BOUNDEDNESS OF MULTILINEAR OPERATORS ON TRIEBEL-LIZORKIN SPACES

PII. S016117120430147X http://ijmms.hindawi.com

© Hindawi Publishing Corp.

BOUNDEDNESS OF MULTILINEAR OPERATORS ON TRIEBEL-LIZORKIN SPACES

LIU LANZHE Received 20 January 2003

The purpose of this paper is to study the boundedness in the context of Triebel-Lizorkin spaces for some multilinear operators related to certain convolution operators. The opera- tors include Littlewood-Paley operator, Marcinkiewicz integral, and Bochner-Riesz operator.

2000 Mathematics Subject Classification: 42B20, 42B25.

1. Introduction. LetTbe a Calderon-Zygmund operator. A well-known result of Coif- man et al. [6] states that the commutator [b, T ]=T (bf )−bT f (whereb∈BMO) is bounded onL^p(Rⁿ)for 1< p <∞; Chanillo [1] proves a similar result whenT is re- placed by the fractional integral operator. In [7, 9], Janson and Paluszy´nski extend these results to the Triebel-Lizorkin spaces and the caseb∈Lipβ(where Lipβis the homogeneous Lipschitz space). The main purpose of this paper is to discuss the bound- edness of some multilinear operators related to certain convolution operators in the context of Triebel-Lizorkin spaces. In fact, we will establish the boundedness on the Triebel-Lizorkin spaces for some multilinear operators related to certain convolution operator only under certain conditions on the size of the operators. As applications, we obtain the boundedness of the multilinear operators related to the Marcinkiewicz inte- gral, Littlewood-Paley operator, and Bochner-Riesz operator in the context of Triebel- Lizorkin spaces.

2. Preliminaries. Throughout this paper, M(f ) will denote the Hardy-Littlewood maximal function off,Mpf=(M(f^p))^1/p forp >0, andQwill denote a cube ofRⁿ with sides parallel to the axes. For a cubeQ, letfQ= |Q|⁻¹

Qf (x)dxand f^#(x)= supx∈Q|Q|⁻¹

Q|f (y)−fQ|dy. For β >0 and p >1, let ˙Fp^β,∞ be the homogeneous Triebel-Lizorkin space. The Lipschitz space ˙∧βis the space of functionsfsuch that

f∧˙β= sup

x,h∈Rⁿ h≠0

∆^[β]h ⁺¹f (x)

|h|^β <∞, (2.1)

where∆^khdenotes thekth difference operator (see [9]).

The operators considered in this paper are following several sublinear operators.

Letmbe a positive integer and letAbe a function onRⁿ. We denote Rm+1(A;x, y)=A(x)−

|α|≤m

1

α!D^αA(y)(x−y)^α. (2.2)

Definition2.1. Letε >0 and letψbe a fixed function which satisfies the following properties:

(1) |ψ(x)| ≤C(1+|x|)⁻⁽ⁿ⁺¹⁾,

(2) |ψ(x+y)−ψ(x)| ≤C|y|^ε(1+|x|)^−(n+1+ε)when 2|y|<|x|. The multilinear Littlewood-Paley operator is defined by

g^A_ψ(f )(x)= _∞

0

F_t^A(f )(x)²dt t

1/2

, (2.3)

where

F_t^A(f )(x)=

Rⁿψt(x−y)Rm+1(A;x, y)

|x−y|^m f (y)dy (2.4) andψt(x)=t⁻ⁿψ(x/t)fort >0. DenoteFt(f )=ψt∗f. Also define

gψ(f )(x)=∞ 0

Ft(f )(x)²dt t

1/2

(2.5) which is the Littlewood-Paleygfunction (see [10]).

LetH be the spaceH= {h:h =(_∞

0 |h(t)|²dt/t)^1/2<∞}. Then, for each fixed x∈Rⁿ,F_t^A(f )(x)may be viewed as a mapping from[0,+∞)toH, and it is clear that

gψ(f )(x)=Ft(f )(x), g_ψ^A(f )(x)=F_t^A(f )(x). (2.6) Definition2.2. Let 0< γ≤1 and letΩ be homogeneous of degree zero onRⁿ such that

Sⁿ⁻¹Ω(x)dσ (x)=0. Assume thatΩ∈Lip_γ(Sⁿ⁻¹), that is, there exists a constantM >0 such that for anyx, y∈Sⁿ⁻¹,|Ω(x)−Ω(y)| ≤M|x−y|^γ. The multi- linear Marcinkiewicz integral operator is defined by

µ^A_Ω(f )(x)= _∞

0

F_t^A(f )(x)²dt t³

1/2

, (2.7)

where

F_t^A(f )(x)=

|x−y|≤t

Ω(x−y)

|x−y|ⁿ⁻¹

Rm+1(A;x, y)

|x−y|^m f (y)dy. (2.8) Denote

Ft(f )(x)=

|x−y|≤t

Ω(x−y)

|x−y|ⁿ⁻¹f (y)dy. (2.9) Also define

µ_Ω(f )(x)= _∞

0

Ft(f )(x)²dt t³

1/2

(2.10) which is the Marcinkiewicz integral (see [11]).

LetHbe the spaceH= {h:h =(_∞

0 |h(t)|²dt/t³)^1/2<∞}. Then, it is clear that µ_Ω(f )(x)=Ft(f )(x), µ_Ω^A(f )(x)=F_t^A(f )(x). (2.11)

Definition2.3. LetB^δ_t(f )(ξ)ˆ =(1−t²|ξ|²)^δ₊f (ξ). Denoteˆ B_δ,t^A (f )(x)=

RⁿB^δ_t(x−y)Rm+1(A;x, y)

|x−y|^m f (y)dy, (2.12) whereB^δ_t(z)=t⁻ⁿB^δ(z/t)fort >0. The maximal multilinear Bochner-Riesz operator is defined by

B_δ,∗^A (f )(x)=sup

t>0

B_δ,t^A (f )(x). (2.13)

Also define

B^δ_∗(f )(x)=sup

t>0

B^δ_t(f )(x) (2.14)

which is the Bochner-Riesz operator (see [7,8]).

LetHbe the spaceH= {h:h =sup_t>0|h(t)|<∞}, then it is clear that

B^δ_∗(f )(x)=B_t^δ(f )(x), B_δ,∗^A (f )(x)=B^A_δ,t(f )(x). (2.15) More generally, we consider the following multilinear operators related to certain convolution operators.

Definition2.4. LetK(x, t)be defined onRⁿ×[0,+∞). Denote that Ktf (x)=

RⁿK(x−y, t)f (y)dy, K_t^Af (x)=

Rⁿ

Rm+1(A;x, y)

|x−y|^m K(x−y, t)f (y)dy.

(2.16)

Let H be the normed space H= {h:h<∞}. For each fixedx ∈Rⁿ, Ktf (x) and K^A_t(f )(x)are viewed as a mapping from[0,+∞)toH. Then, the multilinear operators related toKtis defined by

TAf (x)=K_t^A(f )(x); (2.17) also defineT f (x)= Ktf (x).

It is clear that Definitions2.1,2.2, and2.3are the particular examples ofDefinition 2.4. Note that whenm=0,TAis just the commutator ofKtandA. It is well known that multilinear operators are of great interest in harmonic analysis and have been widely studied by many authors (see [2,3,4,5]). The main purpose of this paper is to consider the continuity of the multilinear operators on Triebel-Lizorkin spaces. We will prove the following theorems inSection 3.

Theorem2.5. Letg_ψ^Abe the multilinear Littlewood-Paley operator as inDefinition 2.1 and let0< β <min(1, ε),1< p <∞, andD^αA∈∧˙βfor|α| =m. Then

(a) g_ψ^Ais bounded fromL^p(Rⁿ)toF˙p^β,∞(Rⁿ),

(b) g_ψ^Ais bounded fromL^p(Rⁿ)toL^q(Rⁿ)for1/p−1/q=β/nand1/p > β/n.

Theorem2.6. Letµ_Ω^Abe the multilinear Marcinkiewicz integral operator as inDefini- tion 2.2and let0< γ≤1,0< β <min(1/2, γ),1< p <∞, andD^αA∈∧˙βfor|α| =m.

Then

(a) µ_Ω^Ais bounded fromL^p(Rⁿ)toF˙p^β,∞(Rⁿ),

(b) µ_Ω^Ais bounded fromL^p(Rⁿ)toL^q(Rⁿ)for1/p−1/q=β/nand1/p > β/n.

Theorem 2.7. Let B^A_δ,∗ be the maximal multilinear Bochner-Riesz operator as in Definition 2.3 and let δ > (n−1)/2, 0< β <min(1, δ−(n−1)/2), 1< p <∞, and D^αA∈∧˙βfor|α| =m. Then

(a) B_δ,∗^A is bounded fromL^p(Rⁿ)toF˙p^β,∞(Rⁿ);

(b) B_δ,∗^A is bounded fromL^p(Rⁿ)toL^q(Rⁿ)for1/p−1/q=β/nand1/p > β/n.

3. Main theorem and proof. First, we will establish the following theorem.

Theorem3.1. Let0< β <1,1< p <∞, andD^αA∈∧˙βfor|α| =m. LetKt,T, and TA be the same as inDefinition 2.4. IfT is bounded onL^q(Rⁿ)forq∈(1,+∞)andTA

satisfies the size condition

K_t^A(f )(x)−K_t^A(f )

x0 ≤C

|α|=m

D^αA_˙

∧β|Q|^β/nM(f )(x) (3.1)

for any cubeQwithsuppf⊂(2Q)^candx∈Q, then (a) TAis bounded fromL^p(Rⁿ)toF˙p^β,∞(Rⁿ),

(b) TAis bounded fromL^p(Rⁿ)toL^q(Rⁿ)for1/p−1/q=β/nand1/p > β/n.

To prove the theorem, we need the following lemmas.

Lemma3.2(see [9]). For0< β <1and1< p <∞, fF˙_p^β,∞≈

sup

Q

1

|Q|^1+β/n

Q

f (x)−fQdx _Lp

≈ sup

·∈Qinf

c

1

|Q|^1+β/n

Q

f (x)−cdx _Lp.

(3.2)

Lemma3.3(see [9]). For0< β <1and1≤p≤ ∞, f˙∧β≈sup

Q

1

|Q|^1+β/n

Q

f (x)−fQdx

≈sup

Q

1

|Q|^β/n 1

|Q|

Q

f (x)−fQ^pdx 1/p

.

(3.3)

Lemma3.4(see [1]). For1≤r <∞andδ >0, let Mδ,r(f )(x)=sup

x∈Q

1

|Q|^{1−δr /n}

Q

f (y)^pdy 1/p

. (3.4)

Suppose thatr < p < δ/nand1/q=1/p−δ/n. ThenMδ,r(f )L^q≤CfL^p.

Lemma3.5(see [9]). LetQ1⊂Q2. Then

fQ₁−fQ₂≤Cf∧˙βQ2^β/n. (3.5) Lemma3.6(see [4]). LetAbe a function onRⁿ andD^αA∈L^q(Rⁿ)for|α| =mand someq > n. Then

Rm(A;x, y)≤C|x−y|^m

|α|=m

1 Q(x, y)˜

Q(x,y)˜

D^αA(z)^qdz 1/q

, (3.6)

whereQ(x, y)˜ is the cube centered atxand having side length5√

n|x−y|. Proof ofTheorem3.1. (a) Fix a cubeQ=Q(x0, l)and ˜x∈Q. Let ˜Q=5√

nQand A(x)˜ =A(x)−

|α|=m(1/α!)(D^αA)Q˜x^α, then Rm(A;x, y)=Rm(A;˜x, y) and D^αA˜= D^αA−(D^αA)Q˜ for|α| =m. Forf1=f χQ˜andf2=f χ_Rn\Q˜,

K_t^A(f )(x)=

Rⁿ

Rm+1A;˜x, y

|x−y|^m K(x−y, t)f (y)dy

=

Rⁿ

Rm+1A;˜x, y

|x−y|^m K(x−y, t)f (y)dy +

Rⁿ

RmA;˜x, y

|x−y|^m K(x−y, t)f1(y)dy

−

|α|=m

1 α!

Rⁿ

K(x−y, t)(x−y)^α

|x−y|^m D^αA(y)f˜ 1(y)dy,

(3.7)

then

TA(f )(x)−TA˜

f2 x0 =K_t^A(f )(x)−K_t^A˜ f2 x0

≤ Kt

RmA;˜x,·

|x−·|^m f1

(x)

+

|α|=m

1 α!

Kt

(x−·)^α

|x−·|^mD^αAf˜ 1

(x)

+K_t^A˜

f2 (x)−K_t^A˜ f2 x0

=A(x)+B(x)+C(x).

(3.8)

Thus, 1

|Q|^1+β/n

Q

TAf (x)−TA˜(f ) x0 dx

≤ 1

|Q|^1+β/n

Q

A(x)dx+ 1

|Q|^1+β/n

Q

B(x)dx+ 1

|Q|^1+β/n

Q

C(x)dx :=I+II+III.

(3.9)

Now, we estimate I, II, and III, respectively. First, forx∈Qandy∈Q, using Lemmas˜ 3.3and3.6, we get

Rm

A;˜x, y ≤C|x−y|^m

|α|=m

sup

x∈Q˜

D^αA(x)−

D^αA Q˜

≤C|x−y|^m|Q|^β/n

|α|=m

D^αA_˙∧

β. (3.10)

Thus, by Holder’s inequality and theL^r boundedness ofT for 1< r < p, we obtain I≤C

|α|=m

D^αA_∧˙

β

1

|Q|

Q

T

f1 (x)dx

≤C

|α|=m

D^αA_∧˙

βT

f1 L^r|Q|^−1/r

≤C

|α|=m

D^αA_∧˙

βf1_Lr|Q|^−1/r

≤C

|α|=m

D^αA_∧˙

βMr(f )(x).˜

(3.11)

Secondly, for 1< r < q, using the inequality (see [9]) D^αA−

D^αA Q˜f χQ˜_Lr ≤C|Q|^1/r+β/nD^αA_∧˙

βMr(f )(x), (3.12) and similar to the proof of I, we obtain

II≤ C

|Q|^1+β/n

|α|=m

T D^αA−

D^αA Q˜ f χQ˜ L^r|Q|^1−1/r

≤C|Q|^{−β/n−1/r}

|α|=m

D^αA−

D^αA Q˜ f χQ˜_Lr

≤C

|α|=m

D^αA_˙

∧βMr(f )(x).˜

(3.13)

For III, using the size condition ofTA, we have III≤C

|α|=m

D^αA_˙

∧βM(f )(˜x). (3.14)

Putting these estimates together, taking the supremum over allQsuch that ˜x∈Q, and using theL^pboundedness ofMr forr < p, we obtain

TA(f )_˙

F_p^β,∞≤C

|α|=m

D^αA_∧˙

βfL^p. (3.15)

This completes the proof of (a).

(b) By the same argument as in the proof of (a), we have 1

|Q|

Q

TA(f )(x)−TA˜

f2 x0 dx≤C

|α|=m

D^αA_∧˙

β

Mβ,r(f )+Mβ,1(f ) , (3.16)

thus,

TA(f ) ^#≤C

|α|=m

D^αA_∧˙

β

Mβ,r(f )+Mβ,1(f ) . (3.17)

Now, usingLemma 3.4, we obtain TA(f )_Lq≤CTA(f ) ^#_Lq

≤C

|α|=m

D^αA_˙∧

βMβ,r(f )_Lq+Mβ,1(f )_Lq ≤CfL^p. (3.18) This completes the proof of (b) and the theorem.

To prove Theorems2.5,2.6, and2.7, it suffices to verify thatg_ψ^A,µ^A_Ω, andB_δ,^A_∗satisfy the size condition in theTheorem 3.1.

Suppose suppf⊂Q˜^candx∈Q=Q(x0, l). Note that|x0−y| ≈ |x−y|fory∈Q˜^c. Forg_ψ^A, we write

F_t^A˜(f )(x)−F_t^A˜(f ) x0

=

Rⁿ\Q˜

ψt(x−y)

|x−y|^m −ψt

x0−y x0−y^m

Rm

A;˜x, y f (y)dy

+

Rⁿ\Q˜

ψt

x0−y f (y) x0−y^m

Rm

A;x, y˜ −Rm

A;˜x0, y dy

−

|α|=m

1 α!

Rⁿ\Q˜

ψt(x−y)(x−y)^α

|x−y|^m −ψt

x0−y x0−y ^α x0−y^m

×D^αA(y)f (y)dy˜

=I1+I2+I3.

(3.19)

By the condition ofψ, we obtain I1≤C

Rⁿ\Q˜

x−x0 x0−y^m+1Rm

A;˜x, y f (y)

× _∞

0

t dt

t+x0−y ²⁽ⁿ⁺¹⁾ 1/2

dy +C

Rⁿ\Q˜

x−x0^ε x0−y^mRm

A;˜x, y f (y)

× _∞

0

t dt

t+x0−y ^2(n+1+ε) 1/2

dy

≤C

Rⁿ\Q˜

x−x0

x0−y^m+n+1+ x−x0^ε x0−y^m+n+ε

Rm

A;˜x, y f (y)dy

≤C

|α|=m

D^αA_˙

∧β|Q|^β/n

× ∞ k=0

2^k+1Q\2˜ ^k+1Q˜

x−x0

x0−yⁿ⁺¹+ x−x0^ε x0−y^n+ε

f (y)dy

≤C

|α|=m

D^αA_∧˙

β|Q|^β/n ∞ k=1

2^−k+2^−kε 1 2^kQ˜

2^kQ˜

f (y)dy

≤C

|α|=m

D^αA_∧˙

β|Q|^β/n ∞ k=1

2^−k+2^−kε M(f )(x)

≤C

|α|=m

D^αA_∧˙

β|Q|^β/nM(f )(x).

(3.20) For I2, by the formula (see [4])

RmA;˜x, y −RmA;˜x0, y =

|η|<m

1

η!Rm−|η|

D^ηA;x, x˜ 0 (x−y)^η (3.21)

andLemma 3.6, we get Rm

A;x, y˜ −Rm

A;˜x0, y ≤C

|α|=m

D^αA_˙∧

β|Q|^β/nx−x0x0−y^m−1. (3.22) Thus, similar to the proof of I1,

I2≤C

Rⁿ\Q˜

Rm

A;˜x, y −Rm

A;˜x0, y

x0−y^m+n f (y)dy

≤C

|α|=m

D^αA_∧˙

β|Q|^β/n ∞ k=0

2^k+1Q\2˜ ^kQ˜

x−x0

x0−yⁿ⁺¹f (y)dy

≤C

|α|=m

D^αA_∧˙

β|Q|^β/nM(f )(x).

(3.23)

For I3, byLemma 3.5, we get D^αA(y)−

D^αA Q˜≤D^αA_˙

∧βx0−y^β. (3.24) Thus, similar to the proof of I1, we obtain

I3≤C

|α|=m

Rⁿ\Q˜

x−x0

x0−yⁿ⁺¹+ x−x0^ε x0−y^n+ε

f (y)D^αA(y)˜ dy

≤C

|α|=m

D^αA_˙

∧β|Q|^β/n ∞ k=1

2^k(β−1)+2^k(β−ε) M(f )(x)

≤C

|α|=m

D^αA_˙∧

β|Q|^β/nM(f )(x).

(3.25)

So,

F_t^A˜(f )(x)−F_t^A˜(f )

x0 ≤C

|α|=m

D^αA_∧˙

β|Q|^β/nM(f )(x). (3.26)

Forµ^A_Ω, we write F_t^A˜(f )(x)−F_t^A˜(f )

x0

≤



_∞

0

|x−y|≤t

Ω(x−y)Rm

A;˜x, y

|x−y|^m+n−1 f (y)dy

−

|x0−y|≤t

Ω

x0−y Rm

A;˜x0, y

x0−y^m+n−1 f (y)dy

2dt t³





1/2

+C

|α|=m



_∞

0

|x−y|≤t

Ω(x−y)(x−y)^α

|x−y|^m+n−1

−

|x₀−y|≤t

Ω

x0−y x0−y ^α x0−y^m+n−1

×D^αA(y)f (y)dy˜

2dt t³





1/2

≤



_∞

0

|x−y|≤t,|x₀−y|>t

Ω(x−y)Rm

A;˜x, y

|x−y|^m+n−1 f (y)dy 2

dt t³





1/2

+



_∞

0

|x−y|>t,|x0−y|≤t

Ω

x0−y Rm

A;x˜ 0, y

x0−y^m+n−1 f (y)dy 2

dt t³





1/2

+



_∞

0

|x−y|≤t,|x0−y|≤t

Ω(x−y)Rm

A;˜x, y

|x−y|^m+n−1 −Ω

x0−y Rm

A;˜x0, y x0−y^m+n−1

×f (y)dy 2

dt t³





1/2

+C

|α|=m



_∞

0

|x−y|≤t

Ω(x−y)(x−y)^α

|x−y|^m+n−1

−

|x₀−y|≤t

Ω

x0−y x0−y ^α x0−y^m+n−1

×D^αA(y)f (y)dy˜

2dt t³





1/2

:=J1+J2+J3+J4.

(3.27)

Thus

J1≤C

Rⁿ\Q˜

f (y)Rm

A;˜x, y

|x−y|^m+n−1

|x−y|≤t<|x₀−y|

dt t³

1/2

dy

≤C

Rⁿ\Q˜

f (y)Rm

A;˜x, y

|x−y|^m+n−1

x0−x^1/2

|x−y|^3/2 dy

≤C

|α|=m

D^αA_˙

∧β|Q|^β/n∞

k=1

2^−k/22^kQ˜⁻¹

2^kQ˜

f (y)dy

≤C

|α|=m

D^αA_∧˙

β|Q|^β/nM(f )(x).

(3.28)

Similarly, we haveJ2≤C

|α|=mD^αA∧˙β|Q|^β/nM(f )(x).

ForJ3, by the inequality (see [11])

Ω(x−y)

|x−y|^m+n−1− Ω x0−y x0−y^m+n−1

≤C x−x0

x0−y^m+n+ x−x0^γ x0−y^m+n−1+γ

, (3.29)

we obtain

J3≤C

|α|=m

D^αA_∧˙

β|Q|^β/n

Rⁿ\Q˜

x−x0

x0−yⁿ+ x−x0^γ x0−y^n−1+γ

×

|x₀−y|≤t,|x−y|≤t

dt t³

1/2

f (y)dy

≤C

|α|=m

D^αA_∧˙

β|Q|^β/n ∞ k=1

2^−k+2^−γk M(f )(x)

≤C

|α|=m

D^αA_˙

∧β|Q|^β/nM(f )(x).

(3.30)

ForJ4, similar to the proof ofJ1,J2, andJ3, we obtain

J4≤C

|α|=m

Rⁿ\Q˜

x−x0

x0−yⁿ⁺¹+ x−x0^1/2

x0−y^n+1/2+ x−x0^γ x0−y^n+γ

×D^αA(y)˜ f (y)dy

≤C

|α|=m

D^αA_˙

∧β|Q|^β/n

× ∞ k=1

2^k(β−1)+2^k(β−1/2)+2^k(β−γ) 1 2^kQ˜

2^kQ˜

f (y)dy

≤C

|α|=m

D^αA_∧˙

β|Q|^β/nM(f )(x).

(3.31)

ForB^A_δ,∗, we write

B_δ,t^A˜(f )(x)−B_δ,t^A˜(f ) x0

=

Rⁿ\Q˜

B_t^δ(x−y)

|x−y|^m −B_t^δ x0−y x0−y^m

Rm

A;˜x, y f (y)dy

+

Rⁿ\Q˜

B_t^δ x0−y x0−y^m

Rm

A;˜x, y −Rm

A;˜x0, y f (y)dy

−

|α|=m

1 α!

Rⁿ\Q˜

B^δ_t(x−y)(x−y)^α

|x−y|^m −B_t^δ

x0−y x0−y ^α x0−y^m

×D^αA(y)f (y)dy˜

=L1+L2+L3.

(3.32)

We consider the following two cases.

Case1(0< t≤l). In this case, notice that (see [8]) B^δ(z)≤c

1+|z| ^{−(δ+(n+1)/2)}. (3.33)

We obtain L1≤Ct⁻ⁿ

Rⁿ\Q˜

f (y)Rm

A;˜x, y x0−y^m

1+|x−y|/t −(δ+(n+1)/2)dy

≤C

|α|=m

D^αA_˙∧

β|Q|^β/n(t/l)^{δ−(n−1)/2}

× ∞ k=1

2k((n−1)/2−δ) 1 2^kQ˜

2^kQ˜

f (y)dy

≤C

|α|=m

D^αA_˙∧

β|Q|^β/nM(f )(x), L2≤Ct⁻ⁿ

Rⁿ\Q˜

f (y)Rm

A;˜x, y −Rm

A;˜x0, y x0−y^m

×

1+|x−y|/t ^{−(δ+(n+1)/2)}dy

≤C

|α|=m

D^αA_˙

∧β|Q|^β/n(t/l)^{δ−(n−1)/2}

× ∞ k=1

2^{k((n−1)/2−δ)} 1 2^kQ˜

2^kQ˜

f (y)dy

≤C

|α|=m

D^αA_˙∧

β|Q|^β/nM(f )(x).

(3.34)

ForL3, similar to the proof ofL1, we get L3≤C

|α|=m

D^αA_∧˙

β|Q|^β/n(t/l)^{δ−(n−1)/2}

× ∞ k=1

2^{k(β−δ+(n−1)/2)} 1 2^kQ˜

2^kQ˜

f (y)dy

≤C

|α|=m

D^αA_˙

∧β|Q|^β/nM(f )(x).

(3.35)

Case2(t > l). In this case, we chooseδ0such thatβ+(n−1)/2< δ0<min(δ, (n+ 1)/2). Notice that (see [8])

(∂/∂z)B^δ(z)≤C

1+|z| −(δ+(n+1)/2). (3.36)

Similar to the proof ofCase 1, we obtain L1≤Ct⁻ⁿ

Rⁿ\Q˜

f (y)Rm

A;˜x, y x0−y^m+1

×x0−x1+x0−y/t ^{−(δ0+(n+1)/2)}dy +Ct⁻ⁿ⁻¹

Rⁿ\Q˜

f (y)RmA;˜x, y x0−y^m

×x0−x1+x0−y/t ^{−(δ0+(n+1)/2)}dy

≤C

|α|=m

D^αA_∧˙

β|Q|^β/n(l/t)^{(n+1)/2−δ0}

× ∞ k=1

2k((n−1)/2−δ0) 1 2^kQ˜

2^kQ˜

f (y)dy

≤C

|α|=m

D^αA_∧˙

β|Q|^β/nM(f )(x), L2≤Ct⁻ⁿ

Rⁿ\Q˜

f (y)RmA;˜x, y −RmA;˜x0, y x0−y^m

×

1+x0−y/t ^{−(δ0+(n+1)/2)}dy

≤C

|α|=m

D^αA_˙

∧β|Q|^β/n(l/t)^{(n+1)/2−δ0}

× ∞ k=1

2^{k((n−1)/2−δ0)} 1 2^kQ˜

2^kQ˜

f (y)dy

≤C

|α|=m

D^αA_∧˙

β|Q|^β/nM(f )(x),

L3≤C

|α|=m

D^αA_∧˙

β|Q|^β/n(l/t)^{(n+1)/2−δ0}

× ∞ k=1

2k(β+(n−1)/2−δ0) 1 2^kQ˜

2^kQ˜

f (y)dy

≤C

|α|=m

D^αA_˙

∧β|Q|^β/nM(f )(x).

(3.37) These yield the desired results.

Acknowledgment. This work was supported by the National Natural Science Foun- dation (NNSF) Grant 10271071.

References

[1] S. Chanillo,A note on commutators, Indiana Univ. Math. J.31(1982), no. 1, 7–16.

[2] W. G. Chen,A Besov estimate for multilinear singular integrals, Acta Math. Sin. (Engl. Ser.) 16(2000), no. 4, 613–626.

[3] J. Cohen,A sharp estimate for a multilinear singular integral inRⁿ, Indiana Univ. Math. J.

30(1981), no. 5, 693–702.

[4] J. Cohen and J. Gosselin,A BMO estimate for multilinear singular integrals, Illinois J. Math.

30(1986), no. 3, 445–464.

[5] J. Cohen and J. A. Gosselin,On multilinear singular integrals onRⁿ, Studia Math.72(1982), no. 3, 199–223.

[6] R. R. Coifman, R. Rochberg, and G. Weiss,Factorization theorems for Hardy spaces in sev- eral variables, Ann. of Math. (2)103(1976), no. 3, 611–635.

[7] S. Janson,Mean oscillation and commutators of singular integral operators, Ark. Mat.16 (1978), no. 2, 263–270.

[8] S. Z. Lu,Four Lectures on RealH^pSpaces, World Scientific Publishing, New Jersey, 1995.

[9] M. Paluszy´nski,Characterization of the Besov spaces via the commutator operator of Coif- man, Rochberg and Weiss, Indiana Univ. Math. J.44(1995), no. 1, 1–17.

[10] A. Torchinsky,Real-Variable Methods in Harmonic Analysis, Pure and Applied Mathemat- ics, vol. 123, Academic Press, Florida, 1986.

[11] A. Torchinsky and S. L. Wang,A note on the Marcinkiewicz integral, Colloq. Math.60/61 (1990), no. 1, 235–243.

Liu Lanzhe: Department of Applied Mathematics, Hunan University, Changsha 410082, China E-mail address:[email protected]

Journal of Applied Mathematics and Decision Sciences

Special Issue on

Intelligent Computational Methods for Financial Engineering

Call for Papers

As a multidisciplinary field, financial engineering is becom- ing increasingly important in today’s economic and financial world, especially in areas such as portfolio management, as- set valuation and prediction, fraud detection, and credit risk management. For example, in a credit risk context, the re- cently approved Basel II guidelines advise financial institu- tions to build comprehensible credit risk models in order to optimize their capital allocation policy. Computational methods are being intensively studied and applied to im- prove the quality of the financial decisions that need to be made. Until now, computational methods and models are central to the analysis of economic and financial decisions.

However, more and more researchers have found that the financial environment is not ruled by mathematical distribu- tions or statistical models. In such situations, some attempts have also been made to develop financial engineering mod- els using intelligent computing approaches. For example, an artificial neural network (ANN) is a nonparametric estima- tion technique which does not make any distributional as- sumptions regarding the underlying asset. Instead, ANN ap- proach develops a model using sets of unknown parameters and lets the optimization routine seek the best fitting pa- rameters to obtain the desired results. The main aim of this special issue is not to merely illustrate the superior perfor- mance of a new intelligent computational method, but also to demonstrate how it can be used e

ectively in a financial engineering environment to improve and facilitate financial decision making. In this sense, the submissions should es- pecially address how the results of estimated computational models (e.g., ANN, support vector machines, evolutionary algorithm, and fuzzy models) can be used to develop intelli- gent, easy-to-use, and/or comprehensible computational sys- tems (e.g., decision support systems, agent-based system, and web-based systems)

This special issue will include (but not be limited to) the following topics:

• Computational methods

: artificial intelligence, neu- ral networks, evolutionary algorithms, fuzzy inference, hybrid learning, ensemble learning, cooperative learn- ing, multiagent learning

• Application fields

: asset valuation and prediction, as- set allocation and portfolio selection, bankruptcy pre- diction, fraud detection, credit risk management

• Implementation aspects

: decision support systems, expert systems, information systems, intelligent agents, web service, monitoring, deployment, imple- mentation

Authors should follow the Journal of Applied Mathemat- ics and Decision Sciences manuscript format described at the journal site

Prospective authors should submit an electronic copy of their complete manuscript through the journal Manuscript Track- ing System at

lowing timetable:

Manuscript Due December 1, 2008 First Round of Reviews March 1, 2009 Publication Date June 1, 2009

Guest Editors

Lean Yu,

Academy of Mathematics and Systems Science, Chinese Academy of Sciences, Beijing 100190, China;

Department of Management Sciences, City University of Hong Kong, Tat Chee Avenue, Kowloon, Hong Kong;

[email protected]

Shouyang Wang,

Academy of Mathematics and Systems Science, Chinese Academy of Sciences, Beijing 100190, China; [email protected]

K. K. Lai,

Department of Management Sciences, City University of Hong Kong, Tat Chee Avenue, Kowloon, Hong Kong; [email protected]

Hindawi Publishing Corporation http://www.hindawi.com