Vous êtes sur la page 1sur 6

Optimization Methods: Optimization using Calculus – Kuhn-Tucker Conditions

1

Module - 2 Lecture Notes – 5

Kuhn-Tucker Conditions

Introduction In the previous lecture the optimization of functions of multiple variables subjected to equality constraints using the method of constrained variation and the method of Lagrange multipliers was dealt. In this lecture the Kuhn-Tucker conditions will be discussed with examples for a point to be a local optimum in case of a function subject to inequality constraints.

Kuhn-Tucker Conditions

It was previously established that for both an unconstrained optimization problem and an

optimization problem with an equality constraint the first-order conditions are sufficient for a global optimum when the objective and constraint functions satisfy appropriate

concavity/convexity conditions. The same is true for an optimization problem with inequality constraints.

The Kuhn-Tucker conditions are both necessary and sufficient if the objective function is concave and each constraint is linear or each constraint function is concave, i.e. the problems belong to a class called the convex programming problems.

Consider the following optimization problem:

Minimize f(X) subject to g j (X) 0 for j = 1,2,…,p ; where X = [x 1 x 2

x n ]

Then the Kuhn-Tucker conditions for X * = [x 1 *

x 2 *

x n * ] to be a local minimum are

f

m

g

+

j

= 1

x

i

x

i

λ

j

λ

j

g

g

j

j

λ

j

=

0

i

=

1, 2,

,

n

=

0

j

=

1, 2,

,

m

0

j

=

1, 2,

,

m

0

j

=

1, 2,

,

m

(1)

D Nagesh Kumar, IISc, Bangalore

M2L5

Optimization Methods: Optimization using Calculus – Kuhn-Tucker Conditions

2

In case of minimization problems, if the constraints are of the form g j (X) 0, then

λ

j

have

to be nonpositive in (1) . On the other hand, if the problem is one of maximization with the

constraints in the form g j (X) 0, then

λ

j

have to be nonnegative.

Example (1)

Minimize

f

2

1

= x

+

2

x

2

2

+

3

x

2

3 subject to the constraints

g

g

1

2

x

= −

1

=

x

1

+

x

2

2

x

2

2

x

3

3

x

3

1 2

8

using Kuhn-Tucker conditions.

Solution:

The Kuhn Tucker conditions are given by

a)

∂ ∂

f

+

λ

1

g

1

g

2

x

i

x

i

x

i

+

λ

2

i.e.

b)

λ g = 0

j

j

i.e.

c)

g

j

i.e.,

0

= 0

λ

1

λ

2

(

x

1

(

x

1

+

4

2

x

6

x

3

x

2

2

x

2

x

1

+

λ λ

1

+

2

2

λ

1

2

λ

1

+

2

3

λ

2

λ

2

2

x

3

3

x

3

12)

8)

=

=

D Nagesh Kumar, IISc, Bangalore

=

0

(2)

=

0

(3)

=

0

(4)

0

(5)

0

(6)

M2L5

Optimization Methods: Optimization using Calculus – Kuhn-Tucker Conditions

3

d)

λ

j

i.e.,

0

x

1

x

1

λ

1

λ

2

+

x

2

2

x

0

0

2

2

x

3

3

x

3

− ≤

12

8

From (5) either λ = 0 or,

1

Case 1:

λ = 0

1

x

1

x

2

2x 12 = 0

3

From (2), (3) and (4) we have x 1 = x 2 =

λ / 2 and x 3 =

2

0

(7)

0

(8)

(9)

(10)

λ

2

/ 2 .

Using these in (6) we get

2

λ

2

+ 8λ = 0,λ = 0 or 8

2

2

From (10),

λ

2

0 , therefore,

λ =0, X * = [ 0, 0, 0 ], this solution set satisfies all of (6) to (9)

2

Case 2:

x

1

x

2

2x 12 = 0

3

Using (2), (3) and (4), we have

λλ λλ

1

2

1

2

2

2

λ+ λ

1

3

2

2

4

3

12

=

0

or,

17λ +12λ = −144 . But conditions (9) and (10) give us λ 0 and λ 0 simultaneously,

1

2

1

2

which cannot be possible with 17λ +12λ = −144 .

1

2

Hence the solution set for this optimization problem is X * = [ 0 0 0 ]

D Nagesh Kumar, IISc, Bangalore

M2L5

Optimization Methods: Optimization using Calculus – Kuhn-Tucker Conditions

4

Example (2)

Minimize

f

2

1

= x

g

g

1

2

x

= −

1

=

x

1

+

80

x

2

0

120

2

+ x +

2

0

60

x

1

subject to the constraints

using Kuhn-Tucker conditions.

Solution

The Kuhn Tucker conditions are given by

a)

∂ ∂

f

+

λ

1

g

1

g

2

g

3

x

i

+

λ

2

+

λ

3

∂∂∂

xxx

iii

i.e.

= 0

b)

λ g = 0

j

j

i.e.

2

2

x

1

x

2

+

+

λ (x

1

1

λ

2

(x

1

60

λ

2

+

=

λ λ

1

+

2

0

+

80)

=

0

x

2

120)

c)

g

j

i.e.,

0

d)

λ

j

i.e.,

0

x

1

x

1

80 0

+ x +120 0

2

=

=

0

0

D Nagesh Kumar, IISc, Bangalore

(11)

(12)

(13)

(14)

(15)

(16)

M2L5

Optimization Methods: Optimization using Calculus – Kuhn-Tucker Conditions

5

λ

1

λ

2

0

0

From (13) either λ = 0 or,

1

(x 1 80) = 0

Case 1:

λ = 0

1

From (11) and (12) we have

x

1

= −

2 2
2
2

λ

30 and

x

2

=−

2 2
2
2

λ

Using these in (14) we get

λ

2

(

)

λ 150 = 0

2

Considering

λ= 0

2

, X *

= [ 30, 0].

;

λ = 0 or 150

2

But this solution set violates (15) and (16)

For

λ = −150 , X * = [ 45, 75].

2

But this solution set violates (15) .

Case 2:

(x

1

80) = 0

Using

x = 80 in (11) and (12), we have

1

λ

2

λ

1

=− 2 x

2

=

2

x

2

220

Substitute (19) in (14), we have

2x x 40 = 0

2

2

(

)

.

For this to be true, either x

2

= 0

or x

2

40 = 0

For x

2

= 0 ,

λ = −220 . This solution set violates (15) and (16)

1

D Nagesh Kumar, IISc, Bangalore

(17)

(18)

(19)

M2L5

Optimization Methods: Optimization using Calculus – Kuhn-Tucker Conditions

6

For x 40 = 0 , λ = −140 and λ = −80 . This solution set is satisfying all equations from

2

1

2

(15) to (19) and hence the desired. Therefore, the solution set for this optimization problem

is

X * = [ 80 40 ].

D

Nagesh Kumar, IISc, Bangalore

M2L5