MODULE ONE

UNCONSTRAINED-OPTIMIZATION

OBJECTIVES:

At the end of lectures on this topic, students should be able to:

1. State and apply the first and second order condition for unconstrained optimization

in a function of one independent variable.

2. State and apply the first and second order conditions for unconstrained optimization

in a function of several independent variables.

3. Apply the concept of unconstrained optimization in practical economics and

business problems.

A. FUNCTIONS OF ONE INDEPENDENT VARIABLE.

1. INTRODUCTION

Let y = f(x) be a continuous function with finite and continuous derivatives: then if f(x)
increases as x increases, y = f(x) is an increasing function, while it is a decreasing
function otherwise.
Let the figure below represents the graph of y = f(x) for positive values of x in a 2-

dimensional space:

FIGURE 1.0

B D F

A C

0 x1 x2 x3 x4 x5

1
 Point B, C, D and E are known as stationary points at each of those points f (x) = 0.

They can also be referred to as critical point or turning points. Note that before and

after the turning points, the slope of the function changes. At point B, for instance, the

slope changes from positive at the left to negative on the right

Point B, D represent the peak, while C and E are the troughs, point B and D are the

relative or local maximum. At each of these points the function changes from

increasing to decreasing. At C and E the function changes from decreasing to

increasing. Points C and E also represent the local or relative minimum.

In general, a function of one independent variable y = f(x) has a relative or local

maximum at the point X=X3 if (x3) > f(x3 + x ) where x is a small change in x. this

implies that f (x3) > f(x) for all values of x in a small neighborhood about point x.

Similarly, f(x) has a minimum value at x = x4 if f(x4) < f(x4 + x ) where x is a small

change in x. that is, the function f(x) is a relative minimum at the point x = x4 if f(x4) <

f(x) for all values of x in a small neighbourhood about x4.

II. ABSOLUTE OR GLOBAL MAXIMUM AND MINIMUM VALUES

A function y = f(x) in a defined domain is said to have absolute or global maximum at the

point x = a if f(a) > f(x) for all values of x in a defined domain. Conversely, f(x) has an

absolute or global minimum, at the point x = c if f(c) < f(x) for all values of x in its defined

domain.

For example, I the figure above, y =f(x) in the domain 0  x  x5 , the point B at which

2
x =x1 and f(x) = f(x) is the global maximum. Similarly, point C is the absolute minimum.

Example1:

Given y =f(x) = 4x-x2. 0  x  x3 ; plot the graph and show the absolute minimum and

maximum points.

Solution:

X 0 1 2 3 -1

Y 0 3 4 3 -5

0 2

The origin where x = 0 and f(0) = 0 is the point of absolute minimum for x  0,2

The point where x = 2, and f(2) = 4, is the point of local and global maximum.

III. ALLIED OR DERIVATIVE CONDITIONS FOR RELATIVE OPTIMA

The conditions must be met to establish relative optima. : ( d 1 ” f   )

a. The first order condition (FOC) or the Necessary condition:

3
 b. The second-order condition (SOC) or the sufficient condition.

Given y = f(x); the FOC is that f’ (x) = 0 critical points

The SOC is that (i) f” (x) < 0 for a maximum and (ii) f” (x) > 0 for a minimum

f’(x) =0 h’(x) <0

decreasing
increasing
decreasing h’(x)
>0
f’(x) <0
increasing
f’(x) >0 h’(x) =
0

Example 2: Find the relative optima of the function y = 4x –x2.

Solution: The FOC for relative optima is: f’ (x) = 0.

For y = 4x –x2; y’ = 4 – 2x and as such, FOC required 4 – 2x = 0.

 2x = 4: and x = 2. :(  ’ “ f   )

The SOC is that: y “> 0 for a minimum and <0 for a maximum

Form the given function; = -2 < 0.

 y = 4x – x2 has a maximum stationary value at x = 2, where y = 4(2) - 22 = 4.

Example 3: Let the cost function for ABC manufacturing Company be y = 40 – 6x + x2.

Find the extreme values.

Solution: FOC: y’ = -6 + 2x = 0.  x = 3.

4
SOC: y” = 2 > 0;  the curve is convex downwards and as such the function has a
minimum point at x = 3, where y is 40 – 6(3) + 32 = 31.

Example 4: Examine the function y = 2x – 3x2 for maximum or minimum.

Solution: FOC: y’ = 2 – 6x = 0;.  x = 1/3.

SOC: y “ = -6 < 0. Thus is a maximum at x = 1/3 and its value is y = 2(1/3)-3(1/3)2 = 2/3 -
1/3 = 1/3.

Example 5: Find the stationary (or turning) points and the stationary values of the function

f(x) = x3 + 3x2 -9x – 5.

Solution: Given f(x) = x3 + 3x3 – 9x – 5, FOC for turning point requires f’ (x) = 0.

Thus FOC for turning point of f(x) is: f’ (x) = 3x2 + 6x – 9 = 0.

 = 3x2 + 6x – 9 = 0 dividing through by 3).

Factorizing we have: (x + 3)(x-1) = 0; )  x = -3 or 1.

SOC: f” (x) = 6x + 6. Evaluating f” (x) at x = -3 and 1 we have:

a) 6(-3) + 6 = -12 < 0  x = 3 is a maximum point.

b) 6(1) + 6 = 12 > 0  x = 1, is a minimum point.

Thus f(x) attains a maximum value at x = -3 where y = (-3)3 + 3(-3)2-9(3) -5 = 22; and

minimum value at x = 1 where y = (1)3 + 3(1) – 9(1) – 5 = 10.

POINT OF INFLEXION

Let us examine what happens if, given by = f (x), both f’(x) equal and f” (x) equal zero x

=c

a) f’ (x) = 0 implies that a stationary point exists at x = c, and

5
 b) f” (x) = 0, implies that such a stationary point is neither a maximum nor a minimum

such a turning point is called a point of inflexion as shown in the figure below:

f’(x) <0 f’(x) >0

f’(x) =0 f’(x) =0

f’(x) <0
f’(x) <0

B. FUNCTIONS OF SEVERAL VARIABLES

A function Z = f(x, y) having 2 independent variables is said to have a relative optimum at

the point x = x0 and y =y0 if f(x0, y0) is greater or less than f(x, y) for all values of x and y

in the neighbourhood of x = x0 and y = y0.

Thus f (x, y) has a maximum value at a point (x, y) = (x0, y0) if:

f (x0 + x , y0 + y ) < f(x0, y0), where x and y are small arbitrary quantities.

Similarly, f(x, y) has a minimum point at (x, y) = (x0, y0) if:

f (x0 + x , y0 + y ) < f(x0, y0), ), where x and y are small arbitrary changes in x and y

about x0 and y0 respectively.

For a stationary value to exist for z = f (x, y); the FOC is that fx = fy = 0; implying that the

total differential dz = fxdx + fydy = 0 at the point x = x0 and y = y0.

The SOC is that:

a) d2z < 0 at (x0, y0) for maximum, for any value of dx and dy (not both zero)

b) d2z>0 at (x0, y0) for minimum, for any value of dx and dy (not both zero).

6
Note that for a function z = f(x, y);

dz = fxdx + fydy; and

d2z =  /  x(fxdx + fydy) dx +  /  y (fxdx +fydy)dy

 d2z =  /  x (fxdx + fydydx) +  /  x (fxdxdy +fydy2): where dx2 = (dx)2 and dy2 = (dy)2

 d2z =fxxdx2 + fyxdydx + fxydxdy + fyydy2 (since  /  x = fx and  /  y = fy)

But by Young’s theorem fyx = fxy and as such:

d2z = fxxdx2 + 2fyxdydx + fyydy2

Thus, for SOC to be satisfied for a maximum: d2z < 0  fxx< 0; fyy<0 and fyyfxx> f2 xy

Similarly for SOC to be satisfied for a minimum: d2z < 0  fxx< 0; fyy<0 and fyyfxx> f2 xy

Also, for SOC to be satisfied for a saddle point (i.e. a point that is neither maximum nor

minimum but just a turning point, like the point of inflexion in a single independent

variable case), fyyfxx < f2 xy

In summary we have the following:

1. Maximum: FOC fx=fy =0

SOC d2z < 0  fxx< 0; fyy<0 and fyyfxx> f2 xy
2. Minimum FOC fx=fy =0
SOC d2z < 0  fxx< 0; fyy<0 and fyyfxx> f2 xy
3. Saddle Point FOC fx=fy =0

SOC fyyfxx < f2 xy

4. No Information FOC fx=fy =0

SOC fyyfxx = f2 xy

Example 6: Find the relative optimum value(s) of the function

Z = f(x. y) = x2 + y2 + 4x – 8y.

Solution: The stationary value of z = f (x, y) requires the following:

7
FOC: fx = fy = 0

fx = 2x + 4 = 0; and fy = 2y – 8 = 0

 x = -2 and y = 4 as the point at which stationary value is obtained.

To test for SOC, we have obtain the 4 second-order partial derivatives of z = f(x, y) as

follows: fxx = 2: fyy=2; fxy = fyx = 0

Since fxx = 2> 0: fyy = 2>0; and fyyfxx = 4> f2 xy= 0; it then follows that the point

(x = -2, y =4) gives a minimum point for the given f(x, y).

Using the values of x, and y in z=f(x, y) we have:

Z = (-2)2 + 42 + 4(-2) -8 (4) = -20.

The minimum point is obtained at (x, y, z) = (-2, 4, -20).

Example7: Find the extreme values of z = f(x, y) = x3 + y3 -3x -27y + 24

Solution: The stationary value of z =f(x, y) requires the following:

FOC: fx = fy = 0

fx = 3x2 – 3 = 0; x2-1=0  x = 1 or -1

fy = 3Y2 – 27 = 0; Y2 – 9 = 0  Y = 3 or -3

Thus, (x, y) = (1, 3); (1, -3); (-1, 3); (-1,-3)

SOC: fxx = 6x; fyy = 6y; fxy = fyx =0

a). When (x, y) = (1, 3): fxx = 6(1) = 6>0; 6(3) = 18>0; fxxfyy = 6(18) = 84 > 0; And f2 xy =

Thus Z satisfies the condition for a Minimum at (1, 3), since fxx >0; fyy > 0 and fyyfxx > f2xy

b). When (x, y) = (1, -3): fxx = (-1) = -6 < 0; fyy = 6(3) = 18 < 0; fxxfyy = 6(-18) = -10 < 0:
and f2xy = 0. We have a saddle point at (x, y) = (1, -3) since fxxfyy >f2xy
c) When (x, y) = (-1, 3): fxx = (-1) = -6 < 0; fyy = 6(3) = 18 > 0; fxxfyy = -6(18) =-84 < 0;

8
And f2xy= 0. we have a saddle point at (x, y) = (-1, 3) Since fxxfyy < f2xy.

d) When (x, y) = (-1,-3): fxx = 6 (-1) = -6 < 0; fyy = 6(-3) = -18, 0; fxxfyy= -6(-18) = 84> 0;

And f2xy = 0. We have a maximum point at (x, y), since fxx, 0; fyy<0 and fyyfxx > f2xy.

C. THE SECOND-ORDER CONDITION: THE HESSIAN APPROACH

The Hessian Matrix approach can be used to determine whether or not the SOC is fulfilled.

This is obtained by:

a) Obtaining the Hessian matrix (H) from the given function. The elements of H are

the second order partial derivatives derived from the function when differentiated

with respect to its independent variables and evaluated at the relevant points.

b) Obtaining the determinant of H i.e. /H/

and evaluating the signs of its principal minors

In the case of 2-independent fxx fxy

Variables x, y; /H/ = fyx fyy

IF the first element on the principal diagonal, i.e. the first principal minor, H1 = fxx is

positive, and the second principal minor, H2 = /H/ = fxx fxy

fyx fyy = fxxfyy – f2 xy > 0,

then the second-order condition for a minimum are met. In other words, the Hessian is

positive definite and meets the SOC for a minimum.

In the first principal minor, H1 = fxx < 0 and H2 > 0, the Hessian is negative definite,

and thus fulfill the SOC for a maximum.

In summary, in terms of Hessian analysis, the function Z = f(x, y) attains:

9
 a) Maximum when fx = fy = 0, and the principal minor of the Hessian

determinant alternate in sign, beginning with a negative sign.

b) Minimum, when fx = fy = 0, and the principal minor of the Hessian

determinant are all positive.

Example 8: Given that Z = f(x, y) = 6x2 – 9x – 3xy – 7y + 5y2, determine the critical values

and test to see that the function attains a minimum or maximum.

Solution: The stationary values of z = f(x, y) requires the following:

FOC: fx = fy = 0

fx = 12x – 9 - 3y = 0  12x – 3y = 9

fy = 10y – 3x -7 = 0;  -3x + 10y = 7

Thus (x, y) = (1, 1)

SOC: fxx = 12; fyy = 10; fxy = fyx = -3

Thus /H/ = fxx fxy 12 -3

fyx fyy -3 10

Taking the principal minors /H/ = 12 > 0, /H2/ = 12(10) - (-3) (-3) = 111>0
The Hessian is positive definite and thus Z is minimized at the critical value (1, 1).

Assignment: Rework Example 7, using the Hessian determinant approach.

D. APPLICATIONS OF UNCONSTRAINED OPTIMISATION

The major significance of optimization in economic and business analysis is towards

achieving certain desired objectives: maximization of utility, profit, and revenue; as well

as minimization of cost and losses, etc. We shall illustrate these with some examples.

10
Example 9: Consider a monopolist, who faces a linear demand function: P= 100-2q.

His total Revenue function is given by R = Pq = (100-2q) q = lOOq - 2q2. Calculate the

revenue maximizing level of output.

Solution:

The revenue maximizing level of output is obtained at the point where dR/dq = 0 (FOC)

and d2R/dq2 <0 (SOC).

DR/dq or R' =100-4q =0. This implies q = 25.

To test whether q=25 is maximum we use the sac: R" = -4<0. Since the sac for maximum

is satisfied, then q=25 maximises revenue.

The maximum revenue level is: R = 100(25) - 2(25)2 = Nl, 250.

Example 10: a) Find the rectangle or largest area whose perimeter is 12 meters.

b.) The Sum of 2 numbers is equal to 6. For which choice(s) of these numbers is their

Product a maximum?

Solution: These are optimization problems.

a). We are required to Maximise Area = Length * Breadth  A = L*B

Note that: Perimeter = 2(L +B) = 12  L + B = 6; Thus B = 6 - L

Expressing A as a function of L alone, we have: A = LB = L (6-L) = 6L - L2

FOC: dA/dL = 6-2L = 0  L = 3 and B = 6- L = 3.

SOC: d2 A/dL2 = -2 <0. This ensures that L=3, b=3 guarantees the maximum required.

b.) Let Land B be the two numbers. We are to maximize A = L *8, but with the additional

requirement that L + B = 6  B = 6-L

11
Expressing A as a function of one variable, we have:

A = L *B = L (6-L) = 6L-L*L

FOC: dA/dL = A' = 6-2L = 0 =>L = 3

SOC: d2 A/dL2 = -2 <0. This guarantees that L=3, b=3 maximum product.

Example 11: A firm has the following Total Cost and Demand Functions:

C= 1/3X3 -7X2 + 111 X-50; and X = 100 - P

a) Write out the total revenue function R

b. Derive the Total Profit,  , in terms of X

c) Determine the profit maximizing level of output

d) What is the maximum profit?

Solution:

a). C =0 1/3X3 -7X2 + l11X - 50

X = 100-P  P = 100 – X2

b).The profit  is the difference between R and C, i.e.  = R – C

 = 1OOX- X2 = (1/3X3 -7X2 + l11X - 50)

= 1OOX - X2 - 1/3X3 + 7X2 - 111 X + 50

= 6X2 -l/3X3 -IIX + 50

c) (Students should solve this and (d) on their own.

Example12:

Let the total cost function of a firm be Z = f(x,y) = 2X2 - xy + 2y2- 6x + 6y + 50; find the

Critical values of x and y and test if the stationary values of the function is maximum or

12
minimum at the critical values of the variables.

Solution:

Z =2X2 - xy + 2y2- 6x - 6y + 50

FOC: fx =fy = 0

Thus: fx = 4x – y – 6 = 0 (1)

fy = -x + 4y – 6 = 0 (2)

Solving equations (1) and (2) simultaneously, we have: y = 2, x = 2

Thus. a critical point exists at (x, y, Z) = (2,2, 22)

To test for sac, we take the second partial derivatives and check the signs of the principal
minors of the ensuing Hessian determinant.
Thus: fxx = 4; fyy = fyx = -1.

H = fxx fyy = 4 -1

fyx fyy -1 4

H1 = 4>0; H2 = 16-1= 15 >0. Thus, H is positive definite and as such Z is minimum

Practice Exercises

1. Compute the FOC and sac to investigate the turning points of the following functions:

(a) 2x - 1 + 8/x (b) x2 - 3 x-I

2. Obtain 2 positive numbers whose sum is 20 and whose product is as large as possible

3. Find the stationary points of the following functions and discriminate between them:

a) x3 -9x2 + 15x + 11 b) (4 - x)3/(2-x);

4. A monopolist has the following linear demand function: P = 100 - 4q; and the total cost

function: C = 50 + 20 q, where P= price, q=quantity of output and C = total cost. You

are required to find:

13
a) The profit-maximising level of output

b) Maximum profit,

c) Price per unit of at which his profit-maximising output is sold.

14