PARTIAL DIFFERENTIATION

OBJECTIVES
Atthe end of this lesson students should be able to
 understand the concept of partial differentiation.
 differentiate a function partially with respect to each of its variables in turn.
 apply Euler’s theorem on homogeneous functions.
 find the partial derivatives of implicit and composite functions.
 apply the concept of partial differentiation in calculating the errors and approximations

1. INTRODUCTION
In many applications in science and engineering, a function of interest depends on
multiplevariables. For instance, the ideal gas law p = RT states that the pressure p is a
function ofboth its density, and its temperature, T. (The gas constant R is a material property
and nota variable). Consider volume of a container in the shape of a right circular cone:
1
V = πr 2h. Here the volume V depends on the two quantities r and h, representing the base
3
radius and altitude of the cone respectively. These are simple example of a function of more
than one variable.
The process of taking the derivative, with respect to a single variable, and holding constant all
of the other independent variables, is called finding (or, taking) a partialderivative.Partial
derivatives are distinguished from ordinary derivatives by using a instead of a d.

2. FUNCTION OF TWO VARIABLES

Let D be a region in the xy – plane. Suppose that f is a function that associates every (x, y)D,
with a unique real number z. We write z = f(x, y) and we call z, a function two variables x and
y. The variables x and y are called independent variables and z is called the dependent variable.

3. NEIGHBOURHOOD OF A POINT (a,b)

Let δ be any positive number. The points (x,y) such that a  δ  x  a+δ, b  δ  y  b+δ
determine a square bounded by the lines x = a  δ, x = a+δ; y = b  δ, y = b+δ. Its centre is at the
point (a,b). This square is called a neighbourhood of the point (a,b). Thus the set
 x, y : a  δ  x  a+δ, b  δ  y  b+δ  is a neighbourhood of the point (a,b).
4. CONTINUITY OF A FUNCTION OF TWO VARIABLES.
We shall briefly introduce the concept of a continuous function in two variables. A real valued
function w = f(x, y) defined on a region D of the xy-plane is said to be continuous at a point
(x0, y0) ∈ D if lim f  x,y  = f  x 0 ,y0  .
 x,y  x0 ,y0 

If f is continuous at every point in D, it is said to be continuous on D.

 xy
 , if  x,y    0,0 
Example 1. Let f  x,y  =  x 2 +y 2
0, if  x,y  =  0,0 

Consider the polar coordinates of the point (x, y), i.e. x = r cos θ and y = r sin θ. Then
xy
 rcosθsinθ . Since r → 0 as (x, y) → (0, 0) and |sin θ cos θ| ≤ 1 for all θ, we have
x 2 +y 2

xy
lim = 0  f  0,0  . Therefore, the function is continuous at the origin.
 x,y  0,0  x 2 +y 2

 xy
 2 2 , if  x,y    0,0 
Example 2.Let f  x,y  =  x +y
0, if  x,y  =  0,0 

Suppose that (x, y) approaches (0, 0) along the line y = x. Then
1
lim f  x,y  = lim f  x,x  
 x,y  0,0 x 0 2
y=x

Similarly, if (x, y) approaches (0, 0) along the line y = −x, we have

1
lim f  x,y  = lim f  x,-x   
 x,y  0,0 x 0 2
y = x

Therefore, the limit does not exist and the function is not continuous at the origin.
Remark: In order for the limit in lim f  x,y  = f  x 0 ,y0  to exist, f(x, y) must approach
 x,y  x 0 ,y0 

f(x0, y0) for each and every path of approach of (x, y)to (x0, y0).
5. PARTIAL DERIVATIVES
Let z = f (x,y) be a function of two variables x and y.Then
 lim f  a+h, b   f  a, b  ,
h 0 h
if it exists, is said to be the first order partial derivative of f w.r.t x at (a,b) and is denoted by
 z 
 x  or f x  a, b  .
  a, b 
 

Similarly,
lim f  a, b + k   f  a, b 
,
k 0 k
if it exists, is said to be the first order partial derivative of f w.r.t y at (a,b) and is denoted by
 z 
  or f y  a,b  .
y
  a,b 
 
 

If the partial derivatives of f exist at each point in the domain of definition of f, then theyare
given by
z f(x+x, y)  f(x, y)
 lim  fx
x x  0 x
z f(x, y+y)  f(x, y)
 lim  fy
y y  0 y
Example 3: Let z = f(x, y) = x3+ y3- 3xy. Then
z z
 3x 2  3 y,  3 y 2  3x.
x y
We see that fxand fy are functions of x and y again and hence may be differentiable wr.t x and
w.r.t y. If the partial derivatives exist, they are called the second order partial derivatives ad are
given as follows:
  z   2 z   z   2 z
   f xx ,    f xy ,
x  x  x 2 y  x  yx
  z   2 z   z   2 z
   f yx ,    f yy .
x  y  xy y  y  y 2

In a similar way one can define third and higher order partial derivatives.
Note:The two second partial derivatives fxy and fyx above, the ones with one partial derivative
with respect to x and one with respect to y, are called mixed partial derivatives.If the partial
order derivatives are continuous then the mixed partial derivatives are equal i.e., the order in
which we differentiate f is immaterial. For instance
 fxy= fyx; fxxy= fxyx= fyxx, fxyy= fyyx= fyxy.
Remark:Just because the order of partial differentiation doesn't (typically) matter as far as the
final resulting higher-order partial derivative is concerned, that doesn't mean that calculating
the partial derivatives in different orders is equally easy. For example, consider
e x cos  x  tan 1 x 
f ( x, y )  xe 
5y

1  log x
If you want to calculate the second partial derivative of f, once with respect to xand once with
respect to y, it would be a painful waste of time to calculate f/x first. If this isn't obvious to
you, you should think about it until it's clear.
What you want to do is calculate the partial derivative with respect to y first, since, then, the
entire right-hand ugly expression will disappear. Hence, we find that
2 f   f  
     5 xe5 y   5e5 y .
yx x  y  x

6. GEOMETRICAL REPRESENTATION OF PARTIAL DERIVATIVES OF THE

FIRST ORDER.
The derivative of a one-variable function can be interpreted graphically as the slope of the
tangent line. Is there also a way to interpret the partial derivatives graphically? Yes.

Geometrically the function z = f(x,y) represents a surface. Thenfx(a,b) denotes the tangent of
the angle which the tangent to the curve , in which the plane y = b parallel to the ZX plane cuts
the surface at the point P(a,b, f(a,b)), makes with X-axis.
Similarly, fy(a, b) denotes the tangent of the angle which the tangent to the curve, in which the
plane x = a parallel to the ZY plane cuts the surface at the point P(a,b, f(a,b)), makes with Y-
axis.
Example 4:Find z/x and z /y if xy + yz + zx = 1.

Solution: xy + yz + zx – 1= 0 (1)
Differentiating (1) partially w.r.to x, we get
z z
y+y  z  x 0
x x
z  yz
  
x  x y

Differentiating (1) partially w.r.to y, we get

z z
x+y  z  x 0
y y

z  xz 
  
y  x y
 2u  2u x
Example 5 : Verify  where u  sin 1
xy yx y
u x u
2
y
Solution :  , 
y xy 3
y y2  x2
 y 2
 x 
2 2

u 1  2u y
 , 
x yx 3
y  x2
y  x2  2
2
2

 2u  2u
 
xy yx

z z
Example 6 : If z = eax+by f  ax  by  prove that b  a  2abz.
x y
z
Solution :  aeax+by f  ax  by   aeax+by f   ax  by 
x
z
 beax+by f  ax  by   beax+by f   ax  by 
y
z z
 b  a  2abz.
x y
  2u  2u 1
Example 7 : If u = f  r  where r = x + y prove that 2  2  f   r   f   r 
2 2 2

x y r
u u r x
Solution :    f r 
x r x r
u 1
2
x f   r  x 2 f   r 
2

 f   r   
x 2 r r3 r2
 2u 1 y 2 f   r  y 2 f   r 
similarly,  f r   
y 2 r r3 r2
 2u  2u 1
  2  f   r   f   r 
x y
2
r

Example 8: If
 y x
u  x 2 tan 1    y 2 tan 1   ; xy  0
x  y

prove that
 2u x 2  y 2
 .
xy x 2  y 2

Solution:
u 1 1 x 1 x
 x2  2 y tan 1    y 2
y 2
y x  y
2
x y2
1 2 1 2
x y
x3 xy 2 x
=   2 y tan 1  
x y
2 2
x y
2 2
 y
x
= x  2 y tan 1   .
 y
 2u   u 
  
xy x  y 
1 1 x2  y 2
= 1  2y  .
x 2 y x2  y 2
1+ 2
y

PROBLEMS:

z z  x2  y 2 
1. Find and if z  tan 1  
x y  x y 
 1
u u
2. If u = 1-2xy+y 
2 -2
, prove that x  y  y 2u 3
x y
y z x u u u
3. If u = + + show that x  y  z  0
z x y x y z

 2u  2u  x2  y 2 
4. Verify  where u  log  
xy yx  xy 
2
 z z   z z 
5. If z  x + y  = x + y show that     4 1  
2 2

 x y   x y 

7. HOMOGENEOUS FUNCTIONS:

A function z = f(x, y) is said to be a homogeneous function of degree n in x and y if we can

write

 y x
z  x n    y n   .
x  y
Example 1:
x3  y3
Let f(x, y) = . Then .
x y

  y 3 
x 1    
3

x y
3 3
  x    y
f ( x, y )     x 2  
x y   y  x
x 1    
  x 
Thus f(x, y) is a homogeneous function of degree 2.

Example 2:
 y
Let f(x, y) = x 3  y 3 log y  y 3 log x  x 2 y sin   . Then
x
  y  y
f ( x, y )  x 3  y 3 log    x 2 y sin  
x x
  y 3  y y  y 
= x 3 1    log    sin   
  x  x x  x  
 y
= x 3   .
x

8. EULER’S THEOREM ON HOMOGENEOUS FUNCTIONS:

Statement:If z = f(x, y) is a homogeneous function of degree n in x and y then

z z
x  y  nz,  x, y in the domain of of f.
x y

Proof: Since z is a homogeneous function of degree n in x and y, we can express it in the form
 y
z  x n  
 x .

z  y   1   y
 x n '    y  2   nx n1  
x x x  x
z  y 1
 x n '      .
y  x x
Hence,
z z  y
x  y  nx n    x n1   y ' y '  nz .
x y x
Corollary: If z is a homogeneous function of degree n, then
2 z 2 z 2 z
2
x2  2 xy  y  n(n  1) z.
x 2 xy y 2
Proof:

Since z is a homogeneous function of degree n, we have

z z
x  y  nz ...........................(1)
x y

Differentiate equation (1) partially with respect to x, we get

 2 z z 2 z z
x   y n .
x 2
x xy x
Multiplying by x, we get
2 z 2 z z
x2  xy  (n  1) x .................(2)
x 2
xy x

Differentiate equation (1) partially with respect to y, we get

2 z  2 z z z
x y 2 n
xy y y y

Multiplying by y, we get
2 z 2 z z
xy  y 2 2  (n  1) y .......................(3)
xy y y

Adding equation (1) and (2), we get

2 z 2 z 2  z
2
 z z 
x 2
 2 xy y  (n  1)  x  y   n(n  1) z.
x 2
xy y 2
 x y 

Example 1:Verify Euler’s theorem for

x  y
z  sin 1    tan 1   .
 y x

Solution:
   
 1  1  1  y    y
z  x sin
0
 tan    x 0   .
  y  x  x
  x 
Hence z is homogeneous function of degree n = 0.
z 1 1 1 y
 .  . 2
x 2
 x  y 1  y  x
2

1    
 y x

1 y
  .
y x
2 2 x  y2
2

Similarly,
z 1  x  x
    2
y y 2  x2  y  x  y
2
 z z
x  y  0  nz.
x y
Hence verified.

Example 2:If
 x y 
u  sin 1  
 x  y 

show that
z z 1
x  y  tan u.
x y 2

Solution:
Let z  sin u. Then
 y
x 1  
 x y   x   x 12  y  .
z   
 x  y   y x
x 1  
 x

Thus z is a homogeneous function of degree ½. Hence by Euler’s theorem

z z 1
x  y  n z  sin u.
x y 2
 (sin u )  (sin u ) 1
x y  sin u.
x y 2
u u 1
 cos u x  cos u y  sin u.
x y 2
u u 1
x  y  tan u.
x y 2

Example 3:If
1
 x y 2
u  cos ec 1  3 
 x  3 y 

prove that
  2u  2u 2 u
2
tan u  13 tan 2 u 
x2  2 xy  y    .
x 2 xy y 2 12  12 12 

Solution: Let

z z 1
x y  z
x y 12
u u 1
 x cos ecu cot u  y cos ecu cot u  cos ecu
x y 12
u u 1 cos ecu 1
x y    tan u ................(1)
x y 12 cos ecu cot u 12

Differentiate equation(1) partially with respect to x, We get

 2u  2u  1 2  u
x  y   1  sec u
x 2 xy  12  x
Multiplying by x, we get

 2u  2u  1  u
x 2
 xy   1  sec2 u   x ...................(2)
x 2
xy  12  x
Differentiate equation (1) partially with respect to y, We get

 2u  2u  1  u
x  y 2   1  sec2 u  .
xy y  12  y
Multiplying by y, we get

 2u 2 u
2
 1  u
xy y   1  sec2 u  . y ...................(3)
xy y 
2
12  y
Adding equation (2) and (3), we get

 2u  2u 2 u  2   u u 
2
1
x 2
 2 xy  y  1  sec u   x
x  y 
x 2 xy y 2  12  y 


  1 
 12
1

1  tan 2 u  
1 
 12 tan u 

 1 13 tan 2 u 
 tan u    
12 12 12 

Exercise :

1. Verify Euler’s theorem for the following functions:

(i) z = ax2 + 2hxy + by2 (ii) z = (x2 + xy + y2)-1.
x2  y 2 u u
2. If u  log then show that x  y  1.
x y x y
 x3  y 3  u u
3. If u  tan 1   then show that x  y  sin 2u.
 x y  x y

 x3  y 3  2 u  2u 2 u
 
2 2
4. If u  tan 1   then show that x  2 xy  y  1  4sin 2 u sin 2u.
 x y  x xy y
2 2

x  y 
3 3
2 u
2
u2
u 1
2
5. If u  tan 1   then show that x  2 xy  y 2 2   tan3 u  tan u  .
 x y  x xy y
2
4

9. TOTAL DERIVATIVES:

Let z = f(x, y).Then the total differential dz is defined as

z z
dz  dx  dy.
x y
10. COMPOSITE FUNCTIONS:

A function of a function is called a composite function.

Let z = f(x, y) where x = (t) and y = (t).

Now, we can express z as s function of t alone by substituting the values of x and y in f(x, y).

Then z is a composite function of t.

Thus the ordinary derivative dz/dt which is called the total derivative of and is given by,
dz z dx z dy
 . 
dt x dt y dt
Total partial derivative theorem for composite function:

If z = f(u, v) where u = (x, y) and v = (x, y), then the partial derivatives of z are given by
z z u z v
   
x u x v x
z z u z v
   
y u y v y

Example 1:
dz
Find given that z  xy 2  x 2 y, x  at 2 , y  2at.
dt
Solution:
dz z dx z dy
   
dt x dt y dt

   
 y 2  2 xy  2at   2 xy  x 2  2a 

 8a3t 3  8a3t 4  8a3t 3  2a3t 4

 16a3t 3  10a3t 4
Example 2:
z z
If z = x 2  y 2 , x  cos uv, y  sin(u  v), find and interms of u and v.
u v
Solution:
z z x z y
   
u x u y u
 2 x  v sin uv  2 y cos(u  v)
 2v sin uv cos uv  2sin(u  v)cos(u  v)
 v sin 2uv  sin 2(u  v).
Similarly,
z z x z y
   
v x v y v
 u sin 2uv  sin 2(u  v).
Example 3:
z z z
If z = f(x, y), x  eu sin v, y  eu cos v prove that x
u v

 y  x2  y 2x
.

Solution:
z z x z y
   
u x u y u
z u z
 e sin v   eu cos v 
x   y  

Similarly,
z z u z
 e cos v    eu cos v 
v x y

z z  z z   z z 
x  y  x x  y   y  y  x 
u v  x y   x y 


 x2  y 2  xz .
Example 4:
H H H
If H  f  y  z, z  x, x  y  , prove that    0.
x y z
Solution:
Let H = f(u, v, w) where u  y  z, v  z  x and w  x  y.
H H u H v H w
      .
x u x v x w x
H H H
  0    1  1.
u v w
Similarly,
H H H H
 1   0    1.
y u v w
H H H H
  1  1   0 .
z u v w

Hence
 H H H
   0.
x y z

Example 5:
2
 z   z   z 
2 2 2
1  z 
If z  f ( x, y ) and x  r cos , y  r sin  , prove that          2   .
 x   y   r  r   

Solution:
z z x z y
    .
r x r y r
z z
  cos   sin  .
x y

Similarly,
z z z
  r sin     r cos  .
 x y

Hence,
2 2 2 2
 z  1  z   z   2 1 2 2   z   2 1 2 2 
   2      cos   2 r sin      sin   2 r cos  
 r  r     x   r   y   r 
2 2
 z   z 
   
 x   y 

Exercise

u v u v z z z z
1. If z  f ( x, y ) and x  e  e , y  e  e , prove that  x y .
u v x y
2 2 2 2
 r  r    
2. If x  r cos  , y  r sin  , find (i )  (ii )  .
2 2 2 2
x y x y
z z
3. If z  eu 2v and u  sin x, y  x 2  y 2 , find and .
x y
11. IMPLICIT DIFFERENTIATION:
When f ( x, y ) is a function of two variables x and y , the equation f ( x, y)  c (where c is a

constant) enables us to obtain values of y corresponding to values of x . Then we say that y

is an implicit function of x . We assume, for simplicity, that the above equation always defines
a unique value of y for each value of x .

Since f is a function of x and y , and y is again a function of x , we can consider f a

composite function of x . Then, its total derivative with respect to x is

df f dx f dy f f dy
    .
dx x dx y dx x y dx
But since f ( x, y)  c , the total derivative of f must be identically 0 . Thus

f f dy dy f / x f
 0    x if f y  0 .
x y dx dx y / y fy

Differentiating again with respect to x , considering f / x and f / y as composite

functions of x , we get
  2 f  2 f dy  f f   2 f  2 f dy 
 2      
d2y  x yx dx  y x  xy y 2 dx 

dx 2  f 
2

 y 
 
2
 2 f  f 
2
 2 f f f  2 f  f 
   
x 2  y 
2
yx x y y 2  x 
 3
 f 
 y 
 
Thus,

f xx  f y   2 f yx f x f y  f yy  f x 
2 2
dy f d2y
 x and  
dx fy dx 2 ( f y )3

dy
Example 1:If x  y , find
y x
.
dx
Solution:
xy  yx
Taking log on both sides, we get
y logx = x logy
f(x, y) = y logx - x logy = 0
y
 log y y y  x log y

dy fx x
  
 .
dx f y log x  x x  x  y log x 
y
2 2 2
Example 2:Prove that if y 3  3ax2  x3  0 , then d y  2a x  0 .
dx 2 y5
Solution:Let

f ( x, y)  y3  3ax 2  x3  0.
We have
f x  6ax  3x 2 , f y  3 y 2 .

f xx  6a  6 x, f xy  0, f yy  6 y
Therefore,
d2y 6( x  a)9 y 4  (3x 2  6ax)2 6 y a2 x2
   2
dx 2 27 y 6 y5
EXERCISE

1. If ax  2hxy  by  2 gx  2 fy  c  0 , prove that

2 2

d 2 y abc  2 fgh  af 2  bg 2  ch 2

dx 2 (bx  by  f )3

2. If A , B , C are the angles of a triangle such that sin A  sin B  sin C is

2 2 2

constant, show that

dA tan B  tan C

dB tan C  tan A

12. ERRORS AND APPROXIMATIONS

Let f ( x, y ) be a continuous function of x and y . If  x and  y are the increments of x
and y , then the new value of f ( x, y ) will be f ( x   x, y   y) . Hence
  f  f ( x   x, y   y)  f ( x, y) .
Expanding f ( x   x, y   y) by Taylor's theorem and supposing  x ,  y to be small
enough that their products, squares, and higher powers can be neglected, we get
f f
f   x   y , approximately.
x y

The value  f is called the error in f due to the errors  x and  y in x and y.  f is called
f

the relative error in f and  f  100 is called the percentage error in f .

f

Example 1:If PV 2  K and if the relative errors in P is 0.05 and in V is 0.025 then
show that error in K is 10% .
Solution:
P V
 0.05 and  0.025
P V
PV 2  K

Taking log on both sides of the equation,

log P  2logV  log K
  log P   2  logV    (log K )
1 1 1 K
 P  2  V   K  0.05  2(0.025) 
P V K K
K
 0.1
K
Thus, error is
K
100   (0.1)  100  10%
K
Example 2:The time T of a complete oscillation of a simple pendulum is given by the formula

T  2 l / g . If g is a constant find the error in the calculated value of T due to an error of

3% in the value of l .
Solution:

T  2 l / g
Taking log on both sides,
 1 l
log T  log 2  log
2 g

1 
 (log T )   (log 2 )    (log l  log g ) 
2 
T 1 l
0 0
T 2 l
T 1 l 1
100   100   3  1.5
T 2 l 2
Thus, the error in T is 1.5%
Example 3:If the sides and angles of a plane triangle ABC vary in such a way that its
a b  c , where
circum-radius remains constant, prove that    a ,  b , and  c
cos A cos B cos C
denote small increments in the sides a, b and c respectively.
Solution:
Let R be the circum-radius of the ABC , then
a b c
R  
2sin A 2sin B 2sin C
 2R sin A, b  2R sin B, c  2R sin C
Differentiating, we get
 a  2R cos A  A,  b  2R cos B  B,  c  2R cos C  C
a b c
    2 R( A   B   C )
cos A cos B cos C
 2R ( A  B  C )  2R ( )  0.