ON A PROPERTY OF HARMONIC MEASURE ON SIMPLY CONNECTED

DOMAINS

CHRISTINA KARAFYLLIA
arXiv:1909.00026v1 [math.CV] 30 Aug 2019

Abstract. Let D ⊂ C be a domain with 0 ∈ D and, for R > 0, let ω̂D (R) denote the harmonic
measure of D ∩{|z| = R} at 0 and ωD (R) denote the harmonic measure of ∂D ∩{|z| ≥ R} at 0. The
behavior of the functions ωD and ω̂D near ∞ determines (in some sense) how large D is. However,
it is not known whether the functions ωD and ω̂D always have the same behavior when R tends
to ∞. Obviously, ωD (R) ≤ ω̂D (R) for every R > 0. Thus, the arising question, first posed by
Betsakos, is the following: Does there exist a positive constant C such that for all simply connected
domains D with 0 ∈ D and all R > 0,
ωD (R) ≥ C ω̂D (R)?
In general, we prove that the answer is negative by means of two different counter-examples. How-
ever, under additional assumptions involving the geometry of D, we prove that the answer is
positive. We also find the value of the optimal constant for starlike domains.

1 Introduction
We will give an answer to a question of Betsakos ([6, p. 788]) about a property of harmonic
measure. For a domain D, a point z ∈ D and a Borel subset E of D, let ωD (z, E) denote the
harmonic measure at z of E with respect to the component of D\E containing z. The function
ωD (·, E) is exactly the solution of the generalized Dirichlet problem with boundary data ϕ = 1E
(see [1, ch. 3], [11, ch. 1] and [24, ch. 4]).
Let D ⊂ C be a domain with 0 ∈ D. For R > 0, we set
ωD (R) = ωD (0, ∂D ∩ {z : |z| ≥ R})
and
ω̂D (R) = ωD (0, {z : |z| = R}) .
The behavior of the functions ωD and ω̂D near ∞ determines (in some sense) how large D is and it
has been studied from various viewpoints. For example, in [28] and [29, p. 111-118] Tsuji proved
bounds for the growth of ω̂D (R) in terms of the size of the maximal arcs on {z : |z| = R}. Tsuji’s
inequalities can be used to obtain estimates for the maximum modulus, means and coefficients of
various classes of p−valent functions (see also [12, ch. 8]). In [13] Hayman and Weitsman used
ω̂D (R) to estimate the means and hence the coefficients of functions when information is known
about their value distribution. With the aid of ωD (R) and ω̂D (R), Sakai [25] gave an integral
representation of the least harmonic majorant of |x|p in an open subset D of Rn with 0 ∈ D and
proved isoperimetric inequalities for it. Essén, Haliste, Lewis and Shea ([9], [10]) also studied the
problem of harmonic majoration in higher dimensions in terms of the geometry of D by using
ωD (R) and ω̂D (R). In [26, p. 1348] Solynin proved an estimate of ωD (R) when D = f (D) and
f is in the class S of functions which are regular and univalent in the unit disk and f (0) = 0,
f 0 (0) = 1. Baernstein [2] proved an integral formula involving ω̂D (R) and Green’s function.
2010 Mathematics Subject Classification. Primary 30C85; Secondary 30F45, 30C35, 31A15.
Key words and phrases. Harmonic measure, conformal mapping, hyperbolic distance, extremal length.
I thank Professor D. Betsakos, my thesis advisor, for his advice during the preparation of this work, and the
Onassis Foundation for the scholarship I receive during my Ph.D. studies.
1
2 On a property of harmonic measure on simply connected domains
In [8] Essén proved that every analytic function f : D → D belongs to the Hardy space H p for
some p > 0 if and only if for some constants q and C, we have ω̂D (R) ≤ CR−q for every R ≥ 1.
With the aid of Essén’s result, Kim and Sugawa [16] proved that the Hardy number, h (D), of a
plane domain D with 0 ∈ D, can be determined by
log ω̂D (R)
h (D) = − lim sup .
R→+∞ log R
In [5] Betsakos studied another problem involving ωD (R). Let B be the family of all simply
connected domains D ⊂ C such that 0 ∈ D and there is no disk of radius larger than 1 contained
in D. It is obvious that if D ∈ B then ωD (R) is a decreasing function of R. In fact, ωD decays
exponentially as it is proved that there exist positive constants β and C such that ωD (R) ≤ Ce−βR ,
for every D ∈ B and every R > 0. The problem studied in [5] is to find the optimal exponent β.
Poggi-Corradini (see [19, p. 33-34], [20], [21]) studied ωD (R) and ω̂D (R) in relation with con-
formal mappings in Hardy spaces. In fact, if D is an unbounded simply connected domain with
0 ∈ D and ψ is a conformal mapping of D onto D, then he proved that
Z +∞ Z +∞
p p−1
ψ ∈ H (D) ⇔ R ωD (R) dR < +∞ ⇔ Rp−1 ω̂D (R) dR < +∞.
0 0
To establish the last equivalence, Poggi-Corradini first proved that there exists a constant M0 > 1
such that for all R > 0,
1
(1.1) ωD (R) ≥ ω̂D (M0 R) .
2
All the results mentioned above are some of the estimates and applications of ωD and ω̂D that
have been made over time. However, it is still unknown whether the functions ωD and ω̂D always
have the same behavior when R tends to ∞. Obviously, by the maximum principle, for every R > 0,
ωD (R) ≤ ω̂D (R)
but all we know about the inverse inequality is (1.1). Thus, a natural question, first posed in [6, p.
788] by Betsakos, is the following:
Question 1.1. Does there exist a positive constant C such that for a class of domains D (such as
simply connected, starlike etc.) with 0 ∈ D and every R > 0,
ωD (R) ≥ C ω̂D (R)?
In this paper we prove that for simply connected domains the answer is negative by means of two
different counter-examples. However, under additional assumptions involving the geometry of the
domains, we prove that the answer is positive and we also find the value of the optimal constant
for starlike domains.
Before presenting our results, we need to introduce some notation. Let D be an unbounded simply
connected domain with 0 ∈ D. The Riemann mapping theorem implies that there exists a conformal
mapping ψ from D onto D with ψ (0) = 0. For R > 0, we set FR = {z ∈ D : |ψ (z)| = R} and
ER = {ζ ∈ ∂D : |ψ (ζ)| ≥ R}, that is, ψ (FR ) = D ∩ {z : |z| = R} and ψ (ER ) = ∂D ∩ {z : |z| ≥ R}.
Note that since ψ (FR ) is a countable union of open arcs in D which are the intersection of D with
the circle {z : |z| = R}, then the preimage of every such arc is also an arc in D with two distinct
endpoints on ∂D (see Proposition 2.14 [22, p. 29]).
So, if N (R) ∈ N ∪ {+∞} denotes the number of components of FR and FRi denotes each of these
components for i = 1, 2, . . . , N (R), then we set ΓiR be the hyperbolic geodesic joining the endpoints
of FRi in D and CR i be the arc of ∂D joining the endpoints of Γi and lying on the boundary of the
R
component of D\ΓiR which does not contain the origin (see Fig. 1).
 Christina Karafyllia 3

Figure 1. Figure 2.

With the notation above, in Section 3, we construct the simply connected domain D of Fig. 2
and prove that there exists a sequence of positive numbers {Rn }n∈N such that

ω̂D (Rn )
lim = +∞
n→+∞ ωD (Rn )

which implies that there does not exist a positive constant C such that ωD (R) ≥ C ω̂D (R) for
every R > 0. As we see in the proof, this result is due to the fact that there does not exist a
positive constant c such that ψ (FRn ) ⊂ {z ∈ D : dD (z, ψ (ΓRn )) < c} for every n ∈ N. Note that
dD (z, ψ (ΓRn )) denotes the hyperbolic distance between z and ψ (ΓRn ) in D which we define in
Section 2. Now we consider the following condition on the simply connected domain D:

Condition (1). There exists a positive constant c such that ψ FRi ⊂ z ∈ D : dD z, ψ ΓiR < c



for every R > 0 and every i ∈ {1, 2, . . . , N (R)}.

This condition means that every arc of D ∩ {z : |z| = R} lies in a hyperbolic neighborhood of
the geodesic joining its endpoints. The arising question is whether the answer to the Question 1.1
is positive for simply connected domains that satisfy Condition (1). However, we prove that this
condition is not enough by constructing, in Section 4, the simply connected domain D of Fig. 3
which comes from a small variation of the domain of Fig. 2. In fact, there exists a sequence of
positive numbers {Rn }n∈N such that, despite the fact that Condition (1) is satisfied, we have again

ω̂D (Rn )
lim = +∞.
n→+∞ ωD (Rn )

∗ ∗ corresponds


This time, this is due to the fact that for every n ∈ N, ψ CR n
6= ψ (ERn ), where CR n
∗
to FRn which denotes the component of FRn such that its image under the mapping ψ intersects
the real axis, that is, ψ FR∗ n ∩ R 6= ∅. So, we consider the following condition:

i


Condition (2). For every R > 0 and every i ∈ {1, 2, . . . , N (R)}, ψ CR coincides with a compo-
nent of ψ (ER ).
4 On a property of harmonic measure on simply connected domains

Figure 3.

This condition means that there does not exist any prime end P of ∂D that is inside the disk
{z : |z| < R} but every arc in D joining 0 to P intersects the circle {z : |z| = R}. Note that in the
first counter-example (Section 3) Condition (2) is satisfied, since for every n ∈ N, ψ (CRn ) = ψ (ERn )
and hence there do not exist such prime ends. However, in the second counter-example (Section
4), there exists prime end P of ∂D that is inside the disk {z : |z| < Rn } but every arc in D joining
0 to P intersects the circle {z : |z| = Rn }. See, for example, the prime end P in Fig. 3. These
two counter-examples show that Conditions (1) and (2) are necessary if we want to give a positive
answer to the Question 1.1. But are they enough? In Section 5, we actually prove that if a simply
connected domain satisfies Conditions (1) and (2), then there exists a positive constant K = K (c)
such that for every R > 0,
ω̂D (R) ≤ KωD (R) .
Moreover, we prove that we can find the value of this constant if we retain Condition (2) and
replace Condition (1) with the following condition:
Condition (3). For every R > 0 and every i ∈ {1, 2, . . . , N (R)}, ψ ΓiR ⊂ D ∩ {z : |z| ≤ R}.

This condition means that for every arc of D ∩ {z : |z| = R}, the hyperbolic geodesic joining its
endpoints lies entirely in D ∩ {z : |z| ≤ R}.
So, having these results in mind, in Section 5, we prove the theorem below which gives a positive
answer to the Question 1.1. First, we remind the notation. Let D be an unbounded simply
connected domain with 0 ∈ D and ψ be a Riemann mapping of D onto D with ψ (0) = 0. For R > 0,
we set FR = {z ∈ D : |ψ (z)| = R} and ER = {ζ ∈ ∂D : |ψ (ζ)| ≥ R}. If N (R) ∈ N ∪ {+∞} denotes
the number of components of FR and FRi denotes each of these components for i = 1, 2, . . . , N (R),
then, for i = 1, 2, . . . , N (R), we set ΓiR be the hyperbolic geodesic joining the endpoints of FRi in D
and CR i be the arc of ∂D joining the endpoints of Γi and lying on the boundary of the component
R
of D\ΓiR which does not contain the origin.
Theorem 1.1. With the notation above, if Conditions (1) and (2) are satisfied, then there exists
a positive constant K = K (c) such that for every R > 0,
ω̂D (R) ≤ KωD (R) .
If Conditions (2) and (3) are satisfied, then for every R > 0,
ω̂D (R) ≤ 2ωD (R) .
 Christina Karafyllia 5

Finally, recall that a domain D in C is called starlike with respect to 0, if for every point z ∈ D,
the segment of the straight line from 0 to z, [0, z], lies entirely in D. In Section 6, we prove that
starlike domains satisfy Conditions (2) and (3) and that 2 is the optimal constant:
Theorem 1.2. Let D be a starlike domain in C. Then for every R > 0,
ω̂D (R) ≤ 2ωD (R)
and the constant 2 is best possible.
In Section 2, we introduce some preliminaries such as notions and results in hyperbolic geometry,
basic properties of harmonic measure and extremal length and relations between them. In Sections
3 and 4, we present the counter-examples of Fig. 2 and 3 respectively, and in Sections 5 and 6, we
prove Theorems 1.1 and 1.2 respectively.

2 Preliminary results
2.1 Results in hyperbolic geometry
For the unit disk D the density of the hyperbolic metric is
2
λD (z) = .
1 − |z|2
If D is a simply connected domain with D 6= C and f be a conformal mapping of D onto D then
the density λD is determined from
2
λD (f (z)) f 0 (z) =
 
1 − |z|2
(see [1, ch. 1], [3, p. 11-28], [18, p. 236]). The hyperbolic distance between two points z, w in D is
defined by  
 z−w 
1 +  1−z w̄ 
dD (z, w) = log  z−w  .
 
1 −  1−z w̄ 
It is conformally invariant and thus it can be defined on any simply connected domain D 6= C as fol-
lows: If f is a Riemann mapping of D onto D and z, w ∈ D, then dD (z, w) = dD f −1 (z) , f −1 (w) .

Also, for a set E ⊂ D, we define dD (z, E) := inf {dD (z, w) : w ∈ E}.

The following theorem is known as Minda’s reflection principle [18, p. 241]. First, we introduce
some notation: If Γ is a straight line (or circle), then R is one of the half-planes (or the disk)
determined by Γ and Ω∗ is the reflection of a hyperbolic region Ω in Γ .
Theorem 2.1. Let Ω be a hyperbolic region in C and Γ be a straight line or circle with Ω ∩ Γ 6= ∅.
If Ω\R ⊂ Ω∗ , then
λΩ∗ (z) ≤ λΩ (z)
for all z ∈ Ω\R. Equality holds if and only if Ω is symmetric about Γ.
A generalization of Theorem 2.1 was proved by Solynin in [27]. Next we present a result of
Beardon and Carne concerning the hyperbolic geodesic joining two points in a simply connected
domain [4, p. 206].
Theorem 2.2. Let D1 , D2 be two simply connected, hyperbolic domains that satisfy the following
conditions:
(1) The domain D2 is a Jordan domain with boundary ∂D2 consisting of a closed arc α, joining
the endpoints a0 and a1 , and the complementary arc β.
(2) The domain D2 is a subset of D1 with α ⊂ ∂D1 and β ⊂ D1 .
6 On a property of harmonic measure on simply connected domains

Then the geodesic γ1 from a0 to a1 in D1 and the geodesic γ2 from a0 to a1 in D2 do not meet in
D2 . Indeed, γ2 separates γ1 and α (see Fig. 4).
Remark 2.1. It follows from the proof that the theorem is still valid if D1 , D2 are simply connected
domains and α, β are arcs of prime ends.

Figure 4.

2.2 Quasi-hyperbolic distance

The hyperbolic distance between z1 , z2 ∈ D can be estimated by the quasi-hyperbolic distance,
δD (z1 , z2 ), which is defined by
|dz|
Z
δD (z1 , z2 ) = inf ,
γ:z1 →z2 γ d (z, ∂D)

where the infimum ranges over all the paths connecting z1 to z2 in D and d (z, ∂D) denotes the
euclidean distance of z from ∂D. Then it is proved that (1/2) δD ≤ dD ≤ 2δD (see [3, p. 33-36],
[19, p. 8]).

2.3 Extremal length

Before we present the definition of extremal length (see [11, ch. 4], [1, ch. 4] and [7, p. 361-385]),
we need to define some notions. A path family in a domain D is a non-empty set Γ of countable
unions of rectifiable arcs in D. An element γ ∈ Γ is called a curve even though γ may not be
connected and may have many self-intersections. By definition, a metric is a non-negative Borel
measurable function ρ on D such that the ρ-area,
Z Z
A (D, ρ) = ρ2 dxdy,
D
satisfies 0 < A (D, ρ) < +∞. When ρ is a metric and Γ is a path family, we define the ρ-length of
Γ by Z
L (Γ, ρ) = inf ρ |dz|.
γ∈Γ γ

Definition 2.1. The extremal length of a path family Γ in D is defined as

L(Γ, ρ)2
λD (Γ) = sup ,
ρ A (D, ρ)
where the supremum is taken over all metrics ρ on D.
The conformal invariance of extremal length is an immediate consequence of its definition (see
[11, p. 130]). Next we state a basic property of extremal length known as the serial rule (see [1, p.
54-55], [7, p. 363] and [11, p. 134-135]).
 Christina Karafyllia 7

Theorem 2.3. Let Γ1 , Γ2 , . . . , Γn be path families contained in mutually disjoint Borel sets D1 , D2 ,
Sn
. . . , Dn respectively, and let Γ be a path family contained in a domain D ⊃ Di . If each γ ∈ Γ
i=1
contains at least one γi ∈ Γi for every i = 1, 2, . . . , n, then
n
X
λD (Γ) ≥ λDi (Γi ).
i=1

Sometimes it is more convenient to use the more special notion of extremal distance. Let D be
a plane domain and E1 , E2 be two disjoint closed sets on ∂D. If Γ is the family of connected arcs
in D that join E1 and E2 , then the extremal length λD (Γ) is called the extremal distance between
E1 and E2 with respect to D and is denoted by λD (E1 , E2 ).
Now let D be a Jordan domain and E1 , E2 be two disjoint arcs on its boundary ∂D. Also, let
F1 , F2 denote the arcs of ∂D\ {E1 ∪ E2 }. If Γ1 is the family of connected arcs in D that join E1
and E2 and Γ2 is the family of connected arcs in D that join F1 and F2 , then we define
s1 = inf l (γ) , s2 = inf l (γ) ,
γ∈Γ1 γ∈Γ2

where l (γ) denotes the euclidean length of γ. Using the notation above, we state the following
inequality, proved in [17, p. 22-24].
Theorem 2.4. Let D be a Jordan domain and E1 , E2 be two disjoint arcs on its boundary ∂D.
The extremal distance between E1 and E2 with respect to D satisfies the inequality
1 (log (1 + 2s1 /s2 ))2
λD (E1 , E2 ) ≥ .
π 1 + 2 log (1 + 2s1 /s2 )
2.4 Harmonic measure
If E ⊂ D\ {0}, then a special case of Beurling-Nevanlinna projection theorem (see [1, p. 43-44],
[11, p. 105] and [24, p. 120]) is the following:

Theorem 2.5. Let E ⊂ D\ {0} be a closed, connected set intersecting the unit circle. Let E ∗ =
{− |z| : z ∈ E} = (−1, −r0 ], where r0 = min {|z| : z ∈ E}. Then,
2 (1 − r0 )
ωD (0, E) ≥ ωD (0, E ∗ ) = arcsin .
π (1 + r0 )
Next theorem states the strong Markov property for harmonic measure which follows from the
probabilistic interpretation of harmonic measure (see [5, p. 282] and [23, p. 88]).
Theorem 2.6. Let D1 and D2 be two domains in C. Assume that D1 ⊂ D2 and let F ⊂ ∂D2 be
a closed set. If σ = ∂D1 \∂D2 , then for z ∈ D1 ,
Z
ωD2 (z, F ) = ωD1 (z, F ) + ωD1 (z, ds) ωD2 (s, F ).
σ

2.5 Harmonic measure and hyperbolic distance

Theorem 2.7. Let Γ be the hyperbolic geodesic joining two points z1 , z2 ∈ ∂D in D. Then
4 −dD (0,Γ)
e−dD (0,Γ) ≤ ωD (0, Γ) ≤ e .
π
For the proof see [15].
8 On a property of harmonic measure on simply connected domains

2.6 Harmonic measure and extremal distance

Let D be a simply connected domain, E be an arc on ∂D and z0 ∈ D. Consider all Jordan arcs
σ ⊂ D that join z0 to ∂D\E and define
λD (z0 , E) = sup λD\σ (σ, E) ,
σ

where the supremum is taken over all such Jordan arcs. Then the following theorem gives a relation
between ωD (z0 , E) and λD (z0 , E) (see [7, p. 368-371], [11, p. 144-146]).
Theorem 2.8. Let D be a simply connected domain, E be an arc on ∂D and z0 ∈ D. Then
8 −πλD (z0 ,E)
e−πλD (z0 ,E) ≤ ωD (z0 , E) ≤ e .
π

3 First counter-example
Hereinafter, we use the notation D (z, r) := {w ∈ C : |w − z| < r} for some z ∈ C and some
r > 0. Let D be the simply connected domain of Fig. 5, namely D = {z − 2 : z ∈ D0 }, where
+∞[ 
 n 1
D0 = z ∈ C\D : |Arg z| < 1 \ z ∈ ∂D (0, e ) : n ≤ |Arg z| ≤ 1 .
40
n=1

Figure 5. The simply connected domain D.

The Riemann mapping theorem implies that there exists a conformal mapping ψ from D onto D
1
such that ψ (0) = 0. Consider the sequence {Rn }n∈N with Rn = en+ 40n − 2 for every n ∈ N. For
n ∈ N, we set FRn = {z ∈ D : |ψ (z)| = Rn } and ERn = {ζ ∈ ∂D : |ψ (ζ)| ≥ Rn }. Also, for n ∈ N,
let ΓRn be the hyperbolic geodesic joining the endpoints of FRn in D and CRn be the arc of ∂D
joining the endpoints of ΓRn and lying on the boundary of the component of D\ΓRn which does
not contain the origin.
Theorem 3.1. With the notation above, the simply connected domain D satisfies the following
properties:
(1) For every n ∈ N, ψ (CRn ) = ψ (ERn ).
(2) There does not exist a positive constant c such that ψ (FRn ) ⊂ {z ∈ D : dD (z, ψ (ΓRn )) < c}
for every n ∈ N.
(3) lim ωω̂D D (Rn )
(Rn ) = +∞.
n→+∞
 Christina Karafyllia 9

Proof. Property (1) is immediate by the construction of D. So, we prove properties (2) and (3)
(for a similar calculation see [14]). If r0 = min {|z| : z ∈ FRn }, then
1 + r0
dD (0, FRn ) = inf {dD (0, z) : z ∈ FRn } = log .
1 − r0
So, applying Theorem 2.5 we have that for every n ∈ N,

2 (1 − r0 ) 2 2
(3.1) ωD (0, FRn ) ≥ arcsin = arcsin e−dD (0,FRn ) ≥ e−dD (0,FRn )
π (1 + r0 ) π π
(see [19, p. 9-10]). Now fix a number n > 1. If f (z) = z +2 and g (z) = Log z, then the composition
h = g ◦ f ◦ ψ is a conformal mapping of D onto h (D) := D0 (see Fig. 6). By (3.1) and the conformal
invariance of the hyperbolic distance, we deduce that
2 2 2
(3.2) ωD (0, FRn ) ≥ e−dD (0,FRn ) = e−dh(D) (h(0),h(FRn )) = e−dD0 (log 2,h(FRn )) ,
π π π
where   
1
h (FRn ) = n + n + iy : |y| < 1
40
(see Fig. 6). Notice that if z ∈ D and gD0 (log 2, z) denotes the Green’s function for D0 (see [11, p.
0

41-43], [24, p. 106-115]), then

1 + e−gD0 (log 2,z)
dD0 (log 2, z) = log
1 − e−gD0 (log 2,z)
(see [3, p. 12-13] and [24, p. 106]). Let bn = n + 401n . For every b0n ∈ h (FRn ) \ {bn }, we infer, by a
symmetrization result, that
gD0 (log 2, bn ) ≥ gD0 log 2, b0n ,

(see Lemma 9.4 in [12, p. 659]). Since

1 + e−x
f (x) = log
1 − e−x
is a decreasing function on (0, +∞), we have that dD0 (log 2, h (FRn )) = dD0 (log 2, bn ). This in
conjunction with (3.2) implies that
2
(3.3) ωD (0, FRn ) ≥ e−dD0 (log 2,bn ) .
π

Figure 6. The simply connected domain D0 and h (FRn ), h (ΓRn ) in case n = 3.

10 On a property of harmonic measure on simply connected domains

Since ΓRn denotes the hyperbolic geodesic joining the endpoints of FRn in D, by Theorem 2.7 and
[7, p. 370],
1 2 2
(3.4) ωD (0, ERn ) = ωD (0, CRn ) = ωD (0, ΓRn ) ≤ e−dD (0,ΓRn ) = e−dD0 (log 2,h(ΓRn )) .
2 π π
Since h preserves the geodesics and D0 is symmetric with respect to the real axis, we deduce that
dD0 (log 2, h (ΓRn )) = dD0 (log 2, cn ), where cn = h (ΓRn ) ∩ R ∈ (n, n + 1) and thus by (3.4) we
conclude that
2
(3.5) ωD (0, ERn ) ≤ e−dD0 (log 2,cn ) .
π
Next we prove that bn < cn . Since cn lies on the geodesic h (ΓRn ), we have that
c


1
ωD0 cn , h ER n
= 1 − ωD0 (cn , h (ERn )) = ,
2
c
where ERn = ∂D\ERn . Also, we observe that
     
c


1 1 1
ωD0 bn , h ER ≥ ωD(n,1) bn , n + iy : ≤ |y| ≤ 1 = ω D , iy : ≤ |y| ≤ 1 ,
n
40n 40n 40n
where the last equality occurs by a translation (see Fig. 7). Applying the Beurling-Nevanlinna
projection theorem [1, p. 43], we get
        
1 1 1 1 1 1
ωD , iy : n ≤ |y| ≤ 1 ≥ ωD , −1, − n ≥ ωD , −1, − ,
40n 40 40n 40 40 40
where !2
1
1−
  
1 1 2 40
ωD , −1, − = arcsin 1 = 0.719987303 > 0.7.
40 40 π 1+ 40
Therefore,
c c





ωD0 bn , h ER n
> 0.7 > 0.5 = ωD0 cn , h ER n

which implies that bn < cn .

Figure 7.

Since log 2, bn and cn lie, in this order, along a geodesic, we have that
dD0 (log 2, cn ) = dD0 (log 2, bn ) + dD0 (bn , cn )
 Christina Karafyllia 11

(see [3, p. 14]). Combining this with (3.3) and (3.5), we infer that
ωD (0, FRn )
(3.6) ≥ edD0 (log 2,cn )−dD0 (log 2,bn ) = edD0 (bn ,cn ) .
ωD (0, ERn )
Using the quasi-hyperbolic distance defined in Section 2, we get
1 cn 1 cn
Z Z
1 dx dx
dD0 (bn , cn ) ≥ δD0 (bn , cn ) = ≥
2 bn d (x, ∂D0 )
q
2 2 bn
2
1
40n + (x − n)2
1 1
arcsinh (40n (cn − n)) − arcsinh (1)
=
2 2
1 1
(3.7) ≥ arcsinh (40n k) − arcsinh (1) ,
2 2
where k is a positive constant whose existence comes from the following fact: Suppose that the
sequence {cn − n}n∈N (or any subsequence of it) tends to 0. By similar estimates as above, we have
  
c


1
ωD0 cn , h ERn ≥ ωD(n,1) cn , n + iy : n ≤ |y| ≤ 1
40
     
1 1
= ωD cn − n, iy : n ≤ |y| ≤ 1 ≥ ωD cn − n, −1, − n
40 40
which implies that
c



lim ωD0 cn , h ER n
= 1.
n→+∞
But this leads to contradiction because, for every n > 1,
c



ωD0 cn , h ER n
= 0.5.
So, cn − n ≥ k for some k > 0.
Now, taking limits in (3.7) as n → +∞, we obtain
(3.8) lim dD0 (bn , cn ) = +∞
n→+∞

which proves property (2). Finally, by (3.6) and (3.8) we infer that
ω̂D (Rn ) ωD (0, ψ (FRn )) ωD (0, FRn )
lim = lim = lim = +∞
n→+∞ ωD (Rn ) n→+∞ ωD (0, ψ (ERn )) n→+∞ ωD (0, ERn )
and thus property (3) holds. So, there does not exist a positive constant C such that for every
R > 0,
ωD (R) ≥ C ω̂D (R) .


4 Second counter-example
Let D be the simply connected domain of Fig. 8, namely D = {z − 2 : z ∈ D0 }, where

D0 = z ∈ C\D : |Arg z| < 1 \K0
and
+∞
[    
n 1 iθ n 1
n+ 40 1
K0 = z ∈ ∂D (0, e ) : n ≤ |Arg z| ≤ 1 ∪ re : e ≤ r ≤ e , |θ| = n .
40 40
n=1
12 On a property of harmonic measure on simply connected domains

Figure 8. The simply connected domain D.

1
Consider the sequence {Rn }n∈N with Rn = en+ 40 − 2 for every n ∈ N. For n ∈ N, we set
FRn = {z ∈ D : |ψ (z)| = Rn } and ERn = {ζ ∈ ∂D : |ψ (ζ)| ≥ Rn }. For n ∈ N, FR∗ n denotes the
component of FRn such that ψ FR∗ n intersects the real axis and Γ∗Rn is the hyperbolic geodesic


∗ be the arc of ∂D joining the endpoints of Γ∗ and
joining the endpoints of FR∗ n in D. Also, let CR n Rn
lying on the boundary of the component of D\Γ∗Rn which does not contain the origin.
Before presenting the second counter-example, we prove a lemma.

Lemma 4.1. Let D be the simply connected domain of Fig. 8. With the notation above, we have
that for every n ∈ N,
ψ Γ∗Rn ⊂ {z ∈ D : |z| ≥ Rn } .

Proof. The conformal

 mapping f (z) = z + 2 maps D conformally onto f (D) = D∗0 (see Fig. 9).
If D1 = D0 ∪ z ∈ C\D (0, Rn + 2) : |Arg z| < 1 , then the hyperbolic geodesic, γRn , joining the
endpoints of f ψ FR∗ n in D1 lies in D1 \D (0, Rn + 2).

Indeed, suppose that γR ∗ 6⊂ D1 \D (0, Rn + 2). Then γR ∗ contains a curve γ0 lying in D1 ∩

n n
D (0, Rn + 2) with endpoints z1 , z2 ∈ D1 ∩ ∂D (0, Rn + 2) (see Fig. 10). Since γR ∗ is the hyperbolic
n
geodesic joining the endpoints of f ψ FR∗ n in D1 , γ0 is the hyperbolic geodesic joining z1 to

z2 in D1 . Notice that D1 is a hyperbolic region in C such that D1 ∩ ∂D (0, Rn + 2) 6= ∅ and

D1 \ {z ∈ C : |z| > Rn + 2} ⊂ D1∗ , where D1∗ is the reflection of D1 in the circle ∂D (0, Rn + 2). So,
applying Theorem 2.1, we get
λD1∗ (z) < λD1 (z) , z ∈ γ0
and thus Z Z
∗ ∗
λD1 (z ) |dz | < λD1 (z) |dz|,
γ0∗ γ0

where γ0∗ is the reflection of γ0 in ∂D (0, Rn + 2). But this leads to contradiction because γ0 is
∗ ⊂ D \D (0, R + 2). Since D ⊂ D , by
the hyperbolic geodesic joining z1 to z2 in D1 . So, γR n

1 n 0 1
Theorem 2.2 we deduce that f ψ ΓRn separates f ψ FR∗ n and γR
∗ ∗ and hence for every n ∈ N,



n

f ψ Γ∗Rn ⊂ {z ∈ D0 : |z| ≥ Rn + 2} .

Since f is a conformal mapping of D onto D0 , we infer that for every n ∈ N,

ψ Γ∗Rn ⊂ {z ∈ D : |z| ≥ Rn } .


 Christina Karafyllia 13

Figure 9. The domain D0 . Figure 10.


Theorem 4.1. With the notation above, the simply connected domain D satisfies the following
properties:
∗


(1) For every n ∈ N, ψ CR n

6= ψ (ERn ).
(2) For every n ∈ N, ψ FR∗ n ⊂ z ∈ D : dD z, ψ Γ∗Rn ≤ 4π .


(3) lim ω̂D (Rn ) = +∞.

n→+∞ ωD (Rn )

Proof. Property (1) is immediate by the construction of D. So, we prove properties (2) and (3).
Fix a number n ∈ N. If f (z) = z + 2 and g (z) = Log z, then the composition h = g ◦ f ◦ ψ
is a conformal mapping of D onto h (D) := D0 (see Fig. 12). By Lemma 4.1, we have that
∗


ψ ΓRn ⊂ {z ∈ D : |z| ≥ Rn } and thus
 
∗

0 1
h ΓRn ⊂ z ∈ D : Re (z) ≥ n + .
40
If ψ (z 0 ) ∈ z ∈ ψ FR∗ n : Im (z) ≥ 0 , then let w := h (z 0 ) and w0 denote the endpoint of h FR∗ n


lying in the upper half-plane. If w1 = h Γ∗Rn ∩ ∂D (w0 , |w − w0 |) and γRn denotes the arc of

∂D (w0 , |w − w0 |) joining w to w1 and lying in z ∈ D0 : Re (z) ≥ n + 40 1


(see Fig. 11), then we
infer that
dD ψ z 0 , ψ Γ∗Rn = dD0 w, h Γ∗Rn ≤ dD0 (w, w1 ) ≤ 2δD0 (w, w1 )






Z 2π
|dz| |dz|
Z Z
rdθ
≤ 2 0
=2 ≤2
γRn d (z, ∂D ) γRn |z − w0 | 0 r
= 4π,
where the last inequality comes from the substitution z − w0 = reiθ , where r = |z − w0 | and
θ ∈ (0, 2π). Therefore, for every ψ (z 0 ) ∈ z ∈ ψ FR∗ n : Im (z) ≥ 0 and every n ∈ N,


dD ψ z 0 , ψ Γ∗Rn ≤ 4π.

Since D is symmetric with respect to the real axis, we have that for every ψ (z 0 ) ∈ ψ FR∗ n and


every n ∈ N,
dD ψ z 0 , ψ Γ∗Rn ≤ 4π

which implies that for every n ∈ N,

ψ FR∗ n ⊂ z ∈ D : dD z, ψ Γ∗Rn ≤ 4π



and hence property (2) holds.

14 On a property of harmonic measure on simply connected domains

Figure 11. Figure 12. The domain D0 .

Next we set SRn = {z ∈ D0 ∩ ∂D (n, 1) : Re (z) ≥ n} as illustrated in Fig. 13. By the conformal
invariance of harmonic measure, we deduce that
ωD 0, ψ FR∗ n ωD0 log 2, h FR∗ n ωD0 log 2, h FR∗ n







ω̂D (Rn )
(4.1) = = ≥ .
ωD (Rn ) ωD (0, ψ (ERn )) ωD0 (log 2, h (ERn )) ωD0 (log 2, SRn )
Applying Theorems 2.8 and 2.3, we have that
∗
ωD0 log 2, h FR∗ n π e−πλD0 (log 2,h(FRn ))



π π(λD0 (log 2,SRn )−λD0 (log 2,h(FR∗ )))
≥ −πλ (log 2,S )
= e n
ωD0 (log 2, SRn ) 8 e D0 Rn 8
π π(λD0 (log 2,h(FR∗ ))+λD0 (h(FR∗ ),SRn )−λD0 (log 2,h(FR∗ )))
≥ e n n n
8
π πλD0 (h(FR∗ ),SRn )
= e n
8
which in combination with (4.1) implies that for every n ∈ N,
ω̂D (Rn ) π ∗
(4.2) ≥ eπλD0 (h(FRn ),SRn ) .
ωD (Rn ) 8

Figure 13. The sets SRn and h FR∗ n in case n = 2.

 Christina Karafyllia 15

If s1 denotes the euclidean distance between h FR∗ n and SRn in D0 and s2 denotes the euclidean


distance between the sets
   
1 1 1
n + iy : n ≤ y ≤ 1 ∪ x + i n : n ≤ x ≤ n +
40 40 40
and    
1 1 1
n + iy : −1 ≤ y ≤ − n ∪ x − i n : n ≤ x ≤ n +
40 40 40
in D0 , then we observe that
2 1
s2 = n and ≤ s1 ≤ 1.
40 2
So, applying Theorem 2.4, we infer that for every n ∈ N,

1 (log (1 + 2s1 /s2 ))2 1 (log (1 + 40n /2))2
λD0 h FR∗ n , SRn ≥


≥ .
π 1 + 2 log (1 + 2s1 /s2 ) π 1 + 2 log (1 + 40n )
Taking limits as n → +∞, we get

1 (log (1 + 40n /2))2
lim λD0 h FR∗ n , SRn ≥


lim = +∞.
n→+∞ π n→+∞ 1 + 2 log (1 + 40n )
This in conjunction with (4.2) gives
ω̂D (Rn )
lim = +∞
n→+∞ ωD (Rn )
which proves property (3). So, there does not exist a positive constant C such that for every R > 0,
ωD (R) ≥ C ω̂D (R) .


5 Proof of theorem 1.1

Proof of Theorem 1.1. Suppose that Conditions (1) and (2) are satisfied. From the proof of The-
orem 1.1 in [14, p. 3-4, 22-23], we deduce that there exists a positive constant K0 = K0 (c) such
that for every R > 0 and every i ∈ {1, 2, . . . , N (R)},
i
ωD 0, FRi ≤ K0 e−dD (0,FR ) .


(5.1)
Theorem 2.7 implies that for every R > 0 and every i ∈ {1, 2, . . . , N (R)},

1
1 i
(5.2) ωD 0, CRi
= ωD 0, ΓiR ≥ e−dD (0,ΓR ) .
2 2
Combining (5.1) and (5.2), we get
i
ωD 0, FRi ωD 0, FRi e−dD (0,FR )



i i
i

= 2 i

≤ 2K0 i
= 2K0 edD (0,ΓR )−dD (0,FR )
ωD 0, CR ωD 0, ΓR e−dD (0,ΓR )
i i
(5.3) = 2K edD (0,ψ(ΓR ))−dD (0,ψ(FR )) .
0

ψ FRi such that

> 0 and some i ∈ {1, 2, . . . , N (R)}. Then
there exist a point zR ∈


Fix some R


dD 0, ψ FRi = dD (0, zR ) and a point wR ∈ ψ ΓiR such that dD zR , ψ ΓiR = dD (zR , wR ).
Therefore,
dD 0, ψ ΓiR − dD 0, ψ FRi = dD 0, ψ ΓiR − dD (0, zR ) ≤ dD (0, wR ) − dD (0, zR )

≤ dD (0, zR ) + dD (zR , wR ) − dD (0, zR ) = dD (zR , wR )

= dD zR , ψ ΓiR < c,



16 On a property of harmonic measure on simply connected domains

where the last inequality comes from Condition (1). This in conjunction with (5.3) implies that for
every R > 0 and every i ∈ {1, 2, . . . , N (R)},
ωD 0, FRi


i

≤ 2K0 ec
ωD 0, CR
or equivalently
ωD 0, FRi ≤ KωD 0, CR i



,
c
where K = 2K0 e . By this relation, the conformal invariance of harmonic measure and Condition
(2), we infer that for every R > 0,
N (R)
X
ωD 0, FRi


ω̂D (R) = ωD (0, ψ (FR )) = ωD (0, FR ) =
i=1
N (R)
X
i


≤ K ωD 0, CR = KωD (0, ER ) = KωD (0, ψ (ER ))
i=1
= KωD (R) ,
and thus there exists a positive constant K = K (c) such that for every R > 0,
ω̂D (R) ≤ KωD (R) .
Now suppose that Conditions (2) and (3) are satisfied. For every R > 0 and every i =
1, 2, . . . , N (R), we have
ωD 0, ΓiR = 2ωD 0, CR i



(5.4) ,

p. 370]). Since by Condition (3), for every R > 0 and every i ∈ {1, 2, . . . , N (R)},
(see [7,
ψ ΓiR ⊂ D ∩ {z : |z| ≤ R}, we infer that
ωD 0, ψ ΓiR ≥ ωD 0, ψ FRi .





(5.5)
The conformal invariance of harmonic measure in conjunction with (5.4), (5.5) and Condition (2)
implies that

N (R) N (R) N (R)

X X X
FRi ΓiR ωD 0, ΓiR






ω̂D (R) = ωD (0, ψ (FR )) = ωD 0, ψ ≤ ωD 0, ψ =
i=1 i=1 i=1
N (R)
X
i


(5.6) = 2 ωD 0, CR = 2ωD (0, ER ) = 2ωD (0, ψ (ER )) = 2ωD (R)
i=1
and thus we have the desired result
ω̂D (R) ≤ 2ωD (R)
for every R > 0. 

6 Proof of theorem 1.2

In the proof of Theorem 1.2 we will use the following result which is an easy computation coming
from the conformal invariance of harmonic measure.
Lemma 6.1. Let a ∈ (0, 1) and b ∈ [0, 1). Then
2 1
ωD\[a,1) (−b, ∂D) = 1 − arctan r .
π (1+a)(1+b)
2
(1−a)(1−b) −1
 Christina Karafyllia 17

Proof of Theorem 1.2. Let D be a starlike domain in C. Using the notation of Theorem 1.1,
we will prove that Conditions (2) and (3) are satisfied. Since D is starlike, Condition (2) is
obviously satisfied and thus we prove Condition (3). Let FRi be a component of FR for some
i i i



i ∈ {1, 2, . . . , N (R)}. Suppose that ψ ΓR 6⊂ D ∩ {z : |z| ≤ R}, then ψ ΓR contains a curve γR
i


lying in D\D (0, R) with endpoints z1 , z
2 ∈ ∂D (0, R) (see Fig. 14). Since ψ ΓR is the hyperbolic
geodesic joining the endpoints of ψ FRi in D, γR i is the hyperbolic geodesic joining z to z in D.
1 2
Notice that D is a hyperbolic region in C such that D ∩ ∂D (0, R) 6= ∅. Since D is starlike, we
have that D\D (0, R) ⊂ D∗ , where D∗ is the reflection of D in the circle ∂D (0, R). So, applying
Theorem 2.1, we get
i
λD∗ (z) < λD (z) , z ∈ γR
and thus Z Z
λD (z ∗ ) |dz ∗ | < λD (z) |dz|,
i ∗ i
γR γR
i∗ i in ∂D (0, R). But this leads to contradiction because γ i is the
where γR is the reflection of γR R
hyperbolic geodesic joining z1 to z2 in D. So, ψ ΓiR ⊂ D ∩ {z : |z| ≤ R} and thus Condition (3)


is satisfied. Theorem 1.1 implies that for every R > 0,
ω̂D (R) ≤ 2ωD (R) .

Figure 14.

z
Now we prove that the constant 2 is best possible. Consider the Koebe function K (z) = (1−z)2
1 1

which maps D conformally onto D0 := C\ −∞, − 4 . For R > 4 , by the conformal invariance of
harmonic measure and Lemma 6.1, we have

ω̂D0 (R) = ωD(0,R)\(−R,− 1 ] (0, ∂D (0, R)) = ωD\(−1,− ] (0, ∂D) = ωD\[ 4R
1 1
,1) (0, ∂D)
4 4R

2 1 2 4R − 1
(6.1) = 1− arctan r = 1 − arctan √ .
π 4R+1
2 π 4 R
4R−1 −1

Using the fact that

  √
−1 2R − 1 4R − 1
K (−R) = ±i
2R 2R
18 On a property of harmonic measure on simply connected domains

and the conformal invariance of harmonic measure, we deduce that

ωD0 (R) = ωD0 (0, (−∞, −R])
   √   √ 
2R − 1 4R − 1 2R − 1 4R − 1
= ωD 0, arc −i , +i
2R 2R 2R 2R
    √ 
2R − 1 4R − 1
= 2ωD 0, arc 1, +i
2R 2R
√
1 4R − 1
(6.2) = arctan ,
π 2R − 1
 √ √  √
4R−1 4R−1
where arc 2R−1 2R−1
denotes the arc of ∂D joining 2R−1 − i 4R−1




2R − i 2R , 2R + i 2R 2R 2R to
√
2R−1
+ i 4R−1


2R 2R counterclockwise. Applying (6.1) and (6.2), we infer that
4R−1
ω̂D0 (R) π − 2 arctan 4√R
lim = lim √ = 2.
R→+∞ ωD0 (R) R→+∞ arctan 4R−1 2R−1
Suppose that there exists a positive constant C < 2 such that for every starlike domain D and
every R > 0, ω̂D (R) ≤ CωD (R). This implies for D0 that
ω̂D0 (R)
2 = lim ≤C
R→+∞ ωD0 (R)

which leads to contradiction. Therefore, the constant 2 is best possible. 

Note that we could also prove Theorem 1.2 by using instead of Minda’s reflection principle and
Theorem 1.1, the strong Markov property for harmonic measure (see Section 2).
Another proof of Theorem 1.2. Let D be a starlike domain in C. Set FR = D ∩ ∂D (0, R), ER =
∂D\D (0, R), LR = ∂D ∩ D (0, R) and D0 = D ∩ D (0, R) as illustrated in Fig. 15. So, we have the
relations
(6.3) ω̂D (R) = ωD0 (0, FR ) = 1 − ωD0 (0, LR )
and
(6.4) ωD (R) = ωD (0, ER ) = 1 − ωD (0, LR ) .

Figure 15. The starlike domain D. Figure 16.

 Christina Karafyllia 19

By Theorem 2.6,
Z
ωD (0, LR ) = ωD0 (0, LR ) + ωD0 (0, ds) ωD (s, LR )
FR

which in conjunction with (6.3) and (6.4) implies that

Z
(6.5) ω̂D (R) = ωD (R) + ωD0 (0, ds) ωD (s, LR ).
FR

If N (R) ∈ N ∪ {+∞} denotes the number of components of FR and FRi denotes each of these
components for i = 1, 2, . . . , N (R), then
Z N (R) Z
X
(6.6) ωD0 (0, ds) ωD (s, LR ) = ωD0 (0, ds) ωD (s, LR ),
FR i
FR
i=1

since FRi are mutually disjoint sets.

Let FRi be a component of FR for some i ∈ {1, 2, . . . , N (R)}. If z1 , z2 denote the endpoints of
FRi such that Arg z1 < Arg z2 , then we set

L∗R = rei Arg z1 : 0 ≤ r ≤ R ∪ rei Arg z2 : 0 ≤ r ≤ R

 

and
D∗ = {z ∈ C : Arg z1 < Arg z < Arg z2 }
as illustrated in Fig. 16. For every s ∈ FRi ,
(6.7) ωD (s, LR ) ≤ ωD∗ (s, L∗R ) .
If θ = Arg z2 − Arg z1 , we consider the conformal mappings f1 (z) = ze−i(Arg z2 −θ/2) , f2 (z) = z π/θ

z−1
and f3 (z) = z+1 . Then the composition f = f3 ◦ f2 ◦ f1 maps D∗ conformally onto D. Since f FRi
is the hyperbolic geodesic joining f (z1 ) to f (z2 ) in D, for every s ∈ FRi ,
1
ωD∗ (s, L∗R ) = ωD (f (s) , f (L∗R )) = .
2
This in combination with (6.7) implies that for every s ∈ FRi ,

1
ωD (s, LR ) ≤ .
2
By this and the relations (6.5) and (6.6) we infer that
N (R) Z N (R) Z
X 1 X
ω̂D (R) = ωD (R) + ωD0 (0, ds) ωD (s, LR ) ≤ ωD (R) + ωD0 (0, ds)
i
FR 2 i
FR
i=1 i=1
N (R)
1 X 1
ωD0 0, FRi = ωD (R) + ωD0 (0, FR )


= ωD (R) +
2 2
i=1
1
= ωD (R) + ω̂D (R)
2
and thus for every R > 0,
ω̂D (R) ≤ 2ωD (R) .
The fact that the constant 2 is best possible is proved as before. 
20 On a property of harmonic measure on simply connected domains

References
[1] L.V. Ahlfors, Conformal Invariants: Topics in Geometric Function Theory, McGraw-Hill, New York, 1973.
[2] A. Baernstein, The size of the set where a univalent function is large, J. d’ Anal. Math. 70 (1996), 157–173.
[3] A.F. Beardon and D. Minda, The hyperbolic metric and geometric function theory, Quasiconformal mappings
and their applications (2007), 9–56.
[4] A.F. Beardon and T.K. Carne, Euclidean and hyperbolic lengths of images of arcs, Proc. London Math. Soc. 97
(2008), 183–208.
[5] D. Betsakos, Harmonic measure on simply connected domains of fixed inradius, Ark. Mat. 36 (1998), 275–306.
[6] , Geometric theorems and problems for harmonic measure, Rocky Mountain J. of Math. 31 (2001), 773–
795.
[7] A. Beurling, The Collected Works of Arne Beurling: Vol. 1, Complex Analysis, Birkhäuser, Boston, 1989.
[8] M. Essén, On analytic functions which are in H p for some positive p, Ark. Mat. 19 (1981), 43–51.
[9] M. Essén K. Haliste J.L. Lewis and D.F. Shea, Harmonic majorization and classical analysis, J. London Math.
Soc. 32 (1985), 506–520.
[10] M. Essén, Harmonic majorization and thinness, Proc. of the 14th Winter School on Abstract Analysis (1987),
295–304.
[11] J.B. Garnett and D.E. Marshall, Harmonic Measure, Cambridge University Press, Cambridge, 2005.
[12] W.K. Hayman, Subharmonic Functions: Volume 2, Academic Press, London, 1989.
[13] W.K. Hayman and A. Weitsman, On the coefficients and means of functions omitting values, Math. Proc.
Cambridge Philos. Soc. 77 (1975), 119–137.
[14] C. Karafyllia, On a relation between harmonic measure and hyperbolic distance on planar domains, Indiana Univ.
Math. J. (to appear).
[15] , On the Hardy number of a domain in terms of harmonic measure and hyperbolic distance, (submitted).
[16] Y.C. Kim and T. Sugawa, Hardy spaces and unbounded quasidisks, Ann. Acad. Sci. Fenn. 36 (2011), 291–300.
[17] O. Lehto and K.I. Virtanen, Quasiconformal Mappings in the Plane, Springer-Verlag, New York Heidelberg
Berlin, 1973.
[18] D. Minda, Inequalities for the hyperbolic metric and applications to geometric function theory, Lecture Notes in
Math. 1275 (1987), 235–252.
[19] P. Poggi-Corradini, Geometric models, iteration and composition operators, Ph.D. Thesis, University of Wash-
ington (1996).
[20] , The Hardy class of geometric models and the essential spectral radius of composition operators, Journal
of Functional Analysis 143 (1997), 129–156.
[21] , The Hardy class of Kœnigs maps, Michigan Math. J. 44 (1997), 495–507.
[22] C. Pommerenke, Boundary Behaviour of Conformal Maps, Springer-Verlag, Berlin, 1992.
[23] S.C. Port and C.J. Stone, Brownian Motion and Classical Potential Theory, Academic Press, New York, 1978.
[24] T. Ransford, Potential Theory in the Complex Plane, Cambridge University Press, Cambridge, 1995.
[25] M. Sakai, Isoperimetric inequalities for the least harmonic majorant of |x|p , Trans. Amer. Math. Soc. 299 (1987),
431–472.
[26] A.Yu. Solynin, The boundary distortion and extremal problems in certain classes of univalent functions, J. Math.
Sci. 79 (1996), 1341-1358.
[27] , Functional inequalities via polarization, St. Petersburg Math. J. 8 (1997), 1015–1038.
[28] M. Tsuji, A theorem on the majoration of harmonic measure and its applications, Tohoku Math. J. (2) 3 (1951),
13–23.
[29] , Potential Theory in Modern Function Theory, Maruzen, Tokyo, 1959.

Department of Mathematics, Aristotle University of Thessaloniki, 54124, Thessaloniki, Greece

E-mail address: karafyllc@math.auth.gr