Abstract
Let be a nonempty closed set such that for some and the -Hausdorff content for all cubes with centre and edge length . For each we give an intrinsic characterization of the trace space of the Sobolev space to the set . Furthermore, we prove the existence of a bounded linear operator such that is the right inverse to the standard trace operator. Our results extend those available in the case for Ahlfors-regular sets .
Bibliography: 36 titles.
Export citation and abstract BibTeX RIS
The research of S. K. Vodopyanov was carried out within the framework of a state assignment of the Ministry of Education and Science of the Russian Federation for the Institute of Mathematics of the Siberian Branch of the Russian Academy of Sciences (project no. 0314-2019-0006). |
§ 1. Introduction
Given let denote the linear space of all functions on with continuous partial derivatives up to order equipped with the standard seminorm. The problem considered by Whitney in 1934 in his famous papers [1] and [2] reads as follows.
Classical Whitney Extension Problem. Let and let be an arbitrary nonempty subset of . How can we decide whether a given function extends to a -function?
Whitney [2] solved this problem completely only in the case . Furthermore, he gave a solution of the analogous problem in the context of the Lipschitz spaces , (see [1]). After [1] and [2], great progress was made by many authors (see [3]–[5] and also the references there). Fefferman gave a complete solution (that is, for all ) of the Classical Whitney Extension Problem [6]–[9] only recently.
Recall that, according to the classical Sobolev embedding theorem (for example, see [10], Ch. I, §1.8.2), in the case , and , for every there exists a representative . This fact enables one to identify each element with its unique continuous representative. This implies that has a well-defined restriction to any given nonempty subset of . As a result, in the case we can consider the analogue of the Classical Whitney Extension Problem, where is replaced by . There is extensive literature devoted to these problems (see [11]–[19]).
In the case when , and , functions in the space do not (in general) have continuous representatives (see Ch. 5, §6 in [20]). Nevertheless, every has a 'sufficiently nice' representative which has a well-defined trace on each set with positive -capacity. Unfortunately, by contrast with the case , in the case Whitney-type problems have been posed and solved only in very particular cases. More precisely, either Ahlfors -regular sets [21]–[23] or special cusps in [24] have been considered. Under minimal restrictions on the corresponding problem is very complicated and has never been considered. In [25], for each Rychkov introduced so-called -thick sets , which are regular with respect to the -Hausdorff content, that is, for all and . For such sets he considered Whitney-type problems in the context of Besov and Lizorkin-Triebel spaces. Given , the class of Ahlfors -regular sets is strictly contained in the class of -thick sets, but the latter is much wider. For example, every path connected subset of is 1-thick but in general is not Ahlfors 1-regular (see Example 2.1 below).
In this paper we solve the following problem.
Problem A. Let , , and let be a -thick closed set. Given a function , how can we decide whether there exists a function such that for -quasi-every ? Consider the -norms of all functions such that for -q.e. . How small can these norms be?
Given a set with , we denote the usual trace space of the space by and let : denote the corresponding trace operator. In our paper we obtain also a solution to the following problem.
Problem B. Let and , and let be a closed -thick set. Does there exist a bounded linear operator such that on ?
In §4 we present solutions to Problems A and B. In §5 we consider simplified versions of these problems, when the set has a porous boundary. In this case the corresponding criterion (the solution to Problem A) can be simplified. In §6 we show that our main results include the corresponding results concerning -spaces obtained in [21], [23] and [24] as particular cases.
Finally we would like to underline that the methods in [25] gave a solution to Problem B only in the case . To the best of our knowledge, Problem A has never been considered in the literature. We introduce new methods which have never been used before. For example, we introduce the concept of a -regular sequence of measures and generalized Calderón-type maximal functions with respect to such sequences. Such tools help us to capture the smoothness properties of functions in the trace space and enable us to solve Problems A and B for every .
The authors wish to acknowledge their gratitude to Professor P. Shvartsman, who read early versions of the manuscript and made many valuable comments.
§ 2. Necessary background and statements of the main results
Throughout the paper will be generic positive constants. These constants can change even in a single string of estimates. The dependence of a constant on certain parameters is shown, for instance, by the notation . We write if there is a constant such that .
Throughout the paper denotes an element of the space . The symbols and will be used to denote multi-indices, that is, elements of the space . Following [23] it will often be convenient to measure distances in in the uniform norm , . Given two subsets and of , set . For any we also set .
The symbols and stand for the closed balls with centre and radius in the standard Euclidean norm and in the uniform norm , respectively (we will also call a cube) Given a number we write () to denote the ball (the cube , respectively). By a dyadic cube we mean an arbitrary half-open cube , , . Given , we let denote the mesh of all dyadic cubes with edge length .
For any set we let and denote the closure and interior of in the topology induced by an arbitrary norm in (recall that all norms in are equivalent). For and we define the -neighbourhood of to be .
Given a Borel measure and a nonempty Borel set , we define the restriction of to . More precisely, we set for every nonempty Borel set .
Let be an arbitrary Borel measure on . Given , for every Borel set with we set
2.1. Ahlfors -regular sets and -thick sets
Given , and we set
where the infimum is taken over all countable covers of by cubes with arbitrary centres and radii . We call the quantity the -Hausdorff content of .
We define the Hausdorff -measure of by . Note that our definitions of Hausdorff contents and measures are slightly different from the classical ones (cf. §5.1 in [26]). We use covers by balls in the -norm, that is, by cubes instead of classical balls. Up to some universal constants both approaches return the same values of the corresponding measures. Note also that in our case coincides with the classical Lebesgue measure on the Borel -algebra of .
Given and , recall the concept of -capacity (see §2.1 of [26]). In what follows we say that some property holds -quasi-everywhere (-q.e. for short) if it holds everywhere outside some set with . The following property summarizes the connections between -capacity and Hausdorff measures (see Theorems 5.1.9 and 5.1.13 in [26]).
Proposition 2.1. Let and let . If , then . Conversely, if , then for every .
We have taken the following definition from [22], Ch. 2, §§1.1 and 1.2.
Definition 2.1. Given , we say that a closed set is Ahlfors -regular (or just a -set for short) if there exists a -measure on , that is, a Borel measure with such that for all and
where the positive constants depend on but do not depend on or .
Remark 2.1. We can show that given a (closed) -set the restriction is a -measure on (see [22], §1.2, Theorem 1, for details).
Using Remark 2.1 we introduce the following important notation. In what follows, given an Ahlfors -regular (closed) set we set (), .
The following proposition will be useful for comparing -sets with -thick sets below.
Lemma 2.1. Let and let be a nonempty index set. Let be a family of -sets with , , such that the set is closed and Ahlfors -regular. Then the following holds.
(1) for every .
(2) Suppose the family is such that for every , , and . Then .
Proof. (1) If there exists for which , then Remark 2.1 and (2.2) with , together with elementary properties of Hausdorff measures, show that for all and all . Hence the right-hand side of (2.2) must be violated for , each and . This contradicts the Ahlfors -regularity of .
(2) Fix a point . By (1), for all . Assume that . Then it is easy to see from our assumptions that for every there is a countable family of different indices such that for all . Hence there is a sequence of distinct points such that for every . Let . It is clear from the construction and Remark 2.1 that for every . Hence the fact that the sets are disjoint and measurable, together with the countable additivity of the Hausdorff measures, gives the following estimate, which contradicts (2.2). Namely, for all
The proof is complete.
To the best of our knowledge the following concept was first introduced in [25].
Definition 2.2. Let . A set is said to be -thick if there exists a constant such that for all and
The following proposition is an immediate consequence of Definition 2.2. We omit the proof.
Proposition 2.2. Let be a -thick set for some . Then
(1) the closure of is -thick;
(2) is -thick for every , , and .
The following lemma exhibits an important relation between the concepts of Ahlfors -regular sets and -thick sets.
Lemma 2.2. Let . Every Ahlfors -regular set is -thick. Furthermore,
Proof. Suppose that is an Ahlfors -regular set. Fix a cube with and . Let be a covering of such that . Clearly, we can assume that for all . For every fix a point . Using Remark 2.1, estimate (2.2), and the subadditivity of we obtain the required estimate:
The proof is complete.
The next result is a direct consequence of Lemma 2.2, Proposition 2.2, (2), and the monotonicity of .
Lemma 2.3. Let be an arbitrary nonempty index set. Let for every . Let be a family of Ahlfors -regular sets and let . Then is -thick and
Now that we have Lemmas 2.1 and 2.3 at our disposal, we can present useful examples which illustrate the huge difference between Definition 2.1 and Definition 2.2.
Example 2.1. Let be a path connected subset of . Then and are -thick. In fact, fix a point . Let be a cube with edge length . Consider two cases.
In the first case there is a point . Hence there is a curve which connects and . Let be an arbitrary covering of for which
We choose an index set such that for every and . Consider the projections , , of our curve and the projections of cubes in the cover onto the th coordinate axes. Since we measure distances in the -norm, there exists for which . By construction the family of closed intervals covers . Hence from (2.7) we derive
In the second case . Since , we have
Combining (2.8) and (2.9) shows that is -thick, and we can take .
It is obvious that this set cannot be Ahlfors 1-regular for . In addition, elementary computations show that for each and the cusp cannot be Ahlfors -regular for .
Example 2.2. Let . This set is -thick as a union of a -thick set (a square) and a -thick set (a line interval). From Lemma 2.1, (1), it follows that is not Ahlfors -regular.
Example 2.3. Let . From Lemma 2.3 it follows that is -thick. It is not Ahlfors -regular by Lemma 2.1, (2).
Example 2.4. We can show that any -domain is -thick. We present only a sketch of the proof. Fix and . Choose an arbitrary so that . Then it easily follows from formulae (1.1) and (1.2) in [27] that there exists a curve and a point such that for some positive constant . The case can be considered similarly to the second case in Example 2.1.
2.2. Regular sequences of measures and Calderón-type maximal functions
The following concept is one of the cornerstones which enable us to solve Problem A.
Definition 2.3. Let be a closed -thick set for some . Let be a sequence of Borel measures such that , . We say that is a -regular sequence of measures on if and only if for some the following properties hold for every :
Remark 2.2. It is clear that there exists a largest positive constant for which (2.11) holds. We denote it by .
Remark 2.3. We show in §3 below that Definition 2.3 is consistent: for every -thick closed set there exists a -regular sequence of measures on .
Definition 2.4. Let be an arbitrary nonzero Radon measure. Let be a closed cube. Given a function , the best approximation to by constants on , normalized with respect to , is defined by .
Remark 2.4. Elementary computations give
(recall (2.1)).
Here and in the sequel we use the following notation. Given a number we set , so that this is the unique integer such that .
The following definition is a far-reaching generalization of the classical concept of a maximal function measuring smoothness first introduced by Calderón [28].
Definition 2.5. Let be a -thick closed set for some . Let be a -regular sequence of measures on . Let for every . Given , we consider the Calderón-type maximal function with respect to . For every
Remark 2.5. We set for brevity. If the set is Ahlfors -regular, we can take for every . Hence, in this case our maximal function is similar to that introduced by Shvartsman [23]. In particular, if , we obtain the Calderón-type maximal function (see [28]).
2.3. Porous sets
Definition 2.6. Let be a closed nonempty subset of and let . For every we define
and call the maximal -porous subset of . We say that is porous if there exists such that for every .
We gather some useful facts about porous sets. The second (see item (2) below) is a special case of Proposition 9.18 in [29].
Proposition 2.3. Let be a closed nonempty subset of and let . Then
(1) is closed for every ;
(2) if is Ahlfors -regular for some , then is porous.
Example 2.5. Let be a continuous function that is strictly increasing and such that and , . Consider the closed single cusp
It is easy to see that the boundary of is porous.
2.4. Trace spaces of Sobolev spaces
Recall that given , and an open set , the Sobolev space is the linear space of all (equivalence classes of) real functions whose generalized partial derivatives on , , belong to . This space is equipped with the norm
The next result, which is a very special case of Theorem 6.2.1 in [26], will help us to define the trace of a Sobolev function consistently on a given 'sufficiently massive' set .
Proposition 2.4. Let and . If , then there exists a set with and a representative of the element such that every point is a Lebesgue point of . If , then there exists a continuous representative of .
In the sequel we call the representative constructed in Proposition 2.4 a good representative of . Recall that, given , a property of is said to hold -quasi-everywhere (-q.e. for short) if it holds everywhere except on a set of -capacity zero.
Definition 2.7. Let and . Let be a good representative of . Given a set with , the trace of on is the class of equivalent (modulo sets of -capacity zero) functions such that for -q.e. .
Definition 2.8. Let and . Let be a continuous representative of . Given a nonempty set the trace of on the set is the pointwise restriction of to .
Below we identify a function and the class of functions each of which coincides -quasi-everywhere with on .
Now using Definitions 2.7 and 2.8 we introduce the following.
Definition 2.9. Let . Given a nonempty set with , we define the trace space of the space as follows:
We equip this space with the usual trace norm
where the infimum is taken over all such that . Furthermore, we define the trace operator which takes and gives back .
Definition 2.10. Let . Let be a nonempty set. Assume that whenever . We say that a map is an extension operator if it is the right inverse for the trace operator, so that on .
Remark 2.6. Let , and let be a -thick set. It is important to underline that Proposition 2.1 and Definition 2.9 clearly imply that for every the trace space is well defined. Furthermore, our definitions immediately implies that the trace operator is linear and bounded.
Remark 2.7. In the case the Sobolev space can be identified with the space of Lipschitz functions, and it is known that the restriction of the latter coincides with the space of Lipschitz continuous functions on and that, furthermore, the classical Whitney extension operator maps linearly and continuously into (for instance, see [20], Ch. 6). Hence in the sequel we will only deal with the case .
2.5. Statements of the main results
As we said above, without loss of generality we can work with the case .
Given a closed set and , we define
Definition 2.11. Let , and . Let be a -thick closed set. Let be a -regular sequence of measures on . For every we define the following nonnegative functionals (with values in ) on the space :
Remark 2.8. From Proposition 2.1 and Lemma 3.6 it follows that all the functionals in (2.15) are well defined on the trace space . More precisely their values remain the same after changing the function on a set of -capacity zero.
Remark 2.9. The symbols , and have not been picked at random. Informally speaking, is the 'Sobolev part' of the trace norm, while we may regard the functionals and as possible variants for the role of a Besov-type seminorm in the trace space. We clarify this in Examples 6.1 and 6.2, respectively.
Now we are ready to formulate our main result, which solves Problems A and B.
Theorem 2.1. Let and . Let be a -thick closed set. Let be a -regular sequence of measures on . Then a function belongs to the trace space if and only if for -q.e.
and for some . Furthermore,
and there exists a bounded linear extension operator .
Remark 2.10. Recall Example 2.1. Consider a path connected closed set . It is obvious that using Theorem 2.1 we obtain an intrinsic description of the trace space of the Sobolev space on in the full range of parameters . We would like to underline that even this particular case of Theorem 2.1 was never considered in the literature.
The results in Theorem 2.1 can be simplified in the case when either or possesses a certain 'plumpness'.
Theorem 2.2. Let and . Let be a -thick closed set. Let be a -regular sequence of measures on . Assume that is porous. Then a function belongs to the trace space if and only if for -q.e.
and . Furthermore,
and there exists a linear bounded extension operator .
§ 3. Main technical tools
The aim of this section is to bring together all the necessary technical results which are essential to the proofs of Theorems 2.1 and 2.2. As well as some very well-known facts, the section contains some new results. We split the section into several subsections for the reader's convenience.
3.1. Maximal functions and potentials
Let and . Given , we introduce the fractional maximal operator
We use the notation , and .
Remark 3.1. Assume that . Then it is easy to see that for every and
The following result is a very particular case of Theorem B in [30].
Theorem A. Let , , and . Let be a Radon measure on such that for some (universal) positive constant
If , then the operator is bounded from into .
The following simple fact will be of use in what follows (see [31], §2.4.3, for instance, for the proof).
Proposition 3.1. Suppose that . Then given a function , there exists a set with such that for every
Let . Given a function , for every cube we define the reduced Riesz potential by
Remark 3.2. It is useful to note a simple relation between Riesz potentials and fractional maximal operators. Suppose that . Then for every and
Now we formulate an important beautiful estimate, which is a special case of one of the implications in Theorem 2.1 in [32]. In fact, in [32] classical (nonreduced) Riesz potentials were considered. Nevertheless, the corresponding proof for reduced Riesz potentials is similar (at any rate, in the case of interest to us).
Given a (nonnegative) Radon measure and parameters and , we define the Wolf potential at the scale by
Theorem B. Let , , , and let be a positive Radon measure on . Assume that . Then there exists a positive constant (independent of ) such that
for every . Moreover, the least possible constant in (3.5) satisfies the inequality
where the positive constant does not depend on .
Recall a classical Poincaré-type inequality (see formula (7.45) in [33]).
Proposition 3.2. Assume that . Then for every cube , ,
The following estimate is well known.
Proposition 3.3. Let and . Then for -q.e. points (for every point in the case ) and every cube
where the positive constant is independent of , and .
To prove this we use Propositions 2.4 and 3.2, and then repeat the simple arguments in the proof of Theorem 5.2 in [34] almost verbatim, with minor modifications. We omit the elementary details.
3.2. Overlappings of sets
Given a nonempty family of nonempty subsets of , we say that the multiplicity of overlapping of the sets is finite if there exists such that for every .
Definition 3.1. Let be a nonempty set in . Let , and let , , be a subset of with the following properties:
(i) for every , ;
(ii) for every there is a point such that .
We call the set a maximal -separated subset of .
The following propositions will be used often in what follows. We omit the elementary proofs.
Proposition 3.4. Let be a family of pairwise disjoint cubes with the same edge length. Then for every there is a positive constant such that the multiplicity of overlapping of the cubes is finite and bounded above by .
Proposition 3.5. Let be a finite Borel measure on . Let be a family of Borel subsets of such that the multiplicity of overlapping of the sets is finite and bounded above by some constant . Then
The following elementary observation is a direct consequence of Definition 3.1.
Lemma 3.1. Let be a nonempty subset of . Let and let be a maximal -separated subset of . Then
(1) ;
(2) the family is pairwise disjoint;
(3) every point belongs to at most cubes in .
3.3. The Whitney decomposition
Recall that we measure the distances in in the uniform norm . For a cube we set . Recall that, unless otherwise stated, all cubes are assumed to be closed.
The following result is a slight modification of the Classical Whitney Decomposition Lemma. Its proof repeats the proof of Theorem 1 in [29], Ch. 6, with minor changes.
Lemma 3.2. For each closed nonempty set there exists a family of closed dyadic cubes with the following properties:
(1) ;
(2) for each
(3) the following inequalities hold:
(4) for each index there exist at most indices such that ;
(5) for every , , and if and only if .
The family of cubes , constructed in Lemma 3.2 is called a Whitney decomposition of the open set , and the cubes are called Whitney cubes. In what follows we also need the part of the Whitney decomposition comprised of the cubes of small edge length. More precisely, we set , where .
The following notation is useful below. Given a closed set , for every set
We call a cube neighbouring to a cube if . Similarly, set for every .
To construct our extension operator we use the following (see [20], Ch. 6, §1.3, for details).
Proposition 3.6. Let be a closed nonempty set and let be the Whitney decomposition of the open set constructed in Lemma 3.2. Then there exists a family of functions with the following properties:
(1) for every ;
(2) and for every ;
(3) for all ;
(4) for every multi-index and every , where the positive constant depends only on .
Definition 3.2. Given a closed nonempty set and , we say that is a nearest point to or a metric projection of onto whenever .
Remark 3.3. Let be a metric projection of onto . Consider the line interval
Consider an arbitrary and a point . We show that .
Clearly, because . Assume that . Then there is a point such that . Using this and the equality we obtain . This contradicts the fact that .
Definition 3.3. Fix a closed nonempty set . For a cube with we call a reflected cube, where is a metric projection of onto .
Remark 3.4. Clearly, a metric projection onto a closed nonempty set exists. It is not unique in general. We will specify an algorithm for choosing only when our constructions require this. Otherwise, given a cube , we fix any metric projection and the cube .
Lemma 3.3. Let be a closed nonempty set and let be a Whitney decomposition of . Then for every there exists a positive constant such that
Proof. Suppose that for some with . In view of (3.10) we have and . Hence . This implies that . Then Lemma 3.2, (5), and arguments based on volume estimates give
The proof is complete.
Lemma 3.4. Let be a closed nonempty set and let be a Whitney decomposition of . Let be a finite Borel measure with . Then for every
where the positive constant depends only on and .
Proof. Consider the family of cubes . Using Vitali's covering theorem (see §1.5 of [3] for details) we find an index set such that all cubes in the family are mutually disjoint and
Note that if for some , then
because and . From this and (3.10) it follows that . Hence, using Lemma 3.2, (5), we obtain
Using this fact, (3.13), (3.14) and Propositions 3.4 and 3.5 we obtain the required estimate
The proof is complete.
Recall Definition 2.1 and Remark 2.1. The following result is a minor modification of Theorem 2.4 in [23] and can be proved analogously.
Theorem C. Let be a closed Ahlfors -regular set in , and let be a Whitney decomposition of . Then there exists a family of Borel sets with the following properties:
(1) for all ;
(2) for all ;
(3) for .
The positive constants and depend only on , and the constants and .
3.4. -regular sequences of measures
The following result is a version of Frostman's theorem adapted for our purposes (cf. Theorem 5.1.12 in [26]). For the reader's convenience we present a detailed proof.
Theorem 3.1. Let be a closed nonempty subset of . Then given , there exists a sequence of Borel measures with , , such that for every the following properties hold:
(1)
(2) for every finite index set let ; then
(3) there exists a function such that and
Proof. Fix a nonnegative integer and let be a measure with constant density that has mass on each that intersects . We now modify in the following way. If for some , we reduce its mass uniformly on until it becomes . On the other hand, if , we leave unchanged on . In this way we obtain a new measure . Using the fact that every cube which has nonempty intersection with contains cubes with the property we have
We repeat this procedure for , obtaining , and after steps we obtain . It follows from this construction that
for every and every dyadic cube , where . Furthermore, it is clear that
Using (3.18) it is easy to see that for every the sequence is bounded for every compact subset of . Then has a subsequence that converges weakly to (see [31], §1.9, Theorem 2), and clearly (recall that is closed).
Fix an arbitrary and an arbitrary Borel set . We compare and . First note that, according to our construction, for every dyadic cube we have
Let be the set of continuous functions with compact support. For every nonnegative function we have
Fix an arbitrary nonnegative . Choosing an appropriate subsequence if necessary and passing to the limit in (3.20) we obtain
Using the Borel regularity of the measures and the Radon-Nikodym theorem and taking (3.21) into account we obtain (3.17).
We show that for every , , and every dyadic cube . Indeed, if is such that , then (3.18) yields
Hence, using the fact that every closed cube with and has nonempty intersection with dyadic cubes , where , we obtain (3.15).
Fix an arbitrary nonempty index set and and fix an arbitrary , . The key observation, which follows directly from our construction, is that every belongs to some dyadic cube , , (or several cubes) such that . We can choose a disjoint covering consisting of maximal dyadic cubes with this property, so that . This gives
where the infimum is taken over all finite or countable coverings of with dyadic cubes . The right-hand side is independent of . Combining this with the definition of the -Hausdorff content we note that is a compact set. This gives
This completes the proof.
The following result shows that Definition 2.3 is consistent.
Corollary 3.1. Let and let be a -thick closed set. Then there exists a -regular sequence of measures on .
Proof. We apply Theorem 3.1 to the set . This gives a sequence of Borel measures with satisfying (3.15)–(3.17). We set for every .
It is sufficient to verify (2.11). Fix some , and let be the index set such that if and only if . It is clear that and . Hence, using Definition 2.2, estimates (3.16) and (3.17) and the monotonicity of the -content we obtain
This completes the verification of the corollary.
The following lemma gives some asymptotic estimates for measures in a fixed -regular sequence. Recall Remark 2.2.
Lemma 3.5. Let and let be a -thick closed set. Let be a -regular sequence of measures on . Then for each and every
Proof. Fix such that . It follows from estimates (2.10)–(2.12) that for every and every
Similarly,
The required estimate (3.23) follows from (3.24) and (3.25), which completes the proof.
Remark 3.5. Recall that a Borel measure on a metric space is called a doubling measure if there exists a constant such that for all and . It is very important to note that the estimate (3.23) does not imply the doubling property of the measures , . Roughly speaking, the point is that, given , in Lemma 3.5 we compare , where and , only with . If we try to compare and for , then we obtain a bad estimate, with the corresponding positive constant depending heavily on .
Lemma 3.6. Let and let be a -thick closed set. Let be a -regular sequence of measures on . Let be a Borel subset of . If , then for every . The converse is false.
Proof. Fix with . Using (2.10) and the definition of Hausdorff measure it is easy to see that for every .
To prove that the converse is false we use the construction from Example 6.3 below. More precisely, in Example 6.3 we build a 1-thick path-connected set with and a -regular sequence of measures on such that every is absolutely continuous with respect to . Hence for every smooth curve with we obtain for all . The proof is complete.
Recall that for every we set . The following theorem will be an important technical tool in the sequel. Recall Remark 2.2.
Theorem 3.2. Let and let be a -thick closed set. Let be a -regular sequence of measures on . Then for every , and every Borel set
The positive constant depends only on and in Remark 2.2.
Proof. Fix , and and , , such that . It is clear that
On the other hand, using (2.10)–(2.12) (we can use these estimates because ) we have
Combining (3.27) and (3.28) we obtain
Hence we obtain (3.26) for .
Fix . For every , , let be a maximal -separated subset of . Clearly, and the cubes are pairwise disjoint. For every take an arbitrary nonempty set and consider the set . It is clear that for every . We use this inclusion, (3.29), Propositions 3.4 and 3.5, and (2.12); then for every we obtain
Fix an arbitrary compact set . Now using the -additivity of the measures and we have
Combining (3.31) and (3.30), we obtain (3.26) for every compact set . To establish (3.26) for a general Borel set it remains to recall that the measures , , and are Radon measures. The proof is complete.
Corollary 3.2. Let and let be a -thick closed set. Let be a -regular sequence of measures on . Assume that for some (and hence every) . Then there exists (independent of , and ) such that
Proof. The estimate (3.32) clearly holds for a simple function due to Theorem 3.2. In the general case we have to construct an increasing sequence of simple functions converging to and use the monotone convergence theorem for integrals (see §1.3 of [31]). This completes the verification of the corollary.
Lemma 3.7. Let and let be a -thick closed set. Let be a -regular sequence of measures on . Assume that for some (and hence every) . Let , , and satisfy and . Then
where the positive constant does not depend on , , or .
Proof. Clearly, . Suppose . Using this we obtain by Lemma 3.5, which obviously implies the first inequality in (3.33). When the corresponding estimate easily follows from (2.11) and (2.12).
The second estimate in (3.33) follows immediately from (2.10)–(2.12). The proof is complete.
3.5. Calderón-type maximal functions
Recall Definition 2.5 and also that for every .
Lemma 3.8. Let and let be a -thick closed set. Let be a -regular sequence of measures on . Assume that for every . Let , , and be such that . Then
The positive constant in (3.34) depends on the constant but does not depend on or .
Proof. Fix . If , then we use the inclusion , Lemma 3.7, Remark 2.4 and the monotonicity of with respect to . We obtain
Now consider the case . We use Remark 2.4 and Lemma 3.7 and note that . This gives
Now (3.34) follows directly from Definition 2.5 and estimates (3.35) and (3.36). The proof is complete.
Lemma 3.9. Let and let be a -thick closed set. Let be a -regular sequence of measures on . Assume that for some (and hence every) . Let , and be such that and . Then
where the positive constant does not depend on or .
Proof. Clearly, and . Since we have and hence . Arguing as in (3.35) and using Remark 2.4 we obtain
(note that ). Clearly, similar inequalities hold true with replaced by . Hence, we get
Let . Arguing as in (3.38) and using Remark 2.4 we have
Combining (3.38) and (3.39) we obtain (3.37), which completes the proof.
Theorem 3.3. Let and let be a -thick closed set. Let be a -regular sequence of measures on . Let for some (and hence every) . Then there exists a positive constant such that for any , with
Proof. Note that and . Using the triangle inequality, (3.37) and the monotonicity of with respect to we see that the left-hand side of (3.40) is bounded above by
here we have set . This completes the proof.
Corollary 3.3. Let and let be a -thick closed set. Let be a -regular sequence of measures on . Let for some (and hence every) . Let be such that and let
Then there exists (independent of , and ) such that
Proof. From Theorem 3.3, using the monotonicity of with respect to we obtain the required estimate
This completes the verification of the corollary.
The following result is crucial in proving the 'direct trace theorem'. Recall Proposition 2.4 and Definitions 2.7 and 2.8.
Theorem 3.4. Let and let be a -thick closed set. Let be a -regular sequence of measures on . Let and . Then for every cube with and
where the positive constant does not depend on , or .
Proof. Fix a cube with and . We consider several cases.
Case 1. Assume that . Using the well-known Sobolev embedding theorem (see [10], §1.8.2) we obtain the estimate required in this case:
Here is the continuous representative of and depends only on and .
Case 2. Now we consider the most complicated case, when and . We set . It is clear that for every . We can rewrite (3.8) as follows. For -quasi-every (and hence for -almost every)
Set . Since lies in a union of at most cubes with edge length , it follows from (2.10) that for and . This gives (recall the definition of the Wolf potential (3.4))
The positive constant in (3.47) does not depend on or .
Now we recall Lemma 3.6 and apply Theorem B with measure (instead of ) and with . Hence, using (2.11), (3.46) and (3.47) we obtain
Case 3. In the case , we choose an arbitrary and use the previous step to obtain (3.45) with instead of . To complete the proof it remains to apply Hölder's inequality.
Corollary 3.4. Let and let be a -thick closed set. Assume that and and set . Then for every and every
Proof. Consider the case . The case is similar. Using Remark 2.4 and Theorem 3.4, for every we have the required estimate
The verification of the corollary is complete.
3.6. Porous sets
Recall Lemma 3.2 and Definitions 2.6 and 3.2. Also recall that we let denote the unique integer such that . We continue to measure distances in in the -norm.
Lemma 3.10. Let be a closed nonempty set in . Let be a Whitney cube in . Then for every and . Furthermore, for every and every .
Proof. Consider the interval . Clearly, because otherwise there exists a point such that . For every consider the point . It follows from Remark 3.3 that . Hence for every the cube lies in . This proves the first claim in the lemma.
Given a number we set . Then from Remark 3.3 we conclude that . On the other hand it is clear that for every . This proves the second claim.
Lemma 3.11. Let be a closed nonempty set in . Let be a Whitney decomposition of . Let and . Then for every there exists a point such that
for every Whitney cube .
Proof. By Definition 2.6 there exists a point such that . We set . Now we prove (3.51). Consider an arbitrary Whitney cube . From (3.10) we have
On the other hand, using (3.10) again, we see that
Combining the above estimates we complete the proof.
§ 4. The main results
Recall that, given , we set . For each closed nonempty set and a Whitney decomposition of we set for each . Throughout this section, unless otherwise stated, we equip with the uniform norm .
4.1. The direct trace theorem
Recall Definitions 2.3 and 2.6 and the notation following Lemma 3.2.
Lemma 4.1. Let , and . Let be a -thick closed set. Let be a -regular sequence of measures on . Assume that for some (and hence every) . Then there is a positive constant independent of such that
Proof. Given , let be a maximal -separated subset of . Using Lemma 3.1, (1), for every we have
Using Lemma 3.11, for every and we choose a point and fix an index such that and (3.51) holds. We define a map as follows:
Since , it follows from (3.51) that . Using this, Lemma 3.1, (2), and Proposition 3.4 for , it is easy to see that there exists a positive constant such that for every fixed and every
It follows from (3.51) that the equality implies that
Combining (3.51), (4.4) and (4.5), for every we obtain
If for some , , then it follows from (3.10) and (3.51) that
This gives the inclusion
Now we use (2.10) and (4.7) and apply Lemma 3.8 for . Then for every we obtain
From (4.6) it follows that
Similarly
Combining (4.2), (4.8), (4.9) and (4.10) we complete the proof of Lemma 4.1.
Lemma 4.2. Let , and . Let be a -thick closed set. Let be a -regular sequence of measures in . Assume that . Then for each there is a positive constant depending only on and such that
Proof. We use (2.11) and then apply Lemma 3.4 with . This gives the required estimate
The proof is complete.
Recall the notion of a good representative of a given element .
Lemma 4.3. Let and . Let be a -thick closed set. Let be a -regular sequence of measures on . Then for every
Proof. Let be the intersection of with the set of all Lebesgue points of the function . It follows from Proposition 2.4 that . For every we have
Clearly, as according to the construction of . Combining this fact with (4.14) and Proposition 2.1 we conclude that it is sufficient to show that as for -a.e. . In fact, applying Theorem 3.4 (for ) and Proposition 3.1 gives
The lemma is proved.
Now we are ready to prove the main result of this subsection. Recall Definitions 2.7–2.9 and 2.11.
Theorem 4.1. Let , and . Let be a -thick closed set, and let be a -regular sequence of measures on . Then the functional is bounded on the trace space .
Proof. It is sufficient to verify that there exists a positive constant (independent of ) such that the inequality
holds for each with .
We fix some with throughout the proof.
Step 1. First of all we estimate from above. Fix an arbitrary . We apply Corollary 3.4, Remark 3.1 and Theorem A for , , and exponent instead of . Then we obtain
Step 2. Fix some . From Lemma 3.2 it follows that for every , and . Applying Corollary 3.4, Remark 3.1, Theorem A for , and , and the fact that the interiors of different Whitney cubes are mutually disjoint gives
Step 3. We estimate from above. Let be a maximal -separated subset of . Consider the family of cubes . It is clear that for every . Using this and Lemma 3.1, (1), we derive the following estimate:
Using Hölder's inequality, (2.10), Lemma 3.1, (2), Proposition 3.4 with and Proposition 3.5 with , we easily obtain
Recall Proposition 2.1 and Lemma 3.6 (also recall that ). Then applying Proposition 3.3 and Remark 3.2 shows that there exists (independent of ) such that for every
Since , we can choose in such a way that . Now we use Lemma 3.1, then Proposition 3.5 with , and finally apply Theorem A with and . As a result, using (4.21) we obtain
Combining (4.19), (4.20) and (4.22) yields
Step 4. Combining Lemmas 4.1 and 4.2, and estimates (4.18) and (4.23) we obtain
Now the required estimate (4.16) follows directly from (4.17), (4.23) and (4.24). The proof is complete.
4.2. The reverse trace theorem
The following pointwise characterization of the functions in the space is given in [35].
Theorem D. Let and . Then if and only if there exist a nonnegative function , a set with and a constant such that
for every with . Furthermore,
where the positive constant does not depend on .
Proof. We only sketch the proof. One implication was established in [35] (see the text before Theorem 1 there); see also [36]. For the reverse implication we have to cover by a countable family of 'sufficiently nicely overlapping' balls of diameter and for every apply Theorem 1 in [35] for . Then we observe that, given a function , the fact that implies for any . This shows that locally integrable weak derivatives exist on and their restrictions to any open ball coincide with the corresponding weak derivatives of . It remains to sum the appropriate analogues of estimate (4.26) for all open balls .
Now we are ready to present our construction of the extension operator.
Definition 4.1. Let be a -thick closed set for some . Let be a -regular sequence of measures on . Assume that for every . For the same family of functions as in Proposition 3.6 we set
where
Remark 4.1. In fact, (4.27) defines not a single extension operator, but a whole family of operators. The reason is that the choice of the -regular sequence of measures is not unique. Furthermore, generally speaking, the choice of the reflected cubes is not unique either.
The following result plays a crucial role. It gives a pointwise estimate of the extension constructed in (4.27). Recall the notation : see (3.12).
Lemma 4.4. Let and let be a -thick closed set. Let be a -regular sequence of measures on . Let for some (and hence every) . Suppose that
Then there exists a positive constant depending only on , , and such that for each , for -a.e. , , the function defined in (4.27) satisfies
where, for every ,
Proof. Fix an arbitrary . It is obvious that we have to consider five cases:
(1) and ;
(2) , and ;
(3) , and ;
(4) and ;
(5) and, furthermore, and either or .
By the symmetry of the left-hand side of (4.29) with respect to and , we can identify cases (2) and (3) up to changes in notation.
Case (1). From (4.28) and Corollary 3.3 it follows that for -almost every
Case (2) ( case (3)). Consider the case when , and . Since , estimate (3.11) and Proposition 3.6, (2), give
Recall the notation (given after Lemma 3.2). From Proposition 3.6, (2), it follows that for every . Hence from (4.27) and (4.32) we have
Fix and set . Now, (3.10) and Proposition 3.6, (2), give
Hence, using (3.10) again we obtain
We use (4.34) and the fact that . Then we obtain
We note that can be much smaller than . Hence we need to estimate with care. Note that . Using the same arguments as in the proof of Corollary 3.3, for every and -a.e. we have
we have also used the fact that . As a result, combining (4.33), (4.35) and (4.36), for -a.e. and all we deduce that
Case (4). Fix and with . From (3.10) it follows that for every and . Furthermore, from (3.11) it follows that in this case. Hence we have
There are two subcases here. In the first for any and ; in the second for some cubes and containing and , respectively.
Consider the first subcase. Using (4.38) and arguing as in (4.33) we have
For fixed and we set
It is clear that and in this subcase. Hence we obtain
On the other hand, using (3.10) we have
Combining (4.40) and (4.41) we easily deduce that
It is clear that and . We take (4.42) into account and apply Theorem 3.3 (which is possible due to (4.42) and the restrictions on ). Then we obtain
Combining (4.39) and (4.43) and using (4.30), for this choice of and we have
Now consider the second subcase. Fix arbitrary and with . Let be a smooth curve with , . We use (4.27), Lagrange's mean value inequality, (4.38), Proposition 3.6, (4), and estimates (3.11). As a result, we obtain (recall that for each )
From (3.10) and (3.11) it follows that
Similarly, for every . As a result, since for (due to (3.10)) we obtain
Using (4.46) we apply Theorem 3.3 and continue with (4.45). This gives
Combining (4.44) and (4.47) we can establish case (4).
Case (5). Fix and such that . Without loss of generality assume that . Then . By (3.10) this implies that for every and
Consider two subcases by analogy with case (4).
In the first subcase for any and . Then the same arguments as in (4.40) together with (4.48) give
By (4.27) and (4.30) this implies that
In the second subcase there are and with . Arguing as in (4.45)) and taking (4.48) into account we obtain
Combining (4.50) and (4.51) we obtain case (5).
The proof of Lemma 4.4 is complete.
Lemma 4.5. Let , and . Let be a -thick closed set. Let be a -regular sequence of measures on . Then there exists a positive constant , depending only on , , and , such that
Proof. Clearly, for every and . Hence, applying Lemma 3.8 (for ) gives
for every . Note that because . Hence using (2.11) and (3.23) for , from (4.53) we derive that
It follows from Lemma 3.3 that for every fixed the multiplicity of overlapping of the sets with is bounded above by some positive constant . Furthermore, from Lemma 3.10 it follows that for each , provided that . This, (4.54) and Lemma 4.2 (with ) give the required estimate
Combining (4.54) and (4.55) we complete the proof.
Lemma 4.6. Let , and . Let be a -thick closed set. Let be a -regular sequence of measures on . Let be a Borel function such that . Let be the function constructed in (4.27) and be the function defined in (4.30). Then
The positive constant in (4.56) is independent of .
Proof. From (4.27) and Proposition 3.6 it is clear that . It follows from (4.30) that for some positive constant independent of we have
Hence it is sufficient to establish that
with a positive constant independent of . It is clear that (4.56) follows from (4.57) and (4.58).
Step 1. We establish the first estimate in (4.58). Using Lemma 3.2, (3)–(5), we obtain
From Lemma 4.5 and (2.15) we clearly have
Using (2.13) and arguing as in (3.39) we obtain
Using Hölder's inequality, (4.60) and Lemma 4.2 for , from (4.61) we obtain
As a result, combination of (4.59), (4.60) and (4.62) gives the first estimate in (4.58).
Step 2. Let be a maximal -separated subset in . Using Definition 2.3, Lemma 3.1, Corollary 3.2 and Propositions 3.4 and 3.5, we obtain the second estimate in (4.58):
The lemma is proved.
Recall the definitions of the trace of a given element on the set (Definitions 2.7 and 2.8). Also recall Definition 4.1.
Theorem 4.2. Let and . Let be a -thick closed set and a -regular sequence of measures on . If for some and
then .
Proof. Let be the set of all points where (4.64) holds. Set . We are going to show that . Fix a cube with and . Using (3.10) and the definition of we see that
From (4.65) it follows that and thus . Then it follows from Lemma 3.2, (4), and Proposition 3.6, (2), (3), that
Using Corollary 3.2 and (4.64), for -q.e. we have
To estimate we need some preliminaries.
From (4.65) we deduce the inequality . Hence for all such that we have
Recall that Whitney cubes have disjoint interiors. Hence, using (4.68) we obtain
Now for every with it follows from (3.10) that
For all such , from (4.70) it follows that
Taking the inclusions (4.71) into account we use (2.13) and then apply Lemma 3.9 for (note that ). Finally, we use (4.70) and the monotonicity of with respect to . This gives
Combining (4.69) and (4.72) we obtain
From (4.69), by Hölder's inequality for sums with exponents and we see that the second term on the right-hand side of (4.73) is bounded above by
Clearly, the first term on the right-hand side of (4.73) tends to zero as for -q.e. points because of (4.64).
In the case we use (4.59) and (4.60) and obtain
for every because . Hence as everywhere in this case. This, with (4.66) and (4.67), shows that -quasi-every point where (4.64) holds is a Lebesgue point of . This fact together with Definition 2.7 proves the claim in the case .
Consider the case . In view of Proposition 2.1, to show that as for -q.e. points it is sufficient to verify that as for -q.e. points .
In the case this is easy. Indeed, since and (4.60) holds, we can estimate from above by a remainder part of a convergent series:
Consider now the case . Fix and define
Fix some . For each point we find with . The family of cubes covers . Using Vitali's covering theorem (see [31], Ch. 1, §1.5.1) we find an at most countable family of disjoint cubes such that . Since the cubes are disjoint, the inclusions in (4.68) imply that for each the cube can have nonempty intersection with at most one cube . This observation together with (4.74) and (4.77) yields
Recall that according to our assumption . This fact, with estimates (4.59) and (4.60), shows that the right-hand side of (4.78) is a remainder part of a convergent series and hence tends to zero as . This clearly implies that , and hence . As a result, the arguments above and (4.73) imply that
Combining (4.66), (4.67) and (4.79) we complete the proof in the case . Theorem 4.2 is proved.
Now we can formulate the reverse trace theorem. Recall our construction of the extension operator (4.27).
Theorem 4.3. Let , , and . Let be a -thick closed set. Let be a -regular sequence of measures on . Let and
Then and
where the positive constant depends only on the parameters , , , and .
Proof. Assume that . Then it is obvious that for all . Consequently, (4.27) yields a well-defined function , whose pointwise restriction to coincides with the original function . Applying Theorem 4.2 gives , and hence .
From Theorem D and Lemmas 4.4 and 4.6 it follows that and
where the positive constant does not depend on . Combining Definition 2.9 and (4.82) we obtain (4.81). The theorem is proved.
4.3. The proof of the main result
Now we are ready to prove the main result in this paper.
Proof of Theorem 2.1. Given , from Theorem 4.1 it follows that
where the positive constant does not depend on . Furthermore, from Lemma 4.3 we deduce that (2.16) holds.
Conversely, assume that . Then we deduce from Theorem 4.3 that and
where the positive constant does not depend on .
Clearly operator constructed in (4.27) is linear. Furthermore, it was mentioned in the proof of Theorem 4.3 that .
Finally, estimates (4.83) and (4.84) obviously imply (2.17) and boundedness of the operator . The proof is complete.
Remark 4.2. As we noted above, while constructing the extension operator we chose a -regular sequence of measures. It is remarkable, however, that both the proofs of Theorems 2.1 and 4.3 and the constants in these proofs depend only on the constant in Remark 2.2 but are independent of the concrete choice of the -regular sequence of measures.
Remark 4.3. We would like to draw the reader's attention to the fact that (in general) it is impossible to obtain (2.16) from the condition alone. The point is that, given a set with , , we cannot claim that . Hence, changing the given function on a set such that , , does not affect the value of , but can violate (2.16).
§ 5. A simplified criterion for sets with porous boundary
In this section we are going to prove Theorem 2.2, which is a simplified version of Theorem 2.1 in the case of sets with porous boundary. Recall Definition 2.6.
Let be a closed set in with porous boundary. Given , for every we set
From Definition 2.6 it is clear that if is porous then there exists a number such that for all
Definition 5.1. Let be a closed nonempty subset of . We set
Lemma 5.1. Let , and . Let be a -thick closed set and a -regular sequence of measures on . Assume that is porous. Then there exists a positive constant depending only on , , and such that for every
Proof. Given , let be an arbitrary maximal -separated subset of . For every we set , .
Step 1. Given and , it is clear that for every . Hence elementary computations (similar to (3.35)) give
Using Lemma 3.1, (1), estimates (2.10) and (5.5) and Remark 2.4 we obtain
Fix some . Arguing as in (3.50), from (5.6) we derive the following estimate:
Step 2. Fix some such that (5.2) holds. Since all the cubes are assumed to be closed, for all and . Let be the set of all such that . Let be the set of all such that . It is clear that for every .
Let and be Whitney decompositions of and , respectively. Let and be the sets of indices corresponding to the cubes with edge length in and , respectively.
For every and we choose a point , and for every we choose a point . Since for every and , we apply Lemma 3.11 and for each we find a point such that
Similarly, for each we find a point such that
Consider a map that takes a pair () and returns an arbitrary index such that (5.8) holds. Similarly, we build a map that takes and returns an arbitrary such that (5.9) holds. Arguing as in (4.6), from (5.8) and (5.9) we derive the existence of a positive constant such that for every and
Step 3. Let , and . From (5.8) it follows that for every . Using this and (3.1) we obtain
Similarly, if , and , then
Combining (5.10) and (5.11) we have
Similarly, from (5.10) and (5.12) we obtain
We combine estimates (5.7), (5.13) and (5.14) and apply Theorem A with , and . This gives
Lemma 5.1 is proved.
Lemma 5.2. Let , and . Let be a -thick closed set with porous boundary . Let be a -regular sequence of measures on . Assume that for some (and hence every) . Then
The positive constant in (5.16) does not depend on .
Proof. Fix an arbitrary positive number . We set . Using Vitali's covering theorem (see [31], §1.5.1) we find an index set such that the cubes belonging to the family are mutually disjoint and
Note that if for some and , then . Using Remark 2.4 and reasoning as in (3.35) it is easy to show that for such and
Since different Whitney cubes have disjoint interiors, we find that for every
Combining (5.17), (5.18) and (5.19) we obtain
It is clear that for every . Using this observation, Remark 2.4, (2.11) and (2.12) and arguing as in (3.35), for every we have
It is clear that . Furthermore, according to our construction of the multiplicity of overlapping of the sets , is finite and independent of . Hence, substituting (5.21) into (5.20) and using Proposition 3.5 we obtain
Now we use Remark 2.4 and Hölder's inequality. Then we use Lemma 3.7 and Lemma 4.2 for . This gives
To complete the proof it is sufficient to put (5.22) and (5.23) together.
Lemma 5.3. Let , , and . Let be a -thick closed set with porous boundary . Let be a -regular sequence of measures on . Assume that for every . Then for each
The positive constant in (5.24) is independent of .
Proof. It is clear from Lemma 3.5 and Remark 2.4 that for every (recall that for such ) we can choose so that and
Given , let . Estimate (5.25) together with Lemmas 4.1, 4.2 and 5.2 allow us to deduce that
The proof is complete.
Proof of Theorem 2.2. Let . It follows from Lemma 5.1 and (4.16) that
where the positive constant does not depend on . Furthermore, from Lemma 4.3 we deduce that (2.18) holds.
Conversely, let . Then, from Lemma 5.3 we deduce that for some , with a positive constant independent of . Hence, from Theorem 4.3 we deduce that and
where the positive constant does not depend on .
Finally, estimates (5.27) and (5.28) obviously imply (2.19). The proof is complete.
§ 6. Example
The aim of this section is to present several useful examples, which show the power of our main results.
Example 6.1. Let be an Ahlfors -regular closed subset of , and let . In this case we can take for every to obtain an -regular sequence of measures on . Hence, taking Remark 2.4 into account, for every we have
To simplify our notation we set in this case. This notation was used in [23].
We establish the following key estimate:
where the positive constant does not depend on .
We combine Theorem C, (2), and Lemmas 4.1 and 4.2. This gives
Adapting the arguments in the proof of Lemma 4.5 to this case (and also using the monotonicity of with respect to ) we easily obtain
Now (6.2) clearly follows from (6.3) and (6.4).
Recall that -a.e. points are Lebesgue points of a function . If we relax the notion of the trace of a given and identify with the class of functions equivalent modulo coincidence -a.e. on , then from (6.2) we deduce the following simplified version of Theorem 2.1.
Let be an Ahlfors -regular set. Then a function belongs to the trace space if and only if
Moreover, the operator constructed in (4.27) is a bounded linear extension operator and
This result is a slight modification of the corresponding result obtained by Shvartsman [23] in the context of first-order Sobolev spaces.
Example 6.2. Let and . Let be an Ahlfors -regular subset of . In this case there exists a simple -regular sequence of measures on . More precisely, set for every . Clearly, . Furthermore, and is porous (see Proposition 2.3).
Note that the measure is Radon. Hence, from Theorem 1 in [31], §1.7.1, we conclude that if , then
for -almost every .
Now we apply Theorem 2.2 and take the above facts into account. If we relax the notion of the trace of a given element and identify with the class of functions equivalent modulo coincidence -a.e. on , we obtain the following simplified version of Theorem 2.2.
Given an Ahlfors -regular set , for some , let . Then a function belongs to the trace space if and only if
Moreover, the operator constructed in (4.27) is a bounded linear extension operator from to and
Note that this result coincides with that obtained in [21] in the context of first-order Sobolev spaces.
In the simplest case when , , this is a classical result. Namely, . This fact together with Theorem 2.2 implies that . This equivalence has motivated us to call a 'Besov-type seminorm'.
Example 6.3. Let be a strictly increasing continuous function such that and for every . Let denote the inverse function, so that on . Consider the closed single cusp . For each we also consider the sets
Recall Definition 5.1. It is clear that coincides with .
For every consider the measure , where
It is clear from (6.7) (recall that is strictly increasing) that
Using the monotonicity of and elementary geometric observations, it is easy to see that for every point and
On the other hand, using (6.7) and the monotonicity and continuity properties of , it is easy to show that for every
Direct computations give
for every .
Combining (6.8)–(6.11) we see that the sequence of measures , possibly after multiplying by a fixed constant (depending on and ), becomes -regular on .
Recall Example 2.1, (2), and Example 2.5. Thus we see that the set is -thick and has a porous boundary. Consider a slightly relaxed definition of the trace of on the set . Namely, we write if for -a.e. . Then from Theorem 2.2 we clearly derive the following criterion.
Let . Then a function lies in the trace space if and only if
Furthermore, the functional gives an equivalent norm in the trace space and the operator in (4.27) is a bounded linear extension operator from to .
Remark 6.1. To the best of our knowledge the results in Example 6.3 are new and could not be obtained using the techniques previously known. However, we have to mention [24], where a similar example was considered under certain additional assumptions on . More precisely, it was assumed there that is Lipschitz. Furthermore, in [24] precise statements were only formulated in the case .