A right inverse of the divergence for planar H\"older-$\alpha$ domains
Ricardo G. Dur\'an, Fernando L\'opez Garc\'ia

TL;DR
This paper constructs right inverses of the divergence operator for planar H"older-$eta$ domains, enabling solutions to the Stokes equations with weaker boundary conditions in weighted Sobolev spaces, especially for domains with cusps.
Contribution
It proves the existence of divergence right inverses in weighted spaces for planar H"older-$eta$ domains, extending classical results to non-Lipschitz domains with cusps.
Findings
Existence of divergence right inverses in weighted spaces for planar H"older-$eta$ domains.
Weak boundary conditions are shown to be equivalent to standard conditions in cusp domains.
Solutions to the Stokes equations are obtained in weighted Sobolev spaces with some $L^r$ integrability, $r<2$.
Abstract
If is a bounded domain, the existence of solutions of for with vanishing mean value, is a basic result in the analysis of the Stokes equations. In particular it allows to show the existence of a solution , where is the velocity and the pressure. It is known that the above mentioned result holds when is a Lipschitz domain and that it is not valid for arbitrary H\"older- domains. In this paper we prove that if is a planar simply connected H\"older- domain, there exist right inverses of the divergence which are continuous in appropriate weighted spaces, where the weights are powers of the distance to the boundary. Moreover, we show that the powers of the distance in the results obtained are optimal.…
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
Taxonomy
TopicsAdvanced Mathematical Modeling in Engineering · Analytic and geometric function theory
Ricardo G. Durán
Departamento de Matemática
Facultad de Ciencias Exactas y Naturales
Universidad de Buenos Aires
1428 Buenos Aires
Argentina.
and
Fernando López García
Departamento de Matemática
Facultad de Ciencias Exactas y Naturales
Universidad de Buenos Aires
1428 Buenos Aires
Argentina.
A right inverse of the divergence for
planar Hölder- domains
Ricardo G. Durán
Departamento de Matemática
Facultad de Ciencias Exactas y Naturales
Universidad de Buenos Aires
1428 Buenos Aires
Argentina.
and
Fernando López García
Departamento de Matemática
Facultad de Ciencias Exactas y Naturales
Universidad de Buenos Aires
1428 Buenos Aires
Argentina.
Abstract.
If is a bounded domain, the existence of solutions of for with vanishing mean value, is a basic result in the analysis of the Stokes equations. In particular it allows to show the existence of a solution , where is the velocity and the pressure.
It is known that the above mentioned result holds when is a Lipschitz domain and that it is not valid for arbitrary Hölder- domains.
In this paper we prove that if is a planar simply connected Hölder- domain, there exist right inverses of the divergence which are continuous in appropriate weighted spaces, where the weights are powers of the distance to the boundary. Moreover, we show that the powers of the distance in the results obtained are optimal.
In our results, the zero boundary condition is replaced by a weaker one. For the particular case of domains with an external cusp of power type, we prove that our weaker boundary condition is equivalent to the standard one. In this case we show the well posedness of the Stokes equations in appropriate weighted Sobolev spaces obtaining as a consequence the existence of a solution for some depending on the power of the cusp.
Key words and phrases:
Divergence Operator, planar Hölder- domains, Stokes equations
1991 Mathematics Subject Classification:
Primary: 26D10 , 35Q30 ; Secondary 76D03
1. Introduction
Let be a bounded open domain. We will use standard notations for Sobolev spaces and, for , will denote the subspace of functions in with vanishing mean value.
The existence of right inverses of the operator is a basic result for the theoretical and numerical analysis of the Stokes equations by variational methods (see for example [BS, BF, GR, T]). This result is also closely connected with the Korn inequality which is fundamental in the analysis of the elasticity equations (see [HP]).
Usually, the problem is stated as follows: for any there exists such that
[TABLE]
and
[TABLE]
where, here and throughout the paper, the letter denotes a generic constant.
This problem, as well as its generalization to the case, has been widely analyzed and several different arguments have been given to prove it under different assumptions on the domain. We refer the reader for example to [ASV, BB, BS, B, GR, DM2]. Recently, in [ADM], the existence of right inverses of the divergence acting on , , was proved for the so called John domains, which form a large class containing properly the Lipschitz domains. Moreover, for the particular case of planar simply connected domains, it is shown in [ADM] that being a John domain is also a necessary condition in the case . In particular, there exist bounded domains and values of for which continuous right inverses of the divergence do not exist.
Actually, this fact was previously well known, indeed, several arguments have been given to show it. For example, in the old paper [F1], Friedrichs proved that, for smooth planar domains, the -norm of the conjugate of a harmonic function (normalized in an appropriate way) is bounded by the -norm of times a constant depending only on the domain. Moreover, he showed that this inequality is not valid if the domain has an external cusp of quadratic type. It is easy to see that the Friedrichs inequality can be deduced from the existence of satisfying (1.1) and (1.2). Therefore, such a cannot exist for that kind of domains. More recently other examples have been given in [GG] and in an unpublished work of Gabriel Acosta. Acosta’s examples are very elementary and applies to external cusps of power type with any power and any . We also refer the reader to [D1] where a particular case has been reproduced and to [ADL].
In view of the above mentioned results, it seems natural to ask whether some weaker results can be proved for a more general class of domains. Moreover, if this is the case, can those results be applied to show the well posedness of the Stokes problem in appropriate Hilbert spaces?
In this paper we give some partial answers to these questions in the particular case of planar simply connected domains. We consider Hölder domains, with . We say that a domain belongs to this class if its boundary is locally the graph of a Hölder function. For these domains we prove the existence of solutions of (1.1) satisfying weaker estimates than (1.2) involving weighted norms where the weights are powers of the distance to the boundary. For general Hölder domains the zero boundary condition will be imposed in a weak way. Afterwards, in some particular examples, we will show that this weak boundary condition agrees with the usual one.
Our approach use some of the ideas of the papers [ADL, GK]. The existence of solutions of the divergence is derived from appropriate Korn type inequalities. The weighted Korn inequalities that we need are slight variants of those obtained in [ADL] but we include the proofs for the sake of completeness.
Although our arguments to derive the existence of right inverses of the divergence are two dimensional, we write the proofs of the Korn type inequalities in the general -dimensional case because they have interest in themselves.
The rest of the paper is organized as follows. In Section 2 we introduce some notations and prove the weighted Korn inequalities. Section 3 deals with our main results concerning the existence of right inverses of the divergence continuous in appropriate weighted norms for Hölder domains. In Section 4 we apply the results of the previous section for the particular case of domains having power type external cusps. We show that in this case our weak zero boundary condition agrees with the usual one. Also in this section we prove optimality of our results. In Section 5 we show how our results can be applied to prove the well posedness of the Stokes equations in appropriate Hilbert spaces. Recall that, if is the velocity and the pressure of a viscous incompressible fluid, the Stokes equations are given by
[TABLE]
The existence of solutions of (1.1) satisfying (1.2) can be used to prove that, for , there exists a unique solution of (1.3) and moreover,
[TABLE]
We will show that our right inverses of the divergence can be used to prove a slightly weaker result for some cuspidal domains. In particular, under some restriction on the power of the cusp, we will obtain the existence of a unique solution of the Stokes equations (1.3) satisfying
[TABLE]
for some which depends on the power of the cusp.
2. Preliminaries and Korn type inequalities
Let be a bounded open subset of and the distance of to the boundary . We will denote by the Banach space given by the norm
[TABLE]
and, analogously, will be the Banach space with norm
[TABLE]
Whenever we will call the subspace of formed by functions of vanishing mean value. Since no confusion is possible we will use the same notations for the norms of vector or tensor fields.
For a vector field defined in we denote by the jacobian matrix, namely, and by its symmetric part (i.e., the linear strain tensor associated with ), that is .
We start by giving a weighted Korn inequality for Hölder domains. The statement given in the following theorem is slightly stronger than the result in Theorem 3.1 of [ADL]. Therefore, we include the proof for the sake of completeness although the arguments are essentially those given in that reference. In particular we will make use of the following improved Poincaré inequality proved in [ADL, Theorem 2.1]. If is a Hölder domain, , a ball and is such that then, for and such that there exists a constant depending only on , and such that,
[TABLE]
Theorem 2.1**.**
Let be a Hölder domain, a ball and . Then, for the following inequality holds,
[TABLE]
where the constant depends only on , and .
Proof.
Following [KO], we can show that there exists such that
[TABLE]
and
[TABLE]
Now, let be such that . For define the linear functions
[TABLE]
and as the vector with components .
Then,
[TABLE]
and, integrating by parts and applying the Hölder inequality we obtain
[TABLE]
where is the dual exponent of .
Therefore, it follows from (2.4) that there exists a constant depending only on , and such that
[TABLE]
Let us now introduce
[TABLE]
Then, in view of the bounds (2.4) and (2.5), it only remains to estimate . But, from (2.3) and the fact that is linear we know that
[TABLE]
and consequently,
[TABLE]
But, if is a harmonic function in , the following estimate holds
[TABLE]
for all . Indeed, this estimate was proved in [D] (see also Lema 3.1 in [ADL], and [KO] for a different proof in the case and ).
Therefore, taking we obtain
[TABLE]
and using the well known identity
[TABLE]
we conclude that
[TABLE]
for any and .
Since (indeed, we have defined in order to have this property), it follows from the improved Poincaré inequality (2.2) that
[TABLE]
Therefore, using (2.6), we obtain
[TABLE]
concluding the proof. ∎
In the following corollary we give a weighted Korn inequality for Hölder domains which can be seen as a generalization of the so-called second case of Korn inequality. To state this inequality we need to introduce the space of infinitesimal rigid motions, namely,
[TABLE]
Corollary 2.1**.**
Let be a Hölder domain and . Then, for the following inequality holds,
[TABLE]
Proof.
Take and as in the previous theorem with . Define and defined by
[TABLE]
with
[TABLE]
It is easy to check that . Now, since , it follows from (2.2) (actually we are using only a weaker standard Poincaré inequality with weights) and Theorem 2.1 that
[TABLE]
and using now the Poincaré inequality in we have
[TABLE]
But,
[TABLE]
and therefore, the so-called second case of Korn inequality applied in gives
[TABLE]
Using this inequality in (2.8) and that we obtain
[TABLE]
which implies (2.7) because . ∎
Remark 2.1**.**
It is possible to prove the above corollary directly, i.e., without using the Korn inequality in the ball , by using a standard compactness argument. Indeed, assuming that (2.7) does not hold and using that is compactly embedded in for any (see [KuOp, Theorem 19.11]) and Theorem 2.1 one obtains a contradiction.
3. Right inverse of the divergence in Hölder domains
This section deals with solutions of divergence in planar simply connected Hölder domains. In what follows we restrict ourselves to the case .
For regular enough bounded domains (for example Lipschitz) it is known that, if , , there exists such that
[TABLE]
and
[TABLE]
where the constant depends only on and .
On the other hand, as we have mentioned in the introduction, it is known that for general Hölder domains this result is not valid. Our main goal is to prove a similar result for this kind of domains but using weighted norms.
We will use the following notation. For a scalar function we write and for a vector field , denotes the matrix which has as it rows. Furthermore, if , denotes the vector field with components obtained by taking the divergence of the rows of .
We will impose the boundary condition in a weak form. To explain this weak condition observe first that to solve the problem it is enough to find a solution of (3.1) such that the restriction to of both components of are constant (whenever the domain is such that this restriction makes sense). Of course, we should replace the estimate (3.2) by
[TABLE]
Afterwards, (3.2) would follow by applying the Poincaré inequality to the solution obtained by adding an appropriate constant vector field to in order to obtain the vanishing boundary condition.
Now, assume that is a Lipschitz domain. Then, if satisfies
[TABLE]
it follows by integration by parts that
[TABLE]
where indicates the tangential derivative of . Therefore and then the restriction of to is constant.
For a general domain the tangential derivative on the boundary might not even be defined and therefore (3.5) would not make sense. However, condition (3.4) is well defined in any domain and this is the condition that we will use. Therefore we introduce the space
[TABLE]
defined by
[TABLE]
and more generally, for any ,
[TABLE]
The proof of the following lemma uses ideas introduced in [GK] with different goals.
For and , denotes the subspace of symmetric tensors in .
Lemma 3.1**.**
Let be a Hölder domain and , with , such that . Then, there exists satisfying
[TABLE]
and
[TABLE]
Proof.
Let the subspace defined as
[TABLE]
Let us see that the application
[TABLE]
defines a continuous linear functional on .
First of all observe that is well defined. Indeed, it is enough to check that the expression on the right of (3.6) vanishes whenever . But, it is known that in that case and therefore
[TABLE]
Now, we want to show that is continuous on . Using again that if and applying Corollary 2.1 we have, for ,
[TABLE]
By the Hahn-Banach theorem the functional can be extended to and therefore, by the Riesz representation theorem, there exists such that
[TABLE]
and
[TABLE]
where depends on the constant in Corollary 2.1. In particular,
[TABLE]
for every . Then, we conclude the proof observing that, since is symmetric, we can replace in (3.7) by . ∎
It is a very well known result that a divergence free vector field is a rotational of a scalar function . Indeed, for smooth vector fields the proof is usually given at elementary courses on calculus in several variables. On the other hand, if the vector field is only in but is Lipstchiz, it is not difficult to see that the vector field can be extended to a divergence free vector field defined in and then, the existence of can be proved by using the Fourier transform. However, we need to use the existence of in the case where the domain and the vector field are both non-smooth. We have not been able to find a proof of this result in the literature and so we include the following lemma.
Lemma 3.2**.**
Let be a simply connected Hölder domain and . Given a vector field such that , there exists such that
[TABLE]
where is a constant depending only on .
Proof.
Take satisfying , where is the unit ball centered at the origin. For , define and, extending by zero to , .
Let be a sequence of Lipschitz simply connected open subsets of such that
[TABLE]
Using that the distance between and is greater than and , it is not difficult to see that in for every .
Then, since , there exists such that . Moreover, adding a constant we can take such that .
Now, by the Poincaré inequality we have, for any , there exists a constant depending only on such that
[TABLE]
for .
Then, there exists such that in and so in and consequently in .
Finally, using Theorem 2.1 of [ADL] we have
[TABLE]
and the Lemma is proved. ∎
We can now state and prove our results on solutions of the divergence on Hölder- domains. As we mentioned above, it is known that for this kind of domains a solution of (3.1) satisfying (3.2) does not exist in general. Therefore, it is natural to look for solutions of (3.1) satisfying a weaker estimate. There are two possibilities: to use a stronger norm on the right of (3.2) or a weaker norm on the left. We will prove both kind of results but, to avoid technical complications while presenting the arguments, we give first a particular case of our results and postpone the generalization.
Theorem 3.1**.**
Let be a bounded simply connected Hölder- domain, . Given , , there exists such that
[TABLE]
and
[TABLE]
Proof.
Take such that
[TABLE]
and
[TABLE]
The existence of such a is well known, for example, since no boundary condition on is required, we can extend by zero and take the solution of problem (3.1) and (3.2) in a ball containing .
To prove the theorem it is enough to show that there exists satisfying and such that
[TABLE]
and
[TABLE]
Indeed, in view of (3.9), will be the desired solution.
But, since has vanishing mean value, we know from Lemma 3.1 that there exists satisfying
[TABLE]
and
[TABLE]
Then,
[TABLE]
for every and therefore .
Now, from Lemma 3.2 we know that there exists such that
[TABLE]
We have to check that , but since is a symmetric tensor we have
[TABLE]
To conclude the proof observe that in view of (3.10), (3.12) and (3.13) we have (3.11) which together with (3.10) yields (3.8). ∎
Now, it is natural to ask whether part or all the weight in the estimate (3.8) can be moved to the right hand side. We will give a positive answer to this question. As we will show, the proof of this more general result is similar to that of Theorem 3.2 but it requires some non-trivial preliminary results. In particular, we will need an extra hypothesis on the domain.
Definition 3.1**.**
For , a compact is an m-set, if there exists a positive constant such that
[TABLE]
for every and , where is the -dimensional Hausdorff measure and is the ball with radius and center .
The reader who is not familiar with Hausdorff measure can think in the particular case that is a rectifiable curve in and . In that case is the length.
We are going to use that Calderón-Zygmund singular integral operators are continuous in weighted -norms, , for weights in the Muckenhaupt class . This is a well known result which can be seen for example in the book [S2].
We state and prove the following lemma in the general n-dimensional case since it does not make any difference with the particular case . Our lemma generalizes the results proved in [DST] for smooth domains. Since the proof is too technical we postpone it for an appendix and continue now with our main results. In what follows we consider the distance to the boundary of , defined for every and not only for .
Lemma 3.3**.**
Let be a bounded domain such that its boundary is an m-set. If , then belongs to the class .
Proof.
See Apendix. ∎
As a consequence we have the following result on weighted estimates for solutions of the divergence.
Lemma 3.4**.**
Let be a bounded domain such that its boundary is a 1-set. Given , , with there exists such that
[TABLE]
and
[TABLE]
Proof.
Extend by zero to . Then, it is well known that
[TABLE]
is a solution of . Moreover, it follows from the theory of singular integral operators (see for example [S2]) that, if ,
[TABLE]
But, since satisfies the hypothesis of Lemma 3.3 with and , and therefore is the desired solution. ∎
We can now give our more general result on solutions of the divergence.
Theorem 3.2**.**
Let be a bounded domain such that its boundary is a 1-set. Given , , if and , there exists such that
[TABLE]
and
[TABLE]
Proof.
Since , it follows from Lemma 3.4 that there exists such that
[TABLE]
and
[TABLE]
The rest of the proof follows as that of Theorem 3.1. Now we have to show that there exists satisfying and such that
[TABLE]
and
[TABLE]
The reader can easily check that the existence of follows by using Lemma 3.1 as in Theorem 3.2. ∎
4. Domains with external cusps
In this section we consider the particular case of the Hölder- domain defined as
[TABLE]
with .
We are going to show that in this case the weaker boundary condition imposed in Theorem 3.2 is equivalent to the standard one, i.e., that the solution of the divergence obtained in that theorem can be modified, by adding a constant vector field, to obtain a solution which vanishes on the boundary in the classic sense.
We will consider the particular case of our general Theorem 3.2. Extension of the arguments to other cases might be possible but it is not straightforward.
Theorem 4.1**.**
Let be the domain defined in (4.1) and . If then, given there exists such that
[TABLE]
and
[TABLE]
with a constant depending only on and .
Proof.
It is easy to see that satisfies the hypotheses of Theorem 3.2. Therefore, it follows from that theorem that there exists which verifies (4.2).
We are going to prove that, for any , there exists a constant such that
[TABLE]
Consequently, can be modified by adding a constant to each of its components to obtain the desired solution. Indeed, the estimate (4.3) will follow form (3.14) by the Poincaré inequality.
Given , let us show first that is constant on . From the definition of we have that
[TABLE]
Now, let be a point in different from the origin and an open ball centered in such that . Taking we have
[TABLE]
where indicates the tangential derivative of . Consequently in the distributional sense on and then, since is a connected set, we conclude that there exists a constant such that on . To simplify notation we assume in what follows that and so, we have to see that .
Now, let be such that
[TABLE]
We decompose as
[TABLE]
It is easy to see that where is the Lipschitz domain
[TABLE]
Thus, we can suppose that . Let now be a sequence satisfying in and let .
It is easy to check that, for ,
[TABLE]
Therefore, integrating and using the Hölder inequality we have
[TABLE]
Thus, using the continuity of the trace in the Lipschitz domain we have
[TABLE]
Now we will show that the sequence defined by
[TABLE]
where , converges to in . Moreover, it is easy to see that .
By symmetry we can assume that . Using the dominated convergence theorem we obtain
[TABLE]
On the other hand,
[TABLE]
and then,
[TABLE]
Thus, using again dominated convergence, it is easy to check that . So, it only remains to analyze .
Now, by the change of variables defined by and using (4.4) it follows that
[TABLE]
An analogous argument can be applied to prove that in .
Consequently, we conclude the proof by observing that belongs to . ∎
In the following theorem we show that the estimate (4.3) is optimal in the sense that it is not possible to improve the power of the distance in the right hand side. Recall that is the dual exponent of .
Theorem 4.2**.**
Let be the domain defined in (4.1). If admits a continuous right inverse for some then, .
Proof.
For define . Then, calling , we have
[TABLE]
and therefore, using that for , , we obtain
[TABLE]
but,
[TABLE]
[TABLE]
where we have used . Therefore,
[TABLE]
where and with constants in the equivalence independent of .
Now, let be a ball such that and such that . From our hypothesis we know that, if , there exists such that
[TABLE]
But, since ,
[TABLE]
and so,
[TABLE]
where we have used that because the support of is contained in . Then,
[TABLE]
Using again that the support of is at a positive distance from the boundary, together with (4.6), it follows that
[TABLE]
On the other hand,
[TABLE]
where for the last inequality we have used (4.7).
Therefore,
[TABLE]
But, an elementary computation shows that
[TABLE]
where and with constants in the equivalence independent of .
Thus, from (4.5), (4.8) and (4.9) we conclude that there exists a constant independent of such that
[TABLE]
therefore, and it follows immediately that . ∎
Remark 4.1**.**
If we put some part of the weight in the left hand side as in (3.14), it is possible to prove a more general result, namely, under some restriction on the exponents in the weights, the difference between the powers in the right and left sides cannot be less than (see [ADLg]).
5. An application to the Stokes equations
In this section we show how our results can be applied to the analysis of the Stokes equations when is the domain defined in (4.1).
We are going to use the well known theory developed by Brezzi (see for example [BF, D2, GR]) but modifying the usual Hilbert spaces and the bilinear form corresponding to the divergence free restriction in the weak formulation of the Stokes equations.
Theorem 5.1**.**
Let be the domain defined in (4.1) with . Then, if , there exists a unique weak solution of the Stokes equations (1.3). Moreover, there exists a constant C depending only on such that
[TABLE]
Proof.
Let us introduce the spaces
[TABLE]
which is a Hilbert space with the norm
[TABLE]
and
[TABLE]
Define the bilinear forms and by
[TABLE]
and
[TABLE]
We are going to show that the problem
[TABLE]
has a unique solution .
Using the Schwarz inequality it is easy to check that the bilinear forms and are continuous and, since , that the linear functional defined by the right hand side of (5.2) is continuous.
Let
[TABLE]
According to Brezzi’s theory it is enough to see that is coercive in and satisfies the inf-sup condition
[TABLE]
Since we can take in the equation and conclude that . Therefore, coerciveness of in follows from the Poincaré inequality.
On the other hand, from Theorem 4.1 we know that given there exists such that
[TABLE]
where is a positive constant which depends only on . Moreover, from the definition of the norm in it follows immediately that
[TABLE]
for another constant depending only on . Then,
[TABLE]
and therefore the inf-sup condition (5.4) is proved.
Summing up we have proved that the problem given in (5.2) and (5.3) has a unique solution . Moreover, it follows also from the general theory that there exists a constant depending only on such that
[TABLE]
Now, define . It easy to see that and moreover, it follows from (5.3) that and from (5.2) that verifies
[TABLE]
Therefore, since , is a solution of the Stokes equations (1.3) in the sense of distributions as we wanted to prove. Finally, since , (5.1) follows immediately from (5.5). ∎
We end this section with a corollary which gives an estimate for the pressure in a standard -norm.
Corollary 5.1**.**
Let be the domain defined in (4.1) with and be the solution of the Stokes equations (1.3). If and then . Moreover, there exists a constant C depending only on such that
[TABLE]
Proof.
We only have to prove that and that
[TABLE]
Observe that for any . Indeed, this follows easily by using that . Then, applying the Hölder inequality with exponent , we have
[TABLE]
but the integral in the right hand side is finite because . So and therefore, (5.6) follows immediately from (5.1). ∎
6. Appendix
To prove Lemma 3.3 we will work with Whitney decompositions. If is a compact non-empty subset of , then can be represented as a union of closed dyadic cubes with pairwise disjoint interior satisfying
[TABLE]
where the edge length of is . The decomposition (6.1) is called a Whitney decomposition of and the collection is called the generation of Whitney cubes. Furthermore, the Whitney cubes satisfy
[TABLE]
where denotes the distance of the cube to and the diameter of (see for example [S]).
For and , denotes the number of Whitney cubes of in the generation contained in .
Lemma 6.1**.**
Let be a compact m-set. Given and , there exists a constant depending only on such that
[TABLE]
Proof.
The idea is to use that the number of Whitney cubes of in the generation contained in a ball is essentially the number of balls of radius necessary to cover .
Let be a Whitney cube in the generation contained in . Then, it is easy to check that
[TABLE]
Suppose there exist balls with , for , satisfying the following properties
[TABLE]
Thus, if is a point satisfying we can conclude that there is , for some , such that
So, using that is a Whitney cube in the generation it follows that
[TABLE]
where is the diameter of . But, cannot contain more than a finite number of Whitney cubes . Then, by (6.2) it follows that
[TABLE]
Thus, to complete the proof we have to show (6.2). Let . For we define the numbers
[TABLE]
and
[TABLE]
Then, using that is an m-set we have
[TABLE]
Thus, using the definition of we obtain (6.2), concluding the proof. ∎
Before proving Lemma 3.3 let us recall the definition of the Muckenhaupt class . For a non-negative function is in if
[TABLE]
where the supremum is taken over all the balls .
Proof of Lemma 3.3: Let be a ball in , its radius and the distance of to .
If , given in we have Then,
[TABLE]
On the other hand, if , there exists such that . Then, without loss of generality, we can assume that is centered at a point of .
Now, from the Whitney decomposition of we have
[TABLE]
where the sum is taken over all Whitney cubes intersecting . There is no loss of generality in assuming that the Whitney cubes are contained in .
Observe that if is contained in then . We call the minimum satisfying this inequality. Then, it is easy to see that .
Now, using that for every and Lemma 6.1 we obtain
[TABLE]
Then, since , we obtain
[TABLE]
and therefore the Lemma is proved.
The reference list from the paper itself. Each links out to its DOI / PubMed record.
- 1[ADL] G. Acosta, R. G. Durán and A. Lombardi , Weighted Poincaré and Korn inequalities for Hölder α 𝛼 \alpha domains , Math. Meth. Appl. Sci. (MMAS) 29 (4), 387-400, 2006.
- 2[AD Lg] G. Acosta, R. G. Durán and F. López García , work in progress.
- 3[AADL] G. Acosta, María G. Armentano, R. G. Durán and A. Lombardi , Nonhomogeneous Neumann problem for the Poisson equation in domains with an external cusp , Journal of Mathematical Analysis and Applications 310 , pp 397-411, 2005.
- 4[ADM] G. Acosta, R. G. Durán and M. A. Muschietti , Solutions of the divergence operator on John Domains , Advances in Mathematics 206 (2), 373-401, 2006.
- 5[ASV] D. N. Arnold, L. R. Scott and M. Vogelius , Regular inversion of the divergence operator with Dirichlet boundary conditions on a polygon , Ann. Scuola Norm. Sup. Pisa Cl. Sci-Serie IV XV , pp. 169-192, 1988.
- 6[B] M. E. Bogovskii , Solution of the first boundary value problem for the equation of continuity of an incompressible medium , Soviet Math. Dokl. 20 , pp. 1094-1098, 1979.
- 7[BB] J. Bourgain and H. Brezis , On the equations d i v Y = f 𝑑 𝑖 𝑣 𝑌 𝑓 div Y\,=\,f and application to control of phases , Journal of the American Mathematical Society 16 (2), pp. 393-426, 2002.
- 8[BS] S. C. Brenner and L. R. Scott , The Mathematical Theory of Finite Element Methods, Springer-Verlag, Berlin, 1994.
