POLONICIMATHEMATICI
92.3(2007)
Proper holomorphi mappings in the spe ial
lass of Reinhardt domains
by
Łukasz Kosiński
(Kraków)Abstra t. A omplete hara terization of proper holomorphi mappings between
domainsfromthe lassofallpseudo onvexReinhardtdomainsin
C 2
withthelogarithmi imageequaltoastriporahalf-planeisgiven.1.Statement of results. Weadoptthestandardnotationsof omplex
analysis. Given
γ = (γ 1 , γ 2 ) ∈ R 2 and z = (z 1 , z 2 ) ∈ C 2 we put |z γ | =
|z γ | =
|z 1 | γ 1 |z 2 | γ 2 whenever it makes sense. The unit dis
in C
is denoted by D
andthesetofproperholomorphi mappingsbetweendomains
D, G ⊂ C nis
denotedby
Prop(D, G).
Inthispaperwe dealwiththosepseudo onvexReinhardt domainsin
C 2
whose logarithmi image is equal to a strip or a half-plane. Observe that
su h domainsarealwaysalgebrai ally equivalent to domainsof theform
D α,r − ,r + := {z ∈ C 2 : r − < |z α | < r + },
where
α = (α 1 , α 2 ) ∈ (R 2 ) ∗ , 0 < r + < ∞, −∞ < r − < r + .
We say that
D α,r − ,r + is of the irrational type if α 1 /α 2 ∈ R \ Q.
In the
other ase itis ofthe rational type.
Re all that if
r − < 0 < r + and α ∈ (R 2 ) ∗ ,
then D α,r − ,r + are so-
alled
elementary Reinhardt domains.
Below we shall give a omplete des ription of all proper holomorphi
mappings from
D α,r −
1 ,r 1 +
toD β,r −
2 ,r + 2
for arbitraryα, β ∈ (R 2 ) ∗ and 0 <
r i + < ∞, −∞ < r − i < r i + , i = 1, 2.
Similar problems have been studied intheliterature.In[Shi1℄and[Shi2℄theproblemofholomorphi equivalen eof
elementary Reinhardt domainswas onsidered. Those results werepartially
extended by A. Edigarian and W. Zwonek [Edi-Zwo ℄ who gave a hara -
2000Mathemati sSubje tClassi ation:32H35,32A07.
Keywordsand phrases:properholomorphi mappings,Reinhardtdomains,elemen-
taryReinhardtdomains.
[285℄
InstytutMatematy znyPAN,2007
terization of proper holomorphi mappings between elementary Reinhardt
domainsof therationaltype.
Set
A(̺ − , ̺ + ) := {z ∈ C : ̺ − < |z| < ̺ + }
for̺ + > 0, ̺ − < ̺ + and
A ̺ := A(1/̺, ̺), ̺ > 1.
Moreover, put
D γ,r := {z ∈ C 2 : 1/r < |z 1 | |z 2 | γ < r}, γ ∈ R ∗ , r > 1, D γ := {z ∈ C 2 : |z 1 | |z 2 | γ < 1}, γ ∈ R ∗ ,
D ∗ γ := {z ∈ C 2 : 0 < |z 1 | |z 2 | γ < 1}, γ ∈ R ∗ .
Notethatif
γ
is rational,i.e.γ = p/q
for some relatively primep, q ∈ Z
,q > 0,
thenD γ,r is biholomorphi
ally equivalent to A r q × C ∗ and D ∗ γ is
D ∗ γ is
biholomorphi allyequivalent to
D ∗ × C.
Indeed, putψ(z 1 , z 2 ) := (z q 1 z p 2 , z 1 m z 2 n )
for(z 1 , z 2 ) ∈ C 2 ,
where
m, n ∈ Z
aresu hthatpm−qn = 1.
One an he kthatthemappingsψ| D γ,r : D γ,r → A r q × C ∗ and ψ| D γ ∗ : D γ ∗ → D ∗ × C ∗ arebiholomorphi
.
Moreover, onemayeasily prove that
D α,r − ,r + isalgebrai allyequivalent to adomain of one ofthe following types:
(i) If
r − > 0
:(a)
A ̺ × C
,α 1 α 2 = 0,
(b)
A ̺ × C ∗,α 1 /α 2 ∈ Q ∗ ,
( )
D γ,̺,γ = α 2 /α 1 ∈ R \ Q
.
(ii) If
r − = 0
:(a)
D ∗ × C
,α 1 α 2 = 0,
(b)
D ∗ × C ∗,α 1 /α 2 ∈ Q ∗ ,
( )
D γ ∗,γ = α 2 /α 1 ∈ R \ Q.
(iii) If
r − < 0
:(a)
D × C
,α 1 α 2 = 0,
(b)
D γ,γ = α 2 /α 1 6= 0.
Ourmain resultisthefollowing:
Theorem 1.
(a)If
α ∈ R \ Q,
then thesetof proper holomorphi mappings fromD α,r
to
D β,R isnon-empty if andonly if
(1)
log R
log r ∈ Z + βZ, α log R
log r ∈ Z + βZ.
(b) Let
α, β ∈ R\Q
and letr, R > 1
be su h thatlog R log r = k 1 + l 1 β and
α log R log r = k 2 + l 2 β
for some integers k i , l i , i = 1, 2.
Then any proper
holomorphi mapping
f : D α,r → D β,R is of one of the following
(2)
( f (z) = (az 1 k 1 z 2 k 2 , bz 1 l 1 z 2 l 2 ),
f (z) = (az 1 −k 1 z 2 −k 2 , bz 1 −l 1 z 2 −l 2 ), z = (z 1 , z 2 ) ∈ D α,r ,
where
a, b ∈ C
satisfy|a| |b| β = 1.
Moreover, any of the mappingsgiven by (2)is proper.
Noti ethatinTheorem 1(a)we do notdemand
β
tobe irrational.Using Theorem 1 we will easily obtain analogous results for domains of
theforms (ii)and (iii)of the irrationaltype.
Theorem2. Let
α, β ∈ R \ Q.
The setof proper holomorphi mappings fromD α ∗ to D ∗ β is non-empty if and only if α = (k 2 + βl 2 )/(k 1 + βl 1 )
for
α = (k 2 + βl 2 )/(k 1 + βl 1 )
forsome
k i , l i ∈ Z
,i = 1, 2.
Moreover, in that ase, ifk 1 + l 1 β > 0,
then anyproper holomorphi mapping
f : D α ∗ → D ∗ β is of the form
(3)
f (z 1 , z 2 ) = (az 1 k 1 z k 2 2 , bz 1 l 1 z 2 l 2 ), (z 1 , z 2 ) ∈ D ∗ α ,
where
a, b ∈ C
satisfy|a| |b| β = 1.
Theorem3. Let
α, β ∈ R \ Q.
Then thesetProp(D α , D β )
isnon-emptyif and only if
α = (k 2 + βl 2 )/(k 1 + βl 1 )
for somek i , l i ∈ Z ≥0 , i = 1, 2.
Moreover, in that ase any proper holomorphi mapping
f : D α → D β is of
the form
(4)
f (z 1 , z 2 ) = (az 1 k 1 z k 2 2 , bz 1 l 1 z 2 l 2 ), (z 1 , z 2 ) ∈ D α ,
where
a, b ∈ C
are su h that|a| |b| β = 1.
Nextweprove the following
Theorem 4. Let
α, β ∈ (R 2 ) ∗ , r + i > 0, r − i < r + i , i = 1, 2.
Assumethat the sets
D α,r −
1 ,r 1 + , D β,r −
2 ,r + 2
are of the same type (either rational orirrational).If thereexistsaproper holomorphi mappingbetweenthem,then
either
r − 1 r 2 − > 0
orr 1 − = r − 2 = 0.
For domainsof dierent typeswe havethefollowing result:
Theorem 5. Let
α, β ∈ (R 2 ) ∗ , r + i > 0, r − i < r i + , i = 1, 2.
If thesets
D α,r −
1 ,r + 1
andD β,r −
2 ,r + 2
are of dierent types, then there is no properholomorphi mapping betweenthem.
Finally, we dis uss the rational ase. As already mentioned, the set of
proper holomorphi mappings between elementary Reinhardt domains of
the rational type was des ribed in [Edi-Zwo ℄. Thus, in order to obtain the
desired hara terization, itsu esto prove the following threetheorems.
Theorem 6. Let
r, R > 1.
IfR 6= r m for any natural number m,
then
Prop(A r × C, A R × C), Prop(A r × C ∗ , A R × C)
and Prop(A r × C ∗ , A R × C ∗ )
are empty. Moreover,for any
m ∈ N :
(a)
Prop(A r × C, A r m × C)
onsistsof the mappings of the formA r × C ∋ (z, w) 7→ (e iθ z εm , a N (z)w N + · · · + a 0 (z)) ∈ A r m × C,
where
θ ∈ R, N ∈ N, ε = ±1
anda 0 , . . . , a N ∈ O(A r )
aresu h that|a 0 (z)| + · · · + |a N (z)| > 0, z ∈ A r .
(b)
Prop(A r × C ∗ , A r m × C)
onsistsof the mappings of theformA r × C ∗ ∋ (z, w) 7→
e iθ z εm , a N (z)w N + · · · + a 0 (z) w k
∈ A r m × C,
where
θ ∈ R
,k, N ∈ N
,0 < k < N
,ε = ±1
anda i ∈ O(A r )
,i = 1, . . . , N
, satisfy|a 0 (z)| + · · · + |a k−1 (z)| > 0
and|a k+1 (z)| +
· · · + |a N (z)| > 0
forz ∈ A r .
( )
Prop(A r × C ∗ , A r m × C ∗ )
onsistsof the mappings of the formA r × C ∗ ∋ (z, w) 7→ (e iθ z m , a(z)w k ) ∈ A r m × C ∗ ,
where
ε = ±1, θ ∈ R, k ∈ N
anda ∈ O(A r , C ∗ ).
Theorem 7. There are no proper holomorphi mappings from
A r × C
to
A R × C ∗ for any r, R > 1.
Theorem 8.
(a)
Prop(D ∗ × C, D ∗ × C)
onsistsof the mappings of the formD ∗ × C ∋ (z, w) 7→ (e iθ z m , a N (z)w N + · · · + a 0 (z)) ∈ D ∗ × C,
where
θ ∈ R
,N ∈ N
,m ∈ N
anda 0 , . . . , a N ∈ O(D ∗ )
are su h that|a 0 (z)| + · · · + |a N (z)| > 0, z ∈ D ∗ .
(b)
Prop(D ∗ × C ∗ , D ∗ × C)
onsistsof the mappings of the formD ∗ × C ∗ ∋ (z, w) 7→
e iθ z m , a N (z)w N + · · · + a 0 (z) w k
∈ D ∗ × C,
where
θ ∈ R
,m ∈ N
,k, N ∈ N
,0 < k < N
anda i ∈ O(D ∗ )
,i = 1, . . . , N
satisfy|a 0 (z)| + · · · + |a k−1 (z)| > 0
and|a k+1 (z)| +
· · · + |a N (z)| > 0
forz ∈ D ∗ .
( )
Prop(D ∗ × C ∗ , D ∗ × C ∗ )
onsistsof the mappings of the formD ∗ × C ∗ ∋ (z, w) 7→ (e iθ z m , a(z)w k ) ∈ D ∗ × C ∗ ,
where
θ ∈ R, k ∈ N
anda ∈ O(D ∗ , C ∗ ).
(d)
Prop(D ∗ × C, D ∗ × C ∗ ) = ∅
.2. Proofs. The following result is probably known. However, we ould
not ndit inthe literature, so we present a proof.
Lemma 9. Let
D ⊂ C n be a domain, α ∈ R \ Q
and let f, g : D → C
be holomorphi mappings with
|f (z)| = |g(z)| α , z ∈ D.
Then eitherf =
g = 0
onD
or there exists a holomorphi bran h of the logarithm ofg
, i.e.a mapping
ψ ∈ O(D)
su h thate ψ = g
onD.
In parti ular, there exists aθ ∈ R
su h thatf = e iθ+αψ on D.
Proof. Comparing multipli ities of the roots of the fun tions
f
andg
omposed with ane mappings we may redu e our onsiderations to the
ase when
f, g : D → C ∗ .
Moreover, we may assume thatg(x ′ ) ∈ R >0 for
some
x ′ ∈ D.
Obviously, there exists an
η ∈ R
su h that the setG η := {z ∈ D : e iη f (z) ∈ g(z) α }
is non-empty. Considering, if ne essary, a mappinge iη f
instead of
f
we mayassumethatη = 0.
Itiseasytoseethat
G 0isanopen-and-
losedsubsetofD
,andsoG 0 = D.
Thus,thereexistsaholomorphi bran hof
g α (alsodenotedbyg α)su
hthat
g α (x ′ ) ∈ R >0.Itfollowsthatthereexistg tforanyt ∈ Q := {k+lα : k, l ∈ Z}.
g α (x ′ ) ∈ R >0.Itfollowsthatthereexistg tforanyt ∈ Q := {k+lα : k, l ∈ Z}.
t ∈ Q := {k+lα : k, l ∈ Z}.
Fixasequen e
(t m ) ∞ m=1 ⊂ Q
onverging to0
.InvirtueofMontel'stheorem,itis lear that
g t m → 1
lo ally uniformly.Put
ψ m := (g t m − 1)/t m .
Thenlim m→∞ ψ m (x ′ ) = log g(x ′ )
and thesequen e
(ψ ′ m ) ∞ m=1 = (g t m −1 g ′ ) ∞ m=1 is lo ally uniformly onvergent with
limit
(1/g)g ′ .
Thus(ψ m ) ∞ m=1
onverges lo
ally uniformly on D
. Denote its
limit by
ψ.
By the Weierstrass theorem,ψ
is holomorphi onD
andψ ′ = lim m→∞ ψ m ′ = (1/g)g ′ .
Let
D ⊂ D e
beanysimply onne tedneighborhoodofx ′ .
Letψ e
beaholo-morphi mapping on
D e
su h thatg| D e = e ψ eand ψ(x e ′ ) = log g(x ′ ).
Itis easy
to seethat
ψ = ψ e
onD, e
andso, bythe identity prin iple,g = e ψ onD.
Lemma 10. Let
0 < r + i , −∞ < r i − < r + i , i = 1, 2, α, β ∈ R.
Let(λ n ) ∞ n=1 ⊂ A(r − 1 , r + 1 ).
Letφ ∈ Prop(D (1,α),r −
1 ,r + 1 , D (1,β),r −
2 ,r + 2 ). Put
v(λ) := |φ 1 (λ, 1)| |φ 2 (λ, 1)| β , λ ∈ A(r − 1 , r + 1 ).
If the sequen e
(λ n ) ∞ n=1 has no a
umulation points in A(r − 1 , r 1 + ),
then
(v(λ n )) ∞ n=1 has no a
umulation points in A(r 2 − , r + 2 ).
A(r 2 − , r + 2 ).
Proof. Assume that
v(λ n ) → q.
It su esto show thatq ∈ ∂A(r − 2 , r + 2 ).
Otherwise
q ∈ A(r − 2 , r + 2 ).
Notethat for anyλ ∈ A(r 1 − , r 1 + )
thefun tion(5)
u λ : C ∋ z 7→ |φ 1 (λe −αz , e z )| |φ 2 (λe −αz , e z )| β
isbounded andsubharmoni , so
u λ is onstant.
Sin e
φ
is proper, the mappingC ∋ z 7→ φ 2 (λ n e −αz , e z ) ∈ C
is non-onstant for any
n ∈ N.
Pi ard's theorem implies that there is a sequen e(z n ) ∞ n=0 ⊂ C
su h that|φ 2 (λ n e −αz n , e z n )| β = 1.
Obviouslyu λ (z) = u λ (1) = v(λ)
forallz ∈ C
andv(λ n ) → q,
so|φ 1 (λ n e −αz n , e z n )| → q.
Inparti ular,the set{φ(λ n e −αz n , e z n ) : n ∈ N}
isrelatively ompa tinD (1,β),r −
2 ,r 2 +
;however,((λ n e −αz n , e z n )) ∞ n=1 has no a
umulation points inD (1,α),r −
1 ,r 1 +
,a ontradi -Corollary11. Let
φ = (φ 1 , φ 2 ) : D α,r → D β,Rbeaproperholomorphi
mapping and let α, β ∈ R >0 , r, R > 1
. Put v(λ) := |φ 1 (λ, 1)| |φ 2 (λ, 1)| β , λ ∈ A r .
Then either
|λ|→1/r lim v(λ) = 1/R, lim
|λ|→r v(λ) = R or lim
|λ|→1/r v(λ) = R, lim
|λ|→r v(λ) = 1/R.
Lemma 12. Let
α ∈ R \ Q, β ∈ R, −∞ < r − i < r i + < ∞, 0 < r i + , i = 1, 2,
and letφ : D (1,α),r −
1 ,r + 1 → D (1,β),r −
2 ,r 2 +
be a holomorphi mapping. Then for anyλ ∈ A(r 1 − , r + 1 )
,φ({(z 1 , z 2 ) ∈ C 2 : |z 1 | |z 2 | α = |λ|})
⊂ {(w 1 , w 2 ) ∈ C 2 : |w 1 | |w 2 | β = |φ 1 (λ, 1)| |φ 2 (λ, 1)| β }.
Proof. Note thatfor any
λ ∈ A(r − 1 , r 1 + )
the fun tionu : C ∋ z 7→ |φ 1 (λe αz , e −z )| |φ 2 (λe αz , e −z )| β
issubharmoni and bounded. Hen e
u
is onstant.Invirtueof Krone ker's theorem, theset
{(|λ|e αz , e −z ) : z ∈ C}
isdensein
{(z 1 , z 2 ) ∈ C 2 : |z 1 | |z 2 | α = |λ|}.
Thus, thereist ∈ R
su h thatφ({(z 1 , z 2 ) ∈ C 2 : |z 1 | |z 2 | α = |λ|}) ⊂ {(w 1 , w 2 ) ∈ C 2 : |w 1 | |w 2 | β = t}.
It iseasy to seethat
t = |φ 1 (λ, 1)| |φ 2 (λ, 1)| β .
Proof of Theorem 1(a). Let
φ : D α,r → D β,R be a proper holomorphi
mapping. Put v(λ) := |φ 1 (λ, 1)| |φ 2 (λ, 1)| β , λ ∈ A r .
Obviously, log v
is a
harmoni fun tion.ApplyingCorollary11andHadamard'stheoremweinfer
that
v(λ) = |λ| log R log r , λ ∈ A r or v(λ) = |λ| − log R log r , λ ∈ A r .
From thisand Lemma12 we easily on lude thatthereis
ε = ±1
su hthat(6)
|φ 1 (z)| |φ 2 (z)| β = |z 1 | ε log R log r |z 2 | εα log R log r , z ∈ D α,r .
Let
z 2 = 1, z 1 = z ∈ A r , ψ i (z) := φ i (z, 1), i = 1, 2.
Thenlog(ψ 1 (z)ψ 1 (z)) + β log(ψ 2 (z)ψ 2 (z)) = ε log R
log r log(zz).
Dierentiatingwithrespe tto
z
we get(7)
ψ 1 ′ (z)
ψ 1 (z) + β ψ 2 ′ (z)
ψ 2 (z) = ε log R log r · 1
z , z ∈ A r .
It follows that
Ind(ψ 1 ◦ γ; 0) + β Ind(ψ 2 ◦ γ; 0) = ε log R log r ,
where
γ
is the unit ir le. Hen elog R log r ∈ Z + βZ. The same argument with
respe t to these ondvariableshows that
α log R log r ∈ Z + βZ.
To provethe onverse, assumethat
log R
log r = k 1 + l 1 β, α log R
log r = k 2 + l 2 β,
where
k i , l i ∈ Z
,i = 1, 2.
Deneφ 1 (z) := z 1 k 1 z 2 k 2 , φ 2 (z) := z 1 l 1 z 2 l 2 for z = (z 1 , z 2 ) ∈ C 2, and φ := (φ 1 , φ 2 ).
Observe that φ| D α,r ∈ Prop(D α,r , D β,R ).
φ := (φ 1 , φ 2 ).
Observe thatφ| D α,r ∈ Prop(D α,r , D β,R ).
Indeed,it iseasy to he k that
(8)
|φ 1 (z)| |φ 2 (z)| β = |z 1 | log R/log r |z 2 | α log R/log r , (z 1 , z 2 ) ∈ D α,r ,
so
φ| D α,r ∈ O(D α,r , D β,R ).
Sin ek 1 l 2 6= k 2 l 1 , φ
is a proper holomorphi mappingfrom(C ∗ ) 2intoitself(see[Zwo ,Theorem2.1℄).Nowweimmediately
on
lude from (8) that φ| D α,r is a proper holomorphi
mapping from D α,r
D α,r
to
D β,R.
Lemma9and (6)leadto the following
Corollary 13. Let
α, β ∈ R \ Q
andφ ∈ Prop(D α,r , D β,R ).
Assumethat
log R
log r = k 1 + l 1 β andα log R log r = k 2 + l 2 β
forsomek i , l i ∈ Z, i = 1, 2.
Then
there are
θ ∈ R, ψ ∈ O(D α,r )
andε ∈ {1, −1}
su h thatφ(z) = (z 1 εk 1 z 2 εk 2 e iθ e −βψ(z) , z 1 εl 1 z 2 εl 2 e ψ(z) ), z ∈ D α,r .
Remark
14.
We may always assume thatε
in Corollary 13 is equalto
1
(repla ing if ne essaryφ
byφ ◦ h,
whereh ∈ Aut(D α,r )
,h(z 1 , z 2 ) :=
(z 1 −1 , z −1 2 )
).To prove Theorem 1(b) we need the following notation. Put
X α,r :=
{z ∈ C 2 : − log r < Re z 1 + α Re z 2 < log r}
andΠ(z 1 , z 2 ) := (e z 1 , e z 2 )
for(z 1 , z 2 ) ∈ C 2 .
It is lear that(X α,r , Π)
istheuniversal overing ofD α,r .
Lemma 15. Let
α, β ∈ R \ Q, r, R > 1
and assume thatlog R log r = k 1 + l 1 β, α log R log r = k 2 + l 2 β, where k i , l i ∈ Z, i = 1, 2.
Let f : D α,r → D β,R
be a
proper holomorphi mapping. Then every ontinuous lifting of the mapping
f ◦ Π : X α,r → D β,R is proper and holomorphi .
Proof. By Corollary 13 and Remark 14 we may assume that
f (z) = (z 1 k 1 z 2 k 2 e −βψ(z)+iθ , z l 1 1 z l 2 2 e ψ(z) ) (z ∈ D α,r ),
whereθ ∈ R
andψ ∈ O(D α,r ).
Let
f e
be any ontinuous lifting off ◦ Π : X α,r → D β,R ,
that is,f : e X α,r → X β,R and f ◦ Π = Π ◦ e f .
Itis obviousthat f e
is holomorphi
.Then
bytheidentityprin
iple
(9)
( f e 1 (z) = k 1 z 1 + k 2 z 2 − βψ(e z 1 , e z 2 ) + iθ + 2µ 1 πi,
f e 2 (z) = l 1 z 1 + l 2 z 2 + ψ(e z 1 , e z 2 ) + 2µ 2 πi, z ∈ X α,r ,
for some
µ i ∈ Z, i = 1, 2.
Suppose that
f e
is not proper, i.e. there is a sequen e(z m ) ∞ m=1 ⊂ X α,r,
z m = (z 1 m , z 2 m )
,m ∈ N
, without any a umulation pointsinX α,r su
hthat
( e f (z m )) ∞ m=1 is
onvergent inX β,R .
Put y 0 := lim m→∞ f (z e m ) ∈ X β,R .
X β,R .
Puty 0 := lim m→∞ f (z e m ) ∈ X β,R .
Obviously,
f (Π(z 1 m , z 2 m )) = Π( e f (z m 1 , z 2 m )) → Π(y 0 ).
Sin ef
is proper,theset
{Π(z m ) : m ≥ 1}
isrelatively ompa tinD α,r .
Thuswemayassumethat
(Π(z m )) ∞ m=1 is
onvergent, sayto w 0 ∈ D α,r .
From(9)wededu
ethat
(k 1 z m 1 +k 2 z 2 m ) ∞ m=1and(l 1 z 1 m +l 2 z 2 m ) ∞ m=1are
onvergentinC 2 .
Thus(z m ) ∞ m=1
(l 1 z 1 m +l 2 z 2 m ) ∞ m=1are
onvergentinC 2 .
Thus(z m ) ∞ m=1
isalso onvergent.
Put
z 0 := lim m→∞ z m .
Now it su es to observe thatΠ(z 0 ) = w 0 ∈ D α,r ,
soz 0 ∈ X α,r ;
a ontradi tion.Nowwe are able to give a des ription of theset of proper holomorphi
mappingsbetween the domains
D α,r and D β,R of theirrationaltype.
Proof ofTheorem1(b). Let
f ∈ Prop(D α,r , D β,R )
.InvirtueofCorollary13 andRemark 14wemayassume that
f (z) = (z 1 k 1 z 2 k 2 e −βψ(z)+iθ , z l 1 1 z l 2 2 e ψ(z) ), z = (z 1 , z 2 ) ∈ D α,r ,
for some
θ ∈ R
andψ ∈ O(D α,r ).
Our aimis to showthatψ
is onstant.To simplifynotation, for
γ ∈ R
putΛ γ : C 2 ∋ (z 1 , z 2 ) 7→ (z 1 + γz 2 , z 2 ) ∈ C 2 .
It is lear that
Λ γ (X γ,̺ ) = S ̺ × C
,̺ > 1,
whereS ̺ := {z ∈ C : − log r <
Re z < log r}.
Moreover,Λ γ isbiholomorphi
withinverse Λ −1 γ = Λ −γ .
Notethatthe mapping
f : X e α,r → X β,R given by
f (z) = (k e 1 z 1 + k 2 z 2 − βψ(e z 1 , e z 2 ) + iθ, l 1 z 1 + l 2 z 2 + ψ(e z 1 , e z 2 ))
isaliftingof
f ◦Π.
ThusLemma15impliesthatf e
isproperandholomorphi . PutH := (H 1 , H 2 ) := Λ β ◦ e f ◦ Λ −1 α : S r × C → S R × C.
Obviously,H
isproperand holomorphi .
Applyingtherelations
log R
log r = k 1 + l 1 β,α log R log r = k 2 + l 2 β
we seethat
(10) H(z) = (z 1 (k 1 + βl 1 ) + iθ, l 1 z 1 + z 2 (l 2 − l 1 α) + ψ(e z 1 −αz 2 , e z 2 )),
z ∈ S r × C.
Hen e for any
z 1 ∈ S r the mapping C ∋ z 7→ H 2 (z 1 , z) ∈ C
is proper
andholomorphi .Consequently,duetothe formofproperholomorphi self-
mappingsof
C,
there isa polynomialp = p z 1 ∈ P(C)
su h thatH 2 (z 1 , z) = p(z).
Therefore, the polynomialq(z) := q z 1 (z) := p(z) − l 1 z 1 − z(l 2 − l 1 α)
satises theequation
(11)
ψ(e z 1 e −αz , e z ) = q(z), z ∈ C.
Noti ethat
{(e z 1 e −α2πim , e 2πim ) : m ∈ N}
isa relatively ompa tsubsetof
D α,r and the sequen
e{q(2πim)} ∞ m=1 isbounded.Thus the polynomialq
q
Put
c(z 1 ) := ψ(e z 1 −αz 2 , e z 2 ), z 1 ∈ S r .
Fix any1 < ̺ < R
and takea onstant
M = M (̺) > 0
su hthat|c(x)| < M
foreveryx ∈ [− log ̺, log ̺].
Let
λ ∈ ̺D \ (1/̺)D
be arbitrary. Note that for anyz 2 ∈ C
we have|ψ(|λ|e −αz 2 , e z 2 )| = |c(log |λ|)| < M.
ApplyingKrone ker's theoremweinfer thattheset{(|λ|e −αz , e z ) : z ∈ C}
isdensein{(z 1 , z 2 ) ∈ C 2 : |z 1 | |z 2 | α = |λ|}.
Consequently,
ψ| D α,̺ isbounded.
Nowitsu es to repeatthe proof of Lemma2.7.1 of [Jar-P1℄inorder
toshowthateveryboundedholomorphi mappingon
D α,̺ (inparti
ular,ψ
)
is onstant.
On the other hand, we have already mentioned in the proof of Theo-
rem 1(a)thatanymapping given by(2)isproper.
Proof of Theorems 2and 3. We proveboththeoremsimultaneously.Let
f : D α → D β (respe
tively,f : D ∗ α → D β ∗)beaproperholomorphi
fun
tion.
We aim at redu ing the situationto that ofTheorem 1.Takeany
r > 1.
From Lemma 12 we seethat for any
t ∈ [0, 1)
(resp.t ∈ (0, 1)
) there isan
s(t) ∈ [0, 1)
(resp.s(t) ∈ (0, 1)
) su hthatf ({(z 1 , z 2 ) ∈ C 2 : |z 1 | |z 2 | α = t}) ⊂ {(w 1 , w 2 ) ∈ C 2 : |w 1 | |w 2 | β = s(t)}.
Note that
s(|λ|) = |f 1 (λ, 1)| |f 2 (λ, 1)| β and the fun
tion v
given byv : D ∋ λ 7→ s(|λ|) ∈ [0, 1]
(resp. v : D ∗ ∋ λ 7→ s(|λ|) ∈ [0, 1]
) is radial and
subharmoni on
D
(inthe se ond asewemayremove thesingularityat0
).Themaximumprin iple applied to
v
impliesthats
isin reasing.In parti ular,
f | D
(1,α),1/r2,1 : D (1,α),1/r 2 ,1 → D (1,β),1/R 2 ,1
is proper forsome
R > 1
. For̺ > 1
putΛ e ̺ : C 2 ∋ (z 1 , z 2 ) 7→ (̺z 1 , z 2 ) ∈ C 2 and
dene
ψ := e Λ R ◦ f ◦ e Λ −1 r | D α,r .
Note thatψ ∈ Prop(D α,r , D β,R )
. ApplyingTheorem 1 we ndthat
log R
log r = k 1 + l 1 β, α log R log r = k 2 + l 2 β and ψ(z 1 , z 2 ) = (az εk 1 1 z 2 εk 2 , bz 1 εl 1 z 2 εl 2 )
for some k i , l i ∈ Z, i = 1, 2, ε = ±1
and a, b ∈ C
satisfying
|a| |b| β = 1.
Oviouslyα = (k 2 + l 2 β)/(k 1 + l 1 β)
andbytheidentityprin iple we obtain
(12) f (z 1 , z 2 ) = (ar εl 1 β z εk 1 1 z 2 εk 2 , br −εl 1 z 1 εl 1 z 2 εl 2 ),
(z 1 , z 2 ) ∈ D α (
resp.(z 1 , z 2 ) ∈ D α ∗ ).
If
f : D α ∗ → D β ∗ ,
then it su es to noti e that|f 1 (z)| |f 2 (z)| β = (|z 1 | |z 2 | α ) εk 1 +εl 1 β , z = (z 1 , z 2 ) ∈ D ∗ α ,
hen eε(k 1 + l 1 β) > 0.
When
f : D α → D β ,
we on lude thatεk i , εl i ≥ 0, i = 1, 2.
Hen ewe easily gettherequired formulas.
On the other hand, one an he k that any of the mappings given in
Theorem 3is proper (sin e
α
isirrational,k 1 l 2 − k 2 l 1 6= 0
).Lemma 16. Let
r + > 0
,r − < r +, t ∈ R.
Suppose that the fun
tion
v : A(r − , r + ) → [−∞, t)
is subharmoni
, radial (i.e. v(|λ|) = v(λ)
, λ ∈
A (r − , r + )
) and harmoni on the set{z ∈ A(r − , r + ) : v(z) 6= −∞}.
Thenthere exist
a, b ∈ R
su h thatv(λ) = a log |λ| + b, λ ∈ A(r − , r + ).
Proof. Itsu estoobserve thatsin e
v
isradial,A(r − , r + ) \ {0} ⊂ {z ∈ A(r − , r + ) : v(z) 6= −∞}
(and next one maypro eedina standard way,i.e.solve aneasy dierential equation).
Proofof Theorem4. First, onsiderthe asewhen
D α,r −
1 ,r + 1
andD β,r − 2 ,r + 2
are of the irrational type. Then we may assume that
α = (1, α 1 )
for someα 1 ∈ R \ Q.
Letv : A(r − 1 , r 1 + ) ∋ λ 7→ log |ψ 1 (λ, 1)| β 1 |ψ 2 (λ, 1)| β 2 ∈ R.
By Lemma12 we seethat
ψ({(z 1 , z 2 ) ∈ C 2 : |z 1 | |z 2 | α = |λ|} ⊂ {(w 1 , w 2 ) ∈ C 2 : |w 1 | β 1 |w 2 | β 2 = e v(λ) }.
Therefore,v
isradial.Observemoreoverthatv
issubharmoni on
A(r − 1 , r 1 + )
and harmoni on{λ ∈ A(r 1 − , r + 1 ) : v(λ) > −∞}.
Sin e
ψ
issurje tive, we on lude that(13)
v(A(r 1 − , r 1 + )) =
(log r 2 − , log r + 2 )
ifr − 2 ≥ 0, [−∞, log r + 2 )
ifr − 2 < 0
(we put
log 0 := −∞
). However, by Lemma 16,v(λ) = a log |λ| + b, λ ∈ A(r − 1 , r 1 + )
,for somea, b ∈ R,
whi h easilynishes theproof inthis ase.Nowsupposethat
D α,r −
1 ,r + 1
andD β,r −
2 ,r + 2
areoftherationaltype;wemayassume that
β = (p, q) ∈ Z 2 and α = (1, α 1 )
for some α 1 ∈ Q.
Applying
Lemma10 one ansee thatthemapping
A(r 1 − , r 1 + ) ∋ λ 7→ ψ 1 (λ, 1) p ψ 2 (λ, 1) q ∈ A(r 2 − , r + 2 )
isproper.Hen ethis asefollows dire tlyfromtheform ofthesetof proper
holomorphi mappingsfrom
A(r 1 − , r 1 + )
toA(r 2 − , r 2 + ).
Proof of Theorem 5. Assume that
D α,r −
1 ,r + 1
is of the rational type andD β,r −
2 ,r + 2
is of theirrationaltype;without loss of generalityα = (1, p/q)
forsome
p, q ∈ Z
andβ = (1, β 2 )
for someβ 2 ∈ R \ Q
.Suppose that
ψ ∈ Prop(D α,r −
1 ,r + 1 , D β,r −
2 ,r + 2 ). Note that for any λ ∈ A (r − 1 , r 1 + )
the mapping
(14)
u λ : C ∗ ∋ z 7→ |ψ 1 (λz p , z −q )| |ψ 2 (λz p , z −q )| β 2
is onstant.Fix
λ 0 andc 6= 0
su
hthatu λ 0 ≡ c.
One
an seethatC ∗ ∋ z 7→
ψ i (λ 0 z p , z −q ) ∈ C ∗ is a proper holomorphi
self-mapping of C ∗ , i = 1, 2.
Therefore,thereare
a i ∈ C ∗andµ i ∈ Z ∗ , i = 1, 2,
su
hthatψ i (λ 0 z p , z −q ) = a i z µ i for z ∈ C ∗,i = 1, 2.
Applying (14)it is
lear that |a 1 | |a 2 | β 2 |z| µ 1 +µ 2 β 2
z ∈ C ∗,i = 1, 2.
Applying (14)it is
lear that |a 1 | |a 2 | β 2 |z| µ 1 +µ 2 β 2
= c
forz ∈ C ∗ .
Inparti ular,β 2 ∈ Q
,a ontradi tion.Now, suppose that there exists
ψ ∈ Prop(D β,r −
2 ,r + 2 , D α,r −
1 ,r + 1 ). Put u(λ) := |ψ 1 (λ, 1)| |ψ 2 (λ, 1)| β 2
for λ ∈ A(r 2 − , r + 2 ).
Applying Lemmas 10 and 12 we nd that
u
satises the assumptions of Lemma 16. Thus, there area, b ∈ R
su h thatlog u(λ) = a log |λ| + b
for
λ ∈ A(r 2 − , r + 2 ).
In parti ular,u
is either stri tly in reasing or stri tlyde reasing. Takeany
̺ − 2,̺ + 2 su
hthat̺ − 2 > max{0, r − 2 }, ̺ + 2 < r 2 + , ̺ − 2 < ̺ + 2 .
̺ − 2 > max{0, r − 2 }, ̺ + 2 < r 2 + , ̺ − 2 < ̺ + 2 .
Put
̺ − 1 := min{u(̺ − 2 ), u(̺ + 2 )}, ̺ + 1 := max{u(̺ − 2 ), u(̺ + 2 )}.
Thenψ| D
β,̺− 2 ,̺ + 2
: D β,̺ −
2 ,̺ + 2 → D (1,α),̺ − 1 ,̺ + 1
isobviouslyaproperholomorphi mapping.InvirtueofTheorem1(a)there
are
k i , l i ∈ Z
,i = 1, 2,
su h thatβ = (k 1 + l 1 α)/(k 2 + l 2 α).
In parti ular,β ∈ Q,
a ontradi tion.Lemma 17. Let
A, B ⊂ C n be domains andassume that B
is bounded.
(a) Amapping
f : A×C ∗ → B ×C
isproperandholomorphi ifandonly ifthere arem ∈ Prop(A, B), k ∈ N, 0 < k < N, N ∈ N, a i ∈ O(A)
,i = 1, . . . , N,
with|a 0 (z)| + · · · + |a k−1 (z)| > 0
and|a k+1 (z)| + · · · +
|a N (z)| > 0
forz ∈ A
,satisfyingf (z, w) =
m(z), a N (z)w N + · · · + a 0 (z) w k
, (z, w) ∈ A × C ∗ .
(b) A mapping
f : A × C → B × C
isproperandholomorphi ifandonly iftherearea 0 , . . . , a N ∈ O(A)
,N ∈ N
,with|a 0 (z)|+· · ·+|a N (z)| > 0
for
z ∈ A
, and there is a proper holomorphi mappingm : A → B
su h that
f (z, w) = (m(z), a N (z)w N + · · · + a 0 (z)), (z, w) ∈ A × C.
( ) A mapping
f : A × C ∗ → B × C
is proper and holomorphi if and only ifthere arem ∈ Prop(A, B), a ∈ O(A, C ∗ )
andk ∈ N
su hthatf (z, w) = (m(z), a(z)w k ), (z, w) ∈ A × C ∗ .
(d) There isno proper holomorphi mapping from
A × C
toB × C ∗.
Proof. First of all,noti e that for any
z ∈ A
themappingw 7→ f 1 (z, w)
∈ C n isbounded on C
(orC ∗),so itis
onstant.
(a)Observe that
C ∗ ∋ w 7→ f 2 (z, w) ∈ C
isproperfor anyz ∈ A.
Thus,for any
z ∈ A
there is a polynomialp(z, ·), p(z, 0) 6= 0,
and a naturalk(z)
su h that
(15)
φ 2 (z, w) = p(z, w)
w k(z) , (z, w) ∈ A × C ∗ .
One an see that there is a
k
su h thatk = k(z)
forz ∈ A
(use Rou he'stheorem).Consequently,
p ∈ O(A × C ∗ )
.Fixanydomain
A ′ ⊂⊂ A
and putA µ :=
z ∈ A ′ : ∂ µ p
∂w µ (z, w) = 0
foranyw ∈ C
.
Theabove onsiderations implythat
S ∞
µ=1 A µ = A ′ .Applying Baire'stheo-
rem we nd that there exists
N ∈ N
su h thatA N has non-empty interior.
By theidentity prin iple,
A N = A.
Thus, there are holomorphi mappings
a 0 , . . . , a N : A → C
su h thatp(z, w) = a N (z)w N + · · · + a 1 (z)w + a 0 (z)
for(z, w) ∈ A × C,
i.e.(16)
f 2 (z, w) = a N (z)w N + · · · + a 1 (z)w + a 0 (z)
w k , (z, w) ∈ A × C.
By properness of
f 2 (z, ·)
we on lude that0 < k < N
, and|a N (z)| + . . . +
|a k+1 (z)| > 0
and|a k−1 (z)| + · · · + |a 0 (z)| > 0
for anyz ∈ A.
Put
m(z) := f 1 (z, 1), z ∈ A.
We laim thatm
isproper.Indeed, take any sequen e
(z n ) ∞ n=1 and assume that it has no a umu-
lation points in
A.
We may assume thata 0 (z n ) 6= 0
for anyn ∈ N
(if ne -essary repla e
a 0 witha 1 et
.).Then there exists a sequen
e (w n ) ∞ n=1 ⊂ C ∗
(w n ) ∞ n=1 ⊂ C ∗
su h that
a N (z n )w n N + · · · + a 1 (z n )w n + a 0 (z n ) = 0
for anyn ∈ N.
Sin ef (z n , w n ) = (m(z n ), 0),
it is obvious that(m(z n )) ∞ n=1 has no a
umulation
pointsinB.
Conversely, one an he k that every mapping
f
dened in this way isproper.
(b)It iseasy to seethat
C ∋ w 7→ f 2 (z, w) ∈ C
is a proper holomorphi mappingforanyz ∈ A.
Fromtheformofsu hmappingswe on ludethatforevery
z ∈ A
the mappingf 2 (z, ·)
is a omplex polynomial.Nowwe pro eed exa tly as intheproofof (a).( )We pro eedsimilarly to theproofs of (a)and (b).
(d)Supposethat
f : A × C → B × C ∗ isa properholomorphi fun tion.
Fix
z ∈ A.
ThenC ∋ w 7→ f 2 (z, w) ∈ C ∗ isproper.
Take
ψ ∈ O(C)
su h thatf 2 (1, ·) = exp ◦ψ.
Observe thatψ
is a properholomorphi self-mapping of the omplex plane, hen e
ψ
is a polynomial.From thisweeasily geta ontradi tion.
Proof of Theorems 6, 7 and 8. These are dire t onsequen es of Lem-
ma17.
Finally, Itake this opportunityto express mydeep gratitudeto Profes-
sor Wªodzimierz Zwonek for introdu ing me to the subje t and numerous
Referen es
[Edi-Zwo℄ A.EdigarianandW.Zwonek,Properholomorphi mappingsinsome lassof
unboundeddomains,KodaiMathJ.22(1999),305-312.
[Jar-P1℄ M.Jarni kiandP.Pug,InvariantDistan esandMetri sinComplexAnal-
ysis,deGruyter,1993.
[Jar-P2℄ ,,First Steps in SeveralComplex Variables:Reinhardt Domains, Eur.
Math.So .Publ.House,2007,toappear.
[Rud℄ W.Rudin,Fun tionTheoryintheUnitBallof
C n
,GrundlehrenMath.Wiss.241,Springer,1980.
[Shi1℄ S.Shimizu,Holomorphi equivalen eproblemfora ertain lassofunbounded
Reinhardtdomains in
C 2
,OsakaJ.Math.28(1991),609621.[Shi2℄ ,Holomorphi equivalen eproblemfora ertain lassof unbounded Rein-
hardtdomainsin
C 2
,II,KodaiMath.J.15(1992),430444.[Zwo℄ W.Zwonek,Onhyperboli ityofpseudo onvexReinhardtdomains,Ar h.Math.
(Basel)72(1999),304314.
InstytutMatematyki
UniwersytetJagiello«ski
Reymonta4
30-059Kraków,Poland
E-mail:lukasz.kosinskiim.uj.edu.pl
Re eived27.7.2007
andinnalform3.8.2007 (1804)