• No results found

Analytic Classes on Subframe and Expanded Disk and the Differential Operator in Polydisk

N/A
N/A
Protected

Academic year: 2020

Share "Analytic Classes on Subframe and Expanded Disk and the Differential Operator in Polydisk"

Copied!
22
0
0

Loading.... (view fulltext now)

Full text

(1)

Journal of Inequalities and Applications Volume 2009, Article ID 353801,22pages doi:10.1155/2009/353801

Research Article

Analytic Classes on Subframe and Expanded Disk

and the

R

s

Differential Operator in Polydisk

Romi F. Shamoyan

1

and Olivera R. Mihi ´c

2

1Department of Mathematics, Bryansk University, 241036 Bryansk, Russia

2Faculty of Organizational Sciences, University of Belgrade, Jove Ili´ca 154, 11000 Belgrade, Serbia

Correspondence should be addressed to Olivera R. Mihi´c,[email protected]

Received 25 November 2008; Revised 17 September 2009; Accepted 28 September 2009

Recommended by Narendra Kumar Govil

We introduce and study new analytic classes on subframe and expanded disk and give complete description of their traces on the unit disk. Sharp embedding theorems and various new estimates concerning differentialRsoperator in polydisk also will be presented. Practically all our results

were known or obvious in the unit disk.

Copyrightq2009 R. F. Shamoyan and O. R. Mihi´c. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

1. Introduction and Main Definitions

Letn ∈ Nand Cn {z z

1, . . . , zn : zk ∈ C, 1 ≤ kn}be then-dimensional space of

complex coordinates. We denote the unit polydisk by

Un{z∈Cn:|zk|<1, 1≤kn}, 1.1

and the distinguished boundary ofUnby

Tn{z∈Cn:|zk|1, 1≤kn}. 1.2

We usem2nto denote the volume measure onUnandmnto denote the normalized Lebesgue

measure onTn.LetHUnbe the space of all holomorphic functions onUn.Whenn1,we

simply denoteU1byU, T1byT, m

2nbym2, mnbym.We refer to1,2for further details. We

denote the expanded disk by

Un z r1ξ, . . . , rnξUn:ξT, rj∈0,1, j 1, . . . , n

(2)

and the subframe by

Un zUn :z

jr, r∈0,1, j 1, . . . , n

. 1.4

The Hardy spaces, denoted byHpUn0< p≤ ∞,are defined by

HpUn

fHUn: sup

0≤r<1 Mp

f, r<, 1.5

whereMppf, r

Tn|frξ1, . . . , rξn|

pdm

nξ, Mf, r maxξ∈Tn|frξ1, . . . , rξn|, r ∈0,1,

fHUn.

Forαj >−1, j 1, . . . , n0 < p <,recall that the weighted Bergman spaceApαUn

consists of all holomorphic functions on the polydisk satisfying the condition

fp

Apα

Un

fz1, . . . , znp n

i1

1− |zi|2αidm2nz<. 1.6

Forαj>−1, j1, . . . , n, p∈0,,the Bergman class on expanded disk is defined by

ApαUn

⎧ ⎨

fHUn:f

p ApαUn

1

0

· · ·

1

0

T

f|z1|ξ, . . . ,|zn|ξpn j1

1−zj2

αj

dzjdmξ<

⎫ ⎬ ⎭,

1.7

and similarly the Bergman class on subframe denoted byApαUnis defined by

Apα

Un

fHUn:fp

ApαUn

Tn 1

0

f|z|ξ1, . . . ,|z|ξnp

1− |z|2αdmnξd|z|<, 1.8

wherep∈0,, α >−1.

Throughout the paper, we write C sometimes with indexes to denote a positive

constant which might be different at each occurrence even in a chain of inequalities but

is independent of the functions or variables being discussed.

The notationA Bmeans that there is a positive constantCsuch thatB/CA

CB.We will write for two expressionsABif there is a positive constantCsuch thatA < CB.

This paper is organized as follows. In first section we collect preliminary assertions. In the second section we present several new results connected with so-called operator of

diagonal map in polydisk. Namely, we define two new maps Sbf and Edf from

subframe U and expanded disk Un

∗ to unit disk Uand, in particular, completely describe

(3)

expanded disk on usual unit diskU on the complex plane. Proofs are based among other things on new projection theorems for these classes.

A separate section will be devoted to the study ofRsdierential operator in polydisk.

It is based in particular on results from the recent paper 3. We will use the dyadic

decomposition technique to explore connections between analytic classes on subframe, polydisk, and expanded disk. We also prove new sharp embedding theorems for classes on subframe and expanded disk. Last assertions of the final section generalize some

one-dimensional known results to polydisk and to the case ofRsoperators simultaneously.

2. Preliminaries

We need the following assertions.

Lemma Asee4. Letαbe a fixedn-tuple of nonnegative numbers and let{Bi}i∈Zbe an arbitrary family of α-boxes in Rn lying in the cubeQn 2π,2πn. There exists a setJ I such that

BiBi, i, iJand for alliIthere existsjJsuch thatBi⊂5Bj.

The following proposition is heavily based on ideas from4.

Proposition 2.1. a Let f be a nonnegative summable function on Tn T × · · · × T. Let

fx1, . . . , xn feix1, . . . , eixn. Let αj ∈ N, rj ∈ 0,1, j 1, . . . , n, x ∈ Rn, gr, x

supB∈Z1/mnB

Bfdmn,Z ZZα,where Z {B Bx, y : yj ≥ 1−rj, 1 ≤ jn},

B→−ϕ,→−y {t∈Rn :|t

iϕi| ≤yi, 1 ≤in},and whereZαis a family of allB→−ϕ,→−yboxes such

thaty1, . . . , ynis proportional to2−α1, . . . ,2−αn, αi ∈N, i1, . . . , nfor some fixedα.Then the

following statements hold:

μEαa μ

zreixU:g→−αr, x> aC a

Tn

fϕdmn

ϕ,

μ1E1αaμ1zr1eix, . . . , rneix

Un:g→−αr1, . . . , rn, x> a

C1 a

Tnf

ϕdmn

ϕ,

2.1

whereμandμ1are any positive Borel measures onUandUn

such that

μΔU μΔCln, l∈0,1, ξT,

μUn∗≤Cl1· · ·ln, lj∈0,1, ξTn,

2.2

where Δ {zUn : 1−lj ≤ |zj| < 1,|argzj−argξj| ≤ lj/2, j 1, . . . , n}.b Let

μz 1− |z|n−2 dm2z, μ1z nk11− |zk|−1/n d|z1| · · ·d|zn|dmξ, n >1.Then

μEαa

C a

Tn

fϕdmn

ϕ, μ1

E1

αa

C

a

Tn

fϕdmn

ϕ. 2.3

Proof. Proofs of all parts are similar. The first part of the lemma connected withμmeasure

and unit disk can be found in4. We will give the complete proof of the second part. To

prove the second part letE0

(4)

Fix a pointzin theE0

αaand associate anα- boxBzsuch that

1

|Bz|

Bz

fdmn> a, BzBz1× · · · ×Bzn, |Bz||Bz1| · · · |Bzn|, Bzj

≥1−rj. 2.4

We use standard covering Lemma A to construct a setJE0αasuch thatα- boxesBz, zJ

pairwise disjoint, we have

z∈J

|Bz|≤ 1 a

z∈J

Bz

fξdmnξ

Tn

fdmn. 2.5

So the lemma will be proved ifμ1E1αaC

z∈J|Bz|.Let

Keiϕ, lξU:ξeiϕ< l, l >0, E1αa z→−r eixUn:g→−α→−r , x> a.

2.6

Let wE0

αa, then we will show w

z∈JKeiϕ1, lz1× · · · × Ke

iϕn, l

zn, ϕj argzj. So E0

αa

z∈JKeiϕ1, lz1× · · · ×Keiϕn, lzn

z∈JKz1,where lzj C|Bzj|, j 1, . . . , n,

and constant Cwill be specified later. Hence we will have μ1E0

αaUn μ1Eα1a

Cz∈Jμ1Kz1Un∗≤Cz∈J|Bz|.

The last estimate follows from inclusionKz

1 ∩Un∗ ⊂ΔlzξUn,

ΔlzξU

n

zUn :zj

1−lzj,1

, argξj

−argzlzj

2 , j1, . . . , n . 2.7

It remains to show the inclusionE0

αa

z∈JKz1.To show this inclusion we note if ξ ξ1, . . . , ξnE0αathen we have|Bξi| ≥1− |ξi|, i1, . . . , n.Using covering Lemma A we haveBξiCB zi, i1, . . . , n, z z1, . . . , znJ.Hence

ξieiϕiξieiargξi eiargξieiϕi≤1− |ξi| argξiϕi≤2C|Bzi|. 2.8

It remains to note that we put abovelzi |Bzi|2C.

bNote that for the case ofUn

∗ we can step by step repeat the same procedure with

E1

αainstead ofE0αaand the condition onμ1will be replaced by weaker condition

μUn∗≤Cl1· · ·ln, lj∈0,1, ξT. 2.9

Note forξT

ΔUn

zUn

∗ : 1−ljzj<1,arg

zj

−argξ≤ minlj

2 , j1, . . . , n . 2.10

(5)

Remark 2.2. InProposition 2.1bnk11− |zk|−1/n can be replaced bynk11− |zk|−tk, t

j∈ 0,1, nj1tj1.

Lemma 2.3. LetzjU, j1, . . . , n, 0< p≤1, s >1/p−2, fHUand Fz1, . . . , zn CUfz1− |z|sdm2z/nj11−zzjs 2/n.Then

|Fz1, . . . , zn|pC

U

fwp1− |w|sp 2p−2dm2w

n

k1|1−zkw|

s 2/np . 2.11

Estimate2.11forn1 can be found in5and in1for general case. The following

lemma is well known.

Lemma 2.4. Let s > 0.Then forIs, Isstβ−α−2dt/t 11 β one hasa Issβ−α−1, β <1 α, s → 0;bIs∼ln 1/s,1 αβ, s → 0;cIsC, 1 α < β, s → 0.

Lemma 2.5see3. Letw|w|ξ, w, zUn,1wzn

k11−wkzk, s∈N∪ {0}, β >0, p∈ 0,.Then one has

Tn Rs

1

1−ξ|w|

p

dmnξC

αj≥0,αjs n

k1

1

1− |wk||zk|pαk β−1

, p > 1

mink αk β. 2.12

Corollary 2.6. Let0< p <, s∈N∪ {0}, l∈0,, γ >1/p l, wUn.Then

Un

Rs 1 1−wzγ

p1− |z|pl−1dm2nz

αj≥0,αjs

n

k1

C

1− |wk|αk γp−pl−1. 2.13

We will need the following Theorems A and B.

Theorem Asee3. aLetfHUn, p0,, α

j>−1, j1, . . . , n, n∈N.Then

U

fz, . . . , zp1− |z|2

n j1αj n−1

dm2z

C

T

0,1n

f|z1|ξ, . . . ,|zn|ξp

n

k1

1− |zk|2αkd|z1| · · ·d|zn|dmξ.

2.14

bLetfHUn, p0,, α >1, nN.Then

U

fz, . . . , zp1− |z|2α n−1dm2zC

Tn 1

0

f|z|ξ1, . . . ,|z|ξnp

(6)

Theorem Bsee3. aLetfHUn, p0,, α

j>−1, j1, . . . , n, n∈N.Then

T

0,1n

f|z1|ξ, . . . ,|zn|ξpn k1

1− |zk|αk n−1/n d|z1| · · ·d|zn|dmξ

C

Un

fz1, . . . , znp n

k1

1− |zk|2αkdm2nz.

2.16

bLetfHUn, p0,, α >1, nN.Then

1

0

Tn

f|z|ξ1, . . . ,|z|ξnp

1− |z|2αdmnξd|z|

C

Un

fz1, . . . , znp

1− |z1|2

α/n−1 1/n

· · ·1− |zn|2α/n−1 1/ndm2nz.

2.17

3. Analytic Classes on Subframe and Expanded Disk

Let us remind the main definition.

Definition 3.1. LetX ⊂HU,Y ⊂HUnbe subspaces ofHUandHUn.We say that the

diagonal ofYcoincides withXif for any functionf, f ∈ Y, fz, . . . , z∈ X, and the reverse is also true for every functiongfromXthere exists an expansionfz1, . . . , zn∈ Y,such that

fz, . . . , z gz.Then we write DiagY X.

Note when DiagY X,then

fXinf

Φ ΦfY, 3.1

whereΦfis an arbitrary analytic expansion offfrom diagonal of polydisk to polydisk.

The problem of study of diagonal map and its applications for the first time was also

suggested by Rudin in 2. Later several papers appeared where complete solutions were

given for classical holomorphic spaces such as Hardy, Bergman classes; see1,4,6,7and

references there. Recently the complete answer was given for so-called mixed norm spaces in

8. Partially the goal of this paper is to add some new results in this direction. Theorems on

diagonal map have numerous applications in the theory of holomorphic functionssee, e.g.,

9,10.

In this section we concentrate on the study of two maps closely connected with

diagonal mapping Sb from subframe into disk Sb : fz1, . . . , znfz, . . . , z where all

|zj| |z| ∈ 0,1and another map Ed : fz1, . . . , znfz, . . . , z from expanded disk

into disk wherez1 |z1|ξ, . . . , zn |zn|ξ, ξT,andf function is from a functional class on

subframeUnor expanded diskUn

.

Note that the study of maps which are close to diagonal mapping was suggested by

(7)

Theorem 3.2. Let0< p <.If a0< p≤1 and β >−1,or

b p > 1 and β > maxn/p− 1,n−1p−1−1, then for every function gApβUn, g|z|ξ, . . . ,|z|ξ Ap

β n−1Uand for every functionfA

p

β n−1Uthere exist gApβUnsuch thatg|z|ξ, . . . ,|z|ξ fz.

Proof. Note that one part of the theorem was proved in3and follows from Theorem A. Let

us show the reverse. Let firstp≤1.Consider the followinggfunction:

grξ1, . . . , rξn

U

fw1− |w|αdm2w

n

k11−rξkwα 2/n

, z rξ1, . . . , rξn, 3.2

whereα >0 can be large enough,is a constant of Bergman from representation formula

see1, obviouslygrξ, . . . , rξ frξby Bergman representation formula in the unit disk.

It remains to note that the following estimate hold byLemma 2.3:

Tn 1

0

g|z|ξ1, . . . ,|z|ξnp1− |z|β|z|d|z|dmnξ

C

1

0

Tn

U

fwp

1− |w| 2p−21−rβdrdm2wdmnξ |1−1w|α 2p/n· · · |1−rξnw|α 2p/n

,

3.3

whereαcan be large enough. Using Fubini’s theorem and calculating the inner integral we

get what we need.

We consider now 1< p <∞case. LetfApβ n−1U, p >1.Then

fz

U

fw1− |w|α

1−wzα 2 dm2w, zU, 3.4

by Bergman representation formula. ObviouslySbg f, as

grξ1, . . . , rξn

U

fw1− |w|α

1−1 2/n· · ·1−rξnwα 2/n

dm2w, 3.5

whereSbg grξ, . . . , rξ, rξUis a map from subframeU {zUn : |zj| |z|}to

diagonal. Using duality arguments and Fubini’s theorem we have

g

ApβUn

U

1− |w|αfw

1

0

Tn

hrξ1, . . . , rξn1−rβdr dmnξ

n

k11−rξkwα 2/n

dm2w,

3.6

(8)

We will need the following assertion. LethLpβUn, β1,, p1,.

Then let

gw1, . . . , wn

1

0

Tn

hrξ1, . . . , rξn1− |z|βdr dmnξ

n

j1

1−rξjwj

β 1 n/n , 3.7

wherewjU, wj |w|ϕj, j 1, . . . , n.We assert thatgbelongs toAp

βUn.

Indeed using H ¨older’s inequality we get

1

0

Tn

gw1, . . . , wnp

1− |w|βdmn

ϕd|w|

C

1

0

1

0

Tn

Tn

|hrξ1, . . . , rξn|p

1−rβp1− |w|β p/pnεp 1

n

k1|1−rξkwk|β n 1/n−2−εp

2

×dr dmnξdmn

ϕd|w| ≤ChLp βUn

, 1< p<,

3.8

whereε >0.We used above the estimate

Tn 1

0

1− |w|β p/pnεp 1dmn

ϕd|w|

n

k1|1−rξkwk|β 1/n−ε−1p

2C1−rβ−βp

, 3.9

r∈0,1, β >n−1p/p−1, β >n/p−1.

Returning to estimate forgAp

βwe have by H ¨older’s inequality

g

Apβ

UnC

U

fwp

1− |w|β n−1dm2w

!1/p

× U

w, . . . , w|p1− |w|β n−1dm2w

!1/p

CfAp β n−1.

3.10

We used the fact that for allΦ∈Apβ,1< p<, SbΦ∈Apβn−1proved before.

The complete analogue ofTheorem 3.2is true for Bergman classes on expanded disk

we defined previously.

Theorem 3.3. Let0≤p <.If

ap∈0,1,→−β β1, . . . , βn, βk>−1, k1, . . . , n, |β|nk1βk,or

bp > 1,→−β β, . . . , β, β > 0, β > 1/n−1p/p−1,then the following assertion holds. For every functionf, fAp

βU n

, Edfz fz, . . . , z, zUbelongs to

Ap|β| n−1and the reverse is also true, for any functionffromAp|β| n−1there exists a function

gApβU

n

(9)

Proof. We give a short sketch of proof ofTheorem 3.3and omit details. Note that the half of the theorem the inclusion EdAp

βU n

∗⊂ Ap|β| n−1 was proved in3and follows directly from

TheoremA.

Forp ≤ 1 we have to use againLemma 2.3and Fubini’s theorem. Forp > 1 we first

provegw1, . . . , wnApβU

n

∗ifhApβU

n

∗and if|β|p/n > supkβk > minkβk > 1/n

1p/p−1

gw1, . . . , wn C

T

1

0

· · ·

1

0

hz1, . . . , znnk11− |zk||β|/nd|zk|dξ

n

k1|1−zkwk||β| 1 n/n

. 3.11

Indeed by H ¨older’s inequality we have

gw1, . . . , wnp

C

⎛ ⎝

T

1

0

· · ·

1

0

|hz1, . . . , zn|p

n

k11− |zk||β|p/nd|zk|dξ

n

k1|1−zkwk||β| 1 n/np−2 εp 2

⎞ ⎠

×

T

1

0

· · ·

1

0

d|zk|dξ

n

k1|1−zkwk|2 εp

p/p

Ah

T

n

k11−ξwkϕ1 εp

p/p

Ah

n

k1

1

1− |wk|1 εp−1/np/p

, where ε >0.

3.12

Hence calculating integrals we finally havegAp

− →

βU n

ChApβU

n.

We used the estimate

T

1

0

· · ·

1

0

n

k11− |wk|βk 1/n−1 εp

p/p

d|wk|dξ

n

k1|1−zkwk||β| 1 n/np−2 εp 2

C

n

k1

1− |zk|−|β|p/n βk, 3.13

which is true under the conditions on indexes we have in formulation of theorem and can

be obtained by using H ¨older’s inequality for n functions. Using this projection theorem

and repeating arguments of proof of the previous theorem we will complete the proof of

(10)

Remark 3.4. Note that Theorems3.2and3.3are obvious forn1.

Remark 3.5. Note that estimates between expanded disk, unit disk, and polydisk can be also obtained directly from Liuville’s formula

1

0

· · ·

1

0

ϕv1· · ·vn n

k1

1−vkpk−1vp21v

p1 p2

3 · · ·v

p1 ···pn−1

n dv1· · ·dvn

n

k

pk

Γp1· · ·pn

1

0

ϕu1−uni1pi−1du, p

i>0, ϕis continuous function on0,1. 3.14

Remark 3.6. The complete description of traces of classes with fA

αUn∗ and fAαUn quasinorms on the unit disk can be obtained similarly by small modification of the proof ofTheorem 3.2,

fp

AαUn 1

0

M· · ·Mf, rp1−rαdr, α >−1, p∈0,,

fp AαUn

1

0

1

0

Mf, r1, . . . , rn

p

1−r1α· · ·1−rnαdr1· · ·drn, α >−1, p∈0,∞ 3.15

Let

ΛαUn

fHUn: sup rj∈0,1,ξ∈T

fr1ξ, . . . , rnξ n

k1

1−rkα<, α >0,

Λα

Un

fHUn sup r∈0,1,ξi∈T

frξ1, . . . , rξn1−rα<, α >0.

3.16

We formulate complete analogues of Theorems3.2and3.3for classesΛαUn andΛαUn.

Theorem 3.7. Let α > 0 and f ∈ ΛαUn. ThenSbf fz, . . . , z is in ΛαU and any

g ∈ΛαUcan be expanded tof, f ∈ΛαUnsuch thatfz, . . . , z gz.The same statement is

true for pairsEd,ΛαUn,EdΛαUn∗ ΛnαU.

Note that one part of statement is obvious. If, for example,f ∈ ΛαUn,then Edf

ΛnαU.On the other side, letg ∈ΛnαU.Then define as above that

fr1ξ, . . . , rnξ

U

gz1− |z|β

n

k11−zrkξβ 2/n

dm2z, 3.17

βis big enough,is a Bergman constant of Bergman representation formula.

Obviously frξ, . . . , rξ grξ, z rξ, zU. Using H ¨older’s inequality for n

functions we getfΛ

(11)

It is natural to question about discrete analogues of operators we considered previously.

LetCk>0, rk1, k → ∞, β≥0.Let

Bn,Ck,rk,β ⎧ ⎨

fHUn∗:fz1, . . . , zn

k1 Ck

n

j1

1

rkzj

1 β/n <,

zj|z| ∈0,1, zjU, j1, . . . , n⎫⎬ ⎭.

3.18

We have for such a function

1

0

1−r1 αMf, rdr

1

0

1−r1 α sup

|z1|r,...,|zn|r

fzdr

k1 Ck

1

0

1−r1 α

rkr1 β

dr

k1

Ckrk−1α−β 1

rk−1

tβ−α−2

t 11 βdt.

3.19

As a consequence of these arguments and using Lemma 2.4 we have the following

proposition, a discrete copy of assertions we proved above.

Proposition 3.8. LetfBn,Ck,rk,β andα > −1.Then 1

01−r 1 αM

f, rdr <if and only

ifk1Ck <if 1 α > β;∞k1Ckln 1/rk −1 <if1 α β;∞k1Ck/rk−1β 1 <if1 α < β.

4. Sharp Embeddings for Analytic Spaces in Polydisk with

R

s

Operators and Inequalities Connecting Classes on

Polydisk, Subframe, and Expanded Disk

The goal of this section is to present various generalizations of well-known one-dimensional results providing at the same time new connections between standard classes of analytic

functions with quazinorms on polydisk and Rs dierential operator with corresponding

classes on subframe and expanded disk.

In this section we also study another two maps connected with the diagonal mapping from polydisk to subframe and expanded disk using, in particular, estimates for maximal

functions fromLemma 2.3which are of independent interest. Note that for the first time the

study of such mappings which are close to diagonal mapping was suggested by Rudin in2.

Later Clark studied such a map in11.

In this section we also introduce the Rs dierential operator as follows see

3, 12, 13. Rsf

k1,...,kn≥0k1 · · · kn 1

sa k1,...,knz

k1

1 · · ·z

kn

n, s ∈ R, where fz

k1,...,kn≥0ak1,...,knz

k1

1 · · ·z

kn

nHUn.

(12)

In the case of the unit ball an analogue ofRsoperator is a well-known radial derivative

which is well studied. We note that in polydisk the following fractional derivative is well

studiedsee1:

Dαfz k1,...,kn≥0

k1 1α· · ·kn 1αak1,...,knz

k1

1 · · ·z

kn

n , 4.1

whereα∈R, fHUn, andD:HUn HUn.Apparently theRsoperator was studied

in 12for the first time. Then in 13, the second author studied some properties of this

operator. In this section we also continue to study theRs operator. We need the following

simple but vital formula which can be checked by easy calculation:

fτξ1, . . . , τξn Cs

1

0

Rsfτξ

1ρ, . . . , τξnρ log

1

ρ

!s−1

dρ, 4.2

where s > 0, τ ∈ 0,1, Cs > 0, ξjT, j 1, . . . , n.This simple integral representation of

holomorphic fz, zUn functions in polydisk will allow us to consider them in close

connection with functional spaces on subframeUn.

The following dyadic decomposition of subframe and polydisk was introduced in1

and will be also used by us:

Uk,l1,...,ln Uk,l1× · · · ×Uk,ln

&

τξ1, . . . , τξn:τ∈ 1−

1 2k,1−

1 2k 1

'

,

k0,1,2, . . .; πlj 2k < ξj

πlj 1

2k , lj−2

k, . . . ,2k1, j 1, . . . , n ,

mIk,lj

m

ξT : πlj 2k < ξj

πlj 1

2k

2−k, m2n

Uk,l1,...,ln

2−2kn,

Uk1,...,kn,l1,...,ln Uk1,l1× · · · ×Ukn,ln

&

τ1ξ1, . . . , τnξn:τj∈ 1− 1

2kj,1− 1 2kj 1

'

,

kj0,1,2, . . .;

πlj

2kj < ξj

πlj 1

2kj , lj−2

kj, . . . ,2kj1, j 1, . . . , n ,

4.3

Mpf, r, 0 < p ≤ ∞ averages in analytic spaces in polydisk can obviously have a mixed

form, for example,MpMf, r1, r2, rj∈0,1, j1,2, p <.In8Ren and Shi described

the diagonal of mixed norm spaces, but the above mentioned mixed case was omitted there.

Our approach is also different. It is based on dyadic decomposition we introduced previously.

Theorem 4.1. Let p ∈ 0,and Hα,np,∞ {fHUn :

1

0· · ·

1

0

TM∞· · ·Mf,

(13)

Proof. Using diadic decomposition of polydisk we have

M

U

fz, . . . , zp1− |z|αdm2z

j,k

Uj,k

fz, . . . , zp1− |z|αdm2z

C

j,k

max

Uj,k

fz, . . . , zp

2−2k2−αk

C

k1≥0

· · · kn≥0

2minkj−1

j−2minkj

max

Uk1,...,kn,j

fz1, . . . , znp

2−α 2k1/n· · ·2α 2kn/n

k1≥0

· · · kn≥0

2minkj−1

j−2minkj

1−2−k1 2

1−2−k1−1

· · ·

12−kn 2

1−2−kn−1

×

Ij,k1,...,kn

fz1, . . . , znp

dr1· · ·drndξ

2k1τ· · ·2knτ

2minkj n

, τα 2 n 1.

4.4

We used above the following estimate which can be found, for example, in1

sup

z∈Uk1,...,kn,j

fz1, . . . , znp

2minkj

n

2k1 ···kn

Uk

1,...,kn,j

fz1, . . . , znpdm2nz, 0< p <,

4.5

whereUk

1,...,kn,jare enlarged dyadic cubessee1and 0< p <, fHU

n, and

Ij,k1,...,kn

ξ:zrξUk1,...,kn,j,

πj

2minkjξ <

πj 1 2minkj ,

Ij,k

1,...,kn

ξ:zrξUk

1,...,kn,j,

πj−1/2 2minkjξ <

πj 1/2

2minkj .

4.6

Note further sincemnIj,k1,...,kn 2−n

minkj,

Ij,k1,...,kn

fz1, . . . , znp

dξC

I1 j,k1,...,kn

M∞· · ·Mf, rpdξ

2−n−1minkj

,

Ij,k

1,...,kn I

1

j,k1,...,kn × · · · ×I

n

j,k1,...,kn, 0< p <.

4.7

We have

MC

k1≥0

· · · kn≥0

2minkj−1

j−2minkj

12−k1 2

1−2−k1−1

· · ·

12−kn 2

1−2−kn−1

I1 j,k1,...,kn

M∞· · ·Mf, rpdξ2−k1α 1/n−1· · ·2−k 1/n−1/n.

4.8

(14)

Hence using the fact thatIj,k1

1,...,kn is a finite covering ofT, finally we have for all 0 <

p <MCfHp,

α,n.One part of theorem is proved.

To get the reverse statement we use the estimate fromLemma 2.3. Then we have

|Fz1, . . . , zn|C

U

fz1− |z|αdm2z

n j1

1−zzj

α 2/np

, zjU, 4.9

for anyα >−1.HenceFz, . . . , z fzand forp <1 byLemma 2.3

FHp,α,n

1

0

1

0

T

M∞· · ·M

U

fzp1− |z|αp 2p−2dm2z

n

j11−zzjα 2/np

×n k1

1−rkα 1/n−1dr1· · ·drndξ1· · ·dξn

U

fzp1− |z|αp 2p−2

1

0

1

0

n

k11−rkα 1/n−1dr1· · ·drn

n−1

k11− |z|rkα 2/np1−rn|z|α 2/np−1

dm2z

U

fzp1− |z|αp 2p−21− |z|−α 2/np α 1/nn−1

×1− |z|−α 2/np α 1/n 1dm2z

U

fzp1− |z|αdm2z, α >−1, 0< p≤1, fApα.

4.10

Let 1 < p <.Then we may assume that againsis large enough. Let 1/p 1/q

1, γi>0, i1,2, γ1 γ2 s 1/n.Then

⎛ ⎜ ⎝

U

fw 1− |w|2s−1dm2w

n

j11−zjws 1/n

⎞ ⎟ ⎠

p

C

U

fwp

1− |w|2s−1dm2w

n

j11−zjwγ1p

n

j1

1−zjγ1p−s 1/n.

4.11

We used estimate

⎛ ⎜ ⎝

U

1− |w|2s−1dm2w

n

j11−zjwγ2q

⎞ ⎟ ⎠

p/q

C

n

j1

(15)

Choosing appropriateγ1, γ2we repeat now arguments that we presented forp≤1 above to get what we need. The proof is complete.

Remark 4.2. The case ofk, 1 < kn, Mp averages can be considered similarly. Note in Theorem 4.1k1.Thus ourTheorem 4.1extends known description of diagonal of classical

Bergman classessee1,7.

In14Carleson as Rudin and Clark showed that in the case of the polydisk one cannot

expect so simple description of Carleson measures as one has for measures defined in the disk. We would like to study embeddings of the type

Un

fzpdμz

!1/p

CRsfX, s≥0, 4.13

whereμis a positive Borel measure onUnandXis a Bergman class on polydisk or subframe.

Theorem 4.3. Letγ1 β1 s−1p n 1p−1, γ1 > 0, β1 > −1, 0 < p ≤1, s > β1

1/n npn/p, s, n∈N, andμis a positive Borel measure onUn.If

Un

fzp1− |z|tdμzRsf Apβ

1U

n, t >−1, 4.14

then

μ

Un sup

w∈Un

Tn 1

0

1− |w|tdμw

n

k1|1−wkwk|p

1− |w|γ1 <, 4.15

and conversely if

μ

Un sup

w∈Un

Tn 1

0

1− |w|tdμw

n

k1|1−wkwk|p ε

1− |w|γ1 ε<4.16

holds for someε >0wherew |w|ξ1, . . . ,|w|ξn, |w| ∈0,1, ξjT, j1, . . . , nthen

Un

fzp1− |z|tdμzRsfAp β1U

n, t >−1. 4.17

Remark 4.4. With another “ε-sharp” embedding theorem, the complete analogue of

Theorem 4.3is true also when we replace the left side byUn|fz|p n

k11− |zk|tkdμz, tk> −1, k1, . . . , n.The proof needs small modification of arguments we present in the proof of

Theorem 4.3.

(16)

Proof ofTheorem 4.3. It can be checked by direct calculations based on formula4.2that the following integral representation holds:

fz1, . . . , zn Cs

Tn 1

0

Rsfρξ1, . . . , ρξ n log

1

ρ

!s−1 dρdm

n

k1

1−ρϕkrξk

, 4.18

s > 0, zk ϕkr, k 1, . . . , n, r ∈ 0,1,z1, . . . , znUn.Using the fact thatMpf, r in

increasing byr, r ∈ 0,1andM1f, τ2 C1τn1−1/pM

pf, τ, τ ∈ 0,1 see1we

get from above by using the estimate

1

0

1−ραdρ

p

1

0

fρp1−ραp p−1dρ, 4.19

wherefis growing measurable,p≤1, α >−1,

Un

fwp1− |w|tdμwUn 1

0

Tn

Rsfwp

1− |w|β1dm

nξd|w|. 4.20

To obtain the reverse implication we use standard test functiongzw 1/nk11−wkzkγ,

zk, wkU, k1, . . . , n, z, wUn, γ >1 andLemma 2.5

Rsg

zwAp β1U

n 1

0

1− |w|β1

αj≥0,αjs

Gz, wd|w|, 4.21

whereGz, w nk11/1− |wk||zk|pαk γ−1, γ >1, p >1/γ. The rest is clear. The proof is complete.

Below we continue to study connections between standard classes in polydisk and corresponding spaces on subframe and expanded disk.

Let

RAps,v

fHUn:RsfApν

Un,

DApγ ,α

fHUn:DγfApαUn

, DγDγ1

z1· · · D

γn

zn,

n

j1

γjγ, γj≥0.

4.22

Let us note that Theorems3.2and3.7of3show that under some restrictions onγ, α, s, ν

the following assertion is true.

For every function f, f ∈ DApγ,α, 0 < p <, f|z|ξ1, . . . ,|z|ξn belongs to RAps,v, 0< p <,and the reverse is also true, for any functionf, f ∈RAps,vthere exists

“an extension”Fsuch that

(17)

The followingTheorem 4.6gives an answer for the same map from polydisk to expanded disk

fz1, . . . , zn−→f|z1|ξ, . . . ,|zn|ξ, ξT, zj∈0,1, j1, . . . , n, 4.24

forffunctions fromHpUn, 1< p <.

We will develop ideas from 4 to get the following sharp embedding theorem for

classes on expanded disk.

Theorem 4.6. Letn >1, 1< p <, μbe a positive Borel measure onUn.Then

1

0

· · ·

1

0

T

fr1ξ, . . . , rnξpdμ

r ξfHpUn 4.25

if and only if11−l 1· · ·

1

1−ln

minjlj/2

−minjlj/2

r ξCl1· · ·l

n, lj>0, j1, . . . , n.

Proof. Obviously ifKUn ∗|fz|

pdμzCfp

Hp,then by putting

f

n

j1

fj, fjz

⎛ ⎜

⎝ 1−lj 1−zjlj

2

⎞ ⎟ ⎠

1/p

, lj1−lj, lj∈0,1, j 1, . . . , n, 4.26

we havefHp const andK ≥ 1/l1· · ·ln 1

1−l1· · · 1

1−ln τ

−τdμr1ξ, . . . , rnξ, τ minjlj, |1−

zjlj| 1− |zj|, |zj| ∈1−lj,1, argzj∈−τ, τ.Hence we get what we need. Now we will

show the sufficiency of the condition.

LetN >0 : 2N1r

jπ <2N 11−rj, j1, . . . , n.

Letz →−r eiϕ Un

, z r1eiϕ, . . . , rneiϕ.Let also Bzk {t t1, . . . , tn : |tjϕ| <

2kj1r

j,1≤jn}andWk1,...,kn Ik1× · · · ×Ikn,where

Ikj

t: 2kj1r

j

tϕ<2kj 11r

j

, 0≤kjN−1,

IN

t: 2N1−rj

tϕ<2N 11−rj

, |t|< π, I−1

t:|t|<1−rj

.

4.27

In what follows we will use notations ofProposition 2.1. Consider the Poisson integral of a

functionf, fH1T, uz C

TnPz, ξfξdmnξ, zUnsee2.

LetEa {zUn

∗ :Eduz> a}.We will show now as in4that

μEaC a

Tn

(18)

Indeed, this will be enough, since the operatorf → EdTnPz, ξfξdmnξisL, L

operator we can apply the Marcinkiewicz interpolation theoremsee16, Chapter 1to assert

that

J

1

0

· · ·

1

0

T

fzpdμzCfHpUn, 1< p <. 4.29

We have as in4forzUn

Eduz

Tn

Pz, wfwdmnw

N

k1,...,kn−1

Wk1,...,kn

Pz, wfwdmnw

C0

N

k1,...,kn0

1 4k1 ··· knn

k11−rk

Bk z

ftdtC

N

k1,...,kn0 1 2|k|gk

r1, . . . , rn, ϕ

N k1,...,kn0

2−|k|gk

r , ϕC1supk2|k|/2gk

r , ϕ,

4.30

where we used the standard partition of Poisson integral. HenceEaα∈ZnEαC22|α|/2a, and so usingProposition 2.1we finally getμEaC/aTnftdt.

Indeed byProposition 2.1we have

μEa

α∈Zn

μEα

C22|α|/2a

C3

α∈Zn

C21

2|α|/2a

Tn

ftdtC4 a

Tn

ftdt.

4.31

Theorem 4.6is proved.

Theorem 4.7. LetfHUn.Then

sup

|zj|<1

Dαfz1, . . . , z n

n

k1

1− |zk|α 1

C

Tn 1

0

Rsfρξ1, . . . , ρξ np

1−ρsp np−1−1dmnξdρ,

4.32

wheres >0, α >0, p≤1, sp np−1>0;

sup

z∈Un Dβfz

1, . . . , zn

αk≥0,αks

n

k1

1

1− |zk|β αk−2s−α/n ⎞ ⎠

−1

Csup

z∈Un

Rsfz1− |z|α,

(19)

where2sα >0, β >1, α >0, s >0,

sup

|zk|<1

Tn

Rsf|z

1|ϕ1, . . . ,|zn|ϕnpdmn

ϕ

αj≥0,αj2s n

k1

1

1− |zk|pαk s 1−1 −1

C

Tn 1

0

D−sfwp

1− |w|td|w|dmn

ϕ,

4.34

wheres >0, 1/s 1< p≤1, tps−1 n 1p−1>−1.

Proof. We use systematically the integral representation4.18,Lemma 2.5, and it is corollary.

The proof of the estimate4.32follows from equality

Dαfz

1, . . . , zn Cs

Tn 1

0

Rsfρξ

1, . . . , ρξn log

1

ρ

!s−1n

k1

1

1−ξϕkρk

α 1dρdmnξ,

4.35

zkϕkρk, k1, . . . , n, s >0, α >0, fHUn,and estimate

Tn 1

0

Rsfρξ

1, . . . , ρξn1−ρ

α

dρdmnξ

p

C

Tn 1

0

Rsfρξ

1, . . . , ρξnp

1−ραp n 1p−1dρdmnξ,

4.36

p≤1, α >0, s >0, fHUn,obtained during the proof ofTheorem 4.3.

The proof of the estimate4.33follows fromLemma 2.5and its corollary and integral

representation4.35.

The proof of the estimate4.34follows from equalityzjρjϕj, j 1, . . . , n

Rsfz1, . . . , z n

Cs

Tn 1

0

D−sfρξ

1, . . . , ρξn log

1

ρ

!s−1

R2sn k1

1

1−ξϕkρρk

s 1dρdmnξ

.

4.37

Indeed using4.36integrating both sides of4.37byTnand usingLemma 2.5we arrive at

(20)

Remark 4.8. All estimates inTheorem 4.7forn1 are well knownsee1, Chapter1.

We present below a complete analogue of Theorems 3.2 and 3.3 for a map from

polydisk to expanded disk. Note the continuation offfunction is done again from diagonal

z, . . . , z.LetzjU, j1, . . . , nand

fz1, . . . , zn Cs

U

fz, . . . , z1− |z|s

n

k11−zkzs 2/n

dm2z, 4.38

whereCsis a constant of Bergman representation formulasee1.

Proposition 4.9. 1 aLet n∈N, p∈0,, α >−1, fHUn.Then

T

1

0

· · ·

1

0

f|z1|ξ, . . . ,|zn|ξpn k1

1− |zk|α n−1/nd|z1| · · ·d|zn|dmξ

C

Un fzp

n

k1

1− |zk|2αdm2nz.

4.39

And reverse is also true:

bLet0< p <, α >−1and

T

1

0

· · ·

1

0

f|z1|ξ, . . . ,|zn|ξpn k1

1− |zk|α n−1/nd|z1| · · ·d|zn|dmξ<, 4.40

then for all functionsfsuch that condition4.38holds fors > maxα 2n/p−2,−1we have

fApαUn.

2 aLetfHpUn, 1< p <, n >1.Then

fAp

−1/nUn

T

1

0

1

0

f|z1|ξ, . . . ,|zn|ξp

n

k1

1− |zk|−1/ndmξd|z1| · · ·d|zm|<,

4.41

and the reverse is also true:

bFor any functionfwith a finite quasinormfAp

−1/nUnsuch that condition4.38holds fors2n−2one hasfHpUn, p >1.

Proof. 1 aProof of estimate4.39follows directly from Theorem B.

bIndeed from4.38and results of1on diagonal map in Bergman classes we have

fAp

αUnCDiagfAptU, tαn 2n−2.It remains to apply Theorem A.

2For the proof ofawe useTheorem 4.6and get the result we need.

For the proof ofbwe use the same argument as in the proof of part1. Namely first

from4.38and from a Diagonal map theorem onHp classes from1we get f

HpUn

CfAp

(21)

Remark 4.10. NoteProposition 4.9is obvious forn1.

We give only a sketch of the proof of the following result. It is based completely on a technique we developed above.

Proposition 4.11. aLetα > 0, n ∈N, n >1, s > nαthen the following assertions are true: If

DαfHpUnthen

Jp,α,sf

Tn 1

0

Rsf|z|ξ1, . . . ,|z|ξ

np1− |z|sp−npα−1d|z|dmnξ<, p≥2. 4.42

IfJp,α,sf<thenDαfHpUn, 1/α 1< p1.

bLetDαfHpUn, p2.Ifs > nα, α >0,thenJp,α,sf<.Moreover the reverse is

also true if condition4.38holds fors2n−2thenDαfp

HpCJp,α,sf.

The proof of first part ofProposition 4.11follows from4.35,4.36andLemma 2.5

directly. The reverse assertion follows from Theorem Bband estimate

Un

Dα1,...,αnfzp

n

k1

1− |zk|αkp−1dm2

nzCfHp, αk>0, k1, . . . , n, p≥2, 4.43

which can obtained from one dimensional result by induction.

The proof of second part ofProposition 4.11can be obtained from Theorem A and

results on diagonal map on Hardy classes HpUn from 1 similarly as the proof of

Proposition 4.9. For n 1 part a is well known e.g., see 1 and b follows froma

since forn1 condition4.38vanishes by Bergman representation formula.

Remark 4.12. Theorem 4.6, Propositions4.9and4.11 give an answer to a problem of Rudin

see2to find traces ofHpUnHardy classes on subvarieties other than diagonalz, . . . , z.

Note that in11Clark solved this problem for subvarietes ofUn based on finite Blaschke

products.

Acknowledgment

The authors sincerely thank Trieu Le for valuable discussions.

References

1 A. E. Djrbashian and F. A. Shamoian,Topics in the Theory of ApαSpaces, vol. 105 ofTeubner Texts in

Mathematics, Teubner, Leipzig, Germany, 1988.

2 W. Rudin,Function Theory in Polydiscs, W. A. Benjamin, New York, NY, USA, 1969.

3 R. Shamoyan and S. Li, “On some properties of a differential operator on the polydisk,”Banach Journal of Mathematical Analysis, vol. 3, no. 1, pp. 68–84, 2009.

4 F. A. Shamoian, “Embedding theorems and characterization of traces in the spacesHpUn, p

0,∞,”Matematicheskii Sbornik, vol. 107, no. 3, pp. 709–725, 1978.

5 J. M. Ortega and J. F`abrega, “Hardy’s inequality and embeddings in holomorphic Triebel-Lizorkin spaces,”Illinois Journal of Mathematics, vol. 43, no. 4, pp. 733–751, 1999.

(22)

7 J. H. Shapiro, “Mackey topologies, reproducing kernels, and diagonal maps on the Hardy and Bergman spaces,”Duke Mathematical Journal, vol. 43, no. 1, pp. 187–202, 1976.

8 G. Ren and J. Shi, “The diagonal mapping in mixed norm spaces,”Studia Mathematica, vol. 163, no. 2, pp. 103–117, 2004.

9 E. Amar and C. Menini, “A counterexample to the corona theorem for operators onH2Dn,”Pacific

Journal of Mathematics, vol. 206, no. 2, pp. 257–268, 2002.

10 R. Shamoyan, “On the action of Hankel operators in bidisk and subspaces inH∞connected with inner functions in the unit disk,”Comptes Rendus de l’Acad´emie Bulgare des Sciences, vol. 60, no. 9, pp. 929–934, 2007.

11 D. N. Clark, “Restrictions ofHpfunctions in the polydisk,”American Journal of Mathematics, vol. 110,

no. 6, pp. 1119–1152, 1988.

12 V. S. Guliev and P. I. Lizorkin, “Classes of holomorphic and harmonic functions in a polydisk in connection with their boundary values,”Trudy Matematicheskogo Instituta imeni V. A. Steklova, vol. 204, pp. 137–159, 1993.

13 R. F. Shamoyan, “On quasinorms of functions from holomorphic Lizorkin Triebel type spaces on subframe and polydisk,” inSympozium Fourier Series and Application, Conference Materials, pp. 54–55, Rostov Na Donu, 2002.

14 L. Carleson, “A counter example for measures bounded inHpfor the bidisc,” Tech. Rep. 7, Institut

Mittag-Leffler, 1974.

15 Z. Wu, “Carleson measures and multipliers for Dirichlet spaces,”Journal of Functional Analysis, vol. 169, no. 1, pp. 148–163, 1999.

References

Related documents