Electronic Journal of Differential Equations, Vol. 2004(2004), No. 35, pp. 1–12. ISSN: 1072-6691. URL: http://ejde.math.txstate.edu or http://ejde.math.unt.edu ftp ejde.math.txstate.edu (login: ftp) EXISTENCE OF SOLUTIONS TO THE ROSENAU AND BENJAMIN-BONA-MAHONY EQUATION IN DOMAINS WITH MOVING BOUNDARY RIOCO K. BARRETO, CRUZ S. Q. DE CALDAS, PEDRO GAMBOA, & JUAN LIMACO Abstract. In this article, we prove the existence of solutions for a hyperbolic equation known as the the Rosenau and Benjamin-Bona-Mahony equations. We study increasing, decreasing, and mixed non-cylindrical domains. Our main tools are the Galerkin method, multiplier techniques, and energy estimates. 1. Introduction To investigate the dynamics of certain discrete systems, Philip Rosenau obtained the equation ut + (u + u2 )x + uxxxxt = 0. The study of this equation in cylindrical domains was done by Mi Ai Park [13], who proved the existence and uniqueness of local and global solutions. The Rosenau equation could be seen as a variant of Benjamin-Bona-Mahony (BBM) equation, ut + (u + u2 )x − uxxt = 0, which models long waves in a non linear dispersive system. In [3], Benjamin-Bona-Mahony proved the existence and uniqueness of global solutions for the BBM equation in cylindrical domains. In this work, we study the existence of solutions for the Rosenau and BBM equations for increasing, decreasing, and mixed noncylindrical domains. We introduce the following notation: Let α, β, γ = β − α, be C 2 -functions of a real variable, such that α(t) < β(t), for all t ≥ 0. We represent the noncylindrical domain by b = {(x, t) ∈ R2 : α(t) < x < β(t), ∀t ≥ 0}, Q b=S and its lateral boundary by Σ 0≤t≤T {α(t), β(t)} × {t}. In the present work we investigate the following two equations: ut + (u + u2 )x + uxxxxt = 0 b in Q u(x, t) = 0 b for (x, t) ∈ Σ ux (x, t) = 0 b for (x, t) ∈ Σ u(x, 0) = u0 (x) for α(0) < x < β(0) 1991 Mathematics Subject Classification. 35M10, 35B30. Key words and phrases. Benjamin-Bona-Mahony equation, Rosenau equation, noncylindrical domains. c 2004 Texas State University - San Marcos. Submitted April 28, 2003. Published March 11, 2004. 1 (1.1) 2 R. K. BARRETO, C. DE CALDAS, P. GAMBOA, & J. LIMACO and ut + (u + u2 )x − uxxt = 0 u(x, t) = 0 EJDE-2004/35 b in Q b for (x, t) ∈ Σ (1.2) 0 u(x, 0) = u (x) in Ω0 . This paper is organized as follows: The next section is devoted to the existence and uniqueness of solution for (1.1) and (1.2), satisfying the hypothesis (H1) α0 (t) ≥ 0 and β 0 (t) ≤ 0 for t ∈ [0, T ]. b decreases in the sense that if t2 > t1 , then the Note that this hypothesis implies Q projection of [α(t2 ), β(t2 )] in the subspace t = 0 is contained in the projection of [α(t1 ), β(t1 )] in the same subspace. In the third section of this article, we study the existence of solutions for (1.1) and (1.2) satisfying the hypothesis (H2) α0 (t) ≤ 0 and β 0 (t) ≥ 0 for t ∈ [0, T ]. b increases Analogously hypothesis (H2) implies that Q In the last section of this article, we study the (1.1) and (1.2), satisfying the hypothesis: b=Q c1 ∪ Q c2 where Q c1 is increasing and Q c2 is decreasing. (H3) Q In the following, by Ω we represent the interval ]0, 1[, Ωt and Ω0 denote the intervals ]α(t), β(t)[ and ]α(0), β(0)[ respectively. We denote, as usual, by (., .), k · k respectively the scalar product and norm in L2 (Ω). In the sequel, wm,x denotes ∂wm ∂ 2 wm ∂ 2 wm ∂x , analogously wm,xx = ∂x2 , wm,xt = ∂t ∂x , etc. 2. Solutions on decreasing domains In this section we study the existence and uniqueness for (1.1) and (1.2) satisfying the hypothesis (H1). Let γ(t) = β(t) − α(t) > 0, for all t ≥ 0. Then 0 < x−α(t) γ(t) < 1, for all t ∈ [0, T ]. With the change of variable u(x, t) = v(y, t) where y = all t ∈ [0, T ], problem (1.1) is transformed into vt + 1 (α0 + γ 0 y) 4γ 0 1 (v + v 2 )y + 4 vyyyyt − vy − 5 vyyyy γ γ γ γ (α0 + γ 0 y) − vyyyyy = 0 in Ω×]0, T [ γ5 v(0, t) = v(1, t) = 0 in ]0, T [ vy (0, t) = vy (1, t) = 0 x−α(t) γ(t) , for (2.1) in ]0, T [ 0 v(y, 0) = v (y) in Ω . Also problem (1.2) is transformed into vt + 1 1 (α0 + γ 0 y) 2γ 0 vy + 3 vyy (v + v 2 )y − vyyt − γ γ2 γ γ (α0 + γ 0 y) + vyyy = 0 in Ω×]0, T [ γ3 v(0, t) = v(1, t) = 0 in ]0, T [ v(y, 0) = v 0 (y) in Ω . Under these conditions, we establish the following existence results. (2.2) EJDE-2004/35 ROSENAU AND BENJAMIN-BONA-MAHONY EQUATIONS 3 Theorem 2.1. For each u0 ∈ H02 (Ω0 ) ∩ H 4 (Ω0 ), there exists a unique function b → R, satisfying u ∈ C 1 ([0, T ]; H 2 (Ωt )) ∩ C(0, T ; H 3 (Ωt ) ∩ H 2 (Ωt )) and u:Q 0 0 Z Z Z ut φ dx dt + (u + u2 )x φ dx dt + uxxt φxx dx dt = 0, b Q 2 for all φ ∈ L (0, T ; H02 (Ωt )), b Q b Q 0 u(x, 0) = u (x), for all x ∈ Ω0 . Theorem 2.2. For each u0 ∈ H01 (Ω0 ) ∩ H 2 (Ω0 ), there exists a unique function b → R, satisfying u ∈ L∞ (0, T ; H 1 (Ωt )), ut ∈ L∞ (0, T ; H 1 (Ωt )) and u:Q 0 0 Z Z Z ut φ dx dt + (u + u2 )x φ dx dt + uxt φx dx dt = 0, b Q 2 for all φ ∈ L (0, T ; H01 (Ωt )), b Q b Q 0 u(x, 0) = u (x), for all x ∈ Ω0 . To prove Theorem 2.1, we need the following lemmas. Lemma 2.3. For each v 0 ∈ H02 (Ω) ∩ H 4 (Ω), there exists a unique function v : Ω×]0, T [→ R, satisfying v ∈ L∞ (0, T ; H02 (Ω) ∩ H 4 (Ω)), vt ∈ L∞ (0, T ; H02 (Ω)), and Z 1 1 [vt ψ + (v + v 2 )y ψ + 4 vyyt ψyy γ γ Ω×]0,T [ − (α0 + γ 0 y) 4γ 0 (α0 + γ 0 y) vy ψ − 5 vyy ψyy + ( ψ)y vyyyy ]dy dt = 0, γ γ γ5 for all ψ ∈ L2 (0, T ; H02 (Ω)), v(y, 0) = v 0 (y), for all y ∈ Ω. Lemma 2.4. For each f ∈ C([0, T ]; H −2 (Ω)), there exists a unique function z : Ω×]0, T [→ R, satisfying z ∈ C([0, T ]; H02 (Ω)) and z + γ14 ∆2 z = f Lemma 2.5. For each v 0 ∈ H02 (Ω) ∩ H 4 (Ω), there exists a unique function v : Ω×]0, T [→ R, satisfying v ∈ C 1 ([0, T ]; H02 (Ω)) ∩ C([0, T ]; H 3 (Ω) ∩ H02 (Ω)) and Z 1 1 [vt ψ + (v + v 2 )y ψ + 4 vyyt ψyy γ γ Ω×]0,T [ − (α0 + γ 0 y) 4γ 0 (α0 + γ 0 y)ψ vy ψ − 5 vyy ψyy + ( )y vyyyy ] dy dt = 0 , γ γ γ5 for all ψ ∈ L2 (0, T ; H02 (Ω)); v(y, 0) = v 0 (y), for all y ∈ Ω. Lemma 2.6. For each v 0 ∈ H01 (Ω) ∩ H 2 (Ω), there exists a unique function v : Ω×]0, T [→ R, satisfying v ∈ L∞ (0, T ; H01 (Ω) ∩ H 2 (Ω)), vt ∈ L∞ (0, T ; H01 (Ω)) and Z 1 1 [vt ψ + (v + v 2 )y ψ − 2 vyyt ψ γ γ Ω×]0,T [ − (α0 + γ 0 y) 2γ 0 (α0 + γ 0 y)ψ vy ψ + 3 vyy ψ − ( )y vyy ]dy dt = 0, γ γ γ3 for all ψ ∈ L2 (0, T ; H01 (Ω)); v(y, 0) = v 0 (y), for all y ∈ Ω. In this article, we prove Theorem 2.1 and Lemmas 2.3, 2.4, 2.5 which correspond to Rosenau Equation. However, we omit the proofs of Theorem 2.2 and Lemma 2.6 which correspond to Benjamin Bona-Mahony Equation; because the proofs are made in a similar way. 4 R. K. BARRETO, C. DE CALDAS, P. GAMBOA, & J. LIMACO EJDE-2004/35 Proof of Lemma 2.3. Let (wi )i∈N be the special basis of H02 (Ω), such that wi,yyyy = λi wi , in Ω wi (0) = wi (1) = wi,y (0) = wi,y (1) = 0, i ∈ N. We denote by Vm the subspace generated by w1 , . . . , wm . Our starting point is to construct the Galerkin approximation of the solution vm ∈ Vm , which is given by the solution of the approximate equation 1 4γ 0 1 2 )y , w) + 4 (vm,yyyyt , w) − 5 (vm,yyyy , w) (vm,t , w) + ( (vm + vm γ γ γ 0 0 0 0 (α + γ y) (α + γ y) +(− vm,yyyyy , w) + (− vm,y , w) = 0 for all w ∈ Vm γ5 γ 0 vm (0) = vm → v 0 in H 4 (Ω) (2.3) First Estimate. Taking w = vm (t) in (V)1, we have: d 1 γ0 (kvm (t)k2 + 4 kvm,yy (t)k2 ) + 5 kvm,yy (t)k2 dt γ γ 0 0 γ α 2 β0 2 (0) − 5 vm,yy (1) = 0 + kvm (t)k2 + 5 vm,yy γ γ γ (2.4) Integrating this equation over [0, t] and using hypothesis (H1), we obtain 1 kvm,yy (t)k2 γ4 Z Z γ2 t 1 γ2 t 1 0 2 2 k + kv (s)k ds + kvm (s)k2 ds ≤ kv 0 k2 + 4 kvyy m,yy γ (0) γ0 0 γ 4 γ05 0 kvm (t)k2 + (2.5) where γ0 , γ1 , γ2 are positive constants, such that γ0 ≤ γ(t) ≤ γ1 , and γ2 = max |γ 0 (t)|, γ1 = max |α0 (t)|. 0≤t≤T 0≤t≤T This implies kvm (t)k2 + 1 kvm,yy (t)k2 ≤ C0 + C1 γ4 Z 0 t [kvm (s)k2 + 1 kvm,yy (s)k2 ] ds γ4 (2.6) where C0 , C1 , . . . denote positive constants. Applying Gronwall inequality, we have the first estimate 1 kvm (t)k2 + 4 kvm,yy (t)k2 ≤ C2 (2.7) γ Second Estimate Taking w = vm,yyyy (t) in the first equation of (2.3), we obtain 1 d 1 [kvm,yy (t)k2 + 4 kvm,yyyy (t)k2 ] 2 dt γ 3γ2 1 1 ≤ kvm,yyyy (t)k2 + kvm,y (t)kkvm,yyyy (t)k 4 2γ0 γ γ (α1 + γ2 ) + kvm,y (t)k2 kvm,yyyy (t)k. γ0 (2.8) EJDE-2004/35 ROSENAU AND BENJAMIN-BONA-MAHONY EQUATIONS 5 From (2.7) and using Schwartz’s inequality and Poincare’s inequality, we obtain 1 d 1 [kvm,yy (t)k2 + 4 kvm,yyyy (t)k2 ] 2 dt γ 1 γ12 3γ2 1 2 2 ≤ kv (t)k + kv (t)k + C3 kvm,yy (t)k2 m,yyyy m,yyyy 2γ0 γ 4 2γ 4 2 + C3 kvm,yy (t)k2 kvm,yyyy (t)k 1 (1 + 2 1 ≤ (1 + 2 ≤ Let c4 = 3γ2 1 ) kvm,yyyy (t)k2 + γ0 γ 4 3γ2 1 ) kvm,yyyy (t)k2 + γ0 γ 4 (α1 +γ2 ) C3 C2 γ0 and let c5 = γ12 (α1 + γ2 ) C3 C2 + C3 C2 kvm,yyyy (t)k 2 γ0 γ12 γ2 1 C3 C2 + 1 C42 + 4 kvm,yyyy (t)k2 . 2 2 2γ γ12 2 C3 C2 + γ12 2 2 C4 . Then d 1 3γ2 1 [kvm,yy (t)k2 + 4 kvm,yyyy (t)k2 ] ≤ C5 + (2 + ) kvm,yyyy (t)k2 . dt γ γ0 γ 4 (2.9) Integrating this inequality over [0, t] and applying Gronwall inequality, we obtain kvm,yy (t)k2 + 1 kvm,yyyy (t)k2 ≤ C6 γ4 (2.10) Third Estimate Taking w = vm,t (t) in (V)1 , we have kvm,t (t)k2 + 1 kvm,yyt (t)k2 γ4 1 4γ 0 2 = − ((vm (t) + vm (t))y , vm,t (t)) + 5 (vm,yyyy (t), vm,t (t)) γ γ (α0 + γ 0 y) (α0 + γ 0 y) −( v (t), v (t)) + ( vm,y (t), vm,t (t)). m,yyyy m,yt γ5 γ5 (2.11) ¿From (2.7), (2.10), and (2.11), we obtain kvm,t (t)k2 + 1 kvm,yyt (t)k2 ≤ C7 . γ4 (2.12) These three estimates permit to pass to the limit in the approximate equation and we obtain a weak solution v in the sense of Lemma 2.3. The uniqueness of solution and the verification of initial data are showed by the standard arguments. Proof of Lemma 2.4. To prove the existence we consider two stages: First stage f ∈ C([0, T ]; H02 (Ω)). Let (wi )i∈N be the special basis of H02 (Ω) Pnused in the proof of Lemma 2.3. Consider the sequence (fn ), such that fn (t) = i=1 (f (t), wi )wi . It is clear that fn → f strongly in C([0, T ]; H02 (Ω)). Pm The approximated solution zm (t) to z + γ14 ∆2 z = f is zm (t) = i=1 gim (t)wi , where gim are solutions of the approximated system (zm (t), wi ) + 1 (∆zm (t), ∆wi ) = (fm (t), wi ) , γ4 i = 1, . . . m (2.13) A priori estimate. Let us prove that (zm ) is a Cauchy sequence in C([0, T ]; H02 (Ω)). In fact, let m and n be positive integer such that m > n and gin (t) = 0 for 6 R. K. BARRETO, C. DE CALDAS, P. GAMBOA, & J. LIMACO EJDE-2004/35 n ≤ i ≤ m. Then zm and zn are solutions of (2.13) in Vm . Consider the Cauchy difference zm − zn . We have from (2.13) that, for i = 1, . . . , m, 1 (zm (t) − zn (t), wi ) + 4 (∆(zm (t) − zn (t)), ∆wi ) = (fm (t) − fn (t), wi ) , (2.14) γ Taking wi = zm (t) − zn (t) in (2.14), using the Cauchy-Schwarz inequality and the equivalent norms, we obtain |zm − zn |C([0,T ];H02 (Ω)) ≤ c|fm − fn |C([0,T ];H02 (Ω)) Then zn → z strongly in C([0, T ]; H02 (Ω)). Therefore, taking limit in (2.13), we obtain z + γ14 ∆2 z = f in C([0, T ]; H −2 (Ω)). Second stage f ∈ C([0, T ]; H −2 (Ω)). By density, there exists a sequence (fn ), fn ∈ C([0, T ]; H02 (Ω)), such that fn → f strongly in C([0, T ]; H −2 (Ω)). Using the first stage we have that there exist a sequence (zn ), zn ∈ C([0, T ]; H02 (Ω)) such that 1 2 ∆ zn = fn in C([0, T ]; H −2 (Ω)). γ4 Consider the Cauchy difference zm − zn , m > n. We obtain 1 zm− zn + 4 ∆2 (zm − zn ) = fm − fn in C([0, T ]; H −2 (Ω)). γ zn + (2.15) (2.16) Composing (2.16) with zm − zn ∈ C([0, T ]; H02 (Ω)). and integrating in Ω, we have |zm − zn |C([0,T ];H02 (Ω)) ≤ c|fm − fn |C([0,T ];H −2 (Ω)) ; therefore, zn → z strongly in C([0, T ]; H02 (Ω)) and taking limit in (2.15) we have z + γ14 ∆2 z = f in C([0, T ]; H −2 (Ω)). The uniqueness of the solutions is showed by the standard arguments. Proof of Lemma 2.5. From Lemma 2.4 we can define the operator B(t) = (I + 1 2 −1 from C([0, T ]; H −2 (Ω)) to C([0, T ]; H02 (Ω)) by B(t)f = z with f ∈ γ4 ∆ ) C([0, T ]; H −2 (Ω)) where z is a solution of z + γ14 ∆2 z = f . Note that B(t) is linear and continuous. By Lemma 2.3, v ∈ L2 (0, T ; H 4 (Ω) ∩ H02 (Ω)) and vt ∈ L2 (0, T ; H02 (Ω)). From Lions-Magenes, Theorems 3.1 and 9.6, chapter I [10], we conclude that v ∈ C([0, T ]; H02 (Ω) ∩ H 3 (Ω)) On the other hand, from the transformed problem (2.1), we obtain 1 (I + 4 ∆2 )vt = f , γ where, 1 (α0 + γ 0 y) 4γ 0 (α0 + γ 0 y) f = − (v + v 2 )y + vy + 5 vyyyy + ( )vyyyyy γ γ γ γ5 (2.17) (2.18) From (2.17), we conclude that f ∈ C([0, T ]; H −2 (Ω)). Then from (2.18) we have that vt = B(t)f , where vt ∈ C([0, T ]; H02 (Ω)) and we get the required result. The proof of Theorem 2.1 follows immediately from Lemma 2.5 and the Change of Variable Theorem. Therefore, we omit it. Observe that Theorem 2.1 in a cylindrical domain, has the regularity u ∈ C 1 ([0, T ]; H02 (Ω)) ∩ C([0, T ]; H 4 (Ω) ∩ H02 (Ω)). In fact, as we consider an additional estimate with wi = um,txxxx , in the Galerkin approximation, that allows us EJDE-2004/35 ROSENAU AND BENJAMIN-BONA-MAHONY EQUATIONS 7 to obtain the regularity ut ∈ L2 (0, T ; H 4 (Ω) ∩ H02 (Ω)). However, in our noncylindrical domain, this is not possible since the transformed problem (III), contains a term vyyyyy that does not allow us to use the estimate with wi = vm,tyyyy in the Galerkin approximation. 3. Solutions on increasing domains In this section we study the existence of solution for the systems (1.1) and (1.2) satisfying the hypothesis (H2). We use the Penalization Method given by Lions b ⊂ Q. We define the function [10]. Let Q =]a, b[×]0, T [ be the cylinder such that Q M : Q → R, by ( b 1 in Q \ Q M (x, t) = b 0 in Q To show the existence result we will use the following Lemma. Lemma 3.1. If u, ut ∈ L2 (0, T ; L2 (a, b)), then Z t 1 1 (M u(s), ut (s))ds ≥ kM (t)u(t)k2L2 (a,b) − kM (0)u(0)k2L2 (a,b) . 2 2 0 Proof. We have Z t 0 Z Z 1 t b (M u(s), ut (s))ds = M (u2 (s))t dξ ds 2 0 a Z 1 = M (u2 (s))t dξ ds. 2 [0,t]×[a,b] From Fubini’s Theorem and recalling the definition of M , it follows that Z t (M u(s), ut (s))ds 0 Z Z −1 1 α(0) α (x) 2 [u (s)]t ds dξ + [u (s)]t ds dξ 2 α(t) 0 a 0 Z Z −1 Z Z t 1 β(t) β (x) 2 1 b + [u (s)]t ds dξ + [u2 (s)]t ds dξ 2 β(0) 0 2 β(t) 0 Z Z 1 α(t) 2 1 α(0) 2 −1 2 = [u (t, ξ) − u (0, ξ)] dξ + [u (α (x), 0) − u2 (0, ξ)] dξ 2 a 2 α(t) Z Z 1 β(t) 2 −1 1 b + [u (β (x), 0) − u2 (0, ξ)] dξ + [u2 (t, ξ) − u2 (0, ξ)] dξ 2 β(0) 2 β(t) Z Z b Z α(t) 1 α(t) 2 ≥ [ u (t, ξ) dξ + u2 (t, ξ) dξ − ( u2 (0, ξ) dξ 2 a β(t) a Z α(0) Z β(t) Z b + u2 (0, ξ) dξ + u2 (0, ξ) dξ + u2 (0, ξ) dξ)] 1 = 2 Z α(t) α(t) Z b Z t 2 β(0) β(t) Z b 1 1 M (t, ξ) u2 (t, ξ) dξ − M (0, ξ)u2 (0, ξ) dξ 2 a 2 a 1 1 = kM (t)u(t)k2L2 (a,b) − kM (0)u(0)k2L2 (a,b) 2 2 = 8 R. K. BARRETO, C. DE CALDAS, P. GAMBOA, & J. LIMACO EJDE-2004/35 which completes the proof. The existence of solution for (1.1) and (1.2), satisfying the hypothesis (H2), is established in the next theorems. b → R, satisfying Theorem 3.2. For each u0 ∈ H02 (Ω0 ), there exists a function u : Q ∞ 2 ∞ 2 u ∈ L (0, T ; H0 (Ωt )), ut ∈ L (0, T ; H0 (Ωt )) and Z Z Z ut φ dx dt + (u + u2 )x φ dx dt + uxxt φxx dx dt = 0 (3.1) b Q b Q b Q for all φ ∈ L2 (0, T ; H02 (Ωt )); u(x, 0) = u0 (x) b → R, satisfying Theorem 3.3. For each u0 ∈ H01 (Ω0 ), there exists a function u : Q ∞ 1 ∞ 1 u ∈ L (0, T ; H0 (Ωt )), ut ∈ L (0, T ; H0 (Ωt )) and Z Z Z 2 ut φ dx dt + (u + u )x φ dx dt + uxt φx dx dt = 0, b Q b Q b Q for all φ ∈ L2 (0, T ; H01 (Ωt )); u(x, 0) = u0 (x) Proof of Theorem 3.2. To prove this result we use the penalization method. For each > 0 we consider the problem 1 1 u,t + (u + u2 )x + u,xxxxt + M u,t − (M u,xt )x = 0 in Q u (a, t) = u (b, t) = u,x (a, t) = u,x (b, t) = 0 in ]0, T [ u (x, 0) = u e0 (x) (3.2) in ]a, b[ Let {wi }i∈N be a basis of H02 (a, b), such that w1 = u e0 . We denote by Vm = [w1 , . . . , wm ] the subspace of H02 (a, b), generated by u e0 , w2 , . . . , wm . We seek um (t) in Vm solution to the approximate problem (um,t , w) + ((um + u2m )x , w) + (um,xxxxt , w) 1 1 + (M um,t , w) − ((M um,xt )x , w) = 0 for all w ∈ Vm um (0) = u0 (x) → u e0 in H02 (a, b) (3.3) First Estimate. Taking w = um in (3.3) and applying Lemma 3.1, we obtain 1 1 k|um (t)k|2 + k|um,xx (t)k|2 + k|M (t)um (t)k|2 + k|M (t)um,x (t)k|2 ≤ c8 , (3.4) 2 where k| · k| denotes the norm in L (a, b). Second Estimate. Taking w = um,t (t) in (3.3) and using (3.4) we have 1 k|um,t (t)k|2 + k|um,xxt (t)k|2 + (M (t)um,t (t), um,t (t)) 1 (3.5) (M (t)um,xt (t), um,xt (t)) 1 ≤ c9 + k|um,t (t)k|2 2 From where we obtain 1 1 k|um,t (s)k|2 + k|um,xxt (s)k|2 + k|M (t)um,t (t)k|2 + k|M (t)um,xt (t)k|2 ≤ c9 EJDE-2004/35 ROSENAU AND BENJAMIN-BONA-MAHONY EQUATIONS 9 From the estimates above, we pass to the limit in the approximate equation, and we obtain that u is solution of the penalized problem Z TZ b Z TZ b Z TZ b u,t v dx dt + (u + u2 )x v dx dt + u,xxt vxx dx dt 0 a 0 a 0 a (3.6) Z Z Z Z 1 T b 1 T b + M u,t v dx dt + M u,xt vx dx dt = 0 0 a 0 a for all v ∈ L2 (0, T ; H02 (a, b)). From (3.4), (3.5) and the Banach-Steinhauss Theorem, we pass to the limit as → 0 in (3.6) and we obtain (3.1). Regularity. From the first estimate, we have Z 1 t (M um,t (s), um (s)) ds ≤ c 0 On the other hand, from Lemma 3.1 we obtain Z 1 t 1 (M um,t (s), um (s)) ds ≥ k|M (t)um (t)k|2 . 0 2 RT Rb 2 Then k|M (t)um (t)k| ≤ 2c. Thus 0 a M (t) u2m (t) dx dt ≤ 2cT or Z TZ b Z TZ b 2 |M (t)u (t)| dx dt ≤ lim inf |M (t)u (t)|2 dx dt ≤ 2cT 0 a 0 2 a 2 Then M u → 0 in L (0, T ; L (a, b)). On the other hand, M u → M u in L2 (0, T ; L2 (a, b)) and M um → M u in 2 b L (0, T ; L2 (a, b)). So, we conclude that M u = 0 a.e. in Q or u = 0 in Q \ Q. Analogously, applying the Lemma 3.1 to um,xt instead of um,t , we obtain: b Since u ∈ L∞ (0, T ; H 1 (a, b)) and ut ∈ L∞ (0, T ; H 1 (a, b)), then u ∈ ux = 0 in Q\ Q. 0 0 C([0, T ]; H01 (a, b)). Therefore, u(t) ∈ H01 (a, b) for all t and u = 0 in ]a, b[\]α(t), β(t)[. From where u(t) ∈ H01 (α(t), β(t)), for all t. Thus u ∈ L∞ (0, T ; H01 (Ωt )). Analogously, ux ∈ L∞ (0, T ; H01 (Ωt )). From these two statements, we have that u ∈ L∞ (0, T ; H02 (Ωt )). From the second estimate, Z TZ b Z TZ b 2 |M (t)um,t (t)| dx dt + |M (t)um,xt (t)|2 dx dt ≤ 2cT. 0 a 0 a By similar arguments, we obtain that ut ∈ L∞ (0, T ; H02 (Ωt )), which prove the regularity of the solution. The proof of Theorem 3.3 is similar to the proof of Theorem 3.2 and is ommitted. Remark 3.4. Theorems 2.1, 2.2, 3.2 and 3.3 are invariable by translation. In fact, the particular problem ut + (u + u2 )x + uxxxxt = 0 b ⊂ Ω×]T0 , T1 [ in Q u(x, t) = 0 b in Σ ux (x, t) = 0 b in Σ u(x, T0 ) = u0 (x) in ΩT0 , with the change of variable u(x, t) = u(x, t−T0 ), can be transformed into a problem of type (1.1). 10 R. K. BARRETO, C. DE CALDAS, P. GAMBOA, & J. LIMACO EJDE-2004/35 4. Solutions on mixed domains b is a mixed domain; i.e., Q b = B1 ∪ B2 where Here we analyze the case when Q b b B1 = {(x, t) ∈ Q : 0 < t ≤ T1 } and B2 = {(x, t) ∈ Q : T1 < t < T }, where B1 is ci by decreasing satisfying (H1), and B2 is increasing satisfying (H2). We define Q ci = int(Bi ), i = 1, 2. i.e., Q c1 = {(x, t) ∈ Q b : 0 < t < T1 } and Q c2 = {(x, t) ∈ Q; b Q T1 < t < T }. b=Q c1 ∪ Q c2 , we consider the following two cases: To find a solution to (1.1) in Q 0 2 c (1) Solution on Q1 : For each u ∈ H0 (Ω0 ) ∩ H 4 (Ω0 ), by Theorem 2.1, there exist u1 solution of b1 u1,t + (u1 + u21 )x + u1,xxxxt = 0 in Q u1 (x, t) = 0 b1 inΣ u1,x (x, t) = 0 u1 (x, 0) = uo (x) ∞ (4.1) b1 inΣ in Ω0 ∞ (0, T1 ; H02 (Ωt )), satisfying u1 ∈ L u1t ∈ L (0, T1 ; H02 (Ωt )); therefore u1 is in 2 C([0, T1 ]; H0 (Ωt )). c2 : For each u0 = u1 (T1 ) ∈ H 2 (ΩT ), by Theorem 3.2, there exist (2) Solution on Q 0 1 u2 solution of b2 u2,t + (u2 + u22 )x + u2,xxxxt = 0 in Q u2 (x, t) = 0 b2 inΣ u2,x (x, t) = 0 u2 (x, T1 ) = u0 (x) ∞ (T1 , T ; H02 (Ωt )), (4.2) b2 inΣ in ΩT1 ∞ satisfying u2 in L u2t in L (T1 , T ; H02 (Ωt )), and u2 in 2 C([T1 , T ]; H0 (Ωt )). b We define u : Q b → R by (3) Solution on Q: ( c1 u1 (x, t), (x, t) ∈ Q u(x, t) = c2 , u2 (x, t), (x, t) ∈ Q where u1 and u2 are solutions of (4.1) and (4.2), respectively. Given that u0 ∈ H02 (Ω0 ), by Remark 3.4, we deduce that the function u defined above is solution of b ⊂ Ω×]0, T [ ut + (u + u2 )x + uxxxxt = 0 in Q u(x, t) = 0 b in Σ ux (x, t) = 0 b in Σ u(x, 0) = u0 (x) (4.3) in Ω0 satisfying u ∈ L∞ (0, T ; H02 (Ωt )), ut ∈ L∞ (0, T ; H02 (Ωt )) and Z Z Z 2 ut φ dx dt + (u + u )x φ dx dt + uxxt φxx dx dt = 0, b Q b Q b Q for all φ ∈ L2 (0, T ; H02 (Ωt )); u(x, 0) = u0 (x). This result is summarized in the following theorem. EJDE-2004/35 ROSENAU AND BENJAMIN-BONA-MAHONY EQUATIONS 11 b a mixed domain defined Theorem 4.1. For each u0 ∈ H02 (Ω0 ) ∩ H 4 (Ω0 ), and Q b → R satisfying u ∈ L∞ (0, T ; H 2 (Ωt )), as above, there exists a function u : Q 0 ∞ 2 ut ∈ L (0, T ; H0 (Ωt )) and Z Z Z 2 ut φ dx dt + (u + u )x φ dx dt + uxxt φxx dx dt = 0, b Q 2 for all φ ∈ L (0, T ; H02 (Ωt )); b Q b Q 0 u(x, 0) = u (x) for all x ∈ Ω0 . In analogous way we have the following result b a mixed domain defined as Theorem 4.2. For each u0 ∈ H01 (Ω0 ) ∩ H 2 (Ω0 ), and Q b above, there exists a unique function u : Q → R, satisfying u ∈ L∞ (0, T ; H01 (Ωt )), ut ∈ L∞ (0, T ; H01 (Ωt )) and Z Z Z ut φ dx dt + (u + u2 )x φ dx dt + uxt φx dx dt = 0, b Q b Q b Q for all φ ∈ L2 (0, T ; H01 (Ωt )); u(x, 0) = u0 (x), for all x ∈ Ω0 . Acknowledgement. We wish to acknowledge the anonymous referee at the Electronic Journal of Differential Equations, for his constructive remarks and corrections on the original manuscript. References [1] Adams, R. A.; Sobolev Spaces, Academic, New York, 1975. [2] Avrin, J. and Goldstein, J. A.; Global existence for the Bejamin-Bona-Mahony equation in arbitrary dimensions, Nonlinear Analysis TMA, 9 (1985), 861-865. [3] Bona, J. L. and Dougalis, V.A.; An initial boundary-value problem for a model equation for propagation of long waves, J. Math. Analysis Appl. 75 (1980), 503-522. [4] Brill, H.; A semilinear Sobolev equation in a Banach space, J. Diff. Eqns., (1977) 412-425. [5] Caldas, C. S. Q., Limaco, J. and Barreto, K.; Linear thermoelastic system in noncylindrical domains, Funckcialaj Ekvacioj, 1 (1999), 115-127. [6] Friedman, A.; Partial Differential Equations, Holt, Rinehart and Winston, New York, 1969. [7] Goldstein, J. A. Semigrups of Linear Operators and Applications, Oxford University, New York, 1985. [8] Goldstein, J. A.; Mixed problems for the generalized Benjamin-Bona-Mahony equation, Nonlinear Analysis TMA, 4 (1980), 665-675. [9] Lions, J. L.; Une remarque sur les problèmes d’ èvolution linéaires dans des domaines non cylindriques, Revue Roumaine de Math. Pures et Appliquèes, V. 9, (1964) 11-18. [10] Lions, J. L. and Magenes E.; Problèmes aux Limites non Homogènes et Applications V. 1, Dunod, Paris 1968. [11] Medeiros, L. A., Limaco J. and Bezerra S.; Vibrations of Elastic Strings, J. of Computational Analysis and Application 4(2002), No. 2, 91-127, No. 3, 211-263. [12] Park, M. A.; On the Rosenau Equation, Mat. Aplic. Comp. V. 9, (1990), 145-152. [13] Rosenau, P. H.; Dynamics of dense discrete systems, Prog. Theoretical Phys., 79 (1988), 1028-1042. Rioco K. Barreto Instituto de Matemática - UFF, Rua Mário Santos Braga s/no , CEP: 24020-140, Niterói, RJ, Brasil E-mail address: rikaba@vm.uff.br Cruz S. Q. de Caldas, Instituto de Matemática - UFF, Rua Mário Santos Braga s/no , CEP: 24020-140, Niterói, RJ, Brasil E-mail address: gmacruz@vm.uff.br 12 R. K. BARRETO, C. DE CALDAS, P. GAMBOA, & J. LIMACO EJDE-2004/35 Pedro Gamboa Instituto de Matemática - UFRJ, Caixa Postal 68530, CEP 21945-970, Rio de Janeiro, RJ, Brasil E-mail address: pgamboa@dmm.im.ufrj.br Juan Limaco Instituto de Matemática - UFF, Rua Mário Santos Braga s/no , CEP: 24020-140, Niterói, RJ, Brasil E-mail address: juanbrj@hotmail.com