• Aucun résultat trouvé

The stationary three-dimensional Navier-Stokes Equations with a non-zero constant velocity at infinity

N/A
N/A
Protected

Academic year: 2021

Partager "The stationary three-dimensional Navier-Stokes Equations with a non-zero constant velocity at infinity"

Copied!
25
0
0

Texte intégral

(1)

HAL Id: hal-00206208

https://hal.archives-ouvertes.fr/hal-00206208

Preprint submitted on 16 Jan 2008

HAL is a multi-disciplinary open access archive for the deposit and dissemination of sci- entific research documents, whether they are pub- lished or not. The documents may come from teaching and research institutions in France or

L’archive ouverte pluridisciplinaire HAL, est destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des établissements d’enseignement et de recherche français ou étrangers, des laboratoires

The stationary three-dimensional Navier-Stokes Equations with a non-zero constant velocity at infinity

Chérif Amrouche, Huy Hoang Nguyen

To cite this version:

Chérif Amrouche, Huy Hoang Nguyen. The stationary three-dimensional Navier-Stokes Equations

with a non-zero constant velocity at infinity. 2008. �hal-00206208�

(2)

with a non-zero onstant veloity at innity

Chérif AMROUCHE

and Huy Hoang NGUYEN

Laboratoire deMathématiques Appliquées

CNRSUMR5142

Université dePauetdesPaysde l'Adour

IPRA-Avenuedel'Université 64013Pau,Frane

herif.amrouheuniv-pau.fr

huy-hoang.nguyenetud.univ-pau.fr

Abstrat- Thispaperisdevoted to somemathematialquestions related

tothe3-dimensionalstationaryNavier-Stokes. Ourapproahisbasedonaom-

binationofpropertiesofOseen problemsin

R 3

.

Keywords: Navier-Stokesequations;Oseenequations;weightedSobolevspaes;

uidmehanis.

AMSlass: 35Q30,76D03,76D05,76D07

1 Introdution

Let

be a bounded open region of

R 3

, not neessarily onneted, with a Lipshitz-ontinuous boundary and let

be the omplement of

. We sup-

posethat

hasanitenumberofonnetedomponentsandeahonneted

omponenthasaonnetedboundary,sothat

isonneted. Theproblemon-

siststhen in ndingaveloityeld u

= (u 1 , u 2 , u 3 )

andthe pressure

π

satisfy

theNavier-Stokessystem:

(N S )

 

 

 

 

−ν∆

u

+

u

.∇

u

+ ∇π =

f

in Ω,

div

u

= 0 in Ω,

u

= 0 on ∂Ω,

u

u

at infinity,

where

ν > 0

, f and u

∈ R 3

are respetively the visosity of the uid, the external foreeld ating onthe uidand agiven onstantvetor. Thethird

equationofthesystemstatesthat theuidadheresatthesurfaeofthebody,

whih is the ommon no-slip ondition. For the last equation, we have two

dierentasesonerningthebehaviorofuatinnity. Ifu

= 0

,theowisat

restatinnityandintheremainingase,ifu

6= 0

,theowispastatinnity.

In this paper, we areinterestedin onsidering the ase

Ω = R 3

and u

6= 0

.

Ourpurposeistostudysomeregularitypropertiesoftheweaksolutionstothe

problem

(N S )

.

(3)

aboutweaksolutions,weightedSobolevspaesandsomeresultsofOseensystem

in weghted Sobolev spaes. In Setion 2, a result about existene of weak

solutions for the problem

(N S)

will be presented. In next setions, we shall

obtain some regularity properties of the weak solution u and the assoiated

pressure

π

. We shall also onsider the identity energy in the last setion. In

thispaper,weuseboldtypeharaterstodenotevetordistributionsorspaes

ofvetordistributions with3omponentsand

C > 0

usuallydenotes ageneri

onstantthevalueofwhih mayhangefrom linetoline.

Nowwereallthemainnotationsandresults,onerningtheweightedSobolev

spaes,whihweshalluselateron.

We dene

D(Ω)

to be thelinear spaeof innite dierentiablefuntions with ompat support on

. Now, let

D

(Ω)

denote thedual spae of

D(Ω)

, often

alled the spae of distributions on

. We denote by

h., .i

the duality pairing

between

D(Ω)

and

D(Ω)

. Remarkthatwhenf isaloallyintegrablefuntion,

thenf anbeidentiedwithadistributionby

h

f

, ϕi = Z

f

(

x

) .ϕ(

x

) d

x

.

Given aBanah spae

B

, with dual spae

B

and a losed subspae

X

of

B

,

wedenote by

B

⊥ X

(or moresimply

X

, ifthere is noambiguityasto the

dualityprodut)thesubspaeof

B

orthogonalto

X

, i.e.

B

⊥ X = X

= {f ∈ B

|∀ v ∈ X, < f, v >= 0} = (B/X)

.

The spae

X

is also alled thepolar spaeof

X

in

B

. In1933, Jean Leray

[13℄whointroduedtheoneptoftheweaksolution:

Denition 1.1. Aweaksolutionto the problem(

N S

)is aeld u

∈ H 1

loc (Ω)

vanishingon

∂Ω

,with

u

∈ L 2 (Ω)

, divu

= 0

in

and

lim

|x|→∞

Z

S

|

u

(

x

u

)| =

|

u

(

x

) −

u

| = 0

where

S

istheunitsphereof

R 3

suhthatforall

ϕ ∈ V (Ω) = {

v

∈ D (Ω), div

v

= 0}

:

ν Z

u

. ∇ϕ d

x

+ Z

u

. ∇

u

. ϕ d

x

= h

f

, ϕi .

Atypialpointin

R 3

isdenotedbyx

= (x 1 , x 2 , x 3 )

anditsnormisgivenby

|

x

| = (x 2 1 + x 2 2 + x 2 3 )

12

.

Wedene theweightfuntion

ρ(

x

) = (1+ |

x

| 2 )

12. For

eah

p ∈ R

and

1 < p < ∞

,theonjugateexponent

p

isgivenbytherelation

1

p + 1

p

= 1

. With

α ∈ R

and

m ∈ N

, weset

k = k(m, p, α) =

 

 

−1, if 3

p + α 6∈ {1, ..., m}, m − 3

p − α, if 3

p + α ∈ {1, ..., m},

andweintroduethedenition oftheweightedSobolevspaes.

Denition1.2. Let

beeither anexteriordomainor

Ω = R 3

. Then,

W α m,p (Ω) = {u ∈ D

(Ω); ∀ λ ∈ N 3 ,

0 ≤ |λ| ≤ k, ρ α−m+|λ| (ln(1 + ρ))

−1

λ u ∈ L p (Ω),

k + 1 ≤ |λ| ≤ m, ρ α−m+|λ|λ u ∈ L p (Ω)}.

(4)

k u k W

αm,p

(Ω) = ( X

0≤|λ|≤k

||ρ α−m+|λ| (ln(1 + ρ))

−1

λ u|| p L

p

(Ω) +

+ X

k+1≤|λ|≤m

||ρ α−m+|λ|λ u|| p L

p

(Ω) ) 1/p .

Wealsodene thesemi-norm:

|u| W

αm,p

(Ω) =

 X

|λ|=m

||ρ αλ u|| p L

p

(Ω)

1/p

.

We note that the logarithmi weight only appears for the ase

3/p + α ∈ {1, ..., m}

andalltheloalpropertiesof

W α m,p (Ω)

oinidewiththoseofthelas-

sialSobolevspae

W m,p (Ω)

. Weset

W

m, p α (Ω) = D(Ω) W

m, p α

(Ω)

andwedenote

thedualspaeof

W m, p α (Ω)

by

W

−α−m,p

(Ω)

,whihisthespaeofdistributions.

When

Ω = R 3

, we have

W α m,p ( R 3 ) = W

m, p α ( R 3 )

. If

3/p + α 6∈ {1, ..., m}

, we

havethealgebraiandtopologialimbeddings

W α m,p (Ω) ֒ → W α−1 m−1,p (Ω) ֒ → ... ֒ → W α−m 0,p (Ω).

Forall

λ ∈ N n

with

|λ| ≥ 0

,themapping

u ∈ W α,β m,p (Ω) → ∂ λ u ∈ W α,β m−|λ|,p (Ω)

is ontinuous. Moreover,if

3

p + α 6∈ {1, ..., m}

, then forany

γ

in

R

suh that

3

p + α − γ 6∈ {1, ..., m}

themapping

u → ρ γ u

is an isomorphismof

W α m,p (Ω)

onto

W α−γ m,p (Ω)

. Note that ifweonly suppose

3

p + α 6∈ {1, ..., m}

,themapping

isontinuous.

We denote by

[q]

the integer part of

q

. For any

k ∈ N

,

P k

(respetively,

P

k

)standsforthespae ofpolynomials(respetively,harmonipolynomials) ofdegree

≤ k

. If

k

isstritlynegativeinteger,wesetbyonvention

P k = {0}.

Let

k

beaninteger,then

P k

isinludedin

W α m,p (Ω)

with

k =

 

 

m − 3 p + α

, if 3

p + α 6∈ Z

, m − 3

p − α − 1, otherwise.

Weintroduethespae

W f 0 1,p (Ω) =

u ∈ W 0 1,p (Ω), ∂u

∂x 1

∈ W 0

−1,p

(Ω)

whihisaBanahspaeequippedwiththefollowingnorm

||u|| W

f1,p

0

(Ω) = ||u|| W

0,p

−1

(Ω) + X 3 i=1

|| ∂u

∂x i

|| L

p

(Ω) + || ∂u

∂x 1

|| W

−1,p

0

(Ω) ,

if

p 6= 3,

(5)

||u||

f

W

1,3

0

(Ω) = ||(

ln

(1 + ρ))

−1

u|| W

0,3

−1

(Ω) + X 3 i=1

|| ∂u

∂x i

|| L

3

(Ω) + || ∂u

∂x 1

|| W

−1,3

0

(Ω) ,

and

W f 0

−1,p

( R 3 )

isitsdualspae. Thepreviousnormisequivalenttothenatural

one and it allows to prove the density of

D(Ω)

in

W f 0 1,p (Ω)

. This result is

announedin[7℄. Weintroduealsothespae

V

(Ω) = n

v

∈ W

1,2 0 (Ω),

divv

= 0

in

Ω o .

Inorder to understandbetter thethe onditionu

uat innityof the

Navier-Stokessystem,weintrodueafollowinglemma(f[8℄):

Lemma1.3. Assume

1 < p < 3

and

u ∈ D

( R 3 )

suhthat

∇u ∈ L p ( R 3 )

. Then

thereexistsauniqueonstant

u

∈ R

suhthat

u − u

∈ W 0 1,p ( R 3 )

,where

u

isdenedby

u

= lim

|x|→∞

1 ω

Z

S

u(σ(|x|)) dσ

where S is the unit sphere of

R 3

and

ω

is the area of

S

. Moreover, we have

u − u

∈ L

3−p3p

( R 3 )

with the estimate

||u − u

||

L

3p

3−p

(

R3

) ≤ C||∇u||

Lp

(

R3

) ,

(1.1)

lim

|x|→∞

Z

S

|u(σ|x|) − u

|dσ = lim

|x|→∞

Z

S

|u(σ|x|) − u

| p dσ = 0

(1.2)

and

Z

S

|u(rσ) − u

| p dσ ≤ Cr p−3 Z

{x∈R3

,|x|>r}

|∇u| p dx.

(1.3)

ReallalsothefollowingSobolevembeddings

W 0 1,p ( R 3 ) ֒ → L p∗ ( R 3 )

where

p∗ = 3p

3 − p

and

1 < p < 3, W 0 1,3 ( R 3 ) ֒ → V M O( R 3 )

where

V M O( R 3 ) = D( R 3 )

||.||BM O

.

Here,

BM O

isthespaeofloallyintegrablefuntionsin

R 3

andsuhthat,on allubes

Q

,

|| f || BMO = sup

Q

1

|Q|

Z

Q

|f (x) − f (Q)|dx < ∞.

Note alsothatif

∇u ∈ L p

with

p > 3

and

u ∈ L r ( R 3 )

forsome

r ≥ 1

,then we

have

u ∈ L

( R 3 )

.

If

isanexteriordomain, wehaveaorollaryasfollows:

Corollary 1.4. Let

Ω ⊂ R 3

be an exterior domain. Assume

1 < p < 3

and

u ∈ D

(Ω)

suhthat

∇u ∈ L p (Ω)

. Then thereexistsaunique onstant

u

∈ R

suhthat

u − u

∈ W 0 1,p (Ω)

andwehave the properties (1.1)-(1.3).

Proof. Let

u ∈ D

(Ω)

suh that

∇u ∈ L p (Ω)

. Then, therestritionof

u

to

Ω R

with asuientlylarge

R

satisfy

u ∈ D

(Ω R )

and

∇u ∈ L p (Ω R ).

Therefore,

wehave

u ∈ W 1,p (Ω R )

and

u| ∂B

R

∈ W 1−1/p,p (∂Ω R )

(see Proposition2.10[4℄).

(6)

Thenthereexists

u 0 ∈ W 1,p (Ω R )

suhthat

u 0 = u

on

Γ

and

u 0 = 0

on

∂B R .

We

extend

u 0

byzerooutside

B R

anddenote

u

0

theextendedfuntionthatbelongs

to thelassialSobolevspae

W 1,p (Ω)

and hasompat support in

Ω R

. Note

that

v = u− u

0

, then

∇v ∈ L p (Ω)

and

v = 0

on

Γ

. Weset that

v = v

in

and

v = 0

outside

Ω.

Thenweandeduethat

v∈ L p ( R 3 )

. Thereforethereexists

auniqueonstant

u

suhthat

v −v

∈ W 0 1,p ( R 3 )

,or

u− u

0 −v

∈ W 0 1,p ( R 3 )

.

Then

u − v

∈ W 0 1,p (Ω)

.

NowweshallintroduethefollowinglemmabyombiningaresultofBabenko

(1973,Proposition3)withTheoremII.5.1[11℄. Theproofofthislemmaanbe

foundin [11℄.

Lemma 1.5. Let

Ω ⊂ R 3

bea Lipshitzexterior domainor

Ω = R 3

. Assume that

u ∈ W 0 1,2 (Ω) and ∂u

∂x 1

∈ L q (Ω) where 1 < q < 2.

Then

u ∈ L 3q (Ω)

andthe following inequalityholds:

||u|| L

3q

(Ω) ≤ C(|| ∂u

∂x 1

|| L

q

(Ω) + ||∇u|| L

2

(Ω) ).

Thenextlemmagivesananotherversionofthisresult.

Lemma1.6. Let

1 < p < 3

. Assumethat

u ∈ f W 0 1,p ( R 3 )

. Then

u ∈ L

44pp

( R 3 ) ∩ L

33pp

( R 3 )

andfollowing inequality holds:

||u||

L

4p

4−p

(

R3

) + ||u||

L

3p

3−p

(

R3

) ≤ C||u||

f

W

1,p

0

(

R3

) .

(1.4)

Proof. We already showed that if

u ∈ W 0 1,p ( R 3 )

with

1 < p < 3

, then

u ∈ L

33pp

( R 3 )

satisfying

||u||

L

3p

3−p

(

R3

) ≤ C||∇u|| L

p

(

R3

) .

Weknowthat

D( R 3 )

isdensein

W 0 1,p ( R 3 )

,thenthereexistsasequene

(ϕ k ) k∈

N

∈ D( R 3 )

whih onverges towards

1

in

W 0 1,p

( R 3 ).

By hypothesis, we dedue

∆u ∈ W 0

−1,p

( R 3 )

. Then, wehave

h∆u, 1i W

−1,p

0

(

R3

)×W

01,p′

(

R3

) = lim

k→+∞ h∆u, ϕ k i W

−1,p

0

(

R3

)×W

01,p′

(

R3

)

= − lim

k→+∞ h∇u, ∇ϕ k i L

p

(

R3

)×L

p′

(

R3

) = 0.

Analogously, sine

D( R 3 )

is dense in

W f 0 1,p ( R 3 )

(see [7℄), then we an dedue

that

∂u

∂x 1

, 1

W

0−1,p

(

R3

)×W

01,p′

(

R3

)

= 0.

Weset

−∆u + ∂u

∂x 1

= f.

(1.5)

Thenbyhypothesisand[5℄,wehave

f ∈ W 0

−1,p

( R 3 )

satisfyingtheompatibility onditionasfollows

hf, 1i W

−1,p

0

(

R3

)×W

01,p

(

R3

) = 0.

(7)

−∆w + ∂w

∂x 1

= f in R 3

(1.6)

has a unique solution

w ∈ L

33pp

( R 3 ) ∩ L

44pp

( R 3 )

suh that

∇w ∈ L p ( R 3 )

,

∂w

∂x 1

∈ W 0

−1,p

( R 3 )

alsosatisfying

||w||

L

3p

3−p

(

R3

) + ||w||

L

4p

4−p

(

R3

) + ||∇w||

Lp

(

R3

) + ∂w

∂x 1

W

0−1,p

(

R3

)

≤ C|| f || W

−1,p

0

(

R3

) .

(1.7)

Weset

z = u−w

. Subtrating(1.5)to(1.6),weget

−∆z + ∂z

∂x 1

= 0

in

R 3

. Sine

z ∈ L 3p/(3−p) ( R 3 )

,then,from Lemma4.1[8℄,wededuethat

z

isapolynomial

andthen

z = 0

. From(1.7),wehave(1.4). Theproofisomplete.

AnalogouslyasinLemma1.6, itiseasyto deduethefollowing.

Lemma 1.7. Let

1 < p < 2

. Assumethat

u ∈ W 0 2,p ( R 3 )

and

∂u

∂x 1

∈ L p ( R 3 )

.

Then we have

u ∈ L

2−p2p

( R 3 ) ∩ L

3−2p3p

( R 3 )

if

1 < p < 3/2

and

u ∈ L s ( R 3 )

for

all

s ≥ 2p

2 − p

if

3/2 ≤ p < 2

.

Denition 1.8. Let

1 < p < ∞

. Let

γ, δ ∈ R

besuh that

γ ∈ [3, 4]

,

γ > p

,

δ ∈ [ 3 2 , 2], δ > p

. Wedenetworeals

r = r(p, γ)

and

s = s(p, δ)

asfollow

1 r = 1

p − 1

γ and 1

s = 1 p − 1

δ .

Remark 1.9. FromDenition1.8, weandeduethat

i) If

1 < p < 3

,then

4p

4 − p ≤ r ≤ 3p 3 − p

,

ii) If

3 ≤ p < 4

,then

4p

4 − p ≤ r < ∞,

iii)If

1 < p < 3/2

,then

2p

2 − p ≤ s ≤ 3p 3 − 2p

,

iv)If

3/2 ≤ p < 2

,then

2p

2 − p ≤ s < ∞

.

Finally, we introdue thepropertiesonerningthe Oseenequations whih

will be useful in the next parts. We onsider the non homogeneous Oseen

problem : givenavetoreld fand afuntion

g

, welook forasolution

(

u

, π)

tothesystem

(OS)

 

−∆

u

+ ∂

u

∂x 1

+ ∇π =

f

in R 3 , div

u

= g in R 3 .

Theorem1.10. [7℄LetrandsbethenumbersgiveninDenition 1.8. Assume

(

f

, g) ∈ L p ( R 3 ) × W f 0 1,p ( R 3 )

.

(i) If

1 < p < 2

, thenProblem

(OS)

has a uniquesolution

(

u

, π) ∈ L s ( R 3 ) ×

(8)

W 0 1,p ( R 3 )

suhthat

u

∈ L r ( R 3 )

,

2

u

∈ L p ( R 3 )

and

u

∂x 1

∈ L p ( R 3 )

. More-

over, the following estimate holds

||

u

||

Ls

(

R3

) + ||∇

u

||

Lr

(

R3

) + ||∇ 2

u

||

Lp

(

R3

) + || ∂

u

∂x 1

||

Lp

(

R3

) + ||π|| W

1,p

0

(

R3

)

≤ C(||

f

||

Lp

(

R3

) + ||g|| W

f1,p

0

(

R3

) ).

(ii) If

2 ≤ p < 3

, then Problem

(OS)

has a solution

(

u

, π) ∈ W 1,r

0 ( R 3 ) × W 0 1,p ( R 3 )

,uniqueuptoanelementof

N 0

,suhthat

2

u

∈ L p ( R 3 )

and

u

∂x 1

∈ L p ( R 3 )

alsosatisfying

K∈ inf

R3

||

u

+ K ||

W1,r

0

(

R3

) + ||∇ 2

u

||

Lp

(

R3

) + || ∂

u

∂x 1

||

Lp

(

R3

) + ||π|| W

1,p

0

(

R3

)

≤ C(||

f

||

Lp

(

R3

) + ||g|| W

f1,p

0

(

R3

) ).

(iii)If

p ≥ 3

,thenProblem

(OS )

hasasolution

(

u

, π) ∈ W 2,r

0 ( R 3 ) ×W 0 1,p ( R 3 )

,

uniqueuptoan elementof

N 1

,suhthat

u

∂x 1

∈ L p ( R 3 )

. Moreover,wehave

(

λ

,µ)∈ inf

N1

(||

u

+ λ||

W2,p

0

(

R3

) + ||π + µ|| W

1,p

0

(

R3

) ) + || ∂

u

∂x 1

||

Lp

(

R3

)

≤ C(||

f

||

Lp

(

R3

) + ||g||

f

W

1,p 0

(

R3

) ).

Theorem1.11. [7℄ Let

r

bethe number given inDenition 1.8. Assumethat

f

∈ W

−1,p

0 ( R 3 )

andsatisesthe ompatibility ondition

∀λ ∈ P [1−3/p

] , h

f

, λi

W−1,p

0

(

R3

W1,p′

0

(

R3

) = 0.

Let

g ∈ L p ( R 3 )

suhthat

∂g

∂x 1

∈ W 0

−2,p

( R 3 )

,satisestheompatibilityondition

∀λ ∈ P [2−3/p

] , ∂g

∂x 1

, λ

W

0−2,p

(

R3

)×W

02,p′

(

R3

)

= 0.

(i) If

1 < p < 4

, thenthe Oseen system

(OS )

has a unique solution

(

u

, π) ∈ L r ( R 3 )× L p ( R 3 )

suhthat

u

∈ L p ( R 3 )

and

u

∂x 1

∈ W

−1,p

0 ( R 3 )

. Moreover,the

following estimateholds

||

u

||

Lr

(

R3

) + ||∇

u

||

Lp

(

R3

) + || ∂

u

∂x 1

||

W−1,p

0

(

R3

) + ||π|| L

p

(

R3

)

≤ C(||

f

||

W−1,p

0

(

R3

) + ||g|| L

p

(

R3

) + || ∂g

∂x 1

|| W

−2,p

0

(

R3

) ).

(ii) If

p ≥ 4

, then the Oseen system

(OS)

has a unique solution

(

u

, π) ∈ f

W 1,p

0 ( R 3 ) × L p ( R 3 )

, uniqueup to an element of

N 0

. Moreover, the following estimate holds

K

inf

∈R3

||

u

+ K ||

Wf1,p

0

(

R3

) + ||π|| L

p

(

R3

)

≤ C(||

f

||

W−1,p

0

(

R3

) + ||g|| L

p

(

R3

) + || ∂g

∂x 1

|| W

−2,p 0

(

R3

) ).

(9)

spaes

WeshallonsidertheNavier-Stokesproblemin

R 3

:

(N S )

 

 

−ν ∆

u

+

u

.∇

u

+ ∇π =

f

in R 3 ,

div

u

= 0 in R 3 ,

u

−→

u

if |x| → ∞,

where uis a onstant vetor in

R 3

. Without lossof generality, we anset u

= λ

e

1

with e

1 = (1, 0, 0)

and

λ ≥ 0.

From now on, we onsider the ase

of axed

λ > 0

. First, weprovetheexisteneof weaksolutionsand then, we

shalltheregularityofthesesolutionsindimention3. Weonsiderthefollowing

lemma.

Lemma 2.1. If

f ∈ W 0

−1,2

( R 3 )

, then there exists F

∈ L 2 ( R 3 )

suh that

f = div

Fin

R 3

withthe estimate

||

F

||

L2

(

R3

) ≤ C||f || W

−1,2

0

(

R3

) .

(2.1)

Additionallysupposethat

f ∈ W 0

−1,p

( R 3 )

,andfurthermoreassumethat

hf, 1i = 0

if

p ≤ 3

2

,thenF

∈ L p ( R 3 )

andwehave theestimate

||

F

||

Lp

(

R3

) ≤ C

||f || W

−1,p

0

(

R3

) .

(2.2)

Proof. If

f ∈ W 0

−1,2

( R 3 )

, from Theorem 9.5 [5℄, there exists a unique

z ∈ W 0 1,2 ( R 3 )

suhthat

∆z = f

in

R 3

and

||z|| W

1,2

0

(

R3

) ≤ C||f || W

−1,2

0

(

R3

) .

We set that F

= ∇z

, but

z ∈ W 0 1,2 ( R 3 )

, from Proposition 9.2 [5℄, we have F

∈ L 2 ( R 3 )

and (2.1). Moreover,if

f ∈ W 0

−1,p

( R 3 )

then there existsaunique

h ∈ W 0 1,p ( R 3 )/ P

[1−

3p

]

suhthat

f = ∆h

in

R 3

and

||h|| W

1,p

0

(

R3

)/

P

[1−3

p]

≤ C

||f || W

−1,p 0

(

R3

) .

Then

∇(z − h)

isharmoniin

L 2 ( R 3 ) + L p ( R 3 )

andonsequently,

∇z = ∇h

and

F

∈ L 2 ( R 3 ) ∩ L p ( R 3 )

withtheestimate(2.2).

We now return to the question of the existene of weak solutions of the

Navier-StokesEquationsin

R 3

. Thenexttheorem iswell known,then wegive hereaskethoftheproof.

Theorem2.2. Givenaforef

∈ W

−1,2

0 ( R 3 )

,theproblem

(N S )

hasaweakso-

lutionusatisfyingu

u

∈ W 1,2

0 ( R 3 )

andthereexistsafuntion

π ∈ L 2 loc ( R 3 )

,

uniqueup toaonstant,suhthat

(

u

, π)

solves the problem

(N S )

inthe sense

of distributionsandwehavethe following estimation

||

u

u

||

W1,2

0

(

R3

) ≤ C||

f

||

W1,2

0

(

R3

) .

(2.3)

Références

Documents relatifs

The proof is based on the weighted estimates of both pressure and kinetic energy for the approximate system which result in some higher integrability of the density, and the method

As a simple consequence of the result on controllability, we show that the Cauchy problem for the 3D Navier–Stokes system has a unique strong solution for any initial function and

in Sobolev spaces of sufficiently high order, then we show the existence of a solution in a particular case and finally we solve the general case by using the

a problem the unknown angular velocity co for the appropriate coordinate system in which the solution is stationary can be determined by the equilibrium

NIRENBERG, Partial regularity of suitable weak solutions of the Navier-Stokes equations. FRASCA, Morrey spaces and Hardy-Littlewood maximal

— We proved in Section 2 of [7] a local existence and uniqueness theorem (Theorem 2.2) for solutions of the Navier- Stokes equations with initial data in a space 7-^1 ^2,&lt;5s ^

We consider a variational formulation of the three-dimensional Navier–Stokes equations with mixed boundary conditions and prove that the variational problem admits a solution

Taking the duality method introduced by Lions &amp; Magenes in [28] and Giga in [20] up again for open sets of class C ∞ (see also Necas [31] chapter 4 that consider the Hilbertian