Multivariate

UNIT 1 MULTIVARIATE CALCULUS-I Calculus-I

Structure
1.0 Aims and Objectives
1.1 Introduction
1.2 Functions of Several Variables
1.2.1 Functions of Two Independent Variables
1.2.2 Level Curves
1.2.3 General Multivariate Functions
1.3 Partial Derivatives
1.3.1 First- Order Partial Derivatives
1.3.2 Second-Order Partial Derivatives
1.4 Total Differentials and Total Derivatives
1.4.1 Total Differentials
1.4.2 Total Derivatives
1.5 The Chain Rule for Multivariate Functions
1.6 Implicit functions
1.7 Homogeneous and Homothetic functions
1.8 Let Us Sum Up
1.9 Answers to Check Your Progress Exercises

1.0 OBJECTIVES
After reading this unit you will be able to:
• Describe functions where a dependent variable depends on more than
one independent variable;
• Explain the concept of a partial derivative;
• Discuss the techniques of total differentiation, and obtain total
derivatives;
• Explain what homogeneous and homothetic functions are; and
• Explain the chain rule with regard to functions of more than one variable.

1.1 INTRODUCTION
In the course on mathematical methods in economics (BECC 102) that you
studied in the previous semester, you learnt about differentiation. However,
there you studied about differentiation of univariate functions, that is,
functions where the dependent variable is dependent on one independent
variable. This Unit takes up the case of function of more than one
independent variable. It also considers differential calculus pertaining to
multivariate functions. The Unit first discusses the concept of function of
several variables in section 1.2 Section 1.3 deals with partial derivatives, 11
Functions of Several
Variables
which means differentiating the function with respect to one independent
variable, keeping others unchanged. Section 1.4 takes up the case of total
differentiation, which means the change in the dependent variable as a result
of changes in all the independent variables. This section also discusses total
derivatives. The next section, section 1.5 discusses the chain rule that you
studied in the previous unit, but also somewhat different and new. The
subsequent section, section 1.6, discusses the differentiation of functions that
are defined implicitly. Of course, the discussion of implicit function is with
regard to functions, which when expressed explicitly, would be multivariate.
This section also discusses the important implicit function theorem. Finally,
in section 1.7 you are introduced to an important type of multivariate
functions, called homogeneous functions. Other than the definition of a
homogeneous function, some important properties are discussed, including a
theorem called Euler’s theorem. Homothetic functions are defined as well,
and their relationship to homogeneous functions is mentioned.

Let us state explicitly that the present unit discusses the content of all the
topics and concepts only in mathematical terms. That is, the mathematics of
these concepts is discussed in the present unit. The next unit, unit 2 is entirely
devoted to the applications of the ideas in this unit to economics, like the
theory of the consumer, producer’s theory, markets, macroeconomics, and so
on. Just understand the content of this unit well, and you will appreciate
better the matter in the next unit, as well as what you learn in your
microeconomics and macroeconomics courses.

1.2 FUNCTION OF SEVERAL VARIABLES

In this section we introduce you to functions where the dependent variable is
a function of more than one independent variable. We begin with the simplest
case where the dependent variable is a function of two independent variables.
Next we discuss the idea of level curves and finally we take up the general
case of n- independent variables.
1.2.1 Functions of Two Independent Variables
In the course on mathematical methods in economics that you studied in the
last semester, we have discussed almost exclusively functions of one variable
�� = �(�)� that is, functions where a variable is dependent on one
independent variable. But in real life one comes across cases where more
than one independent variables influence one dependent variable. For
simplicity, let us begin by considering a function which shows an
independent variable, say z, being a function of two independent variables, x
and y. In notation
z = f ( x, y )

We can define the above function in the following way. A function f of two
real variables x and y with domain D is a rule that assigns a specified number
12 f(x, y) to each point (x,y) in domain D. In the above function, x and y are
 Multivariate
called the independent variables, or arguments of the function f. The variable Calculus-I
z is called the dependent variable. The domain of the function f is the set of
all possible ordered pairs of the independent variables, while the range is the
set of corresponding values of the dependent variable. In some contexts, z is
called the endogenous variable while x and y are called exogenous variables.

Apart from simplicity, one reason for beginning with functions of only two
variables is that we are able to draw diagrams. We can depict f ( x, y )
diagrammatically in three dimensional space by drawing three mutually
perpendicular axes Ox, Oy and Oz. Figure 1.1 below shows such a diagram
depicting a surface in three-dimensional space, and a point

(x0, y0, z0) on that surface.

z0

(x0, y0, z0)

y0 y
x0

Now let us suppose that this surface is traced out by the function z = f ( x, y ) ,
and that z is traced out as (x,y) varies over the xy-plane. Then z0 = f ( x0 , y0 ) .

1.2.2 Level Curves

If we have z = f(x,y), the graph of this function in three-dimensional space
can visualized as being cut by horizontal planes that are parallel to the x y –
plane. The intersections between the planes and the graph can be projected
onto the x y – plane. If the intersecting plane is at z = k then the projection
onto the x y – plane is known as the level curve or contour at height k for the
function f. the contour or level will consist of points that satisfy the equation
f ( x, y ) = k

Think of a map. It shows the geographical location of a place (in terms of

latitudes and longitudes, for example). To show altitude or height, what we
can do (in maps showing physical properties of places), is to draw a set of
13
Functions of Several
Variables
contours or level curves connecting those points on the map that lie at the
same altitude, or elevation above sea level. So on a two-dimensional map
(basically representing direction, distance, area, and so on) we can show the
third dimension (altitude) by drawing a set of level curves.

It is the same idea that we apply to the diagrammatical representation of

functions of two variables. You can think of the graph of a function in three-
dimensional space (as shown on a two-dimensional diagram) as being
represented by horizontal planes that are parallel to the x-y plane. These
intersections between the planes and the graph, we project onto the x-y plane.

At this point, think back to the unit on coordinate geometry that you studied
in course BECC 102. There you studied that the equation of a line parallel to
the y-axis is x = c , while the equation of a line parallel to the x-axis is y = c

In the present context, we are talking of the equation of a plane (because it is

in three dimensions x-y-z). hence the equation of the plane parallel to the xy-
plane will be, in an analogous fashion, z = c . This shows the projection of
the intersection of this plane with the graph at height c for z. since z = f ( x, y )
, this projection is called the level curve for f at height c. This level curve
then consists of all points that satisfy the equation
f ( x, y ) = c

The level curve thus connects points whose functional values are equal. In the
above, the level curve connects all the points for which the value of f is c. it is
the locus of all points, that is, the combination of x and y, for which the value
of z is equal to c.
1.2.3 General Multivariate Functions
In subsection 1.2.1 above, we had discussed about functions of two variables.
We can extend the discussion to functions in which the dependent variable is
a function of several independent variables. Let us denote a list of n variables
by ( x1 , x2 ,...xn ) We say that the variable x is indexed by i, where i =
1,2,…,n. This collection of n-variables (each xi is a real number) is called a
vector. A vector is an ordered n-tuple. This was mentioned in Units 1 and 2
of course BECC 102 that you studied in the last semester. You will study
about vectors in much greater detail in Block 3 of the present course.
Coming back to the vector x = ( x1 , x2 ,...xn ) , suppose a variable z is a function
of all these n variables. We denote this function as z = f (x1 , x2 ,..., xn ) .

One point that presents itself immediately is that now we are in n+ 1

dimension. Suppose n =2 where n is the number of independent variable.
This is what we discussed in subsection 1.2.1. There n + 1 was = 3 and we
could depict f on a diagram. When n is greater than 2, we end up with more
than 3 dimensions, and the function cannot be depicted diagrammatically. So
we have to think of the function conceptually and in an abstract manner.
When a line is generalised to two-dimensions, it is called a plane. Above that
14
 Multivariate
it is called a hyperplane. A general surface in higher dimensions is called a Calculus-I
hypersurface. Just remember that the function z = f ( x, y ) , which can as well
be depicted z = f ( x1 , x 2 ) is simply a special case of z = f (x1 , x2 ,..., xn ) ,
where n = 2.

1.3 PARTIAL DERIVATIVES

In your study of derivatives in the course BECC 102, you came to realise that
differential calculus studies how the dependent variable changes due to a
change in the independent variable, when the change in the independent
variable is infinitesimally small. In that course you studied the derivatives of
functions of a single independent variable. In this unit we have discussed in
the section above about functions where the dependent variable is a function
of several independent variables. So intuitively you can think of exploring the
idea about how to study changes in the dependent variable due to changes in
the independent variables. Consider the simplest multivariate function
z = f ( x1 , x 2 ) . We want to see how changes due to changes in x1 and x2.
Now we can study the change in z when both the independent variables
change together or when one of them changes and the other does not. Extend
the idea to a general multivariate function z = f (x1 , x2 ,..., xn ) . The rest of the
unit is concerned with a study of such changes. In this section we deal with
the study of derivatives of the function due to a change in only one of the
independent variables at a time. Subsequent sections will deal with situations
of all independent variables changing. When we take derivatives of the
function with respect to an independent variable, keeping other independent
variables constant, it is called a partial derivative. Let us begin the study of
partial derivatives
1.3.1 First- Order Partial Derivatives
Consider the function z = f (x1 , x2 ,..., xn ) . Here let us assume that the variables
xi (i = 1,2,…,n) are all independent of one another so that each variable can
individually vary without influencing the other independent variables.
Suppose there is a change in x1 by Δx1 while x2,…xn all remain unchanged
(fixed), there will be a corresponding change in z, namely Δz. The quotient
∆z
in this case can be expressed as
∆x1

∆z f ( x1 + ∆x1 , x2 ,..., xn ) − f ( x1 , x2 ,...xn )

=
∆x1 ∆x1

∆z
If we take the limit of as Δx1 → 0, the limit we obtain is called the partial
∆x1
derivative of with respect to x1. The term partial derivative is used to
indicate that the other independent variables are held constant. We can derive
similar partial derivatives of the function with respect to each of the other
independent variables. The process of taking partial derivatives is known as
partial differentiation.
15
Functions of Several
Variables
We saw in course BECC 102 that derivatives are denoted by the symbol d.
Thus if we have y = f(x) then the derivative of y with respect to x is denoted
dy
. Partial derivatives are denoted by the symbol ∂. This is a variation of the
dx
lower-case Greek letter ‘delta’ δ. Hence we write the above partial derivative
∂z
as We call this “partial derivative of with respect to x1. For the generic
∂x1
∂z
variable xi, we write the partial derivative of with respect to xi as
∂xi

∂
We sometimes write the partial derivative as z . With such a symbol, the
∂x i
∂
part can be taken as a mathematical operator symbol showing “taking
∂x i
partial derivative (of some function)”. Since z is a function of xi (i =
∂f
1,2,…,n), we can also denote the partial derivative by . We have seen
∂xi
dy
earlier that when we have y = f(x), then is also sometimes denoted f ′( x ) .
dx
In the case of partial derivatives, we sometimes use subscripts to denote the
∂f
partial derivatives. Thus is sometimes denoted f i
∂ xi

If we denote a function as , say z =f(x, y, w, v), then we can denote the partial
derivatives as f x , f y , f w , f v rather than f1 , f 2 , f 3 , f 4

To sum up, suppose we have z = f ( x1 , x2 ,..., xn ) .

Then we can depict partial derivative of z with respect to, say x3, by
∂z ∂f ∂
or or f or f 3
∂x 3 ∂x 3 ∂x 3

Let us now take a few examples to see how partial derivatives are computed.
For simplicity let us consider functions of two independent variables. To
compute partial derivatives, remember two things: first, when you take partial
derivative with respect to one variable, treat the other variable(s) as constant;
and second, remember that the derivative of a constant equals zero. Also
remember that if you have a variable multiplied by a constant, then the
derivative of this product with respect to the variable is the product of the
constant and the derivative of this variable. For example, (d [cx]/dx =cdx/dx;
and d[cx2]/dx =2cx). Moreover, the usual rules of differentiation, like sum
and difference rules, product and quotient rules, and the composite function
rules hold in the case of partial derivatives also.
Example 1 Let f ( x, y ) = x 3 y + y 4

Partially differentiating with respect to x,

16
 ∂f ∂ 3 ∂ 4 Multivariate
= x y +
∂x ∂
( )
∂x
y ( ) Calculus-I

The first term on the right-hand side is 3 x 2 y and the second term is zero
(because y is a constant, and so is y raised to the power 4.
∂f
Thus = 3x 2 y
∂x
Similarly
∂f ∂ 3 ∂ 4
=
∂y ∂y
( )
x y +
∂y
( )
y = x3 y + 4 y3

Example 2: Given z = f ( x1 , x2 ) = 4 x12 + x1 x2 + 3x22

To find the partial derivatives, we have to remember that when we are

computing partial derivative with respect to x1, we must treat x2 as constant.
The variable x2 will drop out if it is an additive constant, such as the third
term in the above example, but will be retained if it is a multiplicative
constant, like in the second term in the above example.
Thus we have:
∂z
= f1 = 8 x1 + x 2
∂x1

Similarly for computing the partial derivative with respect to x2, we treat x1
as constant. We obtain
∂z
= f 2 = x1 + 6x 2
∂x 2

Example 3 Given z = f (w, v ) = (w + 4 )(3w + 2v )

We can obtain the partial derivatives by using the product rule. For partial
derivative with respect to w, we hold v constant. We obtain
f w = (w + 4 )(3) + (3w + 2v )(1) = 2(3w + v + 6 )

For partial derivative with respect to v, we hold w constant, and get:

f v = (w + 4 )(2 ) + 0(3w + 2v ) = 2(w + 4 )

Example 4 Let us see in this example how the quotient rule is used
(
Suppose z = (3w − 2v ) / w 2 + 3v )
We have
∂z
= fw =
( )
3 w 2 + 3v − 2 w(3w − 2v ) − 3w 2 + 4wv + 9v
=
∂w w 2 + 3v(2
) w 2 + 3v 2( )
∂z
= fv =
( )
− 2 w 2 + 3v − 3(w − 2v ) − w(2w + 9)
=
∂v (
w 2 + 3v
2
) w 2 + 3v
2
( )
17
Functions of Several
Variables
In the above we considered first-order partial derivatives. These are partial
derivatives of the given function with respect to the arguments. A basic point
about partial derivatives may be made. The first-order partial derivatives are
themselves functions of the arguments of the original function, that is the
first-order partial derivatives are themselves the functions of the independent
variables in the original function.
Suppose we have z = f (x1 , x 2 )

∂f
Then f1 = = g ( x1 , x 2 ) and
∂x1

∂f
f2 = = h( x1 , x2 )
∂x 2

In the above g and h denote functions.

We may make a final point about first-order partial derivatives. Suppose we
∂f ∂f ∂f
have z = f (x1 , x2 ,..., xn ) . Let us compute , ,..., . The collection of
∂x1 ∂x 2 ∂x n
these partial derivatives in a collection of n-real numbers for specific values
of x1 , x2 ,..., xn . This collection is an n-tuple or vector. This vector is called a
gradient vector denoted by ∇ or grad f.

1.3.2 Second-Order Partial Derivatives

In the above sub-section we discussed first-order partial derivatives. Let us
now turn to second-order partial derivatives. Let us consider the function
z = f ( x, y ) . This function can yield two first-order partial derivatives
∂f ∂f
fx = and f y =
∂x ∂y

We mentioned towards the end of the previous sub-section that the first-order
partial derivatives are themselves functions of x and y, that is,
f x = g ( x, y ) and

f y = h ( x, y ) .

This means that the functions g and h can themselves be partially

differentiated with respect to x and with respect to y. Now notice one thing:
f x is itself a partial derivative and it can be partially differentiated with
respect to x, and with respect to y. The same is true of fy
∂
Just as we have g ( x, y ) ,we can differentiate g(x,y) with respect to y.
∂x
Similarly

fy = h(x,y) can be differentiated with respected to x and with respect to y

18
Let us differentiate f x = g ( x, y ) Multivariate
Calculus-I

∂f x ∂  ∂z 
We get =  
∂x ∂x  ∂x 

∂2z
The right hand side is written
∂x 2

The second order partial derivative of z (or the function f) with respect to x is
thus written:
∂f x ∂  ∂z 
f xx or f x . It’s also written or  
x ∂x ∂x  ∂x 

Similarly, we can have the second-order partial derivative of z with respect

∂ ∂ 2 z ∂  ∂z 
to y. We thus have : f yy ≡ f y or =  
∂y ∂y 2 ∂y  ∂y 

This denotes the rate of change of fy with respect to y, while x is held

constant
Please remember that we saw that fx is a function of y, and that fy is a
function of x as well.
f x = g ( x, y ) , and f y = h( x, y )

Hence we can obtain two more second-order partial derivatives:

� ��
1) ��� = �� ����

This is the partial derivative of z first with respect to y and then with
respect to x.
� ��
2) ��� = �� ����

This is the partial derivative of z first with respect to x and then with
respect to y.

Two points may be made here:

�� � �� �
1) f xy is also denoted by and �yx by
���� ����

2) the order of differentiation is from right to left (←). For example f xy

means first differentiate z with respect to. y and then with respect to x.
Similarly �yx means first differentiate z with respect to x and then with
respect to y. The partial derivatives fxy and fyx are called cross-partial
derivatives.

Young's Theorem
In the case of continuous functions with continuous partial derivatives, fxy
=fyx. This is called Young's Theorem and can be stated as under:
19
Functions of Several
Variables
The order of cross partial derivatives does not matter. That is, it does not
matter whether a function is differentiated first with respect to x and then
again differentiated with respect of y or vice versa. Symbolically:
��� = ���

The above was for the case of second-order derivatives

Check Your Progress 1

1) Explain the concepts of (a) A multivariate function (b) Partial derivative.

…………………………………………………………………………….
…………………………………………………………………………….

…………………………………………………………………………….

…………………………………………………………………………….

…………………………………………………………………………….
2) Find the partial derivatives fx and fy of the functions
i) f(x, y) = 5x2 + 6y2
ii) f(x, y) = -10xy2
…………………………………………………………………………….
…………………………………………………………………………….

…………………………………………………………………………….
…………………………………………………………………………….

…………………………………………………………………………….
…………………………………………………………………………….

3) Determine all first-order and second-order derivatives for the function

f(x, y) = 8x3 – 4x2y + 10y2

…………………………………………………………………………….

…………………………………………………………………………….

…………………………………………………………………………….
…………………………………………………………………………….

…………………………………………………………………………….

…………………………………………………………………………….

…………………………………………………………………………….

20
 Multivariate
1.4 TOTAL DIFFERENTIAL AND TOTAL Calculus-I
DERIVATIVES
Let us consider again the function z = f ( x, y )

The partial derivatives gave us a small change in z when there was a small
change in x holding y constant, or a small change in y holding x constant. Let
us now study what happens to z if both x and y were to change. For that we
turn to the study of differentials
1.4.1 Total Differential
Let us begin by considering a function of a single variable, z =f(x). for a
change in x, Δx, the change in z will be Δz. In the course BECC 102, we have
seen dz/dx to be the change in z when there is a small unit change in x. so,
when there is change of Δx in x the change in z will be:
dz
∆z = ∆x
dx
Writing dz for Δz and dx for Δx, we get
dz
dz = dz
dx
This dz on the left-hand side is called the differential of z.

Now let us consider

If z f x, y is a function, then the total differential dz can be expressed as:

z z
dz fxdx f yd y dx d y approximately
x y

∂z
shows
Let us try to understand it in the following manner. In the above,
∂x
the incremental change in z when x changes by a small amount. So the
∂z
change in z due to change of dx in x is dx
∂x
∂z
Similarly, the change in z due to a change of dy in y is dy . Thus the total
∂y
change in z due to change in both x and y of dx and dy respectively is
z z
dz fxdx f yd y dx dy
x y

This is known as the total differential of z.

The expression dz shows the increment in the function z f x, y when

there is an infinitesimal increments in x and well as y . For example

21
Functions of Several
Variables
if z x3 y3 , then total differential can be expressed as
2 2
dz f x dx fydy 3x d x 3 y d y

If we take a general multivariate function

z = f ( x1 , x2 ,..., xn ) then

∂z ∂z ∂z
dz = dx1 + dx2 + ... + dxn , or
∂x1 ∂x 2 ∂x n
n
dz = f1dx1 + f 2 dx2 + ... + f n dxn = ∑ f i dxi (recall the rules and properties of
i =1

summation as discussed in course BECC 102)

Let us now discuss some rules of differentials.

Rules of differentials
The following rules on total differential will be found useful. Let z and w
represent two functions of x and y , then

1) dk = 0 (constant-function rule)
2) d w z dw dz (Sum-difference rule)
= ( f x dx + f y dy ) ± (g x dx + g y dy )

3) d w z w.d z z dw
w(g x dx + g y dy ) + z ( f x dx + f y dy ) (Product rule)

w z.dw wdz
4) d
z z2
z ( f x dx + f y dy ) − w(g x dx + g y dy )
= (Quotient Rule)
z2
5) ( )
d kz n = knz n−1dz (power-function rule)

6) The chain rule:

If z = z(u) and u = u(x) then

d
dz = d (z (u )) =
(z )du
du
Here du is not an arbitrary increase in u but happens to be a differential
of u. we have
d
du = d (u (x )) = (u )dx . Hence
dx
d  d 
dz =  ( z )  (u )dx 
 du   dx 

22
 Multivariate
Example. Calculus-I

If z u n , where u f x, y , then
d n
dz u du nu n 1 du
dx
Let us now solve some problems on total differentials.
1) Find du when u 3x3 2 y2 y3

Answer:. Total differential du is given by:

du f x d x f y d y 9 x 2 d x uy 3 y 2 d y
9x2d x y u 3y d y

2) Find total differentials of the following functions

x2 y2 2
y2
a) u b) w ex c) u log x 2 y2
x2 y2
x2 y2
Answer: a) u , apply quotient rule
x2 y2
z ( f x dx + f y dy ) − w(g x dx + g y dy )
=
z2

(x 2
)( ) (
+ y2 d x2 − y2 − x2 − y2 d x2 + y2 )( )
(x 2
+y 2 2
)
x2 y 2 2 xdx 2 ydy x2 y 2 2 xdx 2 ydy
2
x2 y2
4 xy 2 dx 4 x 2 ydy
2
x2 y2
2
y2
b) w ex
Put u x2 y 2 so that w e u and dw eu du (1)

Also du d x2 d y2 2 xdx 2 ydy (2)

From (11) and (12), we get

2
y2 2
y2 2
y2
dw e x 2 xdx 2 ydy 2 xe x dx 2 ye x dy

c) u log x 2 y2

Let us try it by using the formula:

1 2x 1 2y
du f x dx f y dy dx dy
x2 y2 x2 y2

2x 2y 2 xdx 2 ydy 2 xdu ydy

2 2
dx dy
x y x 2
y 2
x2 y2 x 2
y2
23
Functions of Several
Variables
1.4.2 Total Derivatives
Let us go back to our familiar function of two variables
z f x, y

Here we assume that x and y are independent variables, and also independent
of each other.

Consider the case where x and y are not independent variables but are
dependent variables of other functions, for example,
x = g (t ), y = h(t ) , where t is the independent variable. Thus, z is a function of
x and y; x and y are functions of t. We want to investigate as to what would
be the derivative of z with respect to t. What would be dz/dt?
∂z
We know is the change in z due to a small unit change in x holding y
∂x
∂z dx
constant hence will be the amount of change in u due to a small unit
∂x dt
∂z dx
change in t, that is transmitted through x. similarly is the amount of
∂y dt
change in z due to a small unit change in t transmitted through y.

Hence, the total change in z due to a small unit change in t will be the sum of
these two effects. We write
dz ∂z dx ∂z dx
= +
dt ∂x dt ∂y dt

This we can also express as

dz dx dx
= fx + fy
dt dt dt
dz
This is called the total derivative of z with respect to t.
dt
If we have z = (x, y, u, …) where x = x(t), y = y(t), u = u(t), …
dz ∂z dx ∂z dx ∂z du
Then + + + + ....
dt ∂x dt ∂y dt ∂u dt

Now let us again go back to the function

z f x, y

Now instead of thinking of x and y being dependent variables that are

functions of an independent variable, say, t, let us suppose x is a function of
y. so our given function is
z f x, y

, where x = g ( y )
24
 Multivariate
We can combine the two functions f and g into a composite function: Calculus-I
z = f [g ( y ), y ]

In the above we see that z, x, and y are related as follows: the variable y
affects z through two channels. It can affect z directly through the function f,
and indirectly through the function g. thus the variable y is the ultimate
source of variation in z. The indirect effect of y on z can be represented as
∂z dx
. The direct effect is simply fy
∂x dy

Hence the total derivative of z with respect to y is obtained by combining the

direct and the indirect effects;
dz ∂z dx
= + fy
dy ∂x dy

It can also be obtained alternatively by simply taking total derivative in the

usual manner:
dz ∂z dx ∂z dy
= +
dy ∂x dy ∂y dy

dz ∂z dx ∂z
Thus = +
dy ∂x dy ∂y

Check Your Progress 2

1) a) Explain what you understand by total differential.

b) How are partial derivatives related to total derivatives?

……………………………………………………………………….

……………………………………………………………………….

……………………………………………………………………….

……………………………………………………………………….
……………………………………………………………………….

2) Find total differential of the following functions.

1 1
x
a) u 3x 3
2y 2
b) u c) x AL K
2 2
x y

……………………………………………………………………………

……………………………………………………………………………

……………………………………………………………………………

……………………………………………………………………………

……………………………………………………………………………
25
Functions of Several
Variables 1.5 THE CHAIN RULE FOR MULTIVARIATE
FUNCTIONS
You are aware of the chain rule for function of one variable. It is based on the
function-of-a-function rule. Let us suppose z is a function of y, and y in turn
is a function of x. We can depict this as � = ���(�)�. we know in this case
how to take the derivative of z with respect x: we first take the derivative of z
with respect to y and then multiply this with the derivative of y with respect
to x. We can write this as follows:
�� �� ��
=
�� �� ��
As similar process works in the case of partial derivatives. There is a chain
rule applicable to the case of partial derivatives as well. We only have to take
care in following the rules of partial differentiation. We shall try to make this
clear with the help of some examples.

Just a matter of notation for the examples: unlike earlier, we denote one
argument in the function by z. Earlier, we denoted the dependent variable by
z. Please do not get confused.

Example 1
Suppose we have � = �(�, �, �)and here x, y, z are themselves each a
��
function of some variable, say t. then we can find �� as follows:

�� �� �� �� �� �� ��
= + +
�� �� �� �� �� �� ��
Notice that this is just the total derivative that we used in the previous
section. This gives us occasion to mention that the total derivative is used
when the variables inside the parentheses (the argument of the function) are
themselves functions of only one independent variable. In the example above,
x, y, z are functions of the single variable t. In the next example we consider
a case where variables that are arguments in the given function, are
themselves functions of more than one variable. This is where we use the
chain rule, and have to use partial derivatives for composite function. Just
study the examples below, and it will become clear.

Example 2
If we have � = �(�, �, �) and x, y, z are themselves each a function of
variable r and s then
�� �� �� �� �� �� ��
= + +
�� �� �� �� �� �� ��
and
�� �� �� �� �� �� ��
= + +
�� �� �� �� �� �� ��
26
 Multivariate
The next example could also have been depicted using total derivatives as we Calculus-I
did in the previous section.

Example 3
Suppose � = �(�, �, �) and if y and z themselves depend on x, that is,
� = �(�), � = �(�). Then
�� �� �� �� �� ��
= + +
�� �� �� �� �� ��
Example 4
Let � = �(�, �, �) and if x depend on t, y depends on x, and z depends on
y, that is, � = �(�), � = �(�), � = �(�). Then
�� �� �� �� �� �� �� �� �� ��
= + +
�� �� �� �� �� �� �� �� �� ��
Notice that in the above example, z depends on y, which depends on x, which
depends on t. Thus z and y ultimately depend on t.

Example 5
Suppose � = �(�, �, �) where x = x (r,s), y = (r), z = z(y).Then
�� �� �� �� �� �� �� ��
= + +
�� �� �� �� �� �� �� ��
and
�� �� ��
=
�� �� ��
So now you should be able to grasp the technique of the chain rule in the case
of partial derivatives, and should be able to carefully observe which variable
depends on which other(s), and carry out the required differentiation. You
should be able to see from the way the variables are related to each other how
to use the chain rule, and when to take the total derivatives.

Let us take one final case and see how the chain rule and total derivatives
could be combined to obtain the relevant results.

Example 6
Suppose y = f (x1 , x 2 ) where

x1 = g (w1 , w2 ) and

x 2 = h(w1 , w2 )

The partial derivatives of y with respect to w1 and w2 are calculated by

making use of the composite-function rule (chain rule) that we show below.
Let us begin by taking total differential of y:
y y
dy dx1 dx2
x1 x2
27
Functions of Several Since we have x1 and x2 as functions of w1 and w2, we can compute the other
Variables
total differentials:
∂x1 ∂x
dx1 = dw1 + 1 dw2 and
∂w1 ∂w2

∂x 2 ∂x
dx 2 = dw1 + 2 dw2
∂w1 ∂w2

Substituting for dx1 and dx2 in the equation for the total differential of y, and
collecting the terms for dw1 and dw2 ,we get

 ∂y ∂x1 ∂y ∂x2   ∂y ∂x1 ∂y ∂x2 

dy =  + dw1 +  + dw2
 ∂x1 ∂w1 ∂x2 ∂w1   ∂x1 ∂w2 ∂x2 ∂w2 

1.6 DERIVATIVES OF IMPLICIT FUNCTIONS

Now let us discuss a type of functions called implicit functions. Suppose we
take a simple function z =f (x), for example z = 9x3. Here z is called an
explicit function of x, because the value of the variable z is explicitly
dependent on the value of x. If on the other hand, we were to write the
function as f ( z , x ) = 0 it becomes an implicit function relating z with x. so the
specific function mentioned above can be written in an implicit form as z –
9x3 =0.
We may write a general multivariate function in explicit form as
z = f (x1 , x 2 ,...x n ) . This may be implicitly written as F (z , x1 , x 2 ,..., x n ) = 0 .
Sometimes we may encounter an equation of the form f ( x1 , x 2 ,...x n ) = 0 .
You must realize that it is implicitly defining a function which, when written
explicitly may be of the form x1 = f ( x 2 , x 3 ,... x n ) .

An explicit function can always be expressed in an implicit manner by

moving the f(.) to the left-hand-side of the equation, but it is not always
possible to express an implicit function in an explicit form. When it can be so
expressed will be the topic of study in Block 3 of this course, when we
discuss the implicit-function theorem.

Instead, let us turn to the derivatives of implicit functions. To do this, let us

recall level curves . Take a function f ( x1 , x 2 ) = k . This, you would recall
was the equation of a level curve. This is also an implicit function. We can
solve this single equation in two unknowns and one of the unknowns can be
expressed in terms of the other, say, x 2 = x 2 ( x1 )

We can substitute this back into the implicit function and write
f ( x1 , x 2 ( x1 )) ≡ k

dx2
The slope of any level curve is the derivative . But this is conceptually
dx1
28 true only if in the implicit function we have defined x2 as a function of x1.
 Multivariate
Here we have done so. Calculus-I
dx2
Here our function x 2 = x 2 ( x1 ) is well defined. We can get by
dx1
differentiating the identity f ( x1 , x 2 ( x1 )) ≡ k , with respect to x1, using the
chain rule. We get
∂f dx1 ∂f dx 2 ∂k
+ ≡ ≡0
∂x1 dx1 ∂x 2 dx1 ∂x1

dx 2
,or f 1 + f 2 ≡0
dx1

If we assume x 2 ≠ 0 ,

dx 2 − f 1
≡
dx1 f2

This shows that the slope of a level curve at any point is the ratio of the first-
order partial derivatives evaluated at some particular point on the given level
curve. Let us next consider a general multivariate function.
Given the implicit function f ( x1 , x2 ,..., xn ) = 0 , the partial derivative of the jth
argument of the function with respect to the ith argument of the function xi ,
∂x j
∂xi

is obtained by first finding the total differential f1dx1 + f 2 dx2 + ... + f n dxn = 0 .

Then we divide by dxi:

dx1 dx dx dx
f1 + f 2 2 + ... + f j j + ... + f i + ... + f n n = 0
dxi dxi dxi dxi

We set all differentials other than dxi and dxj equal to zero. Then
dx j
fj + fi = 0
dxi
dx j fi
⇒ =−
dxi fj

1.7 HOMOGENEOUS AND HOMOTHETIC

FUNCTIONS
Let us consider a function F of two variables x and y, that is, F(x, y). This
function is said to be homogeneous of degree r, if for all x and y in the
domain,
F (λx, λy ) ≡ λr F ( x, y ) .

29
Functions of Several
Variables
We can extend this definition to a function of n-variables. Suppose
z = f ( x1 , x2 ,..., x n ) . Then f is said to be homogeneous of degree r if

f (λx1 , λx2 ,..., λxn ) ≡ λr f ( x1 , x2 ,..., x n ) .

F ( x, y ) is linearly homogeneous (homogeneous of degree 1) if and only if

x  x x 
F ( x, y ) = yf   , where f   = F  ,1
 y  y y 
How do we show this? Let us begin by showing the ‘if’ (sufficiency) part.

Given:

 x
F ( x, y ) = yf   ,we have
 y
λx  x   
=F (=
λ x, λ y ) λ yf=
 yf   λ F ( x, y )
 λ=
 λy   y

Now let us show the only if (necessary condition) part:

Given F(x,y) is linearly homogeneous (homogeneous of degree 1). Then

F ( λ x, λ y ) = λ F ( x, y ) , for any λ

1
Put λ = . The we have
y

x  1
F  ,1 = F ( x, y )
y  y
x 
Thus yF  ,1 = F ( x, y )
y 
x   x
But F  ,1 = f   .
y   y
x
Hence, F ( x, y ) = yf  
 y

Differentiation of a Homogeneous Function

A very important property of homogeneous functions is with regard to the
differentiation of homogeneous functions.

Let us have a homogeneous function f(x, y) of degree r. then we have:

f (λx, λy ) ≡ λr f (x, y )

Differentiating with respect to x we have:

∂f
λf x (λx, λy ) = λr f x ( x, y ) , where f x =
∂x
30
 Multivariate
If we divide by λ we get: Calculus-I

f x (λx, λy ) = λr −1 f x ( x, y )

If you notice carefully, the above equation shows that the function f x is
homogeneous of degree r – 1. The above holds for the function fy too, where
fy is the partial derivative with respect to y. the above holds in the case of any
multivariate homogeneous function.
This result says that if a function is homogeneous of degree r, then each of its
partial derivatives is homogeneous of degree r – 1.

Euler’s equation
This is a very important property displayed by homogeneous equations. Let
z f x, y

be a homogeneous function of degree r. then the following relation holds as

an identity:
∂z ∂z
x + y = rz
∂x ∂y

We can prove this using implicit differentiation and the chain rule:
Let us have the homogeneous function
f (λx, λy ) ≡ λr f (x, y ) (a)

Let us partially differentiate the left-hand side of equation (a) with respect to
λ. We get:
∂f (λx, λy ) ∂λx ∂f (λx, λy ) ∂λy
+
∂λx ∂λ ∂λy ∂λ

= xfλx + yfλy (b)

Now let us obtain the partial derivative of the right-hand side of equation (a)
with respect to l. We now get:

∂λr ∂f ( x, y )
f ( x, y ) + λr
∂λ ∂λ

rλr −1 f ( x, y ) + 0
= (c)
= rλr −1 f (x, y )

Since for equation (a), left-hand side = right hand side, therefore equation (b)
= equation (c). Hence we have
xfλx + yfλy = rλr −1 f ( x, y )

31
Functions of Several
Variables
Now λ can be any number. Let λ be 1. Then, we get Euler’s theorem:
∂z ∂z
xf x + yf y = rf ( x, y ) , i.e. x + y = rz `
∂x ∂y

The above can be generalised for functions of more than two variables.

Homothetic Functions
To understand homothetic function, recall the concept of function-of-
function, or composite function. Suppose we have y = f(x) and x = g(w), then
we can write y =f[g(w)] = h(w). We can extend the idea of composite
function to the case of multivariate function. This we indeed did in this unit,
specially when looking at partial derivatives of implicit functions, the chain
rule etc. Using this concept of composite functions in the case of multivariate
functions, let us try to explain what homothetic functions are.

If f(x,y) is a homogeneous function, then any function g[f(x,y)] is a

homothetic function provided g ′ is positive. In other words, any positive
monotonic function of a homogeneous function is a homothetic function.

Check Your Progress 3

1) Find the derivative dz/dt, given

i) Z = x2 – 8xy – y3, where x = 3t and y = 1 – t

ii) Z = 7 u + vt, where u = 2t2 and v = t + 1
…………………………………………………………………………….
…………………………………………………………………………….

…………………………………………………………………………….

…………………………………………………………………………….

…………………………………………………………………………….
2) Consider the following implicit functions. Find dy/dz for each of these
functions.

i) F(x,y) = y – 3x4 = 0
ii) F(x,y) = x2 +y2 -19 =0
…………………………………………………………………………….
…………………………………………………………………………….

…………………………………………………………………………….

…………………………………………………………………………….

…………………………………………………………………………….

32
 Multivariate
3) Determine whether the following functions are homogeneous. If so, of Calculus-I
what degree?

i) f ( x, y ) = xy

ii) f ( x, y ) = x3 − xy + y 3

xy 2
iii) f ( x, y , =
w) + 2 xw
w
…………………………………………………………………………….
…………………………………………………………………………….

…………………………………………………………………………….

…………………………………………………………………………….

…………………………………………………………………………….

1.8 LET US SUM UP

You just finished reading the first unit of this course. This unit was on the
mathematics of multivariate functions. The unit began by discussing the
concept of functions where the dependent variable is a function of more than
one independent variable. These independent variables are called arguments
of the function. You were familiarized with the idea of level curves which
are contours showing diagrammatically the third variable projected in a two-
variable plane. The unit then discussed the important concept of partial
derivatives, where the function is differentiated with respect to one variable,
with the other variables held constant. For second-order cross partial
derivatives, you learnt about Young’s theorem. Moving on from there, the
unit went on to discuss total differentials and total derivatives. We saw that
when we want to investigate what happens to the dependent variable as a
result of changes in all the independent variables we use total differentials
and total derivatives.
Following this, the unit went on to discuss the chain rule. We saw that the
chain rule in the context of partial derivatives and total derivatives is similar
to, and yet different from the chain rule in the context of single-variable
differentiation. Next, the unit discussed implicit functions. You were
familiarized with the difference between explicit and implicit functions.
Partial derivatives for implicit functions were discussed. You learnt about
the implicit function theorem. The unit then considered implicit functions in
the context of level curves. You saw how the derivative of one independent
variable with respect to the other is equal to the negative of the ratio of partial
derivatives of the dependent variable to each of the independent ones.

In the end, the unit discussed homogeneous functions. The definition of

homogeneous functions was given. Certain properties of homogeneous
33
Functions of Several
Variables
functions were also provided. You were introduced to Euler’s theorem.
Finally mention was made of homothetic functions, and their relationship to
homogeneous functions was stated

1.9 ANSWERS/HINTS TO CHECK YOUR

PROGRESS EXERCISES
Check Your Progress 1

1) (i) See section 1.2 (ii) see section 1.3 and answer.

2) (i) fx = 10x ; fy = 12y

iii) fx = -10y2 ; fy = -20x
3) See section 1.3 and answer

Check Your Progress 2

1) See section 1.4 and answer.

2) See section 1.4 and answer.

Check Your Progress 3

1) See section 1.5 and answer.
2) See section 1.6 and answer.
3) See section 1.7 and answer.

34