F.

Leja – Rachunek różniczkowy i całkowy

I. Foltyńska, Z. Szafrański, Z. Ratajczak – Matematyka cz. II
M. Gewert, Z. Skoczylas – Analiza matematyczna 2 – def. , tw., wzory
E. Swokowski – Calculus with Analytic Geometry

FUNCTIONS OF SEVERAL VARIABLES

1. Introduction
The term ordered pair refers to two real numbers, where one is designated as the “first”
number and the other as the “second”. The symbol ( x, y) is used to denote the ordered
pair consisting of the real numbers x, y . Every ordered pair ( x, y) determines a point
P in the xy-plane with coordinates x, y and conversely, to each point in xy-plane
a unique ordered pair may be assigned.

The distance between two points P0 ( x0 , y0 ) and P( x, y ) in a coordinate plane is given

by d ( P0 , P) | P0 P | ( x  x0 ) 2  ( y  y0 ) 2 .

If P0 ( x0 , y0 , z0 ) and P( x, y, z ) then d ( P0 , P) | P0 P | ( x  x0 )2  ( y  y0 )2  ( z  z0 )2 .

(the distance between two points in three dimensions)

A neighbourhood of a point P0 of radius r  0 is a set of points P such that P0

d ( P0 , P)  r .

The neighbourhood of P0 in the xy-plane is a circle of radius r with the centre in

( x0 , y0 ) . In three dimensions there is a sphere of radius r with the centre in

( x0 , y0 , z0 ) .

Deleted neighbourhood
A point P is an interior point of a set D if there exists a circle with centre at P which
contains points of D only.
A point P is called a boundary point of D if every circle with the centre P contains
points that are in D and points that are not in D.
A region is closed if it contains all of its boundary points.
A region is open if it contains none of its boundary points; that is, every point of the
region is an interior point .

1
2. Functions of two variables
Recall that a function f is a correspondence that associates with each element of a set X
a unique element of a set Y. If both X and Y are subsets of  , then f is called a
function of (one) real variable. If X is a subset of a product    then f is called a
function of two (real) variables.

Definition 2.1 Let D be a set of ordered pairs of real numbers. A function f of two
variables is a correspondence that associates each pair ( x, y) in D with a unique real
number z, denoted by f ( x, y) . The set D is called the domain of f. The range of f
consists of all real numbers f ( x, y) , where ( x, y) is in D.

For functions of two variables we may represent the domain D by points in an xy-plane.
1
Example 2.1 Find the domain of f ( x, y )  .
x y

Example 2.2 Find the domain of f ( x, y)  1  x 2  y 2 .

Example 2.3 Find the domain of f ( x, y)  arcsin( x 2  y 2  3) .

By definition, the graph of f is a graph for the equation z  f ( x, y) in the three-

dimensional rectangular coordinate system. Hence, it is usually a surface S of some
type.

3. Graphs and Surfaces

1) Plane: f ( x, y)  z  5  x  y ;
z 5 x; x  y  5, x  0, x  5, y  0,

2) Circular paraboloid (paraboloid of the revolution) z  a( x 2  y 2 )

The graph of the parabola z  ax 2 is revolved about the z-axis. A circular paraboloid
is generated. The z-axis is called the axis of the paraboloid.

2
 x2 y 2
3) Paraboloid has the equation   cz .
a 2 b2

Trace of the surface – a curve formed by the intersection of a surface and a plane.

4) Circumference x 2  z 2  r 2 is revolved about the z-axis. A sphere

x2  y2  z 2  r 2 is generated.

5) ( x  a)2  ( y  b)2  ( z  c)2  r 2 sphere with the centre in (a, b, c) and

radius r.

x2 y 2 z 2
6) Ellipsoid has the equation    1.
a 2 b2 c 2

7) A line z  kx , (k  0) is revolved about the z-axis. A cone z 2  k 2 ( x2  y 2 )

is generated. The z-axis is the axis of the cone. Traces in plane which are
parallel to the xy-plane are circumferences.

x2 y 2 z 2
8) In general, a cone has the equation    0 ( an elliptical cone).
a 2 b2 c2
Traces in plane parallel to the xy-plane are ellipses.

In 2) - 8) we deal with quadratic surfaces.

In 2-space the graph of x 2  y 2  1 is a circle centred at the origin. However, in 3-space

we can interpret the graph of the set ( x, y, z) : x 2


 y 2  1, z ar bit r ar y as a surface
that is a right-circular cylinder.

9) Cylinder
If K is a curve in a plane and l is a line not parallel to the plane, then the
set of all points ( x, y, z ) generated by a moving line traversing K parallel to
l is called a cylinder. This curve K is called the directrix of the cylinder.

3
 x2 y2
- elliptic cylinder   1;
a 2 b2
- parabolic cylinder z  x2 ;
x2 z 2
- hyperbolic cylinder   1.
a2 b2

Thus an equation of a curve in a coordinate plane, when considered in three dimensions,

becomes an equation of a cylinder perpendicular to that coordinate plane.

4. Limits and Continuity

Definition 4.1 Let a function f of two variables be defined at every point ( x, y) in the
interior of a circle centred at (a, b) , except possibly at (a, b) . Then

lim f ( x, y)  g 
( x , y ) ( a , b )

   0   0 ( x, y) 0  ( x  a)2  ( y  b)2    f ( x, y)  g   .

The notation ( x, y ) approaching a point (a, b) is not as simple as in functions of one

variable where x  a . It means x can approach a only from the left or from the right.
In the xy-plane, there are, of course, an infinite number of ways of approaching a point
(a, b) . In order for lim f ( x, y) to exist, we require that f approaches the same
( x , y ) ( a , b )

number g along every possible curve or path trough (a, b) . Put in negative way:
If f ( x, y) does not approach the same number g for two different paths to (a, b) ,

then lim f ( x, y) does not exist.

( x , y ) ( a , b )

Definition 4.2 - Continuity

A function z  f ( x, y) is continuous at (a, b) if f (a, b) is defined, if

lim f ( x, y) exists, and the limit is the same as f (a, b) that means, f is
( x , y ) ( a , b )

continuous at (a, b) if lim f ( x, y)  f (a, b) .

( x , y ) ( a , b )

4
A function z  f ( x, y) is continuous on a region D of the xy-plane if f is continuous
at every point of D.

The sum and product of two continuous functions are continuous; the quotient of two
continuous functions is continuous, except at points where the denominator is zero.
Lastly, the graph of a continuous function is a surface with no breaks.

5. Partial Differentiation
The derivative of a function of one variable y  f (x) is given by

f ( x0  x)  f ( x0 )
f ' ( x0 )  lim .
x 0 x
In exactly the same manner, we can define a derivative of a function of two variables
z  f ( x, y) with respect to each variable.

Definition 5.1
If z  f ( x, y) , then the partial derivative with respect to x is
f f ( x  x, y)  f ( x, y)
 lim
x x  0 x

and the partial derivative with respect to y is

f f ( x, y  y )  f ( x, y )
 lim ,
y y0 y
provided each limit exists.

f
Note that in the definition of , the variable y is fixed. So, the following remark is
x
valid

Remark 5.1
f
To compute , use the laws of ordinary differentiation while treating y as a constant.
x
f
To compute , use the laws of ordinary differentiation while treating x as a constant.
y

5
 
A symbol such as denotes the operation of taking a partial derivative, in this case
x
 3
with respect to x; for example,
x
x  y 2   3x 2 .

f f
The partial derivatives and can be represented by alternative symbols:
x y
f f
 fx  fy .
x y

Example 5.1 Find the first partial derivatives of the function f ( x, y)  x3 y 2 .

f f ( x  x, y)  f ( x, y) ( x  x)3 y 2  x3 y 2
 lim  lim 
x x 0 x x 0 x
x 3  3x 2 x  3x(x) 2  (x) 3  x 3
 y 2 lim  y 2 lim 3x 2  3xx   3x 2 y 2 ,
x 0 x x 0

f f ( x, y  y)  f ( x, y) x 3 ( y  y ) 2  x 3 y 2
 lim  lim 
y y 0 y y 0 y

y 2  2 yy  (y) 2  y 2
 x 3 lim  x 3 lim 2 y  y   2 x 3 y .
y 0 y y 0

Geometric interpretation
Let z  f ( x, y) have partial derivatives at ( x0 , y0 ) . If y is a constant, say y  y 0 , the

curve K is the trace of the surface z  f ( x, y) in the plane y  y 0 . If we define the

slope of the secant through the indicated points P( x0 , y0 , f ( x0 , y0 )) and

R( x0  x, y0 , f ( x0  x, y0 )) as

f ( x0  x, y0 )  f ( x0 , y0 )
,
x
we have
f f ( x0  x, y0 )  f ( x0 , y0 )
( x0 , y0 )  lim .
x x 0 x

6
 f
In other words, ( x0 , y0 ) can be interpreted as the slope of the tangent line at point
x
( x0 , y0 , f ( x0 , y0 )) on a curve K of intersection between the surface z  f ( x, y) and a

f
plane y  y0 . In turn, is the slope of the tangent line at a point on a curve of
y
intersection between surface z  f ( x, y) and a plane x  constant .

Example 5.2 Find the slope of the tangent line at (2,1,4) for z  9  x 2  y 2 .
f
 2 x
x
At (2,1,4) the slope is
f
(2,1)  4 .
x
In the plane x  2 , x is a constant and so we find the partial derivative with respect to
y:
f
 2 y
y
At (2,1,4) the slope is
f
(2,1)  2 .
y

Example 5.3 Find the first partial derivatives of the function:

y f y f x
- f ( x, y )  arctan ;  2 ,  2 ;
x x x  y 2 y x  y 2
f f
- f ( x, y)  x y ;  yx y 1 ,  x y ln x ;
x y

- f ( x, y)  a x
2
sin y
.

6. Second-order partial derivatives

f f
For function of two variables z  f ( x, y) , the partial derivatives and are
x y
themselves functions of x and y. Consequently, we can compute second and higher
f
partial derivatives. Indeed, we can find the partial derivative of with respect to y and
x

7
 f
the partial derivative of with respect to x. The latter types of partial derivatives are
y
called mixed partial derivatives.

For z  f ( x, y) :
Second-order partial derivatives:
  f   2 f   f   2 f
   f xx ( x, y) and    f yy ( x, y ) ;
x  x  x 2 y  y  y 2
Mixed second-order partial derivatives:
  f   2 f   f   2 f
   f xy ( x, y ) and    f yx ( x, y ) .
y  x  yx x  y  xy

Higher-order partial derivatives for functions of three or more variables are defined in a
similar manner. The second and third-order partial derivatives are denoted by f xx , f yy ,

f xxx and so on. The subscript notation for mixed second partial derivatives is f xy or

f xy . Note that

  f   2 f   f   2 f
   f xy ( x, y ) and    f yx ( x, y ) .
y  x  yx x  y  xy

Example 6.1 Find the second partial derivatives of the function f if:
- f ( x, y)  e xy ; f xx  y 2 e xy , f yy  x 2 e xy , f xy  f yx  e xy (1  xy ) ;

- f ( x, y)  x 2 y 3 ;

f ( x, y)  x y ;

In our examples f xy ( x, y)  f yx ( x, y) . The order in which a mixed second partial

derivative is done is irrelevant because the following theorem is true.

Theorem 6.1
Let f be a function of two variables. If the first and the second partial derivatives are
continuous on an open region E, then

8
 2f 2f
(x , y )  (x , y ) , for ( x0 , y 0 )  E .
yx 0 0 xy 0 0

Example 6.2 Find the indicated partial derivative of the function:

3f
- f ( x, y)  e ,xy
; e xy ( xy 2  2 y )
x 2 y

5f
- f ( x, y)  x 5 y 3 , . 360 x 2 y
x 2 y 2 x

7. The Chain Rule

Let functions u  u( x, y) and v  v( x, y) be defined on an region D and let the

corresponding pair (u, v) be in the domain of function z  f (u, v) for each pair ( x, y) .
Then
z  f u( x, y), v( x, y)
defines z as a (composite) function of x and y.

Example 7.1 If
z  u 2  u sin v , u( x, y)  xe 2 y and v( x, y)  xy
then
z  x 2e4 y  xe 2 y sin xy .

Theorem 7.1 – The Chain Rule

If z  f (u, v) is differentiable and u  u( x, y) and v  v( x, y) have continuous first

partial derivatives, then composite function z  f u( x, y), v( x, y) has first partial
derivatives

f f u f v f f u f v
  and   .
x u x v x y u y v y

9
 u
Example 7.2 If f (u, v)  ln , u( x, y)  x sin y and v( x, y)  x cos y use the
v 1
f f
Chain Rule to find and .
x y
u v
 sin y  cos y
x x
u v
 x cos y   x sin y
y x

f v  1 1 1 f v  1  1 1
   
u u v 1 u v u ( v  1) 2
v 1
f 1 cos y f x sin y
   cot x 
x x 1  x cos y y 1  x cos y

8. Directional derivatives

Definition 8.1

Let function f ( x, y) be defined on neighbourhood of ( x0 , y0 ) and v  (vx , v y ) be a unit

vector. The directional derivative of f at P( x0 , y0 ) in the direction of v , denoted by

D f ( x0 , y0 ) , is
v

f ( x0  tv x , y0  tv y )  f ( x0 , y0 )
D f ( x0 , y0 )  lim .
v t 0 t

It is easy to see that the first partial derivatives of f are special cases of the directional
 
derivative. As a matter of fact, if v  i , then vx  1, v y  0 , and the limit in the

definition 8.1 reduces to

f ( x0  t , y0 )  f ( x0 , y0 )
D f ( x0 , y0 )  lim  f x ( x, y) .
i t 0 t
 
Similarly, if v  j , then vx  0, v y  1 , and we obtain

f ( x0 , y0  t )  f ( x0 , y0 )
D f ( x0 , y0 )  lim  f y ( x, y) .
j t 0 t

10
The following theorem provides a formula for finding directional derivatives.


Theorem 8.1 If f ( x, y) is a differentiable function of two variables and v  (v x , v y )

is a unit vector, then

f f
D f ( x0 , y0 )  ( x0 , y0 ) v x  ( x0 , y0 )v y .
v x y

Example 8.1 If f ( x, y)  xy , find the directional derivative of f at the point

 1 3  3
P( x0 , y0 )  (1,2) in the direction of the vector v  , . 1
2 2  2

Example 8.2 If f ( x, y)  e x  y , find the directional derivative of f at the point

  2 2
P( x0 , y0 )  (1,1) in the direction of the vector v   , . 2
 2 2 

Example 8.3 If f ( x, y)  3 x 3  y 3  z 3 , find the directional derivative of f at the point

  1 3 5
P( x0 , y0 , z0 )  (3,4,5) in the direction of the vector v   , , .
 3 3 3 

 9  16 3  25 5
108

Geometric interpretation of directional derivative

f f
The partial derivatives and give the slope of a tangent to the trace, or curve of
x y
intersection, of the surface given by z  f ( x, y) and vertical planes which are,
f
respectively, parallel to the x and y coordinate axes. Equivalently, is the rate
x
 f
of change of the function in the direction given by the vector i and is the rate of
y

change of z  f ( x, y) in the j -direction. So, the directional derivative is the rate of

change of the function z  f ( x, y) in the direction given by vector v .

11
 
Let the plane be perpendicular to the xy-plane that contains a vector v and let a curve
K be the trace of the surface z  f ( x, y) . The directional derivative D f ( x0 , y0 ) is
v

the slope of the tangent to K at point with coordinates ( x0 , y0 , f ( x0 , y0 )) .

D f ( x0 , y0 )  tan 
v

Definition 8.2
If f ( x, y) is a function of two variables, then the gradient of f at point ( x0 , y0 ) is a
vector
 f f 
f ( x0 , y0 )   ( x0 , y0 ), ( x0 , y0 ) .
 x y 
The symbol  , an inverted capital Greek delta, is called “del” or “nabla”. The vector
f is usually read “grad f ”.

Example 8.4 If f ( x, y)  x 3  y 3 , find the gradient of f at the point ( x0 , y0 )  (1,1) .

[3,3]

Example 8.5 If  
f ( x, y)  sin  x 2  y 2 , find the gradient of f at the point

( x0 , y0 )  (3,4) .

 3 4 
 5 , 5 

Interpretation
1. The gradient vector f ( x0 , y0 ) points in the direction in which f increases most
rapidly.
2. f ( x0 , y0 ) is perpendicular to the level curve at point P( x0 , y0 ) ( level curve or

contour line - curve defined by f ( x, y)  c for suitable c )

12
Theorem 8.2
f f 
Let , be continuous functions at point P( x0 , y0 ) and let v be the unit vector,
x y
then

D f ( x0 , y0 )  f ( x0 , y0 )  v .
v


Thus, to find the directional derivative of f in the direction of unit vector v , we may

dot the gradient of f with v .

   
(The symbol u v is read “ u dot v ”. The dot product is also referred to as the scalar
 
product or inner product. It is important to note that u v is a scalar, not a vector.)

Example 8.6
The temperature in a rectangular box is approximated by
T ( x, y, z)  10 cos( x  y)  20 sin( x  z) , 0  x, y, z   .

   
If a mosquito is located at  , ,  , in which direction should it fly to cool off as
2 2 2
rapidly as possible?

We have
f f f
 10 sin( x  y)  20 cos( x  z ) ,  10 sin( x  y ) and  20 cos( x  z ) .
x y z
Hence
f      f      f     
 , ,   20 ,  , , 0 and  , ,   20 .
x  2 2 2  y  2 2 2  z  2 2 2 

Thus the gradient of T is  20,0,20.

To cool off most rapidly, the mosquito should fly in the direction of 20,0,20

13
 9. Differentials
Definition 9.1 Let z  f ( x, y) and let x and y be the increments of x and y ,
respectively. The increment f of z  f ( x, y) is
f  f ( x0  x, y0  y)  f ( x0 , y0 ) .

Note that the increment f represents the change in the value of f if ( x0 , y0 ) changes to

( x0  x, y0  y) .

The next definition extends the concept of differentials to functions of two variables.

Definition 9.2 Let z  f ( x, y) and let x and y be increments of x and y .

The differentials dx and dy of the independent variables x and y are
dx  x and dy  y .
The total differential, or simply the differential df of the dependent variable f is
f f
df ( x0 , y0 )  ( x0 , y0 )dx  ( x0 , y0 )dy .
x y

f
Let f ( x, y) have first partial derivatives on some region D and suppose that and
x
f
are continuous at P0 ( x0 , y0 ) in D. If ( x0  dx, y0  dy) is in D and
y
f  f ( x0  x, y0  y)  f ( x0 , y0 ) then

f  df .
Thus
f ( x0  dx, y0  dy )  f ( x0 , y0 )  df ( x0 , y0 ) .

Example 9.1 Use differentials to approximate (1,02) 4 (0,97) 2 .

f ( x, y)  x 4 y 2 P0 ( x0 , y0 )  (1,1)

f f
 4x3 y 2 (1,1)  4
x x
f f
 2x4 y (1,1)  2
y y

14
f (1,1)  1
2 3 8 6
(1.02) 4 (0.97) 2  f ( x0 , y0 )  df ( x0 , y0 )  1  4   2  ( )  1   1,02
100 100 100 100
f  1,018460419344.

Example 9.2 Use differentials to approximate (3,01) 2  (3,98) 2

f ( x, y)  x 2  y 2 P0 ( x0 , y0 )  (3,4)

f x f 3
 (3,4) 
x x2  y 2 x 5

f y f 4
 (3,4) 
y x2  y2 y 5

f (1,4)  9  16  5

3 1 4  2  5
(3,01) 2  (3,98) 2  f ( x0 , y0 )  df ( x0 , y0 )  5      5  4,99
5 100 5  100  500
f  4,9900400…

The concept of differentials can be extended to functions of more than two variables. For
example, suppose that w  f ( x, y, z ) where f is defined on a some region D. Let the
first partial derivatives exist and be continuous functions at ( x0 , y0 , z0 ) . Then

f ( x0  dx, y0  dy, z0  z)  f ( x0 , y0 , z0 )  df ( x0 , y0 , z0 )

f f f
df  dx  dy  dz
x y z

Example 9.3 Use differentials to approximate

0,981,01 .
1,012,01
xy
f ( x, y , z )  z
 x y y z P0 ( x0 , y0 , z 0 )  (1,1,2)
y

11
f (1,1,2)  1
12

15
f yx y 1 f 1  10
 (1,1,2)  2  1
x yz x 1
f f
 x y ln x  y  z  x y ( z ) y  z 1 (1,1,2)  2
y y
f f
 x y y  z ln y (1) (1,1,2)  0
z z
dx  0,02 dy  0,01 dz  0,01

Thus

0,981,01  f ( x0 , y 0 , z 0 )  df ( x0 , y 0 , z 0 ) 
1,012,01
 1  1  (0,02)  (2)  0,01  0  0,01  1  0,02  0,02  0,96
f  0,960400... .

Exact differentials
Definition 9.3
A differential expression
P( x, y)dx  Q( x, y)dy
is said to be an exact differential if there exists a function f such that
df  P( x, y)dx  Q( x, y)dy

Theorem 9.1
Let P and Q be continuous and have continuous first partial derivatives in some region
D. Then
P( x, y)dx  Q( x, y)dy
is an exact differential if and only if
P Q

y x
for all ( x, y) in the region.

Example 9.4 Determine whether the following differential expression is exact.

sin2 𝑦 cos 𝑥 𝑑𝑥 + sin 2𝑦 sin 𝑥 𝑑𝑦

16
 The expression is exact.

10. Extrema for functions of two variables

Definition 10.1
A function f of two variables is said to have a local maximum [local minimum] at
P( x0 , y0 ) if there exists a neighbourhood of P( x0 , y0 ) such that

f ( x0 , y0 )  f ( x, y)  f ( x0 , y0 )  f ( x, y)
for all other pairs ( x, y) in the neighbourhood.
Geometrically, if a surface S is the graph of f, then the local maxima correspond to the
high points on S and the local minima correspond to the low points on the graph of f.

The local minima and maxima are called the local extrema of f.

Theorem 10.1
If
o f has extremum at point P( x0 , y0 )

f f
o there exist and ,
x y
f f
then ( x0 , y 0 )  0 and ( x0 , y 0 )  0 .
x y

Definition 10.2
If z  f ( x, y) has first partial derivatives, then the solutions of
f f
( x, y )  0 and ( x, y )  0
x y
are called critical points.

The critical points correspond to points where f could possibly have a relative
extremum. In some texts critical points are called stationary points.

The following theorem gives sufficient conditions for ascertaining relative extrema.

17
Theorem 10.2 Second Partial Test for Relative Extrema
Suppose that first and second partial derivatives of function f are continuous at some
neighbourhood of point P( x0 , y0 ) and let

f f
o ( x0 , y 0 )  0 , ( x0 , y 0 )  0 ,
x y

2f 2f
( x , y ) ( x0 , y 0 )
x 2 xy
0 0

o D ( x0 , y 0 )  2  0.
 f 2f
( x0 , y 0 ) ( x0 , y 0 )
yx y 2
Then function f ( x, y) has an extremum at point P( x0 , y0 ) .

2f
f ( x0 , y0 ) is a local minimum if ( x0 , y 0 )  0 and f ( x0 , y0 ) is a local
x 2
2f
maximum if ( x0 , y 0 )  0 .
x 2

Remark. If D( x0 , y0 )  0 , then f ( x0 , y0 ) is not an extremum of the function f.

If D( x0 , y0 )  0 , no conclusion concerning a relative extremum can be drawn.

Example 10.1 Find the extrema for f ( x, y)  x3  y 3  3xy .

f min (1,1)  1

Example 10.2 Find the extrema for f ( x, y)  x 3  3xy 2  51x  24 y .

f min (4,1)  152, f max (4,1)  152

18
 11. Constraint extrema
Sometimes one is asked to find the maximum or minimum of a function subject to a
given side condition or constraint. Then the side condition is used to eliminate one of the
variables in the function so that well-known tests for finding extrema were applicable.
So, consider the following examples:

Example 11.1 Determine whether f ( x, y)  x 2  y 2 subject to xy  4 has an

extremum.
(2,2,8) , (2,2,8) - constrained minima

Example 11.2 Determine whether f ( x, y)  9  x 2  y 2 subject to x  y  3 has an

extremum.
3 3 9
 , ,  - constrained maximum
2 2 2

Example 11.3 Find the point on the plane x  2 y  z  1 closest to the origin.

1 1 1
 , , 
6 3 6

19