Professional Documents
Culture Documents
Kyungkeun Kang
Abstract We study boundary regularity of weak solutions of the three dimensional NavierStokes equations in a half-space. We prove that a weak solution , which is locally in the class with near boundary is H lder continuous up to the o boundary.
Key words. Navier-Stokes equations, Green tensor AMS subject classications. 76D03, 76D05
1 Introduction
In this paper we study the boundary regularity of weak solutions of the Navier-Stokes equations in and : in
where and are assumed to be smooth and compactly supported in and , respectively. The well known result of J. Serrin states that if a weak solution belongs locally to at an interior point where , then is locally bounded near and thus is smooth with respect to space variables in a neighborhood of away from the boundary (see [12]). M. Struwe extended the Serrins result to the critical case, with (see [15]). Y. Giga gave the global version of the Serrin and Struwes result in smooth bounded domains (see [6]). For the local version of the problem, Takahahi proved the following (see [16]). If a weak solution satises locally with and
on
at
$ U
tr us9
q W 5$& #
2 G
IG ' G
7 C Vur $ U
F D @EC
r yx9 $& wv tr c W q #
9 ' G
(1)
(2)
IG
C 7
% $& % #
, then is bounded up to the boundary under the following additional integrability assumptions of gradient of velocity and the pressure
For suitable weak solutions, H. Choe also proved the boundedness of up to the boundary with and under the assumption that for the limiting case that pressure is bounded at the boundary (see [4]). Our goal is in this paper is to improve the local regularity near boundary when a weak solution belongs locally to at a boundary point where without , we any additional assumption. For a given boundary point denote a parabolic cylinder around by where . We prove the following. The proof will be given in Section 3. Main Theorem Let be a weak solution of Navier-Stokes equations (1). Let . Suppose that satises
Then
is H lder continuous in o
, in other words,
Unlike in the interior case, it is not clear that H lder continuity implies higher reguo larity in space variables at a boundary point. Our main tool is an estimate for the Green tensor of the Stokes system in a half-space, which seems to be independent interest. A representation formula for the solution of the Stokes system with nonzero initial and the boundary data or non-zero external force and initial data were found by the S. Ukai (see [18]) and M. Cannone, F. Planchon, and M. Schonbek (see [3]), respectively. In this paper, we prove the following point-wise estimates of the Green tensor for for the half-space where :
where and is any number with . Although such an estimate may probably not be surprising to experts, we were not able to nd it in the literature. The estimate seems to be useful in other situations. The plan of this paper is as follows: In section 2, we rst review some results by E. B. Fabes, B. F. Jones and N. M. Riviere (see [5]) and Solonnikov (see [13]) and use them to derive the Green tensor for the Stokes system in the half-space. With the aid of the representation formula of the Stokes system, we obtain (3). In section 3, we present the proof of the main theorem.
h C r i r r ' r ii
i 5 9
for every
with
i jC 7
'
xr
for some
with
and
' k
i C j@' 7 C a f
(3)
hg C U
' $ U
f gr e
for all
with
h C U
{ z w n m y &
y }r | c { w &n m W { H ~82 G
F D C
2 g r Tb G 2 G c G
'
$& u %# v o r p
7 C 8
2.1 Preliminaries
Let us begin with some denitions and notations used throughout this paper. We denote the whole space and the half-space by and in dimensional space, respectively. Let be a given positive parameter and we denote for simplicity. For given , we denote a half ball with radius around by , namely and we also write a parabolic cylinder around by under the restriction and . Let be a domain. For , denote the usual Sobolev space, i.e. . As usual, is dened the completion of in . Let and . denote the Banach space consisting of all measurable functions with a nite norm
We denote
if there is no confusion.
For a given parabolic domain , indicates the Banach spaces of functions that are H lder continuous with and to variables and , respectively in . o A characteristic function supported in is denoted by . The convolution of two functions is denoted by with subscript. To be precise, suppose and where or . Then and are dened as follows
In particular, in case of , functions are understood as extended ones to the whole space by assigning zero in the lower half-space. We denote by the fundamental solution of the heat equation in , namely . Let and be nonnegative integers and where is integer for each . a multi-index with . we dene
s 9
' r r H '
' cr 1W
tus9 r W '
c W
f gr f r 1 i ' r 1 ~ Vi f P W c W c W c P H ~ i $& %# $ # ' %5@ $ %# B3 4 $ tr %# xc yx9 W 4 er q q %@lm $& # } 1 q $ # c m W { H } r | %m r r c uc o & r p W c SU EP uW j ' c o &n m W j s s W r i c c e%G W ' Yx s G `W
$ %#
C r i r ' ii ii $ uP r i r W '
$& %#
A c W r A 9 7 s$ sc ("U r e
$ %#
c yc
W x W @ c W } c W ~| 9 xr 4 c W 2 G c W 2 G A P sP W q o r 0' o m k & p# 2 ) 1 4$5t&n ' r p P "q %#w2 r 1 & ' $& v c W Sgx9 2 t r W {
r r r ' 9 D ur sB%{ H
)q' c W $%@X # t r ) d
r e
&r p g H i
If there is no confusion, summation convention is understood over repeated indices running from 1 to , for instance . The capital letter is used to denote the generic constant, the value of which may change from line to line. In this section, we nd the Green tensor and their estimates for the following Stokes system in the half-space :
in in
on
as
is to combine the Green tensor for the The strategy for obtaining the Green tensor whole space proved in [5] and the Green tensor for the boundary value problem for the Stokes system in a half-space proved in [13]. Let us rst recall the Green tensor, the fundamental solution of the Stokes system for the whole space (see [5]). Let and be extensions of and in (4) to the whole space by assigning zero to a lower half-space. Here we consider the following Stokes system:
when
q $ %#
in
h C r i r r ii
where and are smooth and compactly supported in and and the pressure tensor Once we nd the Green tensor solving the following Stokes system for each
f f i w r r 1 W t U f xr i xr r r 1 W t c W c c W c t t ' r U i c 1W G ' G r t r s9 P c 1W P c W c P 1 W 9 ' sxr I1 t r 9 q $& # %wv c W 9 ' t ' r 1 t q %# $& r v c W b ` r r t q %# $& c 1 W ' r v c 1 W VSH v W U c R P h C r i r r ' r ii wt $& % # q $& % # r 5$& wv 9 ' xur 1 r 9 r ' s9 j ' ur 1 9 q # c W G c W %G
$ %#
q $& #
9 ' 8
$ %#
Y i ' z i $
$ %#
'
in
(4)
on
(5)
(6)
(7)
(8)
i i f n
c 1W 9 ' c r c 1 W X'
IP r 1 W m v r
c W c r r U Qc
' c 1 W
r f xr i xr r r 1 f f W m ' c i mW ' c W c H f f c W c r r 1 W tU f c xr W i c xr r r 1 W
5
and
r B$ 1 v 1 h r $ 1 { ' r 1 m { c P C c W ' W v c $ js c W P c W 1 c $ js W W
where is the fundamental solution of the heat equation and mations, namely
sum up, for a given and in (4) and (5), (8). Besides when it is restricted to a half-space, it solves
In general, since is not zero, may not be a solution for (4). Next let us set and where and are the Green tensor and the pressure tensor for the whole space mentioned above and is the
q $& 5%#
where
Moreover,
where
and
is in
for
and
with
on
when
in
in
and
, solves
(10) (9) To
9 '
$ Iv Q 1 $ 1 r Iv h P ' wv | v $
! "
$ 1 P i $ 1 P $ r r c B P 1 W $ 1 r r9 P c sxr B W
(14)
$ r { c U 1 W { U 1W
{ { c
F D (C
With the aid of the Green tensor and their estimates for the half-space in [13], the integral can be obtained explicitly, and thus we can calculate point-wise representation of estimates of , which also implies the point-wise estimates of . Before we analyze , let us rst review Solonnikovs results for the boundary value problem of the Stokes system in a half-space. Although it was studied only in three dimensional case, it could be easily generalized to a general dimension . Let us recall some functions and their estimates used later (see [13, page 37, 41]).
and
h C r i r r ' r ii
6
in
q %# $&
in
q %# $&
when
In addition, they satisfy the following relations (see [13, page 40]):
1 W | H r r a 1 W p H r r a a
{ { c
H r a
9 ur
' c 1W '
r U Qc
r 9 xur 1 W r c 1W ` b 1 W cdRSQH v W P
where
and
if
r (' C
where in a half-space:
in
q %# $&
when on
$& Iv q # 9 '
on
) 0
9 '
' (
% &
pr
t i t U h } } i ii jC r i r r ' r
for
} } i x rs9xr W ex P r P 1 W c c H C r i ii P C r i r ii $ r { c U 1 W { c U 1 W { &
1 $
r r H a P
i ii jC r i r ' r
1 $ r u $ 1 Iv $ c W v
$ 1
v 1 $ P xv |
r 4
x r W x P r P 1 W c c
) 5 4
r ' c W ' c 1W H
) 3 21
where
The solution of the problem above is given by (see [13, page 53])
in
q $& %#
(see [13, page 48]):
in
q %# $&
on
i q %wv $& #
(16)
' r H
h C r i r r ' r ii
9 7
1 8
Although the explicit representation formula of a pressure is also available (see [13, page 48, 52]), it is omitted because it is not needed for our purpose. To sum up, and solves the following boundary value problem of the Stokes system:
'
4 A
1@b ` 6 7
Moreover,
in
q $& 5%#
when
1 5P
t '
U i W
' c r 1 W
h C r i r r ii '
$ Iv h 1 ' r 1 wv P c W
where
Since the estimate of has been known, it is required to analyze only that of , from which we have automatically the estimate of . We start with a simple observation on .
for all
Since and (24), it sufces to check if Using the identities (24) and (23), and substituting
r c 1 W X' r 5' k r H r r c 1 W r H
i c r 1 W
C
$ r 1 Y gI C H
BI $
xr
r t (' c r 1 W t r H r c 1 W H {
' c r 1 W r H
' c r 1 W
C DC
r C 2C 1 H
q %2 c r 1 W $& #
{ c U 1 W r 1 r c W 8 H
wt
r yur ' c C W
' r H $ ur i r ii W
(22)
(24)
'
9 D
'
Vb
'
'
wt
t '
r r H
, it
Theorem 2.3 Let be the Green tensor for (6). Then for every nonnegative integer , it satises the following estimate: For all ,
where
and
Proof. We note rst that it sufces to prove (25) for cases of , and because of Lemma 2.2. Moreover, in this proof, we only consider the case because difculty in computations occurs when is close to zero. If , then we can have the same estimate more easily by following similar procedures as the case . In addition, We shall prove (25) only for the case and , i.e.
Indeed, the tangential derivatives (for example ) are easily estimated. If , it could be estimated as the case that and by using the integration by parts successively. To be precise, we recall rst the representation of already mentioned in (17).
Note that the rst term of satises heat equation and the last one holds the Laplace equation. In addition, for the second term, we can use the relation (15). To sum up, in this proof, we consider only the case that , and , which is sufcient for proving (25).
j '
Thanks to the observations above, it sufces to estimate . ready to prove the estimate of green tensor
9 ! ' B HE 9 D ur 9 ' ur 9 g c B P 1 W 1 r F E H r H r r r E r c 1 W 8 & { (' c 1 W 1 a & { X' H C ~C r W E y~GP r P 1 W FE 1 r H & { X' c c C H C ~C r W E y~GP r P 1 W FE 1 r H & { X' c c C H r C { c C r B W E ec AP r B P 1 W HE 1 8 ' H C H
for all integers when
because
1 $ B$ 1 Iv $ i c W u v
q %2 c r 1 W $& #
'
r t c 1W i
} g$ 1 $ 1
9 '
' 9 h 7
$ 1
1 Iv $ h $ Iv h 1 r P v P wv P ' c 1 W |
c r 1 W t
9 7
$ 1Iv v
9 D
r yexr
P 1
. Now we are
(25)
(26)
i P C r i r ii '
, let us recall the repre-
r C r i r h r j ii '
1 Iv $ v |
FE
FE 1
c 1W
r } $ 1
} $ 1 1 U { c U W
{
V X
c U d W
V U
} $ k 1W { c U c P yW { k $ c x U YW { c U k ' { {
4 T
4 W
1 c x U { 1W c U { k
U YW k c U 1 W { k { c P 1 uW k W 1 x c W
U
i c 1W
x U c 1 W
R SI `
c 1W Q Q { 1 $ c W $ U ` {
i c 1W
RBI 4 4 E A 4 4 E
FE 1
r
4A FE 1
where means that they are comparable. According to (20) and Lemma 2.1, it . Since satises different estimates depending on sufces to estimate or , it is required to analyze them separately. Reminding the expression (19), we have
r c 1W
r
C ' p
r 32 p k
P Q k
2 k
U 1 U P 1 k
k) k)
p h h i A 1 k 4 $ 2 # h 1 A 1 h D k 4 $ 2 # k 1
and , for simplicity, in
I p I
$ B#
$& # 2 %331
$ 1 Iv h ' r 1 P wv P c W 9 C @'
P
C '
For
Let us rst consider the rst term the range of time interval.
, using the estimate (18) and relations (27), and (27), we obtain
! SY
FE 1
, which is also split into two parts depending on , the following relations hold: dened as follows: For 10 if (27)
k 1W c { x U c P yW k
4 W
V 4
} B$ 1 } B$ 1 1 9 { { c U W U k 1 { c U W { c U YW { { U { { k x k `3 tU k `8 { { x } B$ 1 k { c U W k
for for
{ } $ 1
4 U R
V
U
i jC 9
c U c P 1 uW k W c { k
V U R "I
U W
1 { c U W c
4 T c
k 1W k i U x k { c U YW { c U c P yW U k k 1W { c U W c { x U c P uW c 1 W k { { Qu & { r W yc v $ P n c k n n x P h r U k yc P 1 W
4 A
9 {{ h r k k c W u r YW ec v c k n n B$ k x P
4 G
b c rk # P v Wx4 P
1W
$ r & k c k x S P 1 W Qu c 1W c r r v $ P v W ' r r 1 Iv r YW 6 ex r P 1 W i c 1 W u Qc 1 W c U k c k | v r k x P $ v 1 r c YW 6 ex r P 1 W k c k wv r k x P {
4 G
Let us rst consider . Note rst that is the fundamental solution of the heat equation. In addition, since and for the spatial variable the decay rate of integrand is using the integration by parts, we get
b
R "I 4
` a 4 G
E 4
FE
E 4
b v W
c 1W
U { c 1W
4 AP 4
Using
by , we have
V U U W
c 1W
R
9 '
} $ 1 $ 1
4 k x4 r kW $ yxGP c c 4
$ r P 1 W 6$ k
1r
4 k x4 r YkW $ exAP c c 4
r P 1 W $ k n
} $ 1 $ 1 r 9 c 1 W x r { } g$ 1 $ 1 1 } $ 1 { c U W
&{ r 5 9 } g$ 1 $ 1 U
and we used
g h
4 W
1 yW W {
4 U
4 V
} $ { c U 1 W g 1 U { 1 $ 1 c U W ' {
k c { x U c P k { c U W { k k $ { c U YW { x U 1 W k c k
4 W
} g$ { c { x U c
4 T
k P 1 uW { W |
x c 1W
x c 1W
Before estimating
, we note that, on
for any
on
i ap
9 @7
fdxTU c 4
p
c P 1 uW W k
9 7
edc
U W k
where we used
where
for
i 5 9
. If , then, using the relation (16), it will be reduced to . To be precise, using the relation ,
1 r v B$
$ r P 1 W $ c r k $ c k k x W ex P $ P ' 6$
4 A
1 1r
$ '
for
(28) 12
Q 1 v Iv 1 1 1 v Iv r YW $ ex r P 1 yc c 1W c k c k Wc U c v uW v $ p v | W x P {
4 G
4 A
4 vp v1 v
U
P Q 1 $ Iv Iv $ 1 v 1 1 1 1 1 v v Iv Iv P %c xv W v $ P ' | v p v $ P v |
! s
C @' 9 c 1W $ $ $ r $ 1 } g 1 $ 1 { c U 1 W c U 1 W { c U 1 W U $ 1 $ 1 { c U W U
{ { k c U c P k k c { cx U YW { k $ 1 k W c { x U Yc U W {
c 1W
i c 1W
i c 1 W ~c 1 W U
where is any number with . Before we analyze the second term to (15), we have the following relation: for ,
4 T
1 W yW $ U 1 W {
4 T p
4k
U c P 1 uW k W
c U 1W c W ' U {
c 1W
p q
U c 1 W
4 k x4 r kW $ yxGP c c 4
i cW
With the aid of the estimate (14) and the relations (27), the rst term, mated as follows.
p I 4 p I4
4 G
{ U k ` k ` { c 1W $ r kP 1 W x r c r k $ c k x W yx P ' 1 c W
$ r P 1 W k
pr
B$ 1 Iv $ 1 v
$ ' c r 1 W
q 1 rr $ Iv
| v $ c 1W
r $ P ' c 1W
$P ! i
Therefore, it sufces to estimate in the case of (17) of , let us denote, for simplicity,
C 0'
We rst estimate
by
, can be esti-
, thanks
(29) 13
p
1 c W U
R
$
4W 4
c U 1W {{ U { B$ 1
c U W ' { {
V U
c U { { k c U W { {
k 1W { cx U c P yW { k YW k { k { { x U c P c k k
4W 4
1 yW x U $ 1 W W { c
4 W
$ r } g 1 $ 1
{ c U k k { c U YW k k { c U YW
p I
B$ k 9 ' r c k x fP 1 W r h h 1 Iv r $ r i x U 1 U k k p k c YW ec P W Iv v 1 Iv r r YW $ c 1W c k v x P 1 W wv c k p k x P {
4 G
We note that the rst and second integrals except for the last one can be estimated as before, because the former has only tangential derivatives and the latter is the second derivative of the fundamental solution of the heat equation. Thus the details are omitted for those cases. Indeed the most difcult one is the last integral. Disregarding the three terms and only considering the last term, with the aid of the integration by parts, we have
4 G
r YW $ ex r P 1 W c k c k k x P
v Iv 1 r YW $ ex r kP 1 W i k c p v k cx P $ 1v Iv 1 v
4 G
where . Thus, it remains to estimate the rst term (13), (21), and (27), we have
where we used facts that is zero at and for the spatial variable the decay rate of integrand is at innity. Let us rst consider the second term. Using the estimate (13) and (21), and properties of (27), we have
. According to estimates 14
9 1 W 1W U '
$ 1 { c U 1 W { c U { c U W U $ c U 1 W { c U c P 1 uW k W { $ { c U 1 W { c U 1 W c U
U W
$ u { c U 1 W { c U c P 1 yW k W k
{ {
U U
$ 1 } $ 1 c 1W U c 1W c 1W
B$ 1 } $ 1 U w c 1 W i 9 $ $ r $ 1 } g 1 { c U 1 W } g 1 k } $ k 1 { c U YW P k
. Thus we have
$ r $ 1 } g 1
{ c U k W
$ 1 I
Using the fact that
i U
q I
Iv $ 1 P r k $ k c YW v k k $ P r k $ k c W
! i
c 1W
$ r } 1 $ 1
9 $ r } 1 $ 1 1 c W U Qc 1 W c 1 W $ 1 1 $ 1 {{ c U W U
where
and we used
. It remains to estimates
. Since
C 5'
Iv 1 1 U r k $ c YW v 1 1Iv ' c v r
k { c U YW k c 5
on
$ P 1 k U 9 YW k
V q
P 1 k
$ 1 $ P 1 k k 1W
5 9
Hence we get
where
for
i 5 9
$ 1 } g$ 1 c r 1 W
for
i 9
} g$ 1 $ 1 { c r 1 W
r
r
. Using 15 ,
16
Remark 2.4 Especially, in the case that , following the same procedure as the proof above, we can show that satises the following estimate, which is a little better than (25):
Remark 2.5 It is interesting to compare the estimates of and . With the aid of Lemma 2.1 and the Youngs inequality, can be estimated as follows: For any with ,
Although (25) is weaker than (31), it will be sufcient for our estimates of the Riesz potentials, which will be used next section for investigating the boundary regularity of the Navier-Stokes equations.
for any with (see theorem 3.1 in [18]). Our point-wise estimate of the Green tensor also gives the simple proof of the estimate (32).
r c yc $& # W x W u yc $& # W x W 2 G c
P C
c P W C } { g W G c %Tb } g c %G W g
Remark 2.6 In [18] Ukai proved the following for : For any with ,
estimates of a solution
} $ 1 1 { & { c U W c r 1 W r H
P j
u
u
xr
' S
9 X7
9 8i '
where
h 9 $ 1 g$ 1 r { { { r 1 t { c W
where we used
t 9 ' ' } g$ $ 1 { c 1 U W $ r } 1 $ 1 { r U r
.
{
r
(30)
(31)
in (4)
(32)
(33)
V V
r 1 wt r r W H 8 $& # % &
IG
{ y un m w&
"I
Proof. Without loss of generality, we assume that by translation. Using the Green tensor for the half-space, can be represented as follows: For each
h C r i r r j ii '
.
r i cr 1W c r r 1 W t U U c 1 W G f t c G zc gr r r 1 W v ' c r 1 W G GW f H G f gr e
hg C U
q # v c "$& w2 r 1 W ' k
1 Tb U 1 c r e%G H ic 1 i U 1 G W G `W 9 '
i c $& # W 82 |
b wG
v x v w
for all
9 j '
x a x y
To complete the proof of Main Theorem, we rst prove the following crucial Lemma.
with
for some
for all
, and
with
and
i 7 jC T
'
9 ' 1 e g xr
c { y w &n m W d 2 G
c o w &n m W 2 G
Then
for every
with
and
$&% #
is smooth and compactly supported in , there exists such that support of contained in . We recall rst estimate of (see Theorem 2.3).
is
IG
r f
f { y w xr c W
f c gr W
f
i cr 1W
r c 1W r U c
{ wn m &
Due to the global energy inequality (34), the last term is bounded by given data. Thus the claim that is bounded in is completed. To sum up, we have proved so far
{ z w n m y &
SI
is obvious that for every with and Therefore, it sufces to consider the rst term . In order to use we split into two parts. For simplicity, we denote and , respectively.
v G V V V v
BI
mW 2 r 1 d c W h P { o G
supported in
where we used
and
. In fact, , we get
are bounded in
because
and
is bounded in
where
. To complete our assertion, we as follows: . Since . According to this observation, for . The good point for the is bounded by and (38) (37) . , and , it 18 is
m
u X
f
H
i G x x g G x x czW i U d ' P f gr W c %G P H
f r f { f xr W c f f f xr f xr { W W 6 c c G
r c e zW `W c G
f
E e
9 $ ' $ f $ r f g P 1 B$ P 1 { P f xr f xr W W c Q c %G H f } $ $ 1 $ 1 f P P { P f gr f gr W W c c G H $ P $ $ } $ X 9 ' P 1 B$ B P 1 } $ P $ 1 $ P 1 f P f xr f xr { i f W W c c %G H f f xr f xr f xr r r 1 t r W b c 1 zW W c G c W c W G ` c H
r 1 t b c W f xr W G c
wYGW c
where
Since is an arbitrary number between and , we may specically choose Utilizing the inequality of the Riesz potential, it is easily checked that
f
c f xr W
f f y gr { y w xr w c W c W G y {y w i {y w w G wYW c G r
and
d u
f
y { y w x w X G u
g s zW c G q $& c l# W d f xr e
where
For an arbitrarily
Applying the inequality of the Riesz potential again, we have with , we take . Using , the right side of the above inequality is estimated as follows: as (40) (39) and 19 , .
E $
20
The last inequality comes from , which is equivalent to . Since and is compactly supported as mentioned earlier, we obtain (39), which automatically implies (40). We note that representation of in (38) can be rewritten as follows: where
is bounded in
pletes our arguments, at the beginning we decompose as a union of sectors, denoted by such that for and for a given which is sufciently small, we get for all . This is possible if is taken sufciently small. By repeating the procedure above at each , we obtain , which eventually leads to . This completes the proof. The easy consequence of the above Lemma is the following. Corollary 3.2 Let and (2) satises . Suppose that a weak solution
solving (1)
Then
$ P $ 1 g$ P 1 f P f r { f xr W { { G c %G H f gr f f xr r r 1 t r W W b c 1 ewYW W Gc G c %G ` c H
'
r gr e f
2l
for every
i 7 jC T
'
xr
for some
with
and
hg C U
h c r q &o ' r F P
q w2 r 1 W ' k $& # v c
y c w &n m W d 2 G
c o w &n m W 2 G
y c { w &n m W 2 G
for all of
, must be also in the same space . However, we still need the smallness of
c { y u&n m W d w
{y w y&n m
between
if
where
is the con-
The main point in this argument is that is an invertible operator on provided that is sufciently small because the operator norm of
czW G c { y w &n m W d i cr 1W
P 5P U W
'
r U c 1 W
y w G
"I
c G zW r c 1W
' $ U P
r c 1W
c G fW wYxr e P G c o w &n m W % y w G
' c r 1 zyzW W c GW cc G
P W
I i
c $
21
Now we are ready to prove the main theorem. The Proof of Main Theorem According to Lemma 3.1 and corollary 3.2, we know . Adopting the expression (38), we show that are H lder o smooth in for space and time variables. We consider rst . Using the estimate of
and
, we have
of is bounded by norm of , which is also nite due to the global energy inequality. Hence it remains to prove H lder continuity for o . For clarity, we recall , which is given as follow:
f gr r r 1 t v f xr r r t v W P %c 1 W
and
. Since
, we have
1 P 1
r u G c 1 W
2l
by
r x c 1W
ic
l@
as
for simplicity. The key fact is that . Hence any partial derivatives
Yy w
l ox
Dl
instead of modied
in
. Thus, a slightly
y l dw
Yy w n m &
' cr 1W
r 1 u W
. To avoid the technical difculties, we modied the in (37) and . To be precisely, only difference is that we take
continuous in
n m
&
r 1W
IG
V IG
@7f f xr e
r r a H
where satisfying
7 e ' o y&n m m w y w
and and
are sufciently large numbers (in fact, it is enough to choose ). This completes the proof.
r x P %c 1 W
f xr r r 1 t W
f xr e
are
22
It can be easily checked that the last integral is independent of and nite because and it is locally integrable and decay rate at innity is . To sum up, we have the following H lder estimate: o
Since it follows the same argument for the case of space variables, the detailed proof is omitted. Since is arbitrary between and , is and for space variable and time variable. Therefore, we conclude that is also H lder continuous with exponent o for every in . This completes the proof. Remark 3.3 Unlike in the interior case, H lder continuity does not imply smoothness in o the spatial variables at boundary, and therefore we leave it as an open problem. In [8], I constructed a weak solution of the Stokes system of which normal derivatives blow up at boundary at a certain time. ACKNOWLEDGEMENT This research was supported in part by NSF Grant No. DMS-9877055. The author expresses his deep gratitude to his advisor, Professor Vladimr Sver k for guidance and en a couragement.
P
} } g $ $ g $ $ P $ ( P P $ ( } P i I P y } $ $ } $ $ P $ ( P P B$ ( P } P
i P ' } {
c { { r W
i } 1 P 1 ' }
{
j t
} |
j G
r x c 1W j G
c r 1 W u
c r 1 W u
{y w n m &
r u P jc 1 W
r x P %c 1 W
% j sr % l V "' j
5 9
'
Let obtain
and
. Changing of variable
i w
} $ } g $ P $ 1 $ P 1 P $ 1 $ P P ' r } g $ $ } $ $ P 1 $ P 1 P 1 $ P 1 P
p q' %
yY
'
23
References
[1] S. AGMON , A. D OUGLIS & L. N IRENBERG Estimates near the boundary for solutions of elliptic partial differential equations satisfying general boundary conditions. II, Comm. Pure Appl. Math. 17 (1964), 3592. [2] L. C AFFARELLI , R. KOHN & L. N IRENBERG Partial regularity of suitable weak solutions of the Navier-Stokes equations, Comm. Pure Appl. Math. 35 (1982), 771 831. [3] M. C ANNONE , F. P LANCHON , & M. S CHONBEK Strong solutions to the incompressible Navier-Stokes equations in the half-space, Comm. in P.D.E. 25 (2000) 903-924. [4] H. C HOE Boundary regularity of weak solutions of the Navier-Stokes equations, Journal of differential equations 149, (1998), 211247. [5] E. B. FABES , B. F. J ONES & N. M. R IVIERE The initial value problem for the Navier-Stokes equations with data in , Arch. Rational Mech. Anal. 45, (1972), 222 240. [6] Y. G IGA Solutions for semilinear parabolic equations in and regularity of weak solutions of the Navier-Stokes system, J. Differential Equations 62, no. 2, (1986), 186 212. [7] E. H OPF Uber die Anfangswertaufgabe f r die hydrodynamischen Grundgleichungen u (German) , Math. Nachr. 4, (1951), 213231. [8] K. K ANG Unbounded normal derivative for the Stokes system near boundary, preprint. [9] O. A. L ADYZHENSKAYA , V. A. S OLONNIKOV & N. N. U RALCEVA Linear and Quasilinear Equations of Parabolic type. Translations of Mathematical Monographs, 23, Amer. Math. Soc., Providence, R.I., 1968. [10] J. L ERAY Sur le mouvement dun liquide visqueux emplissant lespace, Acta Math. 63, (1934), 193248. [11] G. A. S EREGIN Some estimates near the boundary for solutions to the nonstationary linearized Navier-Stokes equations, Zap. Nauchn. Sem. S.-Peterburg. Otdel. Mat. Inst. Steklov. (POMI), 271 (2000), 204223. [12] J. S ERRIN On the interior regularity of weak solutions of the Navier-Stokes equations, Arch. Rational Mech. Anal. 9, (1962), 187195. [13] V. A. S OLONNIKOV Estimates of solutions of nonstationary linearized systems of Navier-Stokes equations, Trudy Mat. Inst. Steklov 70, (1964), 213317. (in English:A.M.S. Translations, Series II 75, 1117).
24
[14] E. M. S TEIN Singular integrals and differentiability properties of functions, Princeton University Press, 1970. [15] M. S TRUWE On partial regularity results for the Navier-Stokes equations, Comm. Pure Appl. Math. 41, (1988), 437458. [16] S.TAKAHASHI On a regularity criterion up to the boundary for weak solutions of the Navier-Stokes equations, Comm. Partial Differential Equations 17, (1992), 261285. [17] R. T EMAN Navier-Stokes equations. Theory and numerical analysis, North Holland, Amsterdam and New York, 1977.
Kyungkeun Kang School of Mathematics University of Minnesota Minneapolis, MN 55455 USA Email:kkang@math.umn.edu
$& %#
[18] S. U KAI A solution formula for the Stokes equation in 40, (1987), 611621.