The Micro-Support of the Complex Defined by a Convolution Operator in Tube Domains

SINGULARITIES AND DIFFERENTIAL EQUATIONS BANACH CENTER PUBLICATIONS, VOLUME 33 INSTITUTE OF MATHEMATICS POLISH ACADEMY OF SCIENCES WARSZAWA 1996

THE MICRO-SUPPORT OF THE COMPLEX DEFINED BY A CONVOLUTION OPERATOR IN TUBE DOMAINS

RYUICHI ISHIMURA and YASUNORI OKADA Department of Mathematics and Informatics, Faculty of Sciences, Chiba University 1-33 Yayoi-cho, Inage-ku, Chiba, 263 Japan E-mail: [email protected]

0. Introduction. In the present report, we consider the convolution operator u∗ defined by a hyperfunction u(x) with compact support, operating on holomorphic functions in tube domains invariant by any real translation. In all what follows we impose the natural condition called (S) introduced by Kawai [Kaw]. This condition is in fact equivalent to the classical property of completely regular growth in pure-imaginary directions ([IOk]). First we have under this condition (S), the existence of solutions in any open tube domain. And conversely, we also have that the existence of solutions in some special tube domain implies the condition (S), meaning that (S) is a sufficient and almost necessary condition for the existence. Secondly, we will define the characteristic set Car(u∗) of u∗ and prove that all solutions of the homogeneous equation u ∗ f = 0 are extended holomorphically to any non-characteristic direction. Finally, by these results, we will conclude under the condition (S) that the micro-support of the complex defined by the operator u∗ is included in the characteristic set Car(u∗). Most of the results are based on our paper [IO] and on [IOk] and we refer to them for the details of proofs. The authors take this occasion to express their thanks to Polish Academy of Sciences and Institute of Mathematics for inviting one of the authors and for all hospitality during his stay in Warszawa, especially to Professor B. Ziemian and also to secretaries of Banach Center.

√ √ 1. Problem and notations. Let τ be the projection Rn × −1Rn → −1Rn and set τ n O = Rτ∗OC ,

1991 Mathematics Subject Classiﬁcation: 35R50, 46F15. The paper is in ﬁnal form and no version of it will be published elsewhere.

[105] 106 R. ISHIMURA AND Y. OKADA

n where we denote by O n the sheaf of holomorphic functions on . It is, in fact, a sheaf √ C C on −1Rn. Let u(x) be a hyperfunction with compact support M on Rn and consider the complex S : 0 → Oτ −→u∗ Oτ → 0. Then to consider the existence and the analytic continuation at the same time, we formulate (cf. Kashiwara-Schapira [KS]): Problem 1.1. Deﬁne the “characteristic set” Car(u∗) to be a natural generalization of the classical case of partial diﬀerential operators and estimate the micro-support SS(S) by Car(u∗). For an analytic functional T ∈ O(Cn)0, let Tˆ(ζ) denote its Fourier-Borel transform: z·ζ Tˆ(ζ) = hT, e iz. In particular, if u is a hyperfunction with compact support, its Fourier-Borel transform √ uˆ(ζ) is an entire function with infra-exponential growth on −1Rn.

2. The condition (S) Definition 2.1 ([IOk]). We say that a subharmonic function s(ξ) on Rm satisﬁes the

condition (S) in the direction ξ0 or simply it satisfies the condition (S)ξ0 if we have (2.2) For every small ε > 0, there exists N > 0 such that for any r > N, we have m ξ ∈ R satisfying |ξ − ξ0| < ε and ρ(r) ˆ∗ s(rξ)/r ≥ hs(ξ0) − ε, ∗ where hs(ξ0) is the regularized growth indicator of s(ξ) (see for example Lelong-Gruman [LG]). We also say that a holomorphic function f(ζ) defined in a part of a cone, outside of a bounded set, satisfies the condition (S) in the direction ζ0 or satisfies the condition n 2n (S)ζ0 if, identifying C with R , the subharmonic function ln |f(ζ)| satisfies the above condition (S). R e m a r k 2.3. In [IOk], R. Ishimura and J. Okada proved that the condition (S) is nothing but the notion of regular growth, classical in the case of one variable, and considered by Lelong-Gruman [LG], for example, in the general case. R e m a r k 2.4. In the case of a hyperfunction with compact support u(x), because ∗ its Fourier transform is of minimal type in real directions, that is, we have huˆ(ξ) = 0 for every ξ ∈ Rn, definition 2.1 is a natural generalization of the definition of Kawai [Kaw]. n n In the sequel, we denote for R > 0 and ζ0 ∈ C by B(ζ0,R) the open ball in C of centre ζ0 and with radius R. Throughout the paper, we will suppose the condition (S) for the entire function f(ζ) = uˆ(ζ) in pure imaginary directions: √ √ (2.5) For every ε > 0, there exists N > 0 such that for any −1η ∈ −1Rn with |η| > N, we can find ζ ∈ Cn which satisfies √ | −1η − ζ| < ε|η|, |f(ζ)| ≥ e−ε|η|. First we have the following division lemma under the condition (S). COMPLEX DEFINED BY CONVOLUTION 107

Lemma 2.6. Let f, g and h ∈ O(Cn) be entire functions of exponential type and M and K be two compact sets in Cn. Suppose that fg = h and that for every ε > 0, there n exist Aε > 0 and Bε > 0 such that for any ζ ∈ C , we have

(2.7) ln |f(ζ)| ≤ Aε + HM (ζ) + ε|ζ|,

(2.8) ln |h(ζ)| ≤ Bε + HK (ζ) + ε|ζ|. Assume also that f satisﬁes the condition (S). Then for any a > 0 there exists a compact n set L ⊂ C satisfying τ(L) ⊂ τ(K + B(0; a)) and for every ε > 0, there exists Cε > 0 such that we have for any ζ ∈ Cn,

(2.9) ln |g(ζ)| ≤ Cε + HL(ζ) + ε|ζ|. S k e t c h o f p r o o f. For any ε > 0 with 0 < ε < 1 , let Γ be a conic ε-neighborhood √ √ √ 2 ε √ n n n of −1R i.e. Γε := {ζ ∈ C | there exists −1η ∈ −1R such that |ζ − −1η| < ε|η|}. First we prove the estimate (2.9) for ζ ∈ Γ : By the condition (S), there exists N > 0 so ε √ that if |ζ| > N, we can ﬁnd ζ0 ∈ Cn satisfying |ζ0 − −1η| < ε|η| and |f(ζ0)| ≥ e−ε|η|. We recall a lemma of Harnack-Malgrange-H¨ormander([H1], Lemma 3.1): Lemma 2.10. If F (ζ),H(ζ) and G(ζ) = H(ζ)/F (ζ) are three holomorphic functions in the open ball B(0; R), and if |F (ζ)| < A, |H(ζ)| < B on B(0; R), then

2|ζ| − R+|ζ| |G(ζ)| ≤ BA R−|ζ| |F (0)| R−|ζ| for all ζ ∈ B(0; R).

0 We apply this lemma for the ball B(ζ ; 3ε|η|). By (2.8), setting k := supz∈K |z|, for any δ > 0 with δ ≤ min(ε, 1), we have √ 00 sup ln |h(ζ )| ≤ Bε + 4(k + 1)ε|η| + HK ( −1η). ζ00∈B(ζ0;3ε|ε|) n In the same way, by setting m := supx∈M |x| and noting that M ⊂ R , using (2.7) we have 00 sup ln |f(ζ )| ≤ Aε + 4(m + 1)ε|η|. ζ00∈B(ζ0;3ε|ε|) So remarking that ζ ∈ B(ζ0; 2ε|η|)(⊂ B(ζ0; 3ε|η|)), by lemma 2.10, we have √ ln |g(ζ)| ≤ Bε + 4(k + 1)ε|η| + HK ( −1η) 2|ζ − ζ0| 3ε|η| + |ζ − ζ0| + (A + 4(k + 1))ε|η| + ε|η| 3ε|η| − |ζ − ζ0| ε 3ε|η| − |ζ − ζ0| √ ≤ Bε + 4Aε + HK ( −1η) + [4(k + 4m) + 25]ε|η|. 0 So for any ε > 0, we can find a constant Cε > 0 such that √ 0 ln |g(ζ)| ≤ Cε + HK ( −1η) + ε|η| n for ζ ∈ Γε. Next we consider the estimate (2.10) for all ζ ∈ C . We remark that in the case n = 1, by the Phragmén-Lindelöfprinciple and the estimate in Γε, we have the desired estimate, and in the general case when n ≥ 1, we can find directly a constant Cε > 0 so that the estimate (2.10) follows. 108 R. ISHIMURA AND Y. OKADA

3. Vanishing of the ﬁrst cohomology group. Under the condition (S), by the standard method as in Malgrange [M], we have the surjectivity: 3.1. Let u(x) be a hyperfunction with compact support. Assume that uˆ(ζ) Theorem √ √ satisﬁes the condition (S). Then for any open set −1ω ⊂ −1Rn, the operator √ √ u∗ : Oτ ( −1ω) → Oτ ( −1ω) is surjective.

Corollary 3.2. We have SS(H1(S)) = 0. Conversely, by means of Favorov [F] and Morzhakov [Mo2] (see also Lelong-Gruman [LG]), we also note that the condition (S) is a necessary condition for the surjectivity in the following sense: √ 3.3. Let −1ω is a strictly convex bounded open domain with C2-boundary √Theorem √ √ in −1Rn. Then if u∗ : Oτ ( −1ω) → Oτ ( −1ω) is surjective, then f(ζ) :=u ˆ(ζ) has to satisfy the condition (S). R e m a r k 3.4. In the case of n = 1, by theorem 3.1 and by the proof of Theorem 7 of Morzhakov [M2] (or of Lelong-Gruman [LG], Theorem 9.35), we can conclude that the condition (S) is nothing but the property of (completely) regular growth (see e.g. Levin √ [L]) in the pure imaginary directions −1Rn. But in the general case of n ≥ 1, we can also prove the equivalence of (S) and the property of regular growth in Lelong-Gruman sense [LG] (see [IOk]).

4. The characteristic set and the continuation of homogeneous solutions. n For the 0-th cohomology group of the complex S, under the condition (S) for u ∈ Bc(R ), we shall now solve the prolongation problem of Oτ -solutions of the homogeneous equation u ∗ f = 0 by the method of Kiselman [Ki] and Sébbar[Sé]. We denote the sheaf of √ √ Oτ -solutions by N , that is, for any open set −1ω ⊂ −1Rn, we set √ √ N ( −1ω) := {f ∈ Oτ ( −1ω) | u ∗ f = 0} √ √ √ √ For an open set −1Ω ⊂ −1 n with −1ω ⊂ −1Ω, the problem is to get a condition √R √ for the restriction map r : N ( −1Ω) → N ( −1ω) to be surjective. √ √ We note that the subspace N ( −1ω) of the (FS) space Oτ ( −1ω), endowed with the induced topology, becomes a closed subspace. We have √ √ √ √ 4.1. Let −1ω and −1Ω be two open subsets of −1 n with −1ω ⊂ √ Proposition R −1Ω. Assume that u satisfies (S). Then the image Im r of the restriction map r is dense √ in N ( −1ω). S k e t c h o f p r o o f . Let E be the space of exponential-polynomial solutions and √ E◦ its polar set in O(Cn)0. We shall show that E is dense in N ( −1ω). To do this, it is sufficient, by Hahn-Banach’s theorem, to prove that any functional T ∈ E◦ ∩ O( n × √ √ R −1ω)0 is orthogonal to N ( −1ω), and by the following lemma we have the desired result: COMPLEX DEFINED BY CONVOLUTION 109

Lemma 4.2 (B. Malgrange [M]). T ∈ O(Cn)0 belongs to E◦ if and only if there exists an entire function g(ζ) satisfying Tˆ(ζ) = g(ζ)û(−ζ). In order to present the continuation theorem, we will prepare the notion of characteristics which is a natural generalization of the case of usual differential operators of 2n−1 n finite order with constant coefficients. We define the sphere at infinity S∞ by C /R+ 2n n 2n−1 n and consider the compactification with directions D = C t S∞ of C . For any ζ ∈ n, ζ 6= 0, we write ζ∞ ∈ S2n−1 for the accumulating point of {tζ | t > 0} at S2n−1, C ∞ √ ∞ i.e. {ζ∞} = (the closure of {tζ | t > 0} in 2n) ∩ S2n−1. We denote by −1Sn−1 the √ D ∞ ∞ 2n−1 pure-imaginary sphere at infinity {(ξ + −1η)∞ ∈ S∞ | ξ = 0}, which is a closed 2n−1 subset of S∞ . For a hyperfunction u with compact support, using the modulus of the Fourier-Laplace transform f =u ˆ of u, we define the characteristics Car(u∗) as follows: for positive ε we set n ε|ζ| Vf (ε) := {ζ ∈ C | e |f(ζ)| < 1}, 2n−1 2n Wf (ε) := S∞ ∩ (the closure of Vf (ε) in D ), [ Wf := the closure of Wf (ε). ε>0 Now we define the characteristic set of u∗. Definition 4.3. With the above notation, we put √ n−1 Car(u∗) := Wf ∩ −1S∞ . We remark that in the case of differential operators of finite order with constant coefficients, the notion of the characteristic set coincides with the usual one defined as the accumulating directions of zeros of the principal symbol. We note that the direction √ √ −1ρ∞ ∈ −1Sn−1 does not belong to Car(u∗) if and only if there exist a conic neigh- ∞ √ 2n borhood Γ ⊂ D of −1ρ, positive N and an infra-linear function λ : R+ → R+ such that f(ζ) satisfies the estimate |f(ζ)| > e−λ(|ζ|) on Γ ∩ Cn ∩ {|ζ| > N}, where λ(t) is said to be infra-linear if for any positive ε, we can find a constant C satisfying (4.1) λ(t) ≤ εt + C for any t > 0. √ √ √ For this Car(u∗) and an open convex −1ω ⊂ −1Rn, we now put −1Ω := the interior of √ √ √ √ √ \ n √ { −1y ∈ −1R | h −1y, −1ηi ≤ H −1ω( −1η)}, √ −1η∞∈Car(u∗)a √ √ here a means the antipodal. Then −1Ω becomes an open convex subset in −1 n √ √ R containing −1ω. Moreover for any compact L in n × −1Ω, we can take compact K √ R in Rn × −1ω such that √ n (4.2) HL(ζ) ≤ HK (ζ) for any ζ ∈ C such that −1η∞ ∈ Car(u∗). 110 R. ISHIMURA AND Y. OKADA

Under the above situation we state the most important theorem of this article. 4.4. Let u be a hyperfunction with compact support, and let a pair of open Theorem√ √ sets −1ω ⊂ −1Ω be as above. Assume that uˆ satisﬁes the condition (S). Then the √ √ ω restriction rΩ : N ( −1Ω) → N ( −1ω) is surjective. √ √ P r o o f. Remarking that the restriction rω : N ( −1Ω) → N ( −1ω) has a dense √ √ Ω image, the transpose trω : N ( −1Ω)0 ← N ( −1ω)0 is injective and we need only show its Ω √ surjectivity. By Hahn-Banach’s theorem, any functional T ∈ N ( −1Ω)0 has an extension √ √ T ∈ O( n × −1Ω)0, and let us choose S ∈ O( n × −1ω)0 so that T and S coincide 1 √R R 1 on N ( −1Ω). To do this, we use the following division lemma.

Lemma 4.5. Let L and K be a pair of compact subsets of Cn satisfying the estimate (4.2) and g(ζ) be an entire function satisfying the estimate

ln |g(ζ)| ≤ HL(ζ) + C0 with a constant C0 >0. Then for any positive ε, there exists a constant Cε >0 and entire functions q(ζ) and r(ζ) which satisfy g(ζ) =u ˆ(−ζ)q(ζ) + r(ζ),

ln |q(ζ)| ≤ HL+supp u(ζ) + ε|ζ| + Cε,

ln |r(ζ)| ≤ HK+supp u(ζ) + ε|ζ| + Cε.

2n−1 P r o o f o f t h e L e m m a . For given positive ε, we divide S∞ into Λ1 ∪ Λ2 as 2n−1 1 2n−1 Λ1 = {ζ∞ ∈ S∞ | HL(ζ) < HK (ζ) + 4 ε|ζ|}, Λ2 = S∞ \ Λ1.

Remark that Λ1 is an open neighborhood of Car(u∗). Since Λ2 is compact, we can take an infra-linear function λ(t) and positive N such that −λ(|ζ|) ζ∞ ∈ Λ2, |ζ| > N implies f(ζ) > e . Moreover we can take this λ so that it is C∞, equal to 1 in a neighborhood of t = 0, and that |λ0(t)| ≤ 1. According to the above, we divide Cn into the following three parts. n γ1 = {ζ ∈ C | ζ∞ ∈ Λ1, |ζ| > N}, n γ2 = {ζ ∈ C | ζ∞ ∈ Λ2, |ζ| > N}, n γ3 = {ζ ∈ C | |ζ| ≤ N}. Now we construct q(ζ) and r(ζ) as follows: g(ζ) q(ζ) := (1 − ϕ(eλ(|ζ|)f(ζ))) + v(ζ), f(ζ) r(ζ) := g(ζ)ϕ(eλ(|ζ|)f(ζ)) − f(ζ)v(ζ),

by choosing a suitable C∞ function v(ζ) on Cn, where ϕ(τ) is a ﬁxed C∞ function on C with 1 (|τ| ≤ 1/2), 0 ≤ ϕ(τ) ≤ 1, ϕ(τ) = 0 (|τ| ≥ 1). COMPLEX DEFINED BY CONVOLUTION 111

The condition for q and r to be holomorphic is given as g(ζ) ∂ϕ ∂ϕ λ0(|ζ|) ∂ϕ (4.3) ∂v¯ = eλ(|ζ|) f(ζ) + f¯(ζ) ζ · dζ¯ + ∂¯f¯ . f(ζ) ∂τ ∂τ¯ 2|ζ| ∂τ¯ We denote the right hand side by w which satisfies ∂w¯ = 0. Let us estimate w. In the λ(|ζ|) λ(|ζ|) outside of {1/2 < e |f(ζ)| < 1}, especially in γ2, we have w = 0 since ϕ(e f(ζ)) is constant. In {1/2 < eλ(|ζ|)|f(ζ)| < 1}, we have ¯ ¯ λ 1 ∂ϕ ∂ϕ ∂ϕ ∂f |w| ≤ |g|e (|ζ|) max + max |dζ¯| + max 2 |τ|≤1 ∂τ |τ|≤1 ∂τ¯ |τ|≤1 ∂τ¯ f ∂ϕ ∂ϕ ≤ eHL(ζ)+λ(|ζ|) max , (|dζ¯| + 2eλ(|ζ|)|∂f|). |τ|≤1 ∂τ ∂τ¯ Since 0 0 (4.4) |∂f|/ exp(Hsupp u(ζ) + ε |ζ|) is bounded for any ε > 0, we have, with some constant C > 0, ε ln |w| ≤ H (ζ) + H (ζ) + 2λ(|ζ|) + |ζ| + C. L supp u 4 ε Remarking that HL(ζ) < HK (ζ) + 4 |ζ| holds in γ1 and that γ2 is compact, we get the estimate ε ln |w| ≤ H (ζ) + H (ζ) + 2λ(|ζ|) + |ζ| + C0 K supp u 2 in the whole Cn, with another constant C0 > 0. By the same argument as in Sébbar[Sé] (pp. 205–206), using the estimate (4.1) for ε/4, we conclude by Theorem 4.4.2 and the proof of Theorem 4.2.5 in Hörmander[H2] that there exists a solution v ∈ C∞(Cn) of (4.4) which satisfies the estimate 3ε ln |v| ≤ H (ζ) + H (ζ) + |ζ| + C00 K supp u 4 for some C00 > 0. Finally let us estimate the growth of q and r constructed from this solution v. For q, we have = 0 if |f(ζ)| ≤ 1/2, |q − v| = |(1 − ϕ)g/f| ≤ 2 exp(HK (ζ) + λ(|ζ|)) if |f(ζ)| > 1/2, which shows the desired result. For r, we have 1 |gϕ| ≤ exp(HL(ζ)) ≤ exp(HK (ζ) + 4 ε|ζ|) in γ1,

|gϕ| = 0 in γ2, which shows the desired result because of (4.3). E n d o f t h e p r o o f o f t h e T h e o r e m . We can choose a compact L ⊂ n × √ R −1Ω with Tˆ (ζ) ≤ Const ·eHL(ζ) and applying this lemma to g = Tˆ , take functionals √1 √ 1 Q ∈ O( n × −1Ω)0 and R ∈ O( n × −1ω)0 which satisfy Tˆ = uˆˇ · Qˆ + Rˆ. Then for R √ R 1 any g ∈ N ( −1Ω), we get

hT, gi = hT1, gi = hQ, u ∗ fi + hR, gi = hR, gi. t ω This shows that T = rΩS. 112 R. ISHIMURA AND Y. OKADA

5. Main theorem and examples. By the definition of micro-support (see [KS], Definition 5.1.2), Theorem 4.4 implies that (5.1) SS(H0(S)) ⊂ Car(u∗). So recalling Remark 5.1.4 in [KS], by (5.1) and Corollary 3.2, we resume our results of the preceding sections in the following final theorem: Theoren 5.1. Under the condition (S), we have SS(S) ⊂ Car(u∗). R e m a r k . For related topics, see the references; especially we cite a series of works of Korobe˘ınik[Ko1]–[Ko3], Napalkov [N], Morzhakov [MO1]–[Mo3], Epifanov [E1]–[E3], Tkachenko [T] and Berenstein, Gay, Méril,Struppa and Vidras ([BeGV], [BeSt1], [BeSt2], [MéSt]and [V]).

References

[Ao] T. Aoki, Existence and continuation of holomorphic solutions of differential equations of infinite order, Adv. in Math. 72 (1988), 261–283. [BeGV] C. A. Berenstein, R. Gay and A. Vidras, Division theorems in the spaces of entire functions with growth conditions and their applications to PDE of infinite order, preprint. [BeSt] C. A. Berenstein and D. C. Struppa, A remark on “convolutors in spaces of holomorphic functions”, in: Lecture Notes in Math. 1276, Springer, 1987, 276–280. [BeSt] —, —, Solutions of convolution equations in convex sets, Amer. J. Math. 109 (1987), 521–544. [BoS] J. M. Bony et P. Schapira, Existence et prolongement des solutions holomorphes des équationsaux dérivéespartielles, Invent. Math. 17 (1972), 95–105. [E1] O. V. Epifanov, Solvability of convolution equations in convex domains, Mat. Za- metki 15 (1974), 787–796 (in Russian); English transl.: Math. Notes 15 (1974), 472–477. [E2] —, On the surjectivity of convolution operators in complex domains, Mat. Zametki 16 (1974), 415–422. (in Russian); English transl.: Math. Notes 16 (1974), 837–841. [E3] —, Criteria for a convolution to be epimorphic in arbitrary regions of the complex plaine, Mat. Zametki 31 (1974), 695–705 (in Russian); English transl.: Math. Notes 31 (1982), 354–359. [F] Yu. Favorov, On the addition on the indicators of entire and subharmonic functions of several variables, Mat. Sb. 105 (147) (1978), 128–140 (in Russian). [H1] L. Hörmander, On the range of convoluton operators, Ann. of Math. 76 (1962), 148–170. [H2] —, An Introduction to Complex Analysis in Several Variables, Van Nostrand, 1966. [I1] R. Ishimura, Théorèmesd’existence et d’approximation pour les équations aux dérivéespartielles linéaires d’ordre infini, Publ. RIMS Kyoto Univ. 32 (1980), 393–415. [I2] —, Existence locale de solutions holomorphes pour les équations différentielles d’ordre infini, Ann. Inst. Fourier (Grenoble) 35 (3) (1985), 49–57. COMPLEX DEFINED BY CONVOLUTION 113

[I3] R. Ishimura, A remark on the characteristic set for convolution equations, Mem. Fac. Sci. Kyushu Univ. 46 (1992), 195–199. [IO] R. Ishimura and Y. Okada, The existence and the continuation of holomorphic solutions for convolution equations in tube domains, Bull. Soc. Math. France 122 (1994), 413–433. [IOk] —, —, Sur la condition (S) de Kawai et la propriétéde croissance régulière d’une fonction sous-harmonique et d’une fonction entière, Kyushu J. Math. 48 (1994), 257–263. [Kan] A. Kaneko, Introduction to Hyperfunctions, Kluwer Acad. Publ., 1988. [KS] M. Kashiwara and P. Schapira, Sheaves on Manifolds, Grundlehren Math. Wiss. 292, Springer, 1990. [Kaw] T. Kawai, On the theory of Fourier hyperfunctions and its applications to partial differential equations with constant coefficients, J. Fac. Sci. Univ. Tokyo Sect. IA Math. 17 (1970), 467–517. [Ki] C. O. Kiselman, Prolongement des solutions d’une équationaux dérivéespartielles àcoefficients constants, Bull. Soc. Math. France 97 (1969), 329–356. [Ko1] Yu. F. Korobe˘ınik, On solutions of some functional equations in classes of functions analytic in convex domains, Mat. Sb. 75 (1968), 225–234 (in Russian); English transl.: Math. USSR-Sb. 4 (1968), 203–211. [Ko2] —, The existence of an analytic solution of an infinite order differential equation and the nature of its domain of analyticity, Mat. Sb. 80 (1969), 52–76 (in Russian); English transl.: Math. USSR-Sb. 9 (1969), 53–71. [Ko3] —, Convolution equations in the complex domain, Mat. Sb. 127 (1699 (1985) (in Russian); English transl.: Math. USSR-Sb. 55 (1985), 171–193. [LG] P. Lelong and L. Gruman, Entire Functions of Several Complex Variables, Grund- lehren Math. 282, Springer, 1986. [L] B. Ja. Levin, Distribution of Zeros of Entire Functions, Trans. Math. Monographs 5, Amer. Math. Soc., 1964. [M] B. Malgrange, Existence et approximation des solutions des équationsaux dérivées partielles et des équationsde convolution, Ann. Inst. Fourier (Grenoble) 6 (1955– 1956), 271–354. [MéSt] A. Méril and D. C. Struppa, Convolutors in spaces of holomorphic functions, in: Lecture Notes in Math. 1276, Springer, 1987, 253–275. [Mo1] V. V. Morzhakov, Convolution equations in spaces of functions holomorphic in n convex domains and on convex compacta in C , Mat. Zametki 16 (1974), 431–440 (in Russian); English transl.: Math. Notes 16 (1974), 846–851. l [Mo2] —, On epimorphicity of a convolution operator in a convex domains in C , Mat. Sb. 132 (174) (1987), 352–370 (in Russian); Math. USSR-Sb. 60 (1988), 347–364. n [Mo3] —, Convolution equations in convex domains of C , in: Complex Anal. and Appl. ’87, Sofia, 1989, 360–364. [N] V. V. Napalkov, Convolution equations in multidimensional spaces, Mat. Zametki 25 (1979), 761–774 (in Russian); English transl.: Math. Notes 25 (1979). [O] Y. Okada, Stability of convolution operators, Publ. RIMS Kyoto Univ., to appear. [Sé] A. Sébbar, Prolongement des solutions holomorphes de certains opérateurs différen- tiels d’ordre infini àcoefficients constants, in: Sém.Lelong–Skoda, Lecture Notes in Math. 822, Springer, 1980, 199–220. 114 R. ISHIMURA AND Y. OKADA

[T] V. A. Tkachenko, Equations of convolution type in spaces of analytic functionals, Izv. Akad. Nauk SSSR Ser. Mat. 41 (1977), 378–392 (in Russian); English transl.: Math. USSR-Izv. 11 (1977), 361–374. [V] A. Vidras, Interpolation and division problems in spaces of entire functions with growth conditions and their applications, Doct. Diss., Univ. of Maryland, 1992. [Z] M. Zerner, Domaines d’holomorphie des fonctions vérifiantune équationaux déri- véespartielles, C. R. Acad. Sci. Paris 272 (1971), 1646–1648.