PARTIAL DIFFERENTIATION

AND
APPLICATIONS OF PARTIAL DIFFERENTIATION

 Introduction to topic :
Here, we are going to discuss continuity & derivatives of functions
having more than one independent variables.

 Definition of function having more than one independent variables:

If three variables x, y, z are related in such a way that z depends upon

the values of x&y, then z is called function of two variables x&y.
It is denoted z = f(x, y).
Similarly, w = g(x, y, z) is a function of three variables x, y, z.
U = f ( x1 , x 2 , x3 ,......., x n ) is a function of n variables x1 , x 2 , x3 ,......., x n .
Limit of f(x,y)

As (x,y) approaches to (a,b),f(x,y) approaches to l then we say that lim f ( x, y )  l

( x , y )  ( a ,b )

Limit of f(x,y) at (0,0)

First path-first x  0,then y  0
Second path-first y  0,then x  0
Third path-take y=mx
Fourth path-take y=mx 2
if in all paths we et same value then limit exists otherwise not exist.

 Definition of continuity of two variable functions:

A function f(x, y) is said to be continous at (a ,b) if

(i) f is defined at (a, b)

(ii) lim f(x, y) exists and

(x, y)→(a, b)

(iii) lim f(x, y) = f(a, b) in whatever manner x&y approach a & b

(x, y)→(a, b) respectively.

 Definition of partial derivatives of first order:

Let z = f(x , y) be a function of two variables .

We can make change in z by three ways.
First: keep y constant and make change in value of x
Second : keep x constant and make change in value of y
Third : make change in value of x&y both simultaneously.
If we make change in value of x , keeping y constant then the relative change in
value of z is called partial derivative of z with respect to x. and it is denoted by

(1)
 f z
or f x or . Similarly, partial derivative of z with respect to y is denoted by
x x
f z
f z  f x   lim f ( x  x, y )  f ( x, y ).
or f y or . Also, x x
y y
x  0 x
f z
 fy   lim f ( x, y  y )  f ( x, y ).
y y
y  0 y

 PARTIAL DIFFERENTIATION OF HIGHER ORDER:

z z
Let z =f(x, y) then and. are first oprder derivatives of z with respect
x y
to x &y.
Second order partial derivatives of z with respect to x & y are denoted by

 2 z   z   2 z   z 
    f ,     f yy
x 2 x  x  y 2 y  y 
xx

2z   z  2z   z 
    f yx ,     f xy .
xy x  y  yx y  x 

Third order partial derivatives of z with respect to x & y are denoted by

3z   2 z  3z   2 z 
    f ,     f yyy ,
x 3 x  x 2  y 3 y  y 2 
xxx

3z   2z  3z   z 

    f ,     f xyy

x y x  xy   y x y  yx 
2 yxx 2

3z   2z  3z   z 

 
 
2 
 f ,     f xxy .
xy x  y  yx y  x 2 
2 yyx 2

 SOME OTHER NOTATIONS TO DENOTE P.D. :

(2)
 let x  f (r , ) & y  g (r , ) (i.e. x & y botharethe functionsof r &  )
then indirectlyx is afunctionof y and viseversa.
In this case,
 x 
  denotes derivativeof x w.r.t. r keeping  cons tan t
 r 
 y 
  denotes derivativeof y w.r.t. r keeping cons tan t
 r 
 x 
  denotes derivativeof x w.r.t. keeping r cons tan t
   r
 y 
  denotes derivativeof x w.r.t.  keepingr cons tan t
   r
if we e lim inate  from above function we will get x as a functionof y & r
or y as a functionof x & r.
 x 
  denotes derivativeof x w.r.t. y keepingr cons tan t
 y  r
 x 
  denotes derivativeof x w.r.t.  keepingr cons tan t
 r  y
And similarly other derivatives can be defined.

 DERIVATIVES OF FUNCTION OF FUNCTION :

If z = f(t) and t = g(x,y) then z = fog(x,y) is a composite function of f & g

If we consider z as a function of t alone then we have ordinary derivative of z
dz
w.r.t. t which is , but if we consider z as a composite function of f &g then z
dt
will be function of two variables .in this case we can find partial derivatives of z
z z
w.r.t. x&y which are , .And relation between above derivatives is
x y
z dz t z dz t
= . & = . .
x dt x y dt y

 CHAIN RULE FOR PARTIAL DIFFERENTIATION :

(A) If u =f(x ,y) and if x = g(t) &y = h(t) then u is indirectly function of t alone
du u dx u dy
Hence,   ………..(1)
dt x dt y dt
du
is called total differential coefficient
dt
(i) If we eliminate dt from both sides we get Total Derivative of u
u u
That can be given by du  dx  dy …………..(2)
x y
(ii) If we replace t by x in above (1) then we get

(3)
 du u u dy
  ………..…...(3)
dx x y dx
In this case u is a function of x &y and y is function of x. so ultimately u
becomes function of x alone.
If u = f (x1, x2, x3, x4,….,xn) and all x1,x2,x3,x4,….xn are functions of a
single variable t.
Then,
du u dx1 u dx2 u dx3 u dxn
    .......... .......... ..  ....(4)
dt x1 dt x 2 dt x3 dt x n dt
u u u u
And du  dx1  dx2  dx3  .......... .......... ....  dxn …..(5)
x1 x2 x3 xn
(B) If z =f(x, y) and if x = g(u ,v) &y = h(u ,v) then z is indirectly function of two
variables u & v.
z z x z y
Hence,  
u x u y u
z z x z y
  …………(6)
v x v y v

 IMPLICIT FUNCTION :

If we can not express x as a function of y or y as a function x even though

we have relationship f (x, y) = 0 ,then the function f(x, y) = 0 is called an implicit
function.
for e.g. x 3  y 3  3axy  0 .

 (A) DIFFERENTIATION OF IMPLICIT FUNCTION :

To find derivative of implicit function we can use following method.

If f(x, y) = 0 is a given implicit function
then assume,
(i) u = f(x, y) and y = g(x)
u f f dy
Hence, by chain rule   but f(x, y) =0 = u
x x y dx
dy
0 fx  fy
dx
dy f
  x
dx fy
(ii) u = f(x, y) and x = h(x)
u f dx f
Hence, by chain rule   but f(x, y) =0 =u
y x dy y
dx
0  f x  fy
dy

(4)
 dx fy
 
dy fx

 (B) PARTIAL DIFFERENTIATION OF IMPLICIT FUNCTION :

f (x, y, z) = 0 represents an implicit function .

To find partial derivative of implicit function we can use following method.
If f(x, y, z) = 0 is a given implicit function
then assume,

(i) u = f(x, y, z) and z = g(x, y)

Hence, by chain rule

u f f z u f f z
  And  
x x z x y y z y
But f(x, y, z) = 0 = u
z z
0 fx  fz . And 0 f y  fz
dx dy
z f z fy
  x   .
dx fz dy fz

(ii) u = f(x, y, z) and y = h(x, z)

Hence, by chain rule

u f f y u f f y
  And  
x x y x z z y z
But f(x, y, z) =0 = u
y y
0 fx  fy And 0 fz  f y
dx dz
y fx y f
  .   z .
dx fy dz fy

(iii) u = f(x, y, z) and x = t (y, z)

Hence, by chain rule

u f f x u f f x
  And  
y y x y z z x z
But f(x, y, z) =0 = u
x x
0 f y  fx 0 fz  fx
dy dz
x fy x f
  .   z
dy fx dz fx

(5)
 .

 HOMOGENEOUS FUNCTIONS:

A function f of two variables x & y is said to be homogeneous function of

degree n if,

 y  x
f tx, ty   t n f ( x, y) or f ( x, y)  x n g   or f ( x, y)  y n h   .
 x  y
2 2
x y
For e.g. u  . is a homogeneous function of degree 3.
x y
2
y 
x  
4

u   x   x 3 g y  or u tx, ty   t
x2 y2
 
3
Since .
  y  x x y
x1    
  x 

 EULER’S THEOREM :

Statement :
If u is a hom ogeneous function of x & y of dergreen then
u u
x  y  nu.
x y
Proof:
Since u = f(x , y) is a homogeneous function of degree n, it can be
Written as

(6)
  y
u ( x, y )  x n g   .......... ...(i )
x

differentiating (i ) partially w.r.t x differentiating (i ) partially w.r.t y

u  y  y   y  u  y  1 
 nxn 1 g    x n g    2   x n g    
x  x  x  x  y  x  x 
 y  y  y
 nxn 1 g    x n  2 y g     x n 1 g   
 x  x  x
hence, hence,
u  y  y u  y
x  nxn g    x n 1 y g   ........( ii) y  x n 1 y g   .........( iii)
x x  x y x

Now, adding (ii) & (iii) we get,

u u  y
x y  nx n g  
x y x
 n u.

 Cor-1: If u is a homogeneous function of x, y of degree n, then

 2u  2u 2  u
2
x2  2 xy  y  n (n  1) u.
x 2 xy y 2

 Cor-2: If z is a homogeneous function of x, y of degree n and z = f(u) ,

u u f (u )
Then, x  y  n .
x y f (u )

 Cor-3: If z is a homogeneous function of x, y of degree n and then

(7)
  2u  2u 2  u
2
x2
 2 xy y  g (u ) [ g (u )  1].
x 2 xy y 2
f (u )
where g (u )  n .
f (u )

 Jacobians:

(1) If u = f(x, y) and v = g(x, y) be differentiable functions then the

Jacobian of u & v with respect to x & y is defined and denoted by
u u
u, v  u, v  x y
is givenby J   .
  x, y  x, y  v v
x x
(2) If u = f(x, y, z) ,v = g(x, y, z) & w = h(x, y, z) be differentiable functions
then the Jacobian of u, v& w with respect to x, y & z is defined and
denoted by
u u u
x y z
u, v, w u, v, w v v v
is givenby J   .
  x, y , z  x, y, z  x y z
w w w
x y z
 Chain rule for Jacobians:

(1) If u = f(x, y) & v = g (x, y) and x = h (r, s) & y = t (r, s) be differentiable

functions then
 u , v   u, v    x, y 
 . .
 r , s   x, y   r , s 
 u , v, w  x, y, z   u , v.w
(2) .  .
 x, y, z   r , s, t   r , s, t 
 Note:
 u , v    x, y 
(1) If J = and J   then J.J’ = 1.
  x, y   u , v 

(8)
  u , v, w   x, y , z 
(2) If J = and J   then J.J’ = 1.
  x, y , z   u, v, w

 POWER SERIES EXPANSIONS :

 Maclaurin’s Expansions for functions of two variables:

Maclaurin series Expansions for the function f(x, y) can be given by

f(x, y) = f(0, 0) + [ x f x (0, 0)  y f y (0, 0)]


1 2
2!

x f xx (0, 0)  2xy f xy (0,0)  y 2 f yy (0,0)  …. (i)

 .......... .......... .......

 Taylor’s Expansions for functions of two variables:

Taylor series Expansions for the function f(x, y) can be given by

f(x, y) = f(a, b) + [(x - a) f x (a, b)  (y - b) f y (a, b)]


1
2!

(x - a) 2 f xx (a, b)  2(x - a) (y - b)f xy (a, b)  (y - b) 2 f yy (a, b)  …(ii)

 .......... .......... .......

 Cor-1:
Putting x = a + h & y = b+ k in above equation (ii) we get,

f(a  h, b k) = f(a, b) + [ h f x (a , b)  k f y (a, b)]


2!

1 2
h f xx (a, b)  2hk f xy (a, b)  k 2 f yy (a, b)  …(iii)

 .......... .......... .......

 Cor-2:
Putting a= x & b= y in above equation (iii) we get,

f(x  h, y  k) = f(x, y) + [ x f x (x , y )  y f y (x, y)]

1 2
2!
 
x f xx (x, y )  2xy f xy (x, y)  y 2 f yy (x, y) ……..(iv) 
 .......... .......... ......
MAXIMA AND MINIMA OF FUNCTIONS OF TWO VARIABLES :

Working rules to find maximum and minimum values of a function f(x , y):

f f
(1) Find and .
x y

(9)
 f f
(2) Solve simultaneous equations  0 and 0
x y
f f
 0 and  0 is necessary condition for function f(x ,y) to be
x y
maximum or minimum.
(3) let (a1 ,b1 ) ,(a2 ,b2 ),…….be solutions of simultaneous equations
f f
 0 and  0 . These points are called optimal points.
x y
2 f 2 f 2 f
(4) Evaluate r  2 , s  , t  2 at above points.
x xy y
(5) Use following table to decide that function is either maxima or minima
at optimal points.

Sr r t - s2 r conclusion
no.
1 >0 <0 Point of maxima
2 >0 >0 Point of minima
3 >0 =0 Further investigation required
4 =0 any Further investigation required
5 <0 any Point of neither Maxima nor Minima Saddle point

(6) points :

Sr Points Definition
no.
1 Optimal point f f
A point where function satisfies  0 and 0
x y
2 Extreme point A point where function takes its Maximum or
Minimum value.
3 Point of maxima A point where function takes its Maximum value.
4 Point of minima A point where function takes its Minimum value.
5 Saddle point A point where function is neither Maxima nor
Minima

(7) values :

Sr Points Definition
no.
1 Optimal value Value of function at optimal point
2 Maximum value Value of function at Point of maxima
3 Minimum value Value of function at Point of minima

 LAGRANGE’S METHOD OF UNDETERMIND MULTIPLIERS :

This method is useful to find extreme values of a function

u = f(x ,y, z)
subject to the condition  (x, y, z) = 0

(10)
 Working rules to find maximum and minimum values of a function
u = f(x ,y, z)
Subject to the condition  (x, y, z) = 0

(1) Form the Lagrangian equation

u = f (x, y, z) +   (x, y, z).
u u u
(2) Solve equations 0 , 0 &  0 along with  (x, y, z) = 0
x y z
and find the values of x, y, z &  .
The values of x, y, z so obtained will give the extreme values of
f (x, y, z).

 TANGENT PLANE TO THE SURFACE:

Let the equation of the surface be f (x, y, z) = 0.

Then the equation of the tangent plane to the surface f (x, y, z) = 0 at the point

 f   f   f 
P(x1, y1, z1 ) is x  x1     y  y1     z  z1    0.
 x  P  y  P  z  P

 NORMAL LINE TO THE SURFACE:

Let the equation of the surface be f (x, y, z) = 0.

Then the equation of the normal line to the surface f (x, y, z) = 0 at the point

P(x1, y1, z1 ) is
x  x1  
 y  y1  
z  z1  .
 f   f   f 
     
 x  P  y   z  P

 Problems Based On Above Discussion :

(1) Find the equations of the tangent plane and the normal line to the surface
(I) z  3x 2  2 y 2 at (1, 2, 11).

(11)
 (II) x 2  2 y 2   3z 2  12 at (1, 2, -1).
(2) show that the surfaces z  xy  2 and x 2  y 2  z 2  3 have the same
tangent plane at (1, 1, -1).
(3) Show that the tangent plane to the surface x 2  y ( x  z ) at any point
passes through origin.
(4) Show that the plane lx +my +nz = p touches the ellipsoid
x2 y 2 z 2
2
 2  2 1, if l 2 a 2  m 2b 2  n 2 a 2  p 2 .
a b c

 ERRORS AND APPROXIMATIONS :

 ERRORS:

(1)  x is an error in x.
(2)  x is called an absoluteerror in x.
x
(3) is called the relative error in x.
x
x
(4)  100 is called the percentage error in x.
x

 Let f(x, y) be a continuous function of x and y .

Let  x &  y be the change in x & y respectively then related change in

f f
f(x, y) is given by f  x  y .approximately.
x y

(12)