You are on page 1of 7

Int. Journal of Math. Analysis, Vol. 3, 2009, no.

18, 871 - 877

A Class of Third Order Parabolic Equations with Integral Conditions


M. Bouzit and N. Teyar Universit e Mentouri Constantine Institut de Math ematiques 25000 Constantine, Algeria bouzit med@yahoo.fr, teyarnadir@gmail.com
Abstract In this paper, we study a mixed problem for a third order parabolic equation with non classical boundary condition. We prove the existence and uniqueness of the solution. The proof of the uniqueness is based on a priori estimate and the existence is established by Fouriers method.

Mathematics Subject Classication: 35B45, 35K35 Keywords: Integral Boundary Condition, Energy Inequalities, Parabolic equation of mixed type

1. INTRODUCTION In the set = (0, T ) (0, 1), we consider the equation 2u 2u x1r x = F (t, x), r 0 and k 0, (1.1) xr+1 xt + k u t2 t To equation (1.1) we attach the initial condition u(0, x) = (x) x (0, 1) (1.2) u (0, x) = (x) x (0, 1) (1.3) t and the integral condition 1 1 u(t, x)dx = 0, 0 xr u(t, x)dx = 0 for t (0, T ) (1.4) 0 Where (x), (x) L2 (0, 1) are known functions which satisfy the compatibility conditions given in (1.4). The boundary value problems with integrals conditions are mainly motivated by the work of Samarskii [3]. Regular case of this problem for second order equations is studied in [4]. The problem where the equation of mixed 2+1 u type contains an operator of the form a(t) 2 t is treated in [17], the operx u u is treated in [4] and [14]. ator of the form x a(t, x) x and x a(t, x) x

872

M. Bouzit and N. Teyar

Two-point boundary value problems for parabolic equations, with an integral condition, are investigated using the energy inequalities method in [8, 9, 10, 11] and the Fouriers method [12]. Three-point boundary value problem with an integral condition for parabolic equations with the Bessel operator is studied in [12]. And recently parabolic and hyperbolic equations with integral boundary condition are treated by Fouriers method in [1, 5]. The presence of nonlocal conditions raises complications in applying standard methods to solve (1.1)-(1.4). Therefore to over come this diculty we will transfer this problem to another which we can handle more eectly. For that, we have the following lemma. Lemma 1. (1.1)-(1.4) is equivalent to the following problem Problem 2u 2u 1 x = F (t, x) xr+1 xt + k u r t2 x t u(0, x) = (x) u (0, x) = (x) (Pr)1 t 1 r 1 u (t, 1) u (t, 0) = 1 x F (t, x)dx 0 F (t, x)dx 0 t t r 1 2u (t, 1) = 0 xr F (t, x)dx xt Proof. Let u(t, x) be a solution of (1.1)-(1.4). Integrating equation (1.1) with respect to x over (0, 1), and taking into account of (1.4), we obtain
u x xt
2

1 0

1 2u dx 0 xt

1 0

F (t, x)dx
1

And so

+ r ( u (t, 1) u (t, 0)) = 0 F (t, x)dx t t 1 To eliminate the second nonlocal condition 0 xr u(t, )d = 0 multiplying both sides of (1.1) by xr and integrating the resulting over (0,1), and taking in account of (1.4), we obtain 1 2u (t, 1) = 0 xr F (t, x)dx xt These may also be written 1 r 1 u (t, 1) u (t, 0) = 1 x F (t, x)dx 0 F (t, x)dx 0 t t r And 1 2u (t, 1) = 0 xr F (t, x)dx xt Let u(t, x) now be a solution of (Pr)1 , it remains to prove that 1 u(t, x)dx = 0, 0 And 1 r x u(t, x)dx = 0 0 We integrate Eq.(1.1) with respect to x , we obtain 1 1 d2 d u(t, x)dx + k dt u(t, x)dx = 0, t (0, T ) dt2 0 0 And it also follows that 1 r 1 r d2 d x u(t, x)dx + k dt x u(t, x)dx = 0, t (0, T ) 0 dt2 0 Introduce now the new function v (t, x) = u(t, x) u0 (t, x), where

2u (t, 1) xt

Third order parabolic equations

873
t

u0 (t, x) = (x) 0 m1 ( )d + (x) 0 m2 ( )d, (x) = x + x2 1 , (x) = 2x x2 , m1 (t) = 0 xr F (t, x)dx, m2 (t) =
1 r

m1 (t)

1 0

F (t, x)dx

into the following problem Then (Pr) 1 is transformed 2 2v v 1 v xr+1 xt + k v x = f (t, x) r t2 x t lv = v (0, x) = (x) (Pr)2 qv = v (0, x) = (x) t v ( t, 1) = v (t, 0) t t 2v (t, 1) = 0 xt Where 1 1 f (t, x) = F (t, x) + ((x) (rx) ) 0 xr Ft (t, x)dx + (rx) 0 Ft (t, x)dx + (t, x), (t, x) = ((r + 1) + 2(r + 2)x)m1 (t) + (2(r + 1) 2(r + 2)x)m2 (t) k(x)m1 (t) k (x)m2 (t) 1 1 1 (x) = (x) + (x) 0 xr F (0, x)dx + (rx) ( 0 xr F (0, x)dx + 0 F (0, x)dx) 2. A PRIORI ESTIMATE we consider (P r )2 as a solution of the operator equation Lv = F , where L = ( , l, q ) , F = (f, , ). The operator L is acting from the Banach space D (L) = E to F where E= v:x
r +1 2

v, x

r +1 2

r +1 v 2v , x 2 xt t r +1

L2 (0, 1) and
2 r +1

v , x 2 xt ,x 2 x 2 v t With respect to the norm 2v 2v 2 r +1 )2 + ( xt ) + ( )2 x 2 ( v E = r x t t2

r +1

3v 2v r 2v , xr+1 x 2 t , x xt t2

L2 (r )

v xr+1 xt

+ sup

0tT

1 0

v 2 xr+1 v 2 + ( v )2 + ( xt ) t

Here F is the Hilbert space with the norm 1 2 2 2 2 2 F 2 F = ((f, , ) F = r f + 0 ( + ( ) + ) Lemma 2. For any function u E , we have 1 r 2 1 r v 2 exp(cT ) 1 r x (v (, x))2 1 x +1 x ( t ) (2.1) 0 8 8 0 8 0 0 with the constant c satisfying c 1 . ) and using elementary inProof. Integrating by parts (exp(ct)xr v, v t equalities yields (2.1). Theorem 1. For (P r )2 We have v E C Lv F , Where C 0 is independent on v . Proof. Let 2v + xr Mv = 2xr v t t2

874

M. Bouzit and N. Teyar

Consider the scalar product ( v, Mv ), and integrating over = (0, ) (0, 1), we get 1 1 ) 0 xr ( v (, x))2 (1+ k ) 0 xr 2 +2k xr ( v )2 + ( v, Mv )L2 ( ) = (1+ k 2 t 2 t 2v 1 r +1 2 v 1 r +1 2v 2 ) + xr ( )2 + 1 x ( xt (, x))2 1 x ( )2 2 xr+1 ( xt t2 2 0 2 0 1 1 (1 + k ) 0 xr ( v (, x))2 (1 + k ) 0 xr 2 + 2k xr ( v )2 + 2 t 2 t 2 2 2 1 1 r v 2 v 2 1 v ) + xr ( ) +2 xr+1 ( xt (, x))2 1 x ( )2 (2.2) 2 xr+1 ( xt t2 2 0 0 2 v + xr ) we obtain We now apply an -inequality to the term ( v, 2xr v t t2 1 1 r v r 2v r 2 r v 2 r 2 ( v, 2x t + x t2 ) = 1 x f + 1 x ( t ) + 22 x f +
2
2

v 2 xr ( ) t2

(2.3)
2v t2

From equation v
1 8 x v xr+1 xt
2

1 4

1 xr

xr+1 xt

2v

+ k v = f (t, x) we have t
v 2 xr ( ) +1 k t2 4
2

Combining inequalities (2.1), (2.2), (2.3) and (2.4) and since (x 1) we obtain 1 1 1 2 4 2 +2 1 + 1 2 f 2 + (1 + k ) 0 2 + 1 ( )2 + 1 4 1 2 2 2 0 8 0 ) (1 + k 2 +( 3 4
exp(cT ) 8 2
2

xr f 2 + 1 4

v 2 xr ( ) (2.4) t2

1 0

xr+1 ( v (, x))2 + ( 74k t


v 2 xr ( ) + t2
2

)
1 0

1 2

1 0

v xr+1 ( xt (, x))2 +

1 ) 8

xr ( v )2 + 2 t
1 8 x

v 2 xr+1 ( xt )
2

v xr+1 xt

xr+1 (v (, x))2 (2.5) 7k 1 3 2 Next choosing as and i , i = 1, 2 as 4 1 8 = k1 0 and 4 2 = k2 0. The left-hand side of (2.5) is independent of , hence replacing the righthand side by its upper bound with respect to , in the interval [0, 1], we obtain the desired inequality. This completes the proof. 3. EXISTENCE AND UNIQUENESS OF SOLUTION We shall establish the existence of solution of (P r )2 . For this we make use of the Fouriers method. Consider the function vn (t, x) = Tn (t)Xn (x) where Xn (x) is an eigenfunction of the BVP 1 d n xr+1 dX kXn = n Xn x r dx dx Xn (1) = Xn (0) d Xn (1) = 0 dx n , n = 1, 2....... is called the eigenvalue corresponding to the eigenfunction the initial problem Xn (1), and Tn (t) is satisfying d2 T n n = fn (t) dt2 n dT dt Tn (0) = n d Tn (0) = n dt

(x) =

n=1

n Xn (x)

Third order parabolic equations

875

(x) = (x) =

n=1

n Xn (x) n Xn (x) fn (t)Xn (x)

n=1

f (t, x) = And Hence


1 0 2 L2 (0,1) 2 L2 (0,1)

And by the Parseval-Steklov equality =


n=1

n=1

2 n, 2 n,

=
n=1 n=1

2 L2 (0,1)

2 ( n) ,

f (t, x)dx =

n=1

2 fn (t). n=1 T 0

f 2 (t, x)dxdt =

fn2 (t).

Then direct computation yields t t s Tn (t) = n + 0 n exp(n )d + 0 0 fn ( ) exp(n s n )d ds 1 r x Xn (x)Xm (x)dx = 0, n = m 0 And 1 r x (x)X (x)dx n = 0 1 xr X 2 (n x)dx
1

n = v (t, x) =

By principle of superposition, the solution of (P r )2 is given by the series

n 0 r 0x (x)Xn (x)dx 1 r 2 0 x Xn (x)dx

Then we have Theorem 2. Let f, L2 (), and H 1 (0, 1). Then the solution v (t, x) of (P r )2 exists and is represented by series (3.1) wich converges in E .
N

n=1

Tn (t)Xn (x).

(3.1)

Proof. Consider the partial sum SN (t, x) = (3.1) then by theorem 1


N n=1 2

n=1

Tn (t)Xn (x) of the series

Tn (t)Xn (x)
N E T 0 n=1

C1

The series

fn2 (t)dt,

n=1 N

T 0 N

fn2 (t)dt + n2 + n2 + (n )2 n2 ,and


N n=1

(3.2)

n=1

n2 ,

n=1

(n )2 converge. There-

fore it follows from (3.2) that the series (3.1) converges in E and accordingly its sum v E.

876 REFERENCES

M. Bouzit and N. Teyar

[1] M.A.Al-kadhi, A class of hyperbolic equations with nonlocal conditions, Int. J. Math. Analysis, 2 (10) (2008), 491-498. [2] Batten, Jr., G.W., Second-order correct boundary conditions for the numerical solution of the mixed boundary problem for parabolic equations, Math. Comp. 17 (1963), 405-413. [3] N. E. Benouar and N. I. Yurchuk, Mixed problem with an integral condition for parabolic equations with the Bessel operator, Dier. Equ. 27 (12) (1991), 1177-1487. [4] Bouziani, A. and Benouar, N.E., Mixed problem with integral conditions for a third order parabolic equation, Kobe J. Math. 15 (1998), 47-58. [5] L.Bougoa, Parabolic equations with nonlocal conditions, Appl. Math. Sciences 1 (21) (2007), 1041-1048. [6] A. Bouziani and N. E. Benouar, Probl` eme mixte avec conditions int egrales pour une classe d equations paraboliques, C. R. Acad. Sci. Paris, S erie 1 321 (1995), 1182. [7] Choi, Y.S. and Chan, K.Y., A parabolic equation with nonlocal boundary conditions arising from electrochemistry, Nonl. Anal. 18 (1992), 317-331. [8] M. Denche and A. L. Marhoune, A three-point boundary value problem with an integral condition for parabolic equations with the Bessel operator, Appl. Math. Lett. 13 (2000), 85-89. [9] M. Denche and A. L. Marhoune, High-order mixed-type dierential equations with weighted integral boundary conditions, Electron. J. Dierential Equations 60 (2000), 1-10. [10] M. Denche and A. L. Marhoune, Mixed problem with integral boundary condition for a high order mixed type partial dierential equation, J. Appl. Math. Stochastic Anal. 16 (1) (2003), 69-79. [11] M. Denche and A. L. Marhoune, Mixed problem with non local boundary conditions for a third-order. partial dierential equation of mixed type, Int. J. Math. Sci. 26 (7) (2001), 417-426.. [12] N. I. Ionkin, A problem for the heat-condition equation with a twopoint boundary Condition, Dier. Uravn. 15 (7) (1979), 1284-1295. [13] A. V. Kartynnik, Three-point boundary-value problem with an integral space-variable condition for a second-order parabolic equation, Dier. Equ. 26 (9) (1990),1160-1166. [14] N. I. Kamynin, A boundary value problem in the theory of the condition with non classical boundary condition, Th. Vychist. Mat. Fiz. 43 (6) (1964), 1006-1024. [15] A.L.Marhoune and M.Bouzit, High order dierential equations with integral boundary condition, FJMS, 2006. [16] A. A. Samarski, Some problems in dierential equations theory, dier.

Third order parabolic equations

877

Uravn, 16(11) (1980), 1221-1228 [17] N. I. Yurchuk, Mixed problem with an integral condition for certain parabolic equations, Dier. Equ. 22 (12), (1986),1457-1463.

Received: November, 2008

You might also like