Journal of Inequalities and Applications Volume 2010, Article ID 647627,14pages doi:10.1155/2010/647627
Research Article
Asymptotic Behavior for a Class of Modified
α
-Potentials in a Half Space
Lei Qiao
1and Guantie Deng
21Department of Mathematics and Information Science, Henan University of Finance and Economics,
Zhengzhou 450002, China
2Laboratory of Mathematics and Complex Systems, School of Mathematical Science,
Beijing Normal University, MOE, Beijing 100875, China
Correspondence should be addressed to Guantie Deng,[email protected]
Received 13 March 2010; Accepted 21 June 2010
Academic Editor: Shusen Ding
Copyrightq2010 L. Qiao and G. Deng. 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.
A class ofα-potentials represented as the sum of modified Green potential and modified Poisson integral are proved to have the growth estimatesRα,l,lx ox
β
n|x|l−2β2h|x|−1at infinity in the
upper-half space of then-dimensional Euclidean space, where the functionh|x|is a positive non-decreasing function on the interval0,∞satisfying certain conditions. This result generalizes the growth properties of analytic functions, harmonic functions, and superharmonic functions.
1. Introduction and Main Results
Let Rnn ≥ 2 denote the n-dimensional Euclidean space with points x
x1, x2, . . . , xn−1, xn x, xn, where x ∈ Rn−1 and xn ∈ R. The boundary and closure of
an open Ω of Rn are denoted by ∂Ω and Ω, respectively. The upper half-space is the set
H {x x, xn ∈ Rn; xn > 0}, whose boundary is∂H. We identify Rn with Rn−1×R
and Rn−1 with Rn−1 × {0}, writing typical pointsx, y ∈ Rn as x x, x
n, y y, yn,
wherex x1, x2, . . . , xn−1, y y1, y2, . . . , yn−1 ∈ Rn−1 and putting x·y
n
j1xjyj
x·yxnyn, |x|
√
x·x, |x|√x·x.
Forx∈Rnandr >0, letB
nx, rdenote the open ball with center atxand radiusrin
Rn.
It is well known that see, e.g., 1, Chapter 6 the positive powers of the Laplace operatorΔcan be defined by
whereα >0,fis a Schwarz function and
Ffξ fξ
Rn
fxe−ixξdx. 1.2
It follows that we can extend definition 1.1 to certain negative powers of −Δ,
−Δ−α/2for 0< α < nand define an operatorI
αby
Iαf −Δ−α/2f F−1
|ξ|−αf, 1.3
where 0< α < nandfis a function in the Schwartz class.
IfIαis defined as the inverse Fourier transform of|ξ|−αin the sense of distributions,
one can show that
Iαx γα|x|α−n, 1.4
whereγαis a certain constantsee, e.g.,1, page 414for the exact value ofγα.
The functionIα is known as the Riesz kernel. It follows immediately from the rules
for manipulating Fourier transforms that any Schwartz functionf can be written as a Riesz potential,
fx Iαgx
Iα∗g
x γα
Rn
gy
x−y n−αdy, 1.5
where 0< α < nandg −Δα/2f.
This Riesz kernelIαinRninspired us to introduce the modified Riesz kernel forH. To
do this, we first set
Eαx
⎧ ⎨ ⎩
−log|x| ifαn2,
|x|α−n if 0< α < n.
1.6
LetGαx, ybe the modified Riesz kernel forH, that is,
Gα
x, yEα
x−y−Eα
x−y∗, x, y∈H, x /y, 0< α≤n, 1.7
where∗denotes reflection in the boundary plane∂Hjust asy∗ y1, y2, . . . , yn−1,−yn.
We define the kernel functionPαx, ywhenx∈Handy∈∂Hby
Pα
x, y ∂Gαx, y
∂yn
yn0 Cα
xn
x−y n−α2, 1.8
We remark thatG2x, yand P2x, yare the classical Green function and classical Poisson kernel forHrespectivelysee, e.g.,2, page 127.
Next we use the following modified kernel functionPα,mx, ydefined by
Pα,m x, y ⎧ ⎪ ⎪ ⎨ ⎪ ⎪ ⎩ Pα
x, y if y <1,
Pα
x, y−
m−1
k0
Cαxn|x|k
y n−α2kC n−α2/2
k
x·y |x| y
if y ≥1, 1.9
wheremis a nonnegative integer;Cωkt ω n−α/2 is the ultrasphericalor Gegenbauer polynomialssee3. The Gegenbauer polynomials come from the generating function
1−2 trr2−ω
∞
k0
Cωktrk, 1.10
where |r| < 1, |t| ≤ 1, and ω > 0. The coefficients Cω
kt are called the ultraspherical or
Gegenbauerpolynomials of degreekassociated withω, each functionCωktis a polynomial of degreekint. Here note thatP2,mx, yis the modified Poisson kernel inH, which has been
used by several authorssee, e.g.,4–8.
Motivated by this modified kernel function Pα,mx, y, it is natural to ask if the
functionGαx, ycan also be modified? In this paper, we give an affirmative answer to this
question.
First we consider the modified kernel function in caseαn2, which is defined by
En,l
x−y
⎧ ⎪ ⎪ ⎨ ⎪ ⎪ ⎩ En
x−y if y <1,
En
x−yR
logy−
l−1
k1
xk
kyk
if y ≥1. 1.11
In case 0< α < n, we define
Eα,l
x−y
⎧ ⎪ ⎪ ⎨ ⎪ ⎪ ⎩ Eα
x−y if y <1,
Eα
x−y−
l−1
k0
|x|k y n−αkC
n−α/2
k
x·y |x| y
if y ≥1, 1.12
wherelis a nonnegative integer,x, y∈H, andx /y.
Then we define the modified kernel functionGα,lx, yby
Gα,l x, y ⎧ ⎨ ⎩
En,l1
x−y−En,l1
x−y∗ if αn2,
Eα,l1
x−y−Eα,l1
x−y∗ if 0< α < n.
Write
Gα,l
x, μ
H
Gα,l
x, ydμy,
Uα,mx, ν
∂H
Pα,m
x, ydνy,
1.14
where μresp.,ν is a nonnegative measure onHresp., ∂H. Here note thatGα,0x, μ is
nothing but the Green potential of general ordersee9–11. Following Fugledesee6, we set
ky, μ
E
ky, xdμx, kμ, x
E
ky, xdμy, 1.15
for a nonnegative Borel measurable functionkonRn×Rnand a nonnegative measureμon a
Borel setE⊂Rn. We define a capacityC kby
CkE sup μRn, E⊂H, 1.16
where the supremum is taken over all nonnegative measuresμsuch thatSμ the support of
μis contained inEandky, μ≤1 for everyy∈H.
Forβ≤1 andδ≤1, we consider the functionkα,β,δdefined by
kα,β,δ
y, xx−nβyn−δGα
x, y forx, y∈H. 1.17
Ifβ δ 1, thenkα kα,1,1 is extended to be continuous onH×Hin the extended sense,
whereHH∪∂H.
Now we will discuss the behavior at infinity of the modified Green potential and modified Poisson integral in the upper-half space, respectively. For related results, we refer the readers to the papers by Mizutasee9, Siegel and Talvilasee8, and Mizuta and Shimomurasee7.
Theorem 1.1. Lethrbe a positive nondecreasing function on the interval0,∞such that
arβ−1hris nondecreasing on0,∞, brβ−2hris nonincreasing on0,∞andlim
r→ ∞rβ−2hr 0,
cthere exists a positive constantMsuch thath2r≤Mhrfor anyr >0.
Letμbe a nonnegative measure onHsatisfying
H
yδ nh y
1 y nl−α−βδ2
dμy<∞. 1.18
ilim|x| → ∞,x∈H−Exn−β|x|−l2β−2h|x|Gα,lx, μ 0;
ii∞i12−in−αβδC
kα,β,δEi<∞,
whereEi{x∈E: 2i≤ |x|<2i1}.
Corollary 1.2. Letμbe a nonnegative measure onHsatisfying
H
yn
1 y nl−α2dμ
y<∞. 1.19
Then there exists a Borel setE⊂Hwith properties
ilim|x| → ∞,x∈H−Exn−β|x|β−l−1Gα,lx, μ 0;
ii∞i12−in−αβ1C
kα,β,1Ei<∞, whereEi{x∈E: 2i≤ |x|<2i1}.
Theorem 1.3. Lethbe defined as inTheorem 1.1andνa nonnegative measure on∂Hsatisfying
∂H
h y
1 y nm−α−β3dν
y<∞. 1.20
Then there exists a Borel setF⊂Hwith properties
ilim|x| → ∞,x∈H−Fx−nβ|x|−m2β−2h|x|Uα,mx, ν 0;
i∞i12−in−αβ1C
kα,β,1Fi<∞, whereFi{x∈F: 2i≤ |x|<2i1}.
Remark 1.4. In the caseml,FE.
Corollary 1.5. Letνbe a nonnegative measure on∂Hsatisfying
∂H
1
1 y nl−α2dν
y<∞. 1.21
Then there exists a Borel setE⊂HsatisfyingCorollary 1.2(ii) such that
lim
|x| → ∞,x∈H−Ex
−β
n |x|β−l−1Uα,lx, ν 0. 1.22
We define the modifiedα-potentials onHby
Rα,l,mx Gα,l
x, μUα,mx, ν, 1.23
The following theorem follows readily from Theorems1.1and1.3.
Theorem 1.6. Lethbe defined as inTheorem 1.1andRα,l,lxdefined by1.23. Then there exists a
Borel setE⊂HsatisfyingCorollary 1.2(ii) such that
lim
|x| → ∞,x∈H−Ex
−β
n |x|−l2β−2h|x|Rα,l,lx 0. 1.24
Remark 1.7. In the caseh|x| |x|1−β0 ≤β ≤ 1, by usingLemma 2.5below, we can easily
show thatCorollary 1.2iiwithα 2 means thatEisβ-rarefied at infinity in the sense of
12. In particular, This condition withα 2,β 1, andh|x| ≡ 1resp.,α 2,β 0, and
h|x| |x|means thatEis minimally thin at infinityresp., rarefied at infinityin the sense of13.
Theorem 1.6is the best possibility as to the size of the exceptional set. In fact we have the following result. The proof of it is essentially due to Mizutasee9, Theorem 2, so we omit the proof here.
Theorem 1.8. LetE⊂Hbe a Borel set satisfyingCorollary 1.2(ii),hdefined as inTheorem 1.1, and
Rα,l,lxdefined by1.23. Then we can find a nonnegative measureλdefined onHsatisfying
H
h y
1 y nl−α−β3dλ
y<∞, 1.25
such that
lim sup
|x| → ∞,x∈E
xn−β|x|−l2β−2h|x|Rα,l,lx ∞, 1.26
wheredλy yndμyy∈Handdλy dνyy∈∂H.
2. Some Lemmas
Throughout this paper, let M denote various constants independent of the variables in questions, which may be different from line to line.
Lemma 2.1. There exists a positive constantMsuch thatGαx, y≤Mxnyn/|x−y|n−α2,where
0< α≤n, x x1, x2, . . . , xn, andy y1, y2, . . . , yninH.
This can be proved by simple calculation.
Lemma 2.2. Gegenbauer polynomials have the following properties: i|Cω
kt| ≤Ckω1 Γ2ωk/Γ2ωΓk1, |t| ≤1;
ii d/dtCωkt 2ωCkω−11t, k≥1;
iii∞k0Cωk1rk 1−r−2ω;
Proof. iandiican be derived from3.iiifollows by takingt 1 in1.10;ivfollows byi,iiand the Mean Value Theorem for Derivatives.
Lemma 2.3. Letlbe a nonnegative integer andx,y ∈Rnα n 2, then one has the following
properties:
i|Ilk0xk/yk1| ≤l−1
k02kxn|x|k/|y|k2; ii|I∞k0xkl1/yk| ≤2l1x
n|x|l;
iii|Gn,lx, y−Gnx, y| ≤Mlk1kxnyn|x|k−1/|y|k1;
iv|Gn,lx, y| ≤M∞kl1kxnyn|x|k−1/|y|k1.
Lemma 2.4see14. Letmbe a nonnegative integer andM >0.
iIf1≤ |y| ≤ |x|/2, then|Pα,mx, y| ≤Mxn|x|m−1/|y|nm−α1.
iiIf|y| ≥2|x|and|y| ≥1, then|Pα,mx, y| ≤Mxn|x|m/|y|nm−α2.
The following lemma can be proved by using Fuglede6, Th´eor`em 7.8.
Lemma 2.5. For any Borel setEinH, we haveCkα,β,1E Ckα,β,1Eand
Ckα,β,δE infλH
resp. infλH ifδ <1 resp., δ1, 2.1
where the infimum is taken over all nonnegative measuresλonH (resp.,H) such thatkα,β,δλ, x≥1
for everyx∈E.
3. Proof of
Theorem 1.1
For any1 >0, there existsR1>2 such that
{y∈H,|y|≥R1}
yδ nh y
1 y nl−α−βδ2dμ
y< 2. 3.1
For fixedx∈Hand|x| ≥2R1, we write
Gα,l
x, μ
H1
Gα
x, ydμy
H2
Gα
x, ydμy
H3
Gα,l
x, y−Gα
x, ydμy
H4
Gα,l
x, ydμy
H5
Gα
x, ydμy
H6
Gα,l
x, y−Gα
x, ydμy
H7
Gα,l
x, ydμy
V1x V2x V3x V4x V5x V6x V7x,
where
H1
y∈H: y ≥R1, x−y ≤
|x|
2
, H2
y∈H: y ≥R1,
|x|
2 < x−y ≤3|x|
,
H3
y∈H: y ≥R1, x−y ≤3|x|
, H4
y∈H: y ≥R1, x−y >3|x|
,
H5 H6
y∈H: 1≤ y < R1
, H7
y∈H: y <1.
3.3
We distinguish the following two cases.
Case 10< α < n. Note thatV1x xβn
H1kα,β,δy, xy
δ
ndμy.In view of1.18, we can find
a sequence{ai}of positive numbers such that limi→ ∞ai∞and∞i1aibi<∞, where
bi
{y∈H:2i−1<|y|<2i2}
yδ nh y
y nl−α−βδ2dμ
y. 3.4
Consider the sets
Ei
x∈H: 2i≤ |x|<2i1, x−nβV1x≥a−i12il−2β2h
2i1−1
, 3.5
fori1,2, . . . .Ifωis a nonnegative measure onHsuch thatSω ⊂Eiandkα,β,δy, ω≤1 for
y∈H, then we have
H
dω
≤ai2−il−2β2h
2i1 x−β
n V1xdωx
ai2−il−2β2h
2i1
{y∈H:2i−1<|y|<2i2}
kα,β,δ
y, ωynδdμy
≤Mai2−il−2β2h
2i1
{y∈H:2i−1<|y|<2i2}
yδndμy
Mai2−il−2β2h
2i1
{y∈H:2i−1<|y|<2i2}
y 2−β h y
y nl−α2
1 y 2−δ
yδ nh y
1 y nl−α−βδ2dμ
y
≤M2−il−2β22i22−β2i2nl−α22−i2−δ
{y∈H:2i−1<|y|<2i2}
h y y nm−α−β3dμ
y
≤M2in−αβδaibi.
So that
Ckα,β,δ
Ei≤M2in−αβδaibi, 3.7
which yields
∞
i1
2−in−αβδCkα,β,δ
Ei<∞. 3.8
SettingE∞i1Ei, we see thatTheorem 1.1iiis satisfied and
lim sup
|x| → ∞,x∈H−Ex
−β
n |x|−l2β−2h|x|V1x≤lim sup i→ ∞ a
−1
i 0. 3.9
Moreover byLemma 2.1,
|V2x| ≤Mxn
H2
yn
x−y n−α2dμ
y
≤Mxn|x|α−n−2
H2
y 2−β
h y y
nl−α1 yδnh y
1 y nl−α−βδ2dμ
y
≤M2xn|x|l−β1h4|x|−1.
3.10
Note thatCω
0t ≡ 1. By iiiandivin Lemma 2.2, we taket x·y/|x||y|,t∗ x·y∗/|x||y∗|inLemma 2.2ivand obtain
|V3x| ≤
H3
l
k1
|x|k
y n−αk2n−αC
n−α2/2
k−1 1
xnyn
|x| y dμ
y
≤Mxn|x|l−1 l−1
k1
1 4k−1C
n−α2/2
k−1 1
H3
y 2−β
h y
yδ nh y
1 y nl−α−βδ2dμ
y
≤M1xn|x|l−β1h4|x|−1.
3.11
Similarly, we have byiiiandivinLemma 2.2
|V4x| ≤
H4 ∞
kl1
|x|k
y n−αk2n−αC
n−α2/2
k−1 1
xnyn
|x| y dμ
y
≤Mxn|x|l
∞
kl1
1 2k−1C
n−α2/2
k−1 1
H4
y 1−β
h y yδ
nh y
1 y nl−α−βδ2dμ
y
≤M1xn|x|l−β1h2|x|−1.
ByLemma 2.1, we have
|V5x| ≤Mxn
H5
yn
x−y n−α2dμ
y
≤Mxn|x|α−n−2
H5
y 2−β h y y
nl−α1 yδnh y
1 y nl−α−βδ2dμ
y
≤Mxn|x|l−1R2−1βhR2 −1.
3.13
Similarly asV3x, we obtain
|V6x| ≤
H6
l
k1
|x|k
y n−αk2n−αC
n−α2/2
k−1 1
xnyn
|x| y dμ
y
≤Mxn l
k1
Ckn−−1α2/21|x|k−1Rl−k1
2
H6
y 1−β h y
yδ nh y
1 y nl−α−βδ2
dμy
≤MRl1xn|x|l−1h1−1.
3.14
Finally, byLemma 2.1, we have
|V7x| ≤Mxn|x|l−1h1−1. 3.15
Combining3.9–3.15, we prove Case1.
Case 2αn2. In this case, the growth estimates ofV1x,V2x,V5xandV7xcan be proved similarly as in Case1. Inequations3.9,3.10,3.13and3.15still hold.
Moreover we have byLemma 2.3iii
|V3x| ≤M
H3
l
k1
kxnyn|x|k−1
y k1 y
l1 y 2−β
h y yδ
nh y
1 y l−βδ2dμ
y
≤Mxn|x|l−1 l
k1
k
4k−1
H3
y 2−β
h y yδ
nh y
1 y l−βδ2dμ
y
≤M1xn|x|l−β1h4|x|−1.
ByLemma 2.3iv, we have
|V4x| ≤M
H4 ∞
kl1
kxnyn|x|k−1
y k1 y
l2 y 1−β
h y yδ
nh y
1 y l−βδ2dμ
y
≤Mxn|x|l
∞
kl1
k
2k−1
H4
y 1−β
h y yδ
nh y
1 y l−βδ2dμ
y
≤M1xn|x|l−β1h2|x|−1.
3.17
Similarly asV3x, we have
|V6x| ≤MRl1xn|x|l−1h1−1. 3.18
Combining3.9,3.10,3.13,3.15, and3.16–3.18, we prove Case2.
Hence we complete the proof ofTheorem 1.1.
4. Proof of
Theorem 1.3
For any2 >0, there existsR2>2 such that
{y∈∂H,|y|≥R2}
h y
1 y nm−α−β3dν
y< 2. 4.1
For fixedx∈Hand|x| ≥2R2, we write
Uα,mx, ν
G1
Pα,m
x, ydνy
G2
Pα,m
x, ydνy
G3
Pα,m
x, y−Pα
x, ydνy
G4
Pα
x, ydνy
G5
Pα,m
x, ydνy
U1x U2x U3x U4x U5x,
4.2
where
G1
y∈∂H: y <1, G2
y∈∂H: 1≤ y < |x|
2
,
G3G4
y∈∂H: |x| 2 ≤ y
<2|x|, G
5
y∈∂H: y ≥2|x|.
First note that
|U1x| ≤Mxn
| x|
2
α−n−2
G1
dνy
≤Mxn|x|α−n−2
G1
y 2−β
h y y
nm−α1 h y
1 y nm−α−β3ν
y
≤Mxn|x|m−1h1−1.
4.4
Write
U2x U21x U22x, 4.5
where
U21x
G2Bn−10,R2
Pα,m
x, ydνy,
U22x
G2−Bn−10,R2
Pα,m
x, ydνy.
4.6
We obtain byLemma 2.4i
|U2x| ≤Mxn|x|m−1
G2
1
y nm−α1dν
y
≤Mxn|x|m−1
G2
y 2−β
h y
h y
1 y nm−α−β3ν
y.
4.7
For|x|>2R2, by4.7we have
|U21x| ≤Mxn|x|m−1R22−βhR2
−1. 4.8
On the other hand,4.7yields that
|U22x| ≤M2xn|x|m−β1h
| x|
2
−1
. 4.9
Combining4.8and4.9, we have
lim
|x| → ∞,x∈Hx
−β
We have byLemma 2.2iii
|U3x| ≤M
G3
m−1
k0
xn|x|k
y n−α2kC
n−α2/2
k 1dν
y
≤Mxn|x|m m−1
k0
1 2kC
n−α2/2
k 1
G3
y 1−β h y
h y y nm−α−β3dν
y
≤Mεxn|x|m−β1h
| x|
2
−1
.
4.11
ByLemma 2.4ii, we obtain
|U5x| ≤Mxn|x|m
G5
1
y nm−α2dν
y
≤Mxn|x|m
G5
y 1−β h y
h y
1 y nm−α−β3ν
y
≤M2xn|x|m−β1h2|x|−1.
4.12
Note thatU4x xβn
G4kα,β,1y
, xdνy. By the lower semicontinuity ofk
α,β,1y, x,
we can prove the following fact in the same way asV1xin the proof ofTheorem 1.1:
lim sup
|x| → ∞,x∈H−F
x−nβ|x|−m2β−2h|x|U4x 0, 4.13
whereF∞i1Fi, Fi{x∈F: 2i≤ |x|<2i1}, and∞i12−in−αβ1Ckα,β,1Fi<∞.
Combining4.4and4.10–4.13, we complete the proof ofTheorem 1.1.
Acknowledgments
This work is supported by The National Natural Science Foundation of China under Grant 10671022 and Specialized Research Fund for the Doctoral Program of Higher Education under Grant 20060027023.
References
1 L. Grafakos,Classical and Modern Fourier Analysis, Pearson Education, Upper Saddle River, NJ, USA, 2004.
2 L. H ¨ormander,Notions of Convexity, vol. 127 ofProgress in Mathematics, Birkh¨auser, Boston, Mass, USA, 1994.
3 G. Szeg˝o,Orthogonal Polynomials, American Mathematical Society, Providence, RI, USA, 4th edition, 1975, American Mathematical Society, Colloquium Publications, Vol. XXII.
5 M. Finkelstein and S. Scheinberg, “Kernels for solving problems of Dirichlet type in a half-plane,” Advances in Mathematics, vol. 18, no. 1, pp. 108–113, 1975.
6 B. Fuglede, “Le th´eor`eme du minimax et la th´eorie fine du potentiel,”Universit´e de Grenoble. Annales de l’Institut Fourier, vol. 15, no. 1, pp. 65–88, 1965.
7 Y. Mizuta and T. Shimomura, “Growth properties for modified Poisson integrals in a half space,” Pacific Journal of Mathematics, vol. 212, no. 2, pp. 333–346, 2003.
8 D. Siegel and E. Talvila, “Sharp growth estimates for modified Poisson integrals in a half space,” Potential Analysis, vol. 15, no. 4, pp. 333–360, 2001.
9 Y. Mizuta, “On the behavior at infinity of Green potentials in a half space,”Hiroshima Mathematical Journal, vol. 10, no. 3, pp. 607–613, 1980.
10 Y. Mizuta, “Boundary limits of Green potentials of general order,” Proceedings of the American Mathematical Society, vol. 101, no. 1, pp. 131–135, 1987.
11 Y. Mizuta, “On the boundary limits of Green potentials of functions,”Journal of the Mathematical Society of Japan, vol. 40, no. 4, pp. 583–594, 1988.
12 H. Aikawa, “On the behavior at infinity of nonnegative superharmonic functions in a half space,” Hiroshima Mathematical Journal, vol. 11, no. 2, pp. 425–441, 1981.
13 M. Ess´en and H. L. Jackson, “On the covering properties of certain exceptional sets in a half-space,” Hiroshima Mathematical Journal, vol. 10, no. 2, pp. 233–262, 1980.