Absolutely continuous operators on function spaces and vector measures

Let (Ω, Σ, μ) be a finite atomless measure space, and let E be an ideal of L0(μ) such that $${L^\infty(\mu) \subset E \subset L^1(\mu)}$$. We study absolutely continuous linear operators from E to a locally convex Hausdorff space $${(X, \xi)}$$. Moreover, we examine the relationships between μ-absolutely continuous vector measures m : Σ → X and the corresponding integration operators Tm : L∞(μ) → X. In particular, we characterize relatively compact sets $${\mathcal{M}}$$ in caμ(Σ, X) (= the space of all μ-absolutely continuous measures m : Σ → X) for the topology $${\mathcal{T}_s}$$ of simple convergence in terms of the topological properties of the corresponding set $${\{T_m : m \in \mathcal{M}\}}$$ of absolutely continuous operators. We derive a generalized Vitali–Hahn–Saks type theorem for absolutely continuous operators T : L∞(μ) → X.


Introduction and terminology
For terminology concerning vector lattices and function spaces we refer the reader to [1], [2], [10]. Throughout the paper we assume that ( , , μ) is a complete finite atomless measure space and L 0 (μ) denotes the corresponding space of μ-equivalence classes of all -measurable real valued-functions defined on . Let E be an M. Nowak (B) Faculty of Mathematics, Computer Science and Econometrics, University of Zielona Góra, ul. Szafrana 4A, 65-516 Zielona Gora, Poland e-mail: M.Nowak@wmie.uz.zgora.pl ideal of L 0 (μ) such that L ∞ (μ) ⊂ E ⊂ L 1 (μ), and let E ∼ and E ∼ n stand for the order dual and order continuous dual of E respectively. Then E ∼ n separates the points of E and it can be identified with the Köthe dual E of E through the mapping E v → ϕ v ∈ E ∼ n , where ϕ v (u) = uvdμ for all u ∈ E. It is known that the Mackey topology τ (E, E ∼ n )(= τ (E, E )) is a locally solid Lebesgue topology. The so-called order-bounded topology τ 0 can be defined on E as the finest locally convex topology on E for which every order interval in E is a bounded set (see [11]). A local base B 0 at zero for τ 0 is the class of all absolutely convex subsets of E that absorb all order bounded sets in E. Then τ 0 coincides with the Mackey topology τ (E, E ∼ ). Note that if u n , u ∈ E and u n → u uniformly on , then u n → u for τ 0 .
From now on we assume that (X, ξ) is a locally convex Hausdorff space (for short, lcHs) and let P ξ denote the set of all ξ -continuous seminorms on X. By X ξ we denote the topological dual of (X, ξ). We denote by σ (L , K ) and τ (L , K ) the weak topology and the Mackey topology on L with respect to a dual pair L , K .
Recall that a linear operator T : E → X is said to be order-bounded (resp. orderweakly compact), if for each u ∈ E + , the set T ( [−u, u]) is ξ -bounded (resp. relatively σ (X, X ξ )-compact) in X (see [6] Following [13] a linear operator T : E → X is said to be absolutely continuous if for each u ∈ E, T (1 A n u) → 0 for ξ whenever μ(A n ) → 0, (A n ) ⊂ . Absolutely continuous operators on Orlicz spaces and Frechét function spaces have been examined by Orlicz and Wnuk (see [12,13]).
In Sect. 2 we study absolutely continuous operators T : E → X. We show that a linear operator T : E → X is absolutely continuous if and only if T is (τ (E, E ∼ n ), ξ )continuous. We characterize relatively compact sets in the space L τ,ξ (E, X ) of all (τ (E, E ∼ n ), ξ )-continuous linear operators T : E → X, provided with the topology of simple convergence. In Sect. 3 we examine the relationships between μ-absolutely continuous vector measures m : → X and the corresponding integration operators T m : L ∞ (μ) → X.

Absolutely continuous operators on function spaces
We start with the following result.

Proposition 2.1 Assume that T : E → X is an absolutely continuous linear operator.
Then T is (τ 0 , ξ)-continuous.
Proof In view of Proposition 1.1 it is sufficient to show that T ( [−u, u]) is ξ -bounded in X for every u ∈ E + . For this purpose one can repeat the proof of Theorem 1 in [13]. Now we present a characterization of absolutely continuous operators on E.

Proposition 2.2
For a linear operator T : E → X the following statements are equivalent: (iv) ⇒(v) It is obvious.
(v)⇐⇒(vi) It is enough to repeat the reasoning in the proof of Proposition 4 in [13] and use Proposition 2.1 and the fact that u n → 0 in E for τ 0 whenever u n → 0 uniformly on .
, ξ )-continuous linear operators from E to X, equipped with the topology T s of simple convergence. Let P ξ be the family of all ξ -continuous seminorms on X. Then T s is generated by the family {q p,u : p ∈ P ξ , u ∈ E} of seminorms, where q p,u (T ) = p(T (u)) for all T ∈ L τ,ξ (E, X ).
The following result will be of importance (see [15,Theorem 2]).
n . Now we can state a characterization of relative T s -compactness in L τ,ξ (E, X ). X ). Then the following statements are equivalent: Let W be an absolutely convex and ξ -closed neighbourhood of 0 for ξ in X. Then the polar W 0 of W (with respect to the dual pair E, E ξ ), is a σ (X ξ , X )-closed and ξ -equicontinuous subset of X ξ (see [1,Theorem 9.21]). Then by Theorem 2.4 the set H = {x • T :

Corollary 2.6
Assume that K is a relatively T s -compact subset of L τ,ξ (E, X ). Then K is uniformly μ-absolutely continuous, i.e., for each u ∈ E and p ∈ P ξ we have Proof In view of Theorem 2.4, K is (τ (E, E ∼ n ), ξ )-equicontinuous. Let p ∈ P ξ and ε > 0 be given. Then there exists a τ (E, E ∼ n )-neigbourhood V of 0 in E such that for each T ∈ K we have p(T (u)) ≤ ε for all u ∈ V. Let u ∈ E and μ(A n ) → 0 and let u n = 1 A n u for n ∈ N. Note that u n → 0(μ) and |u n (ω)| ≤ |u(ω)|μ-a.e. for all n ∈ N. Hence by the Riesz theorem for every subsequence (u k n ) of (u n ) there exists a subsequence (u l kn ) of (u k n ) such that u l kn (ω) → 0μ-a.e. This means that u l kn is a Lebesgue topology. It follows that u n → 0 for τ (E, E ∼ n ). Then there exists n ε ∈ N such u n ∈ V for n ≥ n ε , and hence sup T ∈K p(T (1 A n u)) ≤ ε for n ≥ n ε .

Absolutely continuous vector measures
Let (X, ξ) be a quasicomplete lcHs and m : → X be a ξ -bounded vector measure (i.e., the range of m is ξ -bounded in X ) and m(A) = 0 if μ(A) = 0, A ∈ (in symbols, m μ).
Then the integral udm is well defined and the corresponding integration operator T m : L ∞ (μ) → X given by T m (u) = udm is ( · ∞ , ξ)-continuous and linear, and for each x ∈ X ξ , (see [9], [14,Lemma 6]). Conversely, let T : L ∞ (μ) → X be a ( · ∞ , ξ)-continuous linear operator, and let m(A) = T (1 A ) for A ∈ . Then m : → X is a ξ -bounded vector measure such that m μ (called the representing measure of T ) and T m (u) = T (u) for all u ∈ L ∞ (μ).
An important example of a quasicomplete lcHs is the space L(Y, Z ) of all bounded linear operators between Banach spaces Y and Z , provided with the strong operator topology.
Recall that a vector measure m : → X is said to be μ-absolutely continuous m(A n ) → 0 for ξ whenever μ(A n ) → 0, (A n ) ⊂ (see [5,Definition 3,p. 11]). Now we characterize μ-absolutely continuous measures in terms of the properties of the corresponding integration operators.

Proposition 3.1 Assume that (X, ξ) is a quasicomplete lcHs. Let m :
→ X be a ξ -bounded vector measure such that m μ. Then the following statements are equivalent: and this means that x • T m ∈ L ∞ (μ) ∼ n .
As a consequence of Proposition 3.1 we get the following Pettis type theorem for countably additive measures (see [5, Theorem 1, p. 10]). that (X, ξ) is a quasicomplete lcHs. Let m : → X be a ξ -countably additive measure. Then the following statements are equivalent:

Corollary 3.2 Assume
Let ca( , X ) stand for the space of all ξ -countably additive measures m : → X. By ca μ ( , X ) we denote the subspace of ca( , X ) consisting of all m ∈ ca( , X ) that are μ-absolutely continuous. Denote by T s the topology of simple convergence in ca( , X ). Then T s is generated by the family {q p,A : p ∈ P ξ , A ∈ } of seminorms, where for all m ∈ ca( , X ).

Proposition 3.3 ca μ ( , X ) is a closed set in (ca( , X ), T s ).
Proof Let m ∈ ca( , X ) and m ∈ cl T s (ca μ ( , X )). Then there is a net (m α ) in ca μ ( , X ) such that m α → m for T s , i.e., for each p ∈ P ξ and A ∈ we have Now we establish some terminology (see [14, pp. 92-93]). For p ∈ P ξ let X p = (X, p) be the associated seminormed space. Denote by ( X p , · ∼ p ) the completion of the quotient normed space X/ p −1 (0). Let p : X p → X/ p −1 (0) ⊂ X p be the canonical quotient map.
Given a vector measure m : → X with m μ, let m p : → X p be given by Then m p is a Banach space-valued measure on . We define the p-variation m p of m by where m p denotes the semivariation of m p : → X p . Note that m is ξ -bounded if and only if m p ( ) < ∞ for each ξ -continuous seminorm p on X. Moreover, we have (see [14,Lemma 7]): For a subset M of ca μ ( , X ) let Now we are ready to state a characterization of relative compactness in the space (ca μ ( , X ), T s ) in terms of the topological properties of the set K M (see [8,Theorem 7], [15,Theorem 8], [16, Theorem 2.1]). (X, ξ) be a quasicomplete lcHs. Then for a set M in ca μ ( , X ) the following statements are equivalent: This means that the family M is uniformly μ-absolutely continuous.

Theorem 3.4 Let
(iii) ⇒(iv) Assume that (iii) holds. Then M ⊂ ca μ ( , X ) ⊂ ca( , X ) and M is a uniformly ξ -countably additive set in ca( , X ). Hence by [8,Theorem 7] M is a relatively compact set in (ca( , X ), T s ). Since ca μ ( , X ) is closed in (ca( , X ), T s ), we obtain that M is a relatively compact set in (ca μ ( , X ), T s ).
Open Access This article is distributed under the terms of the Creative Commons Attribution License which permits any use, distribution, and reproduction in any medium, provided the original author(s) and the source are credited.