1: %Date of acceptance : April 20, 2001
2: \documentclass{proc-l}
3:
4: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
5: %%%%%%%%%%%% PICTURES %%%%%%%%%%%%%%%%%%%%%%%%%
6: \usepackage{amssymb,amsfonts,amscd,xypic}
7: \pagestyle{headings}
8: \usepackage{graphics}
9: \usepackage{epsfig}
10: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
11:
12: % \setlength{\textwidth}{15.25cm}
13: % \setlength{\textheight}{18.5cm}
14:
15: \newtheorem{theorem}{Theorem}[section]
16: \newtheorem{lemma}[theorem]{Lemma}
17:
18: \theoremstyle{definition}
19: \newtheorem{definition}[theorem]{Definition}
20: \newtheorem{example}[theorem]{Example}
21: \newtheorem{xca}[theorem]{Exercise}
22:
23: \theoremstyle{remark}
24: \newtheorem{remark}[theorem]{Remark}
25:
26: \renewcommand{\subjclassname}{
27: \textup{2000} Mathematics Subject Classification}
28:
29: \copyrightinfo{2001}{American Mathematical Society}
30:
31: \numberwithin{equation}{section}
32:
33: \begin{document}
34:
35: %\Large
36:
37: \title{On the Hartogs--Bochner phenomenon
38: for CR functions in $P_2(\mathbb{C})$}
39:
40: \author{Roman Dwilewicz}
41:
42: \address{Institute of Mathematics,
43: Polish Academy of Sciences,
44: \'Sniadeckich 8, P.O.~Box 137, 00-950 Warsaw,
45: Poland}
46:
47: \thanks{
48: Partially supported by a grant of the
49: Polish Committee for Scientific Research
50: KBN 2 PO3A 044 15 and by a grant from the French-Polish
51: program ``Polonium 1999''.}
52:
53: \author{Jo\"el Merker}
54:
55: \address{ Laboratoire d'Analyse, Topologie et Probabilit\'es,
56: Centre de Math\'ematiques et Informatique, UMR 6632,
57: 39 rue Joliot Curie,
58: F-13453 Marseille Cedex 13, France.
59: Fax: 00 33 (0)4 91 11 35 52}
60:
61: \subjclass{Primary 32V25. Secondary 32V10, 32V15, 32D15}
62:
63: \commby{Steve Bell}
64:
65: \keywords{Smooth hypersurfaces of the complex projective space,
66: Holomorphic extension of CR functions, Jump formula, Global
67: minimality, One-sided neighborhood}
68:
69: \begin{abstract}
70: Let $M$ be a compact, connected, $\mathcal{C}^2$-smooth and {\it
71: globally minimal} hypersurface $M$ in $P_2(\mathbb{C})$ which divides
72: the projective space into two connected parts $U^{+}$ and $U^{-}$. We
73: prove that there exists a side, $U^-$ or $U^+$, such that every
74: continuous CR function on $M$ extends holomorphically to this
75: side. Our proof of this theorem is a simplification of a result
76: originally due to F. Sarkis.
77: \end{abstract}
78:
79: \maketitle
80:
81: \section{Introduction}
82:
83: Historically, one of the amazing theorems in the theory of holomorphic
84: functions of several complex variables is the theorem of Hartogs of
85: 1906 about extension of functions from a neighborhood of the boundary
86: of a domain to the inside of the domain. This theorem was generalized
87: by Bochner in 1943, namely that it is enough to consider
88: $\mathcal{C}^1$-smooth functions defined just on the boundary of the
89: domain that satisfy the tangential Cauchy-Riemann equations. Since
90: that time many versions and generalizations of the theorem appeared,
91: {\it see} \, for instance, Kohn-Rossi \cite{KR}, Ehrenpreis \cite{E},
92: Ivashkovich \cite{I}, Harvey \cite{H}, Harvey-Lawson \cite{HaLa},
93: Laurent-Thi\'ebaut \cite{La}, Dolbeault-Henkin \cite{DH}, Sarkis
94: \cite{S1}, \cite{S2}, and others - for a review see \cite{La}.
95:
96: Recently, solving the boundary problem in the sense of Harvey and
97: Lawson for the graph of CR functions defined on boundaries of domains
98: in disc-convex K\"ahler manifolds, Sarkis \cite{S1} proved extension
99: of CR-meromorphic mappings with values in $P_2(\mathbb{C})$ (for
100: $P_n(\mathbb{C})$, $n\geq 3$, {\it see} \cite{HaLa}, {\it part
101: II}). As an application, he obtained Theorem 1.1 below. Our
102: contribution to this subject essentially lies in a simplification of
103: his proof, as will appear below.
104:
105: Recall that a CR manifold $M$ (locally embeddable or not) is called
106: {\it globally minimal} if any two points of it can be joined by a
107: piece-wise smooth curve running in complex tangential directions
108: (\cite{Tr}). The result is the following:
109:
110: \begin{theorem}
111: Let $M$ be a compact connected $\mathcal{C}^2$-smooth real
112: hypersurface in $P_2(\mathbb{C})$ that divides the projective space
113: into two open parts $U^-$ and $U^+$. If $M$ is globally minimal, then
114: \begin{enumerate}
115: \item[{\bf (1)}] There exists a side, $U^-$ \text{\rm or} $U^+$, to
116: which every continuous CR function on $M$ extends holomorphically.
117: \item[{\bf (2)}] All holomorphic functions on the other side of $M$
118: which are continuous up to $M$ are constant.
119: \end{enumerate}
120: \end{theorem}
121:
122: \begin{remark}
123: Theorem 1.1 also holds true ({\it cf.} \cite{S1}, \cite{S2}) if,
124: instead of a globally minimal $M$ as above and instead of CR functions
125: on $M$, we consider an arbitrary $\mathcal{C}^2$-smooth hypersurface
126: as above and {\it holomorphic functions in a neighborhood
127: $\mathcal{V}$ of $M$ in $P_2(\mathbb{C})$}. Our proof is valid in
128: $P_n(\mathbb{C})$ for $n\geq 2$. Of course, the implication {\bf (1)}
129: $\Rightarrow$ {\bf (2)} is trivial.
130: \end{remark}
131:
132: \medskip
133: \noindent
134: {\it Summary of the proof.} \ Let $f$ be a continuous CR function on
135: $M$. Using global minimality, we shall extend $f$ to a one-sided
136: neighborhood $\mathcal{V}^\pm(M)$ of $M$ in $P_2(\mathbb{C})$. By
137: deforming $M$ into $\mathcal{V}^\pm(M)$, we shall argue that we can
138: suppose that $M$ and $f$ are $\mathcal{C}^\infty$-smooth in the
139: assumptions of Theorem~1.1, and even that $f$ is holomorphic in a
140: neighborhood of $M$. As the Dolbeault cohomology group
141: $H^{0,1}(P_n(\mathbb{C}))$ vanishes for $n\geq 2$ ({\it see}
142: \cite{GH}, \cite{HeLe}), every $\mathcal{C}^\infty$ CR function $f$ on
143: $M$ can be decomposed as a jump $f = f^+ - f^-$ of some functions
144: $f^\pm$, holomorphic on $U^\pm$ and $\mathcal{C}^\infty$ on
145: $\overline{U^\pm}$. This decomposition property (which holds without
146: assuming global minimality) easily implies that points {\bf (1)} and
147: {\bf (2)} from Theorem 1.1 are equivalent in the $\mathcal{C}^\infty$
148: category ({\it see} Lemma~3.1). Afterwards, using a theorem of
149: Takeuchi \cite{T}, we may embed a strip neighborhood of $M$ in
150: $\mathbb{C}^N$ which then bounds a complex manifold to which $f^-$
151: {\it and} $f^+$ {\it both extend holomorphically} ({\it see}\cite{H}
152: and \cite{HaLa}, {\it part I}), and we easily deduce using the maximum
153: principle that either $f^-$ or $f^+$ is constant.
154:
155: \medskip
156: \noindent
157: {\it Open question.} \ In Theorem 1.1, the question arises whether the
158: assumption that $M$ is globally minimal can be removed. This question
159: appears to have deep relations to some well-known conjectures of
160: foliation theory, especially the (non-)existence of Levi flat
161: hypersurfaces in projective spaces. Indeed, every compact CR manifold
162: $M\subset P_2(\mathbb{C})$ can be decomposed as a disjoint union of CR
163: orbits. Some of them are of dimension 3, and the others are of
164: dimension 2, i.e., Riemann surfaces, which cannot be closedly embedded
165: in $M$. The closure of each such Riemann surface defines a non-trivial
166: lamination of $P_2(\mathbb{C})$ and every other Riemann surface orbit,
167: having common points with the closure, is dense in this
168: lamination. {\it The open question is whether such laminations exist}
169: ({\it cf.}~\cite{Gh}). For $n\geq 3$ (only), this question has been
170: fixed negatively for laminations arising from global foliations of
171: $P_n(\mathbb{C})$, for real analytic Levi-flat hypersurfaces ({\it
172: see} Cerveau \cite{C}, Lins Neto \cite{Li}) and for smooth Levi-flat
173: hypersurfaces ({\it see} Siu \cite{Si}). Thus, as a part of the
174: folklore, we conjecture that every smooth hypersurface $M\subset
175: P_n(\mathbb{C})$, $n\geq 2$, is globally minimal. \endremark
176:
177: \medskip
178: \noindent
179: {\it Acknowledgement.} \
180: We are grateful to the referee for his clever support in fixing some
181: incorrections in the manuscript and for having pointed out to us substantial
182: improvements in the proofs.
183:
184: \section{Decomposition of CR functions}
185:
186: As we are essentially interested in the geometric approach and not to
187: the best suited regularity assumptions, we shall restrict our
188: attention to the $\mathcal{C}^\infty$-smooth category. By a
189: deformation argument, we shall see in \S3 below that the case where
190: $M$ is $\mathcal{C}^2$ and the CR function $f$ on $M$ is only
191: continuous can be reduced to the case where both $M$ and $f$ are
192: $\mathcal{C}^\infty$. We recall that $H^{0,1}(P_2(\mathbb{C}))=0$ (for
193: $\mathcal{C}^\infty$-smooth forms). It is easy to deduce:
194:
195: \begin{lemma}
196: Let $M$ be a $\mathcal{C}^\infty$ real hypersurface in
197: $P_2(\mathbb{C})$ dividing it into two open parts $U^-$ and $U^+$.
198: Every $\mathcal{C}^\infty$-smooth CR function $f$ on $M$ decomposes as
199: $f=f^+-f^-$, where $f^{\pm}\in \mathcal{O}(U^{\pm})\cap
200: \mathcal{C}^\infty(\overline{U^{\pm}})$.
201: \end{lemma}
202:
203: \begin{proof}
204: We can extend $f$ to a $\mathcal{C}^\infty$-smooth function $F$ over
205: $P_2(\mathbb{C})$ in such a way that supp\,$F$ lies in an arbitrarily
206: small neighborhood of $M$ and $\overline \partial F|_M$ vanishes to
207: infinite order. We define
208: $\omega = \overline \partial F$ on $U^{+}$ and $\omega = 0$
209: on $U^{-}$.
210: The form $\omega$ is a $\mathcal{C}^\infty$-smooth $(0,1)$-form. As
211: $H^{0,1}(P_2(\mathbb{C}))=0$, we can solve the equation $\overline
212: \partial u = \omega$ with $u$ of class $\mathcal{C}^\infty$. So we
213: have
214: $$
215: \aligned
216: & \overline \partial (F - u) = 0 \quad \text{on} \quad U^{+}
217: \quad \text{i.e., $F - u$ is holomorphic on $U^{+}$}, \\ & \overline
218: \partial u = 0 \qquad \qquad \text{on} \quad U^{-} \quad \text{i.e.,
219: $u$ is holomorphic on $U^{-}$,}
220: \endaligned
221: $$
222: and furthermore
223: $$
224: F = (F-u) - (-u), \qquad F|_M = f.
225: $$
226: Obviously the components of the decomposition are of class
227: $\mathcal{C}^\infty$.
228: \end{proof}
229:
230: \section{Embedding of a strip neighborhood of $M$ in $\mathbb{C}^N$}
231:
232: Let $M$ be a $\mathcal{C}^\infty$-smooth hypersurface in
233: $P_2(\mathbb{C})$ bounding $U^-$ and $U^+$ as above. As a preliminary
234: to the proof of Theorem~1.1, we may establish an interesting
235: equivalence between {\bf (1)} and {\bf (2)} in the
236: $\mathcal{C}^\infty$ category.
237:
238: \begin{lemma}
239: Let $M$ be a $\mathcal{C}^\infty$ real hypersurface in
240: $P_2(\mathbb{C})$ dividing it into two open parts $U^-$ and
241: $U^+$. Then the following properties are equivalent:
242: \begin{enumerate}
243: \item[{\bf (1')}] There exists a side, $U^-$ \text{\rm or} $U^+$, to
244: which every $\mathcal{C}^\infty$ CR function on $M$ extends
245: holomorphically.
246: \item[{\bf (2')}] All holomorphic functions on the other side of $M$
247: and $\mathcal{C}^\infty$ up to $M$ are constant.
248: \end{enumerate}
249: \end{lemma}
250:
251: \begin{proof}
252: Without loss of generality, we can fix $U^+$ to be this side. Assume
253: therefore that every $\mathcal{C}^\infty$-smooth CR function on $M$
254: extends holomorphically to $U^+$. Let $g\in \mathcal{O}(U^-)\cap
255: \mathcal{C}^\infty (\overline{U^-})$. Then the trace $g\vert_M$
256: extends holomorphically to $U^+$. So $g$ extends holomorphically to
257: $P_2(\mathbb{C})$, whence it is constant; this is {\bf (2')}.
258:
259: Conversely, suppose that $\mathcal{O}(U^-)\cap
260: \mathcal{C}^\infty(\overline{U^{-}})$ consists of constant functions
261: only. Let $f$ be a $\mathcal{C}^\infty$-smooth CR function on $M =
262: \partial U^-$. From Lemma~2.1 we obtain that $f = f^+ - f^-$, where
263: $f^+$ and $f^-$ belong to the evident spaces. By hypothesis, $f^-$ is
264: constant. Consequently, the CR function $f$ holomorphically extends to
265: $U^+$; this is {\bf (1')}.
266: \end{proof}
267:
268: Let now $M$ be an orientable, $\mathcal{C}^2$-smooth, real hypersurface in
269: a complex manifold $X$. By $\mathcal{V}^\pm(M)$ we denote a {\it
270: one-sided} (global) {\it neighborhood} of $M$, {\it i.e.} a connected
271: open set which contains a (local) one-sided neighborhood of $M$ at
272: each point of $M$ and such that $\mathcal{V}^\pm(M) = \hbox{Int}
273: (\overline{\mathcal{V}^\pm(M)})$, so that $\mathcal{V}^\pm(M)$ contains a
274: neighborhood in $X$ of a point $p\in M$ whenever it contains the two
275: (local) one-sided neighborhoods at $p$. Such one-sided neighborhoods
276: are usually constructed by gluing analytic discs to $M$. Then the side
277: where the discs are lying may of course depend on the point $p\in M$
278: and it may vary effectively if $M$ is pseudoconvex somewhere and
279: pseudoconcave elsewhere. Without convexity assumption, we have\,:
280:
281: \begin{theorem}
282: {\rm (\cite{Tr}, \cite{M}, \cite{J})}
283: Let $M$ be an oriented globally minimal
284: $\mathcal{C}^\infty$-smooth embedded hypersurface in a
285: complex manifold $X$. Then there exists a one-sided neighborhood
286: $\mathcal{V}^\pm(M)$, constructed by gluing small
287: analytic discs to $M$, to which all continuous CR functions on $M$ can be
288: holomorphically extended.
289: \end{theorem}
290:
291: \begin{proof}[Proof of Theorem 1.1.]
292: Recall that {\bf (1)} implies {\bf (2)} trivially. We prove {\bf
293: (1)}. At first, we show that we can assume that $M$ and $f$ are
294: $\mathcal{C}^\infty$. So, let $M$ be a compact connected
295: $\mathcal{C}^2$ real hypersurface in $P_2(\mathbb{C})$ dividing
296: $P_2(\mathbb{C})$ into two connected open components $U^+$ and $U^-$
297: and let $f$ be a continuous CR function on $M$. By the theorem above,
298: $f$ extends holomorphically to a one-sided neighborhood
299: $\mathcal{V}^\pm(M)$ of $M$. Let $f$ still denote the resulting
300: extension. Let us smoothly deform $M$ into $\mathcal{V}^\pm(M)$. We
301: get a $\mathcal{C}^\infty$ hypersurface $M'\subset\subset
302: \mathcal{V}^\pm(M)$ arbitrarily close to $M$ and the trace
303: $f':=f\vert_{M'}$ of $f$ on $M'$ defines a $\mathcal{C}^\infty$ CR
304: function (even better, we could assume that $M'$ and $f'$ are real
305: analytic). Notice that as global minimality is a stable property
306: under sufficiently small perturbations, the deformed hypersurface $M'$
307: can be assumed to be globally minimal also. Again, $M'$ divides
308: $P_2(\mathbb{C})$ into two open parts ${U'}^-$ and ${U'}^+$. Suppose
309: that we can show that there exists a side, say ${U'}^+$, to which
310: every $\mathcal{C}^\infty$ CR function $f'$ on $M'$ extends
311: holomorphically. By inspecting the relative geometry of the three open
312: sets $U^+$, $\mathcal{V}^\pm(M)$ and ${U'}^+$, and using the fact that
313: $f$ is holomorphic in the whole of $\mathcal{V}^\pm(M)$, we deduce
314: that every continuous CR function $f$ on $M$ extends holomorphically
315: to $U^+$. In summary, it suffices for us to prove Theorem~1.1 in the
316: case where both $M$ and $f$ are $\mathcal{C}^\infty$-smooth.
317:
318: Thus, let $M$ be a compact connected $\mathcal{C}^\infty$ real
319: hypersurface of $P_2(\mathbb{C})$ which divides it into two open parts
320: $U^-$ and $U^+$. Let $f$ be a $\mathcal{C}^\infty$ CR function on $M$
321: and let $f^+\in \mathcal{O}(U^+)\cap
322: \mathcal{C}^\infty(\overline{U^+})$, $f^-\in \mathcal{O}(U^-)\cap
323: \mathcal{C}^\infty(\overline{U^-})$ be given by Lemma~2.1 so that
324: $f=f^+-f^-$ on $M$. By virtue of Lemma~3.1, it suffices to show that
325: at least one of the two functions $f^+$ or $f^-$ is constant. We then
326: proceed by contradiction.
327:
328: Assume that $f^+$ {\it and} $f^-$ are both nonconstant. By global
329: minimality, $f$, $f^+$ and $f^-$ can be holomorphically extended to a
330: one-sided neighborhood $\mathcal{V}^\pm(M)$ of $M$ constructed by
331: gluing discs. This open set is in fact a one-sided strip open set,
332: although it is not, in general, a tubular neighborhood of
333: $M$. However, we can easily deform $M$ into $\mathcal{V}^\pm(M)$ in a
334: $\mathcal{C}^\infty$ fashion, getting a $\mathcal{C}^\infty$
335: hypersurface $M_0\subset\subset \mathcal{V}^\pm(M)$. Furthermore, we
336: can include $M_0$ in a one-parameter family of manifolds $M_t$,
337: $|t|\leq \varepsilon$, satisfying
338: \begin{enumerate}
339: \item[{\bf 1.}] $M_t\cap M_{t'} = \emptyset$ for all $t\neq t'$,
340: \item[{\bf 2.}] $P_2(\mathbb{C})\backslash M_t = U_t^+\cup U_t^-$,
341: \item[{\bf 3.}] For $t<t'$, we have $M_t\subset\subset U_{t'}^-$ and
342: $U_t^- \subset\subset U_{t'}^-$. Also, $M_{t'}\subset\subset U_{t}^+$
343: and $U_{t'}^+ \subset\subset U_t^+$,
344: \item[{\bf 4.}] \ The strip $S:= \bigcup_{|t|< \varepsilon} M_t$ is an
345: open neighborhood of $M_0$ which is contained in $\mathcal{V}^\pm(M)$.
346: \end{enumerate}
347: The schematic picture of the geometric situation is as follows:
348:
349:
350: \begin{center}
351: \input one-sided.pstex_t
352: \end{center}
353:
354: \noindent
355: Notice that $M_0\subset\subset U_\varepsilon^-$. We shall apply the
356: following theorem due to Takeuchi over the domain $U_\varepsilon^-$.
357:
358: \begin{theorem}
359: {\rm (\cite{T})} Let $U\subset P_2(\mathbb{C})$ be a domain. Then
360: either
361: \begin{enumerate}
362: \item[(i)] Holomorphic functions on $U$ are constant, \text{\rm or}
363: \item[(ii)] Holomorphic functions on $U$ separate points.
364: \end{enumerate}
365: \end{theorem}
366:
367: \begin{lemma}
368: For every $\delta$ with $0 <\delta < \varepsilon$, the holomorphic
369: functions from $\mathcal{O}(U_\delta^-)$ separate points and give a
370: local coordinate system in a neighborhood of each point of
371: $U_{\delta}^-$.
372: \end{lemma}
373:
374: \begin{proof}
375: According to Theorem~3.3 and since there exists by assumption the
376: nonconstant function $f^-|_{U_\varepsilon^-}\in
377: \mathcal{O}(U_\varepsilon^-)$, then $\mathcal{O}(U_\varepsilon^-)$
378: separate points. We can therefore find at least two functions $h_j\in
379: \mathcal{O}(U_\varepsilon^-)$, $j=1,2$, such that the zero locus of
380: the Jacobian of the mapping $h:=(h_1,h_2)$ is a proper, closed,
381: complex analytic subvariety $\Sigma$ of $U_\varepsilon^-$ of complex
382: dimension $\leq 1$. Of course, $h$ gives a local coordinate system at
383: each point of $U_\varepsilon^-$ not in $\Sigma$. Fix $\delta$ with $0<
384: \delta < \varepsilon$. Let now $p\in U_\delta^-\cap \Sigma$ be
385: arbitrary. Then after composing $h$ with an automorphism $A$ of
386: $P_2(\mathbb{C})$ arbitrarily close to the identity, we obtain that
387: $h\circ A$ gives a local coordinate system at $p$. Such an
388: automorphism $A$ moves $U_\varepsilon^-$ a little bit. As $A$ can be
389: chosen to be arbitrarily close to the identity, we can insure that the
390: domain of definition of $h\circ A$ still contains $U_\delta^-$. This
391: completes the proof.
392: \end{proof}
393:
394: Let now $\eta$ with $0< \eta < \delta < \varepsilon$. The above lemma
395: implies that the manifold with boundary $M_\eta\cup
396: U_\eta^-\subset\subset U_\delta^-$ is embeddable into some complex
397: euclidean space $\mathbb{C}^N$, $N\in \mathbb{N}$, through an
398: embedding $\Phi: M_\eta\cup U_\eta^-\to \mathbb{C}^N$ whose
399: components are holomorphic functions defined in $U_{\delta}^-$. The
400: result of the embedding is a complex {\it manifold} \,
401: $\Phi(U_\eta^-)$ with boundary the three-dimensional maximally complex
402: CR manifold $\Phi(M_\eta)$. Notice that $\Phi$ also embeds into
403: $\mathbb{C}^N$ the closed strip $S':=\bigcup_{|t|\leq \eta}M_t$ which
404: provides a tubular neighborhood of $M_0$. By construction, for all
405: $\vert t\vert \leq \eta$, the $\Phi(M_t)$ are oriented maximally
406: complex three-dimensional CR manifolds in $\mathbb{C}^N$. The
407: orientation on $\Phi(M_t)$ is simply the push-forward of the
408: orientation on $M_t$. Further, $\Phi(M_t)$ bounds the complex manifold
409: $\Phi(U_t^-)$.
410:
411: Finally, the holomorphic function $f^+$ induces a nonconstant
412: holomorphic function on the strip $\Phi(S')$, say $g^+ := f^+ \circ
413: \Phi^{-1}|_{\Phi(S')}$. Applying a generalization of Bochner's theorem
414: ({\it see} \cite{H}), we deduce that the CR function
415: $g^+\vert_{\Phi(M_t)}$ extends holomorphically to the complex
416: submanifold $\Phi(U_t^-)$ for all $\vert t \vert \leq \eta$ (the
417: smoothness of $\Phi(U_t^-)$ is needed, otherwise we could only say
418: that the {\it graph} of $g^+\vert_{\Phi(M_t)}$ extends as a complex
419: analytic set by Harvey-Lawson's theorem). Thanks to the maximum
420: principle, we may now derive the desired contradiction easily.
421:
422: Recall that by assumption, $f^+$ is nonconstant. Therefore, by the
423: maximum principle applied to $f^+$ in the projective space over $U^+$,
424: we have for all $t$ with $-\eta \leq t<0$ the strict inequality
425: $$
426: \sup_{z\in M_t} |f^+(z)| > \sup_{z\in M_0} |f^+(z)|.
427: $$
428: because $M_0\subset\subset U_t^+$ and $f^+$ is nonconstant. But on the
429: other hand, by the maximum principle applied in the complex euclidean
430: space $\mathbb{C}^N$ to the nonconstant holomorphic function
431: $g^+|_{\Phi(S')}$ which extends holomorphically to $\Phi(U_\eta^-)$,
432: we have for all $t$ with $-\eta \leq t<0$ the reverse strict
433: inequality
434: $$
435: \sup_{z\in M_t} |f^+(z)|= \sup_{z\in \Phi(M_t)} |g^+(z)| < \sup_{z\in
436: \Phi(M_0)} |g^+(z)|= \sup_{z\in M_0} |f^+(z)|.
437: $$
438: This gives the desired contradiction, which completes the proof of
439: Theorem 1.1.
440: \end{proof}
441:
442: \bibliographystyle{amsplain}
443: \begin{thebibliography}{10}
444:
445: \bibitem{C} D.~Cerveau,
446: \textit{Minimaux des feuilletages alg\'ebriques
447: de $P_n(\mathbb{C})$.}
448: Ann.~Inst.~Fourier (Grenoble) \textbf{43} (1993), 1535--1543.
449:
450: \bibitem{DH} P.~Dolbeault and G.M.~Henkin,
451: \textit{Cha\^{\i}nes holomorphes de bord
452: donn\'e dans $P_n(\mathbb{C})$.}
453: Bull.~Soc. Math.~France \textbf{125} (1997), 383--446.
454:
455: \bibitem{E} L.~Ehrenpreis,
456: \textit{A new proof and an extension of Hartogs' theorem.}
457: Bull.~Amer.~Math.~Soc. \textbf{67} (1961), 507--509.
458:
459: \bibitem{F} B.~Fabre,
460: \textit{Sur l'intersection d'une surface de Riemann
461: avec des hypersurfaces alg\'e\-briques.}
462: C.R.~Acad.~Sci.~Paris S\'er. I Math\'ematiques
463: \textbf{322} (1996), 371--376.
464:
465: \bibitem{Gh} \'E.~Ghys,
466: \textit{Laminations par des surfaces de Riemann.}
467: Dynamique et g\'eom\'etrie complexe (Lyon, 1997), ix, xi, 49--95,
468: Panor. Synth\`eses, 8, Soc. Math. France, Paris, 1999.
469:
470: \bibitem{GH} P.~Griffiths and J.~Harris, \textit{Principles of Algebraic
471: Geometry.} Wiley Classics Library, John Wiley and Sons, 1994.
472:
473: \bibitem{H} R.~Harvey,
474: \textit{Holomorphic chains and their boundaries.}
475: Proceedings of Symposia in
476: Pure Mathematics, \textbf{XXX} (1977), vol1,
477: AMS, Providence, RI, 309--382.
478:
479: \bibitem{HaLa} R.~Harvey and B.~Lawson,
480: \textit{On boundaries of complex analytic varieties.} Ann. of Math., I:
481: \textbf{102} (1975), 233 - 290; II: \textbf{106} (1977), 213--238.
482:
483: \bibitem{HeLe} G.M.~Henkin and J.~Leiterer,
484: \textit{Theory of functions on complex manifolds.}
485: Monographs in Mathematics, \textbf{79}, Birkh\"auser,
486: Basel-Boston, Mass., 1984, 226~pp.
487:
488: \bibitem{I} S.~Ivashkovich,
489: \textit{The Hartogs-type extension theorem
490: for meromorphic maps into compact K\"ahler manifolds.}
491: Invent.~Math. \textbf{109} (1992), 47--54.
492:
493: \bibitem{J} B.~J\"oricke,
494: {\it Some remarks concerning holomorphically convex hulls and envelopes of
495: holomorphy.} Math.~Z. \textbf{218} (1995), 143--157.
496:
497: \bibitem{KR} J.J.~Kohn and H.~Rossi,
498: \textit{On the extension of holomorphic functions
499: from the boundary of a complex manifold.}
500: Ann.~of Math. \textbf{81} (1965), 451--472.
501:
502: \bibitem{La} C.~Laurent-Thi\'ebaut,
503: \textit{Ph\'enom\`ene de Hartogs-Bochner
504: dans les vari\'et\'es CR.}
505: Topics in Complex Analysis,
506: Banach Center Publications, Warszawa {\bf 31}
507: (1995), 233--247.
508:
509: \bibitem{Li} A.~Lins Neto,
510: \textit{A note on projective Levi flats
511: and minimal sets of algebraic foliations.}
512: Ann. Inst. Fourier (Grenoble) \textbf{49}
513: (1999), 1369--1385.
514:
515: \bibitem{M} J.~Merker,
516: \textit{Global minimality of generic manifolds
517: and holomorphic extendibility of CR functions.}
518: Int. Math. Res. Not. (1994), no1, 329--343.
519:
520: \bibitem{S1} F.~Sarkis,
521: \textit{CR meromorphic extension and the non
522: embedding of the Andreotti-Rossi CR structure in the projective space.}
523: Int. J. Math. \textbf{10} (1999), 897--915.
524:
525: \bibitem{S2} F.~Sarkis,
526: \textit{Probl\`eme de Plateau complexe dans les vari\'et\'es kahl\'eriennes.}
527: Preprint, 1999.
528:
529: \bibitem{Si} Y.T.~Siu,
530: \textit{Nonexistence of smooth Levi-flat hypersurfaces in
531: complex projective spaces of dimension $\geq 3$.}
532: Ann. Math. (2) \textbf{151} (2000), no.3, 1217--1243.
533:
534: \bibitem{T} A.~Takeuchi,
535: \textit{Domaines psudoconvexes infinis et
536: la m\'etrique riemannienne dans un espace projectif.}
537: J.~Math.~Soc.~Japan \textbf{16} (1964) 159--181.
538:
539: \bibitem{Tr} J.-M.~Tr\'epreau,
540: \textit{Sur la propagation des singularit\'es
541: dans les vari\'et\'es CR.}
542: Bull. Soc. Math. France
543: \textbf{118} (1990), 403--450.
544:
545: \end{thebibliography}
546:
547: \end{document}
548: