.

Continuity
of duivativa 1

5.6 SOLUTIONS AND ANSWERS

f (a+h, b) - f (a, b) c-c
El) fx (a, b) = lim = lim - = 0.
h-0 * h h-0 h

Similarly fy (a, b) = 0.

E2) By definition fx (0, 0) = lim

f (Q+h, 0) - f (O,O) = lim -
1
h-0 h h-0 h
1
Since lim - does not exist, f, (0, 0) does not exist.
h-0 h
f (0, k) - f (0, 0) 1
Similarly, fy (0, 0) = lim = lim -
k-0 ;r k-0 k
does not exist.
f (h, 0) - f (0, 0) = lim - = 1
h-0
E3) fx (0, 0) = lim
h-0 h h-0 h

f (0, k) - f (0, 0) = lim -0-0

and f,. (0, 0) = MI = 0.
k-0 k k-o k

E4) The function f (x, y) = 2x2 - xy+2y2 is a polynomial in x A d y.

Therefore the partial derivatives exist. By direct differentiation we get,
 af af First Order
E5) a) - = 2x cos (x2 - y) ; - = - cos '(x2- y) Pnrtinl Derlvntlves
ax ay end Differenthbllity

af , af af
c) - = yz cos xz ; - = sin xz ; - = xy cos xz
ax ay az

E6) u = ex cos y, v = ex sin y. Since the functions involved in u and v are

exponential and trigonometric functions, the partial derivatives exist. Then
au au
- = ex cos y and - = -ex sin y and
ax ay
av
-=
av au av
ex sin y and - =-ex cos y. Hence - = - and
ax aY ax ay

F (x+h, Y) - F (x, Y)
E7) Fx (x, y) = lim
h-0 h

= lim
f -
(x+h) + g Or) (f (x) + g (Y))
h-0 h
f (x+h) - f (x)
= lim = f'(x)
h-0 h

Fy (x, y) = lim F (x, y + k ) - F (x, Y)

k-0 k

= lim
f (x)+g (y+k) - (f (x) + g Or))
k-0 k

= lim
gOr+W-g(y)
= g'(y). *

k-0 k
E8) By definition,
= lim (f +g) (a+h, b) - (f +g) (a, b)
(!!) (a, b) h-0 h
 The converse is not necessarily true.
Consider the following functions f and g.
1 . if .(x, Y) f (0, 0)
' f (x, Y) =
0 if (x, y) = (0, 0)
0 if (x, Y) # (0, 0)
1 if '(x, y) = (0, 0)
Then, (f +g) (x, y) = 1 for all (x, y).
In E2) it has been shown that fx (0, 0) does not exist.
Shlarly, we can show that gx (0, 0) aIso does not exist.

H$wever,
a (f +g) exists at (0, 0) since f + g is a constant function.
-
ax
E9) The slope of the tangent at the point (1, 2, 14) to the curve of intersection of
the plane y = 2 and the surface z = 2x2+ 3y2 is given by
(- )(1 ,2J4)

-
az
Now, z = 2x2+3y2. So
Thus, the slope is 4.
ax
= 4x and hence
(- )(1,2,14)
= 4.

E10) To show the continuity at (0, 0) of f, let & > 0 be given a real number.
E
Choose 6 = -.fl Then,

Therefore, l ii-/, = 0 = f (0, 0).

x-0

Hence, f is continuous at (0, 0). Now, by definition

Since, lim docs not exist, fx (0. 0) does not exist.

h-0 h
Similarly, fy (0, 0) does not exist.

Similarly, f, (0, 0) = 0. Now, we have to show that f is continuous at (0, 0).

We have

Ixsin -
1
x
+ y sin -
Y .
s
I 1x1 + Iyl.

This implies that lim = (0, 0) = f (0, 0).

x-0 Y
Therefore, f is continuous at (0, 0).

f (O+h, 0) - f (0, 0)
E12) fx (0, 0) = lirn
h-6 h

= lirn
h-0
Similarly, fy (0, 0) = 0.
 similarly, for points (x, y) such that xy = 0 we get that
fx (x, Y) = fy (x, y) = 0.
Now, suppose x # 0 # y. Then both the partial derivatives at (x, y) exist
since f (x, y) is a quotient of.two differentiable functions in x and y. By
direct differentiation

- 2x5y*+ -
2xy3 4 ~ 5 ~
(x4+y2)2
and

But lim f (x, y) does not exist, because when we put y = x,

x-0
Y-0

x
' l i i f (x, y) = lim - = 0, and when we put
x-0 x-0 x4+x2
Y-0 ,
x4 1,
Y = x2, lim f (x, y) = lim
x-0 x-0
-
x4+x?
= -
2'

El,) a) Since 171 = -

IxI Iy' - 1x1for su y # 0,.
l ~ l
- lim -
XY
= 0 = f (0, 0). Hence f is continuous at (0, 0).
x-0 l ~ l
Y-0

Similarly, fy (0.0) = 0.
b) fx (1, 0) = 0. However, fy (1, 0) does not exist since

( 1 , - ( 1 , I -E ~ - ~1 1
=-= andlim - does not exlsr.
k k lkl k-0 lkl

-
a) f(x, Y) = xeY fx (x, y) = ey and fy (x, y) = xey. Proceeding as in
Example 10, we can show that fx is bounded in any neighbourhood of any

-
point o f It2.Therefore, by Corollary 1, f is continuou$ everywhere.
b) Now, f (x, y) = ,3ky .fx (x, y) = 3y and fy (x, y) = 3x. Both the
partial derivatives fx And fy satisfy the'coniiitiohs of Corollary 1.
Therefore f is continuous everywhere.
El5) a) A constant function of two variables is differentiable everywhere.
. .
b) If f is differentiable at (a,'b) E It2, theil cf (c E R) is.a4so differentiable at
(a, b).
1
I . c) If f and g are differentiable at (a, b) E It2, then. f +_ g is also differentiable
at (a, b).
.Partial Derivatives d) If f and g are differentiable at (a, b) then fg is also differentiable at
(a, b).
Now we shall check the validity of these statements.
a) Let f (x, y) = c be a given constant function. Let (a, b) be any point of
R ~ Thehi
.
f ( a + h , b + k ) - f (a, b) = c - c = 0

= 0.h + 0.k + h+ (h, k) + k$ (h, k).

where 4 (h, k) = 0 = 4 (h, k) for all h and k.
Since (a, b) was any arbitrary point of R
'
, f is differentiable everywhere.
b) Let f be differentiable at (a, b). Then there exist constants A and B and
functions 4 and $ which tend to zero as (h, k) -
(0, 0) such that
f ( a + h , b + k ) - f (a, b) = Ah +Bk + h4 (h, k) + k$ (h, k).
If c # 0, is a constant, then multiplying the above equation by c, we get
(cf) ( a + h , b + k ) - (cf) (a, b) = A'h + B'k + h4'@, k) + k$'(h, k)
where A' = cA, B' = cB, 4' = c4, $' = c$.
Now we have seen in Unit 4 that if 4 and $ tend to zero as
(h, k) -
(0, 0) then 4 ' and $ ' also tend to zero as (h, k) (0, 0).
Therefore cf is differentiable whenever c is a non-zero constant. If c is
-
zero, then cf is a constant function taking every point to zero and hence is
differentiable.
c) Since f is differentiable at (a, b) there exist constants A ' and B' and
functions 4' (h, k) and $' (h, k) which tend to zero as (h, k)
such that
(0. 0) -
f (a+h, b + k ) - f (a, b) = A'h + B'k + hd'(h, k) + k$'(h, k).
.Similarly, since g is differentiable at (a, b) there exist constants A " and
B" and functions 4 " (h, k) and $" (h, k) which tend to zero as (h, k)
(0, 0) such that
-
g ( a + h , b + k ) - g ( a , b ) = A"h + B"k + hr$"(h,k) + k$"(h.'k)
Then, (f 2 g) ( a + h , b + k ) - (f g) (a, b)
= (A' + A R ) h + (B' + B " ) k + h [ 4 ' ( h , k) * 4"(h, k)]
= Ah + Bk + h4 (h, k) + k$ (h, k)
where A = A ' + A " , B = B' + B", 4 (h, k) = df(h, k) 5 4 " (11, k)

Now since A and B are constants and functions 4 and $ tend to zero as
(h, k) -
(0, 0). f + g as well as f - g is differentiable at (a, b).
d) Proceed as in c). Then
fg ( a + h , b + k ) - fg(a, b) = f(a+h, b + k ) g(a+h, b + k )
- f(a + h, b + k) g(a, b) + f(a + h, b + k) g(a, b)
- f(a, b) g(a, b).
= f(a + h, b + k) [g(a + h, b + k) - g(a, b)] + g(a, b) [f(a + h, b + k) - f(a, b)]
= [f(a, b)+A'h+B'k+h+'(h, k)+k$'(h, k)] [ARh + B"k
+h4"(h, k)+k$" (h, k)]+g(a, b) [A'h+B'k+h4'(h, k)+k$'(h, k)]
= Ah + Bk + h4 (h, k) + k$ (h, k),
where A = A " f (a, b) + A'g (a, b), B = B"f (a, b) + B'g (a, b),
4 (h, k) = [ A ' + 4 ' (h, k)] [A"h + B"k + h4" (h, k) + k$"(h, k)]
+ f (a, b) 4 " (h, k) + g (a, b) 4'(h, k)
and (h, k) = [B'+$' (h, k)] [A"h + B"k + h4" (h, k) + k$"(h, k)]
+ f (a, b) $" Ch, k) + g (a, b) $' 01, k).
I
I E16) Here f (h, k) - f (0, 0) = h L + k + h k = 0.h + 1.k + 'hq5 (h, k) + k,b (h, k) First Order

,b (h, k) -
0 as (h, k) -
where 4 (h, k) = h = ,b (h, k) for all h, k, and therefore, 4 (h, k) -- 0 and
(0, 0). Thus, f is differentiable at (0, 0).
Partial Derivatives
a d Differentiability

I E17) Let f (x, y) = cos (x+ y). Cons~der

= cos (l+ h
4
u
+-
4
+ ) - ( + f)
= cos (* + h + k) - cos);(
2 9 ,

= - sin (h+ k)
sin (h + k) ] + k [ l -
= - h - k + h

= Ah + Bk +
[ 1-
h+k
h 4 (h, k) + k,b (h, k),
sin (h + k)
h+k I
sin (h + k)
whereA = - l , B = -1,4(h., k) = ,b(h, k) = 1 -
(h + k)
Now, lirn 4 (h, k)
01. k)--(O.0)

sin (h + k)
= lim ,b (h, k) = 1 - lim
(h. k) - (0. 0) h k - (0 0 h +k

sin (h + k) sin t
since lim = lim - = 1 where t = h + k .
01 k -0 0 h +k t-0 t

Therefore, f is differentiable at (1,

5)
4 4

E18) Suppose, if possible, that f is differentiable at (0, 0). Then there exist
constants A and B and functions 4 and ,b which tend to zero as
(h, k) -
(0, O), such that
f(O+h, O+ k) - f(0, 0) = A h + B k + ht#l (h, k) + k,b (h, k),
hk
i. e.. = Ah + Bk + h 4 (h, k) + k,b (h, k).

Put h = 0, k # 0. Then 0 = Bk + k$ (0, k)

i.e., B + ,b (0, k) = 0.
Taking the limit as k 0, we get B = 0. -
Similarly, by putting h # 0, k = 0, we get A = 0
Now if we let h = k # 0, then we get

la = Ah + Bh + h 4 (h, h) + h* Q h)

1
= A + B + 4 (h, h) + ,b (h, h)

and so taking limit as b - 0 we get A + B =

1
-
d2
which is impossible as A = 0 = B.
Thus we arrive at a contradiction and so f 'can not be differentiabld at (0, 0).
E19) a) You have already seen in Example 3, Unit 4 that
x2 - y Z
lim - does not exist at (0, 0). This shows that the function is
x-0 x2+y2
discontinuous at (0, O), and therefore not differentiable at (0, 0).
 P.rti.l Derivatives
b) If we put y = mx, f (x, y) =. -
L

1+ m 2
. This means that the limit of
f (x, ,y) as (x, y) - (0, 0) along y = mx is -
CI
L

1+ m Z
, and this is different
for different values of m.

Therefore, lirn
(x. y)-(O. 0)
A
x +y
does not exist and hence f is not continuous

at (0, 0).
x2+ y2
c) Whenweputy = mx, m# 1, then lim
(x. y)-(O.
-
0) X - Y
x (1 + m2)
= lim = 0. But f (0, 0) = 1.
X-o 1-m
*
Therefore, f is discontinuous at (0, 0).
d) x4+y4 s x4

This shows that lirn f(x, y) = 0, which is different from f(0, 0)

(x, Y)'(O. 0)

= 3. Thus f is not continuous at (0, 0).

E20) Now If (x, y)( s 1x1 + lyl in all cases.
lirn f(x, y) = f (0, 0) and hence f is continuous at (0, 0). Now, we have
(x. y)-(O. 0)

to show that f is not differentiable at (0, 0). In view of Theorem 5, it a

enough to show that either fx or fy does not exist.
Now fx (0, 0) docs not exist, since
1
hsinli - 0
f 01, 0) - f (0, 0) - = sin -1
h h h
1
and lim sin - does not exist. Therefore f is not differentiable at (0, 0).
h-0 h
/

E21) a) Now fx (0,O) = 0 = fy (0,O). '

y S-x2y3
At (~SY)# (W), fx (x, Y) = (Xz+y2)2 and fy (XJ) =
(x'+Y~~
Putting x = rcos8, y = rsine, we find that
Ifx (x,y) ( = r )sinStkos28sin% 1 s 2r = 2-/,
This implies that lim fx (x,y) = fx (0,O). Hence fx is continuous at
(X.Y) - (0.0)
(0,O) and fy (0.0) exists. Thus fx and fy satisfy all conditions of Theorem
6. Therefore, f is differentiable at (0,0), and hence f is also m d n u o u s at
(0.0).
b) Since I f(x,y) 1 s 1 yl, it can be easily shown that f is continuous at t(0,O).
Now, fx (0,O) = 0 and fy (0.0) = 1.
Suppose, if possible, that f is differentiable at (0,O). Then there exist func-
tions I$ and $, which. tend to zero as (h,k)
-
f(h,k) f(O.0) = 0.h + 1.k + h+(h,k) + k*(h,k).
(0,0), such that -
Let h = k .# 0. Then
1
h sin - = h + hq6 (hh) + h* (h,h)
h

1 1
Therefore, lirn sin - = 1, which contradicts the fact that lirn sin -
h-0 h h-0 - --
h
I 48 does not exist.
 E22) a) Since f, (x,y) = = fy (x.y) it can'be easily seen that f, and fy are Flnt Order
continuous everywhere. Therefore, f is differentiable everywhere. Partial Derlvatlves
and Differentlability
b) f, (x,y) = 2 coshx and fy (x,y) = 3 sinhy.
So fx and fy are continuous everywhere and hence f is differentiable
everywhere.
E23) a) Let f(x,y,z) = k be a constant function. Then

= k-k = ChiAi+ C hi'#+

i= K i= 1
where Ai = 0, i = l , 2, 3.
and di (x,y,z) = 0, i = 1, 2, 3.
So f is differentiable everywhere.
b) Since f is differentiable at (a, b, c) there exists a neighbourhood N1 of
(a, b, C) such that for all hi, i = 1, 2, 3, such that (a + h,, b + h2, c + h3) c
N,, there exist constants A; ,'i = 1, 2, 3, and functions I${ : R' R, -
-
i = 1, 2, 3, which tend to zero as (hl, h2, h3) (0, 0, 0). such that

Similarly, since g is differentiable at (a, b, c) there is a neighbourhood N2

of (a, b, c), constants A,", i = 1, 2, 3, functions 4,", i = 1, 2, 3, such that
3 3
g(a + hl, b + hz, c + h3) - g(a,b,c) = hi A," + hi 6,"
i= 1 i=l

Let M = N1 n N2. Then N is a neighbourhood of the point (a, b, c) such

that for all hi, i = 1, 2, 3, such that
(a+hl, b+h2, c+h3) E N, we have

(f+g) (a+hl, b+h2, c+h3) - (f+g) (a, b, c)

where A; + A;, i =.l, 2, 3, are constants and +i = & + ++,

i = 1, 2, 3, are functions from R' -R which tend to zero as (X.~,Z)-
(a,b,c).
Hence f + g is differentiable at (a,b,c).
c) Can be proved similarly.
E24) Let (a,b,c) be any point and let the function f(x,y,z) = x + 2y + 42 is defined
in a neighbourhood N of (a,b,c). Let hl, hz, h3 be numbers such that the
point (a + hl, b + h,, c + h3) belongs to the neighbourhood N of the point
(a,b,c). Then
i
f(a+ hl, b + h2, c + h3) - f(a,b,c)
= (a+hl) + 2(b+h2) + 4 ( c + h 3 ) - a - 2 b - 4 c
b = hl + 2h2 + 4h3
= A.hl + B.h2 + C.h3 + hl+ + h2$ + h35

where A = 1, B = 2, C = 4 are constants and +(x,y,z) = $(x,y,z) =

5 (x,y,z) = 0. Hence f is differentiable at the point
I (a,b,c) and so everywhere.

Let (a,b,c) be any point.

f(a+ h, b, c) - f(a,b,c)
fx (a,b,c) = lim
h-0 h
= lim
(a+h) b + b c + c (a+h) - (ab+bc+ca) '

h-0 h
bh+ ch
= lim -
h-0 h
 = b+c
Similarly, fy (a,b,c) = c + a,
f, (a.b,c) = a + b .
That is, f, (a,b,c), f, (a,b,c), f, (a,b,c) exist and are continuous.
Therefore f is differentiable at (a,b,c). This is true for all pointssf R ~ .
~ e n c ef is differefitiahle everywhere.
E25) To show that E is not diffe~ntiableat (0,0,0), f is enough to show that
either f,, fy or f, does not exist at (0,0,0).

2
Since lim - does not exist, f, does m t exist at (0,O.O).
r-o r2
Hence f cannot be differentiable at (0,0,0).