• Nie Znaleziono Wyników

Characteristic of monotonicity of Orlicz function spaces equipped with the Orlicz norm

N/A
N/A
Protected

Academic year: 2021

Share "Characteristic of monotonicity of Orlicz function spaces equipped with the Orlicz norm"

Copied!
12
0
0

Pełen tekst

(1)

P. Foralewski, H. Hudzik, R. Kaczmarek, M. Krbec

Characteristic of monotonicity of Orlicz function spaces equipped with the Orlicz norm

Dedicated to Professor Julian Musielak on his 85th birthday

Abstract. First we prove that the property of strict monotonicity of a Köthe space (E,k.kE)and/or of its Köthe dual (E0,k.kE0)can be used successfully to compare the supports of x ∈ E\{θ} and y ∈ S(E0), where < x, y >= kxkE. Next we prove that any element x ∈ S+(E)with µ(T \ supp x) = 0 is a point of order smoothness in E, whenever E is an order continuous Köthe space. Next, the characteristic of monotonicity of Orlicz function spaces equipped with the Orlicz norm is conside- red. In the case when the measure is finite, the Orlicz function Φ vanishes outside zero, supu>0[A|u| − Φ(u)] = ∞, where A = limu→∞Φ(u)

u and Φ satisfies the ∆2- condition at ∞, we give only some lower and upper estimates for this characteristic.

In all others (easier) cases the characteristic is calculated.

2010 Mathematics Subject Classification: 46B42, 46E30.

Key words and phrases: Orlicz space, Orlicz norm, Köthe space, Köthe dual, cha- racteristic of monotonicity, strict monotonicity, point of order smoothness.

1. Introduction. The monotonicity properties of normed lattices have been introduced and studied by Birkhoff [3] in the context of their geometric structure.

Next, Akcoglu and Sucheston [1] discovered the connection of strict and uniform monotonicity to the ergodic theory.

It is well known that monotonicity properties of normed lattices are restrictions of respective rotundity properties to the couples of comparable elements in the positive cone X+of X (see [13]). In consequence, the role of monotonicity properties in the dominated best approximation problems is similar to the role of rotundity properties, when the more general (not dominated) best approximation problems are considered in Banach spaces (see [14], [24], [6]). The problem of calculating the characteristic of monotonicity of a Banach lattice seems to be of great interest because of the result of Betiuk-Pilarska and Prus [2] stating that if a Banach lattice

(2)

X has this characteristic strictly smaller then 1 and X is weakly orthogonal, then X has the weak normal structure. Consequently, X has the weak fixed point property (for the definition we refer to [19]). Therefore, it was quite natural that several papers were devoted to the problem of calculating the exact value or getting good estimates of the characteristic of monotonicity in a concrete specific class of Banach lattices (see [27], [9] or [10]).

It is also worth noticing that monotonicity properties have close relationships to complex rotundities and their applications. Namely, for any real Köthe space E, strict monotonicity and uniform monotonicity of E coincide, respectively, with complex rotundity and complex uniform rotundity of the complexification Ecof E.

This result was obtained independently by Hudzik and Narloch [18] and Lee [25].

In [14] the concepts of lower and upper local uniform monotonicities were inve- stigated and it was shown that these monotonicity properties are different. In turn, Kolwicz and Płuciennik [20] introduced a new monotonicity property of normed lat- tices, called uniform monotonicity in every order interval and they discovered that this property is useful in order to give a criterion for uniform rotundity in every direction of Calder´on-Lozanovski˘ı spaces.

Recently, Wójtowicz [32] proved that if X is a Dedekind σ-complete Banach lattice and its dual X is strictly monotone, then for any x ∈ X\{0} all supporting functionals at |x| are positive.

For more results connected with monotonicity properties we refer to the papers [7, 8, 12, 15].

Let us denote S+(X) = S(X)∩ X+, where X+is the positive cone of a Banach lattice X = (X, ≤, k.k) and S(X) is the unit sphere of X.

A Banach lattice X is said to be strictly monotone, if kyk < kxk for all x, y ∈ X+ such that y ≤ x and y 6= x. A Banach lattice X is said to be uniformly monotone if for any ε ∈ (0, 1) there is δ(ε) ∈ (0, 1) such that kx − yk ≤ 1 − δ(ε) whenever 0≤ y ≤ x, kxk = 1 and kyk ≥ ε.

For a given Banach lattice X, the function δm,X: [0, 1]→ [0, 1] defined by δm,X(ε) = inf{1 − kx − yk : 0 ≤ y ≤ x, kxk = 1, kyk ≥ ε}

is called the lower modulus of monotonicity of X. It is known (see [11]) that δm,X(ε) = inf{1 − kx − yk : 0 ≤ y ≤ x, kxk = 1, kyk ≥ ε}

= inf{1 − kx − yk : 0 ≤ y ≤ x, kxk = 1, kyk = ε}

= inf{1 − kx − yk : 0 ≤ y ≤ x, kxk ≤ 1, kyk ≥ ε}

= inf{1 − kx − yk : 0 ≤ y ≤ x, kxk ≤ 1, kyk = ε}

for any ε ∈ (0, 1). The lower modulus of monotonicity δm,X is a non-decreasing and convex function on the interval [0, 1] (see [23]), so δm,X is continuous on the interval [0, 1). It is also clear that δm,X(ε)≤ ε for any ε ∈ [0, 1]. Obviously, X is uniformly monotone if and only if δm,X(ε) > 0for every ε ∈ (0, 1]. It is also easy to see that a Banach lattice X is strictly monotone if and only if δm,X(1) = 1.

The number ε0,m(X)∈ [0, 1] defined by

ε0,m(X) = sup{ε ∈ [0, 1): δm,X(ε) = 0} = inf{ε ∈ (0, 1): δm,X(ε) > 0},

(3)

(where inf ∅ := 1) is said to be the characteristic of monotonicity of X. A Banach lattice X is uniformly monotone if and only if ε0,m(X) = 0.

Let us recall some useful facts.

Theorem 1.1 ([11, Theorem 5]) For any Banach lattice X the following formula for the characteristic of monotonicity holds true

ε0,m(X) = sup{lim sup

n→∞ kxn− ynk : kxnk = 1 ∧ 0 ≤ yn ≤ xn∀ n ∈ N ∧ kynk → 1}.

The relation between the characteristic of monotonicity and the modulus of monotonicity reveals the following

Theorem 1.2 ([9, Theorem 2.1]) For any Banach lattice X the following equ- ality is true

ε0,m(X) = 1− δm,X(1), where δm,X(1) = lim

ε→1δm,X(ε). Moreover,

δm,X(1− δm,X(ε)) = 1− ε

for arbitrary ε ∈ (ε0,m(X), 1] if ε0,m(X) < 1 as well as also in the case when ε = ε0,m(X) = 1.

2. Some remarks on strict monotonicity and order smoothness in Köthe spaces. Let us denote by (T, Σ, µ) a positive, complete and σ−finite measure space and by L0 = L0(T, Σ, µ) the space of all (equivalence classes of) real-valued and Σ−measurable functions defined on T . For two functions x, y ∈ L0 we write y ≤ x if and only if y(t) ≤ x(t) µ−a.e. in T .

By E = (E, ≤, k · kE)we denote a Köthe space over the measure space (T, Σ, µ), that is, E is a Banach subspace of L0 which satisfies the following conditions:

(i) if |y| ≤ |x|, x ∈ E and y ∈ L0, then y ∈ E and kykE ≤ kxkE, (ii) there exists a function x ∈ E which is strictly positive µ−a.e. in T .

Let E0 be the Köthe dual of E, that is, E0 =

y∈ L0(T, Σ, µ) : x· y ∈ L1(T, Σ, µ)for any x ∈ E , and let E0 be equipped with the norm

kykE0 = sup

kxkE≤1

Z

T

x(t)y(t)dµ

= sup

kxkE≤1

Z

T

|x(t)y(t)| dµ = sup

kxkE≤1kx · ykL1 for any y ∈ E0.

(4)

Remark 2.1 Let us assume that x ∈ E\ {0}, y ∈ E0, kykE0 = 1andR

T

x(t)y(t)dµ = kxkE. Then:

10 supp x⊂ supp y up to a set of measure zero whenever E is strictly monotone, 20 supp y⊂ supp x up to a set of measure zero whenever E0 is strictly monotone, 30 supp x = supp y up to a set of measure zero if both E and E0 are strictly

monotone.

Proof We start from 10. Let us define the functional ξy: E → R by the formula

ξy(z) =< z, y >=

Z

T

z(t)y(t)dµfor any z ∈ E.

It is obvious that kξyk = kykE0 by the definitions of these two norms. Moreover, we have ξy(x) = kxkE, whence ξy

 x kxkE

 = 1 = yk = kykE0. This means that the supremum in the definition of the norm kykE0 is attained at the point

x

kxkE. Therefore, it must be x(t)y(t) ≥ 0 µ-a.e. in T because otherwise defining

˜

x(t) = kxkx(t)Esgn y(t) µ−a.e. in T , we get ξyx) > ξy

 x kxkE

, a contradiction. It is obvious that

ξy(xχsupp x∩supp y) = ξy(x) =kxkE,

whence by the inequality ξy(z) ≤ kξyk kzkE for any z ∈ E, we conclude that kxχsupp x∩supp ykE =kxkE. By the assumption that E is strictly monotone, we get that supp x ⊂ supp y up to a set of measure zero.

20. By the equality ξsupp x∩supp y(x) =kxkE, we get kyχsupp x∩supp ykE0 = 1 =kykE0.

Therefore, the assumption that E0 is strictly monotone gives that supp y ⊂ supp x up to a set of measure zero.

30. This follows by 10 and 20. 

A point x ∈ S+(E), where E is a Köthe space, is said to be order smooth if Grad(x) contains no order interval which is not a singleton. Let us recall that Grad(x) consists of all functionals x ∈ S(E)(the unit sphere of the topological dual Eof E) such that x(x) =kxkE. A Köthe space E is said to be order smooth if all points x ∈ S+(E)are points of order smoothness.

Recall that a Köthe space E is called order continuous if for any element x ∈ E and any sequence (xn) in E+ (the positive cone in E) with 0 ≤ xn ≤ |x| for all n∈ N and xn→ 0 µ-a.e., there holds kxnkE → 0.

Remark 2.2 If E is a Köthe space with an order continuous norm k · kE, then any x∈ S+(E)with µ(T \ supp x) = 0 is a point of order smoothness.

(5)

Proof Let the assumptions about x be satisfied and let y, z ∈ Grad(x) and 0 ≤ y ≤ z. Since E is order continuous, so E= E0 isometrically, whence there exist y, z ∈ S(E0)such that

y(w) = Z

T

w(t)y(t)dµand z(w) = Z

T

w(t)z(t)dµ

for any w ∈ E. Since y, z∈ Grad(x), we have kyk = kzk = y(x) = z(x) = 1.

This yields that x(t)y(t) ≥ 0 and x(t)z(t) ≥ 0 µ−a.e. in T . Since y(t) ≤ z(t) µ−a.e.

in T , we also have x(t)y(t) ≤ x(t)z(t) µ−a.e. in T . Since µ(T \ supp x) = 0, it must be y(t) = z(t) µ−a.e. in T because otherwise y(t) < z(t) and so x(t)y(t) < x(t)z(t) on a set of positive measure, whence

kxkE= y(x) = Z

T

x(t)y(t)dµ <

Z

T

x(t)z(t)dµ = z(x) =kxkE,

a contradiction, which finishes the proof. 

3. Characteristic of monotonicity in Orlicz spaces equipped with the Orlicz norm. A map Φ : R → [0, ∞] is said to be an Orlicz function if Φ is a nonzero function which is convex, even, continuous and vanishing at zero and left continuous on R+. Let us note that the left continuity of an Orlicz function Φ on R+ is equivalent to the fact that limu→b(Φ)Φ(u) = Φ(b(Φ)), where b(Φ) := sup{u >

0 : Φ(u) <∞}. We will also use another parameter a(Φ) := sup{u > 0 : Φ(u) = 0}

of the Orlicz function Φ.

For any Orlicz function Φ we define its complementary function in the sense of Young by the formula

Ψ(u) = sup

v>0{|u|v − Φ(v)}

for all u ∈ R. It is easy to show that Ψ is also an Orlicz function.

Given any Orlicz function Φ we define on L0 = L0(T, Σ, µ) a convex modular IΦby

IΦ(x) = Z

T

Φ(x(t))dµ

(see [4, 22, 26, 28, 29, 31]). The Orlicz space LΦ = LΦ(T, Σ, µ) generated by an Orlicz function Φ is defined as

LΦ={x ∈ L0: IΦ(λx) < +∞ for some λ > 0}.

Orlicz spaces LΦ are usually considered as Banach spaces equipped with the Lu- xemburg norm

kxkΦ= infn

λ > 0 : IΦ x λ

≤ 1o

or with the Orlicz norm

kxkoΦ= sup Z

T

x(t)y(t)dµ

: IΨ(y)≤ 1

 .

(6)

It is known (see [22] for so-called N-functions and [16] in general) that for any Orlicz function Φ the following Amemiya formula

kxkoΦ= inf

k>0

1

k(1 + IΦ(kx))

for the Orlicz norm holds. Moreover, the Amemiya formula for the Luxemburg norm (see [30] or [16]) is the following

kxkΦ= inf

k>0

 max 1

k{1, IΦ(kx)}

 . Therefore, the inequalities

(1) kxkΦ≤ kxkoΦ≤ 2 kxkΦ,

which are true for any x ∈ LΦ are obvious. Orlicz spaces are Köthe spaces under both (the Luxemburg and the Orlicz) norms when they are considered with the natural semi-order ≤.

We say that an Orlicz function Φ satisfies condition ∆2for all u ∈ R+(at infinity) if there is K > 0 such that the inequality Φ(2u) ≤ KΦ(u) holds for all u ∈ R (for all u ∈ R satisfying |u| ≥ u0 with some u0 > 0 such that Φ(u0) <∞). We write then Φ ∈ ∆2(R+) (Φ∈ ∆2(∞)), respectively. Let us note that Φ ∈ ∆2(R+)implies that a(Φ) = 0 and b(Φ) = ∞. Analogously, Φ ∈ ∆2(∞) implies b(Φ) = ∞.

Given any x ∈ LΦ, let K(x) be the set of all positive numbers k such that kxkoΦ= 1

k(1 + IΦ(kx)) and let

k(x) = inf{k > 0: IΨ(p(k|x|)) ≥ 1} , k∗∗(x) = sup{k > 0: IΨ(p(k|x|)) ≤ 1} ,

where p is the right hand-side derivative of Φ. It is known that K(x) = [k(x), k∗∗(x)], whenever k∗∗(x) <∞ and K(x) = [k(x),∞) whenever k(x) <∞ and k∗∗(x) =

∞. In the second case we also have that kxkoΦ = limk→∞1kIΦ(kx). It is obvious that K(x) 6= ∅ if and only if k(x) <∞. It is known that k(x) <∞ (equivalen- tly K(x) 6= ∅) for any x ∈ S(LΦ)if and only if supu>0[A|u| − Φ(u)] = ∞, where A = limu→∞Φ(u)u (see [5]). It is obvious that this condition is satisfied if A = ∞.

However, there are Orlicz functions with A < ∞ and supu>0[A|u| − Φ(u)] = ∞ (see [5]).

Now, we will present some results concerning the characteristic of monotonicity in Orlicz spaces LΦo = (LΦ,k · koΦ). Recall that analogous results for the spaces (LΦ,k·kΦ)have been presented in [9].

Lemma 3.1 Assume that µ(T ) < ∞, Φ is an Orlicz function with a(Φ) > 0 and satisfying the condition ∆2(∞) and let c ∈ (a(Φ), ∞). Then for any ε ∈ (0, 1) there exists δ(ε) ∈ (0, 1) such that if x ∈ LΦ, |x(t)| ≥ c for µ−a.e. t ∈ supp x and IΦ(x)≤ δ(ε), then kxkoΦ≤ ε.

(7)

This lemma follows immediately from [9, Lemma 3.1] and inequalities (1).

Lemma 3.2 Let µ(T ) < ∞, supu>0[A|u| − Φ(u)] = ∞, where A = limu→∞Φ(u) u , a(Φ) > 0 and Φ ∈ ∆2(∞). Then

1a(Φ)

l(Φ)TkoΦ≤ δm,LΦo(1)≤ 1 −a(Φ) k(Φ)TkoΦ, where

k(Φ) := inf{lim infn→∞ k(yn) : yn∈ S+(LΦo)for any n ∈ N and limn→∞µ(supp yn) = 0} and

l(Φ) := inf{k(x) : x∈ S(LΦo)}.

Proof Let us note that δm,LΦo(1) = 1−sup {kx − ykoΦ: 0≤ y ≤ x, kykoΦ=kxkoΦ= 1}.

Assume now that 0 ≤ y ≤ x, kykoΦ=kxkoΦ= 1and let k ∈ K(x). Then k ≥ 1 and, by superadditivity of the modular IΦon L0

+, 1 = kxkoΦ= 1

k(1 + IΦ(kx))

= 1

k(1 + IΦ(k(x− y) + ky))

1

k(1 + IΦ(k(x− y)) + IΦ(ky))

= 1

k(1 + IΦ(ky)) + 1

kIΦ(k (x− y))

≥ kykoΦ+ 1

kIΦ(k(x− y)) = 1 +1

kIΦ(k(x− y)),

whence IΦ(k(x− y)) = 0, and consequently x − y ≤ a(Φ)k χT. Therefore, kx − ykoΦ

a(Φ)

k TkoΦ. By the arbitrariness of the functions x, y ∈ S(LΦo)such that 0 ≤ y ≤ x, we get the inequality

sup{kx − ykoΦ: 0≤ y ≤ x, kykoΦ=kxkoΦ= 1} ≤ a(Φ)

l(Φ) TkoΦ, i.e.

(2) δm,LΦo(1)≥ 1 −a(Φ)

l(Φ) TkoΦ.

By the definition of the number k(Φ) there exists a sequence (yn)n=1in S+(LΦo)such that µ(supp yn)→ 0 and k(yn)→ k(Φ). Let us denote for simplicity k(yn)by kn

and supp yn by An, and define the sequence of functions (xn)n=1 in the following way

xn = yn+a(Φ) kn

χT\An

(8)

for any n ∈ N. It is clear that 1 = kynkoΦ≤ kxnkoΦ for any n ∈ N. Moreover,

kxnkoΦ 1 kn

(1 + IΦ(knxn)) = 1 kn

(1 + IΦ(knyn)) =kynkoΦ, so kxnkoΦ=kynkoΦ= 1. Moreover,

kxn− ynkoΦ= a(Φ)

kn

χT\An

o

Φ

= a(Φ) kn

χT\An

o

Φ a(Φ) k(Φ)TkoΦ

as n → N, which follows by the facts that kn → k(Φ), µ(An)→ 0 as n → ∞ and the element χT is order continuous (because of the assumptions that µ(T ) < ∞ and Φ∈ ∆2(∞), whence b(Φ) = ∞). Consequently, we have

sup{kx − ykoΦ: 0≤ y ≤ x, kykoΦ=kxkoΦ= 1} ≥ limn→∞kxn− ynkoΦ= a(Φ)

k(Φ)TkoΦ, i.e.

(3) δm,LΦo(1)≤ 1 −a(Φ)

k(Φ)TkoΦ.

Combining inequalities (2) and (3), we get our thesis. 

Remark 3.3 The formula for kχAkoΦ is the following:

TkoΦ= µ(T )Ψ−1

 1 µ(T )

 .

Really, it is known (see [21], p. 107) that for any symmetric space E and its Köthe dual E0 the equality kχAkEAkE0 = AkL1 = µ(A) holds for any A ∈ Σ with 0 < µ(A) < ∞. It is also known that the Köthe dual of LΦo = (LΦ,k · koΦ) is (LΨ,k·kΨ). Denoting λ(A) = 1/Ψ−1(1/µ(A))for any A ∈ Σ with 0 < µ(A) < ∞, we have IΨ

 χA

λ(A)

= 1, whence kχAkΨ= λ(A). In consequence, kχAkoΦ· λ(A) = µ(A), whence kχAkoΦ= µ(A)λ(A) = µ(A)Ψ−1(1/µ(A)).

Remark 3.4 Let us note that if µ(T ) < ∞, supu>0[A|u| − Φ(u)] = ∞, where A = limu→∞Φ(u)

u , Φ ∈ ∆2(∞) and Ψ ∈ ∆2(∞), then the constant k(Φ) from Lemma 3.2 is finite. Indeed, in [4, Theorem 1.35(2)] it has been shown that if µ(T ) < ∞, the Orlicz function Φ vanishes only at 0, Φ has finite values on the whole R, limu→0Φ(u)

u = 0, limu→∞Φ(u)

u = ∞ and Ψ ∈ ∆2(∞), then sup{k >

0 : k ∈ K(x), x ∈ S(LΦo)} < ∞. It is easy to show that Theorem 1.35(2) from [4]

remains also true if Φ has finite values on the whole R and supu>0[A|u| − Φ(u)] = ∞ (also if a(Φ) > 0). In consequence, taking any sequence (yn)n=1from S+(LΦo)with limn→∞µ(supp yn) = 0, we get that the sequence (k(yn))n=1 is bounded, whence lim infn→∞ k(yn) <∞ and k(Φ) ≤ lim infn→∞ k(yn).

(9)

Theorem 3.5 Let µ(T ) < ∞ and Φ be an Orlicz function. Then the following statements hold true:

a) If supu>0[A|u| − Φ(u)] = ∞, where A = limu→∞Φ(u)

u , Φ ∈ ∆2(∞) and a(Φ) = 0, then ε0,m(LΦo) = 0.

b) If supu>0[A|u| − Φ(u)] = ∞, Φ ∈ ∆2(∞), Ψ ∈ ∆2(∞) and a(Φ) > 0, then

(4) a(Φ)

k(Φ)TkoΦ≤ ε0,m(LΦo)a(Φ)

l(Φ) TkoΦ,

where the constants k(Φ) and l(Φ) are defined as in Lemma 3.2. In the case when Ψ /∈ ∆2(∞), we have

(5) a(Φ)

k(Φ)TkoΦ≤ ε0,m(LΦo)≤ a(Φ) kχTkoΦ, c) If Φ /∈ ∆2(∞), then ε0,m(LΦo) = 1.

Proof a) Proceeding analogously as in [14] we get that the space LΦo is uniformly monotone (since the condition limu→∞Φ(u)

u = ∞ that was used in [14], can be replaced successfully by supu>0[A|u| − Φ(u)] = ∞), whence we have immediately ε0,m(LΦo) = 0.

b)Now we will show formula (4). Assume that 0 ≤ yn ≤ xn, kxnkoΦ= 1for all n∈ N and kynkoΦ→ 1 as n → ∞. Let kn∈ K(xn)for any n ∈ N. By superadditivity of the modular IΦon L0

+, we have 1 = kxnkoΦ= 1

kn

(1 + IΦ(knxn))

= 1

kn

(1 + IΦ(kn(xn− yn) + knyn))

1

kn

(1 + IΦ(kn(xn− yn)) + IΦ(knyn))

= 1

kn

(1 + IΦ(knyn)) + 1 kn

IΦ(kn(xn− yn))

≥ kynkoΦ+ 1 kn

IΦ(kn(xn− yn)) .

The assumption that kynkoΦ → 1 as n → ∞ and the above inequality yield that

1

knIΦ(kn(xn− yn))→ 0 as n → ∞. Since sup{kn: n∈ N} < ∞ (see Remark 3.4), we have IΦ(kn(xn− yn))→ 0.

Take arbitrary ε > 0 and define the sets

An={t ∈ T : |kn(xn(t)− yn(t))| ≥ (1 + ε)a(Φ)} (∀ n ∈ N).

Obviously, IΦ(kn(xn− yn) χAn)→ 0 as n → ∞, whence by virtue of Lemma 3.1,

(10)

we obtain that kkn(xn− ynAnkoΦ→ 0 as n → ∞. Hence, we have l(Φ)· lim sup

n→∞ k(xn− yn)koΦ ≤ lim sup

n→∞ kkn(xn− yn)koΦ

(6)

≤ lim sup

n→∞

kn(xn− yn) χT\An

o

Φ

+ lim sup

n→∞ kkn(xn− yn) χAnkoΦ

= lim sup

n→∞

kn(xn− yn) χT\An

o

Φ

≤ lim sup

n→∞ (1 + ε)a(Φ) χT\An

o

Φ

≤ (1 + ε)a(Φ) kχTkoΦ. Since ε > 0 was arbitrary, we get the inequality

lim sup

n→∞ kxn− ynkoΦ a(Φ)

l(Φ) TkoΦ.

By the arbitrariness of (xn)n=1 and (yn)n=1with the desired properties and The- orem 1.1, we obtain ε0,m LΦo

a(Φ)l(Φ)TkoΦ.

Moreover, by Theorem 1.2 and Lemma 3.2, we have ε0,m LΦo

= 1− δm,LΦo(1)≥ 1 − δm,LΦo(1)

≥ 1 −



1a(Φ) k(Φ)TkoΦ



=a(Φ)

k(Φ)TkoΦ.

If we do not assume that Ψ ∈ ∆2(∞), then analogously as in the proof of (4), we get k1nIΦ(kn(xn− yn))→ 0 as n → ∞. Since Φ(u)u is a nondecreasing function with respect to u, we obtain that for any x ∈ L0(T, Σ, µ)the function f(k) := 1kIΦ(kx) is nondecreasing with respect to k. Hence

0≤ IΦ(xn− yn) 1 kn

IΦ(kn(xn− yn))

for any n ∈ N, whence IΦ(xn− yn)→ 0 as n → ∞. Proceeding analogously as in (6) and defining An = {t ∈ T : |(xn(t)− yn(t))| ≥ (1 + ε)a(Φ)} for any n ∈ N, we get ε0,m LΦo

≤ a(Φ) kχTkoΦ.

In order to prove c) let us notice that if an Orlicz function Φ does not satisfy condition ∆2(∞), then the Orlicz space LΦo contains an order almost isometric copy of `(see [17]). Consequently, ε0,m LΦo= 1. 

Theorem 3.6 Let µ(T ) = ∞, then ε0,m(LΦo) = 0whenever supu>0[A|u| − Φ(u)] =

∞, where A = limu→∞Φ(u)

u and Φ ∈ ∆2(R+), and ε0,m(LΦo) = 1 if Φ /∈ ∆2(R+).

Proof Condition Φ ∈ ∆2(R+)implies that a(Φ) = 0. Therefore, using the condi- tion supu>0[A|u| − Φ(u)] = ∞ instead of limu→∞Φ(u)

u =∞, we can prove in the same way as in [14] that the space LΦo is uniformly monotone and, in consequence, ε0,m(LΦo) = 0. If Φ /∈ ∆2(R+), then LΦo contains an order almost isometric copy of

`(see [17]), whence ε0,m(LΦo) = 1. 

(11)

References

[1] M.A. Akcoglu and L. Sucheston, On uniform monotonicity of norms and ergodic theorems in function spaces, Rend. Circ. Mat.8(2) (1985), 325–335.

[2] A. Betiuk-Pilarska and S. Prus, Banach lattices which are order uniformly noncreasy, J. Math. Anal. Appl.342 (2008), 1271–1279.

[3] G. Birkhoff, Lattice Theory, Amer. Math. Soc., Providence, RI, 1979.

[4] S. Chen, Geometry of Orlicz spaces, Dissertationes Math.356 (1996), 1–204.

[5] S. T. Chen, Y. A. Cui and H. Hudzik, Isometric copies of `1and `in Orlicz spaces equipped with the Orlicz norm, Proc. Amer. Math. Soc.132(2) (2004), 473–480.

[6] S. T. Chen, X. Hea, H. Hudzik and A. Kamińska, Monotonicity and best approximation in Orlicz–Sobolev spaces with the Luxemburg norm, J. Math. Anal. Appl.344 (2008), 687–698.

[7] M. Ciesielski, A. Kamińska, P. Kolwicz and R. Płuciennik, Monotonicity and rotundity of Lorentz spaces Γp,ω, Nonlinear Analysis, Theory, Methods & Applications 75(5) (2012), 2713–2723.

[8] M. Ciesielski, P. Kolwicz and A. Panfil, Local monotonicity structure of symmetric spaces with applications, J. Math. Anal. Appl.409 (2014), 649–662.

[9] P. Foralewski, H. Hudzik, R. Kaczmarek and M. Krbec, Moduli and characteristics of mono- tonicity in some Banach lattices, Fixed Point Theory and Applications, Volume 2010 (2010), Article ID 852346, doi:10.1155//2010//852346, 22 pages.

[10] P. Foralewski, H. Hudzik, R. Kaczmarek, M. Krbec and M. Wójtowicz, On the moduli and characteristic of monotonicity in Orlicz-Lorentz function spaces, Journal of Convex Analysis 20(4) (2013), 955–970.

[11] H. Hudzik and R. Kaczmarek, Moduli and characteristics of monotonicity in general Banach lattices and in Orlicz spaces in particular, Nonlinear Analysis, Theory, Methods & Applica- tions70 (2009), 3407–3423.

[12] H. Hudzik and A. Kamińska, Monotonicity properties of Lorentz spaces, Proc. Amer. Math.

Soc.123(9) (1995), 2715–2721.

[13] H. Hudzik, A. Kamińska and M. Mastyło, Monotonicity and rotundity properties in Banach lattices, Rocky Mountain J. Math.30(3) (2000), 933–950.

[14] H. Hudzik and W. Kurc, Monotonicity properties of Musielak–Orlicz spaces and dominated best approximation in Banach lattices, Journal of Approximation Theory95 (1998), 353–368.

[15] H. Hudzik, X.B. Liu and T.F. Wang, Points of monotonicity in Musielak–Orlicz function spaces endowed with the Luxemburg norm, Arch. Math.82 (2004), 534–545.

[16] H. Hudzik and L. Maligranda, Amemiya norm equals Orlicz norm in general, Indag. Math., N.S.11(4) (2000), 573–585.

[17] H. Hudzik and M. Mastyło, Almost isometric copies of l in some Banach spaces, Proc.

Amer. Math. Soc.119(1) (1993), 209–215.

[18] H. Hudzik and A. Narloch, Relationships between monotonicity and complex rotundity pro- perties with some consequences, Math. Scand.96 (2005), 289–306.

[19] W. A. Kirk and B. Sims, Handbook of metric fixed point theory, Kluwer Academic Publishers (2001).

(12)

[20] P. Kolwicz and R. Płuciennik, On uniform rotundity in every direction in Calder´on- Lozanovski˘ı sequence spaces, Journal of Convex Analysis14(3), (2007), 621–645.

[21] S. G. Krein, Ju. I. Petunin, E. M. Semenov, Interpolation of linear operators, Translations of Mathematical Monographs, 54. American Mathematical Society, Providence, R.I., 1982.

[22] M. A. Krasnoselski˘ı and Ya. B. Ruticki˘ı, Convex Functions and Orlicz Spaces, Groningen, Nordhoff, 1961 (translation).

[23] W. Kurc, A dual property to uniform monotonicity in Banach lattices, Collect. Math.44(1- 3) (1993), 155–165.

[24] W. Kurc, Strictly and uniformly monotone Musielak–Orlicz spaces and applications to best approximation, Journal of Approximation Theory69(2) (1992), 173–187.

[25] H. Ju Lee, Monotonicity and complex convexity in Banach lattices, J. Math. Anal. Appl.

307(1) (2005), 86–101.

[26] W. A. J. Luxemburg, Banach Function Spaces, Thesis, Delft, 1955.

[27] Y.M. Lü, J.M. Wang and T.F. Wang, Monotone coefficients and monotonicity of Orlicz spaces, Rev. Mat. Complut.12(1) (1999), 105–114.

[28] L. Maligranda, Orlicz Spaces and Interpolation, (Semin´arias de Mathem´atica: vol.5), Cam- pinas (Brazil) Univ. Estadual de Campinas, 1989.

[29] J. Musielak, Orlicz Spaces and Modular Spaces, Lecture Notes in Math. 1034, Springer- Verlag, Berlin-Heidelberg-New York-Tokyo, 1983.

[30] W. Orlicz, A note on modular spaces, I. Bull. Acad. Polon. Sci., S´er. Sci. Math. Astronom.

Phys.9, 157-162 (1961); Reprinted in: W. Orlicz, Collected Papers, PWN-Polish Scientific Publishers, Warszawa, 1142–1147 (1988).

[31] M. M. Rao and Z. D. Ren, Theory of Orlicz Spaces, Marcel Dekker Inc., New York-Basel-Hong Kong, 1991.

[32] M. Wójtowicz, The positive aspects of smoothness in Banach lattices, Positivity17(2) (2013), 257–263.

P. Foralewski

Adam Mickiewicz University, Faculty of Mathematics and Computer Science Umultowska 87, 61-614 Poznań, Poland

E-mail: katon@amu.edu.pl H. Hudzik

Adam Mickiewicz University, Faculty of Mathematics and Computer Science Umultowska 87, 61-614 Poznań, Poland

E-mail: hudzik@amu.edu.pl R. Kaczmarek

Adam Mickiewicz University, Faculty of Mathematics and Computer Science Umultowska 87, 61-614 Poznań, Poland

E-mail: radekk@amu.edu.pl M. Krbec

Institute of Mathematics, Academy of Sciences of the Czech Republic Zitná 25, CZ-115 67 Prague 1, Czech Republic

(Received: 20.11.2013)

Cytaty

Powiązane dokumenty

We present description of Banach envelope of Hardy–Orlicz spaces of an annulus, genereted by Orlicz functions well estimated by

It is interesting that the theory of Orlicz spaces offers subspaces and topologies which are convenient for these purposes (cf. Moreover, Condition В implies

Meskine, Existence of solutions for elliptic equations having natural growth terms in orlicz spaces, Abstr.. Meskine, Strongly nonlinear parabolic equations with natural growth terms

Key words and phrases: normal pregenfunction, Musielak-Orlicz sequence space, completeness, separability.. In what follows α, γ, δ, ε denote positive numbers, and j, k, m, n -

It was proved, under some minor assumptions, that Orlicz space generated by the function Φ and equipped with the Luxemburg norm ∥⋅∥ Φ is conjugate to the Orlicz space generated by

Criteria for k-strict convexity, uniform convexity in every direction, prop- erty K, property H, and property G in Musielak-Orlicz sequence spaces and their subspaces endowed with

ANNALES SOCIETAT1S MATHEMATICAE POLONAE Series I: COMMENTATIONES MATHEMATICAE XXVII (1987) ROCZNIKI POLSKIEGO TOWARZYSTWA MATEMATYCZNEGO.. Séria I: PRACE MATEMATYCZNE

ROCZNIKI PQLSKIEGO TOWARZYSTWA MATEMATYCZNEGO Séria I: PRACE MATEMATYCZNE XXIV (1984)M. Hence in this case p is