Journal of Inequalities in Pure and Applied Mathematics http://jipam.vu.edu.au/ Volume 4, Issue 5, Article 85, 2003 BEURLING VECTORS OF QUASIELLIPTIC SYSTEMS OF DIFFERENTIAL OPERATORS RACHID CHAILI D ÉPATEMENT DE M ATHÉMATIQUES , U.S.T.O. O RAN , A LGÉRIE . chaili@univ-usto.dz Received 12 May, 2003; accepted 27 June, 2003 Communicated by D. Hinton A BSTRACT. We show the iterate property in Beurling classes for quasielliptic systems of differential operators. Key words and phrases: Beurling vectors, Quasielliptic systems, Differential operators. 2000 Mathematics Subject Classification. 35H30, 35H10. 1. I NTRODUCTION The aim of this work is to show the iterate property in Beurling classes for quasielliptic systems of differential operators. This property is proved for elliptic systems in [2]. A synthesis of results on the iterate problem is given in [1]. Q Let (m1 , . . . , mn ) ∈ Zn+ , mj ≥ 1, 1 ≤ j ≤ n , we set µ = nj=1 mj , m = max{mj }, qj = mmj and q = (q1 , . . . , qn ). If α ∈ Zn+ and β ∈ Zn+ , we denote |α| = α1 + α2 + · · · + αn , Q Pn α α1 1 ∂ α αn D = D1 ◦ · · · ◦ Dn , where Dj = i · ∂xj , hα, qi = j=1 αj qj and αβ = nj=1 βjj . Let (Mp )+∞ p=0 be a sequence of real positive numbers such that (1.1) M0 = 1, ∃a > 0, 1 ≤ p j Mp Mp+1 ≤ ≤ ap , p ∈ Z∗+ , Mp−1 Mp Mp−j Mj ≤ Mp ≤ bp Mp−j Mj , p, j ∈ Z+ , j ≤ p, (1.2) ∃b > 0, ∃c > 0, c (1.3) ∀m ≥ 2, ∃d > 0, ∀p, h ∈ Z+ , h ≤ m; (Mpm )m−h (Mpm+m )h ≤ d (Mpm+h )m , (1.4) Mpm ∀m ≥ 2, ∃H > 0, ∀p, h ∈ Z+ , h ≤ p; ≤ H p−h Mhm ISSN (electronic): 1443-5756 c 2003 Victoria University. All rights reserved. 061-03 Mp Mh m . 2 R ACHID C HAILI ∞ Let (Pj (x, D))N coefj=1 be q−quasihomogeneous differential operators of order m with C n ficients in an open subset Ω of R , i.e. X Pj (x, D) = ajα (x) Dα . hα,qi≤m We define the quasiprincipal symbol of the operator Pj (x, D) by X Pjm (x, ξ) = ajα (x) ξ α . hα,qi=m Definition 1.1. The system (Pj )N j=1 is said q−quasielliptic in Ω if for each x0 ∈ Ω we have (1.5) N X |Pjm (x0 , ξ)| = 6 0, ∀ξ ∈ Rn \{0}. j=1 Definition 1.2. Let M = (Mp ) be a sequence satisfying (1.1) – (1.4), the space of Beurling vec- N ∞ tors of the system (Pj (x, D))N j=1 in Ω, denoted BM Ω, (Pj )j=1 , is the space of u ∈ C (Ω) such that ∀K compact of Ω, ∀L > 0, ∃C > 0, ∀k ∈ Z+ , (1.6) kPi1 . . . , Pik ukL2 (K) ≤ CLkm Mkm , where 1 ≤ il ≤ N, l ≤ k. Definition 1.3. Let l = (l1 , . . . , ln ) ∈ Rn+ and M be a sequence satisfying (1.1) – (1.4), we l call anisotropic Beurling space in Ω, denoted BM (Ω), the space of u ∈ C ∞ (Ω) such that ∀K compact of Ω, ∀L > 0, ∃C > 0, ∀α ∈ Z+n , n Y l α <α,l> (1.7) kD uk 2 ≤ CL Mαj j . L (K) j=1 Remark 1.1. If lj = 1, j = 1, . . . , n, we obtain, thanks to (1.2) the definition of isotropic Beurling space BM (Ω), (see [4]). The principal result of this work is the following theorem: Theorem 1.2. Let M and M 0 be two sequences satisfying (1.1) – (1.4) and p 0 X Mhm Mpm+m (1.8) lim = 0. 0 p→+∞ M hm Mpm+m h=0 q Let (Pj )N j=1 be q−quasielliptic system with BM (Ω) coefficients, then N q BM 0 Ω, (Pj )j=1 ⊂ BM 0 (Ω) . 2. P RELIMINARY L EMMAS Let ω be an open neighbourhood of the origin, we set K = k = hα, qi , α ∈ Zn+ and we define X |u|k,ω = kDα ukL2 (ω) , u ∈ C ∞ (ω) , k ∈ K. hα,qi=k If ρ > 0 we set Bρ = x ∈ Rn , J. Inequal. Pure and Appl. Math., 4(5) Art. 85, 2003 n X j=1 (xj ) 2 qj ! 12 <ρ . http://jipam.vu.edu.au/ B EURLING V ECTORS OF Q UASIELLIPTIC S YSTEMS OF D IFFERENTIAL O PERATORS 3 The two following lemmas are in [6]. Lemma 2.1. Let u ∈ C ∞ (Ω), r ∈ K and p ∈ Z+ , then X (2.1) |u|pm+r,ω ≤ |Dα u|r,ω . hα,qi=pm Lemma 2.2. Let k = pm + r < pm + jm, where k, r ∈ K and p, j ∈ Z∗+ , then ∃c(j) > 0, ∀Bρ ⊂ ω, ∀ε ∈]0, 1[, ∀u ∈ C ∞ (ω) , r |u|k,Bρ ≤ ε |u|(p+j)m,Bρ + c(j)ε− jm−r |u|pm,Bρ . (2.2) If a ∈ C ∞ (ω) , we denote [a, Dα ] u = Dα (au) − aDα u and if P is a differential operator, we define [P, Dα ] u = Dα (P u) − P (Dα u) . q Lemma 2.3. Let B be a bounded subset of Rn and a ∈ BM B , then ∀L > 0, ∃C > 0, ∀u ∈ C ∞ B , ∀p ∈ Z∗+ , 1 X X Mpmµ µ α pm−k (2.3) |[a, D ] u|0,B ≤ C L |u|k,B . Mkµ k≤pm−1 hα,qi=pm Proof. Let L > 0, as a ∈ k∈K q BM B , there exists C1 > 0 such that α |D a| ≤ C1 L hα,qi n Y Mαj qj , ∀α ∈ Zn+ , j=1 therefore, with the Leibniz formula, we get n X α Y qj α Dβ u C1 Lhα−β,qi M (2.4) |[a, D ] u|0,B ≤ . α −β j j 0,B β j=1 β<α We need the following easy inequality qj µ ! µ1 µ1 n Y hα, qi µ α αj (2.5) ≤ ≤ . β βj hβ, qi µ j=1 It is easy to check that from condition (1.2) we have c (2.6) l−1 l Y M hj ≤ MPl j=1 j=1 hence n Y hj ≤b (l−1) Pl j=1 hj l Y Mhj , j=1 Mαj −βj j=1 qj µ Pnj=1 qj µ−1 1 M. ≤ c This inequality with (1.2) and (2.5) imply Y Plj=1 qj 1 n qj Mhα,qiµ µ 1 α Mαj −βj ≤ |u|k,B . (2.7) β c Mhβ,qiµ j=1 pm−hβ,qi As the number of α ∈ Z∗+ satisfying hα, qi = pm and α > β, is limited by C2 , where C2 depends only of n, then (2.4) and (2.7) give Plj=1 qj 1 X X 1 Mpmµ µ pm−k α |[a, D ] u|0,B ≤ C1 (C2 L) |u|k,B , c M kµ k≤pm−1 hα,qi=pm k∈K J. Inequal. Pure and Appl. Math., 4(5) Art. 85, 2003 http://jipam.vu.edu.au/ 4 R ACHID C HAILI from which the desired estimate is obtained. 3. L OCAL E STIMATES q Let (Pj )N j=1 be a q−quasielliptic system with coefficients in BM B , where B is a neighbourhood of the origin. The following lemma is a light modification of an analogous lemma in [6, Lemma 2.3]. Lemma 3.1. Let ω be a small neighbourhood of the origin, ρ > 0 and δ ∈]0, 1[, such that B ρ+δ ⊂ ω. Then there exists C > 0, not depending on ρ and δ, such that for any u ∈ C ∞ (ω) , N X X |Pj u|0,Bρ+δ + δ −m+k |u|k,Bρ+δ . (3.1) |u|m,Bρ ≤ C j=1 k≤m−1 k∈K Lemma 3.2. Let ω, ρ and δ be as in Lemma 3.1, then ∃C > 0, ∀L > 0, ∃A > 0, ∀p ∈ Z∗+ , ∀u ∈ C ∞ (ω) (3.2) |u|(p+1)m,Bρ ≤ C N X |Pj u|pm,Bρ+δ + δ −m |u|pm,Bρ+δ + j=1 1 |u| (4e)m (p+1)m,Bρ+δ p X Mpm+m |u|hm,Bρ+δ +A L(p+1−h)m M hm h=0 ! , and (3.3) |u|m,Bρ ≤ C N X 1 |u| + (4e)m m,Bρ+δ |Pj u|0,Bρ+δ + δ −m |u|0,Bρ+δ j=1 ! . Proof. From (2.1) and (3.1) we obtain N N X X X (3.4) |u|(p+1)m,Bρ ≤ C |Pj u|pm,Bρ+δ + |[Pj , Dα ] u|0,Bρ+δ j=1 j=1 hα,qi=pm + X δ −m+k |u|pm+k,Bρ+δ , k≤m−1 k∈K Following the same idea as in the proof of Lemma 2.2 of [2], we get 1 X X M(pm+m)µ µ α 0 pm+m−s (3.5) |[Pj , D ] u|0,Bρ+δ ≤ C L |u|s,Bρ+δ . Msµ s≤pm+m−1 hα,qi=pm s∈K On the other hand, there exists h ∈ Z+ and r ∈ K such that s = hm + r, r < nm − n, (see [6, (1.3)]). As s ≤ pm + m − 1, then h ≤ p. From (2.2) we have (3.6) r |u|s,Bρ+δ ≤ ε |u|(h+n)m,Bρ+δ + C2 ε− nm−r |u|hm,Bρ+δ if s = hm + r, where 0 ≤ h ≤ p − n + 1 and 0 ≤ r < nm − n, and (3.7) r |u|s,Bρ+δ ≤ ε |u|pm+m,Bρ+δ + C2 ε− jm−r |u|hm,Bρ+δ if s = hm + r where h = p + 1 − j, 1 ≤ j ≤ n − 1 and 0 ≤ r ≤ jm − 1. J. Inequal. Pure and Appl. Math., 4(5) Art. 85, 2003 http://jipam.vu.edu.au/ B EURLING V ECTORS OF Q UASIELLIPTIC S YSTEMS OF D IFFERENTIAL O PERATORS 5 Let ε0 ∈]0, 1[ and put ε = ε0 µ1 Msµ M(h+n)mµ and Msµ L−nm+r in (3.6) µ1 L−jm+r in (3.7) . M(p+1)mµ According to (1.3) we obtain for any s satisfying (3.6), ε=ε L−s |u|s,Bρ+δ ≤ ε0 1 (Msµ ) µ 0 −hm L−(h+n)m 0 0−m L |u| + C d ε 2 1 |u|hm,Bρ+δ 1 (h+n)m,Bρ+δ (Mhmµ ) µ M(h+n)mµ µ and for any s satisfying (3.7) , L−s 1 (Msµ ) µ |u|s,Bρ+δ ≤ ε0 −hm L−(p+1)m 00 0−nm L |u| + C d ε 2 1 |u|hm,Bρ+δ . 1 (p+1)m,Bρ+δ (Mhmµ ) µ M(p+1)mµ µ These inequalities and (3.5) give X |[Pj , Dα ] u|0,Bρ+δ ≤ C 0 nε0 |u|(p+1)m,Bρ+δ hα,qi=pm +c (ε0 ) p X L(p+1−h)m h=0 0 0 m −1 Choosing ε = (2CC N n (4e) ) N X X |[Pj , Dα ] u|0,Bρ+δ ≤ J=1 hα,qi=pm M(pm+m)µ Mhmµ µ1 ! |u|hm,Bρ+δ . , then we obtain, with (1.4), 1 1 |u| 2C (4e)m (p+1)m,Bρ+δ +A p X (HL)(p+1−h)m h=0 Mpm+m |u|hm,Bρ+δ . Mhm It follows from this inequality: ∀L > 0, ∃A > 0, (3.8) N X X |[Pj , Dα ] u|0,Bρ+δ ≤ J=1 hα,qi=pm 1 1 |u| 2C (4e)m (p+1)m,Bρ+δ +A p X h=0 L(p+1−h)m Mpm+m |u|hm,Bρ+δ . Mhm It remains the estimate of the third term of the right-hand side of (3.4) . From (2.2) , we have k |u|pm+k,Bρ+δ ≤ ε |u|pm+m,Bρ+δ + C2 ε− m−k |u|pm,Bρ+δ . −1 Setting ε = ε0 δ m−k and choosing ε0 = (2C1 C (4e)m ) , then we obtain X 1 1 0 −m (3.9) δ −m+k |u|pm+k,Bρ+δ ≤ |u|pm,Bρ+δ . m |u|(p+1)m,Bρ+δ + C2 δ 2C (4e) k≤m−1 k∈K The estimates (3.4) , (3.8) and (3.9) imply (3.2) . The estimate (3.3) is obtained from (3.1) and (3.9) with p = 0. J. Inequal. Pure and Appl. Math., 4(5) Art. 85, 2003 http://jipam.vu.edu.au/ 6 R ACHID C HAILI 4. T HE M AIN R ESULT Let R > 0, to every sequence M satisfying (1.1) – (1.4) we define 1 p sup (R − ρ)pm |u|pm,Bρ . σM (u) = Mpm R/2≤ρ<R The following lemma is in [2]. 0 Lemma 4.1. Let ω be as in Lemma3.1, R ∈]0, 1[ such that B R ⊂ ω, M, M two sequences satisfying (1.1) – (1.4) and u ∈ BM 0 ω, (Pj )N j=1 , then for any L > 0, there exists an increasing positive sequence (Cp )+∞ p=0 such that ∀p, l ∈ Z+ , p σM 0 (Pi0 · · · Pil u) ≤ Cp (4.1) 0 Mpm+lm Lpm+lm . 0 Mpm where the sequence (Cp ) is constructed by recurrence, Cp+1 = Cp p 0 X Mhm Mpm+m NC + A 0 Mhm Mpm+m h=0 ! , where C and A are the constants of Lemma 3.2 and C0 is the constant satisfying 0 kPi0 · · · Pil ukL2 (BR ) ≤ C0 Llm Mlm . Theorem 4.2. Let M and M 0 be two sequences satisfying (1.1) – (1.4) and p 0 X Mhm Mpm+m (4.2) lim = 0. 0 p→+∞ Mhm Mpm+m h=0 q Let (Pj )N j=1 be q-quasielliptic system with coefficients in BM (Ω) , then q BM 0 Ω, (Pj )N j=1 ⊂ BM 0 (Ω) . Proof. We must verify (1.7) near every point x of Ω. By a translation of x at the origin, there exists a neighbourhood ω of the origin for which the precedent lemmas are true. Let L > 0 and let (Cp )+∞ p=0 be as in Lemma 4.1, then from (4.2) there exists p0 ∈ Z+ such that Cp+1 ≤ 2N CCp , p ≥ p0 , hence Cp ≤ Cp0 (2N C)p−p0 ≤ Cp0 (2N C)pm+lm , ∀l ∈ Z+ . For p ≤ p0 , this inequality is true because the sequence (Cp )+∞ p=0 is increasing. Let R ∈]0.1[ such that B R ⊂ ω, from (4.1) we obtain 0 Mpm+lm p σM 0 (Pi0 · · · Pil u) ≤ Cp0 (2N CL)pm+lm , ∀p, l ∈ Z+ . 0 Mpm In particular for l = 0, pm R 1 pm p |u|pm,BR/2 ≤ σM , 0 (u) ≤ Cp0 (2N CL) 0 2 Mpm hence |u|pm,BR/2 ≤ Cp0 4N C L R pm 0 Mpm , which can be rewritten as (4.3) 0 ∀L > 0, ∃C > 0, |u|pm,BR/2 ≤ CLpm Mpm . J. Inequal. Pure and Appl. Math., 4(5) Art. 85, 2003 http://jipam.vu.edu.au/ B EURLING V ECTORS OF Q UASIELLIPTIC S YSTEMS OF D IFFERENTIAL O PERATORS 7 The last inequality will allow us to conclude. In fact let k ∈ K, then there exists p ∈ Z+ and r ∈ K, r < nm − n, such that k = pm + r. From (2.2) , (4.3) and (2.6) , we obtain r 0 0 |u|k,BR/2 ≤ εC 0 L(p+n)m M(p+n)m + C 0 C 00 ε− nm−r Lpm Mpm µ1 µ1 r 1 1 0 0 . + C 0 C 00 ε− nm−r Lpm Mpmµ ≤ εC 0 L(p+n)m M(p+n)mµ c c Setting ε= 0 M(pm+r)µ ! µ1 0 M(p+n)mµ L−nm+r , then from (1.3) we get 0 |u|k,BR/2 ≤ C1 Lk Mkµ (4.4) µ1 . By an imbedding theorem of anisotropic Sobolev spaces (see [5]), from (4.4) and (1.2) we obtain µ1 0 sup |Dα u (x)| ≤ C2 (bL)hα,qi Mhα,qiµ . BR/2 The last estimate, with (2.6) gives hα,qi sup |Dα u (x)| ≤ C3 (bL) bhα,qinµ BR/2 n Y ! µ1 Mα0 j qj µ j=1 ≤ C3 (bL)hα,qi bhα,qinµ ≤ C3 b(1+n+mµ) L hα,qi n Y bqj µ(αj qj µ) Mα0 j j=1 n Y Mα0 j qj qj µ ! µ1 , j=1 q from there u ∈ BM 0 BR/2 . As a corollary we obtain from Theorem 1.2, the principal result of [2]. Theorem 1.2 also gives a result of regularity of solutions of differential equations in Beurling classes. Corollary 4.3. Under the assumptions of Theorem 1.2, the following assertions are equivalent: q i) u ∈ D0 (Ω) and Pj u ∈ BM 0 (Ω) , q ii) u ∈ BM 0 (Ω) . q For anisotropic projective Gevrey classes G{s},q (Ω) = BM (Ω), Mp = (p!)s , s ≥ 1, we have the same result. Corollary 4.4. Let s, s0 be such that s0 > s ≥ 1 and (Pj )N j=1 q−quasielliptic system with coefficients in G{s},q (Ω) , then N {s} G Ω, (Pj )j=1 ⊂ G{s},q (Ω) . Corollary 4.5. Let M and M 0 be two sequences satisfying (1.1) – (1.4) and (4.5) p 0 X Mhm Mpm+m lim = 0, 0 p→+∞ Mhm Mpm+m h=0 J. Inequal. Pure and Appl. Math., 4(5) Art. 85, 2003 http://jipam.vu.edu.au/ 8 R ACHID C HAILI and let (Pj )N j=1 be an elliptic system with coefficients in BM (Ω) , then N BM 0 Ω, (Pj )j=1 ⊂ BM 0 (Ω) . R EFERENCES [1] P. BOLLEY, J. CAMUS AND L. RODINO, Hypoellipticité analytique-Gevrey et itérés d’opérateurs, Ren. Sem. Mat. Univers. Politec. Torino, 45(3) (1989), 1–61. [2] C. BOUZAR AND R. CHAILI, Régularité des vecteurs de Beurling de systèmes elliptiques, Maghreb Math. Rev., 9(1-2) (2000), 43–53. [3] C. BOUZAR AND R. CHAILI, Vecteurs Gevrey d’opérateurs différentiels qusihomogènes, Bull. Belg. Math. Soc., 9 (2002), 299–310. [4] J.L. LIONS AND E. MAGENES, Problèmes aux Limites Non Homogènes et Applications, t. 3, Dunod Paris, 1970. (r ,...,rn ) [5] P.I. LIZORKIN, Nonisotropic Bessel potentials. Imbedding theorems for Sobolev spaces Lp 1 with fractional derivatives, Soviet Math. Dokl., 7(5) (1966), 1222–1226. [6] L. ZANGHIRATI, Iterati di operatori quasi-ellittici e classi di Gevrey, Boll. U.M.I., 5(18-B) (1981), 411–428. J. Inequal. Pure and Appl. Math., 4(5) Art. 85, 2003 http://jipam.vu.edu.au/