The
purpose of work:
To find the maximum and minimum of function
depending from many variables.
Theoretical information and examples.
Content
1.13
The Extreme Value of a Function of Several Variables
1.13.1 Definition of an extreme value of a function
1.13.2 Necessary conditions for an extreme value
1.13.3 Sufficient conditions for an extreme value
1.13.4 The case of a function of many variables
1.13.5 Conditional extremum
1.13.6 Largest and smallest values of a function
1.14 Finding the Greatest and Smallest Values of Functions
1.13 The
Extreme Value of a Function of Several Variables
1.13.1 Definition
of an extreme value of a function. We say that a
function f(x, y) has a maximum
(minimum) f(a, b) at the point P(a, b),
if for all points P'(x, y) other than P in a
sufficiently small neighbourhood of P one has the inequality f(a,b)
>f(x, y) [i.e., f(a, b) < f(x,
y)]. The generic term for maximum and minimum of a function is extreme. In a similar fashion, we define the
extreme value of a function of three or more variables.
1.13.2 Necessary
conditions for an extreme value. The points at which a
differentiable function f(x, y) may attain an extremum
(so-called stationary points) are
found by solving the system of equations:
(1)
(necessary conditions
for an extremum). System (1) is equivalent to a single equation, df(x,y)
= 0. In the general case, at the point of the extreme value P(a,
b) of the function f(x, y), df(a, b)
= 0 or df(a,b) does not exist.
1.13.3 Sufficient
conditions for an extreme value. Let P(a,
b) be a stationary point of the function f(x, y),
i.e., df(a,b)=0. Then:
a) if d²f(a,
b) < 0 for dx² + dy² > 0, then f(a,
b) is a maximum of the function f(x, y),
b) if d²f(a, b) > 0 for dx²
+ dy² >0, then f(a, b) is a minimum
of the function f(x, y),
c) if d²f(a, b) changes sign, then f(a,
b) is not an extreme value of f(x, y).
These conditions are equivalent to the
following: Let f 'x(a, b)=f 'y(a,b)=0
and A=f''xx(a,b),B=f''xy(a,b),
C = f''yy(a,b).
We
form the discriminant
.
Then,
1) if
, the function has an extremum at the point P(a,
b), namely a maximum,
if A
< 0 (or C < 0), and a minimum, if A> 0 (or C
> 0)
2) if
, then there is no extremum at P(a, b),
3) if
, the question of an extremum of the function at P(a,
b) remains open (i.e., it requires a further investigation).
1.13.4 The case
of a function of many variables. For a function of
three or more variables, the necessary conditions for the existence of an
extremum are similar to Conditions (1), while the sufficient conditions are
analogous to Conditions a), b), and c) of 1.13.3.
Example 1. Test for an extremum the function
![]()
Solution. Find the partial
derivatives and form a system of Equations (1):

or

Solving the system, we get the
four stationary points:
![]()
We find now the second
derivatives

and form the discriminant
for each stationary point.
1)
For the point P1:

![]()
Thus, there is no extremum at
this point.
2)
For the point P2:
![]()
The function has a minimum which is equal to
the value of the function for
:
![]()
3)
For the point P3:
![]()
There is no extreme value.
4)
For the point P4:
![]()
At the
point P4, the function has the maximum
![]()
1.13.5
Conditional extremum. In the simplest case, the conditional
extremum of a function f(x, y) is a maximum or a
minimum of this function which is attained on the condition that its arguments
are related by the equation j(x,y) = 0 (coupling equation).
In order to find the conditional extremum of a function f(x, y),
given the relationship j(x, y) = 0,
we form the so-called Lagrange function
![]()
where l is an
undetermined multiplier, and we seek the ordinary extremum of this auxiliary
function. The necessary conditions for the extremum reduce to the system of
three equations

with the three unknowns x,
y, l, from which it is,
generally speaking, possible to determine these unknowns.
The question of the existence and character of
a conditional extremurn is solved on the basis of a study of the sign of the
second differential of the Lagrange function

for the given system of values
of x, y, l obtained from (2) under the condition that dx and dy are
interrelated by the equation

In fact, the function f(x, y)
has a conditional maximum, if d²F
< 0, and a conditional minimum, if d²F> 0. As
a particular case, if the discriminant D of the function F(x, y) at a
stationary point is positive, then there is at this point a conditional maximum
of the function f(x, y), if A < 0 (or C
< 0), a conditional minimum, if A > 0 (or C >
0).
In a similar fashion, we find the conditional
extremurn of a function of three or more variables, provided there are one or
several coupling equations (the number of which, however, must be less than the
number of variables). Here, we have to introduce into Lagrange's function as
many undetermined multipliers as there are coupling equations.
Example 2. Find the extremum
of the function
![]()
provided the variables x
and y satisfy the equation
![]()
Solution. Geometrically speaking,
the problem reduces to finding the largest and smallest values of the z-co-ordinate
of the plane
![]()
for points of its intersection
with the cylinder
We
form the Lagrange function
![]()
We
have

The necessary conditions yield the system of
equations:

Solving this system, we find
![]()
Since

it follows that
![]()
If l = 5/2, x = 4/5 and y = 3/5, then d²F
> 0 and, consequently, the function has a conditional minimum at this
point. If l=-5/2, x =
-4/5 and y = -3/5, then d²F < 0, and,
consequently, the function has at this point a conditional maximum.
Thus,
![]()
![]()
1.13.6 Largest and smallest values of a
function. A function, which is
differentiable in a limited closed region, attains its largest (smallest) value
either at a stationary point or at a point of the boundary of the region.
Example 3. Determine the
largest and smallest values of the function

![]()
in the region ![]()
Solution. This region is a
triangle (Fig. 7).
1) We find first the stationary points:
![]()
![]()
whence x= -1, y=
-1 and get the point M(-1, -1).
At M, the value of the function zM
= -1. It is not absolutely necessary to test for an extremum.
2) Investigate the function on the boundaries
of the region. When x = 0, z = y² + y and the problem
reduces to seeking the largest and smallest values of this function of one
argument on the interval
We find that z(larger)x=0
= 6 at the point (0, -3), z(smaller)y=0 =
-1/4 at the point (-1/2, 0).
When y = 0, we get z = x2+x.
Similarly, we find that z(larger)y=0 = 6 at
the point
, z(smaller)y=0=-1/4
at the point (0,-1/2)
When x + y = - 3 or y = -3 -
x, we will have z = 3x² + 9x + 6.
Similarly, we find that z(smaller)x+y=-3=-3/4 at the
point (-3/2, -3/2), z(larger)x+y=-3 coincides with z(larger)x=0
and z(larger)y=0. On the straight line x+y
= -3, we could test the function for a condilional extreme value without
reduction to a function of a single argument.
3) Correlating all the values of the function
z obtained , we conclude that z(larger) = 6 at
the points (0,-3) and (-3,0), z(smaller) = -1 at the
stationary point M.
1.14 Finding the
Greatest and Smallest Values of Functions.
Example 1. Break up a
positive number a into three non- negative numbers so that their product
should have the largest possible value.
Solution. Let the required
numbers be x, y, a - x - y. Seek the maximum of the
function f(x, y) = xy(a - x - y).
The function f(x, y) is to
be considered inside a closed triangle
(Fig. 8)
Solving the system of equations
![]()
![]()
Fig.8
we will have the unique
stationary point (a/3, a/3) for the Interior of the
triangle. Let us test the sufficiency conditions. We have
![]()
Consequently,

Thus, at (a/3, a/3), the
function has a maximum. Since f(x, y)=0
on the border of the triangle, this maximum will be largest, i.e., the product
will be largest if
and the largest
value is a³/27.
Note. The problem can also be solved by the methods of a
conditional extremum by seeking the maximum of the function u=xyz
subject to the condition x + y + z = a.
Control
questions
1.
Give the definition of an extreme value
of a function.
2. What are the necessary conditions
for an extreme value?
3.
What are the sufficient
conditions for an extreme value?
4. Where the function, which is
differentiable in a limited closed region, attains its largest (smallest)
value?