Characterizations for the generalized frac-

tional integral operators on Morrey spaces

ERIDANI, Hendra GUNAWAN, Eiichi NAKAI and Yoshihiro
SAWANO

Abstract. We present some characterizations for the boundedness of the

generalized fractional integral operators on Morrey spaces. The charac-
terizations follow from two key estimates, one for the norm of the charac-
teristic functions of balls, and another for the values of the corresponding
fractional integrals on smaller balls. Our results extends those obtained
in our earlier paper [5]. We prove three theorems about necessary and
sufficient conditions. We show that these theorems are independent by
giving some examples.
Mathematics Subject Classification (2010). Primary 42B35; Secondary ,
26A33, 46E30, 42B20, 43A15.
Keywords. Fractional integrals, bounded operators, Morrey spaces.

1. Introduction
In this paper, for a measurable function ρ : (0, ∞) → (0, ∞), we are interested
in the generalized fractional integral operator Iρ given by the formula
∫
ρ(|x − y|)
Iρ f (x) := f (y) dy, x ∈ Rd ,
R d |x − y| d

for any suitable function f on Rd . This generalized fractional integral op-

erator was initilally investigated in [14]. Nowadays many authors have been
culminating important observations about Iρ especially in connection with
Morrey spaces. Morrey spaces cover Lebesgue spaces as special cases and
seem to describe the behavior of Iρ well. Here to highlight what we shall
prove in this paper, we take up the works [2, 4, 5, 7, 11, 12, 15, 22]. In these
works we proposed several conditions on ρ for Iρ to be bounded on Morrey
spaces. We aim to show that these conditions are necessary as well.
Note that the integral operators such as (1 − ∆)−α and L−α with α > 0
fall under this scope, where L is a suitable elliptic differential operator. Also, if
a measurable function V : Rd → (0, ∞) satisfies the reverse Hölder inequality,
2 ERIDANI, H. GUNAWAN, E. NAKAI and Y. SAWANO

that is, there exist some constants C > 0, q ≫ 1 such that, for all balls
(∫ )1/q ∫
B, B
V (x) q dx
|B| ≤ C B V (x) |B|
dx
, then the operators V γ (−∆ + V )−β
with 0 ≤ γ ≤ β ≤ 1 and V γ ∂j (−∆ + V )−β with 0 ≤ γ ≤ 21 ≤ β ≤ 1,
β − γ ≥ 21 and j = 1, 2, · · · , d also fall under this scope [10]. We refer to
[6, Sections 3 and 4] for a detailed description of these facts. Below, we
∫1
assume that 0 ρ(s) s ds < ∞, so that the fractional integrals Iρ f are well-
defined, at least for characteristic functions of balls. In addition, we shall also
assume that ρ satisfies the growth condition: there exist constants C1 > 0 and
0 < 2k1 < k2 < ∞ such that
∫ k2 r
ρ(s)
sup ρ(s) ≤ C1 ds, r > 0. (1.1)
r/2<s≤r k1 r s
This condition is weaker than the usual doubling condition: there exists a
constant C2 > 0 such that C12 ≤ ρ(r)ρ(s) ≤ C2 whenever r and s satisfy r, s > 0
and 21 ≤ rs ≤ 2. See [24] for some examples and more explanation about these
two conditions.
We note that if ρ(r) = rα , with 0 < α < d, then Iρ = Iα is the
classical fractional integral operator, also known as the Riesz potential, which
is bounded from Lp (Rd ) to Lq (Rd ) if and only if p1 − 1q = αd , where 1 < p, q <
∞ [19]. The necessary part is usually proved by using the scaling arguments.
See [4, 7, 8, 26] for some recent results on the boundedness properties of Iρ .
Our first theorem below characterizes the function ρ for which Iρ is
bounded from Lp (Rd ) to Lq (Rd ) for 1 < p < q < ∞.
Theorem 1.1. Let 1 < p < q < ∞. The operator Iρ is bounded from Lp (Rd )
to Lq (Rd ) if and only if ρ(r) ≤ C rd(1/p−1/q) for all r > 0.
Theorem 1.1 is a corollary of Theorem 1.3 and we prove and deal with
Theorem 1.1 as such. For ρ(r) = rα , Theorem 1.1 reads that the operator
Iρ is bounded from Lp (Rd ) to Lq (Rd ) if and only if αd = p1 − 1q , where
1 < p < q < ∞.
Our primary aim here is to characterize the function ρ for which Iρ is
bounded from one Morrey space to another. We obtain the characterizations
by estimating the norm of the characteristic functions of balls and the val-
ues of the corresponding fractional integrals on smaller balls. Theorem 1.1
extends to the next result on Morrey spaces. For 1 ≤ p < ∞ and 0 ≤ λ < d,
recall that the Morrey space Lp,λ = Lp,λ (Rd ) consists of all locally integrable
functions f on Rd for which
[ ∫ ] p1
1 1
∥f : Lp,λ ∥ := sup (λ−d)/p |B(a, r)|
|f (x)|p dx < ∞.
a∈Rd , r>0 r B(a,r)

Note that Lp,0 (Rd ) = Lp (Rd ) for 1 ≤ p < ∞. See [18] for more information
about these spaces. The following theorem generalizes the previous charac-
terization in Theorem 1.1. Theorem 1.2 follows from Theorem 1.3 and we
prove and deal with Theorem 1.2 as such again.
 Generalized fractional integral operators on Morrey spaces 3

Theorem 1.2. Let 1 < p < q < ∞ and 0 ≤ λ < d. Assume that ρ satisfies
(1.1). Then the operator Iρ is bounded from Lp,λ (Rd ) to Lq,λ (Rd ) precisely
when one of the following equivalent conditions holds.
(a) ρ(r) ≤ C r(d−λ)(1/p−1/q) for all r > 0.
∫r
(b) ρ̃(r) = 0 ρ(s)
s ds ≤ C r
(d−λ)(1/p−1/q)
for all r > 0.
Now we shall state our main result in full generality. We state result for
generalized Morrey spaces. For a certain function ϕ : (0, ∞) → (0, ∞), we say
that a function f belongs to the generalized Morrey space Lp,ϕ = Lp,ϕ (Rd ),
where 1 ≤ p < ∞, if
[ ∫ ]1/p
1 1
∥f : Lp,ϕ ∥ := sup |f (x)| dx
p
< ∞.
a∈Rd , r>0 ϕ(r) |B(a, r)| B(a,r)

Note that if ϕ(r) := r(λ−d)/p for some 1 ≤ p < ∞ and 0 ≤ λ < d, then
Lp,ϕ (Rd ) = Lp,λ (Rd ). In [13, p.446] we justified that ϕ is nonincreasing func-
tion such that t 7→ ϕ(t)p td is a nondecreasing for Lp,ϕ (Rd ) ̸= {0}. We refer
to [12, 16] and [24, Section 1] for more information about these spaces.
In this paper, we shall assume that ϕ : (0, ∞) → (0, ∞) is almost de-
creasing [that is, if r ≤ s, then ϕ(r) ≥ C3 ϕ(s)], and that rd ϕp (r) is almost
increasing, [that is, if r ≤ s, then rd ϕ(r)p ≤ C3 sd ϕ(s)p ]. These two conditions
implies that ϕ also satisfies the doubling condition. Denote by Gp the set of
all functions ϕ : (0, ∞) → (0, ∞) such that ϕ is almost decreasing and that
r 7→ rd/p ϕ(r) is almost increasing. Now we present three different criteria for
∫r
the boundedness of Iρ . Recall that we defined ρ̃(r) = 0 ρ(t) t dt.

Theorem 1.3. Let 1 < p < q < ∞ and ϕ ∈ Gp . Assume

∫ ∞ ∫ ∞
ϕ(s) ϕ(s)ρ(s)
ds ≤ C ϕ(r), ds ≤ C ϕ(r)ρ(r) (r > 0) (1.2)
r s r s
for some constant C > 0. Then Iρ is bounded from Lp,ϕ (Rd ) to Lq,ϕp/q (Rd )
if and only if
ρ̃(r) ≤ C ϕ(r)p/q−1 (r > 0) (1.3)
for some constant C > 0.
Theorem 1.4. Let 1 < p < q < ∞ and ϕ ∈ Gp .
(i) If there exists a positive constant C > 0 such that
∫ r ∫ ∞
ρ(t) ϕ(t)ρ(t)
ϕ(r) dt + dt ≤ Cϕ(r)p/q (r > 0), (1.4)
0 t r t
then Iρ is bounded from Lp,ϕ (Rd ) to Lq,ϕp/q (Rd ).
(ii) Assume in addition that ϕ satisfies
∫ r
ϕ(s)sd/p
ds ≤ Cϕ(r)rd/p (r > 0) (1.5)
0 s
for some constant C > 0 and Iρ is bounded from Lp,ϕ (Rd ) to Lq,ϕp/q (Rd )
then (1.4) holds.
4 ERIDANI, H. GUNAWAN, E. NAKAI and Y. SAWANO

Theorem 1.5. Suppose that ϕ, ψ ∈ G1 .

(i) Assume that there exists a constant C > 0 such that
∫ r ∫ ∞
ρ(t) ϕ(t)ρ(t)
ϕ(r) dt + dt ≤ Cψ(r) (r > 0). (1.6)
0 t r t
Then Iρ is bounded from L1,ϕ (Rd ) to L1,ψ (Rd ).
(ii) If there exists C > 0 such that
∫ r
ϕ(s)sd
ds ≤ Cϕ(r)rd (r > 0) (1.7)
0 s
and Iρ is bounded from L1,ϕ (Rd ) to L1,ψ (Rd ), then (1.6) holds.
Remark 1.6.
(i) The condition (1.4) appeared in [7] originally and it later appeared in a
bilinear estimate of the form g · Iα f (see [23, Theorem 1.6]).
∫r ∫ ∞ ϕ(t)ρ(t) ∫∞
(ii) If ϕ is decreasing, ϕ(r) 0 ρ(t)
t dt + r t dt = 0 ϕ(max(r,t))ρ(t)
t dt.
Hereafter, the letter C denotes a positive constant whose value may
differ from line to line, which may depend on d, ρ, p and q, but not on the
functions f and the variables x.
This paper is organized as follows: In Section 2, we calculate roughly the
image by Iρ of the characteristic function of balls. Based upon a preliminary
result, Lemma 2.1, in Section 2, we consider the boundedness of Iρ in Section
3. Some examples are presented in Section 4.

2. Some norm and integral estimates

Let us first consider the characteristic functions of balls. For every R > 0,
let BR := B(0, R) be the ball centered at 0 with radius R, and χBR be the
∫r
characteristic function of BR . Next, for all r > 0, let ρ̃(r) := 0 ρ(s)
s ds. Also
we write B(x, R) = {y ∈ R : |x − y| < R}.
d

The following lemmas will be used several times in this paper.

Lemma 2.1. There exists C > 0 such that the inequality ρ̃(R/2) ≤ C Iρ χBR (x)
holds whenever x ∈ BR/2 and R > 0.
Proof. Take x ∈ BR/2 . We write the integral out in full:
∫ ∫
ρ(|x − y|) ρ(|x − y|)
Iρ χBR (x) = χ BR (y) dy = dy.
Rd |x − y| BR |x − y|
d d

A geometric observation shows that B(x, R/2) ⊆ B(0, R). Hence, we have
∫ ∫ R/2
ρ(|x − y|) ρ(s)
Iρ χBR (x) ≥ dy = C ds.
B(x,R/2) |x − y| d
0 s
Note that we only use the spherical coordinates to obtain the last integral. □
 Generalized fractional integral operators on Morrey spaces 5

Lemma 2.2. For every R > 0 and a measurable function ϕ : Rd → (0, ∞)

satisfying the doubling condition
1 ϕ(s)
≤ ≤ C (0 < r/2 ≤ s ≤ 2r), (2.1)
C ϕ(r)
∫∞ ∫∞
the inequality C −1 2R ϕ(t)ρ(t)
t dt ≤ Iρ gR (x) ≤ C 2R/3 ϕ(t)ρ(t)
t dt holds when-
ever x ∈ BR/3 , where gR (x) = ϕ(|x|)χBRc (x).

Proof. We prove the right-hand inequality, the left-hand ∫ inequality being sim-
ilar. Writing Iρ gR (x) out in full, we obtain Iρ gR (x) = Rd \BR ϕ(|y|)ρ(|x−y|)
|x−y|d
dy.
Since ϕ satisfies (2.1), it follows that
∫ ∫
ρ(|x − y|) ρ(|y|)
Iρ gR (x) ≤ ϕ(|y|) dy ∼ ϕ(|y|) dy.
Rd \B(x,2R/3) |x − y| d
Rd \B2R/3 |y|d
∫∞
Hence Iρ gR (x) ≤ C 2R/3 ϕ(t)ρ(t)
t dt. It remains to write the most right-hand
side in terms of the spherical coordinates. □

The lemma below gives an estimate for the norm of χBR in Lp,ϕ (Rd ).

Lemma 2.3. Let 1 ≤ p < ∞ and ϕ ∈ Gp . There exists a constant C > 0 such
that C −1 ϕ(R)−1 ≤ ∥χBR : Lp,ϕ ∥ ≤ C ϕ(R)−1 for all R > 0.

Lemma 2.3 is proven in [13, Lemma 3.3] and see [5] as well.

Lemma 2.4. Let 1 ≤ p < ∞ and ϕ ∈ Gp . Assume that there exists a positive
constant C > 0 such that
∫ R
ϕ(t)td/p−1 dt ≤ Cϕ(R)Rd/p (R > 0). (2.2)
0

Then the function x 7→ ϕ(|x|) belongs to Lp,ϕ (Rd ).

Proof. Since there exists a non-increasing function ϕ1 such that ϕ(r) ≤ ϕ1 (r)
and that Lp,ϕ (Rd ) and Lp,ϕ1 (Rd ) are isomorphic [13, p.446], we can assume
that ϕ itself is decreasing. In this case x 7→ ϕ(|x|) is radial decreasing, so that
[ ∫ ]1/p [ ∫ ]1/p
1 1
p
ϕ(|x|) dx ≤ p
ϕ(|x|) dx (a ∈ Rd ).
|B(a, r)| B(a,r) |Br | Br
∫ ∫r
We have |B1r | Br ϕ(|x|)p dx = C r1d 0 ϕ(t)p td−1 dt ≤ Cϕ(r)p , in view of [17,
Lemma 7.1]. Combining these observations, we prove the lemma. □

3. Proof of Theorems 1.3, 1.4 and 1.5

A normalization allows us to assume ∥f : Lp,ϕ ∥ = 1 (resp. ∥f : L1,ϕ ∥ = 1)
in the proof of necessity of Theorems 1.3 and 1.4 (resp. Theorem 1.5).
6 ERIDANI, H. GUNAWAN, E. NAKAI and Y. SAWANO

3.1. Proof of the sufficiency

We remark that (1.4) includes (1.3). We prove the estimate (3.1). Once we
prove (3.1), the estimate (3.1) gives us the boundedness of Iρ from Lp,ϕ (Rd )
to Lq,ϕp/q (Rd ). Here we use the fact that the Hardy-Littlewood maximal
operator M is bounded on Lp,ϕ (Rd ), if ϕ satisfies the integral condition [12].
See [12, 19, 20] for more information about M on the space Lp,ϕ (Rd ).
Lemma 3.1. If we normalize the norm of f by ∥f : Lp,ϕ ∥ = 1, then we have
( )
|Iρ f (x)| ≤ C [M f (x)]p/q + inf ϕ(r)p/q , x ∈ Rd . (3.1)
r>0
[∑ ∑∞ ∫ ]
−1 ρ∗ (2j r)
Proof. We have |Iρ f (x)| ≤ C j=−∞ + j=0 (2j r)d |x−y|<2j r
|f (y)| dy
for given x ∈ Rd and r > 0. Recall that k1 and k2 appeared in the con-
∫k r
dition (1.1) of ρ. Let ρ∗ (r) = k12r ρ(s)
s ds. Let ΣI and ΣII be the first and
second summations above. Then, by the overlapping property [24], we have
−1
∑
ΣI ≤ C ρ∗ (2j r)M f (x) ≤ C ρ̃(k2 r)M f (x) ≤ C ϕ(r)p/q−1 M f (x),
j=−∞
∑∞ ∫ ∞
∗ ρ(s)ϕ(s)
ΣII ≤ C ρ (2j r)ϕ(2j r)∥f : Lp,ϕ ∥ ≤ C ds.
j=0 k1 r s

By (2.2) and the doubling property of ϕ, we obtain ΣII ≤ C ϕ(r)p/q . Hence,

[ ]
|Iρ f (x)| ≤ C ϕ(r)p/q−1 M f (x) + ϕ(r) (r > 0). (3.2)
Thus, we can assume M f (x) > inf r>0 ϕ(r). Otherwise (3.1) is immediate
from (3.2). From the definition of M f (x) and ∥f : Lp,ϕ ∥ = 1, and Hölder’s
inequality, one may observe that inf r>0 ϕ(r) ≤ M f (x) ≤ supr>0 ϕ(r). We
can thus find R > 0 such that 2C1
ϕ(R) ≤ M f (x) ≤ 2ϕ(R) and with this R
we can obtain (3.1). □
Proof of Theorem 1.5. For a ball B(z, r), let f1 = f χB(z,2r) and f2 = f − f1 .
∫
Then a geometric observation shows |Iρ f1 (x)| ≤ B(z,2r) |f (y)| ρ(|x−y|)
|x−y|d
dy for
all x ∈ B(z, 2r). Hence by the Fubini theorem and the normalization,
∫ ∫ ∫
ρ(|x − y|)
Iρ f1 (y) dy ≤ |f (y)| dx dy ≤ C ρ̃(3r)ϕ(3r)rd .
B(z,r) B(z,2r) B(y,3r) |x − y|d
Thus, the estimate for f1 is valid. As for f2 , we let x ∈ B(z, r). Then we have
∫ ∫
ρ(|x − y|) ρ(|x − y|)
|Iρ f2 (x)| ≤ |f (y)| dy ≤ |f (y)| dy
B(z,2r)c |x − y| d
B(x,r) c |x − y|d
and decomposing∑the right-hand side dyadically as we did in the proof of
Theorem 1.3 for II , we obtain
∑∞ ∫ ∫ ∞
ρ(|x − y|) ϕ(t)ρ(t)
|Iρ f2 (x)| ≤ |f (y)| dy ≤ C dt.
j
j=1 B(x,2 r)\B(x,2
j−1 r) |x − y| d
2k1 r t
 Generalized fractional integral operators on Morrey spaces 7

If we use (1.6), then we obtain |Iρ f2 (x)| ≤ Cψ(r). Thus, the estimate for f2
is valid as well. □
Remark 3.2. The proof of the sufficient part is similar to, but not the same as,
that in [7, 15]. In this paper, we do not assume that ρ satisfies the doubling
condition and that ϕ is surjective, as we did in [7].
3.2. Proof of necessity
[∫ ]1/q
Proof of Theorem 1.3. Note that ρ̃(R/2) ≤ C |I χ (x)|q dx
BR/2 ρ BR
by
Lemma 2.1. Notice also that ∥Iρ χBR : Lq,ϕp/q ∥ ≤ C∥χBR : Lp,ϕ ∥ since Iρ is
assumed bounded. If we invoke Lemma 2.1, and the doubling property of ϕ,
we have ρ̃(R/2) ≤ C ϕ(R/2)p/q ∥Iρ χBR : Lq,ϕp/q ∥ ≤ C ϕ(R/2)p/q−1 , for all
R > 0. This completes the proof. □
Proof of Theorem 1.4. If we integrate the conclusion of Lemma 2.1 over BR/2
and use the boundedness of Iρ is bounded from Lp,ϕ (Rd ) to Lq,ϕp/q (Rd ),
∫ r ρ(t)
0 t
dt ≤ Cϕ(r)p/q−1 for r > 0. By virtue of Lemma 2.2, we obtain
∫ ∞ ( ∫ )1/q
ϕ(t)ρ(t) 1
dt ∼ d
q
Iρ g(x) dx ≤ Cϕ(R)p/q ∥Iρ gR : Lq,ϕp/q ∥.
2R t R BR/3
∫∞
Since Iρ is bounded, we obtain 2R ϕ(t)ρ(t) dt ≤ Cϕ(R)p/q ∥gR : Lp,ϕ ∥. Now
∫ ∞ tϕ(t)ρ(t)
we invoke Lemma 2.4 to conclude 2R t dt ≤ Cϕ(R)p/q ≤ Cϕ(2R)p/q .
Thus, Theorem 1.4 is proven. □
Proof of Theorem 1.5. By Lemma 2.1 and ∥χBr : L1,ϕ ∥ ∼ ϕ(r)−1 , we obtain
∫ (r) ψ(r)
ρ̃(r) ∼ r−d Iρ χBr (x) dx ≤ ψ ∥Iρ χBr : L1,ψ ∥ ≤ C .
Br/2 2 ϕ(r)
∫∞ ∫
Meanwhile, by Lemma 2.2 and r ρ(t)ϕ(t) t dt ∼ r−d B2r/3 Iρ gr (x) dx, we have
∫ ∞ ( )
ρ(t)ϕ(t) 2r
dt ≤ Cψ ∥Iρ gr : L1,ψ ∥ ≤ Cψ(r)∥gr : L1,ϕ ∥ ≤ Cψ(r).
r t 3
Thus, Theorem 1.5 is proved. □

4. Examples
In this section, we show by examples that Theorems 1.3. 1.4 and 1.5 have
independent interest. Here and below we write
( )
 1 −β1
 log r (0 < r < e−1 ),
ℓβ1 ,β2 (r) := 1 (e−1 ≤ r ≤ e),

 β2
(log r) (e < r).
This function is used to describe the “log”-growth and “log”-decay properties.
Also, we fix p and q so that 1 < p < q < ∞. As Proposition 4.1 below shows,
generalized Morrey spaces occur natuarally.
8 ERIDANI, H. GUNAWAN, E. NAKAI and Y. SAWANO

Proposition 4.1. [25, Theorem 5.1] Let 1 < p < ∞ and s ∈ (0, d/p). There
∫ r d ℓ−1,0 (r)
exists Cp,s > 0 such that Br |f (y)|dy ≤ Cp,s (1+r) s ∥(1−∆)s/2 f : Lp,d−ps ∥
holds for all r > 0 and f ∈ Lp,d−ps (Rd ) with (1 − ∆)s/2 f ∈ Lp,d−ps (Rd ).
We can improve Proposition 4.1 to p = 1 while we cannot delete ℓ−1,0
because it is necessary condition for this estimate. See Example 5 below. The
following example deals more deeply with Proposition 4.1.
( )
Example 1. Let µ1 , µ2 satisfy µ1 , µ2 ≥ 0. Set α = dp − dq and βi = pq − 1 µi
for i = 1, 2. Define ρ(r) = rα ℓβ1 ,β2 (r), ϕ(r) = r− p ℓµ1 ,µ2 (r) for r > 0. Then
d

this pair (ρ, ϕ) fulfills the assumption (1.2) and (1.3) in Theorem 1.3 but fails
(1.5) in Theorem 1.4. More precisely, since α > 0, we have ρ̃(r) ∼ ρ(r) and
∫ ∞
ϕ(t)ρ(t)
dt ∼ ϕ(r)ρ(r).
r t
Example 1 is an endpoint case of the next example.
( )
Example 2. Let λ satisfy 0 < pq − 1 λ < d and − dp < λ < 0. Take µ1 , µ2
( ) ( )
arbitrarily. Set α = pq − 1 λ and βi = pq − 1 µi for i = 1, 2. Define
ρ(r) = rα ℓβ1 ,β2 (r) and ϕ(r) = rλ ℓµ1 ,µ2 (r) for r > 0. Then this pair (ρ, ϕ)
fulfills the assumption (1.2)–(1.5) in Theorems 1.3 and 1.4. Indeed, ρ̃(r) ∼
∫ ∞ ϕ(t)ρ(t)
ρ(r), r t dt ∼ ϕ(r)ρ(r).
The next example concerns the case when the spaces are close to L∞ (Rd )
and the smoothing order of Iρ is “almost 0”.
( )
Example 3. Let µ1 , µ2 < 0. Set β1 = pq − 1 µ1 + 1 ∈ (1, ∞) and β2 =
( )
p
q − 1 µ2 − 1 ∈ (−1, ∞). Define ρ(r) = ℓβ1 ,β2 (r) and ϕ(r) = ℓµ1 ,µ2 (r)
for r > 0. Then this pair (ρ, ϕ) fulfills the assumptions (1.4) and (1.5) in
Theorem 1.4 but fails (1.2) in Theorem 1.3. More precisely, for all r > 0, we
∫∞
have ρ̃(r) ∼ ℓβ1 −1,β2 +1 (r), and r ϕ(t)ρ(t)
t dt ∼ ℓµ1 +β1 −1,µ2 +β2 +1 (r).
We consider a case when the target space is close to L∞ (Rd ).
Example 4. Let α, β1 , µ1 , µ2 satisfy 0 < α < dp , µ1 + β1 < 1, µ2 < 0. Set
( )
β2 = pq − 1 µ2 − 1 ∈ (−1, ∞). Define ρ(r) = min(1, rα )ℓβ1 ,β2 (r) and ϕ(r) =
max(1, r−α )ℓµ1 ,µ2 (r) for r > 0. Then this pair (ρ, ϕ) fulfills the assumptions
(1.4) and (1.5) in Theorem 1.4 but fails (1.2) in Theorem 1.3. More precisely,
∫∞
ϕ(r)ρ̃(r) ∼ ℓµ1 +β1 ,µ2 +β2 +1 (r) and r ϕ(t)ρ(t)
t dt ∼ ℓµ1 +β1 −1,µ2 +β2 +1 (r) for
r > 0.
Finally, we explain how we improve Proposition 4.1 from our results.
Example 5. Let 0 < s < d. Define ϕ(r) = r−s and ψ(r) = (1 + r)−s ℓ−1,0 (r)
for r > 0. Let ρ(r) = rd Gs (r), where Gs denotes the Bessel kernel, the
kernel of (1 − ∆)s/2 . Observe that ρ̃(r) ∼ min(rs , 1) and hence ϕ(r)ρ̃(r) ∼
 Generalized fractional integral operators on Morrey spaces 9
{
∫∞ log(e/r) (r < 1),
min(1, r−s ). Note also that ϕ(t)ρ(t)
dt ∼ Then we
r t
r Gs (r) (r ≥ 1).
d−s
∫∞
have ϕ(r)ρ̃(r)+ r ϕ(t)ρ(t)
t dt ∼ ψ(r). Hence it follows from Theorem 1.5 that
∥Iρ f : L1,ψ ∥ ≤ C∥f : L1,ϕ ∥, extending Proposition 4.1. This triple (ρ, ϕ, ψ)
fulfills the assumptions (1.6) and (1.7) in Thoerem 1.5 but it fails (1.2) in
Theorem 1.3 and (1.4) in Theorem 1.4.

Acknowledgements. The research was initiated when the first and the second
authors visited Kyoto University under the GCOE 2011 Project.

References
[1] D. R. Adams, “A note on Riesz potentials”, Duke Math. J. 42 (1975), 765–778.
[2] V. Burenkov; and V. Guliyev, Necessary and sufficient conditions for the
boundedness of the Riesz potential in local Morrey-type spaces, Potential Anal.
30 (2009), no. 3, 211–249.
[3] F. Chiarenza and M. Frasca, “Morrey spaces and Hardy-Littlewood maximal
function”, Rend. Mat. 7 (1987), 273–279.
[4] Eridani; H. Gunawan; and E. Nakai, “On generalized fractional integral oper-
ators”, Sci. Math. Jpn. 60 (2004), 539–550.
[5] Eridani; H. Gunawan; and M.I. Utoyo, “A revisit on Nakai’s results on maximal
functions and frational integrals”, submitted.
[6] V. Guliyev; J. Hasanov; and Y. Zeren, “Necessary and sufficient conditions for
the boundedness of the Riesz potential in modified Morrey spaces”, J. Math.
Inequal. 5 (2011), no. 4, 491–506.
[7] H. Gunawan, “A note on the generalized fractional integral operators”, J. In-
dones. Math. Soc. 9 (2003), 39–43.
[8] H. Gunawan and Eridani, “Fractional integrals and generalized Olsen inequal-
ities”, Kyungpook Math. J. 49 (2009), 31–39.
[9] V. Guliyev; and R. Mustafayev, Boundedness of the anisotropic maximal
and anisotropic singular integral operators in generalized Morrey spaces, Acta
Math. Sin. (Engl. Ser.) 27 (2011), no. 12, 2361–2370.
[10] K. Kurata and S. Sugano, “A remark on estimates for uniformly elliptic op-
erators on weighted Lp spaces and Morrey spaces”, Math. Nachr. 209 (2000),
137–150.
[11] Y. Mizuta; E. Nakai; T. Ohno; and T. Shimomura, Boundedness of fractional
integral operators on Morrey spaces and Sobolev embeddings for generalized
Riesz potentials, J. Math. Soc. Japan 62 (2010), no. 3, 707–744.
[12] E. Nakai, “Hardy-Littlewood maximal operator, singular integral operators and
the Riesz potentials on generalized Morrey spaces”, Math. Nachr. 166 (1994),
95–103.
[13] E. Nakai, “A characterization of pointwise multipliers on the Morrey spaces”,
Sci. Math. 3 (2000), 445-454.
[14] E. Nakai, “On generalized fractional integrals”, Taiwanese J. Math. 5 (2001),
587–602.
10 ERIDANI, H. GUNAWAN, E. NAKAI and Y. SAWANO

[15] E. Nakai, “ On generalized fractional integrals on the weak Orlicz spaces”,

BMOφ , the Morrey spaces and the Campanato spaces, Function spaces, inter-
polation theory and related topics (Lund, 2000), 389–401, de Gruyter, Berlin,
2002.
[16] E. Nakai, “Orlicz-Morrey spaces and the Hardy-Littlewood maximal function”,
Studia Math. 188 (2008), no. 3, 193–221.
[17] E. Nakai, “A generalization of Hardy spaces H p by using atoms”, Acta Math.
Sin. (Engl. Ser.) 24 (2008), no. 8, 1243–1268.
[18] J. Peetre, “On the theory of Lp,λ spaces”, J. Funct. Anal. 4 (1969), 71–87.
[19] E. M. Stein, Singular Integrals and Differentiability Properties of Functions,
Princeton University Press, Princeton, New Jersey, 1970.
[20] E. M. Stein, Harmonic Analysis: Real Variable Methods, Orthogonality, and
Oscillatory Integrals, Princeton University Press, Princeton, New Jersey, 1993.
[21] Y. Sawano; S. Sugano; and H. Tanaka, “A note on generalized fractional inte-
gral operators on generalized Morrey spaces”, Bound. Value Probl. 2009, Art.
ID 835865, 18 pp.
[22] Y. Sawano, Generalized Morrey Spaces for Non-doubling Measures, Nonlinear
differ. equ. appl. 15 (2008), 413–425.
[23] Y. Sawano; S. Sugano; and H. Tanaka, “Generalized fractional integral oper-
ators and fractional maximal operators in the framework of Morrey spaces”,
Trans. Amer. Math. Soc. 363 (2011), no. 12, 6481–6503.
[24] Y. Sawano; S. Sugano and H. Tanaka, “Orlicz-Morrey spaces and fractional
operators”, Potential Anal. 36, No. 4 (2012), 517-556.
[25] Y. Sawano; and H. Wadade, On the Gagliardo-Nirenberg type inequality in the
critical Sobolev-Morrey space, J. Fourier Anal. Appl., online.
[26] S. Sugano; and H. Tanaka, “Boundedness of fractional integral operators on
generalized Morrey spaces”, Sci. Math. Jpn. Online 8 (2003), 233–242.

ERIDANI
Department of Mathematics, Airlangga University, Surabaya 60115, Indonesia.
e-mail: [email protected]
Hendra GUNAWAN
Department of Mathematics, Bandung Institute of Technology, Bandung 40132,
Indonesia.
e-mail: [email protected]
Eiichi NAKAI
Department of Mathematics, Ibaraki University, Mito, Ibaraki 310-8512, Japan
e-mail: [email protected]
Yoshihiro SAWANO
Department of Mathematics and Information Scinence, Tokyo Metropolitan Uni-
versity, 1-1 Minami-Ohsawa, Hachioji, 192-0397, Japan.
e-mail: [email protected]