, , , 1 December 2004

The ”Edge-of-the-Wedge” type theorems deal with the continuation of holomorphic functions of several complex variables. The first theorem was discovered and proved by N. N. Bogolyubov in 1956 in connection with quantum field theory and dispersion relations. Since then, there is a long list of papers dealing with this theorem and its generalizations under various assumptions (see [16] , [20] and the references therein).

Here we consider a one-sided version of the ”Edge-of-the-Wedge” type theorem in the spirit of the pioneer work of Malgrange–Zerner [21] . Epstein's survey article [3] gives a historical discussion and motivation for this version of an ”Edge-of-the-Wedge” theorem, as well as its natural connections with theorems on separate analyticity.

The first results in this direction are obtained by Komatsu [10] and Drużkowski [2] , but only for some special cases. Recently, Gonchar [5, 6] has proved a remarkable more general result for the one-dimensional case. In a recent work [14] , the authors are able to generalize Gonchar's result to the higher dimensional case.

However, in all these cases the hypotheses on the function to extend and its domain of definition are, in some sense, rather restrictive and strong. Therefore, the main goal of our work is to establish one-sided ”Edge-of-the-Wedge” type theorems in some more general one-dimensional cases with more optimal hypotheses. Perhaps, this will be the first step towards understanding the higher dimensional case and the manifold settings in its full generality.

This paper is organized as follows.

In Section 2 we gather some necessary notions and auxiliary results. This preparation will enable us, at the end of this section, to formulate the above mentioned results of Drużkowski and Gonchar and to discuss in more details the motivation for our work.

The statements of our main results as well as an outline of their proofs are given in Section 3.

The tools which are needed for the proof of the main results are developed in Sections 4, 5, and 8.

The proofs of the main results are given in Sections 6, 7, 9, and 10.

Section 11 gives examples showing the optimality of our results. Finally, we conclude the article with some remarks and open questions.

Our approach is based on our previous work [14] , the Gonchar–Carleman operator developed in [5, 6] , conformal mapping theory, and a thorough geometric study of harmonic measures.

Acknowledgment. The paper was written while the second author was visiting the Carl von Ossietzky Universität Oldenburg being supported by The Alexander von Humboldt Foundation. He wishes to express his gratitude to these organizations.

2 Preliminaries

In order to recall the classical one-sided versions of the ”Edge-of-the-Wedge” theorem and to discuss in more detail our motivation, we need to introduce some notation and terminology. In fact, we keep the main notation from the previous work [14] .

2.1 Planar domains with partly rectifiable boundary

We collect here some classical facts from the books by Goluzin, Koosis and Pommerenke ([4] , [11] , [13] ).

A Jordan curve is the image

C : = {γ (t), t \in [a, b]}

of a continuous one-to-one map

γ : [a, b] ⟶ C,

where

a, b \in R, a < b .

The set

{γ (t), t \in (a, b)}

is said to be the interior of the Jordan curve. A closed Jordan curve is the image

C : = {γ (t), t \in [a, b]}

of a continuous map

γ : [a, b] ⟶ C,

which is one-to-one in

[a, b)

and which satisfies

γ (a) = γ (b) .

The map

γ

is called a parametrization of

C .

Moreover,

C

is said to be rectifiable if

sup {\sum_{k = 0}^{n - 1} | γ (t_{k + 1}) - γ (t_{k}) |} < \infty,

where the supremum is taken over all possible positive integers

n

and sequences of values

t_{0}, \dots, t_{n} \in [a, b]

such that

t_{0} < t_{1} < \dots < t_{n} .

It is easy to see that this supremum is independent of the choice of a parametrization. It is called the length of

C .

A (rectifiable) Jordan domain is a bounded domain in

C

whose boundary is a (rectifiable) closed Jordan curve.

Consider an open set

D \subset C .

Then

D

is said to be locally rectifiable at a point

ζ \in \partial D

if there is a neighborhood

U

ζ

C

such that

U \cap \partial D

is the interior of a rectifiable Jordan curve. Moreover,

D

is said to be locally rectifiable on a subset

A

\partial D

D

is locally rectifiable at all points of

A .

Now let

D \subset C

be an open set which is locally rectifiable on a set

A \subset \partial D .

We consider the induced topology on

\partial D

with respect to the Euclidean topology of

C .

Therefore, there is an open set

V

\partial D

such that

A \subset V

and

D

is locally rectifiable on

V .

Denote by

mes

the linear measure (i.e. the one-dimensional Hausdorff measure) in

C .

Notice that when

A

is a Jordan curve, then the length of

A

coincides with

mes (A) .

We fix the following terminology: A set

A \subset C

is said to be linearly measurable if it is measurable with respect to the linear measure.

Let

ζ \in \partial D

be a point with the following property: There is a closed neighborhood

V

ζ

such that

V \cap \partial D

is a Jordan curve whose interior contains

ζ .

Then

ζ

is said to be of type I if there is a neighborhood

U

ζ

such that

U \cap D

is a Jordan domain.

Otherwise,

ζ

is said to be of type II. We see easily that if

ζ

is of type II, then there are an open neighborhood

U

ζ

and two Jordan domains

U_{1},

U_{2}

such that

U \cap D = U_{1} \cup U_{2} .

Let

ζ \in \partial D

be a point of type I or II.

ζ

is said to be a point where

\partial D

admits a tangent if there are a continuous one-to-one map

γ : [a, b] ⟶ \partial D

and a point

t_{0} \in (a, b)

such that

γ (t_{0}) = ζ

and the following limit exists

{lim}_{t \to t_{0}} \frac{γ (t) - γ (t_{0})}{t - t_{0}} \cdot \frac{| t - t_{0} |}{| γ (t) - γ (t_{0}) |} = λ .

The tangent line of

D

ζ

is, by definition, the real line

L : = {t λ, t \in R} .

In Proposition 5.6 below we shall see that all the above notions are independent of the choice of a parametrization

γ .

Let

\partial^{*} D

denote the set of all points

ζ \in \partial D

where

\partial D

admits a tangent at

ζ .

Given a linearly measurable set

A \subset \partial D

such that

D

is locally rectifiable on

A,

then it is well-known (see [11,p.68–69] ) that

mes (A \cap \partial^{*} D) = mes (A) .

We define the concept of angular approach regions at every point of

ζ \in \partial^{*} D

as follows. Let

L_{ζ}

be the tangent line of

D

ζ

and let

n_{ζ}

be any element of

L_{ζ} \ {0} .

Then, for any

0 < α < \frac{π}{2},

the Stolz region or angular approach region

A_{α} (ζ)

is given by

A_{α} (ζ) : = {z \in D : \frac{π}{2} - α < | arg (\frac{z - ζ}{n_{ζ}}) | < \frac{π}{2} + α},

where

arg : C ⟶ (- π, π]

is as usual the argument function. Geometrically,

A_{α} (ζ)

is the intersection of

D

with two cones of aperture

2 α

and vertex

ζ .

Let

ζ \in \partial^{*} D

and let

U

be an open neighborhood of

ζ .

We say that a function

f

defined on

U \cap D

admits the angular limit

λ

ζ

{lim}_{z \in A_{α} (ζ), z \to ζ} f (z) = λ,

for all

0 < α < \frac{π}{2} .

We conclude this subsection with a simple example which may clarify the above definitions. Let

G

be the open square whose four vertex are

1 + i,

- 1 + i,

- 1 - i,

and

1 - i .

Define the domain

D : = G \ [- \frac{1}{2}, \frac{1}{2}] .

Then

D

is locally rectifiable on

\partial G \cup (- \frac{1}{2}, \frac{1}{2}) .

Every point of

\partial G

is of type I and every point of

(- \frac{1}{2}, \frac{1}{2})

is of type II. Moreover,

\partial^{*} D = (\partial G \ {1 + i, - 1 + i, - 1 - i, 1 - i}) \cup (- \frac{1}{2}, \frac{1}{2}) .

2.2 Harmonic measure for an open set of $C$

Let

D

be a proper open subset of

C \cup {\infty}

such that the boundary

\partial D

(with respect to

C \cup {\infty}

) is non-polar (and

A

a subset of

\partial D .

) Let

A

be a subset of

\partial D .

Consider the characteristic function

1_{\partial D \ A} (ζ) : = {\begin{matrix} 1, & ζ \in \partial D \ A, \\ 0, & ζ \in A . \end{matrix}

Then the harmonic measure of the set

\partial D \ A

(denoted by

ω (\cdot, A, D)

) is the Perron solution of the generalized Dirichlet problem with boundary data

1_{\partial D \ A} .

In other words, one has

ω (\cdot, A, D) : = {sup}_{u \in \hat{U}} u,

where

U = U (A, D)

denotes the family of all subharmonic functions

u

D

such that

{limsup}_{D ∋ z \to ζ} u (z) \leq 1_{\partial D \ A} (ζ)

for each

ζ \in \partial D .

It is well-known (see, for example, the book of Ransford [15] ) that

ω (\cdot, A, D)

is harmonic on

D .

Let

D \subset C

be an open set which is locally rectifiable on a subset

A

\partial D .

We say that a point

ζ \in \partial^{*} D

is a locally regular point relative to

A

{lim}_{z \to ζ, z \in A_{α} (ζ)} ω (z, A \cap U, D \cap U) = 0

for any

0 < α < \frac{π}{2}

and any open neighborhood

U

ζ .

Obviously,

ζ \in \bar{A} .

If, moreover,

ζ \in A,

then

ζ

is said to be a locally regular point of

A .

The set of all locally regular points relative to

A

is denoted by

A^{*} .

Observe that in general

A^{*} ⊄ A,

A ⊄ A^{*} .

However, if

A

is open, then

A \cap \partial^{*} D \subset A^{*} .

As an immediate consequence of the Subordination Principle for the harmonic measure (see Corollary 4.3.9 in [15] ), one gets

\begin{matrix} {lim}_{z \to ζ, z \in A_{α} (ζ)} ω (z, A, D) = 0, ζ \in A^{*}, 0 < α < \frac{π}{2} . \end{matrix}

(2.1)

We extend the function

ω (\cdot, A, D)

D \cup A^{*}

by simply setting

ω (z, A, D) : = 0, z \in A^{*} .

Geometric properties of the harmonic measure will be discussed in Section 5 below.

2.3 Cross and separate holomorphicity

Let

N \in N, N \geq 2,

and let

D_{j}

be a planar domain which is locally rectifiable on a linearly measurable subset

A_{j}

\partial D_{j},

j = 1, \dots, N .

We define an

N

-fold cross

X,

its regular part

X^{*},

its interior

X^{o},

its edge

A

and its regular edge

A^{*}

\begin{matrix} X & : = & X (A_{1}, \dots, A_{N}; D_{1}, \dots, D_{N}) \end{matrix}

\begin{matrix} : = & ^{N} ⋃_{j = 1} A_{1} \times \dots \times A_{j - 1} \times (D_{j} \cup A_{j}) \times A_{j + 1} \times \dots \times A_{N} \subset C^{N}, \end{matrix}

\begin{matrix} X^{*} & : = & X (A_{1}^{*}, \dots, A_{N}^{*}; D_{1}, \dots, D_{N}), \end{matrix}

\begin{matrix} X^{o} & = & X^{o} (A_{1}, \dots, A_{N}; D_{1}, \dots, D_{N}) \end{matrix}

\begin{matrix} : = & ^{N} ⋃_{j = 1} A_{1} \times \dots \times A_{j - 1} \times D_{j} \times A_{j + 1} \times \dots \times A_{N}, \end{matrix}

\begin{matrix} A & : = & A_{1} \times \dots \times A_{N}, A^{*} : = A_{1}^{*} \times \dots \times A_{N}^{*} . \end{matrix}

\begin{matrix}  \end{matrix}

Moreover, put

ω (z) : = \sum_{j = 1}^{N} ω (z_{j}, A_{j}, D_{j}), z = (z_{1}, \dots, z_{N}) \in (D_{1} \cup A_{1}^{*}) \times \dots \times (D_{N} \cup A_{N}^{*}) .

It is clear that

ω |_{D_{1} \times \dots \times D_{N}}

is harmonic.

For an

N

-fold cross

X : = X (A_{1}, \dots, A_{N}; D_{1}, \dots, D_{N})

define its wedge

\hat{X} : = \hat{X} (A_{1}, \dots, A_{N}; D_{1}, \dots, D_{N}) : = {z = (z_{1}, \dots, z_{N}) \in (D_{1} \cup A_{1}^{*}) \times \dots \times (D_{N} \cup A_{N}^{*}) : ω (z) < 1} .

Then the set of all interior points of the wedge

\hat{X}

is given by

{\hat{X}}^{o} : = {\hat{X}}^{o} (A_{1}, \dots, A_{N}; D_{1}, \dots, D_{N}) : = {z = (z_{1}, \dots, z_{N}) \in D_{1} \times \dots \times D_{N} : ω (z) < 1} .

In particular, if

A_{j}

is an open set of

\partial D_{j}

and

A_{j} \subset \partial^{*} D_{j},

j = 1, \dots, N,

one has

A \subset A^{*}

and

X \subset X^{*} \subset \hat{X} .

We say that a function

f : X ⟶ C

is separately holomorphic on

X^{o}

and write

f \in O_{s} (X^{o}),

if for any

j \in {1, \dots, N}

and

(a^{^{'}}, a^{^{''}}) \in (A_{1} \times \dots \times A_{j - 1}) \times (A_{j + 1} \times \dots \times A_{N})

the function

f (a^{^{'}}, \cdot, a^{^{''}}) |_{D_{j}}

is holomorphic on

D_{j} .

We say that a function

f : X ⟶ C

(resp.

f : A ⟶ C

) is separately continuous on

X

(resp. on

A

) and write

f \in C_{s} (X)

(resp.

f \in C_{s} (A)

), if for any

j \in {1, \dots, N}

and

(a^{^{'}}, a^{^{''}}) \in (A_{1} \times \dots \times A_{j - 1}) \times (A_{j + 1} \times \dots \times A_{N})

the function

f (a^{^{'}}, \cdot, a^{^{''}}) |_{(D_{j} \cup A_{j})}

(resp.

f (a^{^{'}}, \cdot, a^{^{''}}) |_{A_{j}}

) is continuous. In the sequel, for a subset

J^{^{'}}

{1, \dots, N},

we write

J^{^{''}} : = {1, \dots, N} \ J^{^{'}} .

Moreover, one often identifies

z \in C^{n}

with

(z^{^{'}}, z^{^{''}}),

where

z^{^{'}} : = (z_{j})_{j \in J^{^{'}}}

and

z^{^{''}} : = (z_{j})_{j \in J^{^{''}}} .

In this paragraph, suppose that

A_{j} \subset A_{j}^{*},

j = 1, \dots, N .

We say that a function

f : \hat{X} \ (A_{1} \times \dots \times A_{N}) ⟶ C

is superholomorphic on

\hat{X}

and write

f \in O (\hat{X}),

if for every proper subset

J^{^{'}}

{1, \dots, N}

(including

\emptyset

) and any

a^{^{'}} : = (a_{j})_{j \in J^{^{'}}} \in \prod_{j \in J^{^{'}}} A_{j},

the restricted function

f_{a^{^{'}}} : {\hat{X}}^{o} (A_{j}, j \in J^{^{''}}; D_{j}, j \in J^{^{''}}) ⟶ C,

given by

f_{a^{^{'}}} (z^{^{''}}) : = f (a^{^{'}}, z^{^{''}}),

where

J^{^{''}} : = {1, \dots, N} \ J^{^{'}},

is holomorphic.

For any

λ \in C

and

r > 0,

let

B (λ, r)

denote the ball with center

λ

with radius

r .

Using identity ( 2.1 ) we make the following observation: For any

a = (a^{^{'}}, a^{^{''}}) \in \hat{X}

with

a^{^{'}} = (a_{j})_{j \in J^{^{'}}} \in \prod_{j \in J^{^{'}}} A_{j}^{*}

and

a^{^{''}} = (a_{j})_{j \in J^{^{''}}} \in \prod_{j \in J^{^{''}}} D_{j}

and any

0 < α < \frac{π}{2},

there is a neighborhood

U

a

and an

ε > 0

such that

U \cap (\prod_{j \in J^{^{'}}} A_{α} (a_{j}) \times \prod_{j \in J^{^{''}}} B (a_{j}, ε)) \subset {\hat{X}}^{o} .

We say that a function

f : {\hat{X}}^{o} ⟶ C

admits an angular limit

λ

a \in \hat{X}

{lim}_{z \to a, z \in \prod_{j \in J^{^{'}}} A_{α} (a_{j}) \times \prod_{j \in J^{^{''}}} B (a_{j}, ε)} f (z) = λ, 0 < α < \frac{π}{2} .

Throughout the paper, for a subset

M

of an Euclidean space,

C (M)

denotes the space of all continuous functions

f : M ⟶ C

equipped with the sup-norm

| f |_{M} : = {sup}_{M} | f | .

Moreover, a function

f : M ⟶ C

is said to be locally bounded on

M

if, for any point

z \in M,

there are an open neighborhood

U

z

(with respect to the induced topology on

M

) and a positive number

K = K_{z}

such that

| f |_{U} < K .

Finally, for an open set

Ω \subset C^{n},

S ℋ (Ω)

(resp.

P S ℋ (Ω),

O (Ω)

) denotes the set of all subharmonic (resp. plurisubharmonic, holomorphic) functions on

Ω .

2.4 Motivations for our work

We are now able to formulate what, in the sequel, we quote as the classical one-sided version of the ”Edge-of-the-Wedge” theorem.

Theorem 1. (Gonchar [5, 6] ) Let

D_{j} \subset C

be a rectifiable Jordan domain and

\emptyset \neq A_{j}

an open set of the boundary

\partial D_{j},

j = 1, \dots, N .

Then, for any function

f \in C (X) \cap O_{s} (X^{o}),

there is a unique function

\hat{f} \in C (\hat{X}) \cap O (\hat{X})

such that

\hat{f} = f

X .

Moreover, if

| f |_{X} < \infty

then

| \hat{f} (z) | \leq | f |_{A}^{1 - ω (z)} | f |_{X}^{ω (z)}, z \in \hat{X},

where

X

X^{o}

, and

\hat{X}

denote the

N

-fold cross, its interior and its wedge, respectively, associated to the

A_{j}

D_{j}

Theorem 1 admits various generalizations. The following theorem is announced by Gonchar in [5] .

Theorem 2. Let

D_{j} \subset C

be a Jordan domain and let

\emptyset \neq A_{j}

be an open set of the boundary

\partial D_{j}

such that

D_{j}

is locally rectifiable on

A_{j},

j = 1, \dots, N .

Let

f

be a function defined on the

N

-fold cross

X

with the following properties:

(i) $f |_{X^{o}} \in C (X^{o}) \cap O_{s} (X^{o});$
(ii) $f$ is locally bounded on $X;$
(iii) for any $j \in {1, \dots, N}$ there is a function $f_{j}$ defined on $A$ such that for any $(a^{^{'}}, a^{^{''}}) \in (A_{1} \times \dots \times A_{j - 1}) \times (A_{j + 1} \times \dots \times A_{N}),$ the holomorphic function $f (a^{^{'}}, \cdot, a^{^{''}}) |_{D_{j}}$ has the angular limit $f_{j} (a^{^{'}}, a_{j}, a^{^{''}})$ at $a_{j}$ for a.e. $a_{j} \in A_{j}$ and $f_{1} = \dots = f_{N} = f$ a.e. on $A .$

1) Then there is a unique function

\hat{f} \in O ({\hat{X}}^{o})

such that

{lim}_{z \in {\hat{X}}^{o}, z \to ζ} \hat{f} (z) = f (ζ), ζ \in X^{o} .

2) If, moreover,

| f |_{X} < \infty,

then

| \hat{f} (z) | \leq | f |_{A}^{1 - ω (z)} | f |_{X}^{ω (z)}, z \in {\hat{X}}^{o} .

3) If, moreover,

f

is continuous at a point

a \in A,

then

{lim}_{z \in {\hat{X}}^{o}, z \to a} \hat{f} (z) = f (a) .

On the other hand, the following result due to Drużkowski [2] gives a different flavor.

Theorem 3. Let

D_{j} \subset C

be a rectifiable Jordan domain and let

\emptyset \neq A_{j}

be an open connected set of the boundary

\partial D_{j},

j = 1, \dots, N .

Let

f

be a function defined on

X

with the following properties:

(i) $f \in C_{s} (X) \cap O_{s} (X^{o});$
(ii) $f$ is locally bounded on $X;$
(iii) $f |_{A}$ is continuous on $A .$

Then all conclusions of Theorem 1 still hold.

Observe that all these theorems require the following very strong hypothesis:

D_{1}, \dots, D_{N}

are rectifiable Jordan domains and the edge

A

is an open set of

\partial D_{1} \times \dots \times \partial D_{N} .

Moreover, the assumptions on the boundedness and continuity of

f

are rather restrictive.

The question naturally arises whether Theorems 1–3 are still true if

D_{1}, \dots, D_{N}

are open sets in

C

and the edge

A

is not necessarily an open set of

\partial D_{1} \times \dots \times \partial D_{N} .

In addition, if one drops the hypothesis on the local boundedness and the continuity of

f,

can one obtain a holomorphic extension of

f

and what are its properties? These matters seem to be of interest especially when one seeks to generalize Theorems 1–3 to higher dimensions.

The present paper is motivated by these questions. Our first purpose is to generalize Gonchar's theorems to a very general situation, where

D_{1}, \dots, D_{N}

are, in some sense, almost general open subsets of

C

and where the boundary sets

A_{1}, \dots, A_{N}

are almost general subsets of

\partial D_{1}, \dots, \partial D_{N} .

Our second goal is to establish, in this general context, an extension theorem analogous to Drużkowski's theorem with a minimum of hypothesis on

f .

3 Statement of the main results and outline of the proofs

We are now ready to state the first main result.

Theorem A. Let

D_{j} \subset C

be an open set and

A_{j}

a linearly measurable subset of

\partial D_{j}

such that

D_{j}

is locally rectifiable on

A_{j},

mes (A_{j}) > 0,

j = 1, \dots, N .

Let

f : X ⟶ C

be such that:

(i) $f$ is locally bounded on $X$ and $f \in O_{s} (X^{o});$
(ii) $f |_{A}$ is measurable with respect to the $N$ -dimensional Hausdorff measure on $A;$
(iii) for any $j \in {1, \dots, N},$ there is a function $f_{j} : A ⟶ C$ such that, for any $(a^{^{'}}, a^{^{''}}) \in (A_{1} \times \dots \times A_{j - 1}) \times (A_{j + 1} \times \dots \times A_{N}),$ the holomorphic function $f (a^{^{'}}, \cdot, a^{^{''}}) |_{D_{j}}$ (see (i)) has the angular limit $f_{j} (a^{^{'}}, a_{j}, a^{^{''}})$ at $a_{j}$ for a.e. $a_{j} \in A_{j}$ and $f_{1} = \dots = f_{N} = f$ a.e. on $A .$

Then there exists a unique function

\hat{f} \in O ({\hat{X}}^{o})

with the following property:

1) There are subsets

{\tilde{A}}_{1} \subset A_{1} \cap A_{1}^{*}, \dots, {\tilde{A}}_{N} \subset A_{N} \cap A_{N}^{*}

such that

1a) $mes (A_{j} \ {\tilde{A}}_{j}) = 0,$ $j = 1, \dots, N;$ ¹
1b) $\hat{f}$ can be extended to a function (still denoted by) $\hat{f} \in O (\hat{X} ({\tilde{A}}_{1}, \dots, {\tilde{A}}_{N}; D_{1}, \dots, D_{N}))$ which admits the angular limit $\hat{f} (a)$ at every point $a \in \hat{X} ({\tilde{A}}_{1}, \dots, {\tilde{A}}_{N}; D_{1}, \dots, D_{N}) \ A;$
1c) $\hat{f} = f$ on $X ({\tilde{A}}_{1}, \dots, {\tilde{A}}_{N}; D_{1}, \dots, D_{N}) \ A .$

In addition,

\hat{f}

enjoys the following properties:

2) If

| f |_{X} < \infty,

then

| \hat{f} (z) | \leq | f |_{A}^{1 - ω (z)} | f |_{X}^{ω (z)}, z \in {\hat{X}}^{o} .

3) For any

j \in {1, \dots, N}

and

a^{0} = (a^{^{'}}, z_{j}^{0}, a^{^{''}}) \in A_{1}^{*} \times \dots \times A_{j - 1}^{*} \times D_{j} \times A_{j + 1}^{*} \times \dots \times A_{N}^{*},

{lim}_{a \to a^{0}, a \in X} f (a)

exists, then

\hat{f}

admits the angular limit

{lim}_{a \to a^{0}, a \in X} f (a)

a^{0} .

4) For any

a^{0} \in A^{*},

{lim}_{a \to a^{0}, a \in A} f (a)

exists, then

\hat{f}

admits the angular limit

{lim}_{a \to a^{0}, a \in A} f (a)

a^{0} .

5) If

f |_{A}

can be extended to a continuous function defined on

A^{*},

then

f

can be extended to a unique continuous function (still denoted by)

f

defined on

X^{*} : = X (A_{1}^{*}, . . ., A_{N}^{*}; D_{1}, . . ., D_{N})

and

\hat{f}

admits the angular limit

f (a)

at every

a \in X^{*}

and

f_{1} = \dots = f_{N} = f

A \cap A^{*} .

Theorem A has an immediate consequence.

Corollary A'. We keep the hypotheses and the notation of Theorem A. Suppose in addition that

f \in C (X^{o}) .

Then there exists a unique function

\hat{f} \in O ({\hat{X}}^{o})

with the following property:

1a') $\hat{f}$ can be extended to a function (still denoted by) $\hat{f} \in O (\hat{X} (A_{1} \cap A_{1}^{*}, \dots, A_{N} \cap A_{N}^{*}; D_{1}, \dots, D_{N})),$ which admits the angular limit $\hat{f} (a)$ at every point $a \in \hat{X} (A_{1} \cap A_{1}^{*}, \dots, A_{N} \cap A_{N}^{*}; D_{1}, \dots, D_{N}) \ A;$
1b') $\hat{f} = f$ on $(X \cap X^{*}) \ A .$

It is worthy to note that Theorem A and Corollary A' generalize, in some sense, Theorems 1–3.

Now we drop the hypothesis on local boundedness and continuity of

f .

Then the examples of Drużkowski in [2] (see Section 11 below) show that, without these conditions, the extended function

\hat{f}

(if it does exist) is, in general, not continuous on

\hat{X} .

However, our second main result gives a partially positive answer to this question.

Theorem B. Let

D_{j} \subset C

be an open set and

A_{j}

a linearly measurable subset of

\partial D_{j}

such that

D_{j}

is locally rectifiable on

A_{j},

mes (A_{j}) > 0,

and

A_{j} \subset \partial^{*} D_{j},

j = 1, \dots, N .

Let

f : X ⟶ C

satisfy the following properties:

(i) $f |_{A} \in C_{s} (A)$ and $f \in O_{s} (X^{o});$
(ii) for any $j \in {1, \dots, N}$ and for any $(a^{^{'}}, a^{^{''}}) \in (A_{1} \times \dots \times A_{j - 1}) \times (A_{j + 1} \times \dots \times A_{N}),$ the function $f (a^{^{'}}, \cdot, a^{^{''}})$ is locally bounded on $D_{j} \cup A_{j}$ and the (holomorphic) restriction function $f (a^{^{'}}, \cdot, a^{^{''}}) |_{D_{j}}$ has the angular limit $f (a^{^{'}}, a_{j}, a^{^{''}})$ at $a_{j}$ for every $a_{j} \in A_{j} .$

Then there are subsets

{\tilde{A}}_{1} \subset A_{1} \cap A_{1}^{*}, \dots, {\tilde{A}}_{N} \subset A_{N} \cap A_{N}^{*},

and a unique function

\hat{f} \in O ({\hat{X}}^{o})

with the following properties:

mes (A_{j} \ {\tilde{A}}_{j}) = 0,

j = 1, \dots, N;

\hat{f}

can be extended to a function (still denoted by)

\hat{f} \in O (\hat{X} ({\tilde{A}}_{1}, \dots, {\tilde{A}}_{N}; D_{1}, \dots, D_{N}))

which admits the angular limit

\hat{f} (a)

at every point

a \in \hat{X} ({\tilde{A}}_{1}, \dots, {\tilde{A}}_{N}; D_{1}, \dots, D_{N});

\hat{f} = f

X ({\tilde{A}}_{1}, \dots, {\tilde{A}}_{N}; D_{1}, \dots, D_{N}) .

Observe that if

f \in C_{s} (X) \cap O_{s} (X^{o}),

then conditions (i)–(ii) above are fulfilled.

Below we give some ideas how to prove Theorems A and B. Our method consists of two steps. In the first step we suppose that each

D_{j}

is a Jordan domain,

j = 1, \dots, N .

In the second one we treat the general case. The key technique here is to use level sets of the harmonic measure. More precisely,we exhaust each

D_{j}

by the level sets of the harmonic measure

ω (\cdot, A_{j}, D_{j}),

i.e. by

D_{j, δ} : = {z_{j} \in D_{j} : ω (z_{j}, A_{j}, D_{j}) < 1 - δ}

(0 < δ < 1) .

In order to carry out the first step, we improve Gonchar's method [5, 6] and make intensive use of Carleman's formula and of geometric properties of the level sets of harmonic measures.

In the second step we apply some mixed cross type theorems (see [14] ) in order to prove Theorems A and B with

D_{j}

replaced by

D_{j, δ} .

Then we construct the solution for the original domains

D_{j}

by means of a gluing procedure.

Although our results have been stated for the general case

N \geq 2,

the proofs will be presented only in the case

N = 2 .

At the end of Section 9 we will give the ideas how to get the general results.

4 Gonchar–Carleman operator

In this section we reformulate a result due to Gonchar [5, 6] to our context. This result will play an important role for the proof of Theorems A and B. Let

D_{j}

be a rectifiable Jordan domain and let

A_{j}

be a linearly measurable subset of

\partial D_{j}

such that

mes (A_{j}) > 0,

j = 1, \dots, N .

Let

f

be a function defined on

X : = X (A_{1}, \dots, A_{N}; D_{1}, \dots, D_{N})

with the following properties:

(i) $f |_{A}$ is measurable;
(ii) $f \in O_{s} (X^{o});$
(iii) there exist a constant $C > 0$ and $N$ functions $f_{j} : A ⟶ C,$ $j = 1, \dots, N,$ such that for any $(a^{^{'}}, a^{^{''}}) \in (A_{1} \times \dots \times A_{j - 1}) \times (A_{j + 1} \times \dots \times A_{N}),$ ${| f (a^{^{'}}, \cdot, a^{^{''}}) |}_{D_{j}} < C,$ and $f (a^{^{'}}, \cdot, a^{^{''}})$ has the angular limit $f_{j} (a^{^{'}}, a_{j}, a^{^{''}})$ at $a_{j}$ for a.e. $a_{j} \in A_{j},$ and $f_{1} = \dots = f_{N} = f$ a.e. on $A .$

For

j = 1, \dots, N,

let

{\tilde{ω}}_{j}

be the conjugate function of

ω (\cdot, A_{j}, D_{j})

such that

{\tilde{ω}}_{j} (z_{j}^{0}) = 0

for a certain fixed point

z_{j}^{0} \in D_{j}

. Therefore, we can define the holomorphic functions

g_{j} (z_{j}) : = ω (z_{j}, A_{j}, D_{j}) + i {\tilde{ω}}_{j} (z_{j}),

j = 1, \dots, N,

and

g (z) : = \sum_{j = 1}^{N} g_{j} (z_{j}), z = (z_{1}, \dots, z_{N}) \in D_{1} \times \dots \times D_{N} .

Each function

e^{- g_{j}}

is bounded on

D_{j},

j = 1, \dots, N .

Therefore, in virtue of [4,p. 439] , we may define

e^{- g_{j} (a_{j})}

for a.e.

a_{j} \in A_{j}

to be the angular boundary limit of

e^{- g_{j}}

a_{j} .

In virtue of (i), for each positive integer

M,

we define the Gonchar–Carleman operator as follows

\begin{matrix} K_{M} (z) = K_{M} [f] (z) : = \frac{1}{(2 π i)^{N}} \int_{A} e^{- M (g (a) - g (z))} \frac{f (a) d a}{a - z}, z \in D_{1} \times \dots \times D_{N}, \end{matrix}

(4.1)

where

d a : = d a_{1} \dots d a_{N},

a - z : = (a_{1} - z_{1}) \dots (a_{N} - z_{N}) .

Moreover, we can extend

K_{M}

(D_{1} \cup A_{1}) \times \dots \times (D_{N} \cup A_{N})

in the following way. Let

z = (z_{1}, \dots, z_{N}) \in (D_{1} \cup A_{1}) \times \dots \times (D_{N} \cup A_{N})

and

J^{^{'}}

the set of

{j : 1 \leq j \leq N, z_{j} \in A_{j}}

. Set

J^{^{''}} : = {1, \dots, N} \ J^{^{'}} .

In the sequel one often identifies

z

with

(z^{^{'}}, z^{^{''}}),

where

z^{^{'}} : = (z_{j})_{j \in J^{^{'}}}

and

z^{^{''}} : = (z_{j})_{j \in J^{^{''}}} .

Then we define

(4.2) K M 1 , … , M N ( z ) = K M 1 , … , M N [ f ] ( z ) : = { f ( z ) , J ′ ′ = ∅ , 1 ( 2 π i ) N ∫ A ′ ′ e − ∑ j ∈ J ′ ′ M j ( g j ( a j ) − g ( z j ) ) f ( z ′ , a ′ ′ ) d a ′ ′ a ′ ′ − z ′ ′ , J ′ ′ ≠ ∅ ,

where

M_{1}, \dots, M_{N} \in N,

A^{^{''}} : = \prod_{j \in J^{^{''}}} A_{j}

d a^{^{''}} : = \prod_{j \in J^{^{''}}} d a_{j}

, and

a^{^{''}} - z^{^{''}} : = \prod_{j \in J^{^{''}}} (a_{j} - z_{j}) .

Moreover, we will write simply

K_{M} (z)

in place of

K_{M, \dots, M} (z)

for any

M \in N .

This notation is in accordance with ( 4.1 ).

The following Carleman Theorem due to Goluzin and Krylov (see, for example, [1,p.2] ) will be very useful.

Theorem 4.1. Let

D

be a rectifiable Jordan domain and

A

a linearly measurable subset of

\partial D

such that

mes (A) > 0 .

Then for any function

f \in H^{\infty} (D),

any relatively compact subset

V ⋐ D

, and any

ε > 0,

there exists a positive integer

M_{0}

(depending only on the sets

D, A, V

and the number

∥ f ∥_{H^{\infty} (D)}

) such that

| f (z) - K_{M} [f |_{A}] (z) | < ε, z \in V, M \geq M_{0},

where

K_{M}

is given by (4.1) for

N = 1

and

f |_{A}

is the angular boundary limit of

f

A .

(Notice that by [4,p.439] ,

f |_{A}

is linearly measurable).

Proof. For every $z \in D,$ applying the Cauchy formula to the function $w \mapsto e^{- M (g (w) - g (z))} f (w) \in H^{\infty} (D),$ $M \in N,$ we obtain $f (z) = \frac{1}{2 π i} \int_{A} e^{- M (g (a) - g (z))} \frac{f (a) d a}{a - z} + \frac{1}{2 π i} \int_{\partial D \ A} e^{- M (g (a) - g (z))} \frac{f (a) d a}{a - z} .$ Since $| e^{- (g (a) - g (z))} | < 1$ for a.e. $a \in \partial D \ A,$ the theorem follows. □

Theorem 4.2. Let

D_{j}

be a rectifiable Jordan domain and

A_{j}

a linearly measurable subset of

\partial D_{j}

such that

mes (A_{j}) > 0,

j = 1, \dots, N .

Let

f

be a function defined on

X

which satisfies (i)–(iii) above. Then 1) the following limit

K (z) = K [f] (z) : = {lim}_{M \to \infty} K_{M} (z)

exists for all

z \in \hat{X} \cap (D_{1} \cup A_{1}) \times \dots \times (D_{N} \cup A_{N}),

and its limit is uniform on compact subsets of

{\hat{X}}^{o};

2) there is a finite constant

C_{0}

such that

| K (z) | \leq \frac{C_{0} C}{\prod_{j \in J^{^{''}}} dist (z_{j}, \partial D_{j}) (1 - e^{- (1 - ω (z))})}, z \in \hat{X} \cap (D_{1} \cup A_{1}) \times \dots \times (D_{N} \cup A_{N}),

where

dist (z_{j}, \partial D_{j}) : = {inf}_{ζ_{j} \in \partial D_{j}} | z_{j} - ζ_{j} |

and

C

is the constant given in (iii) above.

Proof. We briefly recall the argument of Gonchar in [6] . Write the difference

K_{M + 1} - K_{M}

as follows:

\begin{matrix} K_{M + 1} - K_{M} = \sum_{j = 1}^{N} K_{M}^{j}, \end{matrix}

(4.3)

where

K_{M}^{j} : = K_{M + 1, \dots, M + 1, {\underset{︸}{M, \dots, M}}_{j - 1}} - K_{M + 1, \dots, M + 1, {\underset{︸}{M, \dots, M}}_{j}}, j = 1, \dots, N .

Reasoning as in formulas (6)–(9) in [6] and taking (iii) into account, we see that there is a constant

C_{0}

such that

\begin{matrix} | K_{M + 1} (z) - K_{M} (z) | \leq \sum_{j = 1}^{N} | K_{M}^{j} (z) | \leq \frac{C C_{0}}{\prod_{j \in J^{^{''}}} dist (z_{j}, \partial D_{j})} e^{- M (1 - ω (z))} \end{matrix}

(4.4)

for

z \in \hat{X} \cap (D_{1} \cup A_{1}) \times \dots \times (D_{N} \cup A_{N}) .

Since

| e^{- (1 - ω (z))} | < 1,

the desired conclusion of Part 1) follows immediately from this estimate. Finally, using ( 4.4 ), Part 2) follows. □

The following version of Privalov's Uniqueness Theorem will be also needed.

Theorem 4.3. Let

D \subset C

be a domain which is locally rectifiable on a linearly measurable subset

A

\partial D

with

mes (A) > 0 .

Let

f

be a holomorphic function on

D

such that the angular limits of

f

on the set

A \cap \partial^{*} D

are equal to

0 .

Then

f \equiv 0 .

Proof. Using the hypothesis, one may find a rectifiable Jordan domain $U \subset D$ such that $mes (A \cap \partial U) > 0$ and $f |_{U}$ admits the angular limit $0$ at almost every point of $A \cap \partial U .$ By Privalov's Uniqueness Theorem (see [11] ), $f |_{U} \equiv 0 .$ Hence, $f \equiv 0 .$ □

Finally, the following result will play a key role in the proof of Theorems A and B.

Lemma 4.4. Let

D

be a rectifiable Jordan domain,

(f_{n})_{n = 0}^{\infty}

a sequence of holomorphic functions on

D

, and

Δ

a linearly measurable subset of

\partial D

with

mes (Δ) > 0 .

Assume that

f_{n}

admits the angular limit (denoted by)

f_{n} (t)

at every point

t \in Δ,

n \in N

, ² that

{lim}_{n \to \infty} f_{n} (t) = f_{0} (t),

t \in Δ,

and that

{sup}_{n \geq 0} | f_{n} |_{D} < \infty .

Then

(f_{n})_{n = 1}^{\infty}

converges uniformly on every compact subset of

D

f_{0} .

Proof. Let

g (t, Δ, D) : = ω (t, Δ, D) + i \tilde{ω} (t, Δ, D), t \in D,

where

\tilde{ω} (\cdot, Δ, D)

is the conjugate harmonic function of

ω (\cdot, Δ, D) .

Since

g \in O (D)

is bounded, it follows from [4,p.439] that for a.e.

t \in \partial D,

g

admits an angular limit (denoted by

g (t)

) at

t .

Fix an arbitrary compact subset

H

D .

Recall that

| f_{n} |_{D} < M < \infty,

n \in N .

Consequently, applying Theorem 4.1 we see that, for any

ε > 0,

there is an

M_{ε}

such that, for any

M \geq M_{ε},

\begin{matrix} | f_{n} (τ) - \frac{1}{2 π i} \int_{Δ} e^{- M (g (t, Δ, D) - g (τ, Δ, D))} \frac{f_{n} (t) d t}{t - τ} | < ε, τ \in H, n \in N . \end{matrix}

(4.5)

On the other hand, using that

{lim}_{n \to \infty} f_{n} (t) = f_{0} (t),

t \in Δ,

and applying Lebesgue's Dominated Convergence Theorem, we see that

{lim}_{n \to \infty} \frac{1}{2 π i} \int_{Δ} e^{- M (g (t, Δ, D) - g (τ, Δ, D))} \frac{f_{n} (t) d t}{t - τ} = \frac{1}{2 π i} \int_{Δ} e^{- M (g (t, Δ, D) - g (τ, Δ, D))} \frac{f_{0} (t) d t}{t - τ}

for all

τ \in H

and

M > M_{ε} .

This, combined with ( 4.5 ), implies that

{lim}_{n \to \infty} f_{n} (τ) = f_{0} (τ),

τ \in H .

Hence, by the Montel Theorem the sequence

(f_{n})_{n = 1}^{\infty}

converges uniformly on compact subsets of

D

f_{0} .

This completes the proof. □

5 Properties of the harmonic measure and some elements of Conformal mapping theory

In this section we develop the tools needed for the proofs of Theorems A and B. In the sequel,

D \subset C

is an open set and

A

is a linearly measurable nonempty subset of

\partial D

such that

D

is locally rectifiable on

A .

Observe that under the above assumption

\partial D

is non-polar. Let

\partial^{*} D

be the set of all points

ζ \in \partial D

where

\partial D

admits a tangent. Recall from Subsection 2.1 that

mes (A \cap \partial^{*} D) = mes (A) .

Let

P_{D}

be the generalized Poisson integral of

D .

If, in addition,

A

is a Borel set, then, by Theorem 4.3.3 of [15] , the harmonic measure of

\partial D \ A

is given by

\begin{matrix} ω (\cdot, A, D) = P_{D} [1_{\partial D \ A}] . \end{matrix}

(5.1)

Next, let

E

denote the unit disc of

C .

Then, for an

f \in L^{1} (\partial E),

a point

ζ \in \partial E

is said to be a Lebesgue point of

f

{lim}_{r \to 0} \frac{1}{mes (\partial E \cap B (ζ, r))} \int_{\partial E \cap B (ζ, r)} | f (θ) - f (ζ) | d θ = 0,

where

d θ

is the linear measure defined on

\partial E

. If

A

is a linearly measurable subset of

\partial E

and

f : = 1_{A},

then every point of

A

that is a Lebesgue point of

f

is called a density point of

A .

Proposition 5.1. 1) Let

f \in L^{1} (\partial E) .

Then a.e. points of

\partial E

are Lebesgue points of

f

and

{lim}_{z \to ζ, z \in A_{α} (ζ)} P_{E} [f] = f (ζ), 0 < α < \frac{π}{2},

for every Lebesgue point

ζ

f .

2) Let

A

be a subset of positive linear measure of

\partial E

and

A^{^{'}}

the set of all density points of

A .

Then

mes (A \ A^{^{'}}) = 0

and

{lim}_{z \to ζ, z \in A_{α} (ζ)} P_{E} [1_{\partial E \ A}] = 0, 0 < α < \frac{π}{2},

for every

ζ \in A^{^{'}} .

Moreover, if

N

is a subset of

\partial E

with

mes (N) = 0,

then

{lim}_{z \to ζ, z \in E} P_{E} [1_{\partial E \ A}] = 0, 0 < α < \frac{π}{2},

for all interior points

ζ

A \cup N .

Proof. Part 1) is classical (see, for example, Theorem 5.4.8 in [17] ). The first assertion of Part 2) is also classical. Applying Part 1) to the function $f : = 1_{\partial E \ A},$ the second assertion of Part 2) follows. The last assertion is almost trivial. □

We recall the following well-known result due to Carathéodory and F. and M. Riesz (see [4,p.44,p.420] ).

Theorem 5.2. Let

D \subset C

be a Jordan domain and let

Φ

be a conformal mapping from

D

onto the unit disc

E .

1) Then

Φ

extends to a (unique) homeomorphic map (still denoted by)

Φ

from

\bar{D}

onto

\bar{E} .

If, in addition,

D

is a rectifiable Jordan domain, then, for any linearly measurable subset

A

\partial D,

mes (A) > 0

if and only if

mes (Φ (A)) > 0 .

2) Suppose that

γ : [0, 1] ⟶ C

is a one-to-one real analytic map and

U

is an open set such that

γ ([0, 1]) \subset U \cap \partial D,

and

U \cap D

is a Jordan domain. Then

Φ

extends to a conformal mapping from

D \cup γ ([0, 1])

onto

E \cup Φ (γ ([0, 1]))

(i.e.

Φ^{^{'}} (z)

exists and is nonzero for

z \in γ ([0, 1])

Now we are ready to formulate the following

Definition 5.3. Let

D \subset C

be an open set which is locally rectifiable on a linearly measurable subset

A

\partial D .

A point

ζ \in A

is said to be a density point of

A

³ if one of the following cases happens:

Case 1:

ζ

is of type I. There are an open neighborhood

U

ζ

such that

U \cap D

is a Jordan domain and a conformal mapping

Φ

from

U \cap D

onto the unit disc

E

which extends homeomorphically from

\bar{U \cap D}

onto

\bar{E}

such that

Φ (ζ)

is a density point of the set

Φ (\partial (U \cap D) \cap A) .

In other words,

{lim}_{r \to 0} \frac{mes (Φ (\partial (U \cap D) \cap A) \cap B (Φ (ζ), r))}{mes (\partial E \cap B (Φ (ζ), r))} = 1 .

Case 2:

ζ

is of type II. There are an open neighborhood

U

ζ

such that

U \cap D = U_{1} \cup U_{2}

with Jordan domains

U_{1}, U_{2},

and conformal mappings

Φ_{j}

(j = 1, 2)

from

U_{j}

onto

E

which extends homeomorphically from

\bar{U_{j}}

onto

\bar{E}

such that

Φ_{j} (ζ)

is a density point of the set

Φ_{j} (\partial U_{j} \cap A) .

In other words,

{lim}_{r \to 0} \frac{mes (Φ_{j} (\partial U_{j} \cap A) \cap B (Φ_{j} (ζ), r))}{mes (\partial E \cap B (Φ_{j} (ζ), r))} = 1, j = 1, 2 .

Proposition 5.4. Let

D \subset C

be an open set which is locally rectifiable on a linearly measurable subset

A

\partial D .

1) Then Definition 5.3 is independent of the choice of

U

and

Φ

in Case 1 (resp.

U,

Φ_{1},

Φ_{2}

in Case 2).

2) Let

A^{^{'}}

denote the set of all density points of

A .

Then

mes (A \ A^{^{'}}) = 0 .

Proof. To prove Part 1), let

ζ_{0}

be a density point of

A .

We consider two cases.

Case 1:

ζ_{0}

is of type I. In virtue of Definition 5.3 and the assumption that

ζ_{0}

is of type I, for each

j \in {1, 2}

let

U_{j}

be an open neighborhood of

ζ_{0}

such that

U_{j} \cap D

is a Jordan domain and let

Φ_{j}

be a conformal mapping from

U_{j} \cap D_{j}

onto

E

which extends homeomorphically from

\bar{U_{j} \cap D}

onto

\bar{E} .

Suppose that

Φ_{1} (ζ_{0})

is a density point of

Φ_{1} (\partial (U_{1} \cap D) \cap A) \subset \partial E .

We would like to show that

Φ_{2} (ζ_{0})

is a density point of

Φ_{2} (\partial (U_{2} \cap D) \cap A) \subset \partial E .

Let

V

be an open neighborhood of

ζ_{0}

such that

V \subset U_{1} \cap U_{2},

V \cap D

is a rectifiable Jordan domain and

V \cap \partial D

is connected. We deduce that

Φ_{j} (V \cap \partial D)

is an open arc of

\partial E

(which is obviously real analytic) and

Φ_{j} (V \cap D)

is a Jordan subdomain of

E

for

j \in {1, 2} .

Let

Ψ_{1}

be a conformal mapping from the Jordan domain

Φ_{1} (V \cap D)

onto

E

which extends homeomorphically from

\bar{Φ_{1} (V \cap D)}

onto

\bar{E} .

By Part 2) of Theorem 5.2 and the fact that

Φ_{1} (V \cap \partial D)

is an open real analytic arc, we see that

Ψ_{1}

extends conformally to the arc

Φ_{1} (V \cap \partial D) \subset \partial E,

and

(Ψ_{1} \circ Φ_{1}) (V \cap \partial D)

is an arc of

\partial E .

Next, consider the conformal mapping

Ψ_{2} : Φ_{2} (V \cap D) ⟶ E

given by

Ψ_{2} (z) : = (Ψ_{1} \circ Φ_{1} \circ Φ_{2}^{- 1}) (z), z \in Φ_{2} (V \cap D) .

Since by Part 1) of Theorem 5.2

Φ_{2} (V \cap D)

is a Jordan domain,

Ψ_{2}

extends homeomorphically from

\bar{Φ_{2} (V \cap D)}

onto

\bar{E}

and satisfies the equation

Φ_{2} \circ Φ_{1}^{- 1} = Ψ_{2}^{- 1} \circ Ψ_{1} .

Since

Φ_{2} (V \cap \partial D) \subset \partial E

is an open real analytic arc, applying again Part 2) of Theorem 5.2 , we deduce that

Ψ_{2}

extends conformally to the arc

Φ_{2} (V \cap \partial D) \subset \partial E,

and

(Ψ_{2} \circ Φ_{2}) (V \cap \partial D)

is an arc of

\partial E .

We summarize what has been done so far: We have shown that

Ψ_{2}^{- 1} \circ Ψ_{1}

is conformal from the arc

Φ_{1} (V \cap \partial D)

onto the arc

Φ_{2} (V \cap \partial D)

and

Φ_{2} \circ Φ_{1}^{- 1} = Ψ_{2}^{- 1} \circ Ψ_{1} .

Hence,

| (Φ_{2} \circ Φ_{1}^{- 1})^{^{'}} (η) | > 0,

η \in Φ_{1} (V \cap \partial D) .

Choosing

η_{0} : = Φ_{1} (ζ_{0})

and noting that

(Φ_{2} \circ Φ_{1}^{- 1}) (η_{0}) = Φ_{2} (ζ_{0}),

we deduce that

\begin{matrix} {lim}_{r \to 0} \frac{mes (Φ_{2} (\partial (U_{2} \cap D) \cap A) \cap B (Φ_{2} (ζ_{0}), r))}{mes (\partial E \cap B (Φ_{2} (ζ_{0}), r))} \end{matrix}

\begin{matrix} = & {lim}_{r \to 0} \frac{mes (Φ_{2} (\partial (V \cap D) \cap A) \cap B (Φ_{2} (ζ_{0}), r))}{mes (\partial E \cap B (Φ_{2} (ζ_{0}), r))} \end{matrix}

\begin{matrix} = & {lim}_{r \to 0} \frac{mes (Φ_{2} (\partial (V \cap D) \cap A) \cap B (Φ_{2} (ζ_{0}), | (Φ_{2} \circ Φ_{1}^{- 1})^{^{'}} (η_{0}) | r))}{mes (\partial E \cap B (Φ_{2} (ζ_{0}), | (Φ_{2} \circ Φ_{1}^{- 1})^{^{'}} (η_{0}) | r))} \end{matrix}

\begin{matrix} = & {lim}_{r \to 0} \frac{mes (Φ_{1} (\partial (V \cap D) \cap A) \cap B (Φ_{1} (ζ_{0}), r))}{mes (\partial E \cap B (Φ_{1} (ζ_{0}), r))} \end{matrix}

\begin{matrix} = & 1, \end{matrix}

where the last identity follows from the assumption that

Φ_{1} (ζ_{0})

is a density point of

Φ_{1} (\partial (V \cap D) \cap A) \subset \partial E .

Consequently,

Φ_{2} (ζ_{0})

is a density point of

Φ_{2} (\partial (V \cap D) \cap A) \subset \partial E .

Hence the proof of Part 1) is complete in this first case.

Case 2:

ζ_{0}

is of type II. Then there is an open neighborhood

U

ζ_{0}

such that

U \cap D = U^{^{'}} \cup U^{^{''}},

where

U^{^{'}},

U^{^{''}}

are some Jordan domains. We apply the result of Case 1 to each Jordan domain

U^{^{'}},

U^{^{''}} .

Hence the proof of Part 1) is finished.

It remains to prove Part 2). We may find a sequence

(U_{k})_{k = 1}^{\infty}

of open sets of

C

such that

U_{k} \cap D

is either a rectifiable Jordan domain or the disjoint union of two rectifiable Jordan domains and

A \subset \cup_{k = 1}^{\infty} \partial (U_{k} \cap D) .

Using Part 1) and Proposition 5.1 , we see that almost every point in

(A \cap \partial (D \cap U_{k}))

is a density point of this set, i.e. belongs to

(A \cap \partial (D \cap U_{k}))^{^{'}} .

On the other hand, clearly

(A \cap \partial (D \cap U_{k}))^{^{'}} \subset A^{^{'}} .

Consequently,

mes (A \ A^{^{'}}) \leq \sum_{k = 1}^{\infty} mes ((A \cap \partial (D \cap U_{k})) \ (A \cap \partial (D \cap U_{k}))^{^{'}}) = \sum_{k = 1}^{\infty} 0 = 0 .

Hence,

A^{^{'}}

is linearly measurable and

mes (A \ A^{^{'}}) = 0 .

This completes the proof of Part 2). □

The following classical result will be needed to prove Proposition 5.6 below (see [11,p.60] or [13,p.51] ).

Theorem 5.5. (A theorem of Lindelöf ) Let

D

be a Jordan domain and let

Φ

be a conformal mapping from

E

onto

D .

By Part 1) of Theorem 5.2 we still denote by

Φ

its homeomorphic extension from

\bar{E}

onto

\bar{D} .

Let

ζ = e^{i θ} \in \partial E

and let

β : = {lim}_{t \to θ +} arg (Φ (e^{i t}) - Φ (e^{i θ})) .

Then

\partial D

has a corner ⁴ of opening

π α

(0 \leq α \leq 2)

Φ (ζ)

if and only if

arg \frac{Φ (z) - Φ (ζ)}{z - ζ} ⟶ β - α (θ + \frac{π}{2}) as z \to ζ, z \in E .

The next proposition justifies the coherence of the definitions of the tangent line and the angular approach regions given in Subsection 2.1.

Proposition 5.6. Let

D \subset C

be an open set and let

ζ \in \partial^{*} D .

1) Suppose that

ζ

is either of type I or of type II. For

j \in {1, 2},

let

γ_{j} : [a_{j}, b_{j}] ⟶ \partial D

be a parametrization of one Jordan curve contained in

\partial D

such that

γ_{j} (c_{j}) = ζ

c_{j} \in (a_{j}, b_{j})

. Suppose that the following limit exists

{lim}_{t \to c_{1}} \frac{γ_{1} (t) - ζ}{t - c_{1}} \cdot \frac{| t - c_{1} |}{| γ_{1} (t) - ζ |} = λ_{1} .

Then the following limit also exists

{lim}_{t \to c_{2}} \frac{γ_{2} (t) - ζ}{t - c_{2}} \cdot \frac{| t - c_{2} |}{| γ_{2} (t) - ζ |} = λ_{2},

and

λ_{2} = \pm λ_{1} .

In other words, the tangent line at

ζ

is independent of the choice of parametrizations.

2) Suppose that

ζ

is of type I. Suppose that

U

is a neighborhoods of

ζ

such that

D \cap U

is a Jordan domain. Let

Φ

be a conformal mapping from

D \cap U

onto

E

such that

Φ (ζ) = 1,

where

Φ (ζ)

is defined as in Part 1) of Theorem 5.2 . Then for any

0 < α < β < γ < \frac{π}{2},

and any neighborhood

V

ζ,

there are neighborhoods

R \subset V

ζ

and

S

1

such that

Φ (A_{α} (ζ) \cap R) \subset A_{β} (1) \cap S \subset Φ (A_{γ} (ζ) \cap R) .

3) Suppose that

ζ

is of type II. Suppose that

U

is a neighborhood of

ζ

such that

D \cap U = U_{1} \cup U_{2}

with some disjoint Jordan domains

U_{1}, U_{2}

satisfying

ζ \in \partial^{*} U_{1} \cap \partial^{*} U_{2} .

For

j \in {1, 2},

let

Φ_{j}

be a conformal mapping from

D \cap U_{j}

onto

E

such that

Φ_{j} (ζ) = 1 .

Then for any

0 < α < β < γ < \frac{π}{2},

j \in {1, 2},

and any neighborhood

V

ζ,

there are neighborhoods

R \subset V

ζ

and

S

1

such that

Φ_{j} (A_{α} (ζ) \cap U_{j} \cap R) \subset A_{β} (1) \cap S \subset Φ_{j} (A_{γ} (ζ) \cap U_{j} \cap R) .

Proof. First one proves Part 1). Without loss of generality suppose that

γ_{1} ([a_{1}, b_{1}]) = γ_{2} ([a_{2}, b_{2}]) \subset \partial D .

Then the function

γ : = γ_{1}^{- 1} \circ γ_{2} : [a_{2}, b_{2}] ⟶ [a_{1}, b_{1}]

is a bijective continuous map and

γ (c_{2}) = c_{1} .

Therefore, it is either monotone increasing or monotone decreasing. Observe that

\begin{matrix} {lim}_{t \to c_{2}} (\frac{γ_{2} (t) - ζ}{t - c_{2}} \cdot \frac{| t - c_{2} |}{| γ_{2} (t) - ζ |}) = {lim}_{t \to c_{2}} (\frac{(γ_{1} \circ γ) (t) - ζ}{t - c_{2}} \cdot \frac{| t - c_{2} |}{| (γ_{1} \circ γ) (t) - ζ |}) \end{matrix}

\begin{matrix} = & {lim}_{t \to c_{2}} (\frac{(γ_{1} \circ γ) (t) - ζ}{γ (t) - c_{1}} \cdot \frac{| γ (t) - c_{1} |}{| (γ_{1} \circ γ) (t) - ζ |}) \cdot {lim}_{t \to c_{2}} (\frac{γ (t) - c_{1}}{t - c_{2}} \cdot \frac{| t - c_{2} |}{| γ (t) - c_{1} |}) \end{matrix}

\begin{matrix} = & λ_{1} \cdot ε (γ), \end{matrix}

where

ε (γ) : = {\begin{matrix} 1, & if γ is monotone increasing \\ - 1, & if γ is monotone decreasing \end{matrix} .

Hence,

λ_{2} = \pm λ_{1}

and the proof of Part 1) is complete.

To prove Part 2), let

R^{^{'}}

be an open neighborhood of

ζ

such that

R^{^{'}} \subset U \cap V

and

R^{^{'}} \cap D

is a Jordan domain. Then

Φ (R^{^{'}} \cap D)

is a Jordan subdomain of

E .

Let

Ψ

be a conformal mapping from

E

onto the Jordan domain

R^{^{'}} \cap D

such that

Ψ (1) = ζ .

Then

Φ \circ Ψ

is a conformal mapping from

E

onto the Jordan domain

Φ (R^{^{'}} \cap D)

and

(Φ \circ Ψ) (1) = Φ (ζ) .

By Part 1) of Theorem 5.2 , we have that

\begin{matrix} {lim}_{z \to 1, z \in E} Ψ (z) = Ψ (1) and {lim}_{z \to 1, z \in E} (Φ \circ Ψ) (z) = (Φ \circ Ψ) (1) = Φ (ζ) . \end{matrix}

(5.2)

On the other hand, observe that

ζ \in \partial^{*} (R^{^{'}} \cap D)

ζ \in \partial^{*} D .

Moreover,

Φ (ζ) \in \partial^{*} (Φ (R^{^{'}} \cap D))

because

\partial (Φ (R^{^{'}} \cap D))

contains a neighborhood of

Φ (ζ)

\partial E .

Consequently, applying Theorem 5.5 to the two conformal mappings

Ψ

and

Φ \circ Ψ,

we obtain the existence of the following two limits

\begin{matrix} {lim}_{z \to 1, z \in E} arg (\frac{Ψ (z) - Ψ (1)}{z - 1}) and {lim}_{z \to 1, z \in E} arg (\frac{(Φ \circ Ψ) (z) - (Φ \circ Ψ) (1)}{z - 1}), \end{matrix}

(5.3)

where

arg z

denotes the argument of a complex number

z .

Choose an

α^{^{'}}

such that

α < α^{^{'}} < β .

Then using ( 5.2 ) and ( 5.3 ), it follows that there is an open neighborhood

T

1

such that

\begin{matrix} \begin{matrix} A_{α} (ζ) \cap Ψ (T \cap E) & \subset Ψ (A_{α^{^{'}}} (1) \cap T), & (Φ \circ Ψ) (A_{α^{^{'}}} (1) \cap T) & \subset A_{β} (1) \cap (Φ \circ Ψ) (T \cap E) . \end{matrix} \end{matrix}

(5.4)

Let

R

be a neighborhood of

ζ

such that

R \cap D = Ψ (T \cap E)

and

S

a neighborhood of

1

such that

S \cap E = (Φ \circ Ψ) (T \cap E) .

Then we deduce immediately from ( 5.4 ) that

Φ_{j} (A_{α} (ζ) \cap U_{j} \cap R) \subset A_{β} (1) \cap S .

The remaining inclusion of Part 2) can be proved in exactly the same manner.

For Part 3) one argues as in the proof of Part 2). Hence the proof of the proposition is complete. □

The following two lemmas will be very useful.

Lemma 5.7. Let

D \subset C

be a bounded domain with

C^{2}

smooth boundary and let

A

be a linearly measurable subset of

\partial D .

Then

ω (z, A, D) = P_{D} [1_{\partial D \ A}] (z), z \in D .

Proof. Since

D

is bounded with

C^{2}

boundary, using ( 5.1 ) it is a classical fact (see, for example, [19] ) that

\begin{matrix} ω (z, B, D) = P_{D} [1_{\partial D \ B}] (z) : = \int_{\partial D} P (z, ζ) 1_{\partial D \ B} (ζ) d σ (ζ), z \in D, \end{matrix}

(5.5)

for any Borel subset

B

\partial D .

Here

P (\cdot, \cdot)

is the Poisson kernel of

D

and

d σ

is the Lebesgue boundary measure of

\partial D .

Since

A

is linearly measurable, there are two Borel sets

A_{1},

A_{2}

such that

A_{1} \subset A \subset A_{2}

and

mes (A_{2} \ A_{1}) = 0 .

It follows from definition that

\begin{matrix} ω (z, A_{1}, D) \geq ω (z, A, D) \geq ω (z, A_{2}, D), z \in D . \end{matrix}

(5.6)

On the other hand, ( 5.5 ) and the assumption on

A_{1}

and

A_{2}

imply that

ω (z, A_{1}, D) = ω (z, A_{2}, D) = P_{D} [1_{\partial D \ A}] (z), z \in D .

This, combined with ( 5.6 ), gives the desired conclusion. □

Lemma 5.8. Let

E

be the unit disc,

A

a linearly measurable subset of

\partial E,

and

u

a subharmonic function defined on

E

with

u \leq 1

. Let

α \in (0, \frac{π}{2})

be such that

{limsup}_{z \to ζ, z \in A_{α} (ζ)} u (z) \leq 0 for a.e. ζ \in A .

Then

u \leq ω (\cdot, A, E)

E .

Proof. Fix a point

z_{0} \in E .

We wish to show that

\begin{matrix} u (z_{0}) \leq ω (z_{0}, A, E) . \end{matrix}

(5.7)

As in the proof of Lemma 5.7 , we may assume (without loss of generality) that

A

is a Borel set. Moreover, we may suppose that

u

is continuous at

z_{0} .

Otherwise, one considers a Poisson Modification

v

u

given by

v (w) : = {\begin{matrix} u (w), & w \in E \ B (z_{0}, r) \\ P_{B (z_{0}, r)} [u |_{\partial B (z_{0}, r)}], & w \in B (z_{0}, r) \end{matrix} .

where

r > 0

is chosen so that

B (z_{0}, r) ⋐ E .

Then by the hypothesis on

u

and by Lemma 4.1.3 in [15] , we deduce that

v

is subharmonic on

E,

u \leq v \leq 1

E,

and

v |_{B (z_{0}, r)}

is harmonic.

For

0 < r < 1

put

u_{r} (z) : = u (r z),

z \in \bar{E} .

We obtain

\begin{matrix} u (z_{0}) = {lim}_{r \to 1} u (r z_{0}) & \leq & {limsup}_{r \to 1} P_{E} [u_{r}] (z_{0}) \leq P_{E} [{limsup}_{r \to 1} u_{r}] (z_{0}) \end{matrix}

\begin{matrix} \leq & P_{E} [1_{\partial E \ A}] (z_{0}) = ω (z_{0}, A, E), \end{matrix}

where the second estimate holds because of Fatou's Lemma, the third one follows from the hypothesis on

u,

and the last equality is a consequence of ( 5.1 ). This proves ( 5.7 ). Hence, the proof is complete. □

The following estimate will be crucial for the future development.

Proposition 5.9. Let

D \subset C

be an open set which is locally rectifiable on linearly measurable subsets

A,

N

\partial D

with

mes (N) = 0 .

Then 1) all density points of

A

which are elements of

\partial^{*} D

are locally regular points of

A

and

mes (A \ A^{*}) = 0;

2) for every

ζ \in A^{*}

and every open neighborhood

U

ζ,

it holds

{sup}_{0 < α < \frac{π}{2}} {limsup}_{z \to ζ, z \in A_{α} (ζ)} ω (z, A, D) = 0,

and

mes (A \cap U) > 0;

3) for any interior point

ζ

A \cup N,

we have

{lim}_{z \to ζ, z \in D} ω (z, A, D) = 0 .

Proof. To prove Part 1), let

ζ_{0}

be a density point of

A .

There are two cases to consider according to the type of

ζ_{0} .

We only treat the case when

ζ_{0}

is of type I. The second case where

ζ_{0}

is of type

I I

is almost analogous, and therefore, left to the interested reader.

Since

ζ_{0}

is of type I and

D

is rectifiable at

ζ_{0},

there is an open neighborhood

U

ζ_{0}

such that

U \cap D

is a rectifiable Jordan domain. Fix a conformal mapping

Φ

from

U \cap D

onto

E

which extends to a homeomorphism from

\bar{U \cap D}

onto

\bar{E} .

By Definition 5.3 and Proposition 5.4 ,

η_{0} : = Φ (ζ_{0})

is a density point of the linearly measurable subset

Φ (\partial (U \cap D) \cap A)

\partial E .

Consequently, applying identity ( 5.1 ) and Part 1) of Proposition 5.1 yields that

\begin{matrix} {sup}_{0 < α < \frac{π}{2}} {limsup}_{w \in A_{α} (η_{0})} ω (w, Φ (\partial (U \cap D) \cap A), E) = 0 . \end{matrix}

(5.8)

On the other hand, one has

\begin{matrix} ω (z, \partial (U \cap D) \cap A), U \cap D) = ω (Φ (z), Φ (\partial (U \cap D) \cap A), E), z \in U \cap D . \end{matrix}

(5.9)

Applying Theorem 5.5 to

Φ^{- 1}

η_{0}

and combining ( 5.8 ) and ( 5.9 ), it follows that

{sup}_{0 < α < \frac{π}{2}} {limsup}_{z \in A_{α} (ζ_{0})} ω (z, \partial (U \cap D) \cap A, U \cap D) = 0 .

Hence,

ζ_{0} \in A^{*} .

This proves the first assertion of Part 1). Let

A^{^{'}}

denote the set of all density points of

A .

By Part 2) of Proposition 5.4 ,

mes (A \ A^{^{'}}) = 0 .

On the other hand, we have already shown that

A^{^{'}} \cap \partial^{*} D \subset A^{*} .

Consequently,

mes (A \ A^{*}) \leq mes (A \ (A^{^{'}} \cap \partial^{*} D)) = mes (A \ A^{^{'}}) + mes (A \ \partial^{*} D) = 0,

which completes the proof of the last assertion of Part 1).

Part 2) follows from the definition of regular points and the Subordination Principle (see Corollary 4.3.9 in [15] ).

In virtue of Lemma 5.8 and the second assertion of Part 2) of Proposition 5.1 , the proof of Part 3) follows along the same lines as that of Part 2). □

In the sequel we formulate some important stability property of the angular harmonic measure. Recall that

D

is a proper open subset of

C \cup {\infty}

such that the boundary

\partial D

(with respect to

C \cup {\infty}

) is non-polar. Let

A

be a linearly measurable subset of

\partial^{*} D .

Let

φ : \partial D ⟶ R

be a bounded function. The associated Perron function

H_{D, A} : D ⟶ R

is defined by

\begin{matrix} H_{D, A} φ : = {sup}_{u \in \hat{U}} u, \end{matrix}

(5.10)

where

\hat{U} = \hat{U} (φ, A, D)

denotes the family of all subharmonic functions

u

D

such that

\begin{matrix} {limsup}_{z \to ζ} u (z) & \leq & φ (ζ), ζ \in \partial D \ A, \end{matrix}

\begin{matrix} {limsup}_{z \to ζ, z \in A_{α} (ζ)} u (z) & \leq & φ (ζ), ζ \in A, 0 < α < \frac{π}{2} . \end{matrix}

In the sequel,

\hat{U} (A, D)

will stand for

\hat{U} (1_{\partial D \ A}, A, D) .

Proposition 5.10. (Maximum Principle) Let

u \in S ℋ (D)

be such that

u

is bounded from the above and

\begin{matrix} {limsup}_{z \to ζ} u (z) & \leq & 0, ζ \in \partial D \ A, \end{matrix}

\begin{matrix} {limsup}_{z \to ζ, z \in A_{α} (ζ)} u (z) & \leq & 0, ζ \in A, 0 < α < \frac{π}{2} . \end{matrix}

Then

u \leq 0

D .

Proof. Suppose that $u < M$ for some $M .$ Let $ζ_{0}$ be an arbitrary point of $A .$ Fix a rectifiable Jordan domain $U$ such that $U \subset D$ and $\partial U \cap \partial D$ is a neighborhood of $ζ_{0}$ in $\partial D .$ Applying Lemma 5.8 and Part 3) of Proposition 5.9 to $u |_{U}$ yields that ${limsup}_{z \to ζ, z \in U .} u (z) \leq 0 .$ Since $ζ_{0}$ is an arbitrary point of $A$ , the desired conclusion follows from the classical Maximum Principle (see Theorem 2.3.2 in [15] ). □

Using the above proposition, the corresponding results in Sections 4.1 and 4.2 of [15] with respect to

H_{D, A}

(instead of

H_{D}

) are still valid making the obviously necessary changes. In particular, we have the following (see Corollary 4.2.6 in [15] ):

Proposition 5.11. Let

D

be a proper open subset of

C \cup {\infty}

such that the boundary

\partial D

(with respect to

C \cup {\infty}

) is non-polar. Let

A

be a linearly measurable subset of

\partial^{*} D

and

φ : \partial D ⟶ R

a bounded function which is continuous n.e. on

\partial D .

Then there exists a unique bounded harmonic function

h

D

such that

{lim}_{z \to ζ} h (z) = φ (ζ)

for n.e.

ζ \in \partial D .

Moreover,

h = H_{D} φ = H_{D, A} φ .

In virtue of this result, Theorem 4.3.3 in [15] is still valid in the context of

H_{D, A} .

More precisely,

Proposition 5.12. Let

D

be a proper open subset of

C \cup {\infty}

such that

\partial D

(with respect to

C \cup {\infty}

) is non-polar. Let

A

be a linearly measurable subset of

\partial^{*} D

and

φ : \partial D ⟶ R

a bounded Borel function. Then

H_{D} φ = H_{D, A} φ = P_{D} [φ] .

We also need the following

Proposition 5.13. Let

D

be a proper open subset of

C \cup {\infty}

such that

\partial D

(with respect to

C \cup {\infty}

) is non-polar. Let

A

be a Borel subset of

\partial D

such that

A \subset \partial^{*} D

and

mes (A) = 0 .

Then

P_{D} [1_{A}] \equiv 0

D .

Proof. Suppose without loss of generality that $D$ is locally rectifiable on the interval $[0, 1] \subset \partial D$ and that $A$ is a Borel subset of $[0, 1]$ with $mes (A) = 0 .$ Since $D \subset C \ [0, 1],$ it follows from the Subordination Principle that $P_{D} [1_{A}] \leq P_{C \ [0, 1]} [1_{A}] on D .$ Therefore, it suffices to show that $P_{C \ [0, 1]} [1_{A}] \equiv 0$ on $C \ [0, 1] .$ To this end consider the conformal mapping $Φ (z) : = \sqrt{\frac{1}{z} - 1}$ which maps $C \cup {\infty} \ [0, 1]$ onto $H : = {z \in C : Im z > 0} .$ It is not difficult to show that $P_{C \ [0, 1]} [1_{A}] = P_{H} [1_{Φ (A)}] \circ Φ^{- 1} \equiv 0 .$ This concludes the proof. □

Now we arrive at one of the main results of the section

Theorem 5.14. Let

D \subset C

be a proper open subset of

C \cup {\infty}

such that

\partial D

(with respect to

C \cup {\infty}

) is non-polar. Suppose that

D

is locally rectifiable on a linearly measurable subset

A

\partial D .

Let

B

be a linearly measurable subset of

\partial D

with

B \subset A \cap A^{*}

and

mes (A \ B) = 0 .

Then

ω (z, A, D) = H_{D, B} 1_{\partial D \ B}

for

z \in D .

In particular, for every subset

N

\partial D

with

mes (N) = 0,

we have

ω (z, A \ N, D) = ω (z, A, D)

for

z \in D .

Proof. Replacing $A$ by two Borel sets $A_{1}, A_{2}$ such that $A_{1} \subset B$ and $A \subset A_{2} \subset \partial D$ and $mes (B \ A_{1}) = mes (A_{2} \ A) = 0,$ one gets that $mes (A_{2} \ A_{1}) = 0 .$ Then we conclude by the Subordination Principle and Proposition 5.13 that $ω (z, A_{2}, D) \leq ω (z, A, D) \leq ω (z, B, D) \leq ω (z, A_{1}, D) = ω (z, A_{2}, D), z \in D .$ In virtue of Proposition 5.12 , we have that $ω (z, A_{1}, D) = H_{D, A_{1}} 1_{\partial D \ A_{1}} .$ On the other hand, it follows from the definition that $H_{D, A_{1}} 1_{\partial D \ A_{1}} \geq H_{D, B} 1_{\partial D \ B} \geq ω (\cdot, A, D) on D .$ Combining the above three estimates, the proof of the first assertion of the theorem follows.
The second assertion is a direct consequence of the first one. □

Proposition 5.15. Let

D \subset C

be an open set which is locally rectifiable on a linearly measurable subset

A

\partial D .

Let

{(D_{k})}_{k = 1}^{\infty}

be a sequence of open subsets

D_{k}

D

and

{(A_{k})}_{k = 1}^{\infty}

a sequence of measurable subsets of

A

such that

(i) $D_{k} \subset D_{k + 1}$ and $\cup_{k = 1}^{\infty} D_{k} = D;$
(ii) $A_{k} \subset A_{k + 1}$ and $A_{k} \subset \partial D \cap \partial D_{k}$ and $D_{k}$ is locally rectifiable on $A_{k}$ and $\cup_{k = 1}^{\infty} A_{k} = A;$
(iii) for any point $ζ \in A$ there is an open neighborhood $V = V_{ζ}$ of $ζ$ in $C$ such that $V \cap D$ = $V \cap D_{k}$ for some $k .$

Then

ω (z, A, D) = {lim}_{k \to \infty} ω (z, A_{k}, D_{k}), z \in D .

Remark 5.16. We may always choose sequences

{(D_{k})}_{k = 1}^{\infty}

and

{(A_{k})}_{k = 1}^{\infty}

such that (i)–(iii) are satisfied and the open sets

D_{k}

are bounded. For example, take

D_{k} : = D \cap B (0, k)

and

A_{k} : = A \cap B (0, k),

k \geq 1,

k \in N .

Proof. Using the Subordination Principle it is easy to see that the sequence

{(ω (\cdot, A_{k}, D_{k}))}_{k = 1}^{\infty}

is decreasing and the following limit

u : = {lim}_{k \to \infty} ω (\cdot, A_{k}, D_{k})

exists and defines a subharmonic function in

D .

By the Subordination Principle again, we have

u \geq ω (\cdot, A, D) .

Therefore, it remains to establish the converse inequality. In virtue of (iii) and of Part 2) of Proposition 5.9 , we conclude that

\begin{matrix} {sup}_{0 < α < \frac{π}{2}} {limsup}_{z \to ζ, z \in A_{α} (ζ)} u = 0, ζ \in B, \end{matrix}

(5.11)

where

B : = \cup_{k = 1}^{\infty} A_{k}^{*} .

On the other hand, in virtue of Part 1) of Proposition 5.9 , we have that

mes ((A \cap A^{*}) \ B) \leq \sum_{k = 1}^{\infty} mes (A_{k} \ A_{k}^{*}) = 0 .

Consequently, applying Theorem 5.14 , we deduce from ( 5.11 ) that

u (z) \leq ω (z, A, D),

z \in D .

This completes the proof. □

Next, we introduce a notion which will be relevant for our further study.

Definition 5.17. Let

D, G \subset C

be two open sets such that

G \subset D .

A point

ζ \in \partial^{*} D

is said to be an end-point of

G

D

if, for every

0 < α < \frac{π}{2},

there is an open neighborhood

U = U_{α}

ζ

such that

U \cap A_{α} (ζ) \subset G .

The set of all end-points of

G

D

is denoted by

G^{D} .

The remaining part of this section is devoted to the study of level sets of the harmonic measure. We begin with the following important properties of these sets.

Theorem 5.18. Let

D \subset C

be an open set and

A

a linearly measurable subset of

\partial D

such that

D

is locally rectifiable on

A

and

mes (A) > 0 .

Then, for any

0 < ε < 1,

the ”

ε

-level set”

D_{ε} : = {z \in D : ω (z, A, D) < 1 - ε}

enjoys the following properties:

(i) Let $G_{1},$ $G_{2}$ be arbitrary distinct connected components of $D_{ε},$ then $G_{1}^{D} \cap G_{2}^{D} = \emptyset .$
(ii) For any point $ζ \in A^{*},$ there is exactly one connected component $G$ of $D_{ε}$ such that $ζ \in G^{D} .$
(iii) $G^{D} \cap A$ is linearly measurable and $mes (G^{D} \cap A) > 0$ for every connected component $G$ of $D_{ε} .$

Proof. To prove (i), suppose, in order to reach a contradiction, that

G_{1}^{D} \cap G_{2}^{D} \neq \emptyset .

Fix a point

ζ_{0} \in G_{1}^{D} \cap G_{2}^{D} .

Then, for any

0 < α < \frac{π}{2},

there is an open neighborhood

U_{α}

ζ_{0}

such that

A_{α} (ζ) \cap U_{α} \subset G_{1} \cap G_{2} .

This implies that

G_{1} \cap G_{2} \neq \emptyset .

Hence,

G_{1} = G_{2},

which contradicts the hypothesis that

G_{1} \neq G_{2} .

The proof of (i) is complete.

Next, we turn to the proof of (ii). Fix a

ζ_{0} \in A^{*} .

In virtue of assertion (i), it suffices to show the existence of a connected component

G

D_{ε}

such that

ζ_{0} \in G^{D} .

Applying Part 2) of Proposition 5.9 , we see that, for every

0 < α < \frac{π}{2},

there is an open neighborhood

U_{α}

ζ_{0}

such that

\begin{matrix} A_{α} (ζ_{0}) \cap U_{α} \subset D_{ε} . \end{matrix}

(5.12)

Fix an arbitrary

0 < α_{0} < \frac{π}{2},

and let

G

be the connected component of

D_{ε}

containing

A_{α_{0}} (ζ_{0}) \cap U_{α_{0}} .

Since

(A_{α_{0}} (ζ_{0}) \cap U_{α_{0}}) \cap (A_{α} (ζ_{0}) \cap U_{α}) \neq \emptyset, 0 < α < \frac{π}{2},

we deduce from ( 5.12 ) that

G

also contains

A_{α} (ζ_{0}) \cap U_{α}

for every

0 < α < \frac{π}{2} .

Hence

ζ_{0} \in G^{D} .

The proof of (ii) is finished.

Finally, we prove (iii). First, we may find a sequence

(U_{k})_{k = 1}^{\infty}

of open sets of

C

such that

U_{k} \cap D

is either a rectifiable Jordan domain or the disjoint union of two rectifiable Jordan domains and

A \subset \cup_{k = 1}^{\infty} \partial (U_{k} \cap D) .

Since

A

is linearly measurable, we see that in order to prove the measurability of

G^{D} \cap A,

it is sufficient to check that

G^{D} \cap \partial (D \cap U_{k})

is linearly measurable for every

k \geq 1 .

To prove the latter assertion, fix an

k_{0} \geq 1

and let

U : = U_{k_{0}} .

Let

Φ

be a conformal mapping from

D \cap U

onto

E

which extends to a homeomorphic mapping (still denoted by)

Φ

from

\bar{D \cap U}

onto

\bar{E} .

Applying Part 2) of Proposition 5.6 and using Definition 5.17 , we see that, for any

ζ \in \partial (D \cap U),

ζ \in G^{D}

if and only if

Φ (ζ) \in [Φ (G \cap U)]^{E} .

We shall prove, in the sequel, that

[Φ (G \cap U)]^{E}

is a Borel subset of

\partial E .

Taking this for granted, then

G^{D} \cap \partial (D \cap U)

is also a Borel set. Consequently,

G^{D} \cap A

is linearly measurable.

To check that

[Φ (G \cap U)]^{E}

is a Borel set, put

\begin{matrix} A_{n, m} (η) : = {w \in E \cap A_{(1 - \frac{1}{n}) \cdot \frac{π}{2}} (η) : | w - η | < \frac{1}{m}}, n, m \geq 1, η \in \partial E . \end{matrix}

(5.13)

For any

n, m, p \geq 1,

let

(5.14) T n m p : = { η ∈ ∂ E : A n , m ( η ) ⊂ Φ ( G ∩ U ) and ω ( Φ − 1 ( w ) , A , D ) ≤ 1 − ε − 1 p , ∀ w ∈ A n , m ( η ) } .

We observe the following:

Geometric fact. Let

η_{0} \in \partial E

and

(η_{q})_{q = 1}^{\infty} \subset \partial E

such that

{lim}_{q \to \infty} η_{q} = η_{0} .

Then

A_{n, m} (η_{0}) \subset ⋃_{q = 1}^{\infty} A_{n, m} (η_{q}) .

The proof of this fact follows immediately from the geometric shape of the cone

A_{n, m} (η)

given in ( 5.13 ).

Let

(η_{q})_{q = 1}^{\infty} \subset T_{n m p}

such that

{lim}_{q \to \infty} η_{q} = η_{0} \in \partial E .

Using the above geometric fact, we see that

A_{n, m} (η_{0}) \subset Φ (G \cap U) .

This, combined with ( 5.14 ) and the continuityof

ω (Φ^{- 1} (\cdot), A, D) |_{E},

implies that

η_{0} \in T_{n m p} .

Hence, the set

T_{n m p}

is closed. Clearly, we have

[Φ (G \cap U)]^{E} = ⋂_{n = 1}^{\infty} ⋃_{m = 1}^{\infty} ⋃_{p = 1}^{\infty} T_{n m p}

It follows immediately from this identity that

[Φ (G \cap U)]^{E}

is a Borel set.

Consequently, as was already discussed before,

G^{D} \cap A

is linearly measurable.

To finish assertion (iii), it remains to prove that

mes (G^{E} \cap A) > 0 .

Suppose, in order to reach a contradiction, that

mes (G^{E} \cap A) = 0 .

Consider the following function

u (z) : = {\begin{matrix} ω (z, A, D), & z \in D \ G \\ 1 - ε, & z \in G \end{matrix} .

Then clearly

u \in S ℋ (D)

and

u \leq 1 .

In virtue of assertions (i) and (ii) and Part 2) of Proposition 5.9 , we have that

{sup}_{0 < α < \frac{π}{2}} {limsup}_{z \to ζ, z \in A_{α} (ζ)} u (z) = {sup}_{0 < α < \frac{π}{2}} {limsup}_{z \to ζ, z \in A_{α} (ζ)} ω (z, A, E) = 0, ζ \in (A \cap A^{*}) \ (G^{D} \cap A) .

Consequently, using the notation in ( 5.10 ), we conclude that

u \in \hat{U} ((A \cap A^{*}) \ N, D),

where

N : = G^{D} \cap A .

Since, by our above assumption,

mes (N) = 0,

it follows from Theorem 5.14 that

u \leq ω (\cdot, A, D) .

But on the other hand, one has

ω (z, A, D) < 1 - ε = u (z)

for

z \in G .

This leads to the desired contradiction. Hence, the proof of (iii) is finished. □

Theorem 5.19. Let

D \subset C

be an open set and

A

a linearly measurable subset of

\partial D

such that

D

is locally rectifiable on

A

and

mes (A) > 0 .

For any

0 \leq ε < 1,

let

D_{ε} : = {z \in D : ω (z, A, D) < 1 - ε} .

1) For any subset

N

\partial D

such that

mes (N) = 0,

let

U_{ε} (A, N, D) : = {u \in S ℋ (D_{ε}) : u \leq 1 and {sup}_{0 < α < \frac{π}{2}} {limsup}_{z \to ζ, z \in A_{α} (ζ)} u (z) \leq 0, ζ \in (A \cap A^{*}) \ N} .

Then

U_{ε} (A, N, D) = U_{ε} (A, \emptyset, D) .

2) Define the ”harmonic measure of the

ε

-level set”

ω_{ε} (\cdot, A, D)

ω_{ε} (z, A, D) : = {\begin{matrix} {sup}_{u \in U_{ε} (A, \emptyset, D)} u (z), & z \in D_{ε} \\ 0, & z \in A^{*} \end{matrix} .

Then

ω_{ε} (z, A, D) = \frac{ω (z, A, D)}{1 - ε}, z \in D_{ε} \cup A^{*} .

Proof. Clearly, by definition,

U_{ε} (A, \emptyset, D) \subset U_{ε} (A, N, D) .

To prove the converse inclusion, fix an arbitrary

u \in U_{ε} (A, N, D) .

Consider the following function

\hat{u} (z) : = {\begin{matrix} max {(1 - ε) u (z), ω (z, A, D)}, & z \in D_{ε} \\ ω (z, A, D), & z \in D \ D_{ε} \end{matrix} .

Then

\hat{u} \in S ℋ (D)

and

\hat{u} \leq 1 .

Moreover, in virtue of Part 2) of Proposition 5.9 , we have that

A^{*} \subset (D_{ε})^{D} .

Consequently, for every

ζ \in (A \cap A^{*}) \ N,

(5.15) sup 0 < α < π 2 limsup z → ζ , z ∈ A α ( ζ ) u ^ ( z ) ≤ max { sup 0 < α < π 2 limsup z → ζ , z ∈ A α ( ζ ) u ( z ) , sup 0 < α < π 2 limsup z → ζ , z ∈ A α ( ζ ) ω ( z , A , D ) } .

Observe that the first term in the latter line of ( 5.15 ) is equal to

0

because

u \in U_{ε} (A, N, D) .

In addition, in virtue of Part 2) of Proposition 5.9 , the second term in the latter line of ( 5.15 ) is also equal to

0 .

Hence,

\hat{u} \in \hat{U} ((A \cap A^{*}) \ N, D) .

Consequently, by Theorem 5.14 ,

\hat{u} \leq ω (\cdot, A, D) .

In particular, one has

\begin{matrix} u (z) \leq \frac{ω (z, A, D)}{1 - ε}, z \in D, u \in U_{ε} (A, N, D) . \end{matrix}

(5.16)

On the other hand, as an immediate consequence of Part 2) of Proposition 5.9 , we get that

\frac{ω (\cdot, A, D)}{1 - ε} \in U_{ε} (A, \emptyset, D) \subset U_{ε} (A, N, D) .

This, combined with ( 5.16 ), implies the desired conclusions of Part 1) and Part 2). □

An immediate consequence of Theorem 5.19 is the following Two-Constant Theorem for level sets.

Corollary 5.20. Let

D \subset C

be an open set and let

A

and

N

be two linearly measurable subsets of

\partial D

such that

D

is locally rectifiable on

A,

mes (A) > 0

and

mes (N) = 0 .

Let

0 \leq ε < 1

and put

D_{ε} : = {z \in D : ω (z, A, D) < 1 - ε} .

u \in S ℋ (D_{ε})

satisfies

u \leq M

D_{ε}

and

{sup}_{0 < α < \frac{π}{2}} {limsup}_{z \to ζ, z \in A_{α} (ζ)} u (z) \leq m,

ζ \in (A \cap A^{*}) \ N,

then

u (z) \leq m (1 - ω_{ε} (z, A, D)) + M \cdot ω_{ε} (z, A, D) .

6 Boundary behaviour of the Gonchar–Carleman operator

Before investigating the boundary behavior of the Gonchar–Carleman operator, we first introduce the following notion and study its properties.

6.1 Angular Jordan domains

Let

E

be the unit disc. We begin with the

Definition 6.1. For every closed subset

F

\partial E

and any real number

h

such that

mes (F) > 0

and

{sup}_{x, y \in F} | x - y | < h < 1 - \frac{\sqrt{2}}{2},

the open set

Ω = Ω (F, h) : = ⋃_{ζ \in F} {z \in A_{\frac{π}{4}} (ζ) : | z | > 1 - h}

is called the angular Jordan domain with base

F

and height

h .

We give below a list of properties of such angular Jordan domains.

Proposition 6.2. Let

Ω = Ω (F, h)

be an angular Jordan domain.

1) Then there exist exactly two points

ζ_{1}, ζ_{2} \in F

such that

| ζ_{1} - ζ_{2} | = {sup}_{x, y \in F} | x - y |

and

F \subset [ζ_{1}, ζ_{2}],

where

[ζ_{1}, ζ_{2}]

is the (small) closed arc of

\partial E

which is oriented in the positive sense and which starts from

ζ_{1}

and ends at

ζ_{2} .

2) Write the open set

[ζ_{1}, ζ_{2}] \ F

as the union of disjoint open arcs

[ζ_{1}, ζ_{2}] \ F = ⋃_{j \in J} (a_{j}, b_{j}),

where

(a_{j}, b_{j})

is the (small) open arc of

\partial E

which goes from

a_{j}

b_{j}

and which is oriented in the positive sense, and the index set

J

is finite or countable.

For

j \in J,

we construct the isosceles triangle with the three vertices

a_{j},

b_{j}

and

c_{j}

such that the base of the isosceles triangle is the segment connecting

a_{j}

b_{j},

and

c_{j}

satisfies

arg (\frac{c_{j} - a_{j}}{a_{j}}) = \frac{3 π}{4} and arg (\frac{c_{j} - b_{j}}{b_{j}}) = \frac{- 3 π}{4} .

Let

[a_{j}, c_{j}]

(resp.

[c_{j}, b_{j}]

) denote the segment connecting

a_{j}

c_{j}

(resp.

the segment connecting

c_{j}

b_{j}

). Put

F_{0} : = F \cup ⋃_{j \in J} ([a_{j} c_{j}] \cup [c_{j} b_{j}]) .

Then

F_{0}

is a rectifiable Jordan curve starting from

ζ_{1}

and ending at

ζ_{2} .

3) Let

η_{1}

(resp.

η_{2}

) be the unique point in the circle

\partial B (0, 1 - h)

such that

arg (\frac{η_{1} - ζ_{1}}{ζ_{1}}) = \frac{- 3 π}{4} (resp. arg (\frac{η_{2} - ζ_{2}}{ζ_{2}}) = \frac{3 π}{4})

and that

| η_{1} - ζ_{1} |

(resp.

| η_{2} - ζ_{2} |

) is minimal. Let

F_{1}

(resp.

F_{2}

) denote the segment connecting

η_{1}

ζ_{1}

(resp. the segment connecting

ζ_{2}

η_{2}

Let

F_{3}

be the (small) closed arc of the circle

\partial B (0, 1 - h)

which starts from

η_{2}

and ends at

η_{1}

and which is oriented in the negative sense.

Then

Ω

is a rectifiable Jordan domain and its boundary

Γ

consists of the rectifiable Jordan curve

F_{0},

two segments

F_{1},

F_{2}

and the closed arc

F_{3} .

4) For every

ε \in (0, \frac{h}{4})

define the dilatation

τ_{ε} : E ⟶ E

as follows

τ_{ε} (z) : = (1 - ε) z, z \in E .

Put

Ω_{ε} : = τ_{ε} (Ω) \ \bar{B (0, (1 + ε) (1 - h))} .

Then

Ω_{ε}

is a rectifiable Jordan domain and its boundary

Γ_{ε}

consists of the rectifiable Jordan curve

F_{0 ε} : = τ_{ε} (F_{0}),

a sub-segment

F_{1 ε}

τ_{ε} (F_{1}),

a sub-segment

F_{2 ε}

τ_{ε} (F_{2}),

and a closed arc

F_{3 ε}

\partial B (0, (1 + ε) (1 - h)) .

5) Consider the projection

τ : E \ {0} ⟶ \partial E

given by

τ (z) : = \frac{z}{| z |},

z \in E \ {0} .

For every

ε \in (0, \frac{h}{4})

notice that

F_{0 ε} \cup F_{1 ε} \cup F_{2 ε} = Γ_{ε} \ \partial B (0, (1 + ε) (1 - h)) .

Then the two maps

\begin{matrix} F_{0 ε} \cup F_{1 ε} \cup F_{2 ε} ∋ ζ & \mapsto & τ (ζ) \in \partial E, \end{matrix}

\begin{matrix} F_{3 ε} ∋ ζ & \mapsto & τ (ζ) \in \partial E, \end{matrix}

are one-to-one. In addition, for any linearly measurable subset

A

Γ_{ε},

mes (A) \leq 10 \cdot mes (τ (A)) .

Ω_{ε} ↗ Ω

ε ↘ 0 .

7) For any closed Jordan curve

C

contained in

Ω

there is an

ε > 0

such that

C \subset Ω_{ε} .

mes (F \ Ω^{E}) = 0 .

Proof. All assertions are quite simple using an elementary geometric argument.
Therefore, we leave the details of their proofs to the reader. However, we will give the proof of the fact that $Ω$ is a domain. This proof will clarify Definition 6.1 .
In virtue of the condition on $F$ and $h$ given in Definition 6.1 , we see that ${z \in A_{\frac{π}{4}} (ζ) : | z | > 1 - h},$ $ζ \in \partial E,$ is connected, and that
${z \in A_{\frac{π}{4}} (ζ) : | z | > 1 - h} \cap {z \in A_{\frac{π}{4}} (η) : | z | > 1 - h} \neq \emptyset, \forall ζ, η \in \partial E : | ζ - η | < h < 1 - \frac{\sqrt{2}}{2} .$
Hence, $Ω$ is a domain. □

Theorem 6.3. Let

D \subset C

be an open set which is locally rectifiable on a linearly measurable subset

A

\partial D

with

mes (A) > 0 .

Then, for any

0 \leq ε < 1

and any connected component

G

D_{ε} : = {z \in D : ω (z, A, D) < 1 - ε},

there are an open set

U \subset C,

a conformal mapping

Φ

, and an angular Jordan domain

Ω = Ω (F, h)

such that

(i) $U \cap D$ is either a rectifiable Jordan domain or the disjoint union of two rectifiable Jordan domains;
(ii) $Φ$ maps $E$ conformally onto one connected component of $U \cap D$ (notice that, in virtue of (i), $U \cap D$ has at most two connected components);
(iii) $Φ (F) \subset A \cap A^{*} \cap G^{D}$ and $Φ (Ω) \subset G .$

Proof. We have already shown in the proof of (iii) of Theorem 5.18 that there is a sequence

(U_{k})_{k = 1}^{\infty}

of open sets of

C

such that

U_{k} \cap D

is either a rectifiable Jordan domain or the disjoint union of two rectifiable Jordan domains, and

A \subset \cup_{k = 1}^{\infty} \partial (U_{k} \cap D),

and

mes (A \cap A^{*} \cap G^{D}) > 0 .

Consequently, there is an index

k_{0}

such that

\begin{matrix} mes (A \cap A^{*} \cap G^{D} \cap \partial (D \cap U)) > 0, \end{matrix}

(6.1)

where

U : = U_{k_{0}} .

Suppose without loss of generality that

U \cap D

is a rectifiable Jordan domain. The remaining case where

U \cap D

is the disjoint union of two rectifiable Jordan domains may be proved in the same way. Let

Φ

be a conformal mapping from

E

onto

D \cap U .

By Theorem 5.2 ,

Φ

extends to a homeomorphic map (still denoted by)

Φ

from

\bar{E}

onto

\bar{D \cap U} .

Hence, (i) and (ii) are satisfied.

In virtue of Theorem 5.2 , it follows from ( 6.1 ) that

\begin{matrix} mes (Φ^{- 1} (A \cap A^{*} \cap G^{D} \cap \partial (D \cap U))) > 0 . \end{matrix}

(6.2)

For any

m \geq 1,

let

\begin{matrix} A_{m} : = {η \in \partial E : A_{2, m} (η) \subset Φ^{- 1} (G)}, \end{matrix}

(6.3)

where

A_{2, m} (η)

is given by formula ( 5.13 ).

Using the Geometric fact just after ( 5.14 ), we see that

A_{m}

is closed. On the other hand, applying Proposition 5.6 leads to

Φ^{- 1} (A \cap A^{*} \cap G^{D} \cap \partial (D \cap U)) \subset ⋃_{m = 1}^{\infty} A_{m} .

Therefore, in virtue of ( 6.2 ), there is an index

m_{0}

such that

mes (A_{m_{0}} \cap Φ^{- 1} (A \cap A^{*} \cap G^{D} \cap \partial (D \cap U))) > 0 .

Put

h : = \frac{1}{2 m_{0}} .

By the latter estimate one may find a closed set

F

contained in

A_{m_{0}} \cap Φ^{- 1} (A \cap A^{*} \cap G^{D} \cap \partial (D \cap U))

such that

mes (F) > 0

and

{sup}_{x, y \in F} | x - y | < h .

Since

h = \frac{1}{2 m_{0}},

a geometric argument shows that

{z \in A_{\frac{π}{4}} (ζ) : | z | > 1 - h} \subset A_{2, m_{0}} (ζ), ζ \in \partial E .

This together with ( 6.3 ) implies that

Ω = Ω (F, h) \subset Φ^{- 1} (G) .

Hence, (iii) is verified.

This completes the proof. □

In the sequel, the following uniqueness theorem will play a vital role.

Theorem 6.4. Let

D \subset C

be an open set which is locally rectifiable on a linearly measurable subset

A

\partial D

with

mes (A) > 0 .

Let

N \subset \partial D

with

mes (N) = 0 .

Let

0 \leq ε < 1

and

G

a connected component of

D_{ε} : = {z \in D : ω (z, A, D) < 1 - ε} .

f \in O (G)

admits the angular limit

0

at every point of

(A \cap A^{*} \cap G^{D}) \ N,

then

f \equiv 0 .

Proof. Applying Theorem 6.3 we obtain an open set $U$ in $C,$ a conformal mapping $Φ$ from $E$ onto $D \cap U$ which extends homeomorphically to $\bar{E},$ and an angular Jordan domain $Ω : = Ω (F, h)$ satisfying assertions (i)–(iii) listed in that theorem.
Consider the function $f \circ Φ : Ω ⟶ C .$ By the hypothesis and by Proposition 5.6 , $f \circ Φ \in O (Ω)$ admits the angular limit $0$ at a.e point in $F .$ Since $mes (F) > 0,$ Theorem 4.3 gives that $f \circ Φ \equiv 0$ on $Ω .$ Hence, $f \equiv 0$ on the subdomain $Φ (Ω)$ of $G .$ This proves $f \equiv 0 .$ □

6.2 Main result of the section

The boundary behavior of Gonchar–Carleman operator is described below.

Theorem 6.5. Let

D_{j}

be the unit disc

E

and

A_{j}

a linearly measurable subset of

\partial D_{j}

such that

mes (A_{j}) > 0,

j = 1, \dots, N .

Let

f

be a function defined on

X : = X (A_{1}, \dots, A_{N}; D_{1}, \dots, D_{N})

with the following properties:

(i) $f |_{A}$ is measurable;
(ii) $f \in O_{s} (X^{o});$
(iii) there exist a constant $C > 0$ and $N$ functions $f_{j} : A ⟶ C,$ $j = 1, \dots, N,$ such that for any $(a^{^{'}}, a^{^{''}}) \in (A_{1} \times \dots \times A_{j - 1}) \times (A_{j + 1} \times \dots \times A_{N}),$ ${| f (a^{^{'}}, \cdot, a^{^{''}}) |}_{D_{j}} < C$ and $f (a^{^{'}}, \cdot, a^{^{''}})$ has the angular limit $f_{j} (a^{^{'}}, a_{j}, a^{^{''}})$ at $a_{j}$ for a.e. $a_{j} \in A_{j},$ and $f_{1} = \dots = f_{N} = f$ a.e. on $A .$

Let

0 < δ_{1} < 1,

z^{^{'}} = (z_{2}, \dots, z_{N}) \in (D_{2} \cup (A_{2} \cap A_{2}^{*})) \times \dots \times (D_{N} \cup (A_{N} \cap A_{N}^{*}))

such that

\sum_{j = 2}^{N} ω (z_{j}, A_{j}, D_{j}) < δ_{1},

and let

G

be any connected component of

D_{1, δ_{1}} : = {z_{1} \in D_{1} : ω (z_{1}, A_{1}, D_{1}) < 1 - δ_{1}} .

Then there is an angular Jordan domain

Ω = Ω (F, h)

such that

Ω \subset G,

F \subset A_{1} \cap A_{1}^{*} \cap G^{D_{1}},

and the Gonchar–Carleman operator

K [f]

(see formulas ( 4.1 )–( 4.2 )) satisfies

K [f] (a_{1}, z^{^{'}}) = {lim}_{z_{1} \to a_{1}, z_{1} \in A_{α} (a_{1})} K [f] (z_{1}, z^{^{'}}), 0 < α < \frac{π}{2},

for a.e.

a_{1} \in F .

6.3 Preparatory results

For the proof of Theorem 6.5 we need the following results.

In the sequel, for every function

f \in L^{1} (\partial E, | d ζ |),

let

C [f]

denote the Cauchy integral

C [f] (z) : = \frac{1}{2 π i} \int_{\partial E} \frac{f (ζ) d ζ}{z - ζ}, z \in E .

For a function

F : E ⟶ C,

the radial maximal function

M_{rad} F : \partial E \to [0, \infty]

is defined by

(M_{rad} F) (ζ) : = {sup}_{0 \leq r < 1} | F (r ζ) |, ζ \in \partial E .

Now we are able to state the following classical result (see Theorem 6.3.1 in Rudin's book [16] )

Theorem 6.6. (Korányi-Vági type theorem) There is a constant

C > 0

such that (changes)

\int_{\partial E} | M_{rad} C [f] (ζ) |^{2} | d ζ | \leq C \int_{\partial E} | f (ζ) |^{2} | d ζ |

for every

f \in L^{2} (\partial E, | d ζ |) .

We recall the definition of the Smirnov class

E^{p},

p > 0,

on rectifiable Jordan domains.

Definition 6.7. Let

p > 0

and

Ω

a rectifiable Jordan domain. A function

f \in O (Ω)

is said to belong to the Smirnov class

E^{p} (Ω)

if there exists a sequence of rectifiable closed Jordan curves

(C_{n})_{n = 1}^{\infty}

Ω,

tending to the boundary in the sense that

C_{n}

eventually surrounds each compact subdomain of

Ω,

such that

\int_{C_{n}} | f (z) |^{p} | d z | \leq M < \infty, n \geq 1 .

Next, we rephrase some facts concerning the Smirnov class

E^{p},

p > 0

on rectifiable Jordan domains in the context of angular Jordan domains

Ω (F, h) .

Theorem 6.8. 1) Let

Ω

be a rectifiable Jordan domain. Then, for every

f \in E^{p} (Ω),

p > 0,

f

admits the angular limit

f^{*}

a.e. on

\partial Ω .

2) Let

Ω : = Ω (F, h)

be an angular Jordan domain and let

Γ : = \partial Ω .

For any

0 < ε < \frac{h}{4},

let

Γ_{ε}

be the rectifiable closed Jordan curve defined in Part 4) of Proposition 6.2 . Then

f \in E^{p} (Ω)

{sup}_{0 < ε < \frac{h}{4}} \int_{Γ_{ε}} | f (z) |^{p} | d z | < \infty .

In addition, for an

f \in E^{p} (Ω)

p > 0,

it holds that

\int_{Γ} | f^{*} (z) |^{p} | d z | \leq {sup}_{0 < ε < \frac{h}{4}} \int_{Γ_{ε}} | f (z) |^{p} | d z | .

3) Every

f \in E^{1} (E)

has a Cauchy representation

f : = C [f^{*}] .

Conversely, if

g \in L^{1} (\partial E, | d z |)

and

\int_{\partial E} z^{n} g (z) d z = 0, n = 0, 1, 2, \dots,

then

f : = C [g] \in E^{1} (E)

and

g

coincides with

f^{*}

a.e. on

\partial E .

Proof. For the proof of Parts 1) and 3), see [4,p.438–441] . Taking into account Parts 6) and 7) of Proposition 6.2 , Part 2) also follows from the results in [4,p.438–441] .
Hence, the proof is complete. □

6.4 Proof of Theorem 6.5 .

We only give the proof of this theorem in the case

N = 2 .

The proof for the general case is quite similar, therefore, it is left to the interested reader.

We fix an arbitrary

z_{2}^{0} \in D_{2} \cup (A_{2} \cap A_{2}^{*})

0 < δ_{2} < δ_{1}

such that

ω (z_{2}^{0}, A_{2}, D_{2}) < δ_{2},

and an arbitrary connected component

G

D_{1, δ_{1}} : = {z_{1} \in D_{1} : ω (z_{1}, A_{1}, D_{1}) < 1 - δ_{1}} .

Applying Theorem 6.3 , we may find an angular Jordan domain

Ω : = Ω (F, h) \subset G

such that

F \subset A \cap A^{*} \cap G^{D_{1}} .

Without loss of generality we may assume that

z_{2}^{0} \in D_{2},

since the case where

z_{2}^{0} \in A_{2} \cap A_{2}^{*}

is trivial because of (iii). In the course of the proof, the letter

C

will denote a positive constant that is not necessarily the same at each step.

Applying Theorem 4.1 , we have

\begin{matrix} f (z_{1}, a_{2}) & = & {lim}_{M \to \infty} K_{M} [f |_{A_{1} \times {a_{2}}}] (z_{1}), z_{1} \in D_{1}, a_{2} \in A_{2}, \end{matrix}

\begin{matrix} f (a_{1}, a_{2}) & = & {lim}_{r \to 1 -} f (r a_{1}, a_{2}), a_{1} \in \partial D_{1}, a_{2} \in A_{2} . \end{matrix}

Consequently,

f |_{\partial D_{1} \times A_{2}}

is measurable. In addition, by (iii) this function is bounded.

Therefore, for every

M \in N

we are able to define the function

K_{\infty, M} (\cdot, z_{2}^{0}) : \partial D_{1} ⟶ C,

\begin{matrix} K_{\infty, M} (a_{1}, z_{2}^{0}) : = \frac{1}{2 π i} \int_{A_{2}} e^{M (g_{2} (z_{2}^{0}) - g_{2} (a_{2}))} \frac{f (a_{1}, a_{2}) d a_{2}}{a_{2} - z_{2}^{0}}, a_{1} \in \partial D_{1} . \end{matrix}

(6.4)

Since, in virtue of (ii)–(iii),

f (a_{1}, \cdot) \in O (D_{2})

and

| f (a_{1}, \cdot) |_{D_{2}} < C

for

a_{1} \in A_{1},

it follows from Theorem 4.1 that

\begin{matrix} {lim}_{M \to \infty} K_{\infty, M} (a_{1}, z_{2}^{0}) = f (a_{1}, z_{2}^{0}), a_{1} \in A_{1}, \end{matrix}

(6.5)

and the above convergence is uniform with respect to

a_{1} \in A_{1} .

On the other hand, by ( 6.4 ) we see that

K_{\infty, M} (\cdot, z_{2}^{0})

is measurable and bounded.

In addition, for any

n = 0, 1, 2, \dots,

taking (ii) into account, we have that

\int_{\partial D_{1}} K_{\infty, M} (a_{1}, z_{2}^{0}) a_{1}^{n} d a_{1} = \frac{1}{2 π i} \int_{A_{2}} (\int_{\partial D_{1}} f (a_{1}, a_{2}) a_{1}^{n} d a_{1}) \frac{e^{M (g_{2} (z_{2}^{0}) - g_{2} (a_{2}))} d a_{2}}{a_{2} - z_{2}^{0}} = 0,

where the first equality follows from an application of Fubini's Theorem and the second one from an application of Part 3) of Theorem 6.8 to

f (\cdot, a_{2}),

a_{2} \in A_{2} .

Consequently, in virtue of Part 3) of Theorem 6.8 , we can extend

K_{\infty, M} (\cdot, z_{2}^{0})

{\bar{D}}_{1}

by setting

\begin{matrix} K_{\infty, M} (z_{1}, z_{2}^{0}) : = C [K_{\infty, M} (\cdot, z_{2}^{0})] (z_{1}) = \frac{1}{2 π i} \int_{\partial D_{1}} \frac{K_{\infty, M} (a_{1}, z_{2}^{0}) d a_{1}}{a_{1} - z_{1}}, z_{1} \in D_{1} . \end{matrix}

(6.6)

Then the following identity holds

\begin{matrix} {lim}_{z_{1} \to a_{1}, z_{1} \in A_{α} (a_{1})} K_{\infty, M} (z_{1}, z_{2}^{0}) = K_{\infty, M} (a_{1}, z_{2}^{0}), 0 < α < \frac{π}{2}, \end{matrix}

(6.7)

for a.e.

a_{1} \in \partial D_{1} .

Now we come back to the angular Jordan domain

Ω

. We keep the notation introduced in Proposition 6.2 . Put

K : = K [f]

and

K_{M, M} : = K_{M, M} [f |_{A}]

(see formulas ( 4.1 )–( 4.2 )). For any

0 < ε < \frac{h}{4}

and any

z_{1} \in Γ_{ε},

applying the Cauchy integral formula, we obtain

(6.8) K ∞ , M ( z 1 , z 2 0 ) − K M , M ( z 1 , z 2 0 ) = 1 ( 2 π i ) 2 ∫ ∂ D 1 \ A 1 ∫ A 2 e M ( g 1 ( z 1 ) − g 1 ( t 1 ) ) + M ( g 2 ( z 2 0 ) − g 2 ( t 2 ) ) f ( t 1 , t 2 ) d t 1 d t 2 ( t 1 − z 1 ) ( t 2 − z 2 0 ) = e M ( g 1 ( z 1 ) − ( 1 − δ 1 ) ) ∫ ∂ D 1 p N ( t 1 ) d t 1 t 1 − z 1 .

Using the choice of

G

and the hypothesis on

δ_{1}

and

δ_{2},

it can be checked that

\begin{matrix} | e^{M (g_{1} (\cdot) - (1 - δ_{1}))} |_{G} \leq 1, | p_{N} |_{\partial D_{1}} \leq C e^{- M (δ_{1} - δ_{2})} . \end{matrix}

(6.9)

Therefore, recalling the projection

τ : E \ {0} ⟶ \partial E

is defined in Part 5) of Proposition 6.2 , we estimate

(6.10) ∫ Γ ε | K ∞ , M ( z 1 , z 2 0 ) − K M , M ( z 1 , z 2 0 ) | 2 | d z 1 | ≤ C ∫ Γ ε | M rad C [ p N ] ( τ ( z 1 ) ) | 2 | d z 1 | ≤ 10 C ∫ τ ( F 0 ε ∪ F 1 ε ∪ F 2 ε ) | M rad C [ p N ] ( a 1 ) | 2 | d a 1 | + 10 C ∫ τ ( F 3 ε ) | M rad C [ p N ] ( a 1 ) | 2 | d a 1 | ≤ 20 C ∫ ∂ E | M rad C [ p N ] ( a 1 ) | 2 | d a 1 | ≤ C ∫ ∂ E | p N ( a 1 ) | 2 | d a 1 | ≤ C e − M ( δ 1 − δ 2 ) ,

where the first estimate follows from ( 6.8 )–( 6.9 ) and the definition of the radial maximal function, the second and the third one from Part 5) of Proposition 6.2 , the fourth estimate holds by an application of Theorem 6.6 , and the last one follows from ( 6.9 ).

On the other hand, for any

0 < ε < \frac{h}{4},

(6.11) ∫ Γ ε | K M + 1 , M + 1 ( z 1 , z 2 0 ) − K M , M ( z 1 , z 2 0 ) | 2 | d z 1 | ≤ 2 ∫ Γ ε | K M 1 ( z 1 , z 2 0 ) | 2 | d z 1 | + 2 ∫ Γ ε | K M 2 ( z 1 , z 2 0 ) | 2 | d z 1 | ≤ C e − M δ 0 ,

where the latter estimate follows from the same argument as in the proof of ( 6.8 )–( 6.10 ). We recall from Part 1) of Theorem 4.2 that

{lim}_{M \to \infty} K_{M, M} (z_{1}, z_{2}^{0}) = K (z_{1}, z_{2}^{0}), z_{1} \in Γ_{ε},

and that the convergence is uniform with respect to

z_{1} \in Γ_{ε} .

This, combined with ( 6.10 )–( 6.11 ), implies that

\begin{matrix} \int_{Γ_{ε}} | K_{\infty, M} (z_{1}, z_{2}^{0}) - K (z_{1}, z_{2}^{0}) |^{2} | d z_{1} | \leq C \cdot e^{- M (δ_{1} - δ_{2})}, 0 < ε < \frac{h}{4} . \end{matrix}

(6.12)

Since we have already shown that

| K_{\infty, M} (\cdot, z_{2}^{0}) |_{D_{1}} < \infty,

in virtue of Part 2) of Theorem 6.8 , we deduce from ( 6.12 ) that

K (\cdot, z_{2}^{0}) |_{Ω} \in E^{2} (Ω) .

For every

a_{1} \in \partial D_{1},

let

K (a_{1}, z_{2}^{0})

denote the angular limit of

K (\cdot, z_{2}^{0}) |_{Ω}

a_{1}

(if the limit exists). It follows from ( 6.12 ) and Part 2) of Theorem 6.8 that

{lim}_{M \to \infty} \int_{Γ} | K_{\infty, M} (a_{1}, z_{2}^{0}) - K (a_{1}, z_{2}^{0}) |^{2} | d a_{1} | \leq {sup}_{0 < ε < \frac{h}{4}} \int_{Γ_{ε}} | K_{\infty, M} (z_{1}, z_{2}^{0}) - K (z_{1}, z_{2}^{0}) |^{2} | d z_{1} | \leq {lim}_{M \to \infty} C \cdot e^{- M (δ_{1} - δ_{2})} = 0 .

This, combined with ( 6.5 ) and Part 8) of Proposition 6.2 , implies finally that

K (a_{1}, z_{2}^{0}) = f (a_{1}, z_{2}^{0}), for a.e. a_{1} \in F .

Hence, Theorem 6.5 has been proved.

□

7 Proof of Theorem A in a special case

In this section, we prove Theorem A under the following hypotheses:

D_{1}, \dots, D_{N}

are rectifiable Jordan domains and

| f |_{X} < \infty .

(*)

We first prove that Theorem A under the above hypotheses may be reduced to the case when

D_{1} = \dots = D_{N} = E .

Indeed, let

Φ_{j}

be a conformal mapping from

D_{j}

onto

E

which extends to a homeomorphic map (still denoted by)

Φ

from

\bar{D_{j}}

onto

\bar{E}

j = 1, \dots, N .

Consider the function

\tilde{f}

given by

\tilde{f} (w) : = f (Φ_{1}^{- 1} (w_{1}), \dots, Φ_{N}^{- 1} (w_{N})), w = (w_{1}, \dots, w_{N}) \in X (Φ_{1} (A_{1}), \dots, Φ_{N} (A_{N}); E, \dots, E) .

Then we deduce easily from the hypothesis on

f

that

\tilde{f}

satisfies the hypotheses (i)–(iii) on the cross

X (Φ_{1} (A_{1}), \dots, Φ_{N} (A_{N}); E, \dots, E) .

Let

\hat{\tilde{f}} \in O ({\hat{X}}^{o} (Φ_{1} (A_{1}), \dots, Φ_{N} (A_{N}); E, \dots, E))

denote the extension function of

\tilde{f}

provided by Theorem A in the case where

D_{1} = \dots = D_{N} = E .

Using Proposition 5.6 and Part 1) of Theorem 5.2 , we see that the function

\hat{f}

defined by

\hat{f} (z) : = \hat{\tilde{f}} (Φ_{1} (z_{1}), \dots, Φ_{N} (z_{N})), z = (z_{1}, \dots, z_{N}) \in {\hat{X}}^{o},

satisfies the conclusions of Parts 1)–4). This proves the above reduction.

We give here only the proof of Theorem A for the case

N = 2 .

Since the general case is quite similar and does not require any new ideas, it is therefore left to the interested reader. In summary, from now on we assume that

N = 2,

D_{1} = D_{2} = E,

and

| f |_{X} < \infty .

(\binom{*}{*})

Using hypotheses (i)–(iii) and

(\binom{*}{*}),

we are able to apply Theorem 6.5 and obtain a function

K [f] \in O (\hat{X})

(see the notation in Subsection 2.3). We define the desired extension function

\hat{f}

as follows

\begin{matrix} \hat{f} : = K [f] on \hat{X} \ (A_{1} \times A_{2}) . \end{matrix}

(7.1)

The remaining part of the proof is divided into several steps.

Step 1: Proof of the estimate

| \hat{f} |_{{\hat{X}}^{o}} \leq | f |_{X} .

Proof of Step 1. Let

z^{0} = (z_{1}^{0}, z_{2}^{0})

be an arbitrary point of

{\hat{X}}^{o} .

Then we may find an

δ_{1} \in (0, 1)

such that

0 < ω (z_{2}^{0}, A_{2}, D_{2}) < δ_{1} < 1 - ω (z_{1}^{0}, A_{1}, D_{1}) .

Let

G

be the connected component of

D_{1, δ_{1}} : = {z_{1} \in D_{1} : ω (z_{1}, A_{1}, D_{1}) < 1 - δ_{1}}

that contains

z_{1}^{0}

. By Theorem 6.3 we may find an angular Jordan domain

Ω : = Ω (F, h)

contained in

G

such that

F \subset A_{1} \cap A_{1}^{*} \cap G^{D_{1}} .

In addition, for every

M \in N,

applying Theorem 6.5 to the function

f^{M},

we obtain the function

K [f^{M}] \in O ({\hat{X}}^{o})

with the following property

{lim}_{z_{1} \to a_{1}, z_{1} \in A_{α} (a_{1})} K [f^{M}] (z_{1}, z_{2}^{0}) = f (a_{1}, z_{2}^{0})^{M} = {lim}_{z_{1} \to a_{1}, z_{1} \in A_{α} (a_{1})} {(K [f] (z_{1}, z_{2}^{0}))}^{M}, 0 < α < \frac{π}{2},

for a.e.

a_{1} \in F .

Consequently, arguing as in the proof of Theorem 6.4 , we get

K [f^{M}] (z^{0}) = {(K [f] (z^{0}))}^{M}, M \in N,

Since

z_{0} \in {\hat{X}}^{o}

is arbitrarily chosen, it follows from the latter identity that

\begin{matrix} K [f^{M}] (z) = {(K [f] (z))}^{M}, M \in N, z \in {\hat{X}}^{o} . \end{matrix}

(7.2)

Now we are able to conclude the proof in the same way as in [6,p.23] . More precisely, taking into account ( 7.1 )–( 7.2 ) and Part 2) of Theorem 4.2 , one gets that

| {\hat{f}}^{M} (z) | \leq | K [f^{M}] (z) | \leq \frac{C_{0} | f |_{X}^{M}}{D_{1}) dist (z_{2}, \partial D_{2}) (1 - e^{- (1 - ω (z))})}, z = (z_{1}, z_{2}) \in {\hat{X}}^{o} .

Extracting the

M

th roots of both sides and letting

M

tend to

\infty,

the desired estimate of Step 1 follows.

□

Step 2: We shall prove that

\hat{f}

is the unique function

O ({\hat{X}}^{o})

which verifies Property 1).

Proof of Step 2. First we show that the function

\hat{f}

given by ( 7.1 ) satisfies Property 1). Without loss of generality, it suffices to prove that there is a subset

{\tilde{A}}_{2}

A_{2} \cap A_{2}^{*}

such that

mes ({\tilde{A}}_{2}) = mes (A_{2})

and

\hat{f}

admits the angular limit

f

at every point of

D_{1} \times {\tilde{A}}_{2} .

For any

a_{1} \in A_{1}

put

A_{a_{1}}^{^{'}} : = {a_{2} \in A_{2} : f (a_{1}, \cdot) has an angular limit at a_{2}} .

By hypothesis (iii), we have

mes (A_{a_{1}}^{^{'}}) = mes (A_{2}),

a_{1} \in A_{1} .

Consequently, applying Fubini's Theorem, we obtain that

\int_{A_{1}} mes (A_{a_{1}}^{^{'}}) | d a_{1} | = mes (A_{1}) mes (A_{2}) = \int_{A_{2}} mes ({a_{1} \in A_{1} : a_{2} \in A_{a_{1}}^{^{'}}}) | d a_{2} | .

Hence,

\begin{matrix} mes ({a_{1} \in A_{1} : a_{2} \in A_{a_{1}}^{^{'}}}) = mes (A_{1}) for a.e. a_{2} \in A_{2} . \end{matrix}

(7.3)

The same reasoning also gives that

\begin{matrix} mes ({a_{1} \in A_{1} : f (a_{1}, a_{2}) = f_{1} (a_{1}, a_{2})}) = mes (A_{1}) for a.e. a_{2} \in A_{2} . \end{matrix}

(7.4)

Set

(7.5) A ~ 2 : = { a 2 ∈ A 2 ∩ A 2 * : mes ( { a 1 ∈ A 1 : a 2 ∈ A a 1 ′ } ) = mes ( A 1 ) and mes ( { a 1 ∈ A 1 : f ( a 1 , a 2 ) = f 1 ( a 1 , a 2 ) } ) = mes ( A 1 ) } .

We deduce from ( 7.3 )–( 7.5 ) that

\begin{matrix} mes ({\tilde{A}}_{2}) = mes (A_{2}) . \end{matrix}

(7.6)

Fix an arbitrary point

a_{2}^{0} \in {\tilde{A}}_{2}

and let

(z_{2}^{n})_{n = 1}^{\infty}

be an arbitrary sequence of

D_{2}

such that

{lim}_{n \to \infty} z_{2}^{n} = a_{2}^{0}

and

z_{2}^{n} \in A_{α} (a_{2}^{0})

for some fixed number

0 < α < \frac{π}{2} .

Fix an arbitrary point

z_{1}^{0}

D_{1}

and let

(z_{1}^{n})_{n = 1}^{\infty}

be an arbitrary sequence of

D_{1}

such that

{lim}_{n \to \infty} z_{1}^{n} = z_{1}^{0} .

Clearly, we may find

0 < δ_{1} < 1

such that

\begin{matrix} {sup}_{n \in N} ω (z_{1}^{n}, A_{1}, D_{1}) < 1 - δ_{1} . \end{matrix}

(7.7)

Fix an

δ_{2}

such that

0 < δ_{2} < δ_{1} .

Since

a_{2}^{0}

is locally regular relative to

A_{2}

and

{lim}_{n \to \infty} z_{2}^{n} = a_{2}^{0}

and

z_{2}^{n} \in A_{α} (a_{2}^{0}),

there is a sufficiently large number

N_{0}

with

\begin{matrix} ω (z_{2}^{n}, A_{2}, D_{2}) < δ_{2}, n > N_{0} . \end{matrix}

(7.8)

Let

G

be that connected component of the following open set

D_{1, δ_{1}} : = {z_{1} \in D_{1} : ω (z_{1}, A_{1}, D_{1}) < 1 - δ_{1}}

which contains

z_{1}^{0}

(see ( 7.7 )). Applying Theorem 6.3 , we may find an angular Jordan domain

Ω : = Ω (F, h)

contained in

G

such that

F \subset A_{1} \cap A_{1}^{*} \cap G^{D_{1}} .

Let

V

be a rectifiable Jordan domain with

Ω \subset V \subset G,

z_{2}^{0} \in V

, and

U

some neighborhood of the base

F

Ω

with

V \cap U = Ω \cap U .

In virtue of ( 7.8 ) and of the fact that

V \subset G \subset D_{1, δ_{1}},

we obtain that

\begin{matrix} V \times {z_{2}^{n}} \subset {\hat{X}}^{o}, n > N_{0} . \end{matrix}

(7.9)

Consequently, Theorem 6.5 yields that for any

n > N_{0},

\begin{matrix} f (a_{1}, z_{2}^{n}) = {lim}_{z_{1} \to a_{1}, z_{1} \in A_{α} (a_{1})} \hat{f} (z_{1}, z_{2}^{n}), 0 < α < \frac{π}{2}, \end{matrix}

(7.10)

for a.e.

a_{1} \in F .

Next, for any

n > N_{0}

let

\begin{matrix} F_{n} & : = {a_{1} \in F : a_{2}^{0} \in A_{a_{1}}^{^{'}} and f (a_{1}, z_{2}^{n}) = {lim}_{z_{1} \to a_{1}, z_{1} \in A_{α} (a_{1})} \hat{f} (z_{1}, z_{2}^{n})}, & F_{0} & : = ⋂_{n = N_{0} + 1}^{\infty} F_{n} . \end{matrix}

It follows from ( 7.5 ), ( 7.10 ) and the fact that

a_{2}^{0} \in {\tilde{A}}_{2}

that

mes (F_{n}) = mes (F),

n > N_{0} .

Hence

\begin{matrix} mes (F_{0}) = mes (F) > 0 . \end{matrix}

(7.11)

In virtue of ( 7.9 ), consider the following holomorphic functions on

V

\begin{matrix} h_{n} (t) : = \hat{f} (t, z_{2}^{n}) and h_{0} (t) : = f (t, a_{2}^{0}), t \in V, n > N_{0} . \end{matrix}

(7.12)

Since we have already shown in Step I that

| h_{n} |_{V} \leq | f |_{X} < \infty,

n > N_{0}

n = 0,

applying Part 1) of Theorem 6.8 , we may find a subset

Δ

F_{0}

with

mes (Δ) = mes (F_{0}) > 0

such that

h_{n},

n > N_{0}

(resp.

h_{0}

) admits the angular limit

f_{1} (t, z_{2}^{n})

(resp.

f_{1} (t, a_{2}^{0}))

t \in Δ .

Observe that by ( 7.5 ) and the fact that

a_{2}^{0} \in {\tilde{A}}_{2}

we have that

{lim}_{n \to \infty} f_{1} (t, z_{2}^{n}) = f_{1} (t, a_{2}^{0}) = f (t, a_{2}^{0}) for a.e. t \in Δ .

Using this and ( 7.12 ), we are able to apply Lemma 4.4 to the sequence

(h_{n})_{n = 0}^{\infty} .

Consequently, one gets

{lim}_{n \to \infty} \hat{f} (z_{1}^{n}, z_{2}^{n}) = f (z_{1}^{0}, a_{2}^{0}) .

This shows that

\hat{f}

admits the angular limit

f

at every point of

D_{1} \times {\tilde{A}}_{2} .

Hence,

\hat{f}

satisfies Property 1).

In order to complete Step 2 we need to show the uniqueness of

\hat{f}

. To do this, let

\hat{\hat{f}} \in O ({\hat{X}}^{o})

be a function with the following property: There is a subset

{\tilde{\tilde{A}}}_{j}

A_{j} \cap A_{j}^{*}

(j = 1, 2)

such that

mes (A_{j} \ {\tilde{\tilde{A}}}_{j}) = 0

and

\hat{\hat{f}}

admits the angular limit

f

at every point of

({\tilde{\tilde{A}}}_{1} \times D_{2}) \cup (D_{1} \times {\tilde{\tilde{A}}}_{2}) .

Fix an arbitrary point

z^{0} = (z_{1}^{0}, z_{2}^{0}) \in {\hat{X}}^{o} .

Let

G

be the connected component containing

z_{1}^{0}

of the following open set

{z_{1} \in D_{1} : ω (z_{1}, A_{1}, D_{1}) < 1 - ω (z_{2}^{0}, A_{2}, D_{2})} .

We deduce from the property of

\hat{f}

and

\hat{\hat{f}}

that both holomorphic functions

\hat{f} (\cdot, z_{2}^{0}) |_{G}

and

\hat{\hat{f}} (\cdot, z_{2}^{0}) |_{G}

admit the angular limit

f (\cdot, z_{2}^{0})

at every point of

{\tilde{A}}_{1} \cap {\tilde{\tilde{A}}}_{1} \cap G^{D_{1}} .

Consequently, applying Theorem 6.4 yields that

\hat{f} (\cdot, z_{2}^{0}) = \hat{\hat{f}} (\cdot, z_{2}^{0})

G .

Hence,

\hat{f} (z^{0}) = \hat{\hat{f}} (z^{0}) .

Since

z^{0} \in {\hat{X}}^{o}

is arbitrary, the uniqueness of

\hat{f}

is established. This completes Step 2.

□

Step 3: Proof of Part 2).

Proof of Step 3. Fix

(z_{1}^{0}, z_{2}^{0}) \in {\hat{X}}^{o} .

For every

a_{2} \in A_{2}

we have

| f (a_{1}, a_{2}) | \leq | f |_{A_{1} \times A_{2}}, a_{1} \in A_{1}, and | f (z_{1}, a_{2}) | \leq | f |_{X}, z_{1} \in D_{1} .

Therefore, the Two-Constant Theorem (see Theorem 2.2 in [14] ) implies that

\begin{matrix} | f (z_{1}, a_{2}) | \leq | f |_{A_{1} \times A_{2}}^{1 - ω (z_{1}, A_{1}, D_{1})} | f |_{X}^{ω (z_{1}, A_{1}, D_{1})}, z_{1} \in D_{1}, a_{2} \in A_{2} . \end{matrix}

(7.13)

Let

δ : = ω (z_{1}^{0}, A_{1}, D_{1})

and consider the

δ

-level set

D_{2, δ} : = {z_{2} \in D_{2} : ω (z_{2}, A_{2}, D_{2}) < 1 - δ} .

Clearly,

z_{2}^{0} \in D_{2, δ} .

Recall from Step 2 that

{\tilde{A}}_{2} \subset A_{2} \cap A_{2}^{*},

mes ((A_{2} \cap A_{2}^{*}) \ {\tilde{A}}_{2}) = 0,

and

\begin{matrix} f (z_{1}^{0}, a_{2}) = {lim}_{z_{2} \to a_{2}, z_{2} \in A_{α} (a_{2})} \hat{f} (z_{1}^{0}, z_{2}), 0 < α < \frac{π}{2}, a_{2} \in {\tilde{A}}_{2} . \end{matrix}

(7.14)

Consider the following function

h : D_{2, δ} \cup {\tilde{A}}_{2} ⟶ C

defined by

\begin{matrix} h (t) : = {\begin{matrix} \hat{f} (z_{1}^{0}, t), & t \in D_{2, δ} \\ f (z_{1}^{0}, t), & t \in {\tilde{A}}_{2} \end{matrix} . \end{matrix}

(7.15)

Clearly,

h |_{D_{2, δ}} \in O (D_{2, δ}) .

On the other hand, in virtue of ( 7.15 ) and the result of Step 1, we have

\begin{matrix} | h |_{D_{2, δ}} \leq | \hat{f} |_{{\hat{X}}^{0}} \leq | f |_{X} < \infty . \end{matrix}

(7.16)

In addition, applying Corollary 5.20 and taking ( 7.14 )–( 7.15 ) into account yields

| h (t) | \leq | h |_{{\tilde{A}}_{2}}^{1 - ω_{δ} (t, A_{1}, D_{1})} | h |_{D_{2, δ}}^{ω_{δ} (t, A_{1}, D_{1})}, t \in D_{2, δ},

where, by Theorem 5.19 ,

ω_{δ} (t, A_{2}, D_{2}) = \frac{ω (t, A_{2}, D_{2})}{1 - ω (z_{1}^{0}, A_{1}, D_{1})} .

This, combined with ( 7.13 )–( 7.16 ), implies that

\begin{matrix} | \hat{f} (z_{1}^{0}, z_{2}^{0}) | & = & | h (z_{2}^{0}) | \leq | f |_{A_{1} \times A_{2}}^{1 - ω (z_{1}^{0}, A_{1}, D_{1}) - ω (z_{2}^{0}, A_{2}, D_{2})} | f |_{X}^{ω (z_{1}^{0}, A_{1}, D_{1}) + ω (z_{2}^{0}, D_{2}, A_{2})} . \end{matrix}

Hence Part 2) for the point

(z_{1}^{0}, z_{2}^{0})

is proved.

□

Step 4: Proof of Part 3).

Proof of Step 4. Let

(a_{1}^{0}, z_{2}^{0}) \in A_{1}^{*} \times D_{2}

be such that the following limit exists

λ : = {lim}_{(a_{1}, z_{2}) \to (a_{1}^{0}, z_{2}^{0}), (a_{1}, z_{2}) \in A_{1} \times D_{2}} f (a_{1}, z_{2}) .

We like to show that

\hat{f}

admits the angular limit

λ

(a_{1}^{0}, z_{2}^{0}) .

In virtue of assumption

(\binom{*}{*}),

we may suppose (without loss of generality) that

| f |_{X} < 1 .

For any

0 < ε < \frac{1}{2},

we may find an open neighborhood

A_{a_{1}^{0}}

a_{1}^{0}

A_{1}

and a positive number

r > 0

such that

B (z_{2}^{0}, r) ⋐ D_{2}

and

\begin{matrix} | f (a_{1}, z_{2}) - λ | < ε^{2}, a_{1} \in A_{a_{1}^{0}}, | z_{2} - z_{2}^{0} | \leq r . \end{matrix}

(7.17)

Put

\begin{matrix} δ : = {sup}_{z_{2} \in \bar{B (z_{2}^{0}, r)}} ω (z_{2}, A_{2}, D_{2}) . \end{matrix}

(7.18)

Since

a_{1}^{0} \in A_{1}^{*},

it follows from Part 2) of Proposition 5.9 that

mes (A_{a_{1}^{0}}) > 0 .

Next, consider the level set

D_{1, δ} : = {z_{1} \in D_{1} : ω (z_{1}, A_{a_{1}^{0}}, D_{1}) < 1 - δ} .

In virtue of ( 7.18 ), we can define

\begin{matrix} h (t, z_{2}) : = \hat{f} (t, z_{2}) - λ, t \in D_{1, δ}, z_{2} \in \bar{B (z_{2}^{0}, r)} . \end{matrix}

(7.19)

Clearly,

\begin{matrix} | h |_{D_{1, δ}} \leq 2 | \hat{f} |_{{\hat{X}}^{0}} = 2 | f |_{X} = 2 . \end{matrix}

(7.20)

By ( 7.19 ) and using the result of Step 2, we know that for every

z_{2} \in B (z_{2}^{0}, r)

the holomorphic function

h (\cdot, z_{2}) |_{D_{1, δ}}

admits the angular limit

f (a_{1}, z_{2}) - λ

a_{1}

for

a_{1} \in {\tilde{A}}_{1} \cap A_{a_{1}^{0}},

where

{\tilde{A}}_{1}

is given in Step 2. Consequently, applying Corollary 5.20 and taking ( 7.17 ) and ( 7.20 ) into account, we see that

| h (t, z_{2}) | < ε^{2 (1 - ω_{δ} (t, A_{a_{1}^{0}}, D_{1}))} 2^{ω_{δ} (t, A_{a_{1}^{0}}, D_{1})}, t \in D_{1, δ} .

Let

0 < α < \frac{π}{2} .

In virtue of Theorem 5.19 and the hypothesis that

a_{1}^{0} \in A_{1}^{*},

we deduce that

{lim}_{t \to a_{1}^{0}, t \in A_{α} (a_{1}^{0})} ω_{δ} (t, A_{a_{1}^{0}}, D_{1}) = 0 .

Consequently, there is an

r_{α} > 0

such that

| f (z_{1}, z_{2}) - λ | = | h (z_{1}, z_{2}) | < ε, z_{1} \in A_{α} (a_{1}^{0}) \cap {| z_{1} - a_{1}^{0} | < r_{α}}, z_{2} \in B (z_{2}^{0}, r) .

This completes the above assertion.

Similarly, we can prove that

\hat{f}

admits the angular limit

{lim}_{(z_{1}, a_{2}) \to (z_{1}^{0}, a_{2}^{0}), (z_{1}, a_{2}) \in D_{1} \times A_{2}} f (z_{1}, a_{2})

at any point

(z_{1}^{0}, a_{2}^{0}),

if the latter limit exists. Hence the proof of Step 4 (i.e. Part 3)) is finished.

□

Step 5: Proof of Part 4).

Proof of Step 5. Let

(a_{1}^{0}, a_{2}^{0}) \in A_{1}^{*} \times A_{2}^{*}

be such that the following limit exists

λ : = {lim}_{(a_{1}, a_{2}) \to (a_{1}^{0}, a_{2}^{0}), (a_{1}, a_{2}) \in A_{1} \times A_{2}} f (a_{1}, a_{2}) .

We like to show that

\hat{f}

admits the angular limit

λ

(a_{1}^{0}, a_{2}^{0}) .

Suppose without loss of generality that

| f |_{X} < 1,

and fix an arbitrary

0 < ε < \frac{1}{2} .

Since

(a_{1}^{0}, a_{2}^{0}) \in A_{1}^{*} \times A_{2}^{*},

we may find an open neighborhood

A_{a_{1}^{0}}

a_{1}^{0}

A_{1}

(resp.

an open neighborhood

A_{a_{2}^{0}}

a_{2}^{0}

A_{2})

such that

\begin{matrix} | f (a_{1}, a_{2}) - λ | < ε^{2}, a_{1} \in A_{a_{1}^{0}}, a_{2} \in A_{a_{2}^{0}} . \end{matrix}

(7.21)

By Part 2) of Proposition 5.9 , one gets

mes (A_{a_{1}^{0}}) > 0

and

mes (A_{a_{2}^{0}}) > 0 .

Consider the function

\begin{matrix} h (z_{1}, z_{2}) : = f (z_{1}, z_{2}) - λ, (z_{1}, z_{2}) \in X (A_{a_{1}^{0}}, A_{a_{2}^{0}}; D_{1}, D_{2}) . \end{matrix}

(7.22)

Clearly,

\begin{matrix} | h (z_{1}, z_{2}) | \leq 2, (z_{1}, z_{2}) \in X (A_{a_{1}^{0}}, A_{a_{2}^{0}}; D_{1}, D_{2}) . \end{matrix}

(7.23)

Applying the results of Steps 1–3 to

h,

we obtain the function

\begin{matrix} \hat{h} : = K [h] on {\hat{X}}^{0} (A_{a_{1}^{0}}, A_{a_{2}^{0}}; D_{1}, D_{2}) . \end{matrix}

(7.24)

so that

\hat{h}

admits the angular limit

h

({\tilde{A}}_{a_{1}^{0}} \times D_{2}) \cup (D_{1} \times {\tilde{A}}_{a_{1}^{0}}),

where

{\tilde{A}}_{a_{1}^{0}},

{\tilde{A}}_{a_{1}^{0}}

are given by Step 2. Clearly,

\hat{X} (A_{a_{1}^{0}}, A_{a_{2}^{0}}; D_{1}, D_{2}) \subset \hat{X} (A_{1}, A_{2}; D_{1}, D_{2}) .

Consequently, arguing as in Step 1 and taking into account the above mentioned angular limit of

\hat{h},

we conclude that

\hat{h} = \hat{f} - λ on \hat{X} (A_{a_{1}^{0}}, A_{a_{2}^{0}}; D_{1}, D_{2}) .

Consequently, applying Step 3 and taking into account ( 7.21 )–( 7.24 ) and the inequality

| f |_{X} < 1,

we see that

\begin{matrix} | \hat{f} (z_{1}, z_{2}) - λ | & = | \hat{h} (z_{1}, z_{2}) | \leq | h |_{A_{a_{1}^{0}} \times A_{a_{2}^{0}}}^{1 - ω (z_{1}, A_{a_{1}^{0}}, D_{1}) - ω (z_{2}, A_{a_{2}^{0}}, D_{2})} (2 | f |_{X})^{ω (z_{1}, A_{a_{1}^{0}}, D_{1}) + ω (z_{2}, A_{a_{2}^{0}}, D_{2})} & < ε^{2 (1 - ω (z_{1}, A_{a_{1}^{0}}, D_{1}) - ω (z_{2}, A_{a_{2}^{0}}, D_{2}))} 2^{ω (z_{1}, A_{a_{1}^{0}}, D_{1}) + ω (z_{2}, A_{a_{2}^{0}}, D_{2})} . \end{matrix}

Therefore, for all

z = (z_{1}, z_{2}) \in \hat{X} (A_{a_{1}^{0}}, A_{a_{2}^{0}}; D_{1}, D_{2})

satisfying

\begin{matrix} ω (z_{1}, A_{a_{1}^{0}}, D_{1}) + ω (z_{2}, A_{a_{2}^{0}}, D_{2}) < \frac{1}{3}, \end{matrix}

(7.25)

we deduce from the latter estimate that

\begin{matrix} | \hat{f} (z) - λ | < ε . \end{matrix}

(7.26)

Since

a_{1}^{0}

(resp.

a_{2}^{0}

) is locally regular relative to

A_{a_{1}^{0}}

(resp.

A_{a_{2}^{0}}

), there is an

r_{α} > 0

such that ( 7.25 ) is fulfilled for

z = (z_{1}, z_{2}) \in (A_{α} (a_{1}^{0}) \cap {| z_{1} - a_{1}^{0} | < r_{α}}) \times (A_{α} (a_{2}^{0}) \cap {| z_{2} - a_{2}^{0} | < r_{α}}) .

This, combined with ( 7.26 ), completes the proof. Hence Step 5 (i.e. Part 4)) is finished.

□

Step 6: Proof of Part 5).

Proof of Step 6. In virtue of Step 5, we only need to show that

\hat{f}

admits the angular limit

f

(A_{1}^{*} \times D_{2}) \cup (D_{1} \times A_{2}^{*}) .

To do this let

(a_{1}^{0}, z_{2}^{0}) \in A_{1}^{*} \times D_{2}

and choose an arbitrary

0 < ε < 1 .

Fix a compact subset

K

A_{2} \cap A_{2}^{*}

such that

mes (K) > 0

and a sufficiently large

N

such that

\begin{matrix} ε^{N (1 - ω (z_{2}^{0}, A_{2}, D_{2}))} (2 | f |_{X})^{ω (z_{2}^{0}, A_{2}, D_{2})} < \frac{ε}{2} . \end{matrix}

(7.27)

Using the hypothesis that

f

can be extended to a continuous function on

A_{1}^{*} \times A_{2}^{*},

we may find an open neighborhood

A_{a_{1}^{0}}

a_{1}^{0}

A_{1}^{*}

such that

\begin{matrix} | f (a_{1}, a_{2}) - f (a_{1}^{0}, a_{2}) | \leq ε^{N}, a_{1} \in A_{a_{1}^{0}} \cap A_{a_{1}^{0}}^{*}, a_{2} \in K . \end{matrix}

(7.28)

On the other hand,

\begin{matrix} | f (a_{1}, z_{2}) - f (a_{1}^{0}, z_{2}^{0}) | \leq 2 | f |_{X} < \infty, a_{1} \in A_{a_{1}^{0}} \cap A_{a_{1}^{0}}^{*}, z_{2} \in D_{2} . \end{matrix}

(7.29)

For

a_{1} \in A_{a_{1}^{0}} \cap A_{a_{1}^{0}}^{*},

applying the Two-Constant Theorem to the function

f (a_{1}, \cdot) - f (a_{1}^{0}, \cdot) \in O (D_{2})

and taking ( 7.27 )–( 7.29 ) into account, we deduce that

\begin{matrix} | f (a_{1}, z_{2}^{0}) - f (a_{1}^{0}, z_{2}^{0}) | \leq ε^{N (1 - ω (z_{2}^{0}, K, D_{2}))} (2 | f |_{X})^{ω (z_{2}^{0}, K, D_{2})} < \frac{ε}{2} . \end{matrix}

(7.30)

Since

f (a_{1}, \cdot) |_{D_{2}}

is a bounded holomorphic function for

a_{1} \in A_{1},

there is an open neighborhood

V

z_{2}^{0}

such that

| f (a_{1}, z_{2}) - f (a_{1}, z_{2}^{0}) | < \frac{ε}{2}, a_{1} \in A_{1}, z_{2} \in V .

This, combined with ( 7.30 ), implies that

\begin{matrix} | f (a_{1}, z_{2}) - f (a_{1}^{0}, z_{2}^{0}) | & \leq & | f (a_{1}, z_{2}^{0}) - f (a_{1}^{0}, z_{2}^{0}) | + | f (a_{1}, z_{2}) - f (a_{1}, z_{2}^{0}) | \end{matrix}

\begin{matrix} < & \frac{ε}{2} + \frac{ε}{2} = ε, a_{1} \in A_{a_{1}^{0}}, z_{2} \in V . \end{matrix}

Therefore,

f

is continuous at

(a_{1}^{0}, z_{2}^{0}) .

Consequently, we conclude, by Step 5, that

\hat{f}

admits the angular limit

f (a_{1}^{0}, z_{2}^{0})

(a_{1}^{0}, z_{2}^{0}) .

Similarly, we may also show that

\hat{f}

admits the angular limit

f (z_{1}^{0}, a_{2}^{0})

at every point

(z_{1}^{0}, a_{2}^{0}) \in D_{1} \times A_{2}^{*} .

This completes the proof of the last step.

□

8 Preparatory results

We first develop some auxiliary results. This preparation will enable us to generalize the results of section 7 to the general case considered in Theorem A.

Definition 8.1. Let

Ω \subset C^{n}

be an open set and let

A \subset Ω .

Define

\begin{matrix} h_{A, Ω} & : = & sup {u : u \in P S ℋ (Ω), u \leq 1 on Ω, u \leq 0 on A}, \end{matrix}

\begin{matrix} h_{A, Ω}^{*} (z) & : = & {limsup}_{w \to z} h_{A, Ω} (w), z \in Ω . \end{matrix}

The function

h_{A, Ω}^{*}

is called the plurisubharmonic measure of

A

relative to

Ω .

Proposition 8.2. Let

D \subset C

be an open set which is locally rectifiable on a linearly measurable subset

A

\partial D

with

mes (A) > 0 .

Let

{a_{j}}_{j \in J}

be a finite or countable subset of

A

with the following properties:

(i) For any $j \in J,$ there is an open neighborhood $U_{j}$ of $a_{j}$ such that $D \cap U_{j}$ is either a rectifiable Jordan domain or the disjoint union of two rectifiable Jordan domains;
(ii) $A \subset \cup_{j \in J} U_{j} .$

For any

0 < δ < \frac{1}{2},

define

\begin{matrix} U_{j, δ} & : = {z \in D \cap U_{j} : ω (z, A \cap U_{j}, D \cap U_{j}) < δ}, j \in J, & A_{δ} & : = ⋃_{j \in J} U_{j, δ}, & D_{δ} & : = {z \in D : ω (z, A, D) < 1 - δ} . \end{matrix}

Then: 1)

A \cap A^{*} \subset A_{δ}^{D}

and

A_{δ} \subset D_{1 - δ} \subset D_{δ};

ω (z, A, D) - δ \leq h_{A_{δ}, D}^{*} (z) \leq ω (z, A, D),

z \in D .

Proof. To prove Part 1), let

a \in A \cap A^{*}

and fix an

j \in J

such that

a \in U_{j} .

Then

{lim}_{z \to a, z \in A_{α} (a)} ω (z, A \cap U_{j}, D \cap U_{j}) = 0, 0 < α < \frac{π}{2} .

Consequently, for every

0 < α < \frac{π}{2},

there is an open neighborhood

V_{α} \subset U_{j}

a

such that

ω (z, A \cap U_{j}, D \cap U_{j}) < δ, z \in A_{α} (a) \cap V_{α} .

This proves

A \cap A^{*} \subset A_{δ}^{D} .

To prove the second assertion of Part 1), one invokes the Subordination Principle and obtains for

z \in U_{j, δ},

\begin{matrix} ω (z, A, D) \leq ω (z, A \cap U_{j}, D \cap U_{j}) < δ < 1 - δ . \end{matrix}

(8.1)

Hence,

z \in D_{1 - δ} .

This implies that

A_{δ} \subset D_{1 - δ} .

In addition, since

0 < δ < \frac{1}{2},

it follows that

D_{1 - δ} \subset D_{δ} .

Hence, Part 1) is proved.

We turn to Part 2). Since

A_{δ}

is an open set and, by Part 1),

A \cap A^{*} \subset A_{δ}^{D},

it follows from Definitions 5.17 , 8.1 that

h_{A_{δ}, D}^{*} (z) \leq ω (z, A \cap A^{*}, D), z \in D .

Hence, in virtue of Proposition 5.9 and Theorem 5.14 , it follows that

h_{A_{δ}, D}^{*} (z) \leq ω (z, A, D), z \in D,

which proves the second estimate of Part 2).

To complete Part 2), let

z \in A_{δ} .

Choose

j \in J

such that

z \in U_{j, δ} .

We deduce from ( 8.1 ) that

ω (z, A, D) - δ \leq 0 .

Hence,

ω (z, A, D) - δ \leq 0, z \in A_{δ} .

On the other hand,

ω (z, A, D) - δ < 1,

z \in D .

Consequently, the first estimate of Part 2) follows. The proof of the lemma is finished. □

The next result gives a nice geometric property of the interior of a wedge.

Proposition 8.3. Let

N \in N,

N \geq 2,

D_{j} \subset C

a domain and

A_{j} \subset \partial D_{j}

such that

D_{j}

is locally rectifiable on

A_{j},

mes (A_{j}) > 0,

j = 1, \dots, N .

Put

{\hat{X}}^{o} : = {\hat{X}}^{o} (A_{1}, \dots, A_{N}; D_{1}, \dots, D_{N}) .

Then

{\hat{X}}^{o}

is a domain.

Proof. We only give the proof for the case

N = 2 .

The general case when

N > 2

can be proved in exactly the same way. Fix arbitrary points

z^{0} = (z_{1}^{0}, z_{2}^{0})

and

w^{0} = (w_{1}^{0}, w_{2}^{0})

{\hat{X}}^{o} .

Let

\begin{matrix} ε : = \frac{1}{2} (min {1 - ω (z_{1}^{0}, A_{1}, D_{1}) - ω (z_{2}^{0}, A_{2}, D_{2}), 1 - ω (w_{1}^{0}, A_{1}, D_{1}) - ω (w_{2}^{0}, A_{2}, D_{2})}) . \end{matrix}

(8.2)

For

j, k \in {1, 2}

j \neq k,

let

D_{j, z_{j}^{0}}

(resp.

D_{j, w_{j}^{0}}

) be the connected component containing

z_{j}^{0}

(resp.

w_{j}^{0}

) of the following open set

\begin{matrix} \begin{matrix} {z_{j} \in D_{j} : ω (z_{j}, A_{j}, D_{j}) < 1 - ω (z_{k}^{0}, A_{k}, D_{k}) - ε}, \\ (resp & {z_{j} \in D_{j} : ω (z_{j}, A_{j}, D_{j}) < 1 - ω (w_{k}^{0}, A_{k}, D_{k}) - ε}), \end{matrix} \end{matrix}

(8.3)

where

ε

is given in ( 8.2 ).

By Theorem 5.18 (iii), for

j \in {1, 2},

we may find a point

a_{j} \in A_{j} \cap A_{j}^{*} \cap D_{j, z_{j}^{0}}^{D_{j}}

(resp.

b_{j} \in A_{j} \cap A_{j}^{*} \cap D_{j, w_{j}^{0}}^{D_{j}}

For

j \in {1, 2},

fix a connected open neighborhood

V_{z_{j}^{0}} \subset D_{j}

z_{j}^{0}

and an open neighborhood

V_{a_{j}} \subset D_{j}

a_{j}

such that

\begin{matrix} \begin{matrix} V_{a_{j}} \cap A_{\frac{π}{4}} (a_{j}) & is a domain, & ω (z_{j}, A_{j}, D_{j}) & < ω (z_{j}^{0}, A_{k}, D_{j}) + ε, z_{j} \in V_{z_{j}^{0}}, & ω (z_{j}, A_{j}, D_{j}) & < ε, z_{j} \in V_{a_{j}} \cap A_{\frac{π}{4}} (a_{j}) . \end{matrix} \end{matrix}

(8.4)

Combining ( 8.2 )–( 8.4 ), we see that

V_{z_{1}^{0}} \times V_{z_{2}^{0}}

and

V_{z_{1}^{0}} \times (V_{a_{2}} \cap A_{\frac{π}{4}} (a_{2}))

are in the same connected component of

{\hat{X}}^{o} .

The same argument also shows that

V_{z_{1}^{0}} \times (V_{a_{2}} \cap A_{\frac{π}{4}} (a_{2}))

and

(V_{a_{1}} \cap A_{\frac{π}{4}} (a_{1})) \times (V_{a_{2}} \cap A_{\frac{π}{4}} (a_{2}))

are in the same connected component of

{\hat{X}}^{o} .

Hence,

V_{z_{1}^{0}} \times V_{z_{2}^{0}}

and

(V_{a_{1}} \cap A_{\frac{π}{4}} (a_{1})) \times (V_{a_{2}} \cap A_{\frac{π}{4}} (a_{2}))

are in the same connected component of

{\hat{X}}^{o} .

We apply the same argument as above to

w^{0} .

Consequently, one may find, for

j \in {1, 2},

a connected open neighborhood

V_{w_{j}^{0}} \subset D_{j}

w_{j}^{0}

and an open neighborhood

V_{b_{j}} \subset D_{j}

b_{j}

such that

\begin{matrix} \begin{matrix} V_{b_{j}} \cap A_{\frac{π}{4}} (b_{j}) & is a domain, & ω (z_{j}, A_{j}, D_{j}) & < ω (w_{j}^{0}, A_{k}, D_{j}) + ε, z_{j} \in V_{w_{j}^{0}}, & ω (z_{j}, A_{j}, D_{j}) & < ε, z_{j} \in V_{b_{j}} \cap A_{\frac{π}{4}} (b_{j}), \end{matrix} \end{matrix}

(8.5)

and

V_{w_{1}^{0}} \times V_{w_{2}^{0}}

and

(V_{b_{1}} \cap A_{\frac{π}{4}} (b_{1})) \times (V_{b_{2}} \cap A_{\frac{π}{4}} (b_{2}))

are in the same connected component of

{\hat{X}}^{o} .

Recall that

D_{1}

and

D_{2}

are domains. Then in virtue of ( 8.4 ) and ( 8.5 ), one may find, for

j \in {1, 2},

a Jordan curve

γ_{j} : [0, 1] ⟶ D_{j}

such that

\begin{matrix} γ_{j} (0) \in V_{a_{j}} \cap A_{\frac{π}{4}} (a_{j}) and γ_{j} (1) \in V_{b_{j}} \cap A_{\frac{π}{4}} (b_{j}) . \end{matrix}

(8.6)

Let

\begin{matrix} δ : = \frac{1}{2} (min {1 - {sup}_{t \in [0, 1]} ω (γ_{1} (t), A_{1}, D_{1}), 1 - {sup}_{t \in [0, 1]} ω (γ_{2} (t), A_{2}, D_{2})}) . \end{matrix}

(8.7)

Next, we shrink

V_{b_{2}}

such that

ω (z_{2}, A_{2}, D_{2}) < δ, z_{2} \in V_{b_{2}} .

This, combined with ( 8.4 )–( 8.7 ), implies that

(V_{a_{1}} \cap A_{\frac{π}{4}} (a_{1})) \times (V_{b_{2}} \cap A_{\frac{π}{4}} (b_{2}))

and

(V_{b_{1}} \cap A_{\frac{π}{4}} (b_{1})) \times (V_{b_{2}} \cap A_{\frac{π}{4}} (b_{2}))

are in the same connected component of

{\hat{X}}^{o} .

Similarly, by shrinking

V_{a_{1}}

if necessary, we see that

(V_{a_{1}} \cap A_{\frac{π}{4}} (a_{1})) \times (V_{a_{2}} \cap A_{\frac{π}{4}} (a_{2}))

and

(V_{a_{1}} \cap A_{\frac{π}{4}} (a_{1})) \times (V_{b_{2}} \cap A_{\frac{π}{4}} (b_{2}))

are in the same connected component of

{\hat{X}}^{o} .

Hence,

(V_{a_{1}} \cap A_{\frac{π}{4}} (a_{1})) \times (V_{a_{2}} \cap A_{\frac{π}{4}} (a_{2}))

and

(V_{b_{1}} \cap A_{\frac{π}{4}} (b_{1})) \times (V_{b_{2}} \cap A_{\frac{π}{4}} (b_{2}))

are in the same connected component of

{\hat{X}}^{o} .

In summary, combining the above fact, ( 8.5 ) and the similar conclusion for

z^{0},

we have shown that

V_{z_{1}^{0}} \times V_{z_{2}^{0}}

and

V_{w_{1}^{0}} \times V_{w_{2}^{0}}

are in the same connected component of

{\hat{X}}^{o} .

Consequently,

z^{0}

and

w^{0}

are in the same connected component of

{\hat{X}}^{o},

which completes the proof. □

The main ingredient in the proof of Theorem A is the following mixed cross theorem.

Theorem 8.4. Let

D \subset C^{n}

be a domain of holomorphy,

Ω \subset C

an open set,

A \subset D,

and

B

a linearly measurable subset of

\partial Ω .

Assume that

A = \cup_{k = 1}^{\infty} A_{k}

with

A_{k}

locally pluriregular⁵ compact subsets of

D,

A_{k} \subset A_{k + 1},

k \geq 1 .

In addition,

Ω

is locally rectifiable on

B

with

mes (B) > 0,

and

B \subset B^{*} .

For

0 \leq δ < 1

put

G : = {w \in Ω : ω (w, B, Ω) < 1 - δ} .

Let

X : = X (A, B; D, G)

X^{o} : = X^{o} (A, B; D, G),

and (using the notation of Theorem 5.19 )

{\hat{X}}^{o} = {\hat{X}}^{o} (A, B; D, G) : = {(z, w) \in D \times G : h_{A, D}^{*} (z) + ω_{δ} (w, B, Ω) < 1} .

Let

f : X ⟶ C

be such that

(i) $f \in O_{s} (X^{o});$
(ii) $f$ is measurable and bounded on $X;$
(iii) for any $z \in A,$ ${lim}_{w \to η, w \in A_{α} (η)} f (z, w) = f (z, η), η \in B, 0 < α < \frac{π}{2} .$

Then there is a unique function

\hat{f} \in O ({\hat{X}}^{o})

such that

\hat{f} = f

A \times G

and

{lim}_{z \to z_{0}, w \to η_{0}, w \in A_{α} (η_{0})} \hat{f} (z, w) = f (z_{0}, η_{0}), 0 < α < \frac{π}{2},

for every

z_{0} \in D

and

η_{0} \in B .

Moreover,

| \hat{f} |_{{\hat{X}}^{o}} \leq | f |_{X} .

Proof. First one proves the existence and uniqueness of

\hat{f} .

To do this we argue as in the proof of Theorem 3.5.1 in [7] . For the sake of completeness, we give here a sketchy proof. Fix an

f : X ⟶ C

which satisfies (i)–(iii) above.

Step I: Reduction to the case where

D

is strongly pseudoconvex and

A

is a locally pluriregular compact subset of

D .

One proceeds as in the first and second step in that proof. More precisely, since

D

is a domain of holomorphy, we may find an exhaustion sequence

(D_{k})_{k = 1}^{\infty}

of relatively compact, strongly pseudoconvex subdomains

D_{k}

D

with

A_{k} \subset D_{k} ↗ D .

By reduction assumption, for each

k

there exists an

{\hat{f}}_{k} \in O ({\hat{X}}^{o} (A_{k}, B; D_{k}, G))

such that

{\hat{f}}_{k}

admit the angular limit

f |_{X (A_{k}, B; D_{k}, G)}

X (A_{k}, B; D_{k}, G) .

We would like to show that

{\hat{f}}_{k + 1} = {\hat{f}}_{k}

{\hat{X}}^{o} (A_{k}, B; D_{k}, G) .

Indeed, fix an arbitrary

k_{0} \geq 1

and an arbitrary point

(z_{0}, w_{0}) \in {\hat{X}}^{o} (A_{k_{0}}, B; D_{k_{0}}, G) .

Let

k \in N

such that

k \geq k_{0} .

Let

D

be the connected component containing

z_{0}

of the following open set

{z \in D : h_{A_{k_{0}}, D_{k_{0}}}^{*} (z) < 1 - ω_{δ} (w_{0}, B, Ω)} .

Observe that both functions

{\hat{f}}_{k_{0}} (\cdot, w_{0}) |_{D}

and

{\hat{f}}_{k} (\cdot, w_{0}) |_{D}

are holomorphic and

{\hat{f}}_{k} (z, w_{0}) = f_{k} (z, w_{0}) = {\hat{f}}_{k_{0}} (z, w_{0}), z \in A_{k} \cap D .

Since

A_{k} \cap D

is non-pluripolar, we deduce that

{\hat{f}}_{k_{0}} (\cdot, w_{0}) |_{D} = {\hat{f}}_{k} (\cdot, w_{0}) |_{D} .

Hence,

{\hat{f}}_{k_{0}} (z_{0}, w_{0}) = {\hat{f}}_{k} (z_{0}, w_{0}),

which proves the above assertion.

On the other hand, by Proposition 5.15 one gets

{\hat{X}}^{o} (A_{k}, B; D_{k}, G) ↗ {\hat{X}}^{o}

k ↗ \infty .

Therefore, we may glue

{\hat{f}}_{k}

together to obtain a function

\hat{f} \in O ({\hat{X}}^{o})

such that

\hat{f}

admits the angular limit

f

X

and

\hat{f} = f

A \times G .

The uniqueness of such an extension

\hat{f}

can be proved using the argument given in the previous paragraph.

This completes Step I. Step II: The case where

D

is strongly pseudoconvex and

A

is a locally pluriregular compact subset of

D .

Suppose without loss of generality that

| f |_{X} < 1 .

The key observation is that we are still able to apply the classical method of doubly orthogonal bases of Bergman type.

Next one observes that Lemma 3.5.10 in [7] is still valid in the present context.

Look at Step 3 in that proof. In the sequel, we will use the notation from [7] . Let

μ : = μ_{A, D},

H_{0} : = L_{h}^{2} (D),

H_{1} : =

the closure of

H_{0} |_{A}

L^{2} (A, μ)

and let

(b_{k})_{k = 1}^{\infty} \subset H_{0}

be the basis from Lemma 3.5.10 in [7] ,

ν_{k} : = ∥ b_{k} ∥_{H_{0}}, k \in N,

with the following property:

\begin{matrix} \sum_{k = 1}^{\infty} ν_{k}^{- ε} < \infty, ε > 0 . \end{matrix}

(8.8)

For any

w \in B,

we have

f (\cdot, w) \in H_{0}

and

f (\cdot, w) |_{A} \in H_{1} .

Hence

\begin{matrix} f (\cdot, w) = \sum_{k = 1}^{\infty} c_{k} (w) b_{k}, \end{matrix}

(8.9)

where

\begin{matrix} c_{k} (w) = \frac{1}{ν_{k}^{2}} \int_{D} f (z, w) \bar{b_{k} (z)} d Λ_{2 n} (z) = \int_{A} f (z, w) \bar{b_{k} (z)} d μ (z), k \in N . \end{matrix}

(8.10)

Taking the hypotheses (i)–(iii) into account and applying Lebesgue's Dominated Convergence Theorem, we see that the formula

\begin{matrix} \hat{c_{k}} (w) : = \int_{A} f (z, w) \bar{b_{k} (z)} d μ (z), w \in G \cup B, k \in N; \end{matrix}

(8.11)

defines a bounded function which is holomorphic in

G .

Moreover, by (iii) and ( 8.10 )–( 8.11 ) it follows that

\begin{matrix} {lim}_{w \to η, w \in A_{α} (η)} \hat{c_{k}} (w) = \hat{c_{k}} (η) = c_{k} (η), η \in B, 0 < α < \frac{π}{2} . \end{matrix}

(8.12)

Observe, as in [7] and using ( 8.10 )–( 8.12 ), that we obtain the following estimates

\begin{matrix} \frac{log | \hat{c_{k}} (w) |}{log ν_{k}} & \leq & \frac{log \sqrt{μ (A)}}{log ν_{k}}, w \in G, k \in N, \end{matrix}

\begin{matrix} {limsup}_{w \to η, w \in A_{α} (η)} \frac{log | \hat{c_{k}} (η) |}{log ν_{k}} & \leq & \frac{log \sqrt{Λ_{2 n} (D)}}{log ν_{k}} - 1, η \in B, 0 < α < \frac{π}{2}, k \in N, \end{matrix}

where

Λ_{2 n} (D)

is the volume of

D

with respect to the Lebesgue measure of

C^{n} .

This shows that for any

ε > 0,

there is a sufficiently large

N

such that for all

k \geq N,

\begin{matrix} \frac{log | \hat{c_{k}} |}{log ν_{k}} \leq ω_{δ} (\cdot, B, Ω) + ε - 1 on G . \end{matrix}

(8.13)

Take a compact set

K ⋐ D

and let

1 > α > {max}_{K} h_{A, D}^{*} .

Choose an

ε > 0

so small that

α + 2 ε < 1 .

Consider the open set

G_{K} : = {w \in G : ω_{δ} (\cdot, B, Ω) < 1 - α - 2 ε} .

By ( 8.13 ) there is a constant

C^{^{'}} (K)

such that

\begin{matrix} | \hat{c_{k}} |_{G_{k}} \leq C^{^{'}} (K) ν_{k}^{ω_{δ} (\cdot, B, Ω) + ε - 1} \leq C^{^{'}} (K) ν_{k}^{- α - ε}, k \geq 1 . \end{matrix}

(8.14)

Now we wish to show that

\begin{matrix} \sum_{k = 1}^{\infty} \hat{c_{k}} (w) b_{k} (z) \end{matrix}

(8.15)

converges locally uniformly in

{\hat{X}}^{o} .

Indeed, by ( 8.8 ), ( 8.14 ), and Lemma 3.5.10 in [7] ,

\begin{matrix} \sum_{k = 1}^{\infty} | \hat{c_{k}} |_{G_{K}} | b_{k} |_{K} \leq \sum_{k = 1}^{\infty} C^{^{'}} (K) ν_{k}^{- α - ε} C (K, α) ν_{k}^{α} \leq C^{^{'}} (K) C (K, α) \sum_{k = 1}^{\infty} ν_{k}^{- ε} < \infty, \end{matrix}

(8.16)

which gives the normal convergence on

K \times G_{K} .

Since the compact set

K

and

ε > 0

are arbitrary, the series in ( 8.15 ) converges uniformly on compact subsets of

{\hat{X}}^{o} .

Let

\hat{f}

denote this limit function in ( 8.15 ).

Fix

z_{0} \in D

and

η_{0} \in B .

We choose a compact

K

D

as above. Moreover, we may suppose without loss of generality that

K

contains a neighborhood of

z_{0} .

Let

ε_{0} > 0 .

In virtue of ( 8.16 ), there is an

N_{0}

such that

\begin{matrix} \sum_{k = N_{0} + 1}^{\infty} | \hat{c_{k}} |_{G_{K}} | b_{k} |_{K} < \frac{ε_{0}}{2} . \end{matrix}

(8.17)

On the other hand, in virtue of ( 8.9 )–( 8.12 ), we may find, for any

0 < α < \frac{π}{2},

an open neighborhood

V_{α}

η_{0}

such that

| \sum_{k = 1}^{N_{0}} \hat{c_{k}} (w) b_{k} (z) - \sum_{k = 1}^{N_{0}} c_{k} (η_{0}) b_{k} (z) | < \frac{ε_{0}}{2}, z \in K, w \in A_{α} (η_{0}) \cap V_{α} .

This, combined with ( 8.15 ) and ( 8.17 ), implies that

{limsup}_{z \to z_{0}, w \to η_{0}, w \in A_{α} (η_{0})} | \hat{f} (z, w) - f (z_{0}, η_{0}) | < ε_{0}, 0 < α < \frac{π}{2} .

Since

ε_{0} > 0

and

(z_{0}, η_{0}) \in D \times B

can be arbitrarily chosen, we conclude that

{lim}_{z \to z_{0}, w \to η_{0}, w \in A_{α} (η_{0})} \hat{f} (z, w) = f (z_{0}, η_{0}), (z_{0}, η_{0}) \in D \times B, 0 < α < \frac{π}{2} .

To complete Step II, it remains to show that

\hat{f} = f

A \times G .

To do this, fix an arbitrary

(z_{0}, w_{0}) \in A \times G .

Let

G

be the connected component of

G

containing

w_{0}

Recall that

G = {w \in Ω : ω (w, B, Ω) < 1 - δ} .

Then observe that both functions

\hat{f} (z_{0}, \cdot) |_{G}

and

f (z_{0}, \cdot) |_{G}

admit the same angular limit

f

B \cap G^{Ω} .

Consequently, applying Theorem 6.4 yields that

\hat{f} (z_{0}, \cdot) |_{G} = f (z_{0}, \cdot) |_{G} .

Hence,

\hat{f} (z_{0}, w_{0}) = f (z_{0}, w_{0}),

which proves the above assertion.

This completes the proof of Step II. It remains to prove the estimate

| \hat{f} |_{\hat{X}} \leq | f |_{X} .

In order to reach a contradiction assume that there is a point

z^{0} \in {\hat{X}}^{o}

such that

| \hat{f} (z^{0}) | > | f |_{X} .

Put

α : = \hat{f} (z^{0})

and consider the function

\begin{matrix} g (z) : = \frac{1}{f (z) - α}, z \in X . \end{matrix}

(8.18)

Using the above assumption, it can be checked that

g

satisfies hypotheses (i)–(iii) of Theorem 8.4 . Hence applying the first assertion of the theorem, there is exactly one function

\hat{g} \in O ({\hat{X}}^{o})

with

\hat{g} = g

A \times G .

Therefore, by ( 8.18 ) we have on

A \times G :

g (f - α) \equiv 1 .

Thus

\hat{g} (\hat{f} - α) \equiv 1

{\hat{X}}^{o} .

In particular,

0 = \hat{g} (z^{0}) (\hat{f} (z^{0}) - α) = 1,

a contradiction. Hence the inequality

| \hat{f} |_{\hat{X}} \leq | f |_{X}

is proved. □

Finally, we conclude this section with two uniqueness results.

Proposition 8.5. Let

D_{j} \subset C

be an open set and

A_{j}

a linearly measurable subset of

\partial D_{j}

such that

D_{j}

is locally rectifiable on

A_{j},

A_{j} \subset A_{j}^{*},

mes (A_{j}) > 0,

j = 1, 2 .

Let

{\tilde{D}}_{1} \subset C

be a domain,

D_{1} \cap {\tilde{D}}_{1} \neq \emptyset,

and let

{\tilde{A}}_{1}

be a linearly measurable subset of

\partial {\tilde{D}}_{1}

such that

{\tilde{D}}_{1}

is locally rectifiable on

{\tilde{A}}_{1},

and

mes ({\tilde{A}}_{1}) > 0 .

Put

\begin{matrix} {\hat{X}}^{o} & : = & {\hat{X}}^{o} (A_{1}, A_{2}; D_{1}, D_{2}), \end{matrix}

\begin{matrix} {\hat{\tilde{X}}}^{o} & : = & {\hat{X}}^{o} ({\tilde{A}}_{1}, A_{2}; {\tilde{D}}_{1}, D_{2},) . \end{matrix}

Let

\hat{f} \in O ({\hat{X}}^{o}),

\hat{\tilde{f}} \in O ({\hat{\tilde{X}}}^{o}),

and

z_{1}^{0} \in D_{1} \cap {\tilde{D}}_{1}

be such that both

\hat{f}

and

\hat{\tilde{f}}

admit the same angular limit at

(z_{1}^{0}, a_{2})

for a.e.

a_{2} \in A_{2} .

Then

\hat{f} (z) = \hat{\tilde{f}} (z)

for every

z = (z_{1}^{0}, z_{2}) \in {\hat{X}}^{o} \cap {\hat{\tilde{X}}}^{o} .

Proof. Fix an arbitrary $z_{2}^{0} \in D_{2}$ such that $z^{0} : = (z_{1}^{0}, z_{2}^{0}) \in {\hat{X}}^{o} \cap {\hat{\tilde{X}}}^{o} .$ Let $ε_{0} : = \frac{1}{4} (1 - max {ω (z_{1}^{0}, A_{1}, D_{1}), ω (z_{1}^{0}, {\tilde{A}}_{1}, {\tilde{D}}_{1})}),$ and put $D_{2, ε_{0}} : = {z_{2} \in D_{2} : ω (z_{2}, A_{2}, D_{2}) < 1 - ε_{0}} .$ Applying Theorem 6.4 to $\hat{f} |_{D_{2, ε_{0}}}$ and $\hat{\tilde{f}} |_{D_{2, ε_{0}}},$ it follows that $\hat{\tilde{f}} (z_{1}^{0}, z_{2}^{0}) = \hat{f} (z_{1}^{0}, z_{2}^{0}) .$ Hence, the proof is finished. □

Now we are able to prove the uniqueness stated in Theorem A.

Corollary 8.6. We keep the hypotheses and the notation of Theorem A. Then there is at most one function

\hat{f} \in O ({\hat{X}}^{o})

which satisfies Property 1) of Theorem A.

Proof. It follows immediately from Proposition 8.5 . □

9 Proof of Theorem A

We mainly consider the case

N = 2 .

In this case we proceed by four steps. Recall that by Corollary 8.6 , the function

\hat{f}

satisfying Part 1) is uniquely determined (if exists).

In the first two steps we mainly prove the following assertion:

There are a function

\hat{f} \in O ({\hat{X}}^{o})

and a subset

{\tilde{A}}_{j}

A_{j} \cap A_{j}^{*}

(j = 1, 2)

such that

mes (A_{j} \ {\tilde{A}}_{j}) = 0

and

\hat{f}

admits the angular limit

f

at every point of

({\tilde{A}}_{1} \times D_{2}) \cup (D_{1} \times {\tilde{A}}_{2}) .

(\binom{*}{* *})

In fact, in the first two steps we always assume that

| f |_{X} < \infty .

Using this and the previous observation and taking

(\binom{*}{* *})

for granted, we conclude the proof of Steps 1 and 2 below in exactly the same way as we did in Section 7 starting from Step 2 of that section.

Step 1: Proof of Theorem A for the case where

N = 2,

| f |_{X} < \infty,

and

D_{2}

is a rectifiable Jordan domain.

Proof of Step 1. In virtue of Proposition 8.2 , let

{a_{1 j}}_{j \in J_{1}}

be a finite or countable subset of

A_{1}

with the following properties:

$∙$ For any $j \in J_{1},$ there is an open neighborhood $U_{1 j}$ of $a_{1 j}$ such that $D_{1} \cap U_{1 j}$ is either a rectifiable Jordan domain or the disjoint union of two rectifiable Jordan domains (according to the type of $a_{1 j}$ );
$∙$ $A_{1} \subset \cup_{j \in J_{1}} U_{1 j} .$

For any

0 < δ < \frac{1}{2},

define

\begin{matrix} U_{1 j, δ} & : = {z_{1} \in D_{1} \cap U_{1 j} : ω (z_{1}, A_{1} \cap U_{1 j}, D_{1} \cap U_{1 j}) < δ}, j \in J_{1}, & A_{1, δ} & : = ⋃_{j \in J_{1}} U_{1 j, δ}, & D_{2, δ} & : = {z_{2} \in D_{2} : ω (z_{2}, A_{2}, D_{2}) < 1 - δ} . \end{matrix}

Moreover, for every

j \in J_{1}

let

\begin{matrix} \begin{matrix} X_{j} & : = X (\partial (D_{1} \cap U_{1 j}) \cap A_{1}, A_{2}; D_{1} \cap U_{1 j}, D_{2})), & {\hat{X_{j}}}^{o} & : = {\hat{X}}^{o} (\partial (D_{1} \cap U_{1 j}) \cap A_{1}, A_{2}; D_{1} \cap U_{1 j}, D_{2})), & {\tilde{f}}_{j} & : = f |_{X_{j}} . \end{matrix} \end{matrix}

(9.1)

Using the hypotheses on

f,

we conclude that

{\tilde{f}}_{j},

j \in J_{1},

satisfies (i)–(iii) of Theorem A. Moreover,

| {\tilde{f}}_{j} |_{X_{j}} \leq | f |_{X} < \infty .

Since

D_{2}

is a rectifiable Jordan domain and

D_{1} \cap U_{1 j},

j \in J_{1},

is either a rectifiable Jordan domain or the disjoint union of two rectifiable Jordan domains, we are able to apply the result of Section 7 to

{\tilde{f}}_{j} .

Consequently, we obtain, for

j \in J_{1},

a unique function

{\hat{f}}_{1 j} \in O ({\hat{X_{j}}}^{o}),

a subset

A_{1 j}

A_{1},

a subset

A_{2 j}

A_{2}

such that

\begin{matrix} \begin{matrix} mes (A_{1} \ A_{1 j}) = mes (A_{2} \ A_{2 j}) = 0, & {\hat{f}}_{1 j} & admits the angular limit f on ((\partial (D_{1} \cap U_{1 j}) \cap A_{1 j}) \times D_{2}) \cup (D_{1} \times A_{2 j}) . \end{matrix} \end{matrix}

(9.2)

Put

\begin{matrix} \begin{matrix} {\tilde{A}}_{1} & : = ⋂_{j \in J_{1}} A_{1 j} and {\tilde{A}}_{2} : = ⋂_{j \in J_{2}} A_{2 j}, & X_{δ} & : = X (A_{1, δ}, {\tilde{A}}_{2}; D_{1}, D_{2, δ}), & {\hat{X_{δ}}}^{o} & : = {\hat{X}}^{o} (A_{1, δ}, {\tilde{A}}_{2}; D_{1}, D_{2, δ}) . \end{matrix} \end{matrix}

(9.3)

In virtue of Proposition 8.5 , we are able to collect the family

{({\hat{f}}_{1 j} |_{U_{1 j, δ} \times D_{2, δ}})}_{j \in J_{1}}

in order to obtain a function

{\tilde{\tilde{f}}}_{δ} \in O (A_{1, δ} \times D_{2, δ}) .

Next, consider the function

{\tilde{f}}_{δ} : X_{δ} ⟶ C

given by

\begin{matrix} {\tilde{f}}_{δ} : = {\begin{matrix} {\tilde{\tilde{f}}}_{δ}, & on A_{1, δ} \times D_{2, δ} \\ f, & on D_{1} \times {\tilde{A}}_{2} \end{matrix} . \end{matrix}

(9.4)

In virtue of ( 9.1 )–( 9.4 ), we deduce that

\begin{matrix} mes (A_{1} \ {\tilde{A}}_{1}) = mes (A_{2} \ {\tilde{A}}_{2}) = 0, \end{matrix}

(9.5)

and

\begin{matrix} \begin{matrix} {lim}_{z_{1} \to z_{1}^{0}, z_{2} \to a_{2}^{0}, z_{2} \in A_{α} (a_{2}^{0})} {\tilde{f}}_{δ} (z) & = f (z_{1}^{0}, a_{2}^{0}), 0 < α < \frac{π}{2}, z_{1}^{0} \in D_{1}, a_{2}^{0} \in {\tilde{A}}_{2}, & {lim}_{z_{1} \to a_{1}^{0}, z_{1} \in A_{α} (a_{1}^{0}), z_{2} \to a_{2}^{0}} {\tilde{f}}_{δ} (z) & = f (a_{1}^{0}, z_{2}^{0}), 0 < α < \frac{π}{2}, a_{1}^{0} \in {\tilde{A}}_{1}, z_{2}^{0} \in D_{2, δ} . \end{matrix} \end{matrix}

(9.6)

In virtue of ( 9.4 )–( 9.6 ),

{\tilde{f}}_{δ}

satisfies the hypotheses (i)–(iii) of Theorem 8.4 . Applying this theorem to

{\tilde{f}}_{δ},

we obtain, for every

0 < δ < \frac{1}{2},

a function

{\hat{f}}_{δ} \in O ({\hat{X_{δ}}}^{o}) .

In virtue of ( 9.6 ), we see that

\begin{matrix} \begin{matrix} {\hat{f}}_{δ} & = {\tilde{f}}_{δ} on A_{1, δ} \times D_{2, δ}, & {lim}_{z_{1} \to z_{1}^{0}, z_{2} \to a_{2}^{0}, z_{2} \in A_{α} (a_{2}^{0})} {\hat{f}}_{δ} (z) & = f (z_{1}^{0}, a_{2}^{0}), 0 < α < \frac{π}{2}, z_{1}^{0} \in D_{1}, a_{2}^{0} \in {\tilde{A}}_{2}, & {lim}_{z_{1} \to a_{1}^{0}, z_{1} \in A_{α} (a_{1}^{0}), z_{2} \to a_{2}^{0}} {\hat{f}}_{δ} (z) & = f (a_{1}^{0}, z_{2}^{0}), 0 < α < \frac{π}{2}, a_{1}^{0} \in {\tilde{A}}_{1}, z_{2}^{0} \in D_{2, δ} . \end{matrix} \end{matrix}

(9.7)

We are now in a position to define the desired extension function

\hat{f} .

Indeed, one glues

{({\hat{f}}_{δ})}_{0 < δ < \frac{1}{2}}

together to obtain

\hat{f}

in the following way

\begin{matrix} \hat{f} : = {lim}_{δ \to 0} {\hat{f}}_{δ} on {\hat{X}}^{o} = {\hat{X}}^{o} (A_{1}, A_{2}; D_{1}, D_{2}) . \end{matrix}

(9.8)

Now one has to check that the limit ( 9.8 ) exists and possesses all the required properties. This will be an immediate consequence of the following

Lemma 9.1. For any point

z \in {\hat{X}}^{o}

put

\begin{matrix} δ_{z} : = \frac{1 - ω (z_{1}, A_{1}, D_{1}) - ω (z_{2}, A_{2}, D_{2})}{2} . \end{matrix}

(9.9)

Then

\hat{f} (z) = {\hat{f}}_{δ} (z)

for all

0 < δ \leq δ_{z} .

Proof of Lemma 9.1 . Fix an arbitrary point

z^{0} = (z_{1}^{0}, z_{2}^{0}) \in {\hat{X}}^{o} (A_{1}, A_{2}; D_{1}, D_{2})

and let

δ_{0} : = δ_{z^{0}} .

Let

0 < δ \leq δ_{0} .

Then, by Proposition 8.2 , we see that

ω (z_{2}^{0}, {\tilde{A}}_{2}, D_{2}) < 1 - δ_{0}

and

\begin{matrix} h_{A_{1, δ}, D_{1}}^{*} (z_{1}^{0}) + ω_{δ_{0}} (z_{2}^{0}, {\tilde{A}}_{2}, D_{2}) & \leq & ω (z_{1}^{0}, A_{1}, D_{1}) + \frac{ω (z_{2}^{0}, {\tilde{A}}_{2}, D_{2})}{1 - δ_{0}} \end{matrix}

\begin{matrix} \leq & \frac{ω (z_{1}^{0}, A_{1}, D_{1}) + ω (z_{2}^{0}, A_{2}, D_{2})}{1 - δ_{0}} < 1, \end{matrix}

where the latter estimate follows from formula ( 9.9 ). Consequently,

\begin{matrix} z^{0} \in {\hat{X}}^{o} (A_{1, δ}, {\tilde{A}}_{2}; D_{1}, D_{2, δ_{0}}) . \end{matrix}

(9.10)

On the other hand, using Part 1) of Proposition 8.2 , it is clear that

\begin{matrix} {\hat{X}}^{o} (A_{1, δ}, {\tilde{A}}_{2}; D_{1}, D_{2, δ_{0}}) \subset {\hat{X}}^{o} (A_{1, δ}, {\tilde{A}}_{2}; D_{1}, D_{2, δ}) \cap {\hat{X}}^{o} (A_{1, δ_{0}}, {\tilde{A}}_{2}; D_{1}, D_{2, δ_{0}}) . \end{matrix}

(9.11)

Moreover, in virtue of ( 9.4 ) and ( 9.7 ), we have

\begin{matrix} {\hat{f}}_{δ} = {\tilde{\tilde{f}}}_{δ} = {\hat{f}}_{δ_{0}} on A_{1, δ} \times D_{2, δ_{0}} . \end{matrix}

(9.12)

Next, let

G

be the connected component containing

z_{1}^{0}

of the following open set

{z_{1} \in D_{1} : h_{A_{1, δ}, D_{1}}^{*} (z_{1}) < 1 - ω_{δ_{0}} (z_{2}^{0}, A_{2}, D_{2})}

Observe that, in virtue of ( 9.10 )–( 9.11 ), both functions

{\hat{f}}_{δ} |_{G}

and

{\hat{f}}_{δ_{0}} |_{G}

are holomorphic and

G \cap A_{1, δ}

is a nonempty open set. Therefore, we deduce from ( 9.12 ) that

{\hat{f}}_{δ} = {\hat{f}}_{δ_{0}}

G .

Hence,

{\hat{f}}_{δ} (z^{0}) = {\hat{f}}_{δ_{0}} (z^{0}),

which completes the proof of the lemma.

□

We complete the proof as follows. An immediate consequence of Lemma 9.1 is that

\hat{f} \in O ({\hat{X}}^{o}) .

Next, we apply Lemma 9.1 and make use of ( 9.4 )–( 9.9 ) and of the fact that

{\hat{X_{δ}}}^{o} \to {\hat{X}}^{o}

δ ↘ 0 .

Consequently, we conclude that

\hat{f}

satisfies

(\binom{*}{* *}) .

Taking into account the remark made at the beginning of this section, the proof of the theorem in Step 1 is finished.

□

Step 2: Proof of Theorem A for the case

N = 2

and

| f |_{X} < \infty .

Proof of Step 2. We proceed using Step 1 in exactly the same way as we proved Step 1 using the result of Section 7. Hence, Step 2 is finished.

□

Step 3: Proof of Theorem A for the case when

N = 2

and

A_{1},

A_{2}

are compacts.

Proof of Step 3. Recall from the hypothesis (i) of Theorem A that

f

is locally bounded. Since

A_{1},

A_{2}

are compact, a compactness argument shows that there are a positive number

M_{0}

and an open neighborhood

U_{1}

(resp.

U_{2}

) of

A_{1}

(resp.

A_{2}

) such that

\begin{matrix} | f | < M_{0} on ([U_{1} \cap (D_{1} \cup A_{1})] \times A_{2}) ⋃ (A_{1} \times [U_{2} \cap (D_{2} \cup A_{2})]) . \end{matrix}

(9.13)

For

j \in {1, 2},

fix a sequence

{({\tilde{D}}_{j, k})}_{k = 1}^{\infty}

of subdomains of

D_{j}

such that

\begin{matrix} {\tilde{D}}_{j, k} ⋐ D_{j} and {\tilde{D}}_{j, k} ↗ D_{j} as k ↗ \infty . \end{matrix}

(9.14)

Consider the sequence

{(D_{j, k})}_{k = 1}^{\infty}

of open sets of

D_{j},

the sequence of crosses

{(X_{k})}_{k = 1}^{\infty},

and the sequence of the interior of their wedges

{({\hat{X}}_{k}^{o})}_{k = 1}^{\infty}

defined as

\begin{matrix} \begin{matrix} D_{j, k} & : = {\tilde{D}}_{j, k} \cup (U_{j} \cap D_{j}), k \geq 1, & X_{k} & : = X (A_{1}, A_{2}; D_{1, k}, D_{2, k}), & {\hat{X}}_{k}^{o} & : = {\hat{X}}^{o} (A_{1}, A_{2}; D_{1, k}, D_{2, k}) . \end{matrix} \end{matrix}

(9.15)

Since

f

is locally bounded,

A_{1},

A_{2}

are compact and, by ( 9.14 ),

{\tilde{D}}_{1, k} ⋐ D_{1},

{\tilde{D}}_{2, k} ⋐ D_{2},

a compactness argument shows that for every

k \geq 1,

there is a number

M_{k}

such that

M_{k} \geq M_{0}

and

| f | < M_{k} on ({\tilde{D}}_{1, k} \times A_{2}) \cup (A_{1} \times {\tilde{D}}_{2, k}) .

This, combined with ( 9.13 ) and ( 9.15 ), implies that

\begin{matrix} | f |_{X_{k}} < M_{k}, k \geq 1 . \end{matrix}

(9.16)

On the other hand, in virtue of ( 9.14 ) and ( 9.15 ), we see that the sequences

{(D_{j, k})}_{k = 1}^{\infty}

and

{(A_{j, k} : = A_{j})}_{k = 1}^{\infty}

satisfy the hypotheses of Proposition 5.15 .

Consequently, applying this proposition and taking ( 9.15 ) into account, we obtain

\begin{matrix} {\hat{X}}_{k}^{o} ↗ {\hat{X}}^{o} = {\hat{X}}^{o} (A_{1}, A_{2}; D_{1}, D_{2}) as k ↗ \infty . \end{matrix}

(9.17)

In virtue of ( 9.16 ) and the hypothesis on

f,

we see that

f |_{X_{k}}

satisfies the hypotheses of Step 2. Consequently, applying the result of Step 2 to

f |_{X_{k}},

k \geq 1,

we obtain a unique function

{\hat{f}}_{k} \in O ({\hat{X}}_{k}^{o})

and a subset

{\tilde{A}}_{1, k}

(resp.

{\tilde{A}}_{2, k}

) of

A_{1}

(resp.

A_{2}

) such that

\begin{matrix} \begin{matrix} mes (A_{1} \ {\tilde{A}}_{1, k}) = mes (A_{2} \ {\tilde{A}}_{2, k}) = 0, & {\hat{f}}_{k} & admits the angular limit f on (D_{1, k} \times {\tilde{A}}_{2, k}) \cup ({\tilde{A}}_{1, k} \times D_{2, k}) . \end{matrix} \end{matrix}

(9.18)

Put

\begin{matrix} {\tilde{A}}_{1} : = ⋂_{k = 1}^{\infty} {\tilde{A}}_{1, k} and {\tilde{A}}_{2} : = ⋂_{k = 1}^{\infty} {\tilde{A}}_{2, k} . \end{matrix}

(9.19)

Using ( 9.18 ), we deduce from ( 9.19 ) that

\begin{matrix} \begin{matrix} mes (A_{1} \ {\tilde{A}}_{1}) = mes (A_{2} \ {\tilde{A}}_{2}) = 0, & {\hat{f}}_{k} & admits the angular limit f on (D_{1, k} \times {\tilde{A}}_{2}) \cup ({\tilde{A}}_{1} \times D_{2, k}) . \end{matrix} \end{matrix}

(9.20)

Using ( 9.17 ), ( 9.20 ) and applying Corollary 8.6 , we obtain

\begin{matrix} {\hat{f}}_{k} = {\hat{f}}_{k + 1} on {\hat{X}}_{k}^{o}, k \geq 1 . \end{matrix}

(9.21)

Therefore, we may glue

{\hat{f}}_{k}

together in order to obtain the desired extension function

\hat{f}

\begin{matrix} \hat{f} = {lim}_{k \to \infty} {\hat{f}}_{k} on {\hat{X}}^{o} . \end{matrix}

(9.22)

In virtue of this formula and ( 9.17 ), ( 9.20 )–( 9.21 ), we conclude that

\hat{f}

satisfies

(\binom{*}{* *}) .

Hence, the proof of Part 1) of Theorem A is complete.

Using

(\binom{*}{* *})

and arguing as in Step 3 of Section 7, Part 2) of Theorem A follows. Now we turn to Part 3) of Theorem A. Let

(a_{1}^{0}, z_{2}^{0}) \in A_{1}^{*} \times D_{2}

be such that the following limit exists

λ : = {lim}_{(a_{1}, z_{2}) \to (a_{1}^{0}, z_{2}^{0}), (a_{1}, z_{2}) \in A_{1} \times D_{2}} f (a_{1}, z_{2}) .

We would like to show that

\hat{f}

admits the angular limit

λ

(a_{1}^{0}, z_{2}^{0}) .

Let

k_{0}

be a sufficiently large integer such that

z_{2}^{0} \in D_{2, k_{0}} .

Then in virtue of ( 9.17 ) and ( 9.20 )–( 9.21 ),

\begin{matrix} \hat{f} = {\hat{f}}_{k_{0}} on {\hat{X}}_{k_{0}}^{o} . \end{matrix}

(9.23)

In addition, applying Part 2) of Theorem A established in Step 2 and taking ( 9.16 ) into account, we see that

\begin{matrix} | {\hat{f}}_{k_{0}} |_{{\hat{X}}_{k_{0}}^{o}} \leq | f |_{X_{k_{0}}} < M_{k_{0}} < \infty . \end{matrix}

(9.24)

Consequently, the proof of Step 4 in Section 7 applied to

{\hat{f}}_{k_{0}}

still works in this context making the obviously necessary changes. Hence,

{\hat{f}}_{k_{0}}

and then

\hat{f}

by ( 9.23 ) admits the angular limit

λ

(a_{1}^{0}, z_{2}^{0}) .

This completes the proof of Part 3) of Theorem A. Using

(\binom{*}{* *})

and ( 9.23 )–( 9.24 ) (for a sufficiently large

k_{0}

), the proof of Parts 4) and 5) of Theorem A given in Steps 5 and 6 of Section 7 still works in this context making the obviously necessary changes. This completes Step 3.

□

Step 4: Proof of Theorem A for the case

N = 2 .

Proof of Step 4. For each

j \in {1, 2},

since

A_{j} \subset \partial D_{j}

is linearly measurable, one may find a sequence

{(A_{j, k})}_{k = 1}^{\infty}

of subsets of

A_{j}

such that

(a) $A_{j, k}$ is a compact subset of $\partial D_{j}$ and $mes (A_{j, k}) > 0,$ $k \geq 1;$
(b) $A_{j, k} \subset A_{j, k + 1},$ $k \geq 1;$
(c) $mes (A_{j} \ \cup_{k = 1}^{\infty} A_{j, k}) = 0 .$

For

k \geq 1,

let

\begin{matrix} \begin{matrix} X_{k} & : = X (A_{1, k}, A_{2, k}; D_{1}, D_{2}), & {\hat{X}}_{k}^{o} & : = {\hat{X}}^{o} (A_{1, k}, A_{2, k}; D_{1}, D_{2}) . \end{matrix} \end{matrix}

(9.25)

On the other hand, in virtue of (a)–(c), we see that the sequences

{(D_{j, k} : = D_{j})}_{k = 1}^{\infty}

and

{(A_{j, k})}_{k = 1}^{\infty}

satisfy the hypotheses of Proposition 5.15 . Consequently, applying this proposition and taking ( 9.25 ) into account, we obtain

\begin{matrix} {\hat{X}}_{k}^{o} ↗ {\hat{X}}^{o} = {\hat{X}}^{o} (A_{1}, A_{2}; D_{1}, D_{2}) as k ↗ \infty . \end{matrix}

(9.26)

In virtue of (a) and the hypothesis on

f,

we see that

f |_{X_{k}}

satisfies the hypotheses of Step 3. Consequently, applying the result of Step 3 to

f |_{X_{k}},

k \geq 1,

we obtain a unique function

{\hat{f}}_{k} \in O ({\hat{X}}_{k}^{o})

and a subset

{\tilde{A}}_{1, k}

(resp.

{\tilde{A}}_{2, k}

) of

A_{1}

(resp.

A_{2}

) such that

\begin{matrix} \begin{matrix} mes (A_{1} \ {\tilde{A}}_{1, k}) = mes (A_{2} \ {\tilde{A}}_{2, k}) = 0, & {\hat{f}}_{k} & admits the angular limit f on (D_{1} \times {\tilde{A}}_{2, k}) \cup ({\tilde{A}}_{1, k} \times D_{2}) . \end{matrix} \end{matrix}

(9.27)

Put

\begin{matrix} {\tilde{A}}_{1} : = ⋃_{k = 1}^{\infty} {\tilde{A}}_{1, k} and {\tilde{A}}_{2} : = ⋃_{k = 1}^{\infty} {\tilde{A}}_{2, k} . \end{matrix}

(9.28)

Using ( 9.27 ), we deduce from ( 9.28 ) that

\begin{matrix} \begin{matrix} mes (A_{1} \ {\tilde{A}}_{1}) = mes (A_{2} \ {\tilde{A}}_{2}) = 0, & {\hat{f}}_{k} & admits the angular limit f on (D_{1, k} \times {\tilde{A}}_{2}) \cup ({\tilde{A}}_{1} \times D_{2, k}) . \end{matrix} \end{matrix}

(9.29)

Applying ( 9.26 ), ( 9.29 ), and Corollary 8.6 , it follows that

\begin{matrix} {\hat{f}}_{k} = {\hat{f}}_{k + 1} on {\hat{X}}_{k}^{o}, k \geq 1 . \end{matrix}

(9.30)

Therefore, we may glue

{\hat{f}}_{k}

together in order to obtain the desired extension function

\hat{f}

\begin{matrix} \hat{f} = {lim}_{k \to \infty} {\hat{f}}_{k} on {\hat{X}}^{o} . \end{matrix}

(9.31)

In virtue of this formula and ( 9.26 ), ( 9.29 )–( 9.30 ), we conclude that

\hat{f}

satisfies

(\binom{*}{* *}) .

Hence, the proof of Part 1) of Theorem A is complete.

Using

(\binom{*}{* *})

and arguing as in Step 3 of Section 7, Part 2) of Theorem A follows.

The remaining parts of Theorem A can also be proved using Step 5 and 6 of Section 7 and making the obviously necessary changes.

This completes the last step of Theorem A in the case

N = 2 .

□

Finally, we present a sketch of the argument for the general case

N > 2 .

In order to prove Theorem A in its full generality, we use induction twice. More precisely, we proceed by induction (I) on

N \geq 2 .

Suppose the theorem is true for

N - 1 \geq 2 .

We have to discuss the case of an

N

-fold cross

X : = X (A_{1}, \dots, A_{N}; D_{1}, \dots, D_{N}),

where

D_{1}, \dots, D_{N} \subset C

and

A_{1}, \dots, A_{N}

are linearly measurable subsets of

\partial D_{1}, \dots, \partial D_{N}

such that

D_{j}

is locally rectifiable on

A_{j}

(1 \leq j \leq N) .

We proceed again by induction (II) on the positive integer

j

(0 \leq j \leq N)

such that there are at least

j

rectifiable Jordan domains among the open sets

D_{1}, \dots, D_{N} .

For

j = N

we are reduced to Section 7.

In fact, our proof follows essentially the scheme of the works in [14] , [12] , and that of Sections 7 and the previous proof for

N = 2 .

10 Proof of Theorem B

We will only give the proof of Theorem B for the case when

N = 2

and

D_{1},

D_{2}

are rectifiable Jordan domains. Since the general case can be proved using the scheme of Section 7 and 9, it is left to the interested reader. The proof is divided into two steps.

Step 1: Proof of Theorem B for the case when the slice functions

f (a_{1}, \cdot) |_{D_{2}}

and

f (\cdot, a_{2}) |_{D_{1}}

are bounded for every

a_{1} \in A_{1}

and

a_{2} \in A_{2} .

Proof of Step 1. For any

M \in N

let

\begin{matrix} A_{1 M} : = {a_{1} \in A_{1} : | f (a_{1}, \cdot) |_{D_{2}} \leq M} and A_{2 M} : = {a_{2} \in A_{2} : | f (\cdot, a_{2}) |_{D_{1}} \leq M} . \end{matrix}

(10.1)

Using the assumption of Step 1 and ( 10.1 ), we obtain

\begin{matrix} A_{j M} ↗ A_{j} as M ↗ \infty, j \in {1, 2} . \end{matrix}

(10.2)

Now we would like to show that for

j \in {1, 2},

M \in N,

\begin{matrix} A_{j M} is a closed subset of A_{j} and f |_{A_{j M} \times D_{k}} \in C (A_{j M} \times D_{k}), k \neq j, k \in {1, 2} . \end{matrix}

(10.3)

To do this, suppose without loss of generality that

j = 1,

k = 2,

and fix an arbitrary

M \in N

and a point

a_{1}^{0} \in A_{1} .

Let

(a_{1 n})_{n = 1}^{\infty}

be a sequence in

A_{1 M}

such that

{lim}_{n \to \infty} a_{1 n} = a_{1}^{0} .

Consequently, using the hypothesis (i), we see that

\begin{matrix} {lim}_{n \to \infty} f (a_{1 n}, t) = f (a_{1}^{0}, t), t \in A_{2} . \end{matrix}

(10.4)

On the other hand, it follows from the assumption

(a_{1 n})_{n = 1}^{\infty} \subset A_{1 M}

and the hypothesis of Step 1 that

| f (a_{1 n}, \cdot) |_{D_{2}} \leq M and | f (a_{1}^{0}, \cdot) |_{D_{2}} < \infty .

Combining this and ( 10.4 ), we are able to apply Lemma 4.4 to the sequence

{(f (a_{1 n}, \cdot) |_{D_{2}})}_{n = 0}^{\infty} \subset O (D_{2}),

where

a_{10} : = a_{1}^{0} .

Consequently, the sequence

{(f (a_{1 n}, \cdot))}_{n = 1}^{\infty}

converges uniformly on compact subsets of

D_{2}

f (a_{1}^{0}, \cdot) .

This completes the proof of ( 10.3 ).

On the other hand, by hypothesis (ii), the holomorphic function

f (a_{1}, \cdot)

admits the angular limit

f (a_{1}, a_{2})

a_{2} \in A_{2} .

Hence, it follows that

f |_{A_{1 M} \times A_{2 M}}

is measurable.

Moreover, by ( 10.1 ),

| f |_{X (A_{1 M}, A_{2 M}; D_{1}, D_{2})} \leq M

for every

M \in N .

In addition, in virtue of ( 10.2 ), there exists a sufficiently large integer

M_{0}

such that

mes (A_{j M}) > 0

for

j \in {1, 2}

and

M \geq M_{0} .

Consequently, we are in a position to apply Theorem A to the function

f

restricted to the cross

X (A_{1 M}, A_{2 M}; D_{1}, D_{2})

for

M \geq M_{0} .

Therefore, we obtain a function

{\hat{f}}_{M} \in O ({\hat{X}}^{o} (A_{1 M}, A_{2 M}; D_{1}, D_{2}))

and a subset

{\tilde{A}}_{1 M}

(resp.

{\tilde{A}}_{2 M}

) of

A_{1 M}

(resp.

A_{2 M}

) for

M \geq M_{0},

such that

\begin{matrix} \begin{matrix} mes (A_{1 M} \ {\tilde{A}}_{1 M}) = mes (A_{2 M} \ {\tilde{A}}_{2 M}) = 0, & {\hat{f}}_{M} & admits the angular limit f on ({\tilde{A}}_{1 M} \times D_{2}) ⋃ (D_{1} \times {\tilde{A}}_{2 M}) . \end{matrix} \end{matrix}

(10.5)

Put

\begin{matrix} {\tilde{A}}_{1} : = ⋃_{M = M_{0}}^{\infty} {\tilde{A}}_{1 M} and {\tilde{A}}_{2} : = ⋃_{M = M_{0}}^{\infty} {\tilde{A}}_{2 M} . \end{matrix}

(10.6)

Using ( 10.5 ), we deduce from ( 10.6 ) that

\begin{matrix} \begin{matrix} mes (A_{1} \ {\tilde{A}}_{1}) = mes (A_{2} \ {\tilde{A}}_{2}) = 0, & {\hat{f}}_{M} & admits the angular limit f on ({\tilde{A}}_{1} \times D_{2}) ⋃ (D_{1} \times {\tilde{A}}_{2}) . \end{matrix} \end{matrix}

(10.7)

Applying ( 10.2 ), ( 10.7 ), and Corollary 8.6 , we obtain

\begin{matrix} {\hat{f}}_{M} = {\hat{f}}_{M + 1} on {\hat{X}}^{o} ({\tilde{A}}_{1 M}, {\tilde{A}}_{1 M}; D_{1}, D_{2}), M \geq M_{0} . \end{matrix}

(10.8)

Therefore, we may glue the

{\hat{f}}_{M}

together to obtain the desired extension function

\hat{f}

\begin{matrix} \hat{f} = {lim}_{M \to \infty} {\hat{f}}_{M} on {\hat{X}}^{o} = {\hat{X}}^{o} (A_{1}, A_{2}; D_{1}, D_{2}) . \end{matrix}

(10.9)

Next, for every

j \in {1, 2}

and

M \geq M_{0},

in virtue of ( 10.2 )–( 10.3 ) and ( 10.5 ), one may find a sequence

(F_{j, M, n})_{n = 1}^{\infty}

of compact subsets of

\partial D_{j}

such that

\begin{matrix} \begin{matrix} F_{j, M, n} \subset F_{j, M, n + 1} \subset A_{j}, & mes (F_{j, M, n}) > 0, & mes ({\tilde{A}}_{j M} \ ⋃_{n = 1}^{\infty} F_{j, M, n}) = 0 . \end{matrix} \end{matrix}

(10.10)

Moreover, for any

k \in N,

k \geq 1,

and for any

m \in N,

put

\begin{matrix} \begin{matrix} A_{1 M n m k} & : = {a_{1} \in A_{1 M} : | f (a_{1}, ζ) - f (a_{1}, η) | \leq \frac{1}{2 k^{2}}, \forall ζ, η \in F_{2, M, n} : | ζ - η | < \frac{1}{m}}, & A_{2 M n m k} & : = {a_{2} \in A_{2 M} : | f (ζ, a_{2}) - f (η, a_{2}) | \leq \frac{1}{2 k^{2}}, \forall ζ, η \in F_{1, M, n} : | ζ - η | < \frac{1}{m}} . \end{matrix} \end{matrix}

(10.11)

Since, by hypothesis (i),

f \in C_{s} (A_{1} \times A_{2}),

we deduce from ( 10.10 ) and ( 10.11 ) that

A_{j M n m k}

is a closed subset of

A_{j M}

and

\begin{matrix} A_{j M n m k} ↗ A_{j M} as m ↗ \infty, j \in {1, 2}, k \geq 1 . \end{matrix}

(10.12)

Consequently, there is an

m_{0} : = m_{0} (M, n, k)

such that

mes (A_{j M n m k} \cap F_{1, M, n}) > 0

for any

m > m_{0} .

Now we are in a position to apply Theorem A to the function

f

restricted on the cross

X (A_{1 M n m k} \cap F_{1, M, n}, A_{2 M n m k} \cap F_{2, M, n}; D_{1}, D_{2}) .

Using ( 10.7 )–( 10.9 ) and Corollary 8.6 , we obtain exactly the function

\hat{f}

restricted to

{\hat{X}}^{o} (A_{1 M n m k} \cap F_{1, M, n}, A_{2 M n m k} \cap F_{2, M, n}; D_{1}, D_{2}) .

Let

\begin{matrix} {\tilde{A}}_{j M n m k} : = (A_{j M n m k} \cap F_{j, M, n}) \cap {(A_{j M n m k} \cap F_{j, M, n})}^{*}, \end{matrix}

(10.13)

where

T^{*}

denotes as usual the set of locally regular points relative to

T .

Taking ( 10.11 )–( 10.13 ) into account and arguing as in Step 5 of Section 7, we may show that

\begin{matrix} \begin{matrix} mes ({\tilde{A}}_{j M n m k} \ F_{j, n}) & = 0, & {limsup}_{z \to a, z_{1} \in A_{α} (a_{1}), z_{2} \in A_{α} (a_{2}), z \in {\hat{X}}^{o}} | \hat{f} (z) - f (a) | & < \frac{1}{k}, 0 < α < \frac{π}{2}, \end{matrix} \end{matrix}

(10.14)

for every

a = (a_{1}, a_{2}) \in {\tilde{A}}_{1 M n m k} \times {\tilde{A}}_{2 M n m k} .

Now it suffices to put

{\tilde{A}}_{j} : = ⋂_{k = 1}^{\infty} ⋃_{M = M_{0}}^{\infty} ⋃_{n = 1}^{\infty} ⋃_{m = m_{0} (M, n, k)}^{\infty} {\tilde{A}}_{j M n m k}, j \in {1, 2} .

Combining this and ( 10.14 ), ( 10.12 ), ( 10.2 ), we may check that all the conclusions 1)–3) of Theorem B are satisfied. Hence the proof is complete in this first step.

□

Step 2: The general case.

Proof of Step 2. Arguing as in Section 7, it suffices to treat the case when

D_{1} = D_{2} =

the unit disc

E .

We begin with the following

Definition 10.1. For every closed subset

F

\partial E

and any

n \in N,

n > 1,

define the following open set

Δ = Δ (F, n) : = ⋃_{ζ \in F} {z \in A_{\frac{π}{4}} (ζ) : | z | \geq 1 - \frac{1}{n}} \cup B (0, 1 - \frac{1}{n}) .

The reader should compare this definition with Definition 6.1 . Below we give a list of properties of such open sets.

Proposition 10.2. Let

F

be a closed subset of

\partial E .

1) Let

Δ (F, n)

be as in Definition 10.1 , then

Δ (F, n)

is a rectifiable Jordan domain and

F \subset \partial Δ (F, n) .

Δ (F, n) ↗ E

n ↗ \infty .

3) Consider a locally bounded function

f : E \cup F ⟶ C .

Then

| f |_{Δ (F, n)} < \infty

for every

n \in N,

n > 1 .

4) There holds the following equality

ω (z, F, E) = {lim}_{n \to \infty} ω (z, F, Δ (F, n)), z \in E .

Proof of Proposition 10.2 . Part 1) may be done as in the proof of Proposition 6.2 .

Part 2) is an immediate consequence of Definition 10.1 .

Part 3) follows immediately from the compactness of

F .

The proof of Proposition 5.15 still works in the context of Part 4) making the obviously necessary changes. This completes Part 4).

□

Now we are in a position to complete Step 2. Indeed, first suppose that both

A_{1}

and

A_{2}

are closed. Then for

j \in {1, 2},

consider the sequence of rectifiable Jordan domain

(D_{j n})_{n = 2}^{\infty}

given by

D_{j n} : = Δ (A_{j}, n), n \in N, n > 1 .

For

n \in N,

n > 1,

let

f_{n} : = f |_{X (A_{1}, A_{2}; D_{1 n}, D_{2 n})} .

In virtue of Proposition 10.2 , we are able to apply the result of Step 1 to

f_{n} .

Consequently, we obtain a function

{\hat{f}}_{n} \in {\hat{X}}^{o} (A_{1}, A_{2}; D_{1 n}, D_{2 n}) .

Therefore, we may glue

{\hat{f}}_{n}

together in order to obtain the desired extension function

\hat{f}

\hat{f} = {lim}_{n \to \infty} {\hat{f}}_{n} on {\hat{X}}^{o} = {\hat{X}}^{o} (A_{1}, A_{2}; D_{1}, D_{2}) .

Because of Proposition 10.2 , we can show that

\hat{f}

possesses all the assertions of Theorem B. The case when

A_{1}

and

A_{2}

are only measurable is similar. It suffices to find a sequence

(A_{j n})_{n = 1}^{\infty}

of subsets of

A_{j}

such that

A_{j n}

is compact and

mes (A_{j} \ \cup_{n = 1}^{\infty} A_{j n}) = 0 .

Then we may apply the previous discussion to

f |_{X (A_{1 n}, A_{2 n}; D_{1}, D_{2})}

in order to obtain a function

{\hat{f}}_{n} \in {\hat{X}}^{o} (A_{1 n}, A_{2 n}; D_{1}, D_{2}) .

Finally, the desired extension function

\hat{f}

is defined by

\hat{f} = {lim}_{n \to \infty} {\hat{f}}_{n} on {\hat{X}}^{o} = {\hat{X}}^{o} (A_{1}, A_{2}; D_{1}, D_{2}) .

This completes the proof in this last step.

□

11 Examples and Concluding remarks

The following examples of Drużkowski [2] show the optimality of Theorem A and B. Consider

N = 2,

D_{1} = D_{2} = E,

A_{1} = A_{2} = {t \in \partial E : Re t > 0},

A : = A_{1} \times A_{2},

X : = X (A_{1}, A_{2}; D_{1}, D_{2}),

and

Y : = (D_{1} \cup A_{1}) \times (D_{2} \cup A_{2}) .

Example 1. Define a function

h : Y ⟶ C

as follows

h (z) : = {\begin{matrix} exp (- [Log (1 - z_{1}) + Log (1 - z_{2})] Log \frac{2 + z_{1} z_{2}}{3}), & z_{1} \neq 1, z_{2} \neq 1 \\ 0, & z_{1} = 1 or z_{2} = 1 \end{matrix} .

where

Log

is the principal branch of logarithm.

Put

f : = h |_{X} .

As in [2] observe that

f

is measurable,

f \in C_{s} (X) \cap O_{s} (X^{o}),

| f |_{X} < \infty,

but

f |_{A}

is not continuous at

(1, 1) .

Since

h |_{{\hat{X}}^{o}} \in O ({\hat{X}}^{o}),

using the uniqueness established in Theorem A, we conclude that the solution

\hat{f}

provided by Theorem A and B satisfies

\hat{f} = h |_{{\hat{X}}^{o}} .

In addition, we see that, for

0 < α < \frac{π}{2},

the angular limit of

\hat{f}

(1, 1)

does not exist. Thus the condition in assertion 3) of Theorem A is necessary. Moreover, the sets

{\tilde{A}}_{1},

{\tilde{A}}_{2}

given by Theorem B do depend on

f .

Example 2. Define a function

h : Y ⟶ C

as follows

h (z) : = {\begin{matrix} exp (- (z_{1} - λ) {Log}^{2} \frac{3 + z_{2}}{1 - z_{2}}), & z_{2} \neq 1 \\ 0, & z_{2} = 1 \end{matrix} .

where

z \in Y,

0 < λ \leq \frac{\sqrt{2}}{2} .

Define

f : = h |_{X} .

Then

\hat{f} = h |_{{\hat{X}}^{o}} .

As in [2] observe that

f |_{A}

is continuous,

f \in C_{s} (X) \cap O_{s} (X^{o}),

but

f

is not locally bounded on

X .

In addition, for

\frac{π}{3} < α < \frac{π}{2},

consider the functions

z_{1}^{α, λ}, z_{2}^{α} : [0, 1] \to C

given by

\begin{matrix} z_{2}^{α} (t) & : = & 1 + t e^{i (π - \frac{9 α}{10})}, \end{matrix}

\begin{matrix} z_{1}^{α, λ} (t) & : = & λ + {(Re {Log}^{2} \frac{3 + z_{2}^{α} (t)}{1 - z_{2}^{α} (t)})}^{- 1} + i λ, t \in [0, 1] . \end{matrix}

We may prove that there is an

t_{α, λ} > 0

and a neighborhood

U_{α, λ}

λ + i λ

C

such that

(z_{1}^{α, λ} (t), z_{2}^{α} (t)) \in {\begin{matrix} ((A_{α} (λ + i λ) \cap U_{α, λ}) \times A_{α} (1)) \cap {\hat{X}}^{o}, & 0 < t < t_{α, λ}, λ = \frac{\sqrt{2}}{2} \\ (U_{α, λ} \times A_{α} (1)) \cap {\hat{X}}^{o}, & 0 < t < t_{α, λ}, 0 < λ < \frac{\sqrt{2}}{2} \end{matrix} .

In addition, it can be checked that

{lim}_{t \to 0} (z_{1}^{α, λ} (t), z_{2}^{α} (t)) = (λ + i λ, 1) and {lim}_{t \to 0} | \hat{f} (z_{1}^{α, λ} (t), z_{2}^{α} (t)) | = \infty .

This shows that the assumption of the local boundedness on

f

is necessary in Theorem A. Finally, we conclude the article by some remarks and open questions.

1. We may generalize Theorems A and B in the following directions: Let

D \subset C

be an open set.

1a. We extend $\partial^{*} D$ to the set of all points $ζ \in \partial D$ where $\partial D$ admits a corner at $ζ$ (see [13] ). Note that there are at most countably many points where $\partial D$ admits a corner but not a tangent (see Exercises 3.4 in [13] ).
1b. We extend the notion type of a point in $\partial D$ as follows. A point $ζ \in \partial D$ is said to be of type $m$ ( $m \geq 1$ ) if there are an open neighborhood $U$ of $ζ$ and $m$ disjoint rectifiable Jordan domains $U_{1}, \dots, U_{m}$ such that $U \cap D = \cup_{j = 1}^{m} U_{j} .$
1c. Instead of the notion local rectifiability, we use the weaker notion local Jordan curve. More precisely, $D$ is said to be locally Jordan curve like at a point $ζ \in \partial D$ if there is a neighborhood $U$ of $ζ$ in $C$ such that $U \cap \partial D$ is the union of $m$ disjoint Jordan curves and the interior of at least one of them contains $ζ .$

Since the corresponding statements of Theorems A and B under these generalizations are somewhat complicated and their proofs do not require any new method, we leave the interested reader to develop these ideas.

2. It may be proved that

{\hat{X}}^{o}

provided by Theorem A is the maximal domain of holomorphic extension of the function

f .

We postpone the proof of this result to an ongoing work.

3. It seems to be of interest to consider Theorem A and B under the following general settings: Let

G_{j}

be a complex manifold of dimension

d_{j}

and

D_{j} ⋐ G_{j}

an open set,

j = 1, \dots, N .

Let

A_{j}

be a subset of positive

d_{j}

-dimensional Hausdorff measure of

M_{j},

where

M_{j}

is a real

d_{j}

-dimensional generating submanifold contained in

\partial D_{j},

j = 1, \dots, N,

etc. We postpone this issue to an ongoing work.

4. Does Theorem A still hold if we omit the assumption (ii) ”

f |_{A}

is measurable”?5. Does Theorem B still hold if we omit the assumption that

f |_{A} \in C_{s} (A)

References

L. Aizenberg, Carleman's formulas in complex analysis. Theory and applications, Transl. from the Russian, updated, enlarged and revised. (English) Mathematics and its Applications (Dordrecht). 244. Dordrecht: Kluwer Academic Publishers. xx, 299 p. (1993).
L. M. Drużkowski, A generalization of the Malgrange–Zerner theorem, Ann. Polon. Math., 38, (1980), 181–186.
H. Epstein, Generalization of the ”Edge of the Wedge” Theorem, J. Math. Phys., 1, (1960), 524–531.
G. M. Goluzin, Geometric theory of functions of a complex variable, (English), Providence, R. I.:American Mathematical Society (AMS). VI, (1969), 676 p.
A. A. Gonchar, On analytic continuation from the ”edge of the wedge” theorem, Ann. Acad. Sci. Fenn. Ser. A.I: Mathematica, 10, (1985), 221–225.
A. A. Gonchar, On Bogolyubov's ”edge of the wedge” theorem, Proc. Steklov Inst. Math., 228, (2000), 18–24.
M. Jarnicki, P. Pflug, Extension of Holomorphic Functions, de Gruyter Expositions in Mathematics 34, Walter de Gruyter, 2000.
M. Jarnicki, P. Pflug, An extension theorem for separately holomorphic functions with analytic singularities, Ann. Pol. Math., 80, (2003), 143–161.
M. Klimek, Pluripotential theory, London Mathematical society monographs, Oxford Univ. Press., 6, (1991).
H. Komatsu, A local version of Bochner's tube theorem, J. Fac. Sci., Univ. Tokyo, Sect. I A 19, (1972), 201–214.
P. Koosis, Introduction to $H_{p}$ spaces. With an appendix on Wolff 's proof of the corona theorem, London Mathematical Society Lecture Note Series. 40. Cambridge etc.: Cambridge University Press. XV, 376 p. (1980). (1991), 183–194.
V.-A. Nguyên, A general version of the Hartogs extension theorem for separately holomorphic mappings between complex analytic spaces, preprint of the University of Oldenburg (2004), 33 p.
Ch. Pommerenke, Boundary behaviour of conformal maps, Grundlehren der Mathematischen Wissenschaften, 299, Berlin: Springer-Verlag, 300 p. (1992).
P. Pflug and V.-A. Nguyên, A boundary cross theorem for separately holomorphic functions, Ann. Polon. Math., 84, (2004), no. 3, 237–271.
T. Ransford, Potential theory in the complex plane, London Mathematical Society Student Texts, 28, Cambridge: Univ. Press., (1995).
W. Rudin, Lectures on the edge-of-the-wedge theorem, (English) Conference Board of the Mathematical Sciences. Regional Conference Series in Mathematics. No.6. Providence, R.I.:American Mathematical Society (1971).
W. Rudin, Function theory in the unit ball of $C^{n}$ , (English) Grundlehren der Mathematischen Wissenschaften 241, Springer-Verlag, New York-Berlin, 1980.
W. Rudin, Function theory in polydiscs (English) Mathematics Lecture Note Series. New York-Amsterdam: W.A. Benjamin, Inc., 188 p. (1969).
E. M. Stein, Boundary behavior of holomorphic functions of several complex variables, (English) Mathematical Notes. Princeton, N. J.: Princeton University Press. IX, 72 p. (1972).
V. S. Vladimirov, Methods of the Theory of Functions of Several Complex Variables, (Russian), Moscow: 'Nauka' Publisher, 411 p. (1964).
M. Zerner, Quelques résultats sur le prolongement analytique des fonctions de variables complexes, Séminaire de Physique Mathématique.

Peter Pflug, Carl von Ossietzky Universitat Oldenburg, Fachbereich Mathematik, Postfach 2503, D–26111, Oldenburg, Germany E-mail address : pflug@mathematik.uni-oldenburg.de Vie t-Anh Nguyen, Carl von Ossietzky Universitat Oldenburg, Fachbereich Mathematik, Postfach 2503, D–26111, Oldenburg, Germany E-mail address : nguyen@mathematik.uni-oldenburg.de

1 December 2004

Peter Pflug

Viê ̣ t-Anh Nguyên