• 検索結果がありません。

.)+641)61.6-652)+-5*)5-,4-6+)55-5),5--9-*-,,1/60-4-5.4));61+52)+-5160-716,15

N/A
N/A
Protected

Academic year: 2022

シェア ".)+641)61.6-652)+-5*)5-,4-6+)55-5),5--9-*-,,1/60-4-5.4));61+52)+-5160-716,15"

Copied!
5
0
0

読み込み中.... (全文を見る)

全文

(1)

ISSN: 1821-1291, URL: http://www.bmathaa.org Volume 2 Issue 2(2010), Pages 50-54

ON FACTORIZATION OF TENT SPACES BASED ON LORENTZ CLASSES AND SOME NEW EMBEDDING THEOREMS FOR

ANALYTIC SPACES IN THE UNIT DISK

ROMI SHAMOYAN, WEN XU

Abstract. We provide new assertions on factorization of tent spaces and based on them provide new embedding theorems for some analytic spaces in the unit disk.

1. Introduction

In this note, we provide new assertions concerning strong factorization of so called tent spaces. In order to formulate our results we will need some standard definitions ([3, 4, 5]).

Let

𝑅𝑛+1+ ={(𝑥, 𝑡) :𝑥∈𝑅𝑛, 𝑡 >0}, Γ(𝑥) ={(𝑦, 𝑡)∈𝑅𝑛+1+ :∣𝑥−𝑦∣< 𝑡}

and𝐵(𝑥, 𝑡) =𝐵 be a ball with center 𝑥∈𝑅𝑛. For𝑥∈𝑅𝑛, let

𝐴(𝑓)(𝑥) =𝑁(𝑓)(𝑥) = sup

(𝑦,𝑡)∈Γ(𝑥)

∣𝑓(𝑦, 𝑡)∣,

𝐴𝑞(𝑓)(𝑥) = (∫

Γ(𝑥)

∣𝑓(𝑦, 𝑡)∣𝑞 𝑡𝑛+1 𝑑𝑦𝑑𝑡

)1/𝑞

and

𝐶𝑞(𝑓)(𝑥) = (

sup

𝐵

1

∣𝐵∣

𝑇(𝐵)

∣𝑓(𝑦, 𝑡)∣𝑞 𝑡 𝑑𝑦𝑑𝑡

)1/𝑞

, where𝐵 contains𝑥and𝑇(𝐵) is the tent over𝐵 in𝑅𝑛 (see [3, 4]).

Define spaces𝑇𝑞𝑝,𝑇𝑝 and𝑇𝑞 respectively 𝑇𝑞𝑝={

𝑓 :𝑓 is measurable in𝑅𝑛+1+ satisfying∥𝑓∥𝑇𝑞𝑝=∥𝐴𝑞(𝑓)(𝑥)∥𝐿𝑝(𝑅𝑛)<∞.} , 𝑇𝑝 ={𝑓 :𝑓 is measurable in 𝑅𝑛+1+ with continuous boundary values on𝑅𝑛

such that∥𝑓∥𝑇𝑝 =∥𝐴(𝑓)(𝑥)∥𝐿𝑝(𝑅𝑛)<∞.}

2000Mathematics Subject Classification. 32A18, 32A37.

Key words and phrases. measurable function, analytic function, tent spaces, factorization.

c

⃝2010 Universiteti i Prishtin¨es, Prishtin¨e, Kosov¨e.

Submitted March, 2010. Published May, 2010.

50

(2)

and

𝑇𝑞={𝑓 :𝑓 is measurable in𝑅𝑛+1+ satisfying∥𝑓∥𝑇𝑞 =∥𝐶𝑞(𝑓)(𝑥)∥𝐿 <∞.}.

One of the main results of [3, 4] asserts that

𝑇𝑞𝑝=𝑇𝑝𝑇𝑞,for 0< 𝑝, 𝑞 <∞ (𝐴).

The mentioned equality was for the first time obtained in [3] for 𝑝 > 2, 𝑞 = 2.

Such type strong factorization theorems have numerous applications in the theory of analytic spaces ([2, 4, 6]). We give some results similar in spirit to (A). As we can easily notice mentioned factorization of 𝑇𝑞𝑝 classes were not considered before for𝑝=∞. In this note we, in particular, intend to give an answer to that natural question. On the other hand 𝑇𝑞𝑝 type classes that were defined above are heavily based on classical𝐿𝑝 spaces in 𝑅𝑛. Our next intention is to replace them by their natural extensions the well-known𝐿𝑝𝑞 Lorentz spaces in𝑅𝑛and to prove, if possible, a result similar to (A) equality.

2. Main Results

Let𝐶(𝑛)−1 be the volume of the unit ball ([4]) so that∥𝑃𝑡0𝐿1(𝑅𝑛)= 1, where 𝑃𝑡0(𝑥) =𝐶(𝑛)𝑡−𝑛𝜒𝐵(0,𝑡)(𝑥) and 𝜒𝐵(0,𝑡)(𝑥) is the characteristic function of the set 𝐵(0, 𝑡). For 𝑥∈𝑅𝑛, define

(𝑃0𝜇)(𝑥) =𝐶(𝑛)

Γ(𝑥)

𝑑𝜇(𝑦, 𝑡) 𝑡𝑛 where𝜇is a positive Borel measure in𝑅𝑛+1+ .

Lemma 2.1. Let𝑃0(𝑔)(𝑥, 𝑡) = 𝐶(𝑛)𝑡𝑛

𝐵(𝑥,𝑡)𝑔(𝑦)𝑑𝑦,𝑔∈𝐿1𝑙𝑜𝑐(𝑅𝑛),𝑆(𝜇) =𝑃0(𝑃0𝜇)−𝜏, where0< 𝜏 ≤1 and𝜇is a positive Borel measure on𝑅𝑛+1. Then

1

∣𝐵∣

𝑇(𝐵)

𝑆𝜇(𝑥, 𝑡)𝑑𝜇(𝑥, 𝑡)≤𝐶

(∫

𝑇(𝐵)∩Γ(𝑦)

𝑑𝜇(𝑥, 𝑡) 𝑡𝑛

)1−𝜏

𝐿(𝐵,𝑑𝑦) .

Remark 1: For𝜏 = 1, Lemma 2.1 was proved in [4].

Proof. Letℎ(𝑦) =𝑃0𝜇(𝑦),𝑦∈𝑅𝑛. Modifying proofs in [4] we have

𝑇(𝐵)

𝑆𝜇(𝑥, 𝑡)𝑑𝜇(𝑥, 𝑡)≤𝐶(𝑛)

𝑇(𝐵)

𝐵(𝑥,𝑡)

ℎ(𝑦)−𝜏𝑑𝜇(𝑥, 𝑡) 𝑡𝑛 𝑑𝑦

≤𝐶

𝑅𝑛

ℎ(𝑦)−𝜏

𝑇(𝐵)∩Γ(𝑦)

𝑑𝜇(𝑥, 𝑡) 𝑡𝑛 𝑑𝑦

≤𝐶∣𝐵∣ sup

𝑦∈𝑅𝑛

(∫

𝑇(𝐵)∩Γ(𝑦)

𝑑𝜇(𝑥, 𝑡) 𝑡𝑛

)1−𝜏

.

The proof is complete. □

Let𝑋, 𝑌 and𝑍 be quasinormed subspaces of a class of all measurable functions in𝑅𝑛. For 0< 𝛼≤1, we say𝑋 ⊂𝛼 𝑌 𝑍, if for any𝑢∈𝑋, there exist𝑤∈𝑌, 𝑣∈𝑍 such that𝑢=𝑤⋅𝑣𝛼.

(3)

Let𝑇𝑞∞,∞ be the class of measurable functions𝑓 satisfying

∥𝑓∥𝑇𝑞∞,∞ =

(∫

Γ(𝑦)

∣𝑓∣𝑞 𝑡𝑛+1𝑑𝑥𝑑𝑡

)1/𝑞 𝐿(𝑅𝑛)

<∞.

Theorem 2.2. Let0< 𝑞 <∞,𝑝 >0and0< 𝛼=𝑠/𝑞≤1. Then𝑇𝑞∞,∞𝛼 𝑇𝑝𝑇𝑞. Remark 2: If we replace 𝑇𝑞∞∞ classes in Theorem 2.2 with larger 𝑇𝑞𝑝 classes, then for s=q Theorem 2.2 is known (see [3, 4]).

Proof. We will modify the proof of [3, 4]. As the proof in [4] (Page 316), we have (∫

𝑋

∣𝑓∣−𝑠𝑑𝜈 )−1/𝑠

≤ (∫

𝑋

∣𝑓∣𝑟𝑑𝜈 )1/𝑟

𝑟, 𝑠 >0 (∗)

where𝜈 is a measure in𝑅𝑛. Let us put𝑑𝜈=𝑃𝑡0(𝑥)𝑑𝑥,𝑓 =𝐴𝑞(𝑢) in (∗). Then we have𝑉 = (𝑃0(𝐴𝑞(𝑢))𝑟)1/𝑟≥𝐶(𝑃0(𝐴𝑞(𝑢))−𝑠)−1/𝑠, that is,𝑉−𝑠≤𝐶(𝑃0(𝐴𝑞(𝑢))−𝑠).

Let𝑑𝜇(𝑥, 𝑡) =𝑢(𝑥, 𝑡)𝑞 𝑑𝑥𝑑𝑡𝑡 . Then (𝐴𝑞(𝑢))𝑞 =𝐶𝑃0𝜇and𝑉−𝑠≤𝐶𝑃0(𝑃0𝜇)−𝑠/𝑞= 𝑃0(𝑃0𝜇)−𝛼, where 0< 𝛼=𝑠/𝑞≤1. Since 𝜔𝑞 = 𝑉𝑢𝑞𝑠 from Lemma 2.1 we have

𝑇(𝐵)

𝜔(𝑥, 𝑡)𝑞𝑑𝑥𝑑𝑡 𝑡 ≤𝐶

𝑇(𝐵)

𝑉−𝑠𝑑𝜇(𝑥, 𝑡)

≤𝐶

𝑇(𝐵)

𝑆𝛼(𝜇)𝑑𝜇

≤𝐶∣𝐵∣

𝑇(𝐵)∩Γ(𝑦)

∣𝑢(𝑥, 𝑡)∣𝑞𝑑𝑥𝑑𝑡 𝑡𝑛+1

1−𝑠/𝑞

𝐿(𝐵,𝑑𝑦)

≤𝐶∣𝐵∣∥𝑢∥𝑞−𝑠𝑇∞,∞

𝑞 ,

which proves that for𝑢∈𝑇𝑞∞,∞ and𝜔𝑞 =𝑉𝑢𝑞𝑠, we have𝜔∈𝑇𝑞.

For 𝑢∈ 𝑇𝑞∞,∞, let 𝑉 = (𝑃0(𝐴𝑞(𝑢))𝑟)1/𝑟. Then (see [3, 4]) 𝑁 𝑃0(𝑓)≤ 𝐶𝑀(𝑓) and hence𝑁(𝑉)≤𝐶(𝑀(𝐴𝑞(𝑢))𝑟)1/𝑟, 𝑝 > 𝑟, where𝑀(𝑓) is the Hardy-littlewood maximal function. Thus 𝑉 ∈ 𝑇𝑝 for every 𝑝. Indeed 𝑀 is a bounded operator from 𝐿𝑝(𝑅𝑛) into 𝐿𝑝(𝑅𝑛), 𝑝 >1. Hence 𝑉 ∈ 𝑇𝑝 for every 𝑝 >0. One the other hand if𝜔= (𝑉𝑢𝑞𝑠)1/𝑞, then we can show that for𝑢∈𝑇𝑞∞,∞ and𝑉 ∈𝑇𝑝, 𝑝 >0

( 1

∣𝐵∣

𝑇(𝐵)

𝜔(𝑥, 𝑡)𝑞𝑑𝑥𝑑𝑡 𝑡

)1/𝑞

≤𝐶∥𝑢∥1−𝑠/𝑞𝑇∞,∞

𝑞 for𝑠≤𝑞.

The proof is complete. □

We now turn to another extension of (A). The following facts from the theory of Lorentz classes𝐿𝑝,𝑞(𝑅𝑛) are needed (see [1, 7]).

For𝑞, 𝑝∈(1,∞), the Hardy-Littlewood maximal operator is extended in𝐿𝑝,𝑞(𝑅𝑛) (see [5, 1, 7]) and we have

∥𝑀(𝑓)∥𝐿𝑝,𝑞 ≤𝐶∥𝑓∥𝐿𝑝,𝑞, (1) and

∥𝑀(𝑓)∥𝐿𝑝,∞ ≤𝐶∥𝑓∥𝐿𝑝,∞. (2)

(4)

For𝑓 be a measurable function in𝑅𝑛+1+ , define

∥𝑓∥𝐿𝑇𝑞𝑝,𝑠 =∥𝐴𝑞(𝑓)∥𝐿𝑝,𝑠(𝑅𝑛),

∥𝑓∥𝐿𝑇𝑝,𝑠 =∥𝑁(𝑓)∥𝐿𝑝,𝑠. For 0< 𝑝 <∞, the spaces𝐿𝑇𝑞𝑝,𝑠 and𝐿𝑇𝑝,𝑠 are defined by

𝐿𝑇𝑞𝑝,𝑠 ={𝑓 :𝑓 is measurable in𝑅𝑛+1+ satisfying∥𝑓∥𝐿𝑇𝑞𝑝,𝑠<∞}.

𝐿𝑇𝑝,𝑠={𝑓 :𝑓 is measurable in𝑅𝑛+1+ satisfying∥𝑓∥𝐿𝑇𝑝,𝑠<∞.}.

Theorem 2.3. Let 𝑠≤𝑝≤𝑞 <∞. Then 𝐿𝑇𝑞𝑝,𝑠=𝐿𝑇𝑝,𝑠𝑇𝑞.

Remark 3: For 𝑝= 𝑠, this was obtained in [3, 4] before and it coincides with (𝐴).

Proof. We again use same ideas from [3, 4]. Note first, if ∥𝐴𝑞(𝑓)∥𝐿𝑝,𝑠(𝑅𝑛) < ∞, putting𝑉 = (𝑃0(𝐴𝑞(𝑓))𝑟)1/𝑟as in the previous case we have𝑁 𝑉 ≤𝐶(𝑀(𝐴𝑞(𝑓)𝑟))1/𝑟, 𝑝, 𝑠 > 𝑟, where 𝑀 is the Maximal Hardy-Littlewood operator. By (1) we have

∥𝑁 𝑉∥𝐿𝑝,𝑠≤𝐶∥𝐴𝑞(𝑓)∥𝐿𝑝,𝑠(𝑅𝑛)<∞ for𝑝, 𝑠 >0 since

∥∣𝑓∣𝑟𝐿𝑝,𝑠 =∥𝑓∥𝐿𝑟𝑝,𝑟𝑠, 𝑝, 𝑠, 𝑟 >0.

The proof of the fact that 𝜔 = 𝑉𝑢 ∈ 𝑇𝑞 follows from the same arguments as in [4]. Let us show the reverse with the same restriction on parameters. Let𝜔∈𝑇𝑞, 𝑉 ∈𝐿𝑇𝑝,𝑠. We will show that

(∫

Γ(𝑦)

∣𝜔𝑉∣𝑞 𝑡𝑛+𝑠 𝑑𝑥𝑑𝑡

)1/𝑞 𝐿𝑝,𝑠(𝑅𝑛)

<∞.

By Holder inequality for Lorentz classes (see [5]), the following estimate holds.

𝐷=

(∫

Γ(𝑦)

∣𝜔(𝑥, 𝑡)𝑉(𝑥, 𝑡)∣𝑞 𝑡𝑛+𝑠 𝑑𝑥𝑑𝑡

)1/𝑞 𝐿𝑝,𝑠(𝑅𝑛)

≤𝐶∥𝑁 𝑉∥

𝐿

𝑝1𝜏 𝑞 ,𝑠1𝜏

𝑞

Γ(𝑦)

𝑉𝑞−𝜏𝜔𝑞 𝑡𝑛+𝑠

𝐿𝑝𝑞2,𝑠2

𝑞

=𝐴𝐵, where 𝑝1

1 + 𝑝1

2 = 1𝑝 and 𝑠1

1 + 𝑠1

2 = 1𝑠. Choosing 𝜏 such that 𝜏 𝑝𝑞1 = 𝑝, 𝜏 𝑠𝑞1 = 𝑠, then 𝑝𝑞2 = 𝑠𝑞2 = 1 and 𝐵 ≤ 𝐶∥𝜔∥𝑇𝑞∥𝑁 𝑉∥𝐿𝑞−𝜏 which follows (𝐴). Hence 𝐷 ≤

∥𝑁 𝑉∥𝐿𝑝,𝑠∥𝑁 𝑉∥𝐿𝑞−𝜏,𝑞−𝜏. Note that 𝜏 = 𝑞𝑠𝑠

1 = 𝑝𝑞𝑝

1, 𝑞 −𝜏 = 𝑞(1− 𝑝𝑝

1) = 𝑝 = 𝑞−𝑞𝑝(1𝑝1𝑞). Hence using known embeddings for Lorentz classes (see [1, 7]) we have𝐷≤ ∥𝑁 𝑉∥𝐿𝑝,𝑠(𝑅𝑛)for𝑠≤𝑝. The proof is complete.

□ Let𝔻={𝑧:∣𝑧∣<1}be the unit disk in the complex plane and T ={𝑧:∣𝑧∣= 1}.

Let Γ𝜎(𝜉) be the standard Luzin cone for the unit disk (see [4]), where𝜉∈T and 𝜎 > 1. Based partially on results we provided above we obtained following two assertions (embedding theorems).

(5)

Theorem 2.4. 1) Let 𝜇 be positive Borel measure on 𝔻,𝛽 > 0, 𝑞 >1, 𝑡 > 𝛽+ 1 and let𝑓 be analytic on 𝔻. Then

Γ𝜎(𝜉)

(1− ∣𝑧∣)𝑡∣𝑓(𝑧)∣𝑑𝜇(𝑧) 1− ∣𝑧∣

𝐿𝑞,∞(T)

≤𝐶 sup

∣𝑧∣<1

(∣𝑓(𝑧)∣(1− ∣𝑧∣)𝛽) if and only if

Γ𝜎(𝜉)

(1− ∣𝑧∣)𝑡−1 (∫

𝐷(𝑧,𝜌)

𝑑𝜇(𝑤) )

(1− ∣𝑧∣)−𝛽−2 𝐿𝑞,∞

<∞ where𝐷(𝑧, 𝜌)is Bergman metric ball, 𝜌 >0.

2) Let𝑓 be analytic function in𝔻,𝜇be positive Borel measure on𝔻. Then

Γ𝜏(𝜉)

∣𝑓(𝑧)∣

1− ∣𝑧∣𝑑𝜇(𝑧) 𝐿𝑞,∞(T)

≤𝐶

Γ𝜎(𝜉)

∣𝑓(𝑧)∣(1− ∣𝑧∣)𝛼−1𝑑𝑚2(𝑧) 𝐿𝑞,∞

if and only if

sup

𝑧∈𝔻

[ 𝜇(𝐷(𝑧, 𝜌)) (1− ∣𝑧∣)𝛼+2

]

<∞, 𝜌 >0, 𝑞 >1, 𝛼 >−1.

Acknowledgments. The authors would like to thank the referee for valuable sug- gestions.

References

[1] J. Bergh, J. Lofstrom, Interpolation spaces, Springer, 1976.

[2] W. Cohn, A factorization theorem for the derivative of a function in𝐻𝑝,Proc. Amer. Math.

Soc. 127(1999), 509–517.

[3] R. Coifman, I. Meyer, E. Stein, Some new tent spaces and their applications to harmonic analysis,J. of Funct. Anal.62(1985), 304–335.

[4] W. Cohn, I. Verbitsky, Factorizations of tent spaces and Hankel operators,,J. Funct. Anal.

175(2000), 308–339.

[5] L. Grafakos, Classical and modern fourier analysis, Pearson Education, Inc., Upper Saddle River, NJ, 2004.

[6] C. Horowitz, Factorization theorems for functions in the Bergman spaces,Duke Math. J.44 (1977), 201–213.

[7] E. Stein and G. Weiss, Harmonic analysis in Euclidean spaces, Princeton, 1974.

Romi Shamoyan

Department of Mathematics, Bryansk State University, Russian Federation of Nations.

E-mail address:[email protected]

Wen Xu

Department of Physics and Mathematics, University of Joensuu, P. O. Box 111, FIN- 80101 Joensuu, Finland.

E-mail address:[email protected]

参照

関連したドキュメント