Ark. Mat., 34 (1996), 97-101
@ 1996 by Institut Mittag-Leflter. All rights reserved
Non straightenable complex lines in C 2
Franc Forstneric(1), Josip Globevnik(2) and Jean-Pierre Rosay(1)
Abhyankar and Mob, in [1, (1.6), p. 151], and M. Suzuki, in [11, w proved that if P is a polynomial embedding of C into C 2, then there exists ~b, a polynomial automorphism of C 2, such that ( ~ b o P ) ( C ) = C x {0}. The corresponding and much easier result has been proved for polynomial embeddings of C into C '~, for n > 4 by Z. Jelonek [7] (more generally, Jelonek treats the case of embeddings of C k into C ~, for n > 2 k + 2 ) . The case of polynomial embeddings of C into C 3 seems open. See also
[8].
The main goal of this paper is to show that the above results do not gener- alize to holomorphic embeddings of C. Another goal is an interpolation theorem (Proposition 2 below).
P r o p o s i t i o n 1. Let n > l . There exists a proper holomorphic embedding H: C - ~ C n such that for no automorphism ~ of C ~, ( ~ o H ) ( C ) = C x {0} c C n.
Notice however that it has been proved in [4, (4.1)] that for every R > 0 and s > 0 there exists ~b, an automorphism of C ~, such that I ( r 1 6 2 for every E C, Ir R. So, compact subsets of the complex line H ( C ) can be "approximately straightened".
The above proposition has been known for some time, for n > 3 . In [4, (7.8)]
this is pointed out as being (non stated but) clear in [10]. For n = 2 , we keep the same approach as in [10]. But we can now take advantage of the ground breaking work by Anders@n and Lempert [2], as further developed in [4].
. P r o o f o f P r o p o s i t i o n 1
Proposition 1 is an immediate consequence of the following two propositions:
(1) Partially supported by NSF grant.
(2) Partially supported by a grant from the Ministry of Science of the Republic of Slovenia
98 Franc Forstneric, Josip Globevnik and Jean-Pierre Rosay
P r o p o s i t i o n 2. Let n > l . Let (OLj)jE N be a discrete sequence in C n, (i.e.
I ~ j ] - - ~ - ~ as j--*co). There exists H: C--+C n, a proper holomorphic embedding of C into C n, such that a j E H ( C ) for every j E N .
The case n > 2 is t r e a t e d in [10] (where in addition it is shown that, for n > 2 , the preimages of the points (~j can be arbitrarily prescribed), so the result is new only for n = 2 . We wish to point out that the technique in the proof of Proposition 2, without any substantial modification, allows one to embed C k into C n, so that the image of C k contains an arbitrary given discrete sequence, if 1 < k < n .
P r o p o s i t i o n 3. (Theorem 4.5 in [9].) Let n > l . There exists a discrete se- quence ( a j ) j e N in C n such that for no automorphism r of C n, g ) ( a j ) E C • (for all j 's).
In [9], such sequences are called "non tame". The existence of non tame se- quences is not immediate. Indeed, any sequence is the union of two tame sequences.
An interesting, more detailed, study of non tame sequences is to be found in [6].
All that is left is therefore to prove Proposition 2.
2. P r o o f of P r o p o s i t i o n 2
As already said, our work depends on an extension of the work of Andershn Lempert, as given in [4]. Also, our proof is inspired from [5]. From [4], we shall need only the following:
L e m m a 1. Let K be a polynomially convex compact set in C n ( n > l ) . Let p and q E C n \ K . For every E>O, there exists r an automorphism of C n, such that r and I r for every z E K . In addition we can fix arbitrarily chosen points p l , . . . ,Ps in g (i.e. r
Proof of n e m m a 1. Let ~/: [0, 1 ] - - * c n \ g be an arc, ~/(0)=p, ~ ( 1 ) = q . Apply Theorem 2.1 in [4] to the following situation: In Theorem 2.1 replace K by K U { p } and consider ~ a sufficiently small neighborhood of K U ( p } . Take ~t to be the identity on K , and to be ~ ( z ) = z § near p. Fixing finitely many given points is a trivial addition.
Remark. If K is convex, the lemma is very simple to prove with really elemen- t a r y tools. Without any intent to look for more generality we now simply state the following.
L e m m a 2. Let K be a polynomially convex compact set in C n. Let H be a proper holomorphie embedding of C into C '~. Let R > 0 and L R = { z = H ( ~ ) : ~ E C , I~I<R}. Then the polynomial hull of K U L R is contained in K U H ( C ) .
Non straightenable complex lines in C 2 99
Proof of Lemma
2. Let p E C n. Assume thatp~KUH(C).
Let f be a polyno- mial such that f ( p ) = l , but Ifl < 1 on K . Let g be an entire function which vanishes identically on H ( C ) , but such thatg(p)r
(The existence of such a g follows from Cartan's Theorem A, but in the application we can explicitly exhibit such a g, a polynomial. See the remark at the end of the paper.) Now, for N large enough, we haveIfNy(p)I>SUPKuH(C)IfNgl.
So, p is not in the polynomial hull ofKULR.
We now begin the proof of Proposition 2 itself.
Proof of Proposition
2. We start with the embedding H0: C--*C '~, H0(~)=((, 0) and 0o=0. In the
jth
step of the construction we shall findOj
>0, Q EC, and then construct a proper holomorphic embedding Hi: C--+C n such that:(i) H i ( ~ l ) = a l , / e { 1 , . . . ,j}, (ii) I H j ( ~ ) ] > l a j l - 1 if I(l>Qj,
(iii) I H j ( ( ) - H j _ l ( ( ) l _ < e j < 2 - j for I(I-<Qj with
ej
to be chosen small enough, depending on previous choices,(iv) ej>_0j_l+l.
Once this is done, we set H = l i m
Hj
(uniform convergence on compact sets)9 The inequality]Hj-Hj_ll<2-J
in condition (iii) shows that the sequence of mapsHj
does converge, and that the limit g satisfies: I g ( ~ ) - H j (()1< 1 for I(I ~ ~j+l. Prom (ii) we get that if
~j~-I~l~j+l,
then I H ( ~ ) l > l a J l - 2 . So g is proper9 And (i) implies t h a t H ( ( t ) = a l , for any I.Finally we have to explain the choice of e j, so as to make sure t h a t H is an embedding 9 Let R > 0 , and let G be any holomorphic embedding of C (or of the disk {I(I<R}) into C n, and 0 < r < R . Then, there exists 7 > 0 (depending on G, r and R) such t h a t if
G'
is any holomorphic map from the disk {1(I < R } into C ~ satisfyingIG-G'I
~ , on this disk, then the restriction of G' to the smaller disk {1~1 < r } is an embedding 9 In the ( j - l ) st step of the construction, the radiusOj-1
and the mapHi-1
have been chosen9 In thejth
step the radiusOj
will be chosen first, as will be explained below9 We then apply the above toG=Hj-1, R=~j,
r = Q i _ l , to get~ = ~ i If for every j E N , 9 ~ z = i ez ~ ? i , then the limit map H will be an embedding 9 + ~ Indeed, the restriction of H to any disk {1~1< 0i-1} will be an embedding, since the inequality I H -
H i_ 11 <- ~i
will hold on the disk {1(] < 0i }. A possible choice ofe i
istherefore: e i = 2 - i minl<y (1, ~?l).
Here is a way to find Qj, Q and construct Hi:
We already have H i _ l ( ~ ) = a l for / e { 1 , . . . , j - l } . If
aj=Hj_l(~)
for some( c C , we just take ~j =~,
Hi=Hi_l
and 0i large enough so that (ii) and (iv) hold9 Otherwise (the general case), we choose Qi >- ~Y- 1 + 1 so large that I H i _ 1 (~)1> l ayl for every CEC, I(l>0i. Let F be defined by:F = {z e C~: Izl < l a y ] - l / 2 } U H j - l { l ~ l
<_ Qj}.
100 Franc Forstneric, Josip Globevnik and Jean-Pierre Rosay
T h e polynomial hull of F does not contain a j since a j ~ H j _ I ( C ) , and according to L e m m a 2. Take ~j so t h a t H j _ I ( ~ j ) does not belong to this hull (it is enough to take I~jl large enough). By L e m m a 1 we can find r an a u t o m o r p h i s m of C n, fixing c~1 ,... , a j - 1 , as close as we wish to the identity on F and such t h a t r In particular we take r close enough to the identity on F so t h a t the image of the ball
{IzI<Iajl- 1}
contains the ball {Iz[_<Iajl-1}. So if I~lkt)j then I H j _ l ( ~ ) l > ] a j l , hence I ~ j ( H / _ I ( 4 ) ) I > I c u I - 1 .We set H j = r o H j _ l . Properties (i)-(iv) are immediate to check.
This ends the proof. We just add the following r e m a r k with respect to the proof of L e m m a 2. One has Hj = r ... r176 and H 0 ( ~ ) = ( ~ , 0). So if Zk denotes the k th coordinate function in C n, the functions Zkor -1 . . . r for k E { 2 , . . . , n}, have precisely H j (C) as their c o m m o n zero set.
Note. Y~rther examples, related to Proposition 1, have been given by G. Buz- zard and J. E. Fornaess ([3]). In particular, they give the example of a complex line e m b e d d e d in C 2, whose complement is hyperbolic.
R e f e r e n c e s
1. ABHYANKAR, S. S. and MOH, T. T., Embeddings of the line in the plane, J. Reine Angew. Math. 276 (1975), 148-166.
2. ANDERSI~.N, E. and LEMPERT, L., On the group of holomorphic automorphisms of C n, Invent. Math. 110 (1992), 371-388.
3. BUZZARD, G. and FORNAESS, J. E., An embedding of C in C 2 with hyperbolic complement, Preprint.
4. FORSTNERIC, F. and ROSAY, J-P., Approximation of biholomorphic mappings by automorphisms of C n, Invent. Math. 112 (1993), 323 349. Erratum in Invent.
Math. 118 (1994), 573-574
5. GLOBEVNIK, J. and STENSONES~ S., Holomorphic embeddings of planar domains into C 2, Math. Ann. 303 (1995), 579 597.
6. GRUMAN, L., L' image d'une application holomorphe, Ann. Fac. Sci. Toulouse Math.
12 (1991), 75-101.
7. JELONEK, Z., The extension of regular and rational embeddings, Math. Ann. 277 (1987), 113-120.
8. KRAFT, H., Challenging Problems on A]fine n-space, Bourbaki Seminar 47, 802, June 95, 1994-95.
9. ROSAY, J-P. and RUDIN, W., Holomorphic maps from C n to C n, Trans. Amer. Math.
Soc. 310 (1988), 47-86.
10. ROSAY, J-P. and RUDIN, W., Holomorphic embeddings of C in C ~, in Several Com- plex Variables: Proceedings of the Mittag-LeJ~er Institute, 1987-88 (Fornaess, J. E., ed.), pp. 563-569. Math. Notes 38, Princeton Univ. Press, Princeton, N. J., 1993.
Non straightenable complex lines in C 2 I01
11. SUZUKI, M., Propri@t@s topologiques des polyn0mes de deux variables complexes, et automorphismes alg@briques de l' espace C 2, J. Math. Soc. Japan 26 (1974), 241-257.
Received April 26, 1995 Franc Forstneric
Department of Mathematics University of Wisconsin Madison, WI 53706 U.S.A.
Josip Globevnik
Institute of Mathematics Physics and Mechanics Jadranska 19
61111 Ljubljana Slovenia
Jean-Pierre Rosay
Department of Mathematics University of Wisconsin Madison, WI 53706 U.S.A.