Functions of Several Variables

CHAPTER THREE

FUNCTIONS OF SEVERAL VARIABLES

3.1 Definitions and properties

Definitions:1) A rule that assigns to each element of a set D   n , n  2 a unique real number
is called real valued function of several variables.

n
2) The largest possible set D in  on which f is defined is called the domain of f.

Notation:

 f : D   2   ( real valued function of two variables), denoted by f  x , y .

 f : D   3   ( real valued function of three variables), denoted by f x , y , z .

Example: Find the domain of the following functions of several variables.

a . Let f : D   2   , defined by f  x , y   x 2  y 2 .
b. Let f : D   3   , defined by f  x , y , z   9  x 2  y 2 .
y
c. Let f : D   2   , defined by f  x , y  
x  y2

Solution: a. D   x , y  : x , y     2 (the whole xy-plane)

  
b. D   x , y , z  : 9  x 2  y 2  0   x , y , z  : x 2  y 2  3 2 
(The set of all points inside (and on) a circle of center at the

origin and radius 3)

c. exercise.

Graph and level curves of a function of several variables.

 
The graph of z  f  x , y  is the set  x , y, z  : z  f  x , y , ( x , y )  D f  in  3
. So we see
3
that this graph is a surface in  .
 If we put z  c , c   the set x , y , c  : ( x , y ) D f and c  f  x , y  is called the level
curve for the graph of f.
 The intersection of the horizontal plane z  c , c   with the surface z  f  x , y  is
called the contour curve (trace) of height c on the surface.

Compiled by: Amsalu W/Gebreal Page 1

Functions of Several Variables

 The vertical projection of the contour curve in to the xy-plane is the level curve
f  x , y   c of the function f.
 To get the graph of f, raise the level curves to the plane z  c .

Exercise: Sketch the graph of the following functions of several variables.

a . f x , y   x 2  y 2 b. f  x , y   4  x 2  y 2 c. f  x , y   4 x 2  y 2

3.2 Limit and continuity

Definition: Let f  x , y  be defined on a circular region centered

a at  x 0 , y 0  except possibly at  x 0 , y 0  . Then

lim f  x , y   L   if and only if   0 ,   0 

 x y  x 0, y0 

0 x  x0 2   y  y0 2    f x , y   L   .

Example: 1. Show that lim (3x  2 y)  5 .

 x y 1,1

Solution: Let   0 be given.

Then we need to find   0  0   x  x 0 2   y  y 0 2    f x , y   L   .

Now , f  x , y   L  3x  2 y  5  3 x  3  2 y  2  3( x  1)  2( y  1)  3 x  1  2 y  1

 x 1  x  12  x  12   y  12 & y 1  x  12   y  12 

 

Thus, f x , y   L  3 x  1  2 y  1  3 x  12   y  12 2 x  12   y  12

5 x  12   y  12 
 x  12   y  12   5.

Choose    5 .  lim (3x  2 y)  5

 x y 1,1

 xy 2 
Example: 2. Show that lim  2   0 .
 x y 0 , 0  x  y 2
 

xy 2
Solution: Given   0 . Then we need to find   0  0  x2  y2     .
x2  y2

Compiled by: Amsalu W/Gebreal Page 2

Functions of Several Variables

2
xy 2 xy
Now , 2
x y 2
 2
x  y2
….. x  x2  x2  y2 & y2  x2  y2 

Thus ,
xy 2


x2  y 2 x2  y 2  x2  y2  .
x2  y2 x y2 2

 xy 2 
Choose    .  lim  2   0
 x y  0, 0  x  y 2
 

Note: lim f x , y  fails to exist (doesn’t exist) if f x , y  approaches different values as

 x y  x 0, y0 
x , y  approaches x0 , y0  from different directions.
 5 xy 
Example: Show that lim  2  doesn’t exist.
 x y  0 , 0  x  y 2
 

Solution: i) approach along the x-axis, i.e. x  0 , y  0 .

 5 xy   0 
Then, lim  2   lim  2   0.
 x y 0 , 0  x  y 2
   x y 0, 0  x 

ii) approach along the y-axis, i.e. y  0 , x  0 .

 5 xy   0 
Then, lim  2   lim  2   0.
 x y  0 , 0  x  y 2
   x y 0, 0  y 

iii) approach along the line y  x , i.e. x  0 , y  0 .

 5 xy   5x 2  5
Then, lim  2   lim  2    0.
 x y 0 , 0  x  y 2
   x y 0, 0  2 x  2

iv) approach along the line y  mx , m  0 .

 5 xy   5m  5m
Then, lim  2   lim    , which varies as m varies.
 x y 0 , 0  x  y 2 2 2
   x y 0, 0  1  m  1  m

 5 xy 
Therefore, lim  2  doesn’t exist.
 x y  0 , 0  x  y 2
 

Compiled by: Amsalu W/Gebreal Page 3

Functions of Several Variables

 3x 2 y   x2 y2 
Exercises: 1. Show that a) lim  2   0. b) lim  4  doesn’t exist.
 x y 0 , 0  x  y 2   x y  0 , 0  x  y 4
   

2. Evaluate: lim x
 x y  0, 0 
2
  
 y 2 ln x 2  y 2 .

Continuity

Definition: 1. A function f  x , y  is said to be continuous at x 0 , y 0  if f  x 0 , y 0  is defined and

lim f  x , y    x 0 , y0 .
 x y  x 0, y0 

2. A function f  x , y , z  is said to be continuous at  x 0 , y 0 , z 0  if f  x 0 , y 0 , z 0  is defined and

lim f  x , y , z    x 0 , y0 , z 0 .
x y , z  x 0, y0 , z 0 

3.3 Partial derivatives

Definition: 1. Let f  x , y  be a function of two variables and  x 0 , y 0   D f . Then the first order
partial derivatives of f with respect to x and with respect to y respectively are defined as:

f f x0  h , y 0   f x0 , y 0 
I)
x
x0 , y0   f x x0 , y 0   lim
h0 h
.

f f x 0 , y 0  h   f x0 , y 0 
II) x0 , y0   f y x0 , y 0   lim .
y h0 h

Note: The symbol  is pronounced “del”

2. Let f  x , y , z  be a function of three variables and x 0 , y 0 , z 0   D f . Then the first order

partial derivatives of f with respect to x , with respect to y, and with respect to z respectively
are define as:

f f x 0  h , y 0 , z 0   f x 0 , y 0 , z 0 
I)
x
x0 , y 0 , z 0   f x x0 , y 0 , z 0   lim
h 0 h
.

f f x 0 , y 0  h , z 0   f x0 , y 0 , z 0 
II) x0 , y0 , z 0   f y x0 , y0 , z 0   lim .
y h0 h
f f x 0 , y 0 , z 0  h   f x 0 , y 0 , z 0 
III)
z
x0 , y 0 , z 0   f z x0 , y 0 , z 0   lim
h 0 h
.

Example: Find f x  x , y  , f y  x , y  , f x  x , y  , f x 1, 2 , f y 1 , 2 

xy 2
a) f x , y   2 x 2 y 2  2 y  4 x. b) f  x , y   e  ln( x 2 y ).

Compiled by: Amsalu W/Gebreal Page 4

Functions of Several Variables

Solution: a) By definition,

f x  x, y   lim
f  x  h , y   f  x, y 
 lim
 2
 
2 x  h  y 2  2 y  4 x  h   2 x 2 y 2  2 y  4 x 
h 0 h h0 h
2 2 2
4 xhy  y h  4h
 lim  lim (4 xy 2  y 2 h  4)  4 xy 2  4
h 0 h h 0

 f x 1, 2  20.

And,

f y  x, y   lim
f  x , y  h   f  x, y 
 lim
  
2 x 2 ( y  h) 2  2( y  h)  4 x  2 x 2 y 2  2 y  4 x 
h0 h h0 h
2 2 2
4 x yh  2 x h  2h
 lim  lim (4 x 2 y  2 x 2 h  2)  4 x 2 y  2
h 0 h h 0

 f y 1, 2  10.

b) Exercise.
 To calculate f x  x , y  , regard y as a constant and differentiate with respect to x.
 To calculate f y  x , y  , regard x as a constant and differentiate with respect to y.

Higher order partial derivatives

f f
Suppose that f is a function of two variables x and y . Since and are also functions of
x y
x and y , these functions may themselves have partial derivatives. This gives rise to four
possible second-order derivatives of f, which are defined by:

2 f   f 
 2
    f xx ( Differentiate twice with respect to x )
x x  x 
2 f   f 
 2
    f yy ( Differentiate twice with respect to y )
y y  y 
2 f   f 
     f xy ( Differentiate first with respect to x & then with respect to y )
yx y  x 
2 f   f 
     f yx ( Differentiate first with respect to y & then with respect to x)
xy x  y 

xy 2
Example: Let f  x , y   e  ln( x 2 y ) . Find all the second order partial derivatives.

Solution: First, we find the first order partial derivatives.

Compiled by: Amsalu W/Gebreal Page 5

Functions of Several Variables

f
x , y   f x x , y   y 2 e xy  2 xy2  y 2 e xy  2 and f x , y   f y x , y   2 xye xy  1 .
2 2 2

x xy x y y
2 f
x , y     f     y 2 e xy  2   y 4 e xy  22 .
2 2
Now , i) f xx  x , y   2
x x  x  x  x x
2 f
x , y     f     y 2 e xy  2   2 ye xy  2 xy 3 e xy  2 ye xy (1  xy 2 ).
2 2 2 2
ii ) f xy  x , y  
yx y  x  y  x
2 f
x , y     f     2 xye xy  1   2 xe xy  4 x 2 y 2 e xy  12 .
2 2 2
iii ) f yy  x , y  
2
y y  y  y  y y
2 f
x , y     f     2 xye xy  1   2 ye xy  2 xy 3 e xy  2 ye xy (1  xy 2 ).
2 2 2 2
iv ) f yx  x , y  
xy x  y  x  y
Theorem: Let f be a function of two variables. If f xy  x , y  and f yx  x , y  are continuous on some
open disk, then f xy  x , y   f yx  x , y  on that disk.

3.4 Gradients and directional derivatives

Definition: 1. If f is a function of x and y, then the gradient of f, denote by f  x , y  is defined

by: f  x , y   f x  x , y i  f y  x , y  j.

2. If f is a function of three variables x, y, and z, then the gradient of f, denote by f  x , y , z 

is defined by: f  x , y , z   f x  x , y , z i  f y  x , y , z  j  f z  x , y , z k .

Example: Find the gradient of f at the given point p 0 .

a) f x , y   x ln x  y ; p 0  2 , 3
b ) f  x , y , z   ze  x tan y ; p0 0 ,  ,  2 

x
Solution: a) f  x , y   x ln  x  y .  f x  x , y   ln( x  y )  ; f x  2 , 3  2
x y
x
and f y x , y   ; f y  2 , 3  2
x y
 f  2 , 3  f x  2 , 3i  f y  2 , 3 j  2i  2 j 

b) f  x , y , z   ze  x tan y  f x  x , y , z    ze  x tan y ; f x 0 ,  ,  2   0 ;

; f y  x , y , z   ze  x sec 2 y ; f y 0 ,  ,  2   2
and f z  x , y , z   e  x tan y ; f z 0 ,  ,  2   0
 f 0 ,  ,  2   f x 0 , ,  2 i  f y 0 ,  ,  2  j  f z 0 ,  ,  2   2 j.

Compiled by: Amsalu W/Gebreal Page 6

Functions of Several Variables


Definition: Let U be a unit vector and f be a function of several variables. Then the rate of

change of f with respect to U at a point p 0 is called the directional derivative of f at p 0 in the
 
direction of U denoted by D  f  p 0  and is defined as: D f  p0   f  p0  U
U U

 

Note: If U  0 and non-unit, then U is a unit vector.

U

Example: Find the D f  p0  for

U


a) f x , y   x 2  x 2 y  y 2 , p0 1, 2 &U  1,1.
2 1 2 
b ) f  x , y , z   x 2 y  yz 3  z , p 0 1,  2 , 0  & U   , ,  
3 2 3 

Solution: a) The partial derivatives of f are:

f x  x , y   2 x  2 xy ; f x 1, 2  6 and f y  x , y   x 2  2 y ; f y 1, 2  5
 f 1, 2  6i  5 j.
 
Since, U  1,1, U  12  12  2  1 , U is not a unit vector.


U 1 1
But U  i j is a unit vector.
2 2

Thus, D  f  p 0    f  p 0   U  6 i  5 j    1 i  1 j   11  11 2 .
U U  2 2  2 2

b) The partial derivatives of f are:

f x  x , y , z   2 xy ; f x 1,  2 , 0   4 , f y  x , y , z   x 2  z 3 ; f y 1,  2 , 0   1
and f z  x , y , z   3 yz 2  1 ; f z 1,  2 , 0   1
 f 1,  2 , 0  4i  j  k .

2 2 2
and U  2 1  2
          1 . That is, u is a unit vector. Thus,
3 3  3

U 2 1 2 
D  f  p0   f  p0    4 i  j  k    i  j  k    3 .
U U  3 3 3 

Compiled by: Amsalu W/Gebreal Page 7

Functions of Several Variables

Theorem: Let U be a unit vector and f is partially differentiable at a point p0 , then the maximum
value of D f  p0  is f  p0  .
U

  
Proof: D f  p 0    f  p 0   U   f  p 0   U cos  , where  is the angle between f  p0  and U .

U

 D f  p0 max  f  p0  , when   0 0
U

 D f  p0 min   f  p0  , when   
U

Example: The temperature at each point on a metal plate is given by T  x , y , z  

e x cos y  e y sin x  z . Find the direction for which T decreases most rapidly at 0 , 0 , 0 .

Solution: The partial derivatives of f are:

Tx  x , y , z   e x cos y  e y cos x ; Tx 0 , 0 , 0   2
T y  x , y , z   e x sin y  e y sin x ; T y 0 , 0 , 0   0 and
Tz  x , y , z   1 ; T y 0 , 0 , 0   1
Thus, T 0 , 0 , 0  2i  k .


Therefore, T decreases rapidly along u`  1 5  2i  k  because it decreases rapidly at
opposite to the gradient T 0 , 0 , 0  2i  k and T increases rapidly when it is in the same
direction with the gradient.

Total differential and tangent planes.

Definition: A function f  x , y  is said to be differentiable at  x0 , y0  if there exists a disk D

centered at  x0 , y0  and functions  1 and  2 of two variables such that

f  x , y   f  x0 , y0   f x  x0 , y0  x  x0   f y x0 , y0  y  y0    1  x , y  x  x0 
  2  x , y  y  y 0  , for  x , y  in D where
lim  1 x , y   0 and lim  2 x , y   0
 x , y  x0 , y0   x , y  x0 , y0 

Theorem: A function f  x , y  is differentiable at  x0 , y0  if f x  x0 , y0  and f y  x0 , y 0  exist and

continuous at  x0 , y0  .

Example: Check whether f is differentiable or not.

a) f  x , y   x 2  y 2  xy and x0 , y0   0 , 0  .
Compiled by: Amsalu W/Gebreal Page 8
Functions of Several Variables

b) f x , y   x  xy and x0 , y 0   0 , 0 .
Solution: a)
f  x , y   x 2  y 2  xy  i ) f x  x , y   2 x  y ; f x 0 , 0   0 ii ) f y  x , y   2 y  x ; f y 0 , 0   0.
That is, both f x 0 , 0 and f y 0 , 0  exist and continuous. Therefore, f x , y   x 2  y 2  xy is
differentiable at 0 , 0 .

b. Since f x 0 , 0 doesn’t exist f  x , y   x  xy is not differentiable at 0 , 0 .

For a differentiable function of two variables, z  f  x , y  , we define the differentials dx and dy

to be independent variables. Then the differential dz , also called the total differential, is defined
z z
by: dz  f x  x , y dx  f y  x , y dy  dx  dy .
x y

Similarly, for a differentiable function w  f  x , y , z  of three variables, the total differential, dw

w w w
is defined by: dw  f x  x , y , z dx  f y  x , y , z dy  f z  x , y , z dz  dx  dy  dz .
x y z

Note: sometimes the notation df is used in place of dz (dw).

Example: If z  f x , y   x 2  3 xy  y 2 , find the total differential dz .

z z
Solution: dz  df  f x  x , y dx  f y  x , y dy  dx  dy  2 x  3 y dx  3x  2 y dy .
x y

Tangent plane

Definition: If f  x , y  is differentiable at the point  x0 , y0  , then the tangent plane to the

surface z  f  x , y  at the point p0  x0 , y0 , f  x0 , y0  or  x0 , y0  is the plane:

z  f  x0 , y0   f x  x0 , y 0  x  x0   f y  x0 , y 0  y  y 0  .

Note: The normal line of the plane can be expressed parametrically as:

x  x0  f x  x0 , y 0 t , y  y 0  f y  x0 , y0 t , z  f  x0 , y0   t , t   .

Example: Find an equation for the tangent plane and parametric equation for the normal line to
the surface z  f  x , y   x 2 y at the point 2 ,1, 4  .

Solution: Here p0  2 ,1, 4  . The partial derivatives of f are:

Compiled by: Amsalu W/Gebreal Page 9

Functions of Several Variables

f x  x , y   2 xy ; f x 2 ,1   4 and f y  x , y   x 2 ; f y 2 ,1   4

Therefore, the tangent plane has equation:

z  f  x 0 , y 0   f x  x 0 , y 0  x  x0   f y  x 0 , y 0  y  y 0   4  4 x  2   4 y  1  4 x  4 y  8.

And the normal line has equations:

x  x0  f x  x0 , y 0 t , y  y 0  f y  x0 , y 0 t , z  f x0 , y0   t , t  
x  2  4t , y  1  4t , z  4  t , t  .

Theorem: Assume that F  x , y , z  has continuous first-order partial derivatives and let
c  F  x0 , y0 , z 0  . If F  x0 , y 0 , z 0   0 , then F  x0 , y0 , z0  is a normal vector to the surface

F  x , y , z   c at the point p0  x0 , y0 , z 0  and the tangent plane to this surface at p0  x0 , y0 , z 0  is

the plane with equation:

Fx  x0 , y 0 , z 0  x  x0   Fy  x0 , y 0 , z 0  y  y0   Fz  x0 , y 0 , z 0  z  z 0   0 .

Example: Find an equation of the plane tangent to the sphere x 2  y 2  z 2  4 at the point
 1,1, 2 .
Solution: The sphere is the level surface F  x , y , z   4 , where F  x , y , z   x 2  y 2  z 2 .
 
Now , Fx  x , y , z   2 x ; Fx  1,1, 2  2 ; Fy  x , y , z   2 y ; Fy  1,1, 2  2  

Fz  x , y , z   2 z ; Fz  1,1, 2  2 2 
Therefore, an equation of the plane tangent at  1 , 1 , 2 is:  
Fx  x0 , y 0 , z 0  x  x0   Fy x0 , y0 , z 0  y  y0   Fz  x0 , y0 , z 0  z  z 0   0
 
 2 x   1  2 y  1  2 2 z  2  0   x  y  2 z  4.

3.5 Applications: Tangent plane approximation of values of functions.

Let f  x , y  is differentiable at the point  x0 , y0  . Thus, the equation of the tangent plane to the
graph of f at  x0 , y0  is z  f  x , y   f  x0 , y0   f x  x0 , y 0  x  x0   f y  x0 , y 0  y  y0 
Now, if the point P x , y  is very close to p0  x0 , y0  , then
f  x , y   f  x0 , y 0   f x  x0 , y0  x  x0   f y  x0 , y 0  y  y 0  which can be written in the form:

f  f x  x 0 , y 0 x  f y  x 0 , y 0 y where f  f  x , y   f  x 0 , y 0  , x  x  x 0 , y  y  y 0 .

Compiled by: Amsalu W/Gebreal Page 10

Functions of Several Variables


Example: Approximate ln x 2  y 2 at 0.01, 0.98.
Solution: Let f  x , y   ln x 2  y 2  , P 0.01, 0.98   x , y  , P0 0 ,1   x0 , y 0  . Then
2x 2y
f 0 ,1  ln 0  1  ln 1  0 . Now, f x  x , y   2 2
; f x 0 ,1  0 & f y  x , y   2 ; f y 0 ,1  2.
x y x  y2

Thus , f  x , y   f  x0 , y 0   f x  x0 , y 0  x  x0   f y  x0 , y0  y  y 0 
 f 0.01, 0.98  f 0 ,1  f x 0 ,10.01  0   f y 0 ,10.98  1  0.04

 ln 0.01  0.98  0.04.
2 2

Example: Approximate e 0.2 ln 0.9.

Solution: Let f  x , y   e x ln y .Then f 0 ,1  0 , P 0.2 , 0.9    x , y  , P0 0 ,1   x0 , y 0  .

ex
The first partial derivatives of f are: f x  x , y   e ln y ; f x 0 ,1  0 & f y  x , y   ; f y 0 ,1  1.
x

Now, f  x , y   f 0.2 , 0.9   e 0.2 ln 0.9   f  x0 , y 0   f x  x 0 , y 0  x  x0   f y  x0 , y 0  y  y 0 

 f 0 ,1  f x 0 ,10.01  0  f y 0 ,10.9  1  0.1
 e 0.2 ln 0.9  0.1.

Exercise: Approximate sin  1.962.03 .

3.6 The chain rule

- It is used to differentiate composite functions of one or more variables.

Let y be a differentiable function of x (i.e. y  f  x  ) and x be a differentiable function of t

dy dx
(i.e. x  g t  . Then y  f  x   f  g t   fog t  and  f '  x  ,  g ' t  .
dx dt

dy d dy dx dy dy dx
Now,   fog t   f '  g t g ' t   f '  x g ' t    .Thus,   .
dt dx dx dt dt dx dt

We will now see the version of the chain rule for functions of several variables.

Theorem: (Two-variables chain rule)

If x  xt  and y  y t  are differentiable at t, and if z  f  x , y  is differentiable at

dz z dx z dy
x , y   xt , yt  , then z  f  xt  , yt  is differentiable at t and   where
dt x dt y dt

Compiled by: Amsalu W/Gebreal Page 11

Functions of Several Variables

the ordinary derivatives are evaluated at t and the partial derivatives are evaluated at  x, y  .
Example: Let z  f  x , y   2 x 2  y 2 , x  cos t , y  sin t . Find dz dt .

dz z dx z dy
Solution:    4 x  sin t   2 y cos t  4 cos t sin t  2 sin t cos t  2 cos t sin t
dt x dt y dt

dz
  2 cos t sin t.
dt

Theorem :( Three-variables chain rule)

If x  xt  , y  y t  and z  z t  are differentiable at t, and if w  f  x , y , z  is differentiable at

x , y , z   xt , yt , z t  , then w  f  xt  , yt  , zt  is differentiable at t and

dw w dx w dy w dz
  
dt x dt y dt z dt

where the ordinary derivatives are evaluated at t and the partial derivatives are evaluated at
x , y , z  .
dw
Example: Given that w  x 2  y 2  z 2 , x  e t , y  e  t z  2t . Find .
dt

Solution:

dw w dx w dy w dz x y z
    et  e t  2
dt x dt y dt z dt 2
x y z 2 2 2
x y z 2 2
x  y2  z2
2

e 2t e 2 t 4t e 2t  e 2t  4t
    .
e 2t  e  2t  4t 2 e 2t  e  2t  4t 2 e 2t  e  2t  4t 2 e 2t  e  2t  4t 2

The chain rule for partial derivatives

Theorem :( Two-variables chain rule)

If x  xu , v  and y  y u , v  have first order partial derivatives at the point u , v  , and if
z  f  x , y  is differentiable at  xu , v  , yu , v  , then z  f  xu , v , yu , v  has first partial
derivatives at u , v  given by:

z z x z y z z x z y
  and  
u x u y u v x v y v

Compiled by: Amsalu W/Gebreal Page 12

Functions of Several Variables

z z
Example: Let z  x  y and x  e uv , y  ln uv  . Find and .
u v

Solution:

z z x z y z z x z y
  ;  
u x u y u v x v y v
1 1 1 1
 
 1 ve uv   1  ve uv 
u u
 
 1 ue uv   1  ve uv 
v v
z 1 z 1
  ve uv  ;  ue uv 
u u v v

Theorem: (Three-variables chain rule)

If the functions x  xu , v  , y  y u , v  and z  zu , v  have first order partial derivatives at the
point u , v  , and if the function w  f  x , y , z  is differentiable at  x u , v , y u , v , z u , v   , then
w  f  xu , v  , yu , v , zu , v  has first order partial derivatives at u , v given by:

w w x w y w z w w x w y w z
   and   
u x u y u z u v x v y v z v

w w
Example: Given that w  y ln xz , x  ve u , y  u 2 v 4 , z  ue v . Find and .
u v

Solution:

w w x w y w z y u y
    ve  ln xz 2uv 4  e v  uv 4 u  2 ln uveu e v  1 .
u x u y u z u x z
   
w
Similarly, it can be shown that
u
   
 u 2 v 3 v  4 ln uve u e v  1 (show!)


Exercise: If g u , v   f u 2  v 2 , v 2  u 2  and f is differentiable, show that g satisfies the
g g
equation: v u  0.
u v

Implicit differentiation

If z can’t be written explicitly in terms of x and y, we use implicit differentiation to find

z z
and (assume x and y are independent).
x y

Compiled by: Amsalu W/Gebreal Page 13

Functions of Several Variables

z z
Example: Consider the equation: x 2  y 2  xyz  z 2  10 . Find and .
x y

Solution:

x 2  y 2  xyz  z 2  10 ; x 2  y 2  xyz  z 2  10
 2  2

x

x  y 2  xyz  z 2  10  
y

x  y 2  xyz  z 2  10 
z z z z
 2 x  0  yz  0  xy  2 z 0  0  2 y  xz  xy  2 z 0
x x y y
z  2 x  yz  z  2 y  xz 
   
x xy  2 z y xy  2 z
z  2 x  yz  z  2 y  xz 
  ; 
x xy  2 z y xy  2 z

Theorem: If the equation f x , y   c defines y implicitly as a differentiable function of x, and if

f dy
 0 , then   f x .
y dx f y

Example: Let x 3  y 2 x  3  0 . Find dy dx .

f f
Solution: Let f  x , y   x 3  y 2 x  3  0 . Then  3 x 2  y 2 and  2 xy . Now,
x y
dy 3x 2  y 2 dy 3x 2  y 2
  f x  . Thus,  .
dx f y 2 xy dx 2 xy

Differentiating the given equation gives:

d 3 d
x3  y 2 x  3  0 
dx
 
x  y 2 x  3  0
dx
dy dy 3x 2  y 2
 3 x 2  y 2  2 yx  0  
dx dx 2 xy

3.7 Relative extreme of function of two variables

Definitions: 1. A function f  x , y  is said to have

local/relative maximum value at  x0 , y 0  if there exists

a circular region centered at  x0 , y 0  such that

f  x0 , y0   f  x , y  for each x , y  in the region.

Compiled by: Amsalu W/Gebreal Page 14

Functions of Several Variables

2. A function f  x , y  is said to have local/relative minimum value at  x0 , y 0  if there exists a

circular region centered at  x0 , y 0  such that f x 0 , y 0   f  x , y  for each x , y  in the region.

3. The local maximum and local minimum values together are called extreme values.

Note: In the inequalities in definitions 1 and 2 hold for all points  x, y  in the domain of f

then f has an absolute/global maximum(or absolute/global minimum) at  x0 , y 0  .

Example: 1. Let f  x , y   x 2  y 2  2. Then f 0 , 0  2  f  x , y   x , y 

Thus, f 0 , 0  2 is absolute/global minimum value.

2. Let f  x , y    x 2  y 2  1. Then f 0 , 0   1  f  x , y   x , y 

Thus, f 0 , 0   1 is absolute/global maximum value.

Definitions: 1. A point  x0 , y 0  is called a critical point (stationary point) of f x , y  if

f x  x , y   0 and f y  x , y   0 , or if one of these partial derivatives doesn’t exist.

2. A critical point p0  x0 , y 0  is called a saddle point if f  x0 , y0  is neither local maximum value

nor local minimum value.

Theorem: If f x , y  has a local maximum value or local minimum value at  x0 , y 0  and the

first order partial derivatives of f exist there, then f x  x0 , y 0   0 and f y  x0 , y 0   0 .

Example: Let f  x , y   x 2  y 2  2 x  6 y  14. Find the critical point of f x , y  .

Solution: The first order partial derivatives are: f x x , y   2 x  2 ; f y x , y   2 y  6.

Now, f x  x , y   0 & f y  x , y   0  2 x  2  0 & 2 y  6  0  x  1 & y  3.

Thus, the point 1, 3 is the critical point.

   
 x 2  y 2  2 x  6 y  14  x 2  2 x  1  y 2  6 y  9  4 
 
   2  y 
 1
x 2  4  4 x , y 
 3
 0 0 

That is, 4 is absolute minimum value and it occurs at 1, 3 .

Compiled by: Amsalu W/Gebreal Page 15

Functions of Several Variables

Theorem :( Second derivative test)

Suppose the second partial derivatives of f x , y  are continuous on a disk with center  x0 , y 0  ,
and suppose that f x  x0 , y 0   0 and f y  x0 , y 0   0 (i.e.  x0 , y 0  is a critical Point of f x , y  ).


Let D  f xx  x0 , y 0  f yy  x0 , y 0   f xy  x0 , y 0  . 
2

1. If D  0 and f xx  x0 , y 0   0 , then f  x0 , y0  is a local minimum value.

2. If D  0 and f xx x0 , y 0   0 , then f  x0 , y0  is a local maximum value.
3. If D  0 , then f  x0 , y0  is neither local maximum value nor local minimum value.
(i.e.  x0 , y 0  is a saddle point)

Note: 1. If D=0, the test gives no information. f x , y  could have a local maximum value or

local minimum value at  x0 , y 0  , or  x0 , y 0  could be a saddle point of f x , y  .

2. To remember the formula for D it’s helpful to write it as a determinant.

f xx  x0 , y 0  f xy  x0 , y 0 
D
f yx  x0 , y 0  f yy  x0 , y 0 
 
 f xx  x0 , y0  f yy  x0 , y0   f xy  x0 , y 0 
2

Example: Find the local maximum, local minimum and saddle

p point of f x , y   x 4  y 4  4 xy  1 .

Solution: We first find the critical points:

f x x , y   4 x 3  4 y ; f y x , y   4 y 3  4 x .

Now, f x  x , y   4 x 3  4 y  0 & f y  x , y   4 y 3  4 x  0
 x 3  y  0 & y 3  x  0  x 9  x  0  x  0 ,1,  1.

Thus, the critical points are 0 , 0  1,1 &  1,  1 .

Next we calculate the second partial derivatives and D  x0 , y 0 .

f xx  x , y   12 x 2 , f yy  x , y   12 y 2 , f xy  x , y   4
 
2
D x , y   f xx  x , y  f yy  x , y   f xy  x , y   144 x 2 y 2  16

Now, i) Since D0 , 0   16  0 , it follows that 0 , 0  is a saddle point.

ii) Since D1,1  128  0 and f xx 1,1  12  0 , f 1,1  1 is local minimum.

Compiled by: Amsalu W/Gebreal Page 16

Functions of Several Variables

iii) Since D 1,1  128  0 and f xx  1,  1  12  0 , f 1,1  1 is also local minimum.

2
Example: A rectangular box without a lid is to be made from 12m of

card board. Find the maximum volume of such a box.

Solution: Let the length, width, and height of the box be x, y and z (in meters).

Then the volume, V of the box is: V  xyz .The area of the four sides and the bottom is:

12  xy
2 xz  2 yz  xy  12  z (2 x  2 y)  12  xy  z  .
2x  2 y

 12  xy  12 xy  x 2 y 2
Thus, the expression for V becomes V  xyz  xy    .
 2x  2 y  2x  2 y

The first partial derivatives are:

V


y 2 12  x 2  2 xy  ,
V


x 2 12  y 2  2 xy 
2x  y  2 x  y 
2 2
x y

 y2 12  x 2
 2 xy
0


Now ,
V
 0
V 
  2
2 x  y 
2



 y 2 12  x 2  2 xy  0 
 2
x y x 
12  y 2  2 xy
 0
 
 x 12  y 2  2 xy  0 
 2 x  y 
2

2
 y  0 or 12  x  2 xy  y  0 or x 2  12  2 xy
0
   
 x  0 or 12  y 2  2 xy
0  x  0 or y 2  12  2 xy
 y  0 or x 2  y 2  y  0 or 12  3 x 2  0  y  0 or x  2
     
x  0 ( x  y) x  0  x  0 or y  2

Hence, 0 , 0 and 2 , 2 are critical points.

We compute the second partial derivatives:

  V
  
2
 2
   y 12  x  2 xy     2 xy 2
 
 2 y 3 2 x  y   4 x  y  12 y 2  x 2 y 2  2 xy
2

 x  2 x  y   4 x  y 
2 4
x  x 
  2V 
  2  2 , 2   2 ;
 x 

Compiled by: Amsalu W/Gebreal Page 17

Functions of Several Variables


  V    12 y 2  x 2 y 2  2 xy 3
  
   24 y  2 x 2
 
y  6 xy 2 2 x  y   4 x  y  12 y 2  x 2 y 2  2 xy 3
2

y  x  y  2 x  y 
2 
 4 x  y 
4

  2V  1
   2 , 2    and
 yx  2
  V


   24 x 2  4 x 3 y  2 x 2 y 2
  
    4 x 3
 2

 4 x 2 y 4 x  y   8 x  y  24 x 2  4 x 3 y  2 x 2 y 2 
y  y  y  2 x  y 
2 
 16 x  y 
4

  2V

  2  2 , 2   1
 y 
Next we calculate D2 , 2  :

 
D2 , 2  Vxx 2 , 2V yy 2 , 2  Vxy 2 , 2  7 4  0. Moreover, Vxx 2 , 2  2  0 .
2

12  xy
Therefore, by second derivative test V is maximum when x  2 , y  2 and z  1 .
2x  2 y

Exercises:1. Find the shortest distance from the point 1 , 0 ,  2  to the plane x  2 y  z  4 .

3
2. Find the minimum cost of a rectangular box with volume 48m if the front and back side cost
$1 , the top and bottom cost $ 2 2 , and the two ends cost $ 3 2 .
m2 m m

3.8 Largest and smallest values of a function on a given set

2
Definitions: 1. A closed set in  is one that contains all its boundary points.

2
2. A bounded set in  is one that is contained within some disk.

Theorem: (Extreme value theorem for functions of two variables)

2
If f is continuous on a closed, bounded set D in  , then f attains an absolute maximum value
f  x1 , y1  and an absolute minimum value f x2 , y2  at some points  x1 , y1  and x2 , y2  in D.

 To find the absolute maximum and minimum values of a continuous function f on

a closed , bounded set D, we follow the following procedures:
I. Find the values of f at the critical points of f in D.
II. Find the extreme values of f on the boundary of D.

Compiled by: Amsalu W/Gebreal Page 18

Functions of Several Variables

III. The largest of the values from steps I and II is the absolute maximum value;
and the smallest of these values is the absolute minimum value.

Example: Find the absolute maximum and minimum of the

function f  x , y   x 2  2 xy  2 y on the rectangle

R   x , y  : 0  x  3 , 0  y  2.

Solution: Since f is continuous on a closed, bounded rectangle R, f has both absolute maximum
and absolute minimum.

Step I) We first find the critical points.

f x  x , y   2 x  2 y  0 , f y  x , y   2 x  2  0  x  y & x  1 .

Therefore, the critical point is 1 , 1 and f 1,1  1 .

Step II) We look at the values of f on the boundary of D.

- On L1 , y  0 and f  x , 0   x 2 , 0  x  3 .
f 0 , 0   0 and f 3 , 0  9 are minimum and maximum values of f on L1 respectively.
- On L2 , x  3 and f 3, y   9  4 y , 0  y  2 .
f 3 , 2   1 and f 3 , 0   9 are minimum and maximum values of f on L2 respectively.
- On L3 , y  2 and f  x , 2  x 2  4 x  4 , 0  x  3 .
f 2 , 2  0 and f 0 , 2   4 are minimum and maximum values of f on L3 respectively.
- On L4 , x  0 and f 0 , y   2 y , 0  y  2 .
f 0 , 0   0 and f 0 , 2  4 are minimum and maximum values of f on L4 respectively.

Hence, on the boundary, the maximum value of f is 9 and the minimum value is 0.

Step III) We compare the values from step I and step II.

The absolute maximum value of f on D is f 3 , 0   9 and the absolute minimum value of f on D

is f 0 , 0  f 2 , 2   0.

Exercise: Find the maximum and minimum values attained by the function
f  x , y   xy  x  y  3 at points of the triangle region R in the xy-plane with vertices at 0 , 0 ,
2 , 0 and 0 , 4 .

Compiled by: Amsalu W/Gebreal Page 19

Functions of Several Variables

3.9 Extreme values under constraint conditions: Lagrange’s multiplier.

In example (*), we maximized the volume function V  xyz subject to the condition
2 xz  2 yz  xy  12 . Such a condition on the values of x , y , and z is known as constraint (or
side condition).

Method of Lagrange Multiplier

 To find the maximum and minimum values of f  x , y, z  subject to the constraint

g  x , y, z   k (assuming that these extreme values exist and g  0 on the surface
g  x , y, z   k ), we follow the following two steps:
Step I) Find all values of x , y , z , and  such that f  x , y, z   g  x , y, z   and
g  x , y, z   k .
Step II) Evaluate f at all the points  x , y, z  that result from step I. The largest of these
values is the maximum value of f; the smallest is the minimum value of f.

Note:

1. The number  in equation  is called Lagrange multiplier.

2. If we write the vector equation f  x , y , z   g  x , y , z  in terms of its components,
then the equation in step I become:

f x  x , y , z   g x  x , y , z  , f y  x , y, z   g y  x , y, z  , f z  x , y, z   g z  x , y , z  , g  x , y, z   k

3. For functions of two variables the method of Lagrange multiplier is similar to the method
for functions of three variables. To find the extreme values of f  x , y  subject to the
constraint g x , y   k , we look for values of x, y and  such that f  x , y   g  x , y 
and g x , y   k .
4. In the case of two constraints g  x , y, z   k and h x , y, z   c ,we find the values of f
x , y , z ,  and  (  and  are called Lagrange multipliers) such that
f  x , y, z   g  x , y, z   h x , y, z  . That is,

f x  x , y , z   g x  x , y , z   hx  x , y , z 
f y  x , y, z   g y  x , y, z   h y  x , y, z 
f z  x , y , z    g z  x , y , z    hz  x , y , z 
g  x , y, z   k
h x , y, z   c

Compiled by: Amsalu W/Gebreal Page 20

Functions of Several Variables

2
Example: A rectangular box without a lid is to be made from 12m of card board. Find the
maximum volume of such a box.

Solution: We wish to maximize V  xyz subject to the constraint g  x , y , z   2 xz  2 yz  xy  12

We look for values of x , y , z &  such that V  x , y , z   g x , y, z  & g  x , y, z   12 . That is,
V x  x , y , z   g x  x , y, z  , V y  x , y, z   g y  x , y , z  , V z  x , y , z    g z  x , y , z  , 2 xz  2 yz  xy  12
 yz   2 z  y 
 xz   2 z  x   xyz   2 xz  xy 
 
   xyz   2 yz  xy 
 xy   2 x  2 y   xyz   2 xz  2 yz 
2 xz  2 yz  xy  12 

Observe that   0   0  yz  xz  0  2 xz  2 yz  xy  12

2 xz  xy  2 yz  xy  xz  yz But z  0 z  0  V  0 x  y
  
2 yz  xy  2 xz  2 yz  xy  2 xz But x  0 x  0  V  0  y  2z
2 2 2
If we now put x  y  2 z in 2 xz  2 yz  xy  12 , we get 4 z  4 z  4 z  12
 z 2  1  z  1(as z  0) . Since x , y , and z are all positive, we have z  1, x  2 , and y  2
as before.

Example: Find the extreme values of the function f ( x , y )  x 2  2 y 2 on the circle x 2  y 2  1 .

Solution: Here, we are asked for the extreme values of f subject

to the constraint g  x , y   x 2  y 2  1 .

We solve the equation f  x , y   g  x , y  and g  x , y   1 .That is,

2 x   2 x

f x  x , y    g x  x , y  , f y  x , y    g y  x , y  , g  x , y   1  4 y   2 y
 2 2
x  y  1

Now, 2 x   2 x  x  x  x  x  0  x1     0  x  0 or   1.

If x  0 , then x 2  y 2  1 gives y  1.

If   1, then y  0( from 4 y   2 y ), so x 2  y 2  1 gives x  1.

Thus, f has possible extreme values at the points 0 ,1 , 0 ,  1 , 1, 0  and  1, 0  .

f 0 ,1  2 , f 0 ,  1  2 , f 1, 0  1 & f  1 , 0   1.

Compiled by: Amsalu W/Gebreal Page 21

Functions of Several Variables

Therefore, the maximum value of f on the circle x 2  y 2  1 is f 0 ,  1  2 and the minimum

value is f  1 , 0   1 .

Exercise: Find the points on the sphere x 2  y 2  z 2  4 that are closest to and farthest from
the point 3 ,1 ,  1 .

Example: Find the maximum value of the function f  x , y , z   x  2 y  3z on the curve of

intersection of the plane x  y  z  1and the cylinder x 2  y 2  1 .

Solution: We maximize the function f  x , y , z   x  2 y  3z subject to the constraints

g  x , y , z   x  y  z  1 and h x , y , z   x 2  y 2  1.

The Lagrange condition is:

f  x , y, z   g  x , y, z   h x , y, z 
i.e. f x  x , y , z    g x  x , y, z   hx  x , y , z  ; 1    2x

f y  x , y , z   g y  x , y , z   h y  x , y, z  ; 2    2 y
f z  x , y, z   g z  x , y, z   h z  x , y , z  ; 3
g  x , y, z   k ; x y  z 1
h x , y, z   c ; x2  y2  1

Now, putting   3 in (1    2x ), we get 2x  2  x  1  .

Putting again   3 in ( 2    2y ) gives y  5 2 .

Substituting x  1  and y  5 2 in x 2  y 2  1 , gives:

1 25 29 29
2
 2
1 2   .
 4 4 2

2 5 7
Then x   ,y  and from z  1  x  y , z  1  .
29 29 29

The corresponding values of f are  2  5   7 

 2    3 1    3  29 .
29  29   29 

Hence, the maximum value of f on the given curve is 3  29 .

Exercise: Use Lagrange multipliers to find the maximum and minimum values of the function
f  x , y , z   x  2 y subject to the constraints x  y  z  1and y 2  z 2  4 .

Compiled by: Amsalu W/Gebreal Page 22