Abstract

Let be a complete -finite measure space, be a Young function, and and be Banach spaces. Let denote the Orlicz-Bochner space, and denote the finest Lebesgue topology on . We study the problem of integral representation of -continuous linear operators with respect to the representing operator-valued measures. The relationships between -continuous linear operators and the topological properties of their representing operator measures are established.

1. Introduction and Preliminaries

Throughout the paper, and denote real Banach spaces and and denote their Banach duals, respectively. By and we denote the closed unit ball in and in . Let stand for the space of all bounded operators from and , equipped with the uniform operator norm .

We assume that is a complete -finite measure space. Denote by the -ring of sets with . By we denote the linear space of -equivalence classes of all strongly -measurable functions , equipped with the topology of convergence in measure on sets of finite measure.

Now we recall the basic concepts and properties of Orlicz-Bochner spaces (see [16] for more details).

By a Young function we mean here a continuous convex mapping that vanishes only at 0 and as and as . Let stand for the complementary Young function of in the sense of Young.

Let (resp., ) denote the Orlicz-Bochner space (resp., Orlicz space) defined by a Young function ; that is,Then , equipped with the topology of the normis a Banach space. For a sequence in , if and only if for all . LetLetThen is a -closed subspace of .

Recall that a subset of is said to be solid whenever -a.e. and , imply . A linear topology on is said to be locally solid if it has a local basis at 0 consisting of solid sets (see [4]).

According to [7, Definition 2.2] and [6] we have the following definition.

Definition 1. A locally solid topology on is said to be a Lebesgue topology if for a net in , in the Banach lattice implies in .

In view of the super Dedekind completeness of one can restrict in the above definition to usual sequences in (see [7, Definition 2.2, p. 173]).

Note that, for a sequence in , in if and only if   -a.e. and   -a.e. for some .

For let . Then the family of all sets of the form:where is a sequence of positive numbers and is a local basis at 0 for a linear topology on (see [4, 6] for more details). Using [4, Lemma 1.1] one can show that the sets of the form are convex and solid, so is a locally convex-solid topology.

We now recall terminology and basic facts concerning the spaces of weak-measurable functions (see [8, 9]). Given a function and , let for . By we denote the linear space of the -equivalence classes of all -measurable functions . In view of the super Dedekind completeness of the set is order bounded in for each . Thus one can define the so-called abstract norm byOne can easy check that the following properties of hold: if and only if and , for and , if , for and .

It is known that, for , , the function defined by is measurable andMoreover, for . LetClearly . If, in particular, has the Radon-Nikodym property (i.e., is an Asplund space; see [10, p. 213]), then .

Let stand for the Banach dual of , equipped with the conjugate norm .

Recall that a Young function satisfies the -condition if for some and all . We shall say that a Young function is completely weaker than another (in symbols, ) if for an arbitrary there exists such that for all . Note that a Young function satisfies the -condition if and only if . If , then and it follows that .

Now we present basic properties of the topology on .

Theorem 2. Let be a Young function. Then the following statements hold:(i) and if satisfies the -condition.(ii) is the finest Lebesgue topology on .(iii) is generated by the family of norms .(iv), where for ,(v) is a closed subset of the Banach space .(vi)If has the Radon-Nikodym property, then the space is strongly Mackey; hence coincides with the Mackey topology .

Proof. (i)–(iii) See [4, Theorems 6.1, 6.3 and 6.5].
(iv) In view of [6, Corollary 4.4 and Theorem 1.2], we get , where stands for the order continuous dual of (see [7, 8, 11] for more details). According to [8, Theorem 4.1] .
Using [11, Theorem 1.3] for we have(v) See [12, § 3, Theorem 2].
(vi) See [6, Theorem 4.5].

Let (briefly ) denote the natural mixed topology on ; that is, is the finest linear topology that agrees with on -bounded sets in (see [5, 13, 14] for more details). Then is a locally convex-solid Hausdorff topology (see [14, Theorem 3.2]) and and have the same bounded sets. This means that is a generalized DF-space (see [15]) and its follows that is quasinormable (see [15, p. 422]). Moreover, for a sequence in , in if and only if in and (see [14, Theorem 3.1]).

We say that a Young function increases essentially more rapidly than another (in symbols, ) if for arbitrary , as and .

Theorem 3. Let be a Young function. Then the mixed topology on is generated by the family of norms .

Proof. It is known that the mixed topology on is generated by the family of norms (see [16, Theorem 2.1]). Since for , by [14, (54), p. 97], the mixed topology on is generated by the family of norms .

Since implies , in view of Theorems 2 and 3, we get

The problem of integral representation of bounded linear operators on Banach function spaces of vector-valued functions to Banach spaces in terms of the corresponding operator-valued measures has been the object of much study (see [5, 1724]). In particular, Dinculeanu (see [19, § 13, Sect. 3], [20], [21, § 8, Sect. B]) studied the problem of integral representation of bounded linear operators from to a Banach space . It is known that if , and an operator measure vanishes on -null sets and has the finite -semivariation , then one can define the integral for all . Moreover, if is a bounded linear operator, then the associated operator measure has the finite -semivariation and for all (see [19, § 13, Theorem 1 p. 259], [20, Theorem 4]). The relationships of the -semivariation to the properties of operators from to were studied in [22]. Diestel [23] found the integral representation of bounded linear operators from an Orlicz-Bochner space to a Banach spaces if and a Young satisfies the -condition.

The present paper is a continuation of [5], where we establish integral representation of -continuous linear operators . We study the problem of integration of functions in with respect to the representing operator measures of -continuous linear operators . An integral representation theorem for -continuous linear operators is established (see Theorem 9 below). We study the relationships between -continuous operators and the properties of their representing measures .

2. -Semivariation of Operator Measures

Assume that is an additive measure such that ; that is, if .

Let denote the space of all -valued -simple functions on . Then if , where is a finite pairwise disjoint sequence in and . For and , we can define the integral byNote thatFor , we define a measure by the equality

For and , we define the integral by the equality:Then

Following [23], [19, § 13] one can define the -semivariation of on bywhere the supremum is taken over all finite pairwise disjoint sets in and for such that .

One can observe thatNote thatLet stand for the -semivariation of on ; that is,

The following lemma will be useful.

Lemma 4. Let be a Young function and be a measure with and . Then the following statements hold:(i)If , then there exists a -Cauchy sequence in such that -a.e.(ii)If is a -Cauchy sequence in , then for , is a Cauchy sequence in a Banach space and for every , is a Cauchy sequence in .(iii)If and and are -Cauchy sequence in such that -a.e. and -a.e., then for , one hasand for every , one has

Proof. (i) Let . Then there exists a sequence in such that -a.e. and -a.e. for all (see [21, Theorem 6, p. 4]). Using the Lebesgue dominated convergence theorem, we obtain that for all , so . Hence is a -Cauchy sequence.
(ii) Assume that is a -Cauchy sequence in . Hence for , we haveIt follows that is a Cauchy sequence in . Hence in view of (15), for , is a Cauchy sequence in .
(iii) Note that is a -Cauchy sequence and -a.e. Hence there exists such that . Note that . Hence in and it follows that there exists a subsequence of such that -a.e. Then -a.e., so and for , we getIt follows thatand hence, in view of (15) for every , we have

Following [21, § 13, Definition 1, p. 254], in view of Lemma 4 we have the following.

Definition 5. Let be a Young function and be an additive measure such that and . Then for every and , we can define the integral by the equalityand for , we can define the integral by the equalitywhere is an arbitrary -Cauchy sequence in such that -a.e.

3. Integral Representation of Continuous Operators on Orlicz-Bochner Spaces

For a bounded linear operator let

Proposition 6. Let be a bounded linear operator andThen the following statements hold:(i)For and .(ii).(iii) if with .(iv) is countably additive; that is, if is a pairwise disjoint sequence in with .(v).

Proof. (i) Let . Then for , we have and henceso .
(ii) This follows from (i) because if .
(iii) Assume that with . Then for . By the Lebesgue dominated convergence theorem, we obtain that for every . This means that and by (i), .
(iv) Assume that is a pairwise disjoint sequence in with . Let for . Then and . Hence by (iii) .
Statement (v) is obvious.

Definition 7. Let be a bounded linear operator andThen the measure will be called a representing measure of .

Proposition 8. Let be a -continuous linear operator and be its representing measure. Then there exists a Young function such that and .

Proof. According to Theorem 2 there exist a finite set of Young functions with for and such thatLet for . Then is a Young function with andHence

For a linear operator and , let

Now we can state our main result that extends the classical results concerning the integral representation of operators on Lebesgue-Bochner spaces (see [19, § 13, Theorem 1, pp. 259–261]) to operators on Orlicz-Bochner spaces .

Theorem 9. Let be a -continuous linear operator and be its representing measure. Then for the following statements hold:(i) is a -continuous linear operator.(ii)For , one hasand for , one has(iii)For , the measure defined by the equalityis countably additive.(iv)and for , .(v).(vi)For , one hasand for , one has

Proof. (i) Assume that is a net in such that in . Since is a locally solid topology on , we get in . Hence(ii) In view of Proposition 8 there exists a Young function such that and . Then . Let . Then there exists a sequence in such that -a.e. and -a.e. for all (see [21, Theorem 6, p. 4]). Then in because is a Lebesgue topology. Hence . In view of Lemma 4 we can define the integral by the equalitySince and by (i), is -continuous, we getHenceand for , we have (iii) Let and be a sequence in such that . Then for , and hence -a.e. and -a.e. Hence in because is a Lebesgue topology, and by (i) we get (iv) Note that . To show that , assume that . Choose a sequence in such that -a.e. and -a.e. for all . Since is a Lebesgue topology, we have in and hence . Note that .
Let be given. Choose such that . ThenIt follows that , so . Hence for , we easily get (v) Using (iv) we have (vi) This follows from (ii) and (iv).

For a sequence in , we will write if and for every .

Definition 10. A measure with and is said to be -semivariationally -continuous if whenever , .

Using a standard argument we can show the following.

Proposition 11. Let be an additive measure such that and . Then the following statements are equivalent:(i) is -semivariationally -continuous.(ii)The following two conditions hold simultaneously:(a)For every there exists such that whenever , .(b)For every there exists such that .

The following theorem characterizes -semivariationally -continuous representing measures.

Theorem 12. Let be a -continuous linear operator and be its representing measure. Then the following statements are equivalent:(i) is -semivariationally -continuous.(ii) is -continuous.(iii) if in and .(iv) if , .

Proof. (i) (ii) (iii) See [5, Corollary 2.8 and Proposition 1.1].
(i) (iv) This follows from Theorem 9.

Now assume that is a completely regular Hausdorff space. Let denote the -algebra of Baire sets in , which is the -algebra generated by the class of all zero sets of bounded continuous positive functions on . By we denote the family of all cozero (=positive) in (see [25, p. 108]).

Let be a countably additive measure. Then is zero-set regular; that is, for every and there exists with such that (see [25, p. 118]). It follows that for every and there exist , such that .

We can assume that to be complete (if necessary we can take the completion of the measure space ).

Proposition 13. Assume that is a completely regular Hausdorff space and is a complete finite measure space. Let be a -continuous linear operator and be its representing measure. Then the following statements are equivalent:(i) is -semivariationally -continuous.(ii)For every sequence in such that and there exists a sequence in with such that .(iii)For every sequence in such that and there exists a sequence in with such that

Proof. (i) ⇒ (ii) Assume that (i) holds and is a sequence in such that and . Then there exists a sequence in such that and for .
Let be given. Then in view of Proposition 11 there exists such that if with . Choose such that for . Then for . Since , we can choose such that for . Then for , we getthat is, (ii) holds.
(ii) ⇒ (iii) Assume that (ii) holds and is a sequence in such that and . Then there exists a sequence in with such that . Note that, for with for , by Theorem 9 we haveIt follows that (iii) holds.
(iii) ⇒ (i) Assume that (iii) holds and with . Then there exists a sequence in with such thatAssume on the contrary that (i) fails to hold. Then without loss of generality we can assume thatChoose such thatIn view of (54) there exists a pairwise disjoint set in , for and such that andLet . Then and . Then by (55) we get .
On the other hand, in view of (56) we have . This contradiction establishes that (i) holds.

Corollary 14. Assume that is a completely regular Hausdorff space and is complete finite measure space. Let be a -continuous linear operator and be its representing measure. Then is regular; that is, for every and there exist and with such that .

Proof. In view of Theorem 12   is -semivariationally -continuous. Let and be given. Then by Proposition 11 there exists such that whenever and . By the regularity of one can choose and with such that . Hence , as desired.

4. Compact Operators on Orlicz-Bochner Spaces

The following theorem presents necessary conditions for a -continuous operator to be compact.

Theorem 15. Assume that a Young function such that satisfies the -condition. Let be a -continuous linear operator and be its representing measure. If is compact, then is -semivariationally -continuous.

Proof. Assume that is compact and fails to be -semivariationally -continuous. Then there exist and a sequence in with such that for (see Theorem 9). Hence one can choose a sequence in such thatBy Schauder’s theorem the conjugate mapping is compact. Note that for all , where is a closed subspace of the Banach space (see Theorem 2). Then for every there exists such thatHence we obtain that, for each ,Since is a relatively sequentially compact subset of , there exist a subsequence of and such thatChoose such that for . Hence for ,Using (57) and (61), for , we getand henceOn the other hand, since is supposed to satisfy the -condition, we have that (see [26, Theorem 3, pp. 58-59]). This contradiction establishes that is -semivariationally -continuous.

Corollary 16. Assume that is a Young function such that satisfies the -condition. Let be a -continuous linear operator. Then the following statements are equivalent:(i) is compact.(ii) is -compact; that is, there exists a -neighborhood of in such that is a relatively norm compact set in .(iii)There exists a Young function with such that , is a relatively norm compact set in .

Proof. (i) ⇒ (ii) Assume that (i) holds. Then by Theorems 12 and 15   is -continuous. Since the space is quasinormable, by Grothendieck's classical result (see [15, p. 429]), we obtain that is -compact.
(ii) ⇒ (i) The implication is obvious.
(ii) (iii) This follows from Theorem 3.

5. Topology Associated with the -Semivariation of a Representing Measure

Assume that be a -continuous linear operator. Let be its representing measure. Let us putNote that is a seminorm on . Following [22, 27] let stand for the topology on defined by the seminorm restricted to .

The following theorem characterizes -continuous compact operators in terms of the topological properties of the space (see [22, Theorem 3]).

Theorem 17. Let be a -continuous linear operator and be its representing measure. Then the following statements are equivalent:(i)The space is compact.(ii) is compact.

Proof. (i) ⇒ (ii) Assume that is compact. Let be a sequence in . Without loss of generality we can assume that in for some . Then using Theorem 9 for , we haveIt follows that , where . This means that is compact and hence is compact.
(ii) ⇒ (i) Assume that is compact and is a net in . Since is -compact, without loss of generality we can assume that in for some . In view of the compactness of the conjugate operator , there exists a subset of and such that . On the other hand, since is -continuous, we get in . Hence ; that is, .
Let be given. Then there exist a pairwise disjoint set in and for such that andHenceHence , and this means that the space is compact.

As a consequence of Theorems 17 and 15, we have the following.

Corollary 18. Assume that is a Young function such that satisfies the -condition. Let be a -continuous linear operator and be its representing measure. If the space is compact, then is -semivariationally -continuous.

Conflicts of Interest

The author declares that there are no conflicts of interest.