The articles in this volume were written to commemorate R. Remmert's 60th birthday in 1990. They are surveys, meant

*185*
*68*
*80MB*

*English*
*Pages IX, 310
[319]*
*Year 1998*

- Author / Uploaded
- S.R. Bell
- J.-L. Brylinski
- A.T. Huckleberry
- R. Narasimhan
- C. Okonek
- G. Schumacher
- A. Van de Ven
- S. Zucker

*Table of contents : Front Matter ....Pages i-ix Proper Holomorphic Mappings of Complex Spaces (Steven R. Bell, Raghavan Narasimhan)....Pages 1-38 An Overview of Recent Advances in Hodge Theory (Jean-Luc Brylinski, Steven Zucker)....Pages 39-142 Actions of Groups of Holomorphic Transformations (Alan T. Huckleberry)....Pages 143-196 Stable Bundles, Instantons and C∞-Structures on Algebraic Surfaces (Ch. Okonek, A. Van de Ven)....Pages 197-249 The Theory of Teichmüller Spaces A View Towards Moduli Spaces of Kähler Manifolds (Georg Schumacher)....Pages 251-310*

Complex Manifolds

Springer-Verlag Berlin Heidelberg GmbH

S. R. Bell J.- L. Brylinski A. T. Huckleberry R. Narasimhan C. Okonek G. Schumacher A. Van de Ven S. Zucker

Complex Manifolds With 4 Figures

~in ~ Springer

Consulting Editors of the Series: A.A. Agrachev, A.A. Gonchar, D. L. Kelendzheridze, E. F. Mishchenko, N.M. Ostianu, V. P. Sakharova, A. B. Zhishchenko Editor: Z. A. Izmailova Tide of the Russian edition: Itogi nauki i tekhniki, Sovremennye problemy matematiki, Fundamental'nye napravleniya, Vol. 69, Kompleksnyj analiz - mnogie peremennye 6 Publisher VINITI, Moscow 1991

Second, Corrected Printing 1997 of the First Edition 1990, which was originally published as Several Complex Variables VI, Volume 69 of the Encyclopaedia of Mathematical Seien ces.

Die Deutsche Bibliothek - CIP-Einheitsaufnahme Complex manifolds 1by S. R. Bell ... Ed.: W. Barth; R. Narasimhan. -

ISBN 978-3-540-62995-5 DOI 10.1007/978-3-642-61299-2

ISBN 978-3-642-61299-2 (eBook)

Mathematics Subject Classification (1991): 32Axx, 32CXX, 32Hxx, 32Lxx, 32Mxx, 14Jxx

© Springer-Verlag Berlin HeideIberg 1998 Originally published by Springer-Verlag Berlin HeideIberg New Y ork in 1998 This work is subject to copyright. All rights are reserved. whether the whole or part of the material is concerned. specifically the rights of translation. reprinting. reuse of illustrations. recitation. broadcasting. reproduction on microfilms or in any other way. and storage in data banks. Duplication of this publication or parts thereof is permitted only under the provisions of the German Copyright Law of September 9. 1965. in its eurrent version. and permission for use must a1ways be obtained from Springer-Verlag. Violations are liable for prosecution under the German Copyright Law.

SPIN: 10629660 41/3143-54 3 2 1 0 - Printed on acid-free paper.

A Volume in Honour of Reinhold Remmert

List of Editors and Contributors Editor-in-Chief R.V. Gamkrelidze, Russian Academy of Sciences, Steklov Mathematical Institute, ul. Gubkina 8, 117966 Moscow, Institute for Scientific Information (VINITI), ul. Usievicha 20 a, 125219 Moscow, Russia; e-mail: [email protected]

Consulting Editors W. Barth, Universität Erlangen-Nürnberg, Mathematisches Institut, Bismarckstraße 11/2, D-91054 Erlangen, Germany R. Narasimhan, The University of Chicago, Department of Mathematics, 5734 University Avenue, Chicago, IL 60637, USA

Gontributars S. R. Bell, Purdue University, Department of Mathematics, Mathematical Seiences Building, West Lafayette, IN 47907, USA J.-L. Brylinski, The Pennsylvania State University, Department of Mathematics, 305 McAllister Building, University Park, PA 16802, USA A. T. Huckleberry, Fakultät und Institut für Mathematik, Ruhr-Universität Bochum, Universitätsstraße 150, D-44801 Bochum, Germany R. Narasimhan, The University of Chicago, Department of Mathematics, 5734 University Avenue, Chicago, IL 60637, USA C. Okonek, Universität Zürich, Mathematisches Institut, Winterthurerstraße 190, CH-8057 Zürich, Switzerland G. Schumacher, Fakultät und Institut für Mathematik, Ruhr-Universität Bochum, Universitätsstraße 150, D-44801 Bochum, Germany A. Van de Yen, Mathematisch Institut, Rijksuniversiteit Leiden, Department of Mathematics and Computer Science, P.O. Box 9512, 2300 RA Leiden, The Netherlands S. Zucker, The Johns Hopkins University, Department of Mathematics, Baltimore, MD 21218, USA

Preface

This volume was originally intended to appear in June 1990 as an independent book. During the final stages of production we felt that it would be a valuable addition to this series, and we would reach a wider audience with its publication in English (Springer-Verlag), as well as in Russian (VINITI). I am vecy glad that the authors of the articles and the editors of the volume unanimously agreed to the suggestion about its inclusion in the Encyclopaedia. We hope the volume will thus be of use to a larger section of the mathematical community. August 1990

R. Gamkrelidze

Foreword

This volume is to appear as Reinhold Remmert starts out on bis seventh decade. There are five contributions, each dealing with a different aspect of Camplex Analysis. These essays are not meant to present new results; rather, they are meant to facilitate access to some, at least, of the many branches of the subject by collecting tagether those topics which, in the author's view, seemed the most useful as an introduction to the respective fields. Forty years ago, the fundamental problems of Camplex Analysis still awaited solution, and it was essential to fmd and sharpen the appropriate tools. Today, it is difficult to discem even the main directions ofthe subject and their mutual relations. Hence the essays in this volume are addressed to mathematicians who are not specialists in these aspects of complex analysis. lf complex analysis now appears as a well organized modern branch of mathematics, Reinhold Remmert was arnong the few who made it so. His discoveries, bis high standards and bis books and editorial work have influenced not only complex analysis, but many other fields of mathematics as well. Remmert's work touches, and unites, the most diverse aspects of complex analysis. The subject is of the greatest importance because of its lang and distinguished history and because of its connections with so many branches of science, both inside and outside mathematics. Thus, it is not surprising ifthere is a tendency for different parts of the subject to diverge in different directions. Keeping these tendencies in balance was always one of Remmert's main intentions. We hope that the present volume will reflect at least some ofthe richness and balance of Complex Analyis. The volume is dedicated toReinhold Remmert by bis many friends, students, and colleagues as a token oftheir gratitude and adrniration. June 1990

Wolf Barth Raghavan Narasimhan

Table of Contents

Proper Holomorphic Mappings of Complex Spaces Steven R.Bell and Raghavan Narasimhan 1

An Overview of Recent Advances in Hodge Theory Jean-Luc Brylinski and Steven Zucker 39

Actions of Groups of Holomorphic Transformations Alan T. Huckleberry

143

Stahle Bundles, Instantons and Coo-Structures on Algebraic Surfaces Ch. Okonek and A. Van de Ven

197

The Theory of Teichmüller Spaces A View Towards Moduli Spaces of Kähler Manifolds Georg Schumacher

251

Proper Holomorphic Mappings of Complex Spaces Steven R. Bell* and Raghavan Narasimhan

Contents Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1. 2. 3. 4.

Remmert's Proper Mapping Theorem .... .. .... .................. .... .. Grauert's Direct Image Theorem .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. . Embeddings in ccn .... ...... .. ...... .......... ........... ........ .... .... Proper Holamorphie Maps into Bounded Domains . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. .. . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

• Research supported by NSF Grant DMS 8420754 and an AMS Fellowship

2 4 7 15 23 33

2

Introduction

Introduction In this article, we attempt to describe some of the most important results conceming proper holomorphic mappings between complex spaces. The first two sections deal respectively with the Remmert Proper Mapping Theorem and the Grauert Direct Image Theorem. These are certainly the most important results of what may be called the general theory. Sect. 3 deals with embeddings in CCN. Finally, Sect. 4 treats proper maps between bounded domains, which is, perhaps, the domain of greatest recent and current research. Wehave not dealt with this in as great detail as we might have because of the availability of the excellent survey articles of Bedford [Bed] and Forstneric [FF 1]. We shall be using the terminology of several complex variables without defining the standard notions. For complex spaces, coherent sheaves and Stein and holomorphically convex spaces, one may consult the books of GrauertRemmert [GR 1, GR 2]. For pseudoconvexity, strict (= strong) pseudoconvexity and plurisubharmonic functions, one may consult the notes of FomaessStensenes [FS]. The study of proper holomorphic maps between complex spaces is of great importance for several reasons. [We shall often call them, simply, proper maps.] In the first place, the existence of a non-trivial proper map from one complex space X to another, Y, often places strong restrictions on these spaces. Examples of this are: 1. If X, Y are compact Riemann surfaces and if there is a non-constant holomorphic map TC : X --+ Y, then, the genus of Y is :5: that of X. 2. If X is a Stein space and TC : X --+ Y a proper map, then TC(X) is again a Stein space. [If Y is Stein, then X need only be holomorph-convex in general.] Further, even when proper maps exist, they tend to be "rigid" and rather few in number. For instance, if n ~ 2, a proper map of the unit ball in ccn into itself is an analytic automorphism. On the other band, proper holomorphic mappings do occur in important contexts. We now mention some of these contexts. 1. Finite Mappings. Let X, Y be complex spaces. A holomorphic map TC :X--+ Y is said tobefinite if TC is proper and TC- 1(y) is finite Vy E Y. If TC : X --+ Y is finite, then TC(X) is an analytic set in Y (see Sect. 1 below) and has the following property: if x E X and y = TC(x), then the local ring (!)x.x of X at x is a finite module over the local ring (!)Y,y of Y at y. A finite map TC : X --+ Y is called a (finite) ramified covering if X, Y are reduced, if TC(X) = Y and if dimx X = dimy Y for all x E X, y E Y. In this case, there exists a nowhere dense analytic set B c: Y such that TCI X - TC- 1(B) --+ Y - B isafinite unramified covering and TC- 1(B) is nowhere densein X. The smallest such B is called the branch locus of TC; further, if x E X- TC- 1(B) and y = TC(x), then the induced map TC. : (!)Y,y --+ (!)x.x is an isomorphism (i.e. TC is a local isomorphism at x). If TC :X --+ Y is a ramified covering and Y is irreducible,

Introduction

3

then the cardinality of n:- 1(y) is independent of y E Y - B and is called the degree of n (other names being the number of sheets or multiplicity). Exarr.ples are, of course, provided by finite unramified coverings. Moreover, any irreducible projective variety of dimension n can be realized as a ramified covering of the (complex) projective space JP'I, while any irreducible affine variety of dimension n can be realized as a ramified covering of ccn. Also, any pure dimensional complex space (of dimension n) can locally be realized as a ramified covering of an open set in ccn. (These are versions of the so-called Noether normalization Iemma.) In this connection, we mention a deep theorem of Grauert and Remmert [GR 3] which implies that a ramified covering n : X --+ ccn is an affine variety if the closure in JP'I of the branch locus of n is an analytic set in JP'I. Their theorem can be formulated as follows:

Let S be a connected normal complex space, T c S a proper analytic set. Let Y = S - T, Iet X be a normal complex space and n : X --+ Y, a ramified covering. Then n extends to a ramified covering 1f : X --+ S (in an obvious sense) if and only if the closure B in S of the branch locus B c Y of n is an analytic set in Y. lt is not hard to find a topological extension 1f : X --+ S ; this is done in just the same way that one adds algebraic branch points to the "Riemann domain" of a holomorphic function. There is also a sheaf (!)x of : V -+ S of V into a reduced complex space S ( depending on q>) such that q> o p is holomorphic on p- 1(V) separates the points of V.

2. Grauert's Direct Image Theorem

12

Proof of Theorem 2.4. Let U satisfy the conditions above and Iet Q = p- 1 (U). If S is a reduced complex space, consider the class Hs = Hs(U) of holomorphic maps qJ : Q - S of the form qJ = cp o p where cp is continuous on U. Let

= {(s, s) ES x S} be the diagonal of S and Iet A~ = (qJ x qJ)- 1(As) c Q x Q (qJ E Hs). Let V~ U and W = p- 1 (V) x p- 1{V) c Q x Q. The hypothesis of As

the theorem means that

(where Aa is the diagonal of Q). Since W is relatively compact in Q, there is a finite subset {qJ(X}(X=l, ... ,N of UsHs such that W

n

n

A~ = W n

S,Hs

n N

A~. = W n Aa = Ap-'(V)·

(X= I

Thus, if we replace X by p- 1 (V) and consider the map F = qJ 1 x ... x qJN : p- 1 (V)- S1 x ... x SN, we may assume the following: there is a holomorphic map F : X - S such that two points x, y are equivalent under R if and

only if F(x) = F(y). Further, we may assume that F is proper. To see this, Iet F = cp o p, cp : Y - S. The above assumption means that cp is injective. Let U0 be a relatively compact open set in Y and L = oU0 ; L is compact, so that S' = S -cp(L) is open inS. But cpiUo, considered as a map into S' is proper: if C c S' is compact, cp- 1 (C) n Uo is disjoint from an open set in Y containing oU0 (e.g. cp- 1 (S- C)) and so is closed in Y; since Uo ~ Y,cp- 1 (C) is compact. Wehave only to replace X by p- 1(U0 ) and S by S'. Thus, it remains to prove the following:

Proposition 2.5. Let F : X - S be a proper holomorphic map between reduced complex spaces and let R be the equivalence relation defined by F : xRy F(x) = F(y). Then (X/R,(!Jx/R) is a complex space. Proof of 2.5. Replacing S by F(X) (which is analytic by Remmert's theorem), we may assume that F is onto S. Hence F induces a homeomorphism of X/R onto S, and we may consider (!) = (!Jx/R as a sheaf on S. We shall show below that (8 is (!)8 -coherent. This implies the result: we may assum shrinking S that (!) is generated (everywhere on S) by finitely many sections /J, ... ,fk E r(S,@). Let gj = fj o p and A the image of X in S x ([;k by the holomorphic map x ~--+ (F(x), g 1(x), ... , gk(x)). A is analytic by Remmert's theorem, and we see at once that (S, (!)) is isomorphic to (A, (!)A). To show that (8 is coherent, consider D = (F x F)- 1(As) c X x X (As is the diagonal of S). Let h : D - S be the map (x,y) ~--+ F(x)(= F(y)). If U is open in X and f is holomorphic on U, we define two holomorphic functions /J,h on (U x U) n D by: fi(x,y) = f(x),f2(x,y) = f(y). This defines two (!)8 -morphisms IX 1,1X2 from F.((!Jx) to h.((!JD). Moreover, to say that f is constant on fibres of F means simply that f(x) = f(y) if (x,y) E D, i.e. that

2. Grauert's Direct Image Theorem

ft = f2.

13

Hence ß.i = ker(IX 1 -1X2). Since F.(lPx) and h.(lPD) are es-coherent, so

also is ß.i.

Cartan's theorem implies the existence of the quotient complex space under the action of a properly discontinuous group; this is due again to Cartan. In fact he bad proved this earlier without the use of Grauert's direct image theorem. A group r of homeomorphisms acting on a locally compact space X is said to be properly discontinuous if, for any compact set K c X, the set {y E Tjy(K) n K f 0} is finite.

Theorem 2.6. Let X be a reduced complex space and r a properly discontinuous group of analytic automorphisms of X. Then X Ir carries a natural structure of complex space. If x EX, the isotropy group Tx = {y E Tjy(x) = x} is finite. lt is easy to find a neighborhood U of x such that rx(U) = U and such that if y E r and y(U) n U f 0, then y E rx. We have thus to consider UITx; here the conditions of Theorem 2.5 are easily verified. In fact r x-invariant holomorphic functions on U separate points of U Ir x if U is small enough as is seen by averaging holomorphic functions on U: consider z ~---+ Lyer. f(yz) where f is holomorphic on U (and z E U). Another application of Cartan's theorem is the proof of a result of Remmert on holomorphically convex spaces (which again was first proved, in a slightly weaker form, before the direct image theorem). Recall that a complex space X is called holomorphically convex if for any non-compact set E c X, there is a function f holomorphic on X such that f(E) is unbounded (in o f) E C00 (Q 2). Since the components of f are bounded holomorphic functions, we have 18a.fj/8za.l ~ ca.d 1(z)-ia.i (by the Cauchy inequalities) where the constant ciX is independent of z E !2 1 and dj(z) denotes the distance of z from the boundary of QjU = 1, 2). Since q> vanishes to infinite order on 8!22, to show that the derivatives of u · (cp o f) are bounded, it suflices to check that d2{f(z)) ~ Cdr (z) (with the constant C independent of z E Qt). To do this, Iet p be a strictly plurisubharmonic function defining 8!2 1. Then p o f- 1 is subharmonic and < 0 on !22, so that, by the Hopf Iemma (see next paragraph), p o f- 1(z) ~ -Cd2 (z). Since lp(w)l ~ Cd 1 (w), the required inequality follows and completes the proof. The Hopf Iemma asserts the following. If Q is a bounded domain with C 2 boundary in Rn, a E 8Q and v(a) is theinward normal to 8Q at a, then, for any subharmonic function u on Q with u < 0 on Q, there is a constant c > 0 suchthat -Ii u(x) YaE 8Q m--. 0. The Iemma follows easily. Of course, this inequality is only of interest when limu(x) = 0, x -+ a along v(a). E. Hopf [Hop] used this argument to prove a strong maximum principle for general second order linear elliptic operators.

4. Proper Holomorphic Maps into Bounded Domains

27

The proof sketched above of Fefferman's theorem can be generalized to several other situations, notably to the case of proper maps between certain pseudoconvex domains. This was done by Bell and Catlin [BC] and by Diederich and Fomaess [DF 2]. Before stating this theorem, we give a very simple example to illustrate the behavior. Let p, q > 0 be integers, j(z1, z2) = (zf, zi). f maps the smooth (weakly) pseudoconvex domain E = {z E 0 and a constant C > 0 so that d2(f(z))N s Cd 1(z). If Q 1 is strictly pseudoconvex, the argument given earlier shows that we may take N = 1. In general, we use the following theorem of Diederich-Fomaess [DF 1] (see Range [Ra] for this argument): there is a C00 boundary defining function p for Q 1 and an integer N > 0 such that the function p = -(-p)fr is continuous on Q 1, and plurisubharmonic on Q 1• (It is, of course, < 0 on Q, and = 0 on oQJ.) We consider the function A.( w) = sup{p(z); z E f- 1 ( w)}; it is clearly a continuous plurisubharmonic function < 0 on Q 2 - V2 and so extends to a plurisubharmonic function on Q 2. Since f is proper, we see that A. is an exhaustion function on Q 2 which extends continuously to oQ 2 and = 0 on oQ2 . We may apply the Hopf Iemma to conclude that A.(w) s -Cd2 (w) and the desired inequality follows easily. Thus, we have shown that u · (h o f) is smooth on :01 if h is holomorphic on Q2 and smooth on Q 2 • We can no Ionger conclude that u =I= 0 on Q 1 and the last step in the proof is a division theorem. One shows that the order of vanishing of u on oQ 1 is bounded, and that this, together with the fact that u · Jj is smooth on :01 for all integers k ~ 0 (and all j = 1, ... , n) implies that u divides ufj in the ring ofholomorphic functions which are smooth on :01.

28

4. Proper Holomorphic Maps into Bounded Domains

Thus f extends smoothly to Q 1• Detailsofthis argumentwill be found in [BC] and [DF2]. This immediately raises the question of when the Bergman projection preserves the space of functions smooth up to the boundary. In one variable, this holds for any smooth bounded domain as follows from Spencer's formula I

Pq>

0

0

= q>- 4 oz G ozq>

where G is the classical Green's operator for the Dirichlet problem, i.e. G satisfies LIGu = u and Gu = 0 on the boundary. Since Gis known to preserve smoothness up to the boundary, the same is true of P. Spencer's formula might give the impression that the Bergman projection is also too closely tied to harmonic functions to be of use in several variables. But, as mentioned earlier, it is tied rather to a i3-problem. If Q is a smooth plane domain, the i3-problem can be described as follows. Given IX E C00 (Q), the Cauchy integral furnishes a solution of ~ = IX with u E C00 (Q). Since u is only determined up to addition of a holomorphic function, it is natural to consider that solution of ~ =IX which is L 2 (Q)-orthogonal to holomorphic functions in L 2• The i3-problem is to prove that this particular solution is in C00 (Q) when IX is. The solution operator A in this case is given in terms of Green's operator: A = 4/zG. In several variables, the Situation is more complicated. Let Q be a bounded domain in

where N is the so called i3-Neumann operator. The best results to date on the regularity properties of the 'ä-Neumann operator have been obtained by Catlin [Ca 1,2,3] who uses the machinery of D'Angelo [D'A] to measure the degree

4. Proper Holomorphic Maps into Bounded Domains

29

of weak pseudoconvexity. Kohn's article [K 3] contains an excellent survey of this subject. The ideas in the proof sketched earlier of Theorem 4.6 have other applications. We shall show how they can be used to prove the following fundamental theorem of Alexander [All]; for a very elegant, direct proof, see the book of Rudin [Ru]. In the restoftbis article, we denote by Bn the unit ball {z E 1 (proved, at least for n = 2, by Poincare). One

30

4. Proper Holomorphic Maps into Bounded Domains

sometimes adds the misleading comment that there is no Riemann mapping theorem in several variables. We next describe two positive mapping theorems. It is not hard to see that the only plane domains Q =!= CC for which the autornorphisrn group is transitive are those biholornorphic to the disc. Bun Wong [Wo] showed that the corresponding staternent is true for strictly pseudoconvex dornains in CC", n > 1. Later, Rosay [Ro] generalized it to the following staternent.

Theorem 4.10. A bounded domain in CC" with C2 boundary whose automorphism group is transitive is biholomorphic to the unit ball in CC"; i.e. a bounded homogeneous domain with C2 boundary is biholomorphic to the ball. lt should be rernarked that bounded homogeneaus dornains are always (without any assurnptions on the boundary) dornains of holornorphy, i.e. pseudoconvex. There are, of course, rnany such dornains without srnooth boundary. The next theorern, due toB. Fridman [Fri], rnay clairn tobe an approximate Riemann rnapping theorern.

Theorem 4.11. Let D be a bounded pseudoconvex domain in CC" diffeomorphic to the unit ball B". Given e > 0, there exist domains D, c: D,D2 c: B" whose boundaries are within e of those of D and B" respectively such that Q 1 and D2 are biholomorphically equivalent. Let us mention also a recent result of Bedford and Pincuk [BP] which confirrns, at least partially, a conjecture of Greene and Krantz [GK].

Theorem 4.12. A ( weakly) pseudoconvex domain in CC2 with real analytic boundary whose automorphism group is non-compact is biholomorphic to a domain of the form {(z, w) E CC2 11zl 2 + lw1 2P < 1},p > 0 an integer. We have described above the approach to boundary regularity based on the Bergman projection and results which can be proved using related techniques. Different rnethods have been found. Pincuk has used the method of nonhomogeneaus dilations which he introduced in connection with Theorem 4.8 to give a new and elernentary proof of the Wong-Rosay theorern. Further, he and Hasanov [PH] have used it in conjunction with a CIJ version of the "edge of the wedge" theorem to give a direct elernentary proof of Fefferrnan's theorern which gives results about extensions in dornains with Ck boundary (k::::;; oo). Lempert [Lern 2] obtained these results independently using bis study of extrernal discs for the Kobayashi rnetric. Other noteworthy elernentary proofs of Fefferrnan's theorem were given by Nirenberg, Webster and Yang [NWY] and by Lernpert [Lern 1]. The survey article of Forstneric [FF 1] contains an excellent surnrnary of these various rnethods. If we consider the questions of extension of holornorphic rnaps past the boundary, the Schwarz refl.ection principle gives the definitive answer in one

4. Proper Holomorphic Maps into Bounded Domains

31

variable. In several variables, the simple geometric picture of refiection is lost; it has been restored only for strictly pseudoconvex domains. The first, and most basic, theorem was proved by Lewy [Le] and Pincuk [P 2].

Theorem 4.13. Let Q1, Q2 be bounded strictly pseudoconvex domains in 1. Then, any proper holomorphic map from B" to BN which is cN-n+I up to the boundary is a rational map. Moreover, Cima and Suffridge [CS 2] have shown that rational proper holomorphic maps cannot have singular points on the boundary, and so extend holomorphically past the boundary. lt is a natural question, and one of very great interest, to find analogues of the embedding theorems for Stein manifolds (Sect. 3) in the case of bounded domains. Of course, there is no single standard domain into which one might attempt to map them; the two obvious candidates are BN and ..1N (the ball and the polydisc). Here, the situation is more complicated than embedding in C[N; one has the following result of N. Sibony [Si].

Theorem 4.17. There exists a bounded pseudoconvex Reinhardt domain in ([2 having C00 smooth boundary which does not admit a proper holomorphic map into any bounded (euclidean) convex open set in C[N for any N. lt tums out, again rather surprisingly, that strictly pseudoconvex domains do admit proper maps into both BN and ..1N. These positive results really go back to the construction by Aleksandrov [All] (in 1982) of inner functions on the ball B"; these are bounded holomorphic functions whose radial Iimits have absolute value 1 almost everywhere on aB"; it bad long been believed that inner functions did not exist when n > 1. A slightly weaker statement bad been obtained by Hakim and Sibony [HM]. Using the ideas of these constructions, Aleksandrov [A 2] (for the ball) and Lew [Lel] proved the following theorem.

Theorem 4.18. Let Q be a bounded strictly pseudoconvex domain in ([" with smooth boundary. Then Q can be properly embedded in ..1N for some N. Lew [Le 2] and Forstneric [FF 3] have also shown that strictly pseudoconvex domains Q can be embedded into BN for some N. Some of these embeddings do not even extend continuously to ao, while some are continuous but not differentiable. There have been several refinements of these theorems. It is of course natural to ask about the smallest codimension that can be allowed in these theorems. The analogue of Bishop's theorem (Theorem 3.2 part 2) would be that there are proper maps in codimension 1 also in this context. One of the present authors long believed that such maps did not exist. However, Berit Stensenes [St] proved the following theorem.

References

33

Theorem 4.19. Any bounded strictly pseudoconvex domain in