CONTROL PROBLEMS FOR NONLINEAR ... - Semantic Scholar

2 downloads 0 Views 148KB Size Report
Jin-Mun Jeong∗ and Han-Geul Kim. Abstract. This paper deals with the approximate controllability for the non- linear functional differential equations with time ...

J. Appl. Math. & Computing Vol. 23(2007), No. 1 - 2, pp. 445 - 453

CONTROL PROBLEMS FOR NONLINEAR RETARDED FUNCTIONAL DIFFERENTIAL EQUATIONS

Jin-Mun Jeong∗ and Han-Geul Kim

Abstract.

This paper deals with the approximate controllability for the nonlinear functional differential equations with time delay and studies a variation of constant formula for solutions of the given equations. AMS Mathematics Subject Classification : 35B37, 93C20. Key words and Phrases : Retarded functional differential equation, existence, regularity, approximate controllability

1. Introduction Let H and V be two real separable Hilbert spaces such that V is a dense subspace of H. In this paper we investigate the approximate controllability for the following nonlinear functional differential equation on H:  Z 0   dx(t) + Ax(t) = g(t, s, x(t), x(t + s))µ(ds) + (Bu)(t), dt −h   x(0) = g 0 , x(s) = g 1 (s) s ∈ [−h, 0).

0 < t ≤ T,

(NE) Let the principal operator A be given as a single valued, monotone operator, which is hemicontinuous and coercive from V to V ∗ . Here V ∗ stands for the dual space of V . If the nonlinear term belongs to L2 (0, T ; V ∗ ), the basic assumption made in these investigations is taken from the regularity result for the quasiautonomous differential equation(see Theorem 2.6 of Chapter III in [1]). Most studies have been devoted to the semilinear system without time delay, and the paper treating the nonlinear system with delay are not many. The Received March 10, 2006. ∗ Corresponding author. This work was supported by the Pukyong National University Research Foundation in 2005. c 2007 Korean Society for Computational & Applied Mathematics and Korean SIGCAM

445

446

J. M. Jeong and H. G. Kim

regularity of solution of the semilinear functional differential equations with unbounded delays has been surveyed in Jeong, Kwun and Park [2] and Vrabie [3]. The approximate controllability for semilinear systems has been also studied in [2]. The existence of solutions for a class of nonlinear evolution equations in the case in which A is nonlinear were developed in many references [4-7]. Ahmed and Xiang [4] gave some existence results for the initial value problem in case where the nonlinear term is not monotone, which improving Hirano’s result[7]. We will first establish a variation of constant formula for solutions of the given equation with a nonlinear operator A on L2 (0, T ; V )∩W 1,2 (0, T ; V ∗ ) under some general condition of the Lipschitz continuity of the nonlinear operator, which is reasonable and widely used in case of the nonlinear system. The main research direction is to find conditions on the nonlinear term such that the regularity result of (NE) is preserved under perturbation. In order to prove control problem, as in [2] we must assume the uniform boundedness of the nonlinear term, although we have some remark on this hypothesis. Since we apply the Leray-Schauder degree of mapping theorem in the proof of the main theorem, we need some compactness hypothesis. We make the natural assumption that the embedding V ⊂ H is compact. Then the embedding L2 (0, T ; V ) ∩ W 1,2 (0, T ; V ∗ ) ⊂ L2 (0, T ; H) is compact in view of J. P. Aubin’s result [5], and we show that the mapping which maps a contol u to the mild solution of (NE) is a compact operator from L2 (0, T ; H) to itself. 2. Assumptions and Preliminaries If H is identified with its dual space, we may write V ⊂ H ⊂ V ∗ densely and the corresponding injections are continuous. The norms on V , H and V ∗ will be denoted by || · ||, | · | and || · ||∗ , respectively. Thus, in terms of the intermediate theory we may assume that (V, V ∗ ) 12 ,2 = H where (V, V ∗ ) 12 ,2 denotes the real interpolation space between V and V ∗ . The duality pairing between the element v1 of V ∗ and the element v2 of V is denoted by (v1 , v2 ), which is the ordinary inner product in H if v1 , v2 ∈ H. For the sake of simplicity, we may consider ||u||∗ ≤ |u| ≤ ||u||,

u ∈ V.

We note that a nonlinear operator A is said to be hemicontinuous on V if w − lim A(x + ty) = Ax t→0

for every x, y ∈ V where ”w − lim” indicates the weak convergence on V .

Control problems for nonlinear retarded functional differential equations

447

Let A : V −→ V ∗ be given a monotone operator and hemicontinuous from V to V ∗ such that (Au − Av, u − v) ≥ ω1 ||u − v||2 − ω2 |u − v|2 ,

(A)

||Au||∗ ≤ ω3 (||u|| + 1) for every u, v ∈ V where ω2 is a real number and ω1 , ω3 are some positive constants. It is well known that A is maximal monotone and R(A) = V ∗ where R(A) denotes the range of A. Let L and B be the Lebesgue σ-field on [0, ∞) and the Borel σ-field on [−h, 0], respectively. Let µ be a Borel measure on [−h, 0] and g : [0, ∞) × [−h, 0] × V × V → H be a nonlinear mapping satisfying the following: (i) For any x, y ∈ V the mapping g(·, ·, x, y) is strongly L × B-measurable; (ii) There exist positive constants L0 , L1 , L2 such that ˆ|| + L2 ||y − yˆ||, |g(t, s, x, y) − g(t, s, x ˆ, yˆ)| ≤ L1 ||x − x

(G)

|g(t, s, 0, 0)| ≤ L0 for all (t, s) ∈ [0, ∞) × [−h, 0] and x, x ˆ, y, yˆ ∈ V . Let U be some Banach space and the controller operator B be a bounded linear operator from the Banach space L2 (0, T ; U ) to L2 (0, T ; H). Let us concern with the quasi-autonomous differential equation   dx(t) + Ax(t) = k(t), 0 < t ≤ T, dt (E)  x(0) = g 0 where A is given satisfying the hypotheses mentioned in section 2. From Theorem 2.6 of Chapter III in [1] we can obtain results of the linear case result. Let g 0 ∈ H and k ∈ L2 (0, T ; V ∗ ). Then there exists a unique solution x of (E) belonging to C([0, T ]; H) ∩ L2 (0, T ; V ) ∩ W 1,2 (0, T ; V ∗ ) and satisfying   ||x||C([0,T ];H)∩L2 (0,T ;V )∩W 1,2 (0,T ;V ∗ ) ≤ C2 |g 0 | + ||k||L2 (0,T ;V ∗ ) , where C2 is a constant. Under assumptions mentioned above, we obtain the following result on the solvability of (NE) by virtue of Theorem 3.1 of [6].

448

J. M. Jeong and H. G. Kim

Lemma 2.1. Let x ∈ L2 (−h, T ; V ), T > 0. Then the nonlinear term G(·, x) defined by (2.1) belongs to L2 (0, T ; H) and n √ ||G(·, x)||L2 (0,T ;H) ≤ µ([−h, 0]) L0 T

(2.1) o

+ (L1 + L2 )||x||L2 (0,T ;V ) + L2 ||g 1 ||L2 (−h,0;V ) . Moreover, if x1 , x2 ∈ L2 (−h, T ; V ), then ||G(·, x1 ) − G(·, x2 )||L2 (0,T ;H) ≤ µ([−h, 0]) n o × (L1 + L2 )||x1 − x2 ||L2 (0,T ;V ) + L2 ||x1 − x2 ||L2 (−h,0;V ) .

(2.2)

Proof. From (G) it is easily seen that n √ ||G(·, x)||L2 (0,T ;H) ≤ µ([−h, 0]) L0 T o + L1 ||x||L2 (0,T ;V ) + L2 ||x||L2 (−h,T ;V ) n √ o ≤ µ([−h, 0]) L0 T + (L1 + L2 )||x||L2 (0,T ;V ) + L2 ||x||L2 (−h,0;V ) . The proof of (2.2) is similar.  Proposition 2.1. Let the assumptions (A), and (G) be satisfied. Then, for every u ∈ L2 (0, T ; U ) and (g 0 , g 1 ) ∈ H × L2 (0, T ; V ) the equation (NE) has a unique solution x ∈ L2 (0, T ; V ) ∩ C([0, T ]; H) ∩ W 1,2 (0, T ; V ∗ ) and there exists a constant C1 depending on T such that   ||x||L2 ∩C∩W 1,2 ≤ C1 1 + |g 0 | + ||g 1 ||L2 (0,T ;V ) + ||u||L2 (0,T ;U ) .

(2.3)

The mapping H × L2 (0, T ; V ) × L2 (0, T ; V ∗ ) 3 (g 0 , g 1 , k) 7→ x ∈ L2 (0, T ; V ) ∩ C([0, T ]; H) is continuous. Futhermore, there exists a constant C > 0 such that ||x1 − x2 ||L2 (0,T ;V )∩C([0,T ];H)   ≤ C |g10 − g20 | + ||g11 − g22 ||L2 (−h,0;V ) + ||k1 − k2 ||L2 (0,T1 ;V ∗ )

(2.4)

Control problems for nonlinear retarded functional differential equations

449

3. Control problems for nonlinear case In this section we are interested in the approximate controllability for the nonlinear functional control system (NE) on H. In what follows we assume that the embedding V ⊂ H is compact. For h ∈ L2 (0, T ; H) and let xh be the solution of the following equation with B = I:   dx(t) + Ax(t) = G(t, x) + h(t), 0 < t, (CP) dt  x(0) = 0, x(s) = 0 − h ≤ s ≤ 0. Theorem 3.1. Let us define the solution mapping S from L2 (0, T ; V ∗ ) to L2 (0, T ; V ) ∩ W 1,2 (0, T ; V ∗ ) by (Sh)(t) = xh (t),

h ∈ L2 (0, T ; V ∗ ).

(3.1)

Then, the mapping h 7→ Sh = xh is compact from L2 (0, T ; V ∗ ) to L2 (0, T ; H) Proof. With the aid of Lemma 2.1 and Proposition 2.1 ||Sh||L2 (0,T ;V )∩W 1,2 (0,T ;V ∗ ) = ||xh || ≤ C2 ||G(·, xh ) + h||L2 (0,T ;V ∗ ) o n √ ≤ C2 µ([−h, 0]) L0 T + (L1 + L2 )||x||L2 (0,T ;V ) + C2 ||h||L2 (0,T ;V ∗ ) h √ i ≤ C2 µ([−h, 0]) L0 T + (L1 + L2 ){C1 (1 + ||h||L2 (0,T ;V ∗ ) )} + C2 ||h||L2 (0,T ;V ∗ ) . Hence, if h is bounded in L2 (0, T ; V ∗ ), then so is xh in L2 (0, T ; V )∩W 1,2 (0, T ; V ∗ ). Since V is compactly embedded in H by assumption, the embedding L2 (0, T ; V ) ∩ W 1,2 (0, T ; V ∗ ) ⊂ L2 (0, T ; H) is compact in view of Theorem 2 of J. P. Aubin [5]. Hence, the mapping h 7→ Sh = xh is compact from L2 (0, T ; V ∗ ) to L2 (0, T ; H).  Let A and G be the Nemitsky operators corresponding to the maps A and G, which are defined by A(x)(·) = Ax(·) and G(h)(·) = G(·, xh ), respectively. Then since the solution x belongs to L2 (−h, T ; V ) ∩ W 1,2 (0, T ; V ∗ ) ⊂ C([0, T ]; H), it is represented by xh (t) =

Z

t

((I + G − AS)h)(s)ds, 0

(3.2)

450

J. M. Jeong and H. G. Kim

From Theorem 4.1, it follows that G is a compact mapping from L2 (0, T ; V ∗ ) to L2 (0, T ; H) and so is AS from L2 (0, T ; V ∗ ) to itself. The solution of (NE) is denoted by x(T ; g, u) associated with the nonlinear term g and a control u at the time T . Definition 3.1. The system (NE) is said to be approximately controllable at time T if the reachable set {x(T ; g, u) : u ∈ L2 (0, T ; U )} is dense in H, that is, Cl{x(T ; g, u) : u ∈ L2 (0, T ; U )} = H. We assume that (G1) g is uniformly bounded: there exists a constant Mg such that |g(t, s, x, y)| ≤ Mg , for all x, y ∈ V . Then it holds that |G(t, x)| ≤ Mg µ([−h, 0]), and hence √ ||G(h)||L2 (0,T ;H) = ||G(·, xh )||L2 (0,T ;H) ≤ Mg µ([−h, 0]) T .

(3.3)

Theorem 3.2. Let the assumptions (G1) and let us assume that

Then

o n Cl y : y(t) = Bu(t), a.e. u ∈ L2 (0, T ; U ) = L2 (0, T ; H).

(B)

n o Cl (G + I − AS)h : h ∈ L2 (0, T ; V ∗ ) = L2 (0, T ; V ∗ ).

(3.4)

Thus, the system (CP) is approximately controllable at time T . Proof. Let us fix T0 > 0 such that  ω3  2ω2 T0 e − 1 > 0. 4ω1 ω2 Let z ∈ L2 (0, T0 ; V ∗ ) and r be a constant such that n o z ∈ Ur = x ∈ L2 (0, T0 ; V ∗ ) : ||x||L2 (0,T0 ;V ∗ ) < r . Putting N 2 :=

 1  2ω2 T0 −1 , e 4ω1 ω2

(3.5)

Control problems for nonlinear retarded functional differential equations

Take a constant d > 0 such that o n √ r + ω3 N Mg µ([−h, 0]) T 0 + ω3 (1 − ω3 N )−1 < d.

451

(3.6)

Noting that L2 (0, T0 ; H) is a dense subspace of L2 (0, T0 ; V ∗ ), for every h ∈ L (0, T0 ; V ∗ ), consider the following equation:   dx(t) + Ax(t) = G(t, x) + h(t), 0 < t, (3.7) dt  x(0) = 0, x(s) = 0 − h ≤ s ≤ 0. 2

Taking scalar product on both sides of (3.7) by x(t) 1 d |x(t)|2 + ω1 ||x(t)||2 ≤ ω2 |x(t)|2 + |G(t, x) + h(t)| |x(t)|. 2 dt Integrating on [0, t], we get Z t 1 |x(t)|2 + ω1 ||x(s)||2 ds ≤ 2 0 Z t e2ω2 (t−s) |G(s, x) + h(s)| |x(s)|ds, +

(3.8)

0

and e−ω2 t |x1 (t)| ≤

Z

t

e−ω2 s |G(s, x) + h(s)|ds.

(3.9)

0

Putting H1 (s) = |G(s, x) + h(s)| and combining (3.8) with (3.9) we obtain Z t 1 2 |x(t)| + ω1 ||x(s)||2 ds ≤ 2 0 Z t Z s e2ω2 (t−s) H1 (s) eω2 (s−τ ) H1 (τ )dτ ds 0 0 Z t Z s e−ω2 s H1 (s) e−ω2 τ H1 (τ )dτ ds = e2ω2 t Z

0 t

Z

0 s

1 d e−ω2 τ H1 (τ )dτ 2 ds 0 0 Z t 2 1 = e2ω2 t e−ω2 τ H1 (τ )dτ 2 0 Z t Z t 1 ≤ e2ω2 t e−2ω2 τ dτ H1 (τ )2 dτ 2 0 0 Z t 1 2ω2 t (e − 1) H1 (s)2 ds. = 4ω2 0 = e2ω2 t

2

ds

(3.10)

452

J. M. Jeong and H. G. Kim

Noting that √ ||G(·, x)||L2 (0,T0 ;H) ≤ Mg µ([−h, 0]) T 0 , it follows from (3.10) that ||Sh||L2 (0,T0 ;V ) = ||x||L2 (0,T0 ;V )

(3.11)

√ ≤ N (||h||L2 (0,T0 ;V ∗ ) + Mg µ([−h, 0]) T 0 ).

In order to prove (3.4) we will use the topological degree theory for the equation z = λ(G − AS)h + h, 0 ≤ λ ≤ 1. (3.12) Let h be the solution of (3.12). Then since z ∈ Ur and from (3.11), (A) we have ||h||L2 (0,T ;V ∗ ) ≤ ||z|| + ||ASh|| + ||Gh|| √ ≤ r + ω3 (||Sh|| + 1) + Mg µ([−h, 0]) T 0 n p o ≤ r + ω3 N ||h||L2 (0,T0 ;V ∗ ) + Mg µ([−h, 0]) T0 + ω3 p + Mg µ([−h, 0]) T0 , and hence, o n √ ||h|| ≤ r + ω3 N Mg µ([−h, 0]) T 0 + ω3 (1 − ω3 N )−1 < d. It follows that h ∈ / ∂Ud where ∂Ud stands for the boundary of Ud . Thus, from the homotopy property of topological degree theory, there exists h ∈ Ud such that the equation z = (G + I − AS)h holds. By virtue of the assumption (B), there exists a sequence {un } in L2 (0, T0 ; U ) such that Bun 7→ h in L2 (0, T0 ; V ∗ ). Then by Proposition 2.1 we have that x(·; Bun ) 7→ xh in L2 (0, T0 ; V ) ∩ C([0, T0 ]; H). Let y ∈ H. We can choose g ∈ W 1,2 (0, T0 ; V ∗ ) such that g(0) = x0 and g(T0 ) = y and from the equation 0 (3.12) there is h ∈ L2 (0, T0 ; V ∗ ) such that g = (G+I −AS)h. By the assumption (B) and Proposition 2.1 there exists u ∈ L2 (0, T0 ; U ) such that ||xh − xBu ||L2 (0,T0 ;V )∩C([0,T0 ];H) ≤ C3 ||h − Bu||L2 (0,T0 ;V ∗ ) for some constant C3 > 0. Thus, we have Z T0 n o |y − xh (T )| = ((G + I − AS)h)(s) − ((G + I − AS)Bu)(s) ds 0

≤ ||xh − xBu ||L2 (0,T0 ;V )∩C([0,T ];H) ≤ C1 ||h − Bu||L2 (0,T ;V ∗ ) .

Control problems for nonlinear retarded functional differential equations

453

Therefore, the system (CP) is approximately controllable at time T0 . Since the condition (3.5) is independent of initial values, we can solve the equation in [T0 , 2T0 ] with the initial value x(T0 ). By repeating this process, the approximate controllablility for (NE) can be extended the interval [0, nT0 ] for natural number n, i.e., for the initial x(nT0 ) in the interval [nT0 , (n + 1)T0 ]. 

References 1. V. Barbu, Nonlinear Semigroups and Differential Equations in Banach space, Noordhoff Leiden, Netherland, 1976. 2. J. M. Jeong, Y. C. Kwun and J. Y. Park, Approcimate controllability for semilinear retarded functional differential equations, J. Dynamics and Control Systems 5(3) (1999), 329–346. 3. I. I. Vrabie, An existence result for a class of nonlinear evolution equations in Banach spaces, Nonlinear Analysis, T. M. A. 7 (1982), 711–722. 4. N. U. Ahmed and X. Xiang, Existence of solutions for a class of nonlinear evolution equations with nonmonotone perturbations, Nonlinear Analysis, T. M. A. 22(1) (1994), 81–89. 5. J. P. Aubin, Un th` eor´ eme de compasit´ e, C. R. Acad. Sci. 256 (1963), 5042–5044. 6. H. Br´ezis, Op´ erateurs Maximaux Monotones et Semigroupes de Contractions dans un Espace de Hilbert, North Holland, 1973. 7. N. Hirano, Nonlinear evolution equations with nonmonotonic perturbations, Nonlinear Analysis, T. M. A. 13(6) (1989), 599–609.

Jin-Mun Jeong received his BS from Kyungnam University( MSC at Pusan National University) and Ph. D. at Osaka University in Japan. Since 1992 he has been at Pukyong National University and became Professor in 2004. His research center on evolution equations, nonlinear functional differential equations and its related control problems. Division of Mathematical Sciences, Pukyong National University, Pusan 608-737, Korea e-mail: [email protected] Han-Geul Kim received his BS and MSC from Pukyong National University(PKNU)in 2006 and continuous her study in PKNU under the direction of Jin-Mun Jeong. Her interests cover partial differential equations and their optimal control problems. Division of Mathematical Sciences, Pukyong National University, Pusan 608-737, Korea e-mail: [email protected]

Suggest Documents