ON SOME PROBLEMS IN SEVERAL

COMPLEX VARIABLES AND CR GEOMETRY

Xiaojun Huang1

0. Introduction. Geometric analysis in several complex variables and Cauchy Riemann

geometry has been a quite active subject in pure and applied mathematics for decades. The
classical work of Poincaré, Hartogs, Levi, E. Cartan, etc at the beginning of the century
laid down part of the frame work of such a study. Since then, more and more attention
has be paid to this field, as people discover its deep relationship with many fundamental
problems in classical analysis and geometry, and its importance for partial differential
equations, algebraic and symplectic geometry, topology, mathematical physics. In this
note, we would like to present some results and problems along the lines of this research.
Limited to the page numbers, we will focus our discussion on some of the problems, which
the author’s recent work has touched upon. We will also have to leave out many important
references, which can be found, say, in the survey paper of Forstneric [Fr1], the author’s
papers [Hu1] [Hu2] [Hu3] [Hu6], and the recent reserach monographs of Baouendi-Ebenfelt-
Rothschild [BER2], D’Angelo [DA1] and Abate [Ab].

1.A Rigidity problem of holomorphic maps between balls. In an important devel-

opment of several complex variables, Poincaré discovered in 1907 that any biholomorphic
map between two pieces of the boundary of the unit ball B2 in C2 is the restrict of a certain
automorphism of B2 . This phenomenon fails obviously in the setting of one complex vari-
able and reveals for the first time a strong super-rigidity property of holomorphic mappings
in several complex variables. Poincaré’s result was later extended to higher dimensions by
Tanaka and Chern-Moser. In the 19700s, Alexander [Alx1-2] and Webster [We1] provided
two more beautiful rigidity theorems for mappings between balls and algebraic domains
in complex spaces of the same dimension. Their work has motivated many of the recent
exploitations of rigidity phenomena of holomorphic objects. Here, we will first constraint
ourselves to problems for mappings between balls in complex spaces of different dimen-
sions. (For more recent development on mappings between algebraic sets, we refer the
reader to [BER2] and the references therein.)

1991 Mathematics Subject Classification. 32, 53.

Key words and phrases. Rigidity problems, elliptic complex tangents, Levi flat submanifolds, reflection
principles, Kobayashi extremal mappings, iterates of holomoprhic mappings.
1 Supported in part by NSF-9970439 and an NSF postdoctoral fellowship

Typeset by AMS-TEX
1
2 XIAOJUN HUANG

In 1978, Webster in [We3] first took up the problem of considering proper holomorphic
mappings from Bn into Bn+1 with n ≥ 3 by using the Cartan-Chern-Moser CR projective
bundle theory. He proved that any such a map must be linear whenever it is in C 3 (Bn ).
Here, Bn denotes the unit ball in Cn centered at the origin. (Also, we recall that two
maps f, g : Bn → BN are called equivalent if there are elements σ ∈ Aut(Bn ) and
τ ∈ Aut(BN ) such that f = τ ◦ g ◦ σ. In particular, we call f linear or a totally geodesic
embedding if f is equivalent to the standard ‘big circle’ embedding sending (z1 , · · · , zn )
into (z1 , · · · , zn , 0, · · · , 0).)
In [Fa1], Faran classified proper maps from B2 into B3 , which are in the smooth class
C 3 (Bn ). Different from the linearity phenomenon of Webster, he surprisingly showed
that up to automorphsims, there are four different maps. These results were proved for
mappings which are only C 2 -regular up to the boundary in 1983 by Cima-Suffridge [CS].
In the same paper, they first conjectured that any proper holomorphic map from Bn
into BN with N ≤ 2n − 2 must be linear once it is in C 2(Bn ). Notice that there are
non-linear proper rational maps from Bn into B2n−1 . Cima-Suffridge’s conjecture was
verified in [Fa2] when the maps are holomorphic up to the boundary, by reducing it to
the problem concerning mappings between projective spaces. Forstneric [Fo2] then in his
work carried out a systematic study on the reflection principle for mappings between balls
in different complex spaces [Fr2]. His work together with Faran’s result gives the super-
rigidity property for mappings from Bn into BN with N ≤ 2n − 2, when the map is
C N −n+1-smooth up to the boundary. In a different direction, due to the discovery of inner
functions over the ball, it is known that there are many proper holomorphic mappings
from Bn into Bn+1 which are continuous up to the boundary but not rational. (See [Fr1],
[Ste], [Theorem 2, Dr]). This opens up a very interesting but also a difficult subject to
clarify the minimal boundary regularity required for mappings between balls to guarantee
the rigidity phenomenon. It particular, it has been wondered for years if there is a fixed
number t independent of the codimension such that the above described rigidity holds for
mappings whose boundary regularity is C t up to the boundary. In [Hu1], we introduced a
new approach for such a study and were able to obtain the following:
Theorem 1.1. Let M1 and M2 be two open connected pieces of the boundaries of the unit
balls Bn and BN , respectively. Let F be a twice continuously differentiable CR mapping
from M1 and M2 . If N ≤ 2n − 2, then F extends to a linear map from Bn into BN , unless
it is constant.
Here, we mention that a bounded function defined over a real analytic hypersurface
M is a CR function when it is annihilated by CR vector fields along M. In particular,
when M does not contain any germs of complex hypersurface, then g is CR if and only if
g extends holomorphically to a certain side of M by a result of Baouendi-Treves [BT]. A
mapping from M is called a CR map if all of its components are CR functions.
Ideas and results developed in [Hu1] have some other applications. For instance, the
following comes as an immediate Corollary of the work in [Hu1]:
Corollary 1.2. Let M1 and M2 be two open connected pieces of the boundaries of the unit
balls Bn and BN , respectively. Let F be a twice continuously differentiable CR mapping
 PROBLEMS IN SCV AND CR GEOMETRY 3

from M1 and M2 . If N = 2n − 1, then F must be rational.

The proof of Corollary 2.1 can be seen as follows: As in [Hu1], we may assume without
loss of generality that M1 and M2 are open pieces of the Heisenberg hypersurfaces con-
taining the origin, and F = (f, φ, g) satisfies the normalization condition as in [(3.1), Hu1].
By [Lemma 5.3, Hu1], we have that f = z + iw 2 a
(1)
(z) + owt (3) and g = w + owt (4) with <
Pn−1 (2) ∗∗
a (z), z > |z| = j=1 |φj (z)| . For each (z0 , w0 ) ∈ M1 , define F(z
(1) 2 2
0 ,w0 )
as in [Hu1]. (See
∗∗
notation following [Lemma 4.1, Hu1]. Also, we notice that Fp0 only depends on p0 and the
∗∗
choice of (Cl )p0 ). Then we claim that if F 6≡ (z, 0, w), after replacing F by a certain F(z 0 ,w0 )
(2)
with (z0 , w0 ) ≈ 0 (if necessary), it must hold that {φj }n−1 j=1 is a linearly independent family
(2) P (2)
over C. Indeed, suppose not. We can then assume that φn−1 ≡ n−2 j=1 µj φj for some com-
P
plex numbers µj , without loss of generality. Notice that n−1 (2) 2
j=1 |φj | can now be written as
Pn−2 (2)
j=1 χj (z)φj (z) for certain holomorphic polynomials χj (z). Applying [Lemma 3.2, Hu1]
Pn−2 (2) (2)
to the equation < z, a(1) (z) > |z|2 = j=1 χj (z)φj (z), it follows that a(1) (z), φj (z) ≡ 0
for each j. Since this phenomenon also holds for Fp∗∗ 0
for each p0 and (Cl )p0 , it follows
from [Proposition 4.2, Lemma 5.3, Hu1] that F ≡ (z, 0, w). This is a contradiction. Next,
(2) P
write φj = 1≤k≤l≤n−1 ajkl zk zl . Write ξkl = (a1kl , · · · , an−1 kl ). By the above, the family
of vectors {ξkl }kl has maximum rank n − 1. Hence, we can find {kj } and {lj } such that
det(ajki li )1≤i,j≤n−1 6= 0. Together with the normalization condition, we thus conclude that
the following is also linearly independent: {L1 fe, · · · , Ln−1 fe, Lk1 Ll1 fe, · · · , Lkn−1 Lln−1 fe}|0 .
√
(Here Lj = 2 −1zj ∂w ∂
+ ∂z∂ j ). Now, applying [Theorem 1, Fa3] and a result of Forstneric
[Fo2], we get the proof of Corollary 2.1.
The main idea introduced in [Hu1] depends on the following two steps (a): Derive a
jet relation for the normalized map (b): Push, by using the huge automorphisms, the
normalized equation into a partial differential equation, from which one tries to dig out
the possible rigidity property of the maps. This idea should be also useful to attack the
following two long standing open questions:
Problem 1.3. Let M1 ⊂ Cn and M2 ⊂ CN be open pieces of spheres and let f be a CR
mapping from M1 into M2 (1 < n < N). Can one find a fixed constant t, independent of
N − n, such that whenever f is C t over M1 , f is then rational? Furthermore, can one
take t = 2 or even 1?
Problem 1.4. Let M1 , M2 and f be as above. Suppose that f is rational. What is the
possible maximum degree of f? When n = 2, is then the degree of f bounded by 2N − 3?
Problem 1 was solved by Forstneric [Fr2] when f is C N −n+1-regular, by a reflection
principle argument. He also obtained a rough bound estimate for the degree of F in terms
of n and N. The key point of Question 1 here is whether one can get a number t which
is independent of the codimension. The second part of Problem 2 was first conjectured by
D’Angelo [Da1], who also verified it for monomial maps when N = 4 [Da1-4]. If solved
affirmatively, results in Problem 2 should be proven to be very useful for the classification
4 XIAOJUN HUANG

problem of mappings between balls, which is certainly one of the most basic problems in
complex analysis (see [Fr1] [Da1]).
1.B Rigidity of holomorphic mappings with group actions. This problem has been
motivated by the well-known super-rigidity problem in differential geometry and ergodic
theory: Let G1 and G2 be two non-compact semi-simple Lie groups with dim(G2 ) ≥
dim(G1 ). Suppose Γj ⊂ Gj are lattices (with certain density or co-compactness properties)
and there is a injective homomorphism φ : Γ1 → Γ2 . Can one then extend φ to a global
homomorphism from G1 into G2 ? This question has been quite well understood when the
real ranks of the groups are at least 2 by the work of Mostow, Margulis and Mok-Siu-
Yeung (see [MSY], for example). In the rank one case, there are essentially four cases to
be considered: The automorphism groups of real, complex and quaternionic hyperbolic
spaces, together with that of the hyperbolic Cayley plane. By the work of many people
(see [Cor] for related references) including Mostow, Gromov-Piastetski-Shaprio, Johnson-
Milnor, Corlette, the important remaining open case is when G1 = Aut(Bn ) and G2 =
Aut(BN ) with 1 < n < N, Γ1 co-compact in G1 and φ(Γ1 ) convex-cocompact. (Notice
that when n = 1, certain kind of counter-examples have been constructed by Mostow).
This problem, after applying the harmonic mapping theory, is reduced to the following
(see [Yue] or [Cor]):
Problem 1.5. Let Bn ⊂ Cn and BN ⊂ CN be the unit balls, and let f be a proper
holomorphic embedding from Bn into BN (n, N > 1). Suppose that there is a discrete
subgroup Γ ⊂ Aut(BN ) such that Γ fixes M = f(Bn ) and acts co-compactly over M. Is f
then a linear embedding?
The result of Gromov, etc, says that the supper-rigidity fails in case the groups come
from the isometry groups of real hyperbolic spaces; and the result of Corlette states the
supper-rigidity indeed holds for the isometry groups of the hyperbolic balls in the quater-
nionic space Qn with n ≥ 2 or hyperbolic Cayley plane (see [Cor] for a more precise and
detailed account on this matter). Hence, it is very interesting to find out what happens
for the automorphism groups Aut(Bn ) of balls in complex spaces with n > 1? Does the
quaternionic space Qn with n ≥ 2 enjoy more rigidity than Cn with n ≥ 2 in this regard?
A solution to Problem 3 is desirable, not just only from the point view of complex analysis,
but also from the ergodic theory, dynamics and Lie group theory point of view.
We mention that this question has been answered in the affirmative in the work of Cao-
Mok [CaoM] in the small codimensional case (N ≤ 2n − 1) by using some pure Kähler
Geometry arguments. Notice that the map f in problem 1.3 is always very nice along the
orbit. Hence, it may be possible to consider some kind of the Moser formal theory along
each orbit and then derive a certain kind of difference equation as in [Hu1] so that the
argument we developed in [Hu1] can be made into prospective. To make such a process
workable, one needs a very good mastering of the dynamic property of hyperbolic elements
of Aut(Bn ).
1.C Rigidity of algebraic domains and their mappings. This study was initiated in
a celebrated paper of S. Webster [We1], who first proved that any biholomorphic map be-
tween two algebraic Levi non-degenerate hypersurfaces must be an algebraic map. Recent
 PROBLEMS IN SCV AND CR GEOMETRY 5

deep developments of his theorem can be found in the work of [BR2], [BER], [Zat], etc. In
[HJ], we proved a sup-rigidity theorem concerning algebraic spheric domains, which claims
all bounded algebraic spherical domains in Cn with n > 1 must be biholomorphically
equivalent to the ball.
The motivation for our work [HJ] with Ji can be described as follows: After the fun-
damental work of Chern-Moser [CM] and Fefferman, people started to attack the complex
structures of domains by using the boundary CR geometry. A very natural question is
then to ask if one can characterize the ball by its boundary CR geometry. More pre-
cisely, is any bounded spherical domain biholomorphic to the ball? Indeed, the work of
Krushilin-Pinchuk-Vitsushili in the 1970’s shed the light toward the positive side of such
a probem. Suprisingly, in 1980, Burns-Schnider [BS], provided a counter-example to this.
They showed that there are many bounded real analytic spherical domains in Cn , which
are topolicaly equivalent to the torus—hence, can not be the ball. A naive observation
is as follows: If the boundary is spherical, it should be flat in a certain sense. Hence, to
make it compact, one has to use a periodic function to define it. (The defining functions
of Burns-Schnider’s examples are trignometric functions). The domain then can not be
algebraic. It is indeed such a (non-rigorous) intuitive which led us to the work in [HJ].
Theorem 2 of [HJ] also tells the set of all bounded strongly pseudoconvex algebraic
domains indeed forms a strictly smaller subset of all real analytic domains; and indicates
that there are many rigidities for algebraic domains which analytic domains do not share.
It seems to us our understanding to the specific holomorphic properties of algebraic do-
mains is still at a very early stage, comparing to the role of complex algebraic geometry in
complex analytic geometry (see [We4]). An interesting question arising naturally is first to
develop a systematic invariant theory to distinguish strongly pseudoconvex algebraic hy-
persurfaces from strongly pseudoconvex real analytic hypersurfaces. Such an investigation
might motivate out new interesting problems, which can lead to a further understanding of
rigidities for algebraic domains. Perhaps, a good start is to find a motivated proof whether
M1 = {(z, u + iv) ∈ C2 : v = e|z| − 1} is holomorphically equivalent to an algebraic
2

hypersurface. Peter Ebenfelt recently communicated to us that by the result in [BHR1]

(see Theorem 3.5 in the next section), his example in [E] on a Levi non-flat real analytic
hypersurface M in C2 can not be biholomorphic to any algebraic hypersurface, due to the
presence of non-trivial holomorphic curves and non-degenerate smooth but not analytic
CR mappings from his M to other Levi non-flat algebraic hypersurafces (see [E]).
When one deals with strongly pseudoconvex hypersurfaces, a totally different argument
is needed. An idea to carry out this study might be to use the Cartan-Chern-Moser
connections and curvatures, which and whose functional relations are algebraic for algebraic
strongly pseudoconvex hypersurfaces. It is very interetsing to see what such an implication
would really be.

2.A Smooth and analytic structures of a Bishop submanifold in Cn and local

complex Plateau problems. Let M be a real k-manifold in Cn with p ∈ M. One of the
fundamental problems in complex analysis is to understand the holomorphic invariants of
(1,0) √
M near p. When p is a totally real point, namely, Tp = CTpM ∩ −1CTpM = {0},
6 XIAOJUN HUANG

M near p basically has the character of the standard Rk in Cn . When the complex
(1,0)
tangent space Tp M 6= 0, the situation can be much more complicated. Thus far, only
sufficiently non-degenerate complex tangents have been studied. Recall that p is called a
point with an elliptic complex tangent if the Levi-form of M in a certain sense is positive
at p ([CM], [Bis]). In this subject, there are two cases which are of particular interest–
namely, k = 2n− 1 and k = n. When M is a hypersurface (k = 2n− 1), p is a point with an
elliptic complex tangent if and only if M is strongly pseudoconvex near p. In this case, the
geometry of M near p has been relatively quite well understood and has been successfully
used for the study of the equivalence problem as in the work of Cartan and Chern-Moser
[CM]. When k = n with p ∈ M an elliptic complex tangent, one has the least possible
complex structure near p, and away from a thin set, M is totally real. In this situation,
the existence of ‘good geometry’, which can be used for the equivalence problem, is still
not completely understood in some important situation.
Bishop [Bis] in 1965 first introduced the second order biholomorphic invariant, now
called the Bishop invariant for the case n = k, which corresponds to the Levi-eigenvalue in
the hypersurface case. He showed that M near such a p bounds many complex curves and
has a non-trivial local hull of holomorphy Mf. Bishop conjectured that the local hull Mf is
a Levi flat submanifold which is foliated by pairwise disjoint embedded complex analytic
disks attached to M. Also, he proposed the study of the fine structure of M f near p. In
the case of n = 2, Kenig-Webster later showed, in their paper [KW1], that M f is indeed a
2
smooth-up-to-the boundary Levi-flat hypersurface in C foliated by a family of analytic
disks. (See also the work done in [BG] [BK] for the global case.) When n > 2, Kenig-
Webster [KW2] showed that for each l, M bounds a Levi flat submanifold M fl , which is C l
up to M. But they left open whether all these M fl are the same and give the local hull of
holomorphy. In the work of Moser-Webster [MW], it was shown that M f is real analytic
across M in case M is a real analytic submanifold and when the Bishop invariant attached
at p does not vanish. But Moser-Webster’s argument does not apply in the case of the
vanishing Bishop invariant. (See also the work done in [Mos] where the case some special
case in the vanishing Bishop situation invariant was studied.) In the author’s joint paper
with Krantz [HK], in case M ⊂ C2 , we studied the real analytic structure of M f in the
vanishing Bishop case, settling an open question of Moser in [Mos]. Recently, In [Hu2],
with a more subtle analysis of the singular Bishop equation and formal theory argument,
we established the above mentioned results in any dimensions, and showed that the M fl ’s
constructed by Kenig-Webster in [KW2] are indeed the same. The effort in [Hu2] provides,
together with the work of the above mentioned mathematicians, a complete solution to an
old problem initiated in the work of Bishop [Bis]:
Theorem 2.1. Let M ⊂ Cn be a smooth submanifold of real dimension n. Let p ∈ M
be an elliptic complex tangent point. Then M bounds, near p, a unique Levi-flat CR
submanifold M f of real dimension n + 1. Moreover, M f is foliated by embedded complex
analytic disks and Mf extends smoothly across the boundary M. Also, M f gives the local
f
hull of holomorphy of M. When M is real analytic, M is real analytical across M and M
can be flattened (namely, M can be biholomorphically transformed into Rn−1 × C).
 PROBLEMS IN SCV AND CR GEOMETRY 7

Notice that in the case of dimension two, when p is elliptic, after a holomorphic change
of coordinates, we can always assume that p = 0 and M near p is defined by an equation
of the form: w = zz + λ(z 2 + z 2 ) + o(|z|2 ), λ ∈ [0, 1/2). Here λ is the Bishop invariant.
The nice property of the local hull of holomorphy M f provides some important information
for the holomorphic structure of M. However, a further understanding of the whole set of
invariants of surfaces of this type is of considerable interest and desirable because of the
following reasons: First, from the point view of complex analysis, they can be viewed as the
simplest higher codimensional analogy of strongly pseudoconvex hypersurfaces; secondly,
they have a rich complex structure at the elliptic complex tangent and have trivial complex
structure elsewhere, namely they can also be viewed as the simplest models where one sees
the CR singularity; thirdly, from the celebrated work of Moser-Webster [MW], one sees
here a tremendous interaction of complex analysis with the classical dynamics problems in
Celestial Mechanics [SM]– An understanding of such a problem may provide information
and motivation to some convergence problems in Mechanics.
In [MW], the normal form was established for the elliptic complex tangent with non-
vanishing Bishop invariant, and M is showed to posses only ‘two and a half’ biholomorphic
invariants near p. However, the situation with vanishing Bishop invariant is quite different
and M may support infinitely many biholomorphic invariants, as conjectured by Moser
and Webster [MW] [Mos]. As is known, one of the major difficulties to the study of the
normal form theory is to provide the convergence proof for certain formal power series. In
the hypersurface case, Chern-Moser [CM] circumvented this in their work by applying the
intrinsic CR geometry of the manifolds (namely, CR structure bundles, the Cartan-Chern-
Moser chains, etc). In the non-vanishing Bishop invariant case with k = n = 2, Moser-
Webster [MW] obtained this by ingeniously using the geometry of a pair of involutions
(intertwined by the natural conjugation operator) defined over the complexification of M,
and then reducing the problem to the normalization problems of these involutions. The
convergence proofs can then be done by a very clever use of the majorant methods as in the
work, dealing with the classical mechanics, of Birkhoff, Siegel, Moser [SM]. In the vanishing
Bishop invariant case, namely, λ = 0, the aforementioned Moser-Webster involutions do not
exist anymore and other geometry has to been employed for the equivalence problem. In
[HK] and [Hu2], we showed that M can be mapped into the Levi-flat analytic hypersurface
R×C and bounds a family of holomorphic disks depending real analytically on a parameter.
This may give some very useful information for the convergence proof of the normal-form
related formal series. For instance, in [Hu2] using this geometry, we were able to handle the
convergence proof of the normalized formal solution of an important functional equation
as in Theorem 2.2:

Theorem 2.2. Let M be an elliptic Bishop n-manifold in Cn near the origin. For any
real analytic function G(z, z) defined near 0, there is a holomorphic solution F to the
functional equation: Im(F (z)) = G(z, z), where z ∈ M.

We think this should also be one of the basic tools to answer the following fundamen-
tal question of Moser-Webster [MW] and Moser [Mos] in this subject of several complex
variables:
8 XIAOJUN HUANG

Problem 2.3. Let E ⊂ C2 be a real analytic submanifold of real dimension 2 defined

by w = |z|2 + o(|z|2 ). (That is, 0 ∈ E is an elliptic complex tangent point with Bishop
invariant λ = 0). What are the biholomorphic invariants of E at 0? Is E biholomorphic
to a real algebraic surface?

2.B Topology of pseudoconvex domains in C2 . We now mention some other related

global problems, which are related to our work decribed above. Probably, the simplest
non-trivial case of the above local consideration is the filling problem of a general 2-
submanifold M in C2 by holomorphic disks. Through the work of Bedford-Gaveau [BG],
Bedford-Klingenberg [BK], Gromov [Gr] and Eliashberg [Eli], etc, certain positive results
have been obtained in case M is a topological 2-sphere in generic position with a certain
convexity assumption. However, when M does not have these properties, the problem is far
from being understood. A natural question is to understand the geometric and topological
obstructions to the filling process. Another interesting problem in this field is to ask what
kind of Levi-flat hypersurface Mf a totally real 2-torus M in C2 can bound. We notice that
Alexander’s [Alx3] recent solution to the Bennequin problem reveals that M f is usually not
3
smooth. We also remark that, in case M is the graph of a torus in R , the problem is
equivalent to studying the regularity problem of the Dirichlet problem of a certain totally
degenerate non-linear elliptic differential equation [TS].
The combining effort of the two dimensional filling result of Bedford-Gaveau [BG] and
the regularity result of Kenig-Webster[KW1] gives the following Theorem (BGKW): Any
generic two-sphere S ⊂ C2 with only two elliptic complex tangents and contained in the
boundary of a certain strongly pseudoconvex domain bounds a unique smooth Levi-flat
holomorphically convex hypersurface M which is smooth up to S. This is a deep achieve-
ment in the subject of several complex variables, whose proof is quite non-trivial. Hence, it
will not be surprised to us if one finds some deep and important applications. For us, it is
desirable to see if they can be applied to work on various problems concerning the topology
of pseudoconvex domains. For instance, we believe that together with the Morse Theory
and a subtle use of Theorem (BGKW), one should be able to get a complete understanding
of the following problem (see [Fr3] for certain studies related to this problem):
Problem 2.4. Let D be a simply-connected smoothly bounded pseudoconvex domain in C2 .
What can we say about its second fundamental group? In particular, can D be homotopic to
the unit 2-sphere? What can we say about the topology of ∂D? Is it also simply connected?
Problems of this type are crucial for the understanding of many problems in both
Geometry and Topology. Let us describe two of them here: There is a famous theorem
in algebraic geometry, which says any polynomial embedding from C1 into C2 can be
straightened. However, it is a wide open question to answer whether any polynomial
embedding from C2 into C3 can be straightened. Closely related to such a problem (see
[Za]) is to understand the structure of a contractible affine algebraic surface M. It is known
that M is isomorphic to C2 if it has simply connected topology at infinity. Now, suppose
that M is not isomorphic to C2 . An important problem left open in algebraic geometry is
then to understand how far M can be holomorphically away from C2 (see problems asked
 PROBLEMS IN SCV AND CR GEOMETRY 9

in [Za]). (The existence of such an M can be found in [Ra]). Along these lines, it was
shown by Ramanujam and Zaidenberg that M can be exhausted by a sequence of strongly
pseudoconvex domains Ωj of M such that each Ωj is contractible but its boundary has a
huge perfect fundamental group. Hence, if Problem 5 can be solved affirmatively, then we
see that any bounded domain in M, when large enough, cannot be (biholomorphically)
embedded in C2 . Another motivation for Problem 5 is as follows: In topology, there is a
famous ball D called the Mazur ball in C2 , which is homeomorphic to the unit ball and
has a smooth boundary. However, π1 (∂D) is non-trivial. A natural question for complex
analysts may be to understand if a Mazur ball can be realized as a pseudoconvex domain
in C2 .
One idea to work on the last part of Problem 5 is to use the Morse theory and Theorem
(BGKW). Namely, one will try to deform any 2-sphere embedded in D, touching a certain
fixed boundary point p to ∂D, relative to p. Use the Morse theorem to understand the
character of the critical points of the Morse defining function and use Theorem (BGKW)
to pave a way for the the 2-sphere to get rid of the critical points with index 2. When
n > 2, this process goes through very easily. (See §5 of [HJ].)

3. Regularity of CR mappings between real analytic hypersurfaces in Cn . One

of the fundamental properties for holomorphic functions in one complex variable is the
Schwarz reflection principle, which states that any holomorphic function f defined over a
certain side of a real analytic curve γ and continuous up to γ must extend holomorphically
across γ once f(γ) is also contained in a real analytic curve. A natural consideration of
this phenomenon in higher dimensions seems to replace γ by real analytic hypersurfaces.
However, a moment thought indicates that the situation is not that simple anymore: Let
M1 = M2 = {(z, w) ∈ C2 : Im(w) = Re(w)|z|2 } and F = (e−1/w , 0). Then F (M1 ) ⊂
1/3

M2 , F extends holomrophically to a certain side of M1 and smooth up to M1 , but F does

not extend holomorphically across M1 . An easy investigation shows that the failure of
the reflection principle in this example is due to the existence of non-trivial holomorphic
curves inside M1 and the degeneracy of the map. Hence, this suggests one to formulate
the following problems. (Here we recall a real analytic hypersurface is of D’Angelo finite
type if it does not contain any non-trivial germs of holomorphic curves):
Problem 3.1. Let M1 and M2 be two real analytic hypersurfaces in Cn of finite D’Angelo
type. Let f be a continuous CR mapping from M1 into M2 . Is f then real analytic over
M1 ?
Problem 3.2. Let M1 and M2 be sufficiently non-degenerate hypersurfaces but of infinite
type. Let f be a sufficiently smooth CR map from M1 into M2 , which is also sufficiently
non-degenerate. Is then f real analytic?
Indeed, although a lot of attention has been paid by many mathematicians in the past
twenty-five years, whether the above holds or not still remains to be unsolved questions in
the subject.
The most important contribution along this direction is contained in the the celebrated
work of C. Fefferman, which, in particular, indicates that any CR homeomorphism between
10 XIAOJUN HUANG

strongly pseudoconvex real analytic hypersurfaces is a smooth CR diffeomorphism. The

work of Chern-Moser [CM] [Jac] and Lewy [Le], Pinchuk [Pi] can then be used to give the
analyticity for such a map. An extensive survey on this subject with a fairly complete list
of references, after the work of Fefferman and Chern-Moser, can be found in the survey
paper of Forstneric [Fr1]. Hence, we will only focus on the development after the survey
paper [Fr1].
Regarding Problem 3.1 and before [Hu3] and [BHR1], there had been many studies for
maps which either have some kind of Hopf lemma property or extend properly to a certain
side through the work of many people (See [Fr1] for a detailed survey and related refer-
ences). However, very little had been known about the regularity of general CR mappings
between non-strongly pseudoconvex hypersurfaces. Indeed, the following theorem proved
in [Hu3] was, at the time, new under the additional assumption that the map is smooth
or the hypersurfaces are pseudoconvex. (This result is due to Baouendi-Bell-Rothschild
[BBR] when f is smooth and does not vanish to infinite order at any point, and is due to
Diederich-Fornaess-Ye [DFY] when the map is a CR homoemorphism.)
Theorem 3.3. Let M1 and M2 be two (connected) real analytic hypersurfaces in C2 which
do not contain any non-trivial holomorphic curves inside. Let f be a non-constant con-
tinuous CR mapping from M1 into M2 . Then f is real analytic along M1 . Moreover, f
extends locally properly to a neighborhood of M1 .
For the general dimensions, we state here a result obtained in the joint work with Pan
[HP] and Baouendi, Rothschild [BHR2], which, together with the theorem of Baouendi-
Rothschild [BR1] and Diederich-Fornaess [DF1], gives the following:
Theorem 3.4. Let M1 and M2 be two real analytic hypersurfaces of finite D’Angelo type
in Cn (n > 1). Suppose that f is a smooth CR mapping from M1 into M2 . If f is real
analytic at most away from a certain compactly supported subset of M1 , then f is actually
real analytic over M1 . In particular, any proper holomorphic map between two bounded
real analytic domains in Cn admits a holomorphic extension across the boundary of the
source domain if it is C ∞-smooth up to the boundary of the source domain.
In regard to Problem 3.2, using the strengthened version of the Artin approximation
theorem, we were able to obtain in a joint work with Baouendi-Rothschild [BHR1] the
following, which essentially gives a fairly nice solution of Problem 3.2 in the algebraic
category:
Theorem 3.5. Let M1 and M2 be two connected algebraic hypersurfaces in Cn . Assume
that M1 is holomorphically non-degenerate. Then there is an integer k depending only on
the degrees of M1 and M2 such that any CR mapping f from M1 into M2 , which is C k -
smooth and whose Jacobian is not identically zero, must be real analytic over M1 .
Recall that a hypersurface M is called holomorphically non-degenerate if there is no
non-trivial holomorphic vector field tangent to M ([Sta], [BHR1]). In [HBR1], examples
have also been given to show that the above theorem fails if M1 and M2 are just assumed
to be real analytic, or if the map is not assumed to be sufficiently smooth. Also, the result
is no longer true if M1 is allowed to be holomorphically degenerate. Notice that one of
 PROBLEMS IN SCV AND CR GEOMETRY 11

the major features of Theorem 3.3 is that no one-sided holomorphic extension is apriori
known. (This seems to be the first reflection principle of this type.)
Concerning Problem 3.1, by Baouendi-Treves [BT], one knows that the map there ex-
tends holomorphically to a certain side D. However, it is usually very difficult to know
whether a certain type of the Hopf lemma still holds for the normal component of the
map or whether the map extends properly to D so that many results developed for the
global case can be applied. In fact, it is often the case that one has to get the regularity
before understanding the Hopf lemma property and the local properness. For example,
the following seems to be a non-trivial open question: Is any continuous CR mapping from
an open piece of the sphere in C3 near the origin into M2 = {(z1 , z2 , w = u + iv) ≈ 0 :
v = |z1 |2 − |z2 |2 + 100|z2 |4 } a constant map? (The difficulty lies in that we do not know
if such an f extends properly to a certain side.)
Theorem 3.3 gave a complete solution to Problem 3.1 in the two dimensional case. Since
it fails for hypersurfaces of infinity type and fails for non-continuous bounded CR maps, it
can be regarded as a natural extension of the classical Schwarz reflection principle in C2 .
We say a few words about the proof of Theorem 3.1 (more detailed account on the closely
related history and references can be found in [Hu3] and [Hu4]):
By [BT], the map in Theorem 3.1 is known to extend holomorphically to a certain side
D of M1 with f ∈ C(D ∪M1 )∩Hol(D). Hence, one needs to construct the extension of the
map to another side Dc . Since it is difficult to get the direct holomorphic extension of the
map to the Dc -side (unless the target hypersurface is strongly pseudoconvex), one wishes
to split the proof into two parts, as suggested in many papers (see [BJT], [DF1], [DFY], for
example): Part (a): Construct the multiple valued extension of the map f to the Dc -side;
and Part (b): Prove the single value property assuming the multiple-valued extension.
Part (b) follows from the Malgrange theorem as in the work of [BJT] [BBR] [DF1] when
the map is assumed to be smooth. Apriori to our work, there had been the work done
by Diederich-Fornaess [DF2], which gave a solution for (b) in the CR homeomorphism
case (or, globally, biholomorphic case.) Their approach, however, is difficult to extend to
proper and CR mappings. Our proof of Part (b) for proper and general CR mappings is
different from that in [DF2]. Concerning the step (a), at the time, there had been the work
done by Diederich-Fornaess-Ye [DFY] and an unpublished note distributed by Diederich-
Pinchuk (on Part (a) for biholomorphic maps, by using the above mentioned Part (b)
result in [DF2].) Also, it had been clear that once (b) was settled, the proof of Part
(a) for mappings which are assumed apriori to be proper from D could be achieved even
by slightly modifying the above mentioned work for Part (a) in the biholomorphic case.
However, the situation for the study of Part (a) for general CR mappings was different,
partially because the previously established global results for the pseudoconvex domains
by Baouendi-Bell-Rothschild, etc, (see [BC]) and the pseudoconcave preserving property of
proper mappings were no longer applicable. We succeeded in writing up a proof of Part (a)
for general CR mappings by using a continuity method, which was also used in the papers
of Webster [We2] and Diederich-Fornaess-Ye [DFY]. The major difficulty in the continuity
method, as usual, was to get the closeness. The key point which enabled us to get (a)
for general CR mappings was a way to use the the geometric size of the Segre varieties to
12 XIAOJUN HUANG

control the rate of blowing-up of the multiple-valued extension of f to the Dc -side, from
which the closeness could be concluded. (See Lemma 3.1 of [Hu3], for example). This idea
also works for the n-dimensional pseudoconvex case and was further exploited in [Hu4].
We now say a few words about the work in [Hu4], which is concerned with the problem
of Bell and Bell-Catlin [BC]:
Problem 3.6. Let M1 and M2 be two pseudoconvex real analytic hypersurfaces of finite
D’Angelo type in Cn . Let f be a continuous CR mapping from M1 into M2 . Is then f real
analytic over M1 .
Regarding Problem 3.6, Bell-Catlin [BC] showed that f is real analytic if for each
p ∈ M1 , Fp = f −1 (f(p)) is a compact subset of M1 . On the other hand, if one would have
already known that f is non-constant and real analytic, then Qq = Qp for each q ∈ Fp
and Fp would be discrete. (Here Qq is the Segre variety of M1 associated to q). Indeed,
the work in [Hu4] essentially tells that we need only to verify the Bell-Catlin condition on
Ep = Qp ∩ Fp . This, in particular, gives the following corollary: The answer to Problem
6 is affirmative if M1 has the following locally finite stratification M1 = ∪j M1j , where for
each q ∈ M1j , (Qq \ {q}) ∩ O(q) ⊂ ∪l<j M1l . It seems one way to get such a stratification is
to use Catlin’s multi-type function C(p). By an induction argument, it suffices to assume
that C(q) ≤ C(p) and f ∈ C ω (M1 \ Vp ) (Vp = {q : C(q) = C(p)}). Since, by Catlin [Ca], Vp
is contained in a CR manifold of holomorphic dimension 0, namely, Vp is a kind of strongly
pseudo-convex along the CR direction contained in Vp , this makes it very plausible to get
a stratification on Vp which has the above described property and thus answer completely
Problem 3.6.
The implication of the work in [Hu3] [BHR1] may also give some hints to work on
the following problem of Alinhac-Baouenid-Rothschild, which, if solved affirmatively, will
give immediately with the famous theorem of Baouendi-Rothschild [BR1] and Diederich-
Fornaess [DF1] the analyticity result for mappings in Problem 3.1 in an important case
where the maps are assumed to be smooth.
Problem 3.7. Let f be a smooth non-constant CR-mapping from M1 into M2 , two smooth
real hypersurfaces of finite D’Angelo-type in Cn (n > 1). Does the normal component of
f have non-vanishing normal derivative at each point of M1 ?
4. Extremal disks and applications to the study of holomorphic self-mappings. Let
D be a domain in Cn , and let f ∈ Hol(D, D) be a holomorphic self-mapping. An interesting
question is then to consider the asymptotic behavior of the sequence {f k } of iterates of f
and its interaction with the invariant objects associated to D (for example, the Kobayashi
metric and Kähler-Einstein (Cheng-Yau) metric of D, etc.) In [Hu6], we were able to ob-
tain an exact Denjoy-Wolff style theorem for contractible strongly pseudoconvex domains
in Cn . That is, we proved the following
Theorem 4.1. Let D ⊂ Cn be a smoothly bounded contractible strongly pseudoconvex
domain. If f is a holomorphic self-mapping from D to itself, then the sequence {f k }
converges uniformly on compacta to a boundary point if and only if f has no fixed point
inside D.
 PROBLEMS IN SCV AND CR GEOMETRY 13

Theorem 4.1 confirms affirmatively an open question in this subject (see [Ab]). We
should mention that certain special cases were previously studied by many authors. Here
we refer the reader to the book of Abate [Ab] and a paper of Ma [Ma] for a detailed list of
related references.
Different from the arguments introduced for such a problem in the previous work (see
[Ab] [Ma], for instance), our proof of the above theorem is based on a very detailed in-
vestigation of the localization phenomenon of Kobayashi extremal mappings near strongly
pseudoconvex points [Hu6], [Hu7], and the deformation theory of Lempert [Lem1], which
themselves rely strongly on the understanding of certain non-linear singular integral equa-
tions (called the Riemann-Hilbert equations). Also the methods and results developed in
[Hu6] [Hu7] can be used for some other problems related to the dynamical property of
holomorphic self-maps.
Along these lines, it is desirable to further investigate the boundary behavior of the
Kobayashi metric and the Kähler-Einstein (Cheng-Yau) metric of a bounded smooth pseu-
doconvex domain, as well as their interaction with the iterates of proper holomorphic self
mappings of a pseudoconvex domain of finite type. As is known, in this situation, one is led
to the study of the boundary invariants of pseudoconvex domains [Ca] and the boundary
behavior of complex Monge-Ampère equations [CY]. To be precise, we mention the the
following well-known open question:
Problem 4.2. (a). Is any proper holomorphic self-mapping of a bounded smooth pseu-
doconvex domain of finite D’Angelo type in Cn (n > 1) an automorphism? (b). Let D
be a bounded smooth pseudoconvex domain of finite D’Angelo type. What are the precise
boundary behaviors of the Kobayashi metric, Bergman metric, and Kähler-Einstein metric
of D? Is the Kobayashi metric complete?

References
[Ab] M. Abate, Iteration Theory of Holomorphic maps on Taut Manifolds, Mediterranean Press, Ren-
dre, Cosenza (1989).
[Alx 1] H. Alexander, Proper holomorphic maps in Cn , Ind. Univ. Math. Journal 26 (1977), 137–146.
[Alx2] H. Alexander, Proper holomorphic maps in Cn , Ind. Univ. Math. Journal 26 (1977), 137-146.
[Alx 3] H. Alexander, Gromov’s method and Bennequin’s problem, Invent. Math. 125 (1996), 135–148..
[ABR] S. Alinhac, S. Baouendi and L. Rothschild, Unique continuation and regularity of holomorphic
functions at the boundary, Duke Math. J. 61 (1990), 635–653.
[BBR] S. Baouendi, S. Bell and L. Rothschild, Mappings of three-dimensional CR manifolds and their
holomorphic extension, Duke Math. J. 56 (1988), 503–530.
[BER1] S. Baouendi, P. Ebenfelt and L. Rothschil, Algebraicity of holomorphic mappings between real
algebraic sets in Cn , Acta Math. 177 (1996), 225-273.
[BER2] S. Baouendi, P. Ebenfelt and L. Rothschild, Princeton Mathematical Series, vol. 47, Princeton,
New Jersey, 1999.
[BHR1] S. Baouendi, X. Huang and L. Rothschild, Regularity of CR mappings between algebraic hyper-
surfaces, Invent. Math. 125 (1996), 13–36.
[BHR2] S. Baouendi, X. Huang and L. Rothschild, Nonvanishing of the differential of holomorphic map-
pings at boundary points, Math. Res. Lett. 2 (1995), 737–750.
[BJT] S. Baouendi, H. Jacobowitz and F. Treves, On the analyticity of CR mappings, Ann. of Math.
122 (1985), 365-400.
14 XIAOJUN HUANG

[BR1] S. Baouendi and L. Rothschild, Germs of CR maps between real analytic hypersurfaces, Invent.
Math. 93 (1988), 481-500.
[BR2] M.S. Baouendi and L. P. Rothschild, Geometric properties of smooth and holomorphic mappings
between surfaces in complex spaces, J. Diff. Geom. 31 (1990), 473-499.
[BR3] S. Baouendi and L. Rothschild, Mappings of real algebraic hypersurfaces, Jour. of AMS 8 (1995),
997-1015.
[BR4] S. Baouendi and L. Rothschild, A general reflection principle in C2 , Journal of Funct. Analysis
99 (1991), 409-442.
[BT] S. Baouendi and F. Treves, About the holomorphic extension of CR functions on real hypersurfaces
in complex space, Duke Math. J. 51 (1985), 77–107.
[BG] E. Bedford and B. Gaveau, Envelopes of holomorphy of certain 2-spheres in C2 , Amer. J. Math.
105 (1983), 975-1009.
[BK] E. Bedford and W. Klingenberg, On the envelopes of holomorphy of a 2-sphere in C2 , Journal of
AMS 4 (1991), 623-655.
[Be] S. Bell, Analytic hypoellipticity of the ∂-Neumann problem and extendibility of holomorphic map-
pings, Acta Math. 147 (1981), 109-116.
[BC] S. Bell and D. Catlin, Regularity of CR mappings, Math. Z. (1988), 357-368.
[Bis] E. Bishop, Differentiable manifolds in complex Euclidean space, Duke Math. J. 32 (1965), 1-21.
[BKr] D. Burns and S. Krantz, On a new boundary rigidity theorem for holomorphic mappings, Jour of
Amer. Math. Soc. 7 (1994), 667-680.
[BS] D. Burns and S. Shnider, Spherical hypersurfaces in complex manifolds, Invent. Math. 33 (1976),
223-246.
[CaM] H. Cao and N. Mok, Holomorphic immersions between compact hyperbolic space forms, Invent.
Math. 100 (1990), 49–61.
[Ca] D. Catlin, Boundary invariant of pseudoconvex domains, Ann of Math 120 (1984), 529-586.
[CD] D. Catlin and J.P. D’Angelo, A stabilization theorem for Hermitian forms and applications to
holomorphic mappings, Math. Res. Lett. 3 (1996), 149–166.
[CS] J. Cima and T. J. Suffrige, A reflection principle with applications to proper holomorphic map-
pings, Math Ann. 265 (1983), 489-500.
[CKS] J. Cima, S. Krantz, and T. J. Suffrige, A reflection principle for proper holomorphic mappings of
strictly pseudoconvex domains and applications, Math. Z. 186 (1984), 1-8.
[CY] S. Y. Cheng and S. T. Yau, On the existence of a complex Kähler metric on non-compact complex
manifolds and the regularity of Fefferman’s equation, Comm. Pure Appl. Math. 33 (1980), 507–
544.
[CM] S. S. Chern and J. K. Moser, Real hypersurfaces in complex manifolds, Acta Math. 133 (1974),
219-271.
[CMS] B. Coupet, F. Meylan and A. Sukhov, Holomorphic maps of algebraic CR manifolds, Internat.
Math. Res. Notices (1999), 1-29.
[Cor] K. Corlette, Archimedean superrigidity and hyperbolic geometry, Ann. of Math. (2) 135 (1992),
165–182.
[DA 1] J. P. D’Angelo, Several Complex Variables and the Geometry of Real Hypersurfaces, CRC Press,
Boca Raton (1993).
[DA2] J. D’Angelo, Real hypersurfaces, orders of contact, and applications,, Ann. Math. 115 (1982),
615-637.
[DA3] J. P. D’Angelo, Polynomial proper holomorphic mappings between balls, Duke Math J. 57 (1988),
211-219.
[DA4] J. P. D’Angelo, Proper holomorphic mappings between balls in different complex spaces, Mich.
Math. Jour. 38 (1991), 53-65.
[DF1] K. Diederich and J.E. Fornaess, Proper holomorphic mappings between real analytic pseudoconvex
domains in Cn , Math. Ann. 282 (1988), 681-700.
[DF2] K. Diederich and E. Fornaess, Applications holomorphes propres entre domaines a bord analytique
reel, C-R-Paris-Ser.I-Math. 307 (1988), 321- 324.
 PROBLEMS IN SCV AND CR GEOMETRY 15

[DFY] K. Diederich, J. Fornaess, and Z. Ye, Biholomorphisms in Dimension 2, J. Geom. Anal. 4 (1994),
539-552.
[Dr] A. Dor, Proper holomorphic maps between balls in one co-dimensio, Ark. Math. 28 (1990), 49-100.
[Fa1] J. Faran, Maps from the two ball to the three ball, Invent Math. 68 (1982), 441-475.
[Fa2] J. Faran,, The linearity of proper holomorphic maps between balls in the low codimension case,,
J. Diff. Geom. 24 (1986.), 15-17.
[Fa3] J. Faran, A reflection principle for proper holomorphic mappings and geometric invariants, Math.
Z. 203 (1990), 363-377.
[Fr1] F. Forstneric, A survey on proper holomorphic mappings, Proceeding of Year in SCVs at Mittag-
Leffler Institute, Math. Notes, Princeton, NJ: Princeton University Press 38 (1992).
[Fr2] F.Forstneric, Extending proper holomorphic mappings of positive codimension, Invent. Math. 95,
(1989.), 31-62.
[Fr3] F. Forstneric, Complex tangents of real surfaces in complex surfaces, Duke Math. J. 67 (1992),
353–376.
[E] P. Ebenfelt, on the unique continuation problem for CR mappings into nonminimal hypersurfaces,
J. Geom. Anal. 6 (1996), 385–405.
[Eli] Y. Eliashberg, Filling by holomorphic discs and its applications, Geometry of Low-Dimensional
Manifolds, London Math. Soc. Lecture Notes 151 (1997).
[GR] M. Gromov, Pseudo holomorphic curves in symplectic geometry, Invent Math. 82 (1985), 307-347.
[HS] M. Hakim and N. Sibony, Fonctions holomorphes bornes sur la boule unite de Cn , Invent. Math.
67 (1982), 213–222.
[Hu1] X. Huang, On a linearity problem of proper holomorphic mappings between balls in complex spaces
of different dimensions, Jour of Diff. Geom. 51 (No. 1) (1999), to appear.
[Hu2] X. Huang, On an n-manifold in Cn near an elliptic complex tangent, J. Amer. Math. Soc. 11
(1998), 669–692.
[Hu3] X. Huang, Schwarz reflection principle in complex spaces of dimension two, Comm. in PDE 216
(1996), 1781-1828.
[Hu4] X. Huang, A Removable singularity property of CR mappings between real analytic hypersurfaces
in Cn , Comm. in PDE (1999), (to appear).
[Hu5] X. Huang, On the mapping problem for algebraic real hypersurfaces in complex spaces of different
dimensions, Annales de L’Institut Fourier 44 (1994), 433-463.
[Hu6] X. Hunag, A non-degeneracy property of extremal mappings and iterates of holomorphic map-
pings, Annali Scoula Norm. Sup. Pisa, Series IV. Vol. XXI (1994), 399-419.
[Hu7] X. Huang, A preservation principle of extremal mappings and its applications, Illinois J. of Math.
38 (1994), 283-303.
[HJ] X. Huang and S. Ji, Global extension of local holomorphic maps and the Riemann mapping
theorem for algebraic domains, Math. Res. Lett. 5 (1998), 247–260.
[HP] X. Huang and Y. Pan, Proper holomorphic mappings between real analytic domains in Cn , Duke
Math. J. 82 (1996), 437-446.
[HK] X. Huang and S. Krantz, On a problem of Moser, Duke Math. J. 78 (1995), 213-228.
[Jac] H. Jacobowitz, An introduction to CR structures Mathematical Surveys and Monographs, Amer-
ican Mathematical Society, Providence, RI 32 (1990).
[KW1] C. Kenig and S. Webster, The local hull of holomorphy of a surface in the space of two complex
variables, Invent. Math. 67 (1982), 1-21.
[KW2] C. Kenig and S. Webster, On the hull of holomorphy of an n-manifold in C n , Annali Scoula Norm.
Sup. de Pisa IV 11 No. 2 (1984), 261-280.
[Lem1] L. Lempert, La metrique de Kobayashi et la representation des domain sul la boule, Bull. Soc.
Math. France 109 (1981), 427-474.
[Lem2] L. Lempert, A precise result on the boundary regularity of biholomorphic mappings, Math Z. 193
(1986), 559-579.
[Le] H. Lewy, On the boundary behavior of holomorphic mappings, Acad. Zaz. Lincei. 35 (1977), 1-8.
[Ma] D. Ma, On iterates of holomorphic maps, Math. Z 207 (1991), 417-428.
16 XIAOJUN HUANG

[MSY] N. Mok, Y. T. Siu and S. K. Yeung, Geometric superrigidity, Invent. Math. 113 (1993), 57-83.
[Mos] J. Moser, Analytic surfaces in C2 and their local hull of holomorphy, Annales Aca -demiæFennicae
Series A.I. Mathematica 10 (1985), 397-410.
[MW] J. Moser and S. Webster, Normal forms for real surfaces in C2 near complex tangents and
hyperbolic surface transformations, Acta Math. 150 (1983), 255-296.
[Pi] S. Pinchuk, On the proper holomorphic mappings of real analytic pseudoconvex domains, Siberian
Math. 15 (1974), 909-917.
[PS] S. Pinchuk and Tsyganov, Maps between strongly pseudoconvex hypersurfaces, Akad. Nauk. SSSR
53 (1989), 1120-1129.
[Ra] C. Ramanujam, A topological characterisation of the affine plane as an algebraic variety, Ann.
of Math. 94 (1971), 69-88.
[Ros1] J. P. Rosay, A remark on a paper by H. Alexander ‘on the Benneque Problem’, Invent. Math.
116 (1996), 625-627.
[Ros2] J. P. Rosay, A counter-example to a unique continuation problem, Preprint (1998).
[Sta] N. Stanton, Infinite CR automorphism of real heypersurfaces, Amer. Jour. of Math. 118 (1996),
209-233.
[SM] C. L. Siegle and J. Moser, Lectures on Celestial Mechanics, Classics in Mathematics, Springer-
verlag (1995).
[ST] Z. Slodkowski and G. Tomassini, Geometric properties of solutions of the Levi curvature equation
in C2 , J. Funct. Anal. 138 (1996), 188–212.
[Ste] B. Stensones, Proper holomorphic mappings between the balls, Jour. of Geom. Analysis (to appear)
(1994).
[TU] A. Tumanov, Extension of CR functions into a wedge from a manifold of finite type, Math.
USSR-Sb. 64 (1989), 129-140.
[We1] S. Webster, On the mapping problem for algebraic real hypersurfaces, Invent. Math. 43 (1977),
53-68.
[We2] S. Webster, Biholomorphic mappings and the Bergmann kernels off diagonal, Invent. Math. 51
(1979), 155-169.
[We3] S. Webster, On mappings an (n+1)-ball in the complex space, Pac. J. Math. 81 (1979), 267-272.
[We4] S. Webster Surveys on mappings bewteen analytic and algebraic hypersurfaces, Preprint (1995).
[Yue] C. Yue, Dimension and rigidity of quasi-Fuchsian representations, Ann. of Math. 143 (1996),
331–355.
[Za] M. Zaidenberg, Open problems on open algebraic varieties, Montreal (1994).
[Zat] D. Zaitsev, Mappings between algebaric sets, Preprint (1998).

Department of Mathematics, Rutgers University, New Brunswick, NJ 08903, USA