Rend. Sem. Mat. Univ. Pol. Torino Vol. 59, 4 (2001) N.M. Tri ON LOCAL PROPERTIES OF SOME CLASSES OF INFINITELY DEGENERATE ELLIPTIC DIFFERENTIAL OPERATORS Abstract. We give necessary and sufficient conditions for local solvability and hypoellipticity of some classes of infinitely degenerate elliptic differential operators. 1. Introduction We deal with local properties of differential operators 1 ∂ G ca,b = X 2 X 1 + i c(x)|x|−4 + (a(x) − b(x))|x|−3 e− |x| , ∂y √ where (x, y) ∈ R2 , i = −1, the functions a(x), b(x), c(x) satisfy a+ ∈ C if x > 0 a(x) = =: a, a− ∈ C if x < 0 b+ ∈ C if x > 0 b(x) = =: b, b− ∈ C if x < 0 c+ ∈ C if x > 0 c(x) = =: c c− ∈ C if x < 0 (1) and X1 = X2 = 1 ∂ ∂ − i b(x)sign(x)|x|−2e− |x| , ∂x ∂y 1 ∂ ∂ − i a(x)sign(x)|x|−2e− |x| . ∂x ∂y We will assume that Re a+ · Re a− · Re b+ · Re b− 6= 0. The form (1) is motivated by [1], [2] (see also [3])where the authors studied the hypoellipticity and the local solvability of finitely degenerate differential operators. Hypoellipticity and local solvability of G ca,b were investigated in [4] when a+ = a− = −1, b+ = b− = 1, c+ = c− , or a+ = a− = −1, b+ = b− = 1, c+ + c− = 2, and in [5] where a+ = a− = −eiϕ , b+ = b− = eiϕ , c+ = (1 − λ+ )eiϕ , c− = (1 − λ− )eiϕ , where ϕ ∈ [0, π/2), λ+ , λ− ∈ C. For those cases in [1], [2], [4] and [5], we have given another proof 277 278 N.M. Tri of non-hypoellipticity, recently in [6], [7], [8], by constructing explicit non-smooth solutions of homogeneous equations. In [9], by the same method we give the values of a+ , a− , b+ , b− , c+ , c− where G ca,b is not hypoelliptic. Here we shall use the method of constructing parametrixes. Since G ca,b is elliptic away from the line x = 0, we will consider only the case when |x| ≤ 1. Throughout the paper we denote by C a general positive constant which may vary from place to place. The paper is organized as follows. In section 2 we consider the case Re a+ < 0, Re a− < 0, Re b+ > 0, Re b− > 0, which directly generalizes the results of Hoshiro-Yagdjian. In section 3 we investigate the case Re a+ > 0, Re a− > 0, Re b+ > 0, Re b− > 0. In section 4 we deal with the case Re a+ > 0, Re a− < 0, Re b+ > 0, Re b− < 0. Finally, in section 5 we consider the case Re a+ < 0, Re a− > 0, Re b+ > 0, Re b− > 0. The results in sections 3-5 have completely new characteristics comparing with those considered in [4] and [5]. 2. The case Re a+ < 0, Re a− < 0, Re b+ > 0, Re b− > 0 T HEOREM 1. Assume that Re a+ < 0, Re a− < 0, Re b+ > 0, Re b− > 0. Then G ca,b is not hypoelliptic (nor local solvable) if and only if or c+ = k, a+ − b+ c+ = −k − 1, a+ − b+ where k and l are non-negative integers. c− =l a− − b− c− = −l − 1 a− − b− Proof. I) In this part we prove the hypoellipticity and the local solvability by constructing right and left parametrices. Let us consider the Fouruer transform of u with respect to y Z û(x, η) = +∞ e−iyη u(x, y)d y. −∞ By this transform G ca,b become 1 1 d d d Ĝ ca,b x, ,η = + ax −2 sign(x)e− |x| η + bx −2 sign(x)e− |x| η dx dx dx 1 = (c|x|−4 + (a − b)|x|−3)e− |x| η = 1 d2 d + (a + b)x −2sign(x)e− |x| η 2 dx dx + abx −4e− |x| η2 + (b − c)|x|−4e− |x| η − (a + b)|x|−3 e− |x| η. − 2 We would like to study solutions of the equation d c (2) Ĝ a,b x, , η û(x, η) = 0. dx 1 1 279 On local properties 1 − |x| Put û(x, η) = xe−be η f (z), where z = (b − a)e following confluent hypergeometric equation for f (z) 1 − |x| η. Then we will have the d 2 f (z) d f (z) c + (1 − z) − f (z) = 0. 2 dz dz b−a The equation has two linearly independent solutions c c , 1, z , f 2 (z) = e z 9 1 − , 1, −z f 1 (z) = 9 b−a b−a z where 9(α, γ , z) is the Tricomi function (see [10], p. 255). Therefore we have the following pair of solutions of the equation (2) − 1 1 c+ − |x| −b+ ηe |x| , if x ≥ 0 : û + (x, η) := xe 9 , 1, (b − a )ηe + + 1 b+ − a+ − 1 1 c+ − |x| + −a+ ηe |x| u 2 (x, η) := xe , 9 1− , 1, (a+ − b+ )ηe b+ − a+ (3) − 1 1 c− − |x| −b− ηe |x| , 1, (b − a )ηe 9 , if x ≤ 0 : û − (x, η) := xe − − 1 b− − a− − 1 1 c− − |x| −a− ηe |x| , 1, (a − b )ηe û − (x, η) := xe . 9 1 − − − 2 b− − a− ± As x → 0 we have the following asymptotics for û ± 1 (x, η), û 2 (x, η) û + 1 (x, η) ≈ 1 − x (log[(b+ − a+ )η] + 8(c+ /(b+ − a+ )) − 28(1)) , 0 (c+ /(b+ − a+ )) û + 2 (x, η) ≈ 1 − x (log[(a+ − b+ )η] + 8(1 − c+ /(b+ − a+ )) − 28(1)) , 0 (1 − c+ /(b+ − a+ )) û − 1 (x, η) ≈ −1 − x (log[(b− − a− )η] + 8(c− /(b− − a− )) − 28(1)) , 0 (c− /(b− − a− )) û − 2 (x, η) ≈ −1 − x (log[(a− − b− )η] + 8(1 − c− /(b− − a− )) − 28(1)) , 0 (1 − c− /(b− − a− )) (4) where 8(z) is the Gauss polygramma function (8(z) = 0(z)/ 0 0 (z)). Define D + (x, η) + − (respectively D − (x, η)) as the Wronskian of û + 1 (x, η), û 2 (x, η) (resp. û 1 (x, η), − û 2 (x, η)). Similarly (as in [4], Proposition 2) it is not difficult to see that − ctg πc+ − i sin πc+ if arg(b+ − a+ )η ∈ (0, π] b+ −a+ b+ −a+ D + (0, η) = πc πc + + − ctg if arg(b+ − a+ )η ∈ (−π, 0] b+ −a+ + i sin b+ −a+ − ctg πc− − i sin πc− if arg(b− − a− )η ∈ (0, π] b− −a− b− −a− D − (0, η) = πc πc − − − ctg if arg(b− − a− )η ∈ (−π, 0]. b− −a− + i sin b− −a− 280 N.M. Tri + Note that D + (0, η) · D − (0, η) 6= 0 when η 6= 0. Therefor û + 1 (x, η), û 2 (x, η) are − linearly independent solutions of (2) when x ≥ 0, η 6= 0, and û 1 (x, η), û − 2 (x, η) are linearly independent solutions of (2) when x ≤ 0, η 6= 0. Next for η > 0 we define the following pair of solutions of (2) ( û + 1 (x, η) if x ≥ 0 (5) u + (x, η) = u + (η)û − (x, η) + c u + (η)û − (x, η) if x ≤ 0 c1,− 1 2,− 2 + v (x, η) = (6) + ( + + v (η)û + (x, η) + c v (η)û + (x, η) if x ≥ 0 c1,+ 1 2,+ 2 û − 1 (x, η) if x ≤ 0 + + + u u v , cv + + where coefficients c1,− , c2,− , c1,+ 2,+ are chosen such that u (x, η), v (x, η) are continously differentiable at x = 0 c− − log (a − b )η − log (b − a )η − 8 1 − − − + + b− −a− u+ c1,− (η) = c+ c− − 0 b+ −a+ 0 1 − b− −a− D (0, η) + + 48(1) −8 b+c−a + , + c− + − (0, η) 0 b+c−a 0 1 − D b− −a− + + u (η) = c2,− + + v c1,+ (η) = + + v c2,+ (η) = + + log (b+ − a+ )η + log (b− − a− )η + 8 b+c−a + c− c+ − 0 b− −a− 0 b+ −a+ D (0, η) − − 48(1) 8 b−c−a − , c+ − − (0, η) 0 b−c−a 0 D b+ −a+ − log (a+ − b+ )η + log (b− − a− )η + 8 1 − − + 0 b−c−a 0 1 − b+c−a D + (0, η) − + − 8 b−c−a − 48(1) − , c+ − + (0, η) 0 1 − D 0 b−c−a b+ −a+ − c+ b+ −a+ + − log (b+ − a+ )η − log (b− − a− )η − 8 b+c−a + c− c+ 0 b− −a− 0 b+ −a+ D + (0, η) − −8 b−c−a + 48(1) − . − + 0 b−c−a 0 b+c−a D + (0, η) − + 281 On local properties For η < 0 we define the following pair of solutions of (2) ( û + − 2 (x, η) if x ≥ 0 − (7) u (x, η) = − u− u c1,− (η)û − 1 (x, η) + c2,− (η)û 2 (x, η) if x ≤ 0 − v (x, η) = (8) ( − − − v (η)û + (x, η) + c v (η)û + (x, η) if x ≥ 0 c1,+ 2,+ 2 1 (x, η) if x ≤ 0 û − 2 − − − u u v , cv − − where the coefficients c1,− , c2,− , c1,+ 2,+ are chosen such that u (x, η), v (x, η) are continously differentiable at x = 0 c− − log (a − b )η − log (a − b )η − 8 1 − − − + + b− −a− u− c1,− (η) = c− c+ − 0 1 − b+ −a+ 0 1 − b− −a− D (0, η) + −8 1 − b+c−a + 48(1) + , + c− + − (0, η) 0 1 − D 0 1 − b+c−a b− −a− + u− c2,− (η) = + v− c1,+ (η) = + − v c2,+ (η) = + log (a+ − b+ )η + log (b− − a− )η + 8 1 − c+ − − 0 1 − 0 b−c−a b+ −a+ D (0, η) − − 8 b−c−a − 48(1) − , c+ − − (0, η) 0 b−c−a 0 1 − D b −a − + + c+ b+ −a+ + log (a+ − b+ )η + log (a− − b− )η + 8 1 − b+c−a + c− c+ + 0 1 − b− −a− 0 1 − b+ −a+ D (0, η) − 8 1 − b−c−a − 48(1) − , c+ − + (0, η) 0 1 − b−c−a 0 1 − D b −a − + + + − log (b+ − a+ )η − log (a− − b− )η − 8 b+c−a + c+ c− 0 1 − b− −a− 0 b+ −a+ D + (0, η) − −8 1 − b−c−a + 48(1) − . − + 0 1 − b−c−a 0 b+c−a D + (0, η) − + 282 N.M. Tri Put W ± (0, η) = u ± (0, η)vx± (0, η) − v ± (0, η)u ± x (0, η). From (5), (6), (7), (8) we deduce that + v W + (0, η) = c2,+ (η)D + (0, η), − u W − (0, η) = c1,− (η)D − (0, η). + − We see that W ± (0, η) = 0 for |η| > C if and only if a+c−b = k, a−c−b = l or + − c+ c− = −k − 1, = −l − 1 where k and l are non-negative integers. Hence a+ −b+ a− −b− ± ± u (x, η), v (x, η) are two linearly independent solutions of (2) for |η| > C if and + − + − only if a+c−b 6= k, a−c−b 6= l or a+c−b 6= −k − 1, a−c−b 6= −l − 1 where k and l + − + − are non-negative integers. Now if we denote the Wronskian of u ± (x, η), v ± (x, η) by W ± (x, η) then by the Liouville theorem we have Z x ± ± p(s, η)ds , W (x, η) = W (0, η) exp − 0 where p(s, η) = Hence (a+ + b+ )x −2 sign(x)e−1/|x|η if x ≥ 0, (a− + b− )x −2 sign(x)e−1/|x|η if x ≤ 0. −1/|x| v + (η)D + (0, η)e −η(a+ +b+ )e−1/|x| W + (0, η)e−η(a++b+ )e = c2,+ if x ≥ 0, W + (x, η) = + (0, η)e −η(a− +b− )e−1/|x| = c v + (η)D + (0, η)e −η(a− +b− )e−1/|x| W 2,+ if x ≤ 0, −1/|x| u − (η)D − (0, η)e −η(a+ +b+ )e−1/|x| W − (0, η)e−η(a++b+ )e = c1,− if x ≥ 0, W − (x, η) = − (0, η)e −η(a− +b− )e−1/|x| = c u − (η)D − (0, η)e −η(a− +b− )e−1/|x| W 1,− if x ≤ 0. + − Since Re a+ < 0, Re a− < 0, Re b+ > 0, Re b− > 0, if a+c−b 6= k, a−c−b 6= l + − c− c+ or a+ −b+ 6= −k − 1, a− −b− 6= −l − 1 where k and l are non-negative integers, then u + (−1, η), v + (1, η) exponentially increase when η → +∞, and u − (−1, η), v − (1, η) exponentially increase when η → −∞. Hence we construct the Green function as follows + G (x, x 0 , η) if η ≥ C, G(x, x 0 , η) = G − (x, x 0 , η) if η ≤ −C, where + 0 for η ≥ C : G (x, x , η) = − 0 ( for η ≤ −C : G (x, x , η) = ( v + (x,η)u + (x 0 ,η) W + (x 0 ,η) v + (x 0 ,η)u + (x,η) W + (x 0 ,η) v − (x,η)u − (x 0 ,η) W − (x 0 ,η) v − (x 0 ,η)u − (x,η) W − (x 0 ,η) if x ≤ x 0 , if x 0 ≤ x, if x ≤ x 0 , if x 0 ≤ x. 283 On local properties + v (η)/W + (0, η)| < C and using the asymptotic behaviors of By noting that |c2,+ 9(α, γ , z) at zero and at infinity we can show that (in the similar way as in [4]) Z 1 Z 1 −1 −1 |G + (x, x 0 , η)|d x ≤ C ; − 0 |G (x, x , η)|d x ≤ C ; Z 1 Z 1 −1 −1 |G + (x, x 0 , η)|d x 0 ≤ C , |G − (x, x 0 , η)|d x 0 ≤ C , Z 1 ∂ G + (x, x 0 , η) ∂ G + (x, x 0 , η) | | |d x ≤ C ; |d x 0 ≤ C , ∂x ∂x −1 −1 Z 1 Z 1 ∂ G − (x, x 0 , η) ∂ G − (x, x 0 , η) | | |d x ≤ C ; |d x 0 ≤ C. ∂x ∂x −1 −1 Z 1 More general, for an arbitrary natural number n we will have Z Z |Dηn G + (x, x 0 , η)|d x ≤ Cη−n ; Z 1 |Dηn G − (x, x 0 , η)|d x ≤ C|η|−n ; Z 1 1 −1 1 −1 −1 −1 |Dηn G + (x, x 0 , η)|d x 0 ≤ Cη−n , |Dηn G − (x, x 0 , η)|d x 0 ≤ C|η|−n . Finally if we define the operator Q Qu(x, y) = Z 1 Z 1 Z ∞ 0 eiη(y−y ) φ(η)G(x, x 0 , η)u(x 0 , y 0 )d y 0 d x 0 dη, −1 −1 −∞ where φ(η) is a cut-off function φ(η) ∈ C ∞ (R), φ(η) = 0 if |η| ≤ C, φ(η) = 1 if |η| ≥ 2C, then Q will serve a right parametrix for G ca,b , and Q ∗ will serve a left c̄ parametrix for G c∗ a,b = −G b̄,ā . Hence the hypoellipticity and local solvability follow. II) In this part we will prove the theorem in the non-hypoellipticity and non-local solv+ − able cases. We argue only the cases when a+c−b = k, a−c−b = l, where k, l are + − non-negative numbers. The other case can be treated similarly. By the theorem of Hörmander if G ca,b is local solvable at the origin for some its neighborhood ω, there exist constants C, m such that Z X X (9) | f vd xd y| ≤ C sup |Dxα D βy f | |Dxα D βy G c∗ a,b v| α+β≤m α+β≤m for all f, v ∈ C0∞ (ω). Functions which violate this inequality will be constructed. For large λ let fλ = F(λ2 x, λ2 y)λ5 , where function F(x, y) ∈ C0∞ (R) and (10) Z ∞ Z ∞ −∞ −∞ F(x, y)d xd y = 1. 284 N.M. Tri Next it is easy to see that the function ( −1/|x| 0 xe−b̄+ηe L k (b̄+ − ā+ )ηe−1/|x| if x ≥ 0, U (x, η) = −1/|x| 0 xe−b̄−ηe L l (b̄− − ā− )ηe−1/|x| if x ≤ 0, where L 0k (z) = Put 1 z n −z n n! e Dz (e z ) are the Laguerre polynomials, solve the equation (2). Z ∞ 2 g(λρ)eiλ yρ U (x, λ2 ρ)dρ, vλ = χ(x)χ(y) 0 R where g ∈ C0∞ (0, ∞), χ ∈ C0∞ (−∞, ∞), g(ρ)dρ = 1, χ(x) = 1, when |x| ≤ , with a fixed small enough positive number . Now we will show that the inequality (9) does not hold for f = ∂x f λ and v = vλ with the parameter λ large positive enough. Indeed, for λ large enough, f λ ∈ C0∞ (ω) and Z Z Z Z ∂vλ ∂ fλ vλ (x, y)d xd y = f λ (x, y)d xd y = A + B, (11) − ∂x ∂x where A= Z Z Z f λ (x, y)χ(y)χx (x)g(λρ)U (x, λ2 ρ)eiλ B= Z Z Z fλ (x, y)χ(y)χ(x)g(λρ)Ux (x, λ2 ρ)eiλ 2 yρ 2 yρ dρd xd y, dρd xd y. It is not difficult to show that limλ→∞ B = 1 and for every positive number N, there exists a number C N such that |A| ≤ C N (1 + λ)−N . Next it is easy to check that the function Z ∞ wλ (x, y) = g(λρ)eiλ 2 yρ U (x, λ2 ρ)dρ 0 solves the equation G c̄b̄,ā wλ (x, y) = 0. Hence G c∗ a,b χ(x)χ(y)wλ (x, y) = Q(x, y, D x , D y )wλ (x, y) where Q(x, y, Dx , D y ) is a first order differential operator with coefficients vanishing in S = [−, ] × [−, ]. Next by similar argument in [2], [5] it is shown that |Dxα D βy wλ (x, y)| ≤ C N,m (1 + λ)−N for any (x, y) ∈ / S . Therefore we deduce that X −N (12) sup |Dxα D βy G c∗ . a,b vλ | ≤ Cn (1 + λ) α+β≤m Finally we see that (10), (11), (12) contradict (9). R EMARK 1. The following cases Re a+ > 0, Re a− < 0, Re b+ < 0, Re b− > 0; Re a+ > 0, Re a− > 0, Re b+ < 0, Re b− < 0; Re a+ < 0, Re a− > 0, Re b+ > 0, Re b− < 0 can be considered analogously. 285 On local properties 3. The case Re a+ > 0, Re a− > 0, Re b+ > 0, Re b− > 0 T HEOREM 2. Assume that Re a+ > 0, Re a− > 0, Re b+ > 0, Re b− > 0. Then G ca,b is not hypoelliptic nor local solvable at the origin. Proof. We separate the proof into some cases I) The non-resonance case a+ 6= b+ , a− 6= b− . We retain all notations used previously. For η > 0 we define the following solution of (2) V (x, η) = u + (x, η) is defined as − in (5). Note that when x ≤ 0 the solutions û − 1 (x, η), û 2 (x, η) exponentially decrease when η → +∞. + A) If b+c−a ∈ / Z− ∪ 0 then we set f = fλ , v = vλ as in section 2, with U (x, η) + + ∈ Z− ∪ 0 then we set f = ∂x fλ , v = vλ as in section replaced by V (x, η). B) If b+c−a + 2, with U (x, η) replaced by V (x, η). Then in a similar way as in section 2 we can contradict (9) by using f, v with large enough λ. II) The resonance case a+ = b+ , a− = b− . A) When c+ 6= 0, c− 6= 0 by taking the limit when a+ → b+ , a− → b− in (3) it is easy to see that the following pair is solutions of (2) (see [10], p. 266) 1 − |x| 1 1 − |x| )2 1 1 9 21 , 1, 4 c+ ηe− |x| 2 , − 1 1 − 1 1 1 −a+ ηe |x| +2(c+ ηe |x| ) 2 9 1 , 1, −4 c ηe − |x| 2 , ũ + (x, η) := xe + 2 2 1 1 1 − |x| − |x| 1 1 −a− ηe −2(c− ηe ) 2 9 1 , 1, 4 c ηe − |x| 2 , when x ≤ 0 : ũ − − 1 (x, η) := xe 2 − 1 − 1 1 1 1 −a− ηe |x| +2(c− ηe |x| ) 2 9 1 , 1, −4 c ηe − |x| 2 . ũ − (x, η) := xe − 2 2 when x ≥ 0 : −a+ ηe ũ + 1 (x, η) := xe −2(c+ ηe Next for η > 0 let us define the following solution of (2) ( + ũ 1 (x, η) if x ≥ 0, Ṽ (x, η) = − Ṽ Ṽ c1,− (η)ũ − 1 (x, η) + c2,− (η)ũ 2 (x, η) if x ≤ 0, Ṽ Ṽ where c1,− (η), c2,− (η) are chosen as in section 2. Now we can repeat the proof in section 2. + B) When c+ 6= 0, c− = 0 then we have solutions ũ + 1 (x, η), ũ 2 (x, η) when x ≥ 0, and e−a− ηe 1 |x| , xe−a− ηe 1 |x| when x ≤ 0. C) When c+ = 0, c− 6= 0 then we have solutions e−a+ ηe − and ũ − 1 (x, η), ũ 2 (x, η) when x ≤ 0. D) When c+ = 0, c− = 0 then we have solutions e−a+ ηe 1 ηe |x| 1 ηe |x| 1 |x| 1 |x| , xe−a+ ηe , xe−a+ ηe 1 |x| 1 |x| when x ≥ 0, when x ≥ 0, and e−a− , xe−a− when x ≤ 0. III) The half-resonance case a+ 6= b+ , a− = b− . This case can be treated by using the solutions in part I) when x ≥ 0, and the solutions in part II) when x ≤ 0. 286 N.M. Tri IV) The half-resonance case a+ = b+ , a− 6= b− . This case can be treated by using the solutions in part II) when x ≥ 0, and the solutions in part I) when x ≤ 0. R EMARK 2. The following case Re a+ < 0, Re a− < 0, Re b+ < 0, Re b− < 0 can be considered analogously. 4. The case Re a+ > 0, Re a− < 0, Re b+ > 0, Re b− < 0 T HEOREM 3. Assume that Re a+ > 0, Re a− < 0, Re b+ > 0, Re b− < 0. Then G ca,b is always hypoelliptic and local solvable at the origin. Proof. We consider only the non-resonance case a+ 6= b+ , a− 6= b− . The other cases (resonance and half-resonance) can be treated analogously. Next for η > 0 we define the following pair of solutions of (2) (13) (14) + ( + ( U (x, η) = V (x, η) = û + 1 (x, η) if x ≥ 0, + + − U U c1,− (η)û − 1 (x, η) + c2,− (η)û 2 (x, η) if x ≤ 0, û + 2 (x, η) if x ≥ 0, + + − V V c1,− (η)û − 1 (x, η) + c2,− (η)û 2 (x, η) if x ≤ 0, + + + + U U V V where the coefficients c1,− , c2,− , c1,− , c2,− are chosen as in section 2 such that + + U (x, η), V (x, η) are continuously differentiable at x = 0. For η < 0 we define the following pair of solutions of (2) (15) (16) − ( − ( U (x, η) = V (x, η) = − − − + U U c1,+ (η)û + 1 (x, η) + c2,+ (η)û 2 (x, η) if x ≥ 0, û − 1 (x, η) if x ≤ 0, − − + V V c1,+ (η)û + 1 (x, η) + c2,+ (η)û 2 (x, η) if x ≥ 0, û − 2 (x, η) if x ≤ 0, − − − U , cU , c V , c V are chosen such that U − (x, η), where the coefficients c1,+ 2,+ 1,+ 2,+ − V (x, η) are continuously differentiable at x = 0. Since Re a+ > 0, Re a− < 0, Re b+ > 0, Re b− < 0, then U + (−1, η), V + (−1, η) exponentially increase when η → +∞, and U − (1, η), V − (1, η) exponentially increase when η → −∞. Therefore we construct the Green function as follows ( + G (x, x 0 , η) if η ≥ C, 0 G(x, x , η) = G − (x, x 0 , η) if η ≤ −C, On local properties 287 where, for η ≥ C: + + 0 + 0 + V (x, η)U (x , η) − V (x , η)U (x, η) if x ≤ x 0 , + 0 + 0 D (x , η) G (x, x , η) = 0 if x 0 ≤ x, and for η ≤ −C: − − 0 − 0 − V (x, η)U (x , η) − V (x , η)U (x, η) if x ≤ x 0 , − 0 − 0 D (x , η) G (x, x , η) = 0 if x 0 ≤ x. Now the rest of the proof goes through as in section 2. We leave the details to the readers. R EMARK 3. The following case Re a+ < 0, Re a− > 0, Re b+ < 0, Re b− > 0 can be treated analogously. 5. The case Re a+ < 0, Re a− > 0, Re b+ > 0, Re b− > 0 T HEOREM 4. Assume that Re a+ < 0, Re a− > 0, Re b+ > 0, Re b− > 0. Then G ca,b is not hypoelliptic nor local solvable at the origin. Proof. We argue only for the non-resonance case a+ 6= b+ , a− 6= b− . The halfresonance case can be treated analogously. Put ( + ũ 1 (x, η) if x ≥ 0, Ũ (x, η) = − Ũ Ũ c1,− (η)ũ − 1 (x, η) + c2,− (η)ũ 2 (x, η) if x ≤ 0, Ũ Ũ where c1,− (η), c2,− (η) are chosen as in section 2. Since Re b+ > 0, Re a− > 0, Re b− > 0 the solution Ũ (x, η) exponentially decreases when η → +∞. Now the rest of the proof goes as in section 2. R EMARK 4. The seven following cases Re a+ < 0, Re a− < 0, Re b+ 0, Re b− < 0; Re a+ > 0, Re a− < 0, Re b+ > 0, Re b− > 0; Re a+ > 0, Re a− 0, Re b+ > 0, Re b− < 0; Re a+ < 0, Re a− < 0, Re b+ < 0, Re b− > 0; Re a+ 0, Re a− > 0, Re b+ < 0, Re b− > 0; Re a+ > 0, Re a− < 0, Re b+ < 0, Re b− 0; Re a+ < 0, Re a− > 0, Re b+ < 0, Re b− < 0; can be considered analogously. > > > < References [1] G ILIOLI A. AND T REVES F., An example of solvability theory of linear PDE’s, Am. J. Math. 96 (1974), 367–385. 288 N.M. Tri [2] M ENIKOFF A., Some examples of hypoelliptic partial differential equations, Math. Ann. 221 (1976), 176–181. [3] H ELFFER B. AND RODINO L., Opérateur différentiels ordinaires intervenant dans l’étude de l’hypoellipticité, Bollettino U.M.I. 14 B-5 (1977), 491–522. [4] H OSHIRO T., Some examples of hypoelliptic operators of infinitely degenerate type, Osaka J. Math. 30 (1993), 771–782. [5] R EISSIG M. AND YAGDJAN K., An interesting connection between hypoellipticity and branching phenomena for certain differential operators with degeneracy of infinite order, Rend. Mat. Appl. 15 7 (1995), 481–510. [6] T RI N.M., Remark on non-uniform fundamental solutions of a class of differential operators with double characteristics, J. Math. Sci. Univ. Tokyo 6 (1999), 437–452. [7] T RI N.M., Note on necessary conditions of hypoellipticity of some classes of differential operators with double characteristics, Kodai Mathematical Journal 23 (2000), 281–297. [8] T RI N.M., Non-smooth solutions for a class of infinitelydegenerate elliptic differential operators, Vietnam Journal of Mathematics 28 (2000), 159–172. [9] T RI N.M., Some examples of non-hypoelliptic infinitely degenerate elliptic differential operators, Matematicheskie Zametki 71 (2002), 517–529. [10] BATEMAN H. AND E RDELYI A., Higher trascendental functions, vol.I, McGrawHill, New York 1953. AMS Subject Classification: 35H05, 35D10, 35B65. Nguyen Minh TRI Institute of Mathematics P. O. Box 631, Boho 10000 Hanoi, VIETNAM e-mail: triminh@thevinh.ncst.ac.vn Lavoro pervenuto in redazione il 02.11.1999 e, in forma definitiva, il 30.05.2000.