American J ournal of I ndustrial and Business Management, 2012, *, **-**

doi:10.4236/ajibm.2012.***** Published Online *** 2012 (http://www.scirp.org/journal/ajibm)

Copyright 2012 Sces. AJ I BM
Sufficient Fritz J ohn Type Optimality Criteria and Duality
for Control Problems
I. Husain and Santosh K. Srivastav
Department of Mathematics,
J aypee University of Engineering and Technology, Guna (M.P), India
ihusain11@yahoo.com, santosh_00147@rediffmail.com

ABSTRACT
Sufficient Fritz John optimality conditions are obtained for a control problem in which objective
functional is pseudoconvex and constraint functions are quasiconvex or semi-strictly quasiconvex.
A dual to the control problem is formulated using Fritz John type optimality criteria instead of
Karush-KuhnTucker optimality criteria and hence does not require a regularity condition. Various
duality results amongst the control problem and its proposed dual are validated under suitable
generalized convexity requirements. The relationship of our duality results to those of a nonlinear
programming problem is also briefly outlined.
Keywords: Sufficient Optimality criteria; Control problem; Fritz John type dual, Nonlinear pro-
gramming problem, Generalized convexity.

AMS Mathematics Subject Classification: Primary 90C30, Secondary 90C11, 90C20, 90C26.

1. Introduction
Optimal control models represent a variety of common
situations, notably, advertising investment, production
and inventory, epidemic, control of a rocket, etc. The
optimal planning of a river system which is an invincible
resource of nature, where it is needed to make the best
use of the water, can also be modelled as an optimal con-
trol problem. Optimal control models are also potentially
applicable to economic planning and to the world models
of the Limits to Growth kind in general.
Optimality criteria for any optimization problem are of
great significance and lay the foundation of the concept
of duality. Fritz John optimality criteria for a control
problem were first derived by Berkovitz [1]. Subse-
quently Mond and Hanson [2], who first investigated
duality in optimal control pointed out that from Fritz
John optimal criteria, Karush- Kuhn Tucker optimality
criteria can be deduced if normality of the solution of a
control problem which replaces a regularity conditions is
assumed. Later, treating a nondifferentiable control
problem as a nondifferentiable mathematical program-
ming problem in an infinite-dimensional space, Chandra
et al [3], obtained Fritz John as well as Ka-
rush-Kuhn-Tucker optimality criteria.
For a nondifferentiable control problem Using
Karush-KuhnTucker optimality criteria, they
formulated Wolfe type dual and derived usual duality
results under appropriate convexity assumptions.
In this research exposition, sufficient Fritz John criteria
are derived for a differentiable control problem in which
objective functional is pseudoconvex and constraint
functions are quasiconvex or semi-strictly pseudoconvex.
A number of duality results are proved for relating the
solution of the control problem with that of its proposed
dual under suitable generalized convexity requirements
.The relationship of our duality results to those of a non-
linear programming problem is indicated.

Sufficient Fritz J ohn Type Optimality Criteria and Duality for Control Problems

Copyright 2011 SciRes. POS
2. Control Problem and Related Preliminaries
Let
n
R denotes a n-dimensional Euclidean space, I=[a, b]
be a real interval and
n m
f : I R R R be a conti-
nuously differentiable with respect to each of its argu-
ments. For the function ( ) f t,x,u where
n
x: I R is
differentiable with its derivativex and
m
u: I R is the
smooth function, denote the partial derivatives of f by
t
f
,
x
f and
u
f , where
, ,..., , ,..., , :
1 1
T
m u
T
n x t
u
f
u
f
f
x
f
x
f
f
t
f
f |
.
|

\
|
c
c
c
c
= |
.
|

\
|
c
c
c
c
=
c
c

. ) ,..., ( ) ,..., (
1
1 T
m
T n
u u u and x x x = =
For m-dimensional vector function ( ) g t,x,u the gredient
with respect toxis
,
,...,
,....,
1
1
1
|
|
|
|
|
|
|
.
|

\
|
c
c
c
c


c
c
c
c
=
n
p
n
n
p
x
x
g
x
g
x
g
x
g
g
a n p matrix of first
order derivatives.
Here ( ) u t is the control variable and ( ) x t is the state
variable, u is related to x via the state equation
( ) x h t,x,u = .Gradients with respect to uare defined
analogously.
A control problem is to transfer the state vector from an
initial state ( ) x a =

to a final state ( ) x b = so as to
minimize a functional, subject to constraints on the con-
trol and state variables.

A control problem can be stated formally as,
(CP):
b
x X,u U
a
Minimize f (t,x,u)dt,
e e
}

subject to
( ) ( ) ( ) 1 x a ,x b = =


( ) ( ) 2 h t,x,u x,t I = e
( ) ( ) 0 3 g t,x,u ,t I s e

( ) i f is as before,
n m p
g: I R R R and
n m n
h: I R R R are continuously differentiable
functions with respect to each of its arguments.

(ii) X is the space of continuously differentiable state
functions
n
x: I R such that ( ) x a , = ( ) x b =
equipped with the norm x x Dx

= + , and uis the
space of piecewise continuous control functions
m
u: I R has the uniform norm .

and
(iii) The differential equation (2) for xwith the initial
conditions expressed as
t
a
x(t ) x( a) h( s,x( s),u( s))ds,t I , = + e
}

may be written
as
( ) Dx H x,u , = where the mapH : X U
( ) ( )
n n
C I ,R ,C I ,R being the space of continuous func-
tions from
n
I R , defined by ( ) H x,u =
( ) ( ) ( )
h t,x t ,u t .

Following Craven [4], the control problem can be
expressed as,
(ECP): ( )
x X,u U
Minimize F x,u
e e

subject to
( ) Dx H x,u , =

( ) G x,u S, e
WhereG is function from X U into
( )
p
C I ,R given
by
( ) ( ) ( ) ( ) ( ) G x,u t g t,x t ,u t = fromx X,u U e e , and
t I e ; S is the convex cone of functions in
( )
p
C I ,R

whose components are non-negative; thus S has inte-
rior points.
Necessary optimality conditions for existence of exter-
mal solution for a variational problem subject to both
equality and inequality constraints were given by valen-
tine [5]. Invoking Valentines [5] results, Berkovitz [1]
obtained corresponding necessary optimality criteria for
the above control problem (CP). Here we state the Fritz
John type optimality conditions derived by Chandra et. al
[3] in of the following proposition which will be re-
quired in the sequel.

Proposition1:
(Necessary optimality conditions):

If ( ) x,u X U e an optimal solution of (CP) and the
Frechet derivatives ( ) ( ) ( )
'
x u
Q D H x,u H x,u =

is
surjective, then there exist Lagrange multipliersr R e ,
and piecewise smooth functions
p
y: I R and
n
z: I R satisfying, for all t I e ,
Sufficient Fritz J ohn Type Optimality Criteria and Duality for Control Problems

Copyright 2011 SciRes. POS
( )
T
x x
T
x
rf (t,x,u) y(t ) g (t,x,u)
z(t ) h (t,x,u) z(t ) 0, t I 4
+
+ + = e

( )
T
u u
T
u
rf (t,x,u ) y(t ) g (t,x,u )
z(t ) h (t,x,u ) 0,t I 5
+
+ = e

( )
T
y(t) g(t,x,u) 0,t I 6 = e
( ) ( ) ( ) r, y t 0,t I 7 > e

( ) ( ) ( ) ( ) r, y t ,z t 0,t I 8 = e

The above conditions will become Karush-Kuhn-Tucker
conditions if 0 r > . Therefore, if we assume that the op-
timal solutions( ) x,u is normal, then without any loss of
generality, we can set 1 r = .Thus from the above we
have the Karush-Kuhn-Tucker type optimality conditions
T T
x x x
f (t,x,u) y(t) g (t,x,u) z(t) h (t,x,u)
z(t ) 0,t I ,
+ +
+ = e

T T
u u u
f (t,x,u) y(t) g (t,x,u) z(t) h (t,x,u) 0,t I , + + = e
T
y(t) g(t,x,u) 0,t I , = e
y( t ) 0,t I . > e
Using these optimality conditions, Mond and Hanson [2]
constructed following Wolfe type dual.
(CD):
( ) { }
b
T T
a
Maximize f (t,x,u) y(t) g(t,x,u) z(t) h(t,x,u) x) dt + +
}

subject to
T T
x x x
f (t,x,u) y(t ) g (t,x,u) z(t ) h (t,x,u)
z(t ) 0,t I
+ +
+ = e
,
T T
u u u
f (t,x,u) y(t ) g (t,x,u) z(t) h (t,x,u) 0,t I + + = e
y(t ) 0,t I > e .
In [6], [CP] and (CD) are shown to from a dual pair if f ,
g and hare all convex in xand u. Subsequently,
Mond and Smart [6] extended this duality under genera-
lized invexity.
As a follows up, Husain et al [7] formulated the follow-
ing dual (CD) to the primal problem (CP) in the spirit of
Mond and Weir [8].
(CD): Maximize
}
b
a
dt u x t f ) , , (

subject to
( ) ( ) x a ,x b = =
T T
x x x
f (t,x,u) y(t ) g (t,x,u) z(t) h (t,x,u)
z(t ) 0,t I
+ +
+ = e

T T
u u u
f (t,x,u) y(t ) g (t,x,u) z(t ) h (t,x,u) 0,t I + + = e
( ) ( )
( )
b
T T
a
y(t ) g(t,x,u) z t h(t,x,u) x dt 0, + >
}

( ) y t 0,t I . > e
They proved sufficiency of the optimality criteria and
duality for the pair of dual problems (CP) and (CD)
under pseudoinvexity of
b
a
f dt
}
and quasi-invexity of
( ) ( )
b
T T
a
y g z h x dt +
}
.
2. Sufficiency of Fritz Type optimality Criteria
Before proceeding to the main results of this section, we
formulate the following definitions which will be re-
quired in the forthcoming analysis:
Definitions:
( )
n m
i : R R I R, the functional
b
a
dt
}
is said strict pseudoconvex. If all ( ) ( ) x,u x,u , =

( ) ( ) ( ) ( ) { }
b
x u
a
x x t,x,u u u t,x,u dt 0 + >
}
( ) ( )
b b
a a
t,x,u dt t,x,u dt >
} }
.
Equivalently
( ) ( )
( ) ( ) ( ) ( ) { }
b b
a a
b
x u
a
t,x,u dt t,x,u dt
x x t,x,u u u t,x,u dt 0


s
+ <
} }
}
( )
m
ii For y: I R , the functional
b
T
a
y gdt
}
is
semi-strictly pseudoconvex if
b
T
a
y gdt
}
is strictly
pseudoconvex for all ( ) y t 0 > , ( ) y t 0. =
If ( ) ( )
T
t,x,u , y t and ( ) g t,x,u are independent of t
and u then the above definitions reduce to those of [6].

Sufficient Fritz J ohn Type Optimality Criteria and Duality for Control Problems

Copyright 2011 SciRes. POS
Theorem 1 (Sufficiency): If
I
f dt
}
is pseudoconvex,
( )
T
I
y t gdt
}
is semi-strictly pseudoconvex and
( ) ( )
T
I
z t h x dt
}

is quasiconvex, and if there exist r R e
and piecewise smooth
m
y: I R and
n
z : I R such
that from (4)-(8) are satisfied, then ( ) x,u is an optimal
solution of (CP).
Proof: Suppose that ( ) x,u is not optimal for (CP) i.e.
there exist ( ) ( ) x,u x,u = such that
( ) ( )
I I
f t,x,u dt f t,x,u dt <
} }

This, by pseudoconvexity of
I
f dt
}
implies
( ) ( ) ( ) ( )
{ }
( ) ( ) ( ) ( )
{ }
( )
T T
x u
I
T T
x u
I
x x f t,x,u u u f t,x,u dt 0
and
x x rf t,x,u u u rf t,x,u dt 0 9
+ <
+ s
}
}

with strict inequality in (9) if r 0 > .
Feasibility of ( ) x,u for (CP) together with (6) implies,
( ) ( ) ( ) ( )
T T
I I
y t g t,x,u dt y t g t,x,u dt s
} }

which by semi-strict pseudoconvexity of ( )
T
I
y t gdt,
}
implies
( ) ( ) ( ) ( )
( ) ( ) ( )
( )
( )
T
x
T
u I
x x y t g t,x,u
dt 0 10
u u y t g t,x,u

s
`
+
)
}

with strict inequality in (10) if some
( ) { }
i
y t 0,i 1,2,...,m > e .
Also
( ) ( ) ( ) ( ) ( ) ( )
T T
I I
z t h t,x,u x dt z t h t,x,u x dt =
} }


This, in view of quasiconvexity of ( ) ( )
T
I
z t h x dt
}


yields
( ) ( ) ( ) ( ) ( )
( ) ( ) ( )
( ) ( ) ( ) ( )
( )
( ) ( ) ( )
( ) ( )
T T
x
T
I u
T T
x
T T
I
u
x x z t h t,x,u x x z t
0 dt
u u z t h t,x,u
x x z t h t,x,u z t
dt
u u z t h t,x,u
t b
x x z t
t a

+

>
`
+
)



=
`

+
)
=
+
=
}
}

(By integrating by parts)
( ) ( ) ( ) ( )
( )
( ) ( ) ( )
( )



=
`

+
)
}
T T
x
T T
I
u
x x z t h t,x,u z t
dt 11
u u z t h t,x,u

(Using (1))

Combining (9), (10) and (11), we have
( )
( ) ( ) ( )
( ) ( ) ( )
( )
( ) ( ) ( )
( ) ( ) ( ) ( )
| + |
|
|
+

\ .
<
`
+ | |

+ |

|
+
\ . )
}
T x x
x
I u u
u u
rf t,x,u y t g t,x,u
x x
z t h t,x,u z t
dt 0
rf t,x,u y t g t,x,u
u u
z t h t,x,u z t h t,x,u

This contradicts (4) and (5). Hence ( ) x,u is an optimal
solution of (CP).

4. Fritz Type Duality

The following is the Fritz john type dual to the problem
(CP):
( ):
r
F CD Maximize ( , , )
I
f t x u dt
}

subject to
( ) ( ) , ( ) 12 x a x b = =

( ) ( ) ( ) , , ( ) , , ( ) ( ) 0, 13
T T
x x x
r f t x u y t g t x u z t h z t t I + + = e

( ) ( ) ( ) , , ( ) , , ( ) 0 , 14
T T
u u u
r f t x u y t g t x u z t h t I + + = e
( ) ( ) ( )
T
I
y t g t,x,u dt 0, 15 >
}
( ) ( ) ( ) ( ) ( ) , , 0 16
T
I
z t h t x u x t dt >
}


( ) ( , ( )) 0 , 17 r y t t I > e

( ) ( , ( ), ( )) 0 , 18 r y t z t t I = e

Theorem2 (Weak duality): Assume that
1
( ) A satisfies( , ) x u is feasible for (CP) and ( , , , ) u r y z
is feasible for (F
r
CD).
Sufficient Fritz J ohn Type Optimality Criteria and Duality for Control Problems

Copyright 2011 SciRes. POS
2
( ) A :
( ) , ,
I
f t x u dt
}
is pseudo-convex,
( ) ( ) , ,
T
I
y t g t x u dt
}
is semi-strictly pseudo-convex and
( ) ( ) ( ) , ,
T
I
z t h t x u x dt
}

is quasi-convex.

Then
inf ( ) ( )
r
CP Sup F CD >


Proof: Suppose
( , , ) ( , , )
I I
f t x u dt f t x u dt <
} }
This, because of pseudo-convexity of ( ) , ,
I
f t x u dt
}

yields
{ } ( )
{( ) ( , , ) ( ) ( , , )} 0
( ) ( , , ) ( ) ( , , ) 0 19
+ <
+ s
}
}
T T
x u
I
T T
x u
I
x x f t x u u u f t x u dt
and
x x rf t x u u u rf t x u dt

with strict inequality in the above with 0 r > .From the
constraints of (CP) and (F
r
CD), we have
( ) ( , , ) ( ) ( , , )
T T
I I
y t g t x u dt y t g t x u dt s
} }

which by semi-strictly pseudo-convexity of
( ) ( ) . ,
T
I
y t g dt
}
implying
( ) {( )( ( ) ) ( )( ( ) )} 0 20
T T
x u
I
x x y t g u u y t g dt + s
}

with strict inequality with ( ) 0
i
y t > ,t I e ,
{1,2,3..... }, i m e Also, we have
( ) ( ( , , ) ) ( ) ( ( , , ) )
T T
I I
z t h t x u x dt z t h t x u x dt s
} }

Using quasi-convexity of ( ) ( ( , , ) )
T
I
z t h t x u x dt
}

in the
above, we have
( ) {( ) ( ) ( ) ( ) } 0 21
T
T T
x u
I
x x h x x z t u u h dt + + s
}



which as earlier becomes
{( ) ( ) ( ) }
T
T
x u
I
x x zh z u u zh dt = +
}

Combining (19), (20) and (21), we have
( )
( ) ( )
0 22
( ) ( )
T T
x x x
T T T
I u u u
x x rf y g zh z
u u rf y g z h
+ +

<
`
+ + +

)
}


From (13) and (14), we get
{( ) ( )}
{( ) ( )} 0
.
T
T
x x x
I
T T T
u u u
I
x x rf y g zh z dt
u u rf y g z h dt
i e
+ +
+ + + =
}
}


( )
( )
( )
0 23
( ) ( )
T
T
x x x
T T T
I
u u u
x x rf y g zh z
u u rf y g z h

+ +

=
`
+ + +

)
}



The relation (22) and (23) are in contradiction, Thus

( , , ) ( , , )
I I
f t x u dt f t x u dt >
} }

Implying


inf( ) ( ).
r
CP sup F CD >


Theorem3 (Strong Duality): If ( , ) x u is an optimal
solution of (CP), then there exist r R e and piecewise
smooth :
m
y I R and :
k
z I R such that ( , , , , ) x u r y z
is feasible for (F
r
CD) and objective values are equal. If
hypotheses of Theorem2 hold, then ( , , , , ) x u r y z is an
optimal solution of (F
r
CD).

Proof: Since( , ) x u is an optimal solution of (CP) by
Proposition 1, there exist r R e , piecewise smooth
:
m
y I R and :
k
z I R such that

( ) , 24
T T
x x x
rf y g z h z t I + + = e

( ) 0, 25
T T
y u u
rf y g z h t I + + = e

( ) ( ) ( , , ) 0, 26
T
y t g t x u t I = e

( ) ( , , ) 0, 27 g t x u t I s e

( ) ( , , ) 0, 28 h t x u t I = e

( ) ( , ( )) 0, 29 r y t t I > e

( ) ( , ( ), ( )) 0, 30 r y t z t t I > e


The relation (26) implies
( ) ( ) ( , , ) 0 31
T
I
y t g t x u dt =
}

and the relation (28) along ( ) y t 0,t I > e gives
( ) ( ) ( , , ) 0 32
T
I
z t h t x u dt =
}

The relation (24), (25), (29), (30), (31) and (32), yields
the feasibility of ( ) r,x,u, y,z for (F
r
CD). Equality of
objective functionals of (CP) and (F
r
CD) is obvious from
their formulations.
Consequently the optimality for (F
r
CD) follows, given
the pseudo-convexity of the ,
I
f dt
}
semi-strict pseudo-
convexity of ,
T
I
y gdt
}
and quasi-convexity of
( ) ( ) ,
T
I
z t h x dt
}
by Theorem 2.

Theorem4 (Strict -Converse duality): Assume that
Sufficient Fritz J ohn Type Optimality Criteria and Duality for Control Problems

Copyright 2011 SciRes. POS
1
( ) A : f dt
}
is strictly pseudo-convex, ( )
T
y t gdt
}
is
semi-strictly pseudo-convex and ( ) ( )
T
z t h x dt
}
is
quasi-convex.
2
( ): A
0 0
( , ) x u is an optimal solution of (CP) and

(A
3
):( , , , , ) x u r y z is an optimal solution of (F
r
CD).
Then ( , ) x u is an optimal solution of (CP) with
0 0
( , ) x u =( , ), x u t I e .

Proof: we suppose
0 0
( , ) ( , ) x u x u = and exhibit a con-
tradiction. Since
0 0
( , ) x u is an optimal solution of (CP)
by theorem (Strong Duality) that there exist
( , ( ), ( )), r y t z t t I e wherer R e and piecewise smooth
:
m
y I R and piecewise smooth :
m
y I R and
:
k
z I R such that( , , , , ) x u r y z is also an optimal
solution for (FrCD), it follows that

0 0
( , , ) ( , , )
I I
f t x u dt f t x u dt =
} }

By strict pseudo-convexity of gives, this implies

0 0
{( ) ( ) } 0
T T
x u
I
x x f u u f dt + <
}

and multiplying the above by r 0

( )
0 0
{( ) ( ) } 0 33
T T
x u
I
x x r f u u r f dt + s
}

with strict inequality if 0. r > From the constraints of
(CP) and (FrCD),we have
( )
0 0
( ) ( , , ) ( ) ( , , ) 34
T
I I
y t g t x u dt y t g t x u dt s
} }

Also
( )
0 0 0
( ) ( ( , , ) )
( ) ( ( , , ) ) 35
T
I
T
I
z t h t x u x dt
z t h t x u x dt

s
}
}

By semi-strict pseudoconvexity of ( )
T
I
y t gdt
}
and from
(34), we have
( )
0 0
{( ) ( ) ( ) ( )} 0 36
T
T T T
x u
I
x x y g u u y g dt + s
}

with strict inequality in the above if,
( ), , {1, 2,3,..... }
i
y t t I i m e e
By quasi-convexity of ( ) ( ) ,
T
I
z t h x dt
}
and from (35),
we get
0 0
0
( )( ) ( ) ( )
0 (37)
( ) ( ( ) )
x
T T
I u
x x zh x x z t
dt
u u z t h


s
`
+

)
}

As earlier, this reduces to

0 0
{( )( ( ) ( )) ( ) ( ( ) )} 0
T T
x u
I
x x z t h z t u u z t h dt + s
}

combining (33), (36), and (37), we have

0
0
( )( ( ) ( ))
0
( ) ( ( ) ( ) )
T T
x x
T T T
I u u u
x x r f y g z t h x
dt
u u r f y t g z t h
+ +

<
`
+ + +

)
}


This contradicts the feasibility of ( , , , , ) x u r y z for
(F
r
CD), hence( , ) x u is an optimal solution of (CP) and
0 0
( , ) ( , ) x u x u = .

Theorem 5 (Converse duality): Let ( , , , ) x r y z be an
optimal solution of (F
r
CD), Assume
1
( ) A ( )
I
f t dt
}
is pseudo-convex, ( )
T
I
y t gdt
}
is
semi-strictly pseudo-convex and ( )
T
I
z h x dt
}
is qua-
siconvex.
2
( ) A The set { ( ) , ( ) ( )}
T T
x x
y t g z t h z t
or
{ ( ) , ( ) }
T T
u u
y t g z t h is linearly independent.
3
( ) A ( ) ( ) ( ) 0, ( ) 0, ,
T
I
v t M t v t dt t t I = = e
}
for some
column vector ( ) v t and where

( ) ( ) ( ) ( )
( )
( ) ( )
T T T T
xx xx xx xu xu xu
T T T
ux ux ux uu uu uu
rf y t g z t h rf y t g z t h
M t
rf y g zh rf y t g z t h
| |
+ + + +
| =
|
+ + + +
\ .

and

( ) ( ) ( )
4
A z a 0 z b = =
Then( , ) x u is optimal for (CP).



Proof: By Proposition 1, there exist R, R, e e
R, R, e e piecewise smooth
m
: I R and
n
: I R such that
( ) ( ) ( )
( )
( ) ( ) ( )
( )
( )
( )
( ) ( )
( )
( )
T T T
x xx xx xx
T T T
ux ux ux
T T
x x
f t r f y t g z t h
t r f y t g z t h
y t g z t h z t 0, t I 38

+ + +
+ + +
+ + = e

( ) ( ) ( )
( )
( ) ( ) ( )
( )
( )
T T T
u ux ux ux
T T T
uu uu uu
f t r f y t g z t h
t r f y t g z t h 0, t I 39

+ + +
+ + + = e

( ) ( ) ( )
T T
x u
t f t f 0 40 + + =
( ) ( ) ( ) ( )
T T
x u
t g t g g t 0, t I 41 + + + = e

( ) ( ) ( ) ( )
T T
x u
t h t h h x 0, t I 42 + + = e
( ) ( )
T
I
y t gdt 0 43 =
}

Sufficient Fritz J ohn Type Optimality Criteria and Duality for Control Problems

Copyright 2011 SciRes. POS
( ) ( ) ( )
T
I
z t h x dt 0 44 =
}

( ) r 0 45 =
( ) ( ) ( )
T
t y t 0,t I 46 = e
( ) ( ) ( ) r, , , , t 0,t I 47 > e
( ) ( ) ( ) ( ) ( ) r, , , , t , t , t 0,t I 48 = e
Multiplying (41) by ( ) y t and integrating, and then
using (43) and (46), we have
( ) ( )
( )
( ) ( )
( ) { }
T T T T
x u
I
t y t g t y t g dt 0 + =
}

which can be written as
( ) ( ) ( )
( )
( )
( )
T
x
T
I u
y t g
t , t dt 0 49
y t g

| |
|
=
|
\ .
}

Multiplying (42) by ( ) z t ,and then integrating we get
( ) ( )
( )
( ) ( )
( ) ( )
( ) ( ) ( ) ( )
T T T T
x u
I
T T
I I
t z t h t z t h dt
t z t dt z t h x dt 0, t I ,


+ +
+ = e
}
} }

d dx
where and x ,
dt dt

= =

Using (44), this yields

( ) ( )
( )
( ) ( )
( ) ( )
( ) ( )
T T T T
x u
I
T
I
0 t z t h t z t h dt
t z t dt, t I

= +
+ e
}
}

( ) ( )
( )
( ) ( )
( ) ( )
T T T T
x u
I
0 t z t h t z t h dt = +
}

( ) ( ) ( ) ( )
I
t b
z t t t z t dt
t a

=
+
=
}

(by integrating by parts)

which in view of (A
4
), implies

( ) ( ) ( )
( )
( ) ( )
( )
T T T T
x u
I
t z t h z t t z t h dt 0
(
+ =
(

}


This can be written as
( ) ( )
( )
( ) ( )
( )
( )
T
x T T
T
I u
z t h z t
t , t dt 0 50
z t h

| |

|
=
|
\ .
}


Using (13) in (38) and (14) in (39), we have
( ) ( )
( )
( ) ( ) ( )
( ) ( ) ( )
( )
( ) ( ) ( )
( )
T
T
x x
T T T
xx xx xx
T T
ux ux ux
r y t g r z(t) h z t
r t r f y t g z t h
r t r f y t g z t h 0

+
+ + +
+ + + =

These can be combined as
( )
( )
( )
( )
( ) ( )
( )
T T
x x
T T
u u
y t g z t h z t
r r
y t g z t h

| | | |

| |
+
| |
\ . \ .

( ) ( ) ( ) ( )
( ) ( ) ( ) ( )
( )
( )
T T T T
xx xx xx ux ux ux
T T T T
ux ux ux uu uu uu
r f y t g z t h r f y t g z t h
rf y t g z t h r f y t g z t h
r t
0
r t

| |
+ + + +
|
+
|
+ + + +
\ .
| |
= |
|
\ .

Pre-multiplying this by
( ) ( )
( )
T T
r t ,r t
and then inte-
grating we have

( ) ( ) ( )
( )
( )
( )
( ) ( ) ( )
( )
( ) ( )
( )
T
x T T
T
u I
T
x T T
T
u
I
y t g
r r t ,r t dt
y t g
z t h z t
r r t ,r t dt
z t h


| |
|

|
\ .
| |

|
+
|
\ .
}
}


( ) ( )
( )
( ) ( ) ( ) ( )
( ) ( ) ( ) ( )
( ) ( ) ( )

| |
+ + + +

|
+ =
`
|
+ + + +
\ .



)
}
T T
T T T T
xx xx xx ux ux ux
T T T T
ux ux ux uu uu uu I
r t ,r t
r f y t g z t h r f y t g z t h
dt 0
rf y t g z t h r f y t g z t h
r t ,r t



using (49) and (50) , this implies
( ) ( )
( )
( ) ( ) ( ) ( )
( ) ( ) ( ) ( )
( ) ( ) ( )

| |
+ + + +

|
=
`
|
+ + + +
\ .



)
}
T T
T T T T
xx xx xx ux ux ux
T T T T
ux ux ux uu uu uu
I
r t ,r t
r f y t g z t h r f y t g z t h
dt 0
rf y t g z t h r f y t g z t h
r t ,r t



which in view of (A
2
) implies
( ) ( ) ( ) ( )
t r t ,r t 0, t I = = e implies
( ) ( ) ( ) r t 0,r t 0,t I 51 = = e
In view of (A
3
), the equality constraint implies
r 0i.e r 0. = > consequently, we have
( ) ( ) ( ) t 0,t I , t 0,t I , 52 = e = e

Using (52), along with r >0, we have

Sufficient Fritz J ohn Type Optimality Criteria and Duality for Control Problems

Copyright 2011 SciRes. POS
( )
( )
( ) ( )
( )
T T
x x
y t g z t h z t 0
r r


| | | |
+ =
| |
\ . \ .

This, in view of (A
3
),
( ) 0, 0 53
r r

= =

If 0, = (53), implies 0 = = , Thus
( ) ( ) ( ) ( )
r, , , , t , t , t 0, = contradiction.
Hence 0 > and consequently 0 > and 0 > .
From (41) and (42), we have
( )
( )
g t,x,u 0,t I
h t,x,u 0,t I
s e
= e

These relations yield the feasibility of ( ) x,u for (CP)
and objective functionals of (CP) and (FrCD) are equal
there. Hence under the stated convexity hypotheses, by
Theorem 2,( ) x,u is an optimal solution of (CP).

5. Mathematical Programming Problems
If the problems (CP) and (F
r
CD) are independent of t and
x, these problems reduce to essentially to the static cases
of nonlinear programming problem. Letting b a = 1,
the problems (CP) and (F
r
CD) become the pair of dual
nonlinear programming problems formulated by Husain
and Srivastav [9].
(CD
0
): Minimize ( ) f u
subject to

( )
( )
g u 0,
h u 0.
s
=

(FrCD): Maximize ( ) f u
subject to

( ) ( ) ( )
( )
( )
( )
( )
T T
u u u
T
T
rf u y g u z h u 0,
y g u 0,
z h u 0,
r, y 0,
r, y,z 0.
+ + =
>
>
>
=

Wheref is pseudoconvex, ( )
T
y g . is semi-strictly pseu-
doconvex and
T
z h is quasiconvex. If only inequality
constraint in (CD
0
) is given, then (CP
0
) and (F
r
CD
0
) be-
come a pair of dual the nonlinear programming problems
considered by Weir and Mond [10].

6. Conclusion:
In this paper, sufficient optimality conditions are derived
for a control problem which appears in various real life
situations under generalized convexity assumptions. In
order to formulate the dual to this control problem, Fritz
John optimality conditions are used instead of
Karush-Kuhn-Tucker optimality condition and hence the
requirement of regularity condition is eliminated.
Various duality results are obtained and the linkage of
our duality results to those of a nonlinear programming
problem is indicated. Our results can be seen in the set-
ting of multiobjective control problems.


7. Acknowledgement:
The authors acknowledge anonymous referees for their
valuable comments which have improved the presenta-
tion of this research paper.
REFERENCES
[1] L.D.Berkovitz, Variational Methods in Problems of Con-
trol and Programming, J. Math. Anal. Appl. Vol 3, no.1, 1961,
pp. 145-169.

[2] B.Mond and M.A.Hanson, Duality for Control Problems,
SIAM J ournal on Control, Vol. 6, 1965, pp.28-35.

[3] S. Chandra, B. D. Craven, and I. Husain A class of non-
differentiable control problems J ournal of Optimization
Theory and Applications, Vol. 56, Issue 2, 1988, pp. 227-243.

[4] B. D. Craven, Mathematical Programming and Control
Theory, Chapman and Hall, London, 1978.

[5] F.A.Valentine, The Problem of Lagrange with Differential
Inequalities as Added Side Conditions, Contribution to the
Calculus of Variation, 1933-37, University of Chicago Press,
Chicago, Illinois,1937,pp. 407-448.

[6] B.Mond and I. Smart, Duality and Sufficiency in Control
Problems with invexity, J.Math. Anal.Appl. 136, No. 15,1988
325-333, 1988.

[7] I. Husain, A. Ahmed and B. Ahmad Sufficiency and dual-
ity in control problems with generalized invexity J ournal of
Applied Analysis Vol.14 No.1 (2008), pp.27-42.
[8] B.Mond and T. Weir Generalized Concavity and Duality in
Generalized Concavity in Optimization and Economics, (S.
Schaibl and W.T.Ziemba Eds.), Academic Press, New York,
263-279, 1981.

[9] Iqbal Husain and Santosh k. Shrivastav Fritz J ohn Duality
in the Presence of Equality and Inequality Constarints Applied
Mathematics, Vol.3, 2012, pp. 1023-1028.

[10] T.Weir and B.Mond Sufficient Fritz J ohn Optimality
Conditions and Duality for Nonlinear Programming Problems
Opsearch, Vol.23 No.3, 1986, pp.129-141.