You are on page 1of 4

International Journal of Advanced Research and Publications

ISSN: 2456-9992

Optimality Boundary Conditions For Stochastic


Elliptic System Involving Higher Order Operator
A. M. Abdallah
Basic Science Department, Higher Technological Institute,
Tenth of Ramadan, Egypt, PH-00201002621367
ahmedabdallah6236@yahoo.com

Abstract: In this present work, we deal with the stochastic elliptic systems where the model of the stochastic elliptic system is higher order
operator. We prove the existence and uniqueness of the state variable process of these systems. Then, we give the necessary and sufficient
for the optimality of the system.

Keywords: Optimal Stochastic Control, Optimality conditions, Stochastic analysis

1. Introduction where A maps onto ,


Here, we consider the following stochastic elliptic system is the conjugate space of .
involving higher order operator. Then,
b (u ,  )    (  ) u ( x ) ( x )dxdp
m
(  ) u ( x)  W ( x)
m
(1.3)
 (1.1)  G
m 1
 u  u and the linear form
u ( x)   ...  m 1
 0
 n n L ( )   W ( x ) ( x )dxdp
(1.4)
 G
First Eq. is defined on G, Dirichlet condition is defined on
, where G is a bounded, continuous domain in where W(x) is the white noise. Hence, the abstract
with boundary  G , u ( x )  u ( x , w )   0m (  , F ,  , G ) Variational problem associated with Eq.(1.1) can be written
as
is a state process and W(x) is a white noise.
Eq. (1.1) represents the state process equation and

2
{
n

    xi
2
is the Laplacian operator.
i 1

We also prove the existence and uniqueness of the optimal By the Lax-Milgram lemma, we prove the following
stochastic control of distributed and boundary types, and we theorem.
discuss the necessary and sufficient conditions of the
optimality. In the following subsection, existence and Theorem 1.
uniqueness for solution of the state process equations are The bilinear form (1.3) satisfies the stochastic coerciveness
discussed. condition, and then there exists a unique solution
of the system (1.1), conversely, if there
1.1 Existence and Uniqueness for the state process of the exists a unique solution such that
system then we get the system (1.1).
In this subsection, we study the existence and uniqueness of b (u ,  )  L ( ),
solution for Eq.(1.1). Proof.
The proof of existence and uniqueness uses the Lax-
Since Milgram lemma. It is necessary to show that the bilinear
m
form (1.3) is continuous and coercive. We divide the proof
 ( , F ,  , G )  L ( , F ,  , G )   ( , F ,  , G )
m 2
0
(1.2) into several steps.
The model of the stochastic system (1.1) is given by: (a): Continuity: Applying Green’s formula, we find out
m m
A  (  ) , Au (x)  W (x)
m

 (  ) u ( x ) ( x )dx   u 2  dx   F (u ,  ) d  G
m 2

The elliptic operator A in the state equation (1.1) is a G G G

bounded second order self adjoint stochastic elliptic partial Obviously, we can rewrite Eq.(1.3) by stochastic Green’s
differential operator. For this operator we define the bilinear formaula
form:
m m

on , by: b (u ,  )  E (   2 u ( x ) 2  ( x ) d x )
G

b (u ,  )  ( A u ,  ) , u,   m
( , F ,  , G ) m 1 
L
2
(  , F , ,G ) 0 hence F ( u ,  )  (   ) u  0
n

Volume 2 Issue 5, May 2018 1


www.ijarp.org
International Journal of Advanced Research and Publications
ISSN: 2456-9992

on G by virtue Dirchlet condition. Then, by Cauchy Cartesian product as follows:


Schwartz inequality
2
 ( , F ,  , G ) 
m 1 m 1 m

b (u ,  )  E (   u ( x ) d x ) E (    ( x ) d x ) ,
2 2 2 2  0 

  ( , F ,  , G )     ( , F ,  , G )  .
m m
G G
since 1  0   0 
2
u  E (  u (x) dx) 2 ,
m Since the bilinear form b(u,.) is continuous and stochastic
m
 0 (  , F , ,G ) 2
coercive on   ( , F ,  , G ) 
m
, and the linear
then G
 0 
form is also continuous on
b (u ,  )  u  , (1.5)

m
( , F ,  , G ) 
2
, by Lax Milgram lemma

m
(  , F , ,G ) 
m
(  , F , ,G )
 0 
0 0
there exist a unique solution
Thus, the linear form L(.) is continuous on 2
u   ( , F ,  , G ) 
m
, such that
 0 

m
(  , F , ,G ) , by using Cauchy Schwartz
0
b (u ,  )  L ( )   [ H
m
( , F ,  ; G )] .
2 (1.8)
inequality, we find out 0

Conversely, when b ( u ,  )  L (  )    [H
m
( , F ,  ; G )] .
2
L ( )   W ( x ) ( x )dxdp
and u  [ H m (  , F ,  ; G ) ] 2 ,
0
0

 G

1 1 integrating Eqn. (1.1) on G and taking expectation, we find


  (  (W ( x ) ) d x ) 2 d p  (  (  ( x ) ) d x ) 2 d p
2 2

 u  dxdp   W ( x ) ( x )d xd p
m m
 G  G
 G  G
 W 2  2
L L
By applying stochastic Green’s formula:
from Eq.(1.2), we have
  ( ) u  dxdp   F (u ,  ) d  G d p    W ( x ) ( x )d xd p ,
m
u  u ,
L
2
(  , F , ,G ) 
m
(  , F , ,G )  G  G  G
0

Then, we get
on G and F(u(x), (x)) = 0 by virtue of Dirichlet condition
L ( )  W m  m (1.6) u(x) = 0. By comparison of two sides, we deduce the system
H 0 H 0 (1.1), which completes the proof. Under the above
consideration, using the theorems 1.1, 1.2 of [4], we can
(b): Coerciveness: From definition of the norm on
formulate the following Dirichlet problem, which define the
state process of our control problem. Now, we formulate the
m
control problem with adding the control in the region G and
 ( , F ,  , G )
0 we determine the cost functional.
2 2
 E (  W
m
u dx) 1.2 Formulation of the boundary control problem with
m
H
0 G boundary observation
We add the control on the boundary to formulate the initial
 E (  W  W dx)
m m
boundary value Dirichlet problem for stochastic elliptic
G systems and we investigate necessary conditions for an
optimal boundary control problem. The space
 E (  W  W dx)  E (  F ( u ,  )d  G )
m m

is the space of controls. For a control


G G
, the state process of the system is
By Green’s formula, we obtain given by the solution of the following system:
2
 E (  (   ) u .u d x )  b ( u , u )
m
u (  ) u ( y )  W ( y )
m


m
H
0 G m 1 (1.9)
 u  u
 u ( y )   ...  m 1
 y
Therefore,  n n
2 The observation equation is given by , and the
b (u , u )  c u (Stochastic coerciveness) cost functional is given by:
m
H
(1.7) 0
C ( y)  E[  ( u ( y )   d )d x ]    M zydxdp (1.10)
It is easy to construct the following Sobolev spaces G  G

2 where
 ( , F ,  , G ) 
m
 0 
by the 2-times Then, the control problem is defined by such

Volume 2 Issue 5, May 2018 2


www.ijarp.org
International Journal of Advanced Research and Publications
ISSN: 2456-9992

that: Example 2.
C ( y )  C ( z ) z  Yad ;
For m = 2 , it is proved that the state of the system is given
where is a closed convex subset from .
 u( y)  W ( y)
The cost functional (1.10) can be rewritten as: 2


C ( y)   (u ( y )  u (0 )  u (0 )   d ) d xd p  
2

u ( y )  0
M yzdxdp
 G  G

where
b (u ,  )   u ( x ) ( x )d xd p
2

 ( y, z)   (( u ( y )  u ( 0 ))  ( u ( z )  u ( 0 )) ) d x d p
2 2
 G
 G
2. Acknowledgments
 (1.11)
 M yzdxdp The author would like to express his gratitude to professor
 G Dr Ahmed Said Okb El Bab, Department of Mathematics,
Al Azhar Univeristy, Cairo, Egypt and Professor Dr Claudio
Roberto Ávila, Departamento Acadêmico De Mecânica,
L(z)   ((  d  u ( 0 ))  ( u ( z )  u ( 0 )) ) d x d p
2 2
(1.12) Curitiba, Brazil for suggestion the problem and critically
 G reading the manuscript.

Theorem 2. References
If the state u ( y ) is given by Eq.(1.1) and if the cost
[1] A. M. Abdallah “Optimality Conditions for non-
functional is given by Eq.(1.10), then there exists a
cooperative stochastic systems,” Lambert cooperation
unique optimal control y  Y a d such that
publishing : Germany, 2018.
C ( y )  C ( z ) z  Yad ;
[2] A. M. Abdallah “On Stochastic Elliptic System
Moreover, it is characterized by the following equation
Involving Higher Order Operator With Dirichlet
Conditions,” Journal of AMO - Advanced Modeling
(  ) h( y )  u ( y )   d
m
and Optimization, vol. 20, no. 1, 2018.

h( y)  0
[3] J. Fleckinger and M. L. Lapidus, “Eigen values of
h elliptic boundary value problems with an indefinite we
The inequaity
 (
vA
 M y )( z  y ) d x d p  0 weight function,” Transactions of the American
 G Mathematical Society, vol. 295, pp. 305-324, 1986.
represents the necessary and sufficient condition for [4] J. L. Lions, Optimal control of a system governed by
optimality equation. partial differential equations, Springer-Verlag: New
York, 1971.
Proposition 1.
[5] A. S. Okb El Bab, Abd-Allah Hyder, and A. M.
If the constraints are absent, i.e., when Y  ,Y then the ad Abdallah, “Boundary control of stochastic elliptic
equalityh  M y  0 , z  y the differential problem of finding systems involving Laplace operator,” Journal of AMO -
the vector-function satisfies the following relations, the Advanced Modeling and Optimization, vol. 19, pp. 53-
equation 61, 2017.

 Au  W [6] A. S. Okb El Bab, Abd-Allah Hyder, and A. M.


 Abdallah, “Distributed control of stochastic elliptic
 h (1.13)
u   0 systems involving Laplace operator,” Electronic
 M Journal of Mathematical Analysis and Applications,
vol. 5, pp. 191-199, 2017.
represents the state process equation for the system
[7] I. V. Sergienko and V. S. Deineka, “Optimal control of
(1.1) without constraints.
an elliptic system with conjugation conditions and
Neumann boundary conditions,” Journal of Cyberne-
Example 1.
tics and Systems Analysis, vol.40 (2004), pp. 865-882,
2004.
For m = 1 , it is proved in [5] that the state of the system is
given

 (   )u ( y )  W ( y )

u ( y)  0

b (u ,  )      u ( x ) ( x )dxdp
 G

Volume 2 Issue 5, May 2018 3


www.ijarp.org
International Journal of Advanced Research and Publications
ISSN: 2456-9992

Author Profile
Author received the B.S. and M.S. degrees in Functional
Analysis (Optimal Control) from
Zagazig Univeristy, Al Azhar
Univeristy in 2007 and 2017,
respectively. During 2009-now, he
stayed in Basic Science
Department, Higher Technological
Institute, Tenth of Ramadan City,
Egypt.

Volume 2 Issue 5, May 2018 4


www.ijarp.org

You might also like