Econ1050

Calculus 4 - Lagrange
Lecture 9

2
This lecture (Differentiation)

Constrained Optimisation

first order conditions (Lagrange method)
second order conditions (Algebraic method)
meaning of Lagrange multiplier
second order conditions (Matrix method)

3
What does Optimisation mean?
Finding point(s) at which
utility or profit is maximised or
costs are minimised
constrained by
technology available
competitors activity
legal restrictions on
pollution
money (budget)
4
Constrained Optimisation
Maximise
Subject to
m py x p = +
) , ( y x u

x
-
x
-
y
u
y
Find x* and y*
5
Methods
1. Use substitution

But sometimes substitution is not practical
when there are many equality constraints or
the constraint function is complicated, so use

2. Lagrange multipliers
6
Example 1
A firm has 3 factories each producing the
same item. Let x, y and z represent the
respective output quantities from the 3
factories in order to fill an order for 2000 units
in total. Cost functions are
C
1
(x) = 200 + 0.01x
2

C
2
(y) = 200 + y + y
3
/300
C
3
(z) = 200 + 10z
Total cost = C
1
+ C
2
+ C
3

Find output that minimises total cost.
7
Solution 1 Using substitution
z = 2000 x y (constraint)
total C = 200 + 0.01x
2
+

200 + y + y
3
/300
+ 200 + 10(2000 x y)
= 20600 + 0.01x
2
10x + y
3
/300 9y
partial derivatives foc
C
x
= 0.02x 10 = 0 so x = 500
C
y
= 0.01y
2
9 = 0 : y = 30 so z = 1470
soc (and total C = 17920)
C
xx
= 0.02 C
yy
= 0.02y both >0 at (500,30,1470)
C
xy
= 0 hence C
xx
C
yy
(C
xy
)
2
> 0 min point
8
Format
Constrained optimisation problem may be
expressed as an objective function and
constraint:

Minimise C = C
1
(x) + C
2
(y) + C
3
(z)
Objective function
s.t. x + y + z = 2000
Equality constraint
9
Lagrange method
allows critical points to be determined without
substitution
combines the objective function and the
constraint into a single function called the
Lagrange function or Lagrangian
symbol or u or
can find max or min of using usual methods

10
Lagrange function general format
Either
= f(x,y,z) + (c g(x,y,z))
or = f(x,y,z) (g(x,y,z) c)
is called the Lagrange multiplier
is a constant

Important to be consistent about method of setting
out Lagrange function
11
Lagrange method
1. Form = Lagrange function
2. State first order necessary conditions

x
=
y
=
z
=

= 0
3. Solve the above equations
simultaneously to find the critical point
4. Use second order conditions to identify
the type of critical point
12
Example 2

The total revenue function for two goods x and
y is given by TR = 36x 3x
2
+ 56y 4y
2

and the firm is subject to a budget constraint:
5x + 10y = 80

Show that y = 5.5, x = 5 are the quantities of
goods for which first order conditions for
max revenue are satisfied using the
Lagrange method.
13
Solution 2
Problem: Max TR = 36x 3x
2
+ 56y 4y
2

s.t. 5x + 10y = 80
Lagrange fn:
= 36x 3x
2
+ 56y 4y
2

(5x + 10y 80)
First order conditions:

x
= 36 6x 5 = 0 ..(i)

y
= 56 8y 10 = 0 ..(ii)

= 5x 10y + 80 = 0 .(iii)
14
Solution 2 continued
From (i) = (36 6x)/5
From (ii) =(56 8y)/ 10
equate the two expressions for
2(36 6x) = 56 8y
y =1.5x 2
substitute y =1.5x 2 in (iii)
5x +15x 20 80 = 0
x=5 and y = 5.5 and = (36-30)/5=1.2
15
Example 3

Find the values of L and K which satisfy
first order conditions for minimum cost
when labour costs $25 per unit, capital
costs $50 per unit when the production
constraint is 240 units of output and the
production function is Q = 12L
0.5
K
0.5
16
Solution 3
Constrained optimisation problem is
Min Cost = 25L + 50K
s.t 12L
0.5
K
0.5
= 240
Lagrange function:
= 25L + 50K (12L
0.5
K
0.5
240)
First order conditions:

L
= 25 12(0.5L
-0.5
K
0.5
) = 25 6L
-0.5
K
0.5
= 0

.(i)

K
= 50 12(0.5L
0.5
K
-0.5
) = 50 6L
0.5
K
-0.5
= 0

.(ii)

= 240 12L
0.5
K
0.5
= 0 .(iii)
17
Solution 3 continued
From (i) 25 = 6 L
-0.5
K
0.5
so =25 / 6L
-0.5
K
0.5

From (ii) 50 = 6 L
0.5
K
-0.5
so = 50 / 6L
0.5
K
-0.5

Equate: 25 = 50
6L
-0.5
K
0.5
6L
0.5
K
-0.5

L
0.5
K
-0.5
= 2L
-0.5
K
0.5

L = 2K
Sub above into (iii) 240 = 12(2K)
0.5
K
0.5

20 = 2
0.5
K, so K = 102, L = 202

2 3
25
2
1
6
50
K 2
K
6
50
L
K
6
50
K L 6
50
5 . 0 5 . 0 5 . 0
5 . 0 5 . 0
=
|
.
|

\
|
=
|
.
|

\
|
=
|
.
|

\
|
=

and =
18
Second order conditions - Algebraic
Identify convexity or concavity and hence
whether a min or maximum exists at the critical
point (x
0
, y
0
)

For f(x, y) if
f
xx
and f
yy
s 0 f
xx
f
yy
(f
xy
)
2
max
f
xx
and f
yy
0 f
xx
f
yy
(f
xy
)
2
min

- these can be applied to the Lagrange function
19
Example 4
1. Verify that the critical point from example 2
is a maximum

2. Verify that the critical point from example 3
is a minimum
20
Solution 4(1)

x
= 36 6x 5 = 0 ..(i)

y
= 56 8y 10 = 0 ..(ii)

= 5x 10y + 80 = 0 .(iii)
so
xx
= -6
yy


= 8
yx
=0

f
xx
& f
yy
< 0 and f
xx
f
yy
-(f
xy
)
2
0
Hence conditions for maximum are satisfied.

21
Solution 4(2)

L
= 25 6L
-0.5
K
0.5
= 0 .(i)

K
= 50 6L
0.5
K
-0.5
= 0 .(ii)

= 240 12L
0.5
K
0.5
= 0.(iii)

LL
= 3L
-1.5
K
0.5

KK
=3L
0.5
K
-1.5

KL
= 3L
-0.5
K
-0.5
at K = 102; L = 202; and =(25/3)(2)

LL
=325(32)
-1
(202)
-1.5
(102)
0.5
= 0.4419

KK
=325(32)
-1
(202)
0.5
(102)
-1.5
= 1.768

KL
= 325(32)
-1
(202)
-0.5
(102)
-0.5
= 0.8839
22
Continued

LL
and
KK
both > 0

LL

KK
(
KL
)
2
= 0.44191.768 (-0.8836)
2

> 0
Hence conditions for a minimum are satisfied
at K = 102; L = 202
23
Meaning of
or u = f(x,y) (g(x,y) c)
is the coefficient of c which indicates how
much changes when c changes by 1 unit.
is the rate at which the optimal value of the
objective function changes w.r.t. an increase in
the constraint c
24
Example 5 Interpreting
Given the utility function U = 5xy, find the
change in the maximum level of utility when
the budget constraint 5x + y = 30 is increased
by one unit (to 31).





25
Solution (Example 5)


Sub and


15 0 30 = = +
When C changes by 1 unit (from 30 to 31), the
maximum level of utility changes by 15 units.
into
30 5 ( 5 + = y x xy

= = = y y
x
0 5 5
x x
y
5 0 5 = = =
0 30 5 = + = y x

26
Recall matrix approach
Critical point (x
0
, y
0
.

) identified from first
order conditions
2
nd
order conditions for f(x, y.)
|H
1
| < 0; |H
2
|>0; |H
3
| <0.. for max
|H
1
| > 0; |H
2
|> 0; |H
3
| >0.. for min

where the determinants are found at the values of critical point

How can this be extended for constrained
optimisation ?
27
Recall matrix approach
Critical point (x
0
, y
0
.

) identified from first
order conditions
2
nd
order conditions for f(x, y.)
|H
1
| < 0; |H
2
|>0; |H
3
| <0.. for max
|H
1
| > 0; |H
2
|> 0; |H
3
| >0.. for min

where the determinants are found at the values of critical point

How can this be extended for constrained
optimisation ?
28
Second order conditions
Use bordered Hessian matrix, notation
Contains all the second order partial derivatives of
remember that

is equivalent to the constraint, so 2

u u
u u
=
(
(
(

u u u
u u u
u u u
=

0 g g
g
g
H
y x
y yy yx
x xy xx
y x
y yy yx
x xy xx
using u since not
in Word equations
(
(
(

u u
u u =
(
(
(

u u u
u u u
u u u
=

yy yx y
xy xx x
y x
yy yx y
xy xx x
y x
g
g
g g 0
H
30
2
nd
order conditions using H

For max

For min

- evaluated at the critical point


0 H >
0 H <
31
For Example 4 now use matrix
instead of the algebraic method
1. Verify that the critical point is a maximum.
u= 36x 3x
2
+ 56y 4y
2
(5x + 10y 80)

First order conditions:
u
x
= 36 6x 5 = 0 (1)
u
y
= 56 8y 10 = 0 (2)
u

= 80 5x 10y = 0 (3)








= H
(
(
(

=
(
(
(

u u
u u =
8 0 10
0 6 5
10 5 0 0
yy yx y
xy xx x
y x
g
g
g g
H
-5(-40)+10(60) > 0, hence maximum
32
For Example 4 now use matrix
instead of the algebraic method
2. Determine second order conditions for minimum cost
using bordered Hessian matrix.

=25L + 50K (12L
0.5
K
0.5
240 )

L
= 25 6L
-0.5
K
0.5
= 0 .(i)

K
= 50 6L
0.5
K
-0.5
= 0 .(ii)

= 12L
0.5
K
0.5
+ 240 = 0 .(iii)

33
Solution

LL
= 3L
-1.5
K
0.5

KK
=3 L
0.5
K
-1.5

KL
= 3L
-0.5
K
-0.5
At K = 102; L= 202; =(25/3)(2),

LL
= 0.4419
KK
= 1.768
KL
= 0.8839 (from last week)

L
= 6L
-0.5
K
0.5
= -6 (202)
-0.5
(102)
0.5
= 4.243

k
= 6 (202)
0.5
(102)
-0.5
= 8.485




(
(

=
768 . 1 8839 . 0 485 . 8
8839 . 0 4419 . 0 243 . 4
485 . 8 243 . 4 0
H
= H
0
12456 . 4 485 . 8 15 243 . 4
8839 . 0 4419 . 0
485 . 8 243 . 4
485 . 8
768 . 1 8839 . 0
485 . 8 243 . 4
243 . 4
<
=

=
34
Example 6
Problem: Maximise f(x,y) = y e
-x

subject to e
x
+ e
y
= 6
Write the Lagrangian and first order conditions
Show that the critical point is given by x = ln 2
and y = ln 4.
Use the bordered Hessian matrix to prove this
maximises the objective function.
Determine the maximum value of f(x,y).
If the constraint limit changes from 6 to 7 what
effect will this have on the maximum value of
the objective function?

35
Solution (Example 6)
=y e
-x
(e
x
+ e
y
-6)

x
= e
-x
- e
x
=0, so = e
-2x

y
= 1 - e
y
=0, so = e
-y

= -e
x
- e
y
+6 =0 and y=2x
Sub in (iii) e
x
+ e
2x
-6 = 0
e
x
+ (e
x
)
2
-6 = 0
(e
x
-2)(e
x
+3)=0
e
x
=2 (cannot be -3) so x = ln2; y =2ln2 =ln4
= e
-2x
= e
-2ln2
=
36
2
nd
order derivatives:

xx
= -e
-x
e
x
= --.2 = -1

yy
= - e
y
= -.e
ln4
= -1

xy
=0

x
= g
x
= e
x
= 2
y
=g
y
= e
y
= 4







Thus conditions for maximum are satisfied.
Maximum value of objective function
= y-e
-x
= ln4- = 0.88629
When the constraint limit changes from 6 to 7, the maximum
value of the objective function will change by , that is 0.25.
0 ) 4 ( 4 ) 2 ( 2
0 4
1 2
4
1 4
0 2
2
1 0 4
0 1 2
4 2 0
H > + =

=
37
Next week
Revision