Download as pdf or txt
Download as pdf or txt
You are on page 1of 13

J. Math. Anal. Appl.

456 (2017) 1049–1061

Contents lists available at ScienceDirect

Journal of Mathematical Analysis and Applications


www.elsevier.com/locate/jmaa

Multiplication operators on the Bergman spaces of polygons


Hansong Huang a,∗ , Dechao Zheng b,c
a
Department of Mathematics, East China University of Science and Technology, Shanghai, 200237,
China
b
Center of Mathematics, Chongqing University, Chongqing, 401331, China
c
Department of Mathematics, Vanderbilt University, Nashville, TN, 37240, United States

a r t i c l e i n f o a b s t r a c t

Article history: In this paper, we will show that multiplication operators on the Bergman space of
Received 19 May 2017 a polygon and some von Neumann algebras induced by these operators have deep
Available online 19 July 2017 connections with the geometry of the polygon.
Submitted by N. Young
© 2017 Elsevier Inc. All rights reserved.
Keywords:
Bergman space
Local inverse
Polygon
Multiplication operator

1. Introduction

Let Ω be a bounded domain in C. Let L2 (Ω) be the Hilbert space of functions on Ω which are square
integrable with respect to the normalized area measure. The Bergman space L2a (Ω) is the subspace of L2 (Ω)
consisting of all holomorphic functions over Ω. Let H ∞ (Ω) be the set of all functions holomorphic on Ω.
For each function h in L∞ (Ω), we define the Toeplitz operator Th with symbol h on L2a (Ω) given by

Th f = P (hf ), f ∈ L2a (Ω),

where P is the orthogonal projection from L2 (Ω) onto L2a (Ω). If h is a bounded and analytic in Ω, Th is
just the multiplication operator

Th (f ) = hf.

Let V ∗ (h, Ω) denote the von Neumann algebra {Th , Th∗ } that consists of all bounded operators on L2a (Ω)
to commute with both Th and Th∗ . Each orthogonal projection in V ∗ (h, Ω) corresponds to a reducing subspace

* Corresponding author.
E-mail addresses: hshuang@ecust.edu.cn (H. Huang), dechao.zheng@vanderbilt.edu (D. Zheng).

http://dx.doi.org/10.1016/j.jmaa.2017.07.026
0022-247X/© 2017 Elsevier Inc. All rights reserved.
1050 H. Huang, D. Zheng / J. Math. Anal. Appl. 456 (2017) 1049–1061

of Th [18]. Thus the problem that Th has no nontrivial reducing subspace is equivalent to that V ∗ (h, Ω) is
trivial, i.e., V ∗ (h, Ω) = CI. Many people have investigated reducing subspaces of Toeplitz operators and
some von Neumann algebras induced by Toeplitz operators ([4,8–11,13,14,19–23,25]).
As mentioned in [5], on the unit disk D, the following theorem was obtained by Cowen and Wahl by
using some ideas in [4] and [22] where contain an analogous remarkable theorem on the Hardy space. More
details on the theorem are contained in [12].

Theorem 1.1. Suppose that φ is in H ∞ (D), and there exists a point λ in D such that the inner part of
φ − φ(λ) is a finite Blaschke product. Then there exist a finite Blaschke product B and an H ∞ -function ψ
such that φ = ψ(B) and {Tφ } = {TB } hold on the Bergman space.

For each λ ∈ D and a non-constant function φ ∈ H ∞ (D), φ − φ(λ) has finitely many zeros on the unit
disk and is continuous on D. Thus the inner part of φ is a finite Blaschke product. By Theorem 1.1, there
exist a finite Blaschke product B and a function ψ ∈ H ∞ (D) such that φ = ψ(B) and {Tφ } = {TB } , and
hence

V ∗ (φ, D) = V ∗ (B, D).

In fact, for each finite Blaschke product B with order greater than one, V ∗ (B, D) is a nontrivial abelian von
Neumann algebra with finite dimension ([7,8,14]).
A natural question is whether an analogue to Theorem 1.1 can be established on general domains Ω.
For a simply-connected domain Ω in C (Ω = C), the Riemann mapping theorem implies that there is
a biholomorphic map σ from Ω to D. One may conjecture that the Riemann map naturally induces an
analogue to Theorem 1.1 on Ω. It is not surprising that this conjecture is true on the domain Ω with
analytic boundary. In fact, if the boundary of Ω is an analytic closed curve, the Riemann map σ and its
inverse map are analytic on ∂Ω and ∂D respectively [2,6], and hence

H ∞ (Ω) ◦ σ −1 = H ∞ (D).

However, if ∂Ω has a cusp point, the Riemann map is possibly not analytic on the boundary ∂Ω, and hence
H ∞ (Ω) ◦σ −1 may not be contained in H ∞ (D). In this paper we will show some unexpected phenomena even
on polygons by using the interplay of geometry of planar domains, complex analysis and operator theory.
An n-gon is a polygon with n sides. Let Σ denote the body of a polygon and Σ the closure of Σ. If h is
holomorphic on some neighborhood of Σ and h (a) = 0 for some point a on Σ, we say that h has a critical
point at a. One of our main results is the following theorem which implies that for a non-equilateral triangle
Σ, the von Neumann algebra V ∗ (h, Σ) is trivial for almost all function h that are holomorphic on some
neighborhood of the closure of Σ.

Theorem 1.2. Suppose Σ is a q-gon for a prime q. If there is a non-constant holomorphic function h on
some neighborhood of Σ, without critical point at vertices of Σ, such that the von Neumann algebra V ∗ (h, Σ)
is nontrivial, then all interior angles of Σ must be equal; that is, the q-gon is equiangular.

Theorem 1.2 reveals some rigidity of polygons if the von Neumann algebra V ∗ (h, Σ) is nontrivial for some
h holomorphic on some neighborhood of Σ. On the other hand, on the unit disk D, for many non-constant
functions h ∈ H ∞ (D), V ∗ (h, D) is nontrivial [14]. It is sharp that q is prime in Theorem 1.2. For example, let
Σ be a rhombus (a quadrilateral with equal edges) with center at 0. Then the unitary operator defined by

U f (z) = f (−z)

gives a nontrivial element in V ∗ (z 2 , Σ) and hence V ∗ (z 2 , Σ) is nontrivial.


H. Huang, D. Zheng / J. Math. Anal. Appl. 456 (2017) 1049–1061 1051

In the case that Σ is a triangle or a quadrilateral, the following theorems say that the symmetry of
geometry of Σ leads to that V ∗ (h, Σ) is nontrivial.

Theorem 1.3. Suppose that Σ is a triangle. Then there is a non-constant holomorphic function h on some
neighborhood of Σ, without critical point at vertices of Σ, such that the von Neumann algebra V ∗ (h, Σ) is
nontrivial if and only if Σ is an equilateral triangle.

Theorem 1.4. Suppose Σ is a quadrilateral. Then there is a non-constant holomorphic function h on some
neighborhood of Σ, without critical point at vertices of Σ, such that the von Neumann algebra V ∗ (h, Σ) is
nontrivial if and only if Σ is a parallelogram.

For two domains Ω1 and Ω2 in C and a proper map ψ : Ω1 → Ω2 , let Z denote the set of critical points
of ψ. It is known that for each point w ∈ Ω2 \ψ(Z), the cardinality of ψ −1 (w) is a constant integer, which
is called the multiplicity of ψ [16]. For example, a finite Blaschke product B is a proper map from D to D,
and its multiplicity is the order of B. The following theorem can be viewed as a regular n-gon-version of
Theorem 1.1.

Theorem 1.5. Suppose Σ is a regular n-gon. Suppose h is holomorphic on some neighborhood of Σ, and h
has no critical point at vertices of Σ. If V ∗ (h, Σ) is nontrivial, then there is an 
h ∈ H ∞ (Σ) and a proper
map ψ : Σ → Σ such that

h=
h ◦ ψ, (1.1)

V ∗ (h, Σ) = V ∗ (ψ, Σ), and the multiplicity of ψ is a factor of n.

Using Theorem 1.5 we have a surprising example on the equilateral triangle Σ centered at zero. Theo-
rem 1.5 gives

V ∗ (z 3 , Σ) = V ∗ (z 3k , Σ), k > 1.

To see this, we observe that there are three unitary operators Uj in V ∗ (z 3 , Σ) defined by

Uj f (z) = f (eij 3 z)

for f in L2a (Σ) and j = 1, 2, 3. Thus the dimension of V ∗ (z 3 , Σ) is at least 3. The above theorem tells us that
there are a function h in H ∞ (Σ) and a proper map ψ from Σ to Σ with multiplicity s such that z 3k = h ◦ ψ
and

V ∗ (z 3k , Σ) = V ∗ (ψ, Σ).

Letting G be the Riemann map from Σ to D, we have that B = G ◦ ψ ◦ G−1 is a Blaschke product of order s,
which is a factor of 3, and V ∗ (ψ, Σ) is isomorphic to V ∗ (B, D). Since the dimension of V ∗ (B, D) ≤ s ≤ 3 [7,9],
the dimension of V ∗ (z 3k , Σ) is less than or equal to 3. On the other hand, V ∗ (z 3k , Σ) contains V ∗ (z 3 , Σ).
This gives that dim V ∗ (z 3k , Σ) ≥ 3. Therefore

V ∗ (z 3k , Σ) = V ∗ (z 3 , Σ).

This does not hold on the unit disk:

V ∗ (z 3 , D) = V ∗ (z 3k , D), k > 1.
1052 H. Huang, D. Zheng / J. Math. Anal. Appl. 456 (2017) 1049–1061

This indicates that multiplication operators on the Bergman space of a polygon are quite different from
multiplication operators on the Bergman space of the unit disk and the geometry of polygons plays an
important role.
This paper is arranged as follows. Section 2 contains some notations and lemmas. In Section 3, we will
present proofs of Theorems 1.2, 1.3 and 1.4. Section 4 contains the proof of Theorem 1.5.

2. Some lemmas

In this section we will introduce local inverses to show some properties of them and obtain some lemmas
which will be used in the proofs of our main results.
Let Ω be a domain of the complex plane and f be a holomorphic function on Ω. If ρ is a continuous map
defined on some sub-domain V of Ω such that ρ(V ) ⊆ Ω and

f (ρ(z)) = f (z), z ∈ V,

then ρ is called a local inverse of f on V ([22]). Clearly, ρ is necessarily holomorphic and the identity map
ρ0 (z) = z is always a trivial local inverse. We are interested in nontrivial local inverses. For example, let
f (z) = z 2 on D. Then both the identity map ρ0 (z) = z and ρ1 (z) = −z are local inverses of f on D. But
for some other functions, their local inverses are only locally defined, and so we need extend analytically
them to a subdomain of D. To do so, we introduce an analytic extension of a locally analytic function. For
two functions f1 analytic on a domain Ω1 and f2 analytic on a domain Ω2 , suppose that the intersection
of Ω1 and Ω2 is nonempty, f1 is called an analytic extension of f2 if f1 and f2 are equal on Ω1 ∩ Ω2 [17,
Chapter 16]. The following useful lemma is contained in [3,22]. We include a proof for completeness.

Lemma 2.1. Let B be a finite Blaschke product. Then each local inverse ρ of B defined on a disk of some point
on ∂D extends analytically to a univalent function on a neighborhood of the unit circle ∂D and ρ(∂D) = ∂D
and there is a local inverse ρ1 on ∂D such that for each local inverse σ on ∂D there is a positive integer
k such that σ = ρ1 k on ∂D. Moreover, if there is a point λ ∈ ∂D such that ρ(λ) = λ, then ρ must be the
identity map on ∂D.

Proof. Suppose that B is a finite Blaschke product and ρ is a local inverse of B defined on a disk of w0
on ∂D. We must show that ρ extends analytically to a function on a neighborhood of the unit circle ∂D.
The following ideas are contained in [8,12]. We will show that a finite Blaschke product B with order n
acts on ∂D just like z n . Since B is a rational function, it is holomorphic on the complex plane except for
finitely many poles outside D. As the order of B is n, we see that for each fixed w0 in ∂D, the equation

B(z) = B(w0 )

has n roots. The Bochner theorem [24] gives that B has no critical point on ∂D, and hence B −1 (B(w0 ))
consists of exactly n different points on the unit circle

w0 , · · · , wn−1 ,

which is ordered in anti-clockwise direction. By the Implicit Function Theorem, there is a neighborhood O
of w0 where lives a unique local inverse ρj of B satisfying

ρj (w0 ) = wj , 0 ≤ j ≤ n − 1.

To choose O to be small enough, we have that ρ equals some ρj on O.


H. Huang, D. Zheng / J. Math. Anal. Appl. 456 (2017) 1049–1061 1053

We will show that all ρj can extend analytically to some neighborhood of ∂D, and such extensions must
be local inverses of B. To do this, write wn = w0 for convenience. For each j(0 ≤ j ≤ n − 1), let γj denote
the arc on ∂D connecting wj with wj+1 in anti-clockwise direction. Since B is conformal on ∂D and B(γj )
is a closed subarc of ∂D, B maps γj \{wj+1 } onto ∂D. Since B|∂D is an n-to-1 map from ∂D to ∂D, letting
Tj be the restriction of B on γj \{wj+1 }, Tj is also injective. Define


⎪ T1−1 ◦ B(z), z ∈ γ0


⎨ T2−1 ◦ B(z), z ∈ γ1
ρ̃1 (z) = ..

⎪ .


⎩ T −1 ◦ B(z), z ∈ γ
n n−1 .

Since ρ̃1 extends to a local inverse of B at each point of ∂D and ∂D is compact, ρ̃1 extends analytically to
a neighborhood of ∂D. Noting that

ρ2 (z) = ρ˜1 ◦ ρ̃1 (z),


ρ3 (z) = ρ̃1 ◦ ρ˜1 ◦ ρ̃1 (z),
..
.
ρn−1 (z) = ρ̃1 ◦ ρ̃1 ◦ . . . ρ̃1 (z)

on some neighborhood of w0 , we have that all ρj extend analytically on a neighborhood of ∂D. For each
local inverse σ, σ is equal to ρk for some k and hence σ = ρ̃k1 .
Since ρ̃1 is bijective from ∂D onto ∂D and univalent on a neighborhood of ∂D, the extension of each ρj
(1 ≤ j ≤ n) is also bijective from ∂D onto ∂D and univalent on a small neighborhood of ∂D.
To prove the last part of the lemma, we note that for any given λ ∈ ∂D, ρ0 (λ), · · · , ρn−1 (λ) on ∂D are n
solutions of the following equation

B(w) = B(λ).

Since B does not have any critical points on ∂D, we have that ρi (λ) = ρj (λ) if i = j. Thus a local inverse
ρ of a finite Blaschke product B on ∂D is uniquely determined by the value ρ(λ). So if ρ(λ) = λ for a local
inverse ρ of B and some λ on ∂D, we conclude that ρ is the identity map. 2

A Jordan curve in the complex plane is the image of a continuous injective map from the unit circle
∂D into C. A Jordan domain is a domain bounded by a Jordan curve. The Riemann mapping theorem
states that there is a conformal map from the unit disk D onto a simply connected domain distinct from the
complex plane; moreover, such a map is unique up to a composition of a function in Aut(D). Caratheodory’s
theorem [1] says that if Ω is a Jordan domain, then the inverse f of Riemann map from D onto Ω extends
to a continuous bijective map F from D onto Ω and the function F maps ∂D bijectively onto ∂Ω. By
Caratheodory’s theorem, we see that if three boundary values of f are assigned, then f is necessarily unique
[17, pp. 290–291].
For two domains Ω1 and Ω2 in the complex plane, an analytic map h : Ω1 → Ω2 is called proper if for
each compact subset E of Ω2 , pre-image h−1 (E) of E under h is also compact. In fact, all proper maps from
D to D are exactly finite Blaschke products [15, p. 164].
Suppose Ω is a Jordan domain. For a point ξ ∈ ∂Ω, if N is a neighborhood of ξ, then N ∩ Ω is called an
inside-neighborhood of ξ. Similarly, an inside-neighborhood of ∂Ω is the intersection of a neighborhood of
∂Ω and Ω. We have the following.
1054 H. Huang, D. Zheng / J. Math. Anal. Appl. 456 (2017) 1049–1061

Proposition 2.2. Suppose that Σ1 and Σ2 are two Jordan domains and φ : Σ1
→ Σ2 is a proper map. Then
each local inverse ρ of φ extends to a continuous bijection from the closure of an inside-neighborhood of ∂Σ1
to another one of ∂Σ1 , and ρ(∂Σ1 ) = ∂Σ1 .

Proof. Suppose that Σ1 and Σ2 are two Jordan domains and φ : Σ1


→ Σ2 is a proper map. By the Riemann
mapping theorem, there are two conformal maps F1 and F2 , which map the unit disk to Σ1 and Σ2 ,
respectively. Since φ : Σ1
→ Σ2 is proper, F2−1 ◦ φ ◦ F1 is a proper map from D onto D. Thus it is a finite
Blaschke product [15]. By Lemma 2.1, each local inverse of F2−1 ◦ φ ◦ F1 is univalent on a neighborhood of
∂D and maps ∂D onto ∂D. Noting that F2−1 ◦ φ ◦ F1 is a finite Blaschke product, we have that its local
inverse maps an inside-neighborhood of ∂D to another one. Caratheodory’s theorem says that F1 is bijective
from ∂D onto Σ1 and F2 is bijective from ∂D onto Σ2 . Using the mapping

τ
→ F1 ◦ τ ◦ F1−1

from the set of local inverses of F2−1 ◦ φ ◦ F1 to the set of local inverses of φ, we obtain that each local inverse
ρ of φ extends to a continuous bijection from the closure of an inside-neighborhood of ∂Σ1 to another of
∂Σ1 , and ρ(∂Σ1 ) = ∂Σ1 . 2

For h ∈ C(∂Ω), let G(h, ∂Ω) denote the group of all continuous maps ρ : ∂Ω → ∂Ω satisfying h ◦ ρ = h.
We have the following.

Lemma 2.3. Suppose Σ is a Jordan domain whose boundary has zero area measure and h is a non-constant
holomorphic function on some neighborhood of Σ. If V ∗ (h, Σ) is nontrivial, then there is a nontrivial local
inverse ρ of h such that ρ is in G(h, ∂Σ), and extends to a continuous bijection from the closure of an
inside-neighborhood of ∂Σ to another one of ∂Σ.

Proof. Suppose V ∗ (h, Σ) is nontrivial and h is not constant. Let F be a conformal mapping from D onto Σ.
Define an operator U from L2a (Σ) to L2a (D) by

U : g
→ g ◦ F F  , g ∈ L2a (Σ).

Then U is a unitary operator from L2a (Σ) onto L2a (D) and has the following property

U Th = Th◦F U.

This immediately gives

V ∗ (h, Σ) = U ∗ V ∗ (h ◦ F, D)U.

Since V ∗ (h, Σ) is nontrivial, we have that V ∗ (h ◦ F, D) is nontrivial. Noting that h ◦ F (∂D) = h(Σ) has
zero area measure and h ◦ F is analytic on D, we can find a point λ in D such that h ◦ F − h ◦ F (λ) has
no zero on ∂D and has only finitely many zeros in D. Since h ◦ F is continuous on D, the inner part of
h ◦ F − h ◦ F (λ) is a finite Blaschke product. This gives that h ◦ F satisfies the assumption in Theorem 1.1.
Thus there is a finite Blaschke product B with order B ≥ 2 such that {Th◦F } = {TB } and a function
φ ∈ H ∞ (D) satisfying

h ◦ F = φ ◦ B.

Letting B = F ◦ B ◦ F −1 we have that B  is a proper map from Σ onto Σ. For a nontrivial local inverse τ
−1  By Proposition 2.2, ρ is a continuous
of B, letting ρ = F ◦ τ ◦ F , then ρ is a nontrivial local inverse of B.
H. Huang, D. Zheng / J. Math. Anal. Appl. 456 (2017) 1049–1061 1055

bijection on ∂Σ and extends to a continuous bijection from the closure of an inside-neighborhood of ∂Σ to


another one of ∂Σ, and ρ(∂Σ) = ∂Σ. This completes the proof. 2

For a function φ analytic at a point a, let mφ (a) denotes the multiplicity of the zero of φ − φ(a) at a;
equivalently, there is a function g analytic at a such that g(a) = 0 and

φ(z) − φ(a) = (z − a)mφ (a) g(z).

Lemma 2.4. Let Σ be a polygon. Suppose that φ is analytic on a neighborhood of ∂Σ. Assume ρ maps the
closure of an inside-neighborhood of ∂Σ bijectively to another one, ρ is a local inverse of φ, and ρ ∈ G(φ, ∂Σ).
Then for each a ∈ ∂Σ,

mφ (a) α = mφ (ρ(a)) β,

where α and β denotes the interior angles at a and ρ(a) respectively.

Proof. Let a be a fixed point on ∂Σ and n = mφ (a). Since φ is analytic a, we can choose a small number
ε > 0 so that

Vε = Σ ∩ {z : |z − a| < ε}

is an angular domain with the angle α and

φ(z) − φ(a) = (z − a)n g(z)

where g is a holomorphic map on {z : |z − a| < ε} with g(a) = 0. We may choose ε to be small enough so
that

φ(z) − φ(a) = gn (z),



where g(z) = (z − a) n g(z) is univalent on {z : |z − a| < ε}. Thus φ maps Vε to an angular domain with
the vertex at φ(a) and with the angle nα. Here as we count multiplicity, it is allowed that nα > 2π.
On the other hand, noting that ρ maps an inside-neighborhood of ∂Σ bijectively to another inside-
neighborhood, we have that ρ maps an inside-neighborhood of a bijectively to an inside-neighborhood
of ρ(a). The above discussion gives that φ(ρ) maps Vε to an angular domain with the angle mφ (ρ(a)) β.
Since φ(ρ) = φ we conclude

mφ (a) α = mφ (ρ(a)) β

to complete the proof. 2

3. Polygons

In this section, we will study the problem when some special von Neumann algebras induced by multi-
plication operators are nontrivial on the Bergman space over a polygon Σ. Under a mild condition we will
show that such von Neumann algebras are all trivial if polygons do not have some symmetries. The proofs
of Theorems 1.2, 1.3, and 1.4 will be presented.
Let a1 , a2 , · · · , an denote the vertices of a polygon in anti-clockwise direction, and let α1 , α2 , · · · , αn be
their interior angles respectively. The order of the vertices gives an orientation of ∂Σ. For any positive
integer q, let [n] denote the element n + qZ in the group Zq ≡ Z/qZ of integers modulo q.
1056 H. Huang, D. Zheng / J. Math. Anal. Appl. 456 (2017) 1049–1061

Lemma 3.1. Suppose that Σ is a q-gon and ρ is a continuous bijection on ∂Σ and keeps the orientation
of ∂Σ. If ρ maps the vertices {aj }qj=1 of ∂Σ to itself, then there exists an integer n in {1, · · · , q} such that
for each j,

ρ(aj ) = a[j+n] .

If q is a prime and ρ does not fix any vertex of ∂Σ, then

aj = ρkj (a1 )

for a positive integer kj .

Proof. Since ρ maps vertices of Σ into vertices of Σ, and ρ is also bijective on ∂Σ, ρ is bijective on the
vertices of Σ. Thus there are nj in {1, 2, · · · , q} such that

ρ(aj ) = anj

j = 1, · · · , q.
Let γj denote the ordered edge of Σ from the vertex aj to the vertex aj+1 . Since ρ is bijective and keeps
the orientation {a1 , a2 , · · · , an , a1 }, ρ(γj ) must be made of ordered edges from anj to anj+1 . Thus ρ(γj ) is
the ordered edge of ∂Σ. So anj+1 = a[nj +1] . By induction, we obtain

ρ(aj ) = a[j+n1 −1] = a[j+n]

where n = n1 − 1.
If n equals 0, then ρ fixes every vertex of ∂Σ. If n does equal 0 and q is prime, then we have

Z/qZ = {[1], [n + 1], [2n + 1], · · · , [(q − 1)n + 1]}.

Thus for each j, we can find kj in {0, 1, · · · , q − 1} such that [j] = [kj n + 1]. So we obtain

aj = ρkj (a1 )

to complete the proof. 2

Proposition 3.2. Suppose that Σ is a q-gon for a prime q. If there is a non-constant holomorphic function h
on some neighborhood of Σ such that the von Neumann algebra V ∗ (h, Σ) is nontrivial, then there is at least
one interior angle of Σ equal to a rational multiple of π.

Proof. Suppose that Σ is a q-gon for a prime q, and there is a non-constant holomorphic function h on
some neighborhood of Σ such that the von Neumann algebra V ∗ (h, Σ) is nontrivial. By Lemma 2.3, there
is a nontrivial local inverse ρ satisfying the condition of Lemma 2.4.
If ρ maps some vertex aj to a point on ∂Σ, which is not a vertex, then by Lemma 2.4, there are two
positive integers m and n satisfying

m αj − n π = 0.

This forces that αj is a rational multiple of π.


If ρ maps vertices of ∂Σ into vertices of ∂Σ, since ρ is bijective on ∂Σ, ρ is bijective on the vertices of
∂Σ. Thus there are nj in {1, 2, · · · , q} such that
H. Huang, D. Zheng / J. Math. Anal. Appl. 456 (2017) 1049–1061 1057

ρ(aj ) = anj

j = 1, · · · , q. Let αj (1 ≤ j ≤ q) be the interior angle at the vertex aj of Σ. Letting mj = mh (aj ), by


Lemma 2.4 we have

mj αj = mnj αnj , 1 ≤ j ≤ q. (3.1)

By Lemma 3.1, we have

ρ(aj ) = a[j+n] , 1 ≤ j ≤ q,

for some integer n, and for each j, we can find kj in {0, 1, · · · , q − 1} such that

aj = ρkj (a1 ).
q
Using Identity (3.1), by induction, we have that αj is a rational multiple of α1 . Since the sum j=1 αj is
(q − 2)π, α1 is a rational multiple of π. This completes the proof. 2

The above proof immediately gives

Proposition 3.3. For a triangle Σ, if there is a non-constant holomorphic function h on some neighborhood
of Σ such that the von Neumann algebra V ∗ (h, Σ) is nontrivial, then each interior angle of Σ is a rational
multiple of π.

Now we are ready to present the proof of Theorem 1.2. We use V (Σ) to denote the set of all vertices
of ∂Σ.

Proof of Theorem 1.2. By Lemma 2.3 there is a nontrivial local inverse ρ of h such that ρ is in G(h, ∂Σ) and
extends to a continuous bijection from the closure of an inside-neighborhood of ∂Σ to another one of ∂Σ.
First we will show that ρ maps vertices V (Σ) onto vertices V (Σ). If this is not true, for some j, ρ(aj ) is not
in {a1 , · · · , aq }. Then the interior angle at ρ(aj ) equals π. Noting that h (aj ) = 0, by Lemma 2.4, we have
that there is a positive integer m satisfying

αj = mπ,

which is a contradiction.
Since h is holomorphic on some neighborhood of Σ and none of {a1 , · · · , aq } is a critical point of h,

mh (aj ) = 1, 1 ≤ j ≤ q.

Lemma 3.1 gives that for each j, we can find kj in {0, 1, · · · , q − 1} such that

aj = ρkj (a1 ).

Then by Lemma 2.4 again, αj equals α1 for each j. This gives

α1 = α2 = · · · = αq ,

to complete the proof. 2


1058 H. Huang, D. Zheng / J. Math. Anal. Appl. 456 (2017) 1049–1061

Proof of Theorem 1.3. Suppose that Σ is a triangle. If there is a non-constant holomorphic function h
on some neighborhood of Σ, without critical point at vertices of Σ, such that the von Neumann algebra
V ∗ (h, Σ) is nontrivial, Theorem 1.2 implies that Σ is an equilateral triangle.
Conversely, without loss of generality, assume that 0 lies in the center of Σ. Let h(z) = z 3 . Then h has a

nontrivial local inverse ρ(z) = ei 3 z. Define the unitary operator U by

U f (z) = f (ei 3 z)

for f in L2a (Σ). Such operator U is a nontrivial element in V ∗ (h, Σ). This completes the proof. 2

Proof of Theorem 1.4. Suppose that Σ is a parallelogram. Without loss of generality, we may assume that
Σ is centered at 0. Since Σ is symmetric about 0, it is invariant under the reflection about 0: ρ(z) = −z.
Then ρ(z) is a nontrivial local inverse of z 2 . Define

U f (z) = f (−z), z ∈ Σ, f ∈ L2a (Σ).

Thus U is a unitary operator commuting with Tz2 , and so U is in V ∗ (h, Σ). This implies that V ∗ (z 2 , Σ) is
nontrivial.
Conversely, suppose that there is a non-constant holomorphic function h on some neighborhood of Σ,
without critical point at vertices of Σ, such that the von Neumann algebra V ∗ (h, Σ) is nontrivial. Lemma 2.3
gives that there is a nontrivial local inverse ρ of h such that ρ extends to a continuous bijection from the
closure of an inside-neighborhood of ∂Σ to another one of ∂Σ. By Lemma 2.4, we have that ρ maps vertices
{a1 , a2 , a3 , a4 } of Σ onto {a1 , a2 , a3 , a4 } and keeps the orientation (in anti-clockwise direction). We have
three cases: ρ(a1 ) = a2 , ρ(a1 ) = a3 , or ρ(a1 ) = a4 .
In the first case that ρ(a1 ) = a2 , Lemma 3.1 gives

ρ(aj ) = aj+1 , j = 1, 2, 3.

By the identity

h ◦ ρ(z) = h(z), z ∈ ∂Σ,

and

h (aj ) = 0, 1 ≤ j ≤ 4,

using Lemma 2.4 we have

α1 = α2 = α3 = α4 .

This gives that Σ is a rectangle, a special form of a parallelogram.


In the second case that ρ(a1 ) = a3 , Lemma 3.1 gives ρ(a2 ) = a4 . By Lemma 2.4, we have

α1 = α3 and α2 = α4 .

Thus Σ is a parallelogram.
In the last case that ρ(a1 ) = a4 , by Lemma 3.1 again we have that ρ(a2 ) = a1 and ρ(a3 ) = a2 . Lemma 2.4
gives

α1 = α2 = α3 = α4 .

This gives that Σ is a rectangle. This completes the proof. 2


H. Huang, D. Zheng / J. Math. Anal. Appl. 456 (2017) 1049–1061 1059

4. Proof of Theorem 1.5

In this section we will present the proof of Theorem 1.5.

Proof of Theorem 1.5. Let Σ be a regular n-gon. Let h be holomorphic on some neighborhood of Σ. Suppose
that no vertex of Σ is a critical point of h and V ∗ (h, Σ) is nontrivial.
Let G be a Riemann mapping from Σ onto D and let φ = h ◦ G−1 . Since h is a non-constant function in
H ∞ (Σ), φ satisfies the assumption in Theorem 1.1. Thus there are a finite Blaschke product B of order s
and a function φ in H ∞ (D) such that

{Tφ } = {TB }

and

φ = φ ◦ B.

In fact, B is completely determined by some local inverses of φ [22]:

s
B= σj ,
j=1

where σj are some local inverses of φ. Let

ψ = G−1 ◦ B ◦ G.

Then ψ is a holomorphic proper map from Σ to Σ, with the multiplicity s. If we let 


h = φ ◦ G, we have

h = φ ◦ G = φ ◦ B ◦ G = (φ ◦ G) ◦ ψ = 
h ◦ ψ.

Using the Riemann map G, we define a unitary operator U from L2a (Σ) to L2a (D) by

U : f
→ f ◦ G−1 (G−1 ) , f ∈ L2a (Σ).

The operator has the following properties:

U Th = Tφ U

and

U TB◦G = TB U. (4.1)

Since {Tφ } = {TB } , we have

{Th } = {TB◦G } = {Tψ } . (4.2)

Combining (4.1) and (4.2) gives

V ∗ (h, Σ) = V ∗ (ψ, Σ) ∼
= V ∗ (B, D).

Since V ∗ (h, Σ) is nontrivial, V ∗ (B, D) is nontrivial. Thus the order s of B must be greater than 1, so B has
a nontrivial local inverse. Lemma 2.1 implies that the group of local inverses of B has a generator τ1 with
1060 H. Huang, D. Zheng / J. Math. Anal. Appl. 456 (2017) 1049–1061

order s. Since the map τ


→ G−1 ◦ τ ◦ G is an isomorphism from the group of local inverses of B to the
group of local inverses of ψ, letting ρ1 = G−1 ◦ τ1 ◦ G, ρ1 is a generator of the group of local inverses of ψ.
Note that the multiplicity of ψ is the minimal positive integer l satisfying ρl1 = id and hence equal to s. To
finish the proof, we need to show that s is a factor of n.
Lemma 2.3 implies that ρ1 is a bijective continuous map from an insider-neighborhood of ∂Σ onto another
one. Since h is holomorphic on some neighborhood of Σ and none of vertices of Σ is a critical point of h, by
Lemma 2.4 we obtain that ρ1 maps vertices {a1 , · · · , an } to vertices {a1 , · · · , an }. Let ρ1 (a1 ) = a1+m for
some m in {1, · · · , n − 1}, and let d = gcd(m, n). Then we have

m = m d, n = n d

for two positive integers m and n , and gcd(m , n ) = 1. Let ρj denote the j-time iteration of ρ1 on ∂Σ.
Then by Lemma 3.1 we have

ρ1 (ak ) = a[k+m] , k = 1, · · · , n.

A computation gives

ρj (ak ) = a[k+jm] , j, k = 1, · · · , n.

Thus ρj (a1 ) = a[1+jm] , j, k = 1, · · · , n, and so Lemma 2.1 gives that ρj is the identity map if and only if
a1 = a[1+jm] , equivalently, 1 + jm = 1 mod n. Hence ρj is the identity map if and only if n is a factor of
j as gcd(m , n ) = 1. Since s is the minimal positive number such that ρj = ρj1 is the identity map, we have
that s is the minimal positive number j such that n divides j. This gives s = n and hence s is a factor
of n. This completes the proof. 2

We give an example.

Example 4.1. Let Σ be a regular n-gon centered at zero and n be a factor of n. Then dim V ∗ (z n , Σ) = n .
 
To see this, note that z n is a holomorphic proper map from Σ to its image and z n is of multiplicity n . By
Theorem 1.5, we have

dim V ∗ (z n , Σ) ≤ n .

Besides, we can define unitary operators Uj in V ∗ (z n , Σ) by

Uj f (z) = f (eij n z), j = 1, · · · , n .


Therefore,

dim V ∗ (z n , Σ) ≥ n ,

and hence dim V ∗ (z n , Σ) = n .

Acknowledgments

This work is partially supported by NSFC (11271387, 11531003, 11471113, 11571064), Chongqing Natu-
ral Science Foundation (cstc 2013jjB0050), Simons Foundation Grant #196300 and CSC (201406745016),
and by Shanghai Center for Mathematical Sciences (11CG30). The first author thanks the Department of
Mathematics, Vanderbilt University for hospitality during his visit.
H. Huang, D. Zheng / J. Math. Anal. Appl. 456 (2017) 1049–1061 1061

References

[1] L. Ahlfors, Complex Analysis, McGraw–Hill Book Co., 1966.


[2] S. Bell, S. Krantz, Smoothness to the boundary of conformal maps, Rocky Mountain J. Math. 17 (1987) 23–40.
[3] G. Cassier, I. Chalendar, The group of the invariants of a finite Blaschke product, Complex Var. Theory Appl. 42 (2000)
193–206.
[4] C. Cowen, The commutant of an analytic Toeplitz operator, Trans. Amer. Math. Soc. 239 (1978) 1–31.
[5] C. Cowen, Commutants of finite Blaschke product multiplication operators on Bergman spaces, http://www.math.iupui.
edu/~ccowen/Talks/Wabash1405.pdf.
[6] S. Choy, S. Pai, On the regularity of the Riemann mapping function in the plane, http://maths.sogang.ac.kr/shcho/
pdf/P18.pdf.
[7] R. Douglas, S. Sun, D. Zheng, Multiplication operators on the Bergman space via analytic continuation, Adv. Math. 226
(2011) 541–583.
[8] R. Douglas, M. Putinar, K. Wang, Reducing subspaces for analytic multipliers of the Bergman space, J. Funct. Anal. 263
(2012) 1744–1765, arXiv:1110.4920v1 [math.FA].
[9] K. Guo, H. Huang, On multiplication operators of the Bergman space: similarity, unitary equivalence and reducing sub-
spaces, J. Operator Theory 65 (2011) 355–378.
[10] K. Guo, H. Huang, Multiplication operators defined by covering maps on the Bergman space: the connection between
operator theory and von Neumann algebras, J. Funct. Anal. 260 (2011) 1219–1255.
[11] K. Guo, H. Huang, Geometric constructions of thin Blaschke products and reducing subspace problem, Proc. Lond. Math.
Soc. 109 (2014) 1050–1091.
[12] K. Guo, H. Huang, Multiplication Operators on the Bergman Space, Lecture Notes in Math., vol. 2145, Springer, Heidel-
berg, 2015.
[13] K. Guo, S. Sun, D. Zheng, C. Zhong, Multiplication operators on the Bergman space via the Hardy space of the bidisk,
J. Reine Angew. Math. 629 (2009) 129–168.
[14] J. Hu, S. Sun, X. Xu, D. Yu, Reducing subspace of analytic Toeplitz operators on the Bergman space, Integral Equations
Operator Theory 49 (2004) 387–395.
[15] W. Rudin, Function Theory in Polydiscs, Benjamin, New York, 1969.
[16] W. Rudin, Function Theory in the Unit Ball of Cn , Grundlehren Math. Wiss., vol. 241, Springer, New York, 1980.
[17] W. Rudin, Real and Complex Analysis, 3rd edition, McGraw–Hill Book Co., New York, 1987.
[18] W. Rudin, Functional Analysis, Int. Ser. Pure Appl. Math., McGraw–Hill, Inc., New York, 1991.
[19] S.L. Sun, Y. Wang, Reducing subspaces of certain analytic Toeplitz operators on the Bergman space, Northeast. Math. J.
14 (1998) 147–158.
[20] S. Sun, D. Zheng, C. Zhong, Classification of reducing subspaces of a class of multiplication operators via the Hardy space
of the bidisk, Canad. J. Math. 62 (2010) 415–438.
[21] S. Sun, D. Zheng, C. Zhong, Multiplication operators on the Bergman space and weighted shifts, J. Operator Theory 59
(2008) 435–452.
[22] J. Thomson, The commutant of a class of analytic Toeplitz operators, Amer. J. Math. 99 (1977) 522–529.
[23] J. Thomson, The commutant of a class of analytic Toeplitz operators II, Indiana Univ. Math. J. 25 (1976) 793–800.
[24] J. Walsh, On the location of the roots of the jacobian of two binary forms, and of the derivative of a rational function,
Trans. Amer. Math. Soc. 19 (1918) 291–298.
[25] K. Zhu, Reducing subspaces for a class of multiplication operators, J. Lond. Math. Soc. 62 (2000) 553–568.

You might also like