TOPICS IN COMPLEX ANALYSIS BANACH CENTER PUBLICATIONS, VOLUME 31
INSTITUTE OF MATHEMATICS POLISH ACADEMY OF SCIENCES
WARSZAWA 1995
A NOTE ON COEFFICIENT MULTIPLIERS (H
p, B) AND (H
p, BM OA)
M A R I A N O W A K
Institute of Mathematics, M. Curie-Sk lodowska University Pl. Marii Curie-Sk lodowskiej 1, 20-031 Lublin, Poland
1. Introduction and statement of results. For a function f analytic in U = {z : |z| < 1} let
M
p(r, f ) = 1 2π
2π
R
0
|f (re
iθ)|
pdθ
1/p, 0 < p < ∞, M
∞(r, f ) = max
0≤θ≤2π
|f (re
iθ)|, where 0 ≤ r < 1.
The Hardy class H
p, 0 < p ≤ ∞, is the space of those f for which kf k
p= sup
0≤r<1
M
p(r, f ) < ∞.
A function f ∈ H
1is said to be in the space BM OA iff its boundary function f (e
iθ) is of bounded mean oscillation.
The Bloch space B consists of all analytic function in U for which kf k
B= sup
z∈U
(1 − |z|)|f
0(z)| < ∞ The proper inclusions:
H
∞⊂ BM OA ⊂ \
0<p<∞
H
p, BM OA ⊂ B are well-known (e.g. [3]).
A complex sequence {λ
n} is called a multiplier of a sequence space A into a sequence space B if {λ
na
n} ∈ B whenever {a
n} ∈ A. A space of analytic functions in U can be regarded as a sequence space by identifying each function with its
1991 Mathematics Subject Classification: Primary 30D55.
The paper is in final form and no version of it will be published elsewhere.
[299]
300
M. NOWAKsequence of Taylor coefficients. The set of all multipliers from A to B will be denoted by (A, B).
Recently Mateljevic and Pavlovic ([4], see also [5]) have characterized the multiplier spaces (H
1, B) and (H
p, BM OA), 1 ≤ p ≤ 2. They have proved the following theorems:
Theorem A. Let 1 ≤ p ≤ 2 and 1/p + 1/q = 1. Then g ∈ (H
p, BM OA) if and only if
M
q(r, g
0) ≤ c
1 − r , 0 < r < 1, where c denotes a constant.
Theorem B. (H
1, B) = (H
1, BM OA) = B
Here we extend the above theorems by describing the spaces (H
p, B), 0 < p
< ∞ and (H
p, BM OA), 0 < p < 1.
Let c denote a general constant not necessarily the same in each case. We have Theorem 1. If 1 ≤ p < ∞ and 1/p + 1/q = 1 then
g ∈ (H
p, B) if and only if there is a constant c such that
(1) M
q(r, g
0) ≤ c
1 − r , 0 < r < 1.
Theorem 2. If 0 < p < 1 , n is an integer such that 1/p < n + 1 then (H
p, H
∞) = (H
p, BM OA) = (H
p, B)
(2)
=
g : M
∞(r, g
(n)) < c (1 − r)
n+1−1/p= A
n. Note that for 0 < p ≤ 2, (H
p, B) = (H
p, BM OA).
2. Proof of Theorem 1. For f (z) = P
∞n=0
f (n)z b
n, g(z) = P
∞n=0
b g(n)z
nanalytic in U define
(3) h(z) = f ? g(z) =
∞
X
n=0
f (n) b b g(n)z
n. Then
(4) h(r
2e
iϕ) = 1 2π
2π
R
0
f (re
iθ)g(re
i(ϕ−θ))dθ, 0 < r < 1.
Assume that g satisfies (1) and f ∈ H
p, 1 ≤ p < ∞. Differentiating (4) with respect to ϕ we obtain
rh
0(r
2e
iϕ) = 1 2π
2π
R
0
f (re
iθ)g
0(re
i(ϕ−θ))e
−iθdθ.
COEFFICIENT MULTIPLIERS
301
Hence by H¨ older’s inequality
(5) M
∞(r
2, h
0) ≤ CM
p(r, f )M
q(r, g
0), which implies h ∈ B.
To prove the converse suppose that g is an analytic function such that f ?g ∈ B whenever f ∈ H
p, 1 < p < ∞. Without loss of generality we may assume that f (0) = 0. Then f
1(z) = f (z)/z also belongs to H
pand kf
1k
p= kf k
p. It follows from the closed graph theorem that T
g(f ) = f ? g is a bounded linear operator from H
pto B. So there is a constant c such that for any f ∈ H
p(6) kT
g(f )k
B= sup
0≤r<1 0≤ϕ≤2π
(1 − r
2)
2π
R
0
e
−iθr f (re
iθ)g
0(re
i(ϕ−θ))dθ
≤ ckf k
p.
This implies (7)
2π
R
0
f (re
iθ)
re
iθg
0(re
−iθ)dθ
≤ ckf k
p1 − r
2, 0 ≤ r < 1.
Let W (e
iθ) = P
nk=−n
a
ke
ikθbe a trigonometric polynomial with kW k
Lp[0,2π]≤1.
It follows from the M. Riesz theorem that its analytic projection w(e
iθ) = P
nk=0
a
ke
ikθsatisfies
kwk
Lp[0,2π]≤ A
pkW k
Lp[0,2π]≤ A
p. Also note that
(8)
2π
R
0
W (e
iθ)g
0(r
2e
−iθ)dθ
=
2π
R
0
w(re
iθ)g
0(re
−iθ)dθ
≤ cA
p1 − r
2.
If we denote g
0r2(z) = g
0(r
2z), 0 < r < 1, then taking the supremum over all W with kW k
Lp[0,2π]≤ 1 we get
kg
r02k
q= M
q(r
2, g
0) ≤ c 1 − r
2, and this proves Theorem 1.
3. Proof of Theorem 2. The following property of integral means is well known (cf. [1], p. 80): if 0 < p ≤ ∞, β > 0 and f is analytic in U then
M
p(r, f ) = O
1
(1 − r)
βif and only if M
p(r, f
0) = O
1
(1 − r)
β+1Hence the set A
nin formula (2) does not depend on n if only n + 1 > 1/p.
Now assume that 0 < p < 1. It was proved by Duren and Shields [2] that (H
p, H
∞) = A
n. So to prove our theorem it is enough to show that (H
p, B) ⊂ A
n. Suppose that g is an analytic function such that f ? g ∈ B whenever f ∈ H
p. Then the closed graph theorem implies
kf ? gk
B≤ ckf k
p.
302
M. NOWAKFor g(z) = P
∞k=0
ˆ g(k)z
kwe define D
ng(z) =
∞
X
k=0
(k + 1)
nb g(k)z
k. Let f (z) = P
∞k=0
(k + 1)
nz
k. Then we have
kD
ng
rk
B= kg ? f
rk
B≤ ckf
rk
p, where f
r(z) = f (rz), 0 < r < 1.
Because f (z) = P
n(z)/(1 − z)
n+1, where P
nis a polynomial of degree n, (9) kD
ng
rk
B≤ c
1 (1 − rz)
n+1p
= O
1
(1 − r)
n+1−1/p. Hence
(10) sup
0<ρ<1