Escolar Documentos
Profissional Documentos
Cultura Documentos
=
=
, a a scalar.
1.6. Let ( )
1 2
, f x x be a concave function, and assume ( )
2
1 2
, x x D R e c .
a. Must a global maximum exist in D for ( )
1 2
, f x x ?
b. If we find
( )
* *
1 2
, x x is a local maximum, can we conclude it is a global maximum?
c. Define a new function: ( ) ( ) ( )
1 2 1 2
, , g x x H f x x where : H R R . Suppose H is differentiable
and 0
dH
df
> everywhere. Is the function ( )
1 2
, g x x necessarily concave?
d. Same assumptions as (b) and (c). Will
( )
* *
1 2
, x x be a local maximum for ( )
1 2
, g x x ? Will it be a
global maximum? Explain your answer.
1.7. Consider the problem of
( ) ( )( ) ( ) { }
, ,
. . 1 ; 2 10; , , 0
x y z
Max xyz s t x y z w x y x y z + + s + s >
a. Set up the Lagrangean function and find the FOC.
b. Solve for the optimal values of the variables { } , , x y z as well as for the Lagrangean multipliers for
each constraint. Does each constraint always bind? Relate your answers to the value of w.
1.8. Consider the problem of
( ) ( ) { }
, ,
ln ln . . 10; , 0
x y z
Max x y s t x y x y + + s > :
a. What is the solution to this constrained optimization problem? Find the solution using the
Lagrangean function and the Kuhn-Tucker (K-T) conditions. Is there a unique solution to the K-T
conditions?
b. Suppose the domain is rewritten as: ( )
3
10 0 x y > . Does this change the domain of the
original problem or the point in the domain that maximizes the objection function?
i. Write the Langrangean as: ( ) ( ) ( )
3
ln 10 L ln x y x y = + +
If
( )
* *
, x y solves the constrained optimization problem, can you find a
*
so that
( )
* * *
, , x y
solve the Kuhn-Tucker conditions? Explain your answer.
c. Leave the domain as originally given, but write the objective function as:
( ) ( ) ( ) ( ) ( )
3
, ln ln 2ln 5 f x y x y = + . Is this function concave? How does the point
that solves the constrained optimization problem compare to the solution to (a)?
i. If you write the Lagrangean function, does every ( ) , , x y ' ' ' that solves the Kuhn-Tucker
conditions represent a maximum to the constrained optimization problem? If not, why not?