In the paper, we present a necessary and sufficient condition for the existence of a sequence of measurable functions with finite values, which converge to any given essential bounded function in the topology of essential supremum in a Banach space. A new convergence method is proposed, which allows for the discovery of an essential bounded function F that is valued in a Banach space. Generally speaking, there exists a Banach-valued essential bounded function F which Fn can't converge to F in the topology of essential supremum for any sequence of finite-valued measurable function.
Citation: Yiheng Hu, Gang Lyu, Yuanfeng Jin, Qi Liu. Exploration of indispensable Banach-space valued functions[J]. AIMS Mathematics, 2023, 8(11): 27670-27683. doi: 10.3934/math.20231416
[1] | Saak Gabriyelyan . Limited type subsets of locally convex spaces. AIMS Mathematics, 2024, 9(11): 31414-31443. doi: 10.3934/math.20241513 |
[2] | Mahmut Karakuş . Spaces of multiplier σ-convergent vector valued sequences and uniform σ-summability. AIMS Mathematics, 2025, 10(3): 5095-5109. doi: 10.3934/math.2025233 |
[3] | Xinwei Su, Shuqin Zhang, Lixin Zhang . Periodic boundary value problem involving sequential fractional derivatives in Banach space. AIMS Mathematics, 2020, 5(6): 7510-7530. doi: 10.3934/math.2020481 |
[4] | Imran Ali, Faizan Ahmad Khan, Haider Abbas Rizvi, Rais Ahmad, Arvind Kumar Rajpoot . Second order evolutionary partial differential variational-like inequalities. AIMS Mathematics, 2022, 7(9): 16832-16850. doi: 10.3934/math.2022924 |
[5] | Heng Yang, Jiang Zhou . Compactness of commutators of fractional integral operators on ball Banach function spaces. AIMS Mathematics, 2024, 9(2): 3126-3149. doi: 10.3934/math.2024152 |
[6] | Narjes Alabkary . Hyper-instability of Banach algebras. AIMS Mathematics, 2024, 9(6): 14012-14025. doi: 10.3934/math.2024681 |
[7] | Hasanen A. Hammad, Hassen Aydi, Maryam G. Alshehri . Solving hybrid functional-fractional equations originating in biological population dynamics with an effect on infectious diseases. AIMS Mathematics, 2024, 9(6): 14574-14593. doi: 10.3934/math.2024709 |
[8] | Muneerah Al Nuwairan, Ahmed Gamal Ibrahim . Nonlocal impulsive differential equations and inclusions involving Atangana-Baleanu fractional derivative in infinite dimensional spaces. AIMS Mathematics, 2023, 8(5): 11752-11780. doi: 10.3934/math.2023595 |
[9] | Sara Smail, Chafika Belabbaci . A characterization of Wolf and Schechter essential pseudospectra. AIMS Mathematics, 2024, 9(7): 17146-17153. doi: 10.3934/math.2024832 |
[10] | Ateq Alsaadi, Manochehr Kazemi, Mohamed M. A. Metwali . On generalization of Petryshyn's fixed point theorem and its application to the product of n-nonlinear integral equations. AIMS Mathematics, 2023, 8(12): 30562-30573. doi: 10.3934/math.20231562 |
In the paper, we present a necessary and sufficient condition for the existence of a sequence of measurable functions with finite values, which converge to any given essential bounded function in the topology of essential supremum in a Banach space. A new convergence method is proposed, which allows for the discovery of an essential bounded function F that is valued in a Banach space. Generally speaking, there exists a Banach-valued essential bounded function F which Fn can't converge to F in the topology of essential supremum for any sequence of finite-valued measurable function.
The property of Lp(Ω,F,μ;B) will be discussed, here (Ω,F,μ) is a σ-finite measure space, and B is a real Banach space. For 1≤p<∞, Lp(Ω,F,μ;B) is a linear space of all B-valued Bochner Lp integral function with the norm given by the formula
‖F‖Lp(Ω,F,μ;B)≡(∫Ω‖F(ω)‖pBdμ(ω))1p. |
If p=∞, L∞(Ω,F,μ;B) is a linear space of all B-valued essential bounded function with norm defined by letting
‖F‖L∞(Ω,F,μ;B)≡infE∈Fμ(E)=0(supω∈Ec‖F(ω)‖B). |
If B=R, when p∈[1,∞], it is known that there exists a sequence of finite-valued simple measurable function {Fn,n≥1} such that
limn→∞‖F−Fn‖Lp(Ω,F,μ;R)=0, |
for any F∈Lp(Ω,F,μ;R) (see [1,2]). If B is a general Banach space, p∈[1,∞), there exists a sequence of finite-valued simple measurable function Fn such that
limn→∞‖F−Fn‖Lp(Ω,F,μ;B)=0, |
for any F∈Lp(Ω,F,μ;B), and there exists a sequence of countable-valued simple measurable function Fn such that
limn→∞‖F−Fn‖L∞(Ω,F,μ;B)=0, |
for any F∈L∞(Ω,F,μ;B) (see[3]).
The difference between infinite dimension Banach space B and R is that the closed ball of R is compact set and the closed ball of B is non-compact set (see[4]), which makes the property of L∞(Ω,F,μ;B) is very different form L∞(Ω,F,μ;R).
Convergence methods of Banach-valued function were defined in serval ways. For example, Zheng and Cui[5] investigated that l∞(X)- evaluation uniform convergence of operator series can be described completed by the essential bounded subset of l∞(X). Here X is a Banach space,
l∞(X)≡{(xj):xj∈X,supj∈N‖xj‖<∞}, |
and l∞(X) equip the norm of
‖xj‖∞≡supj∈N‖xj‖. |
León-Saavedra considered unconditionally convergence of a series ∑ixi in a Banach space. [6] showed that a series is unconditionally convergent if and only if the series is weakly subseries convergent with respect to a regular linear summability method. Furthermore, this paper unifies several versions of the Orlicz-Pettis theorem that incorporate summability methods. [7] give a another version of the Orlicz-Pettis theorem within the frame of the strong ρ-Cesàro convergence. [8] unified several results which characterize when a series is weakly unconditionally Cauchy (wuc) in terms of the completeness of a convergence space associated with the wuc series. [9] gave a new characterization of weakly unconditionally Cauchy series and unconditionally convergent series through the strong ρ-Cesàro summability is obtained.
In this work, we will present a necessary and sufficient condition for the existence of Fn in L∞(Ω,F,μ;B) for F∈L∞(Ω,F,μ;B), by constructing a sequence of finite-valued measurable functions that converge to F in some sense. A counterexample is also discussed to demonstrate that there exists F∈L∞(Ω,F,μ;B) for which Fn cannot converge to F in the norm topology of L∞(Ω,F,μ;B) for any sequence of finite-valued measurable functions Fn.
The following definitions are about Banach-valued measurable function.
Definition 2.1. [10] If (Ω,F) is a measurable space, B is a Banach space, Ω1,⋯,Ωn∈F are pairwise disjoined nonempty sets, x1,⋯,xn∈B, then the map
F(ω)=n∑i=1xiIΩi(ω), |
is called finite-valued simple function. And the map
F(ω)=∞∑i=1xiIΩi(ω), |
is called countable-valued simple function. A map F:Ω→B is called measurable if ∀A∈B(B),F−1(B)∈F. F is called strongly measurable if there is a sequence of finite-valued simple function Fn such that ∀ω∈Ω,
limn→∞‖F(ω)−Fn(ω)‖B=0. |
Definition 2.2. [11] Let F:Ω→B be a map, for all f∈B∗, the function f(F(ω)) is measurable function on (Ω,F,μ), then F is called weak measurable function on (Ω,F,μ).
The following theorem describes the relationship weak and strong measurable.
Theorem 2.1. (Pettis)[11] Let F:Ω→B be a map, the following assertions are equivalent:
(1) F is strongly measurable.
(2) F is weakly measurable and F(Ω) is almost separable.
By Theorem 2.1, if B is separable space, then F is strongly measurable if and only if it's weakly measurable.
Then the definition of Bochner LP-space is given as follows.
Definition 2.3. [3,10] Let (Ω,F,μ) be a measure space, and let F:Ω→B be a finite-valued simple function with a form of
F(ω)=n∑i=1xiIΩi(ω). |
If ∑ni=1μ(Ωi)<∞, then the Bochner integral of F is defined by
∫ΩF(ω)dμ(ω)=n∑i=1xiμ(Ωi). |
And let F:Ω→B be a strongly measurable function. If there exists a p∈[1,∞) such that
∫Ω‖F(ω)‖pBdμ(ω)<∞, |
then F is called Lp-integrable on (Ω,F,μ). The linear space of all Lp-integrable function with the following seminorm
‖F‖Lp(Ω,F,μ;B)≡(∫Ω‖F(ω)‖pBdμ(ω))1p, |
is denoted by Lp(Ω,F,μ;B). If the function
ω↦‖F(ω)‖B |
is essential bounded, then F is called essential bounded. The linear space of all essential bounded function with the following seminorm
‖F‖L∞(Ω,F,μ;B)≡ess sup{‖F(ω)‖B:ω∈Ω}, |
is denoted by L∞(Ω,F,μ;B).
The following theorems show that the collection of finite-valued function is dense in Lp(Ω,F,μ;B) if p∈[1,∞), and the collection of countable-valued function is dense in L∞(Ω,F,μ;B).
Theorem 2.2. [3] Let (Ω,F,μ) be a measurable space, F:Ω→B is a strongly measurable function, p∈[1,∞), then the following statements are the same in meaning:
(1) F∈Lp(Ω,F,μ;B).
(2) There exists a sequence of finite-valued simple function Fn such that
limn→∞∫Ω‖Fn(ω)−F(ω)‖pBdμ(ω)=0. |
Theorem 2.3. [3] Let (Ω,F,μ) be a measurable space, F:Ω→B be a strongly measurable function, then the following statements are synonymous:
(1) F∈L∞(Ω,F,μ;B).
(2) There exists a sequence of countable-valued simple function Fn such that
limn→∞infE∈Fμ(E)=0(supω∈Ec‖Fn(ω)−F(ω)‖B)=0. |
Theorem 3.1. If (Ω,F,μ) is a measure space, and B is a real Banach space, F∈L∞(Ω,F,μ;B), then the following assertions are equivalent:
(1) There exists a sequence of finite-valued simple function Fn such that
limn→∞‖F−Fn‖L∞(Ω,F,μ;B)=0. |
(2) There exists a measurable set ˜Ω∈F such that μ(˜Ω)=0 and F(˜Ωc) is a sequential compact set.
Proof. If (1) holds, suppose
‖F−Fn‖L∞(Ω,F,μ;B)<12n, |
and
Fn=Kn∑i=1xinIEin, |
where {Ein}Kni=1 are pairwise disjoined and ∪Kni=1Ein=Ω. By the definition of essential bounded, there exists ˜En∈F such that μ(˜En)=0 and
supω∈˜Ecn‖Fn(ω)−F(ω)‖B<1n. |
Considering
˜Ω≡⋃n∈N+(Kn⋃i=1Ein∩˜Ecn)c. |
Then μ(˜Ω)=0. Let
ω∈˜Ωc⊂Kn⋃i=1Ein∩˜Ecn, |
then there exist i=1,⋯,Kn such that
‖xin−F(ω)‖B≤supω∈˜Ecn‖Fn(ω)−F(ω)‖B<1n. |
Therefore, {xin}Kni=1 is a 1/n− web of F(˜Ωc). By the arbitrary of n, F(˜Ωc) is a sequential compact set.
If condition (2) is satisfied, then ∀n∈N+, there exists a finite 1/n− web {xin}Kni=1 of F(˜Ωc). Let
Ein≡{ω∈˜Ωc:‖xin−F(ω)‖B<1n}. |
Let ˜E1n=E1n, and for i>1, defined by
˜Ein≡Ein∖(i−1⋃j=1Ejn). |
Now, let's define a finite-valued function
Fn=Kn∑i=1xinI˜Ein; |
then
‖F−Fn‖L∞(Ω,F,μ;B)≤supω∈˜Ωc‖Fn(ω)−F(ω)‖B<1n. |
By the arbitrary of n, (1) holds.
From now on, suppose B is real Banach space which dual space B∗ is separable, and (Ω,F,μ) is complete measure space. Then B is separable. Let {fn}n∈N+ be countably dense subset of B∗. We define a new convergence.
Definition 3.1. Let Fk,k=1,2,⋯ be a sequence of B-valued strongly measurable function on (Ω,F,μ), we say Fk weakly converge to a B-valued function F almost uniformly if there exists E∈F such that μ(E)=0 and for all weak neighborhood of origin W, there exists N∈N+ such that ∀k>N,
F(ω)−Fk(ω)∈W,∀ω∈Ec. |
We say Fi is a almost uniformly weak Cauchy sequence if there exists E∈F such that μ(E)=0 and for all weak neighborhood of origin W, there exists N∈N+ such that ∀i,j>N,
Fi(ω)−Fj(ω)∈W,∀ω∈Ec. |
Theorem 3.2. (1) Fk weakly converge to F almost uniformly if and only if there exists E∈F such that μ(E)=0, and for all f∈B∗, then
limk→∞supω∈Ec|f(Fk(ω))−f(F(ω))|=0. |
(2) Fk is a almost uniformly weak Cauchy sequence if and only if there exists E∈F such that μ(E)=0, and for all f∈B∗, then
limk→∞supp∈N+supω∈Ec|f(Fk+p(ω))−f(Fn(ω))|=0. |
Proof. We have just proven (1), and likewise, (2) can be demonstrated. Suppose there exists E∈F such that μ(E)=0, and for all weak neighborhood of origin W, there exists N∈N+ such that ∀k>N,
F(ω)−Fk(ω)∈W,∀ω∈Ec. |
Let f∈B∗, given m∈N+, consider the set
Vm≡{x∈B:|f(x)|<1m}. |
Then, for N∈N+ such that ∀k>N,
F(ω)−Fk(ω)∈Vm,∀ω∈Ec. |
That is
supω∈Ec|f(Fk(ω))−f(F(ω))|<1m. |
Let k→∞,
lim supk→∞supω∈Ec|f(Fk(ω))−f(F(ω))|≤1m. |
By the arbitrary of m,
limk→∞supω∈Ec|f(Fk(ω))−f(F(ω))|=0. |
Suppose there exists E∈F such that μ(E)=0, and for all f∈B∗, we have
limk→∞supω∈Ec|f(Fk(ω))−f(F(ω))|=0. |
Given a weak neighborhood of origin W, by the definition of weak topology, there exists g1,⋯,gm∈B∗ and ϵ>0 such that
V≡{x∈B:|g1(x)|<ϵ,⋯,|gm(x)|<ϵ}⊂W. |
Then, for i=1,⋯,m,∃Ni∈N+ such that
supω∈Ec|gi(Fk(ω))−gi(F(ω))|<ϵ,∀n>Ni. |
Let N=max(N1,⋯,Nm), then ∀k>N, we have
Fk(ω)−F(ω)∈V⊂W,∀ω∈Ec. |
Therefore, Fk weakly converge to F almost uniformly.
Theorem 3.3. Let Fk,k=1,2,⋯ be a sequence of B-valued strongly measurable function. If Fk weakly converge to F almost uniformly, then F is strongly measurable.
Proof. By Theorem 2.1, it is sufficient to prove that F is weakly measurable. If Fk weakly converge to F almost uniformly, then there exists E∈F such that μ(E)=0, and ∀f∈B∗,
limk→∞supω∈Ec|f(Fk(ω))−f(F(ω))|=0. |
Therefore, f(FkIEc) pointwise converge to f(FIEc). By the arbitrary of f, FIEc is weakly measurable, thus it is strongly measurable. Because μ(E)=0 and (Ω,F,μ) is complete, FIE is strongly measurable. In summary, F=FIEc+FIE is measurable.
Theorem 3.4. If Fk weakly converge to F′ and F″ almost uniformly, then F′=F″,μ−a.e..
Proof. If Fk weakly converge to F′ and F″ almost uniformly, then there exist E′,E″∈F such that μ(E′)=μ(E″)=0, and
limk→∞supω∈E′c|f(Fk(ω))−f(F(ω))|=limk→∞supω∈E″c|f(Fk(ω))−f(F(ω))|=0. |
Then
μ({ω∈Ω:F′(ω)≠F″(ω)})≤μ(E′∪E″)≤μ(E′)+μ(E″)=0. |
Therefore, F′=F″,μ−a.e..
Theorem 3.5. Let F,Fk,k=1,2,⋯∈L∞(Ω,F,μ;B), then Fk weakly converge to F almost uniformly if and only if
(1) supk∈N+‖Fk‖L∞(Ω,F,μ;B)<∞.
(2) There exists E∈F such that μ(E)=0, and ∀n∈N+,
limk→∞supω∈Ec|fn(Fk(ω))−fn(F(ω))|=0. |
Proof. Suppose Fk weakly converge to F almost uniformly, since (2) is self-evident, we will focus on demonstrating (1). By the conditions, there exists a E∈F such that μ(E)=0 and
limk→∞supω∈Ec|f(Fk(ω))−f(F(ω))|=0. |
In addition, we can suppose supω∈Ec‖F(ω)‖<∞ and supω∈Ec‖Fk(ω)‖<∞(k∈N+). Fixed f∈B∗, then there exists k0∈N+ such that for k≥k0,
supω∈Ec|f(Fk(ω))−f(F(ω))|<1. |
For k≥k0,
supω∈Ec|f(Fk(ω))|≤supω∈Ec|f(Fk(ω))−f(F(ω))|+supω∈Ec|f(F(ω))|≤1+‖f‖B∗supω∈Ec‖F(ω)‖B<∞. |
Therefore,
sup{|f(Fk(ω))|:k∈N+,ω∈Ec}≤max{‖f‖B∗supω∈Ec‖F1(ω)‖B,⋯,‖f‖B∗supω∈Ec‖Fk0(ω)‖B,1+‖f‖B∗supω∈Ec‖F(ω)‖B}<∞. |
By Uniform Boundedness Principle,
sup{‖Fk(ω)‖B:k∈N+,ω∈Ec}<∞. |
Thus, supk∈N+‖Fk‖L∞(Ω,F,μ;B)<∞. Now we suppose (1) and (2) are true, then there exists a E∈F such that μ(E)=0 and ∀n∈N+,
limk→∞supω∈Ec|fn(Fk(ω))−fn(F(ω))|=0. |
We can assume
M≡supω∈Ec‖F(ω)‖B+supk∈N+supω∈Ec‖Fk(ω)‖B<∞. |
Fixed f∈B∗, then ∀ϵ>0,∃n0∈N+ such that
‖f−fn0‖B∗<ϵ2M. |
Then ∀k∈N+,∀ω∈Ec,
|f(Fk(ω))−f(F(ω))|≤|f(Fk(ω))−fn0(Fk(ω))|+|fn0(Fk(ω))−fn0(F(ω))|+|fn0(F(ω))−fn0(F(ω))|≤‖f−fn0‖B∗‖Fk(ω)‖B+|fn0(Fk(ω))−fn0(F(ω))|+‖f−fn0‖B∗‖F(ω)‖B<ϵ+|fn0(Fk(ω))−fn0(F(ω))|. |
By the arbitrary of ω,
supω∈Ec|f(Fk(ω))−f(F(ω))|<ϵ+supω∈Ec|fn0(Fk(ω))−fn0(F(ω))|. |
Therefore,
lim supk→∞supω∈Ec|f(Fk(ω))−f(F(ω))|<ϵ+lim supk→∞supω∈Ec|fn0(Fk(ω))−fn0(F(ω))|≤ϵ, |
By the arbitrary of ϵ,
lim supk→∞supω∈Ec|f(Fk(ω))−f(F(ω))|=0. |
Finally, by the arbitrary of f, Fk weakly converge to F almost uniformly
Theorem 3.6. Suppose F is a B-valued strongly measurable function on (Ω,F,μ), then F∈L∞(Ω,F,μ;B) if and only if there exists a sequence of finite-valued simple function Fk such that Fk weakly converge to F almost uniformly.
Proof. Suppose F∈L∞(Ω,F,μ;B), then there is a E∈F such that μ(E)=0, and
M≡supω∈Ec‖F(ω)‖B<∞. |
By Banach-Alaoglu theorem, F(Ec) is weak relatively compact sets. Let k∈N+, and
Vk≡{x∈B:|f1(x)|<1k,⋯,|fk(x)|<1k}. |
Then there exist ∀k∈N+,∃{xik}Nki=1⊂F(Ec) such that
{F(ω):ω∈Ec}⊂Nk⋃i=1(xik+Vk). |
Let
Eik≡{ω∈Ec:F(ω)−xik∈Vk}. |
and ˜E1k=E1k, for i>1, we can define
˜Eik≡Eik∖(i−1⋃j=1Ejk). |
We can construct a finite-valued measurable function
Fk=Kk∑i=1xikI˜Eik. |
Because Fk(Ec)⊂F(Ec),
sup{‖Fk(ω)‖B:ω∈Ec,k∈N+}≤M<∞. |
For all n∈N+, for k>n,
supω∈Ec|fn(Fk(ω))−fn(F(ω))|<1k. |
Therefore,
limk→∞supω∈Ec|fn(Fk(ω))−fn(F(ω))|=0. |
By Theorem 3.5, Fk weakly converge to F almost uniformly.
Consider the possibility that a sequence of finite-valued simple function Fk that weakly converge to F almost uniformly, then there exists a E∈F such that μ(E)=0 and ∀f∈B∗,
limn→∞supω∈Ec|f(Fk(ω))−f(F(ω))|=0. |
Then there exists a k0∈N+ such that
supω∈Ec|f(Fk0(ω))−f(F(ω))|<1. |
Because Fk0 is finite-valued function,
supω∈Ec|f(F(ω))|≤supω∈Ec|f(Fk0(ω))−f(F(ω))|+supω∈Ec|f(Fk0(ω))|≤1+‖f‖B∗supω∈Ec‖Fk0(ω)‖B<∞. |
By Uniform Boundedness Principle,
supω∈Ec‖F(ω)‖B<∞. |
Therefore, F∈L∞(Ω,F,μ;B).
Now we will proof L∞(Ω,F,μ;B) is complete in the sense of weak convergence almost uniformly.
Theorem 3.7. Let Fk,k=1,2,⋯∈L∞(Ω,F,μ;B) be a almost uniformly weak Cauchy sequence, then ∃F∈L∞(Ω,F,μ;B) such that Fk weakly converge to F almost uniformly.
Proof. Suppose Fk∈L∞(Ω,F,μ;B) is a almost uniformly weak Cauchy sequence, then there exists a E∈F such that μ(E)=0 and for all k∈N+,
supω∈Ec‖Fk(ω)‖B<∞. |
And ∀f∈B∗,
limn→∞supp∈N+supω∈Ec|f(Fn+p(ω))−f(Fn(ω))|=0. |
Fixed f∈B∗,∃k0∈N+ such that ∀k>k0,
supω∈Ec|f(Fk(ω))−f(Fk0(ω))|<1. |
For i=1,⋯,k0,
supω∈Ec|f(Fi(ω))|≤‖f‖B∗supω∈Ec‖Fi(ω)‖B. |
Therefore,
sup{|f(Fk(ω))|:k∈N+,ω∈Ec}≤1+‖f‖B∗max(supω∈Ec‖F1(ω)‖B,⋯,supω∈Ec‖Fk0(ω)‖B). |
By Uniform Boundedness Principle, there exists M∈(0,∞) such that
{Fk(ω):k∈N+,ω∈Ec}⊂{x∈B:‖x‖B≤M}. |
Fixed ω∈Ec, {Fk(ω)}k∈N+ is a bounded sequence, which means it's a weak relatively compact sequence. Therefore exists a subsequence {Fki(ω)}i∈N+ and F(ω)∈B such that Fki(ω) weakly converge to F(ω). Because {Fk(ω)}k∈N+ is a weak cauchy sequence, ∀f∈B∗,∀ϵ>0,∃K∈N+ such that ∀m,n≥K,
|f(Fm(ω))−f(Fn(ω))|<ϵ2. |
Meanwhile, ∃i0∈N+ such that ki0>N and
|f(Fki0(ω))−f(F(ω))|<ϵ2, |
when k>ki0,
|f(Fk(ω))−f(F(ω))|≤|f(Fk(ω))−fki0(F(ω))|+|f(Fki0(ω))−f(F(ω))|<ϵ. |
which means Fk(ω) weakly converge to F(ω). By Mazur Theorem,
F(ω)∈¯co{Fk(ω)}k∈N+⊂{x∈B:‖x‖B≤M}, |
where ¯co{Fk(ω)}k∈N+ is convex hull of {Fk(ω)}k∈N+ in norm topology of B. Therefore, F is essential bounded, and
‖F‖L∞(Ω,F,μ;B)≤supω∈Ec‖F(ω)‖B≤M. |
Finally, Fk weakly converge to F almost uniformly can be show. Fixed f∈B∗, there exists a {Fki}i∈N such that
supω∈Ec|f(Fki(ω))−f(Fki−1(ω))|<12i. |
Then
f(Fk0(ω))+∞∑i=1(f(Fki(ω))−f(Fki−1(ω)))=f(F(ω)). |
Let j→∞,
supω∈Ec|f(Fkj(ω))−f(F(ω))|≤∞∑i=j+1supω∈Ec|f(Fki(ω))−f(Fki−1(ω))|≤∞∑i=j+112i→0. |
∀ϵ>0,∃K∈N+ such that ∀m,n≥K,
supω∈Ec|f(Fm(ω))−f(Fn(ω))|<ϵ2. |
Meanwhile, ∃i0∈N+ such that ki0>N and
supω∈Ec|f(Fki0(ω))−f(F(ω))|<ϵ2. |
Therefore, for k>ki0, we have
supω∈Ec|f(Fk(ω))−f(F(ω))|≤supω∈Ec|f(Fk(ω))−fki0(F(ω))|+supω∈Ec|f(Fki0(ω))−f(F(ω))|<ϵ, |
which means Fk weakly converge to F almost uniformly.
Finally, we will provide a counterexample to demonstrate that there exists an F∈L∞(Ω;B) for which Fn cannot converge to F in the norm topology of L∞(Ω;B) for any sequence of finite-valued measurable functions Fn.
Let H=L2([−π,π],B([−π,π]),l), where l is Lebesgue measure. It is obvious that H is real separable Hilbert space and H′=H by Riesz Representation Theorem, which means the dual space of H is separable.
Let n∈N,
h0≡1√2π,h2n−1(x)=1√πsinnx,h2n(x)=1√πcosnx,n∈N+. |
Then {hn}n∈N is the unit orthogonal basis of H, and
‖hn−hm‖2H=2,∀n≠m. |
A measure space ([0,1],B([0,1]),l) is given. Let C be the Cantor set of [0,1], and the countable connected component of Cc denote by {En}n∈N+.
F≡∞∑n=1hnIEn. |
Then F∈L∞([0,1],B([0,1]),l;H), and
‖F‖L∞([0,1],B([0,1]),l;H)=1. |
However, given any zero measure set E, F(Ec)={hn}n∈N, which means F(Ec) in not a sequential compact set. Therefore, any sequence of finite-valued measurable function cannot converge to F in norm topology of L∞([0,1],B([0,1]),l;H). Let
Fk≡h0IC+k∑n=1hnIEn. |
Then Fk are finite-valued measurable functions. Given m∈N+, for k>m,
supω∈Cc⟨hm,F(ω)−Fk(ω)⟩H=supω∈Cc⟨hm,∞∑n=k+1hnIEn(ω)⟩H=0, |
and l(C)=0, which means Fk weakly converge to F almost uniformly.
In this paper, we give a necessary and sufficient condition for the existence of a sequence of finite-valued measurable function which converge to F∈L∞(Ω,F,μ;B) in topology of essential supremum and give a new convergence which makes any F∈L∞(Ω,F,μ;B) can find sequence of finite-valued measurable function which converge to F. On this basis, we can raise some valuable problems:
(1) Definition 3.1 and the proof of Theorem 3.2 to Theorem 3.6 depend on the separability of B∗, if we can extended the conclusion to the condition that B∗ is a general Banach space?
(2) If we can use a topology to convergence defined in Definition 3.1. For example, let f∈B∗, we define the seminorm pf by
pf(F)=‖f(F)‖L∞(Ω,F,μ;R). |
If we can prove that pf(F)=0 for all f implies ‖F‖L∞(Ω,F,μ;B)=0, then the the family P≡{pf:f∈B∗} can determine a new topology to characterize the convergence.
(3) Based on (2) and Theorem 3.2, we guess the following assertions are equivalent:
(a) Fk weakly converge to F almost uniformly.
(b) For all f∈B∗, we have
limn→∞‖f(Fk)−f(F)‖L∞(Ω,F,μ;R)=0. |
(c) There exists E∈F such that μ(E)=0, and for all f∈B∗, we have
limn→∞supω∈Ec|f(Fk(ω))−f(F(ω))|=0. |
Here F,Fk,k=1,2,⋯∈L∞(Ω,F,μ;B).
In the work, a necessary and sufficient condition for the existence of a sequence of finite-valued measurable function which converge to any given F∈L∞(Ω,F,μ;B) is given. A new convergence is defined. In this convergence, any F∈L∞(Ω,F,μ;B) has a sequence of finite-valued measurable function which converge to F. Finally, a counterexample is also given to show that there exists F∈L∞(Ω,F,μ;B) for which Fn cannot converge to F in the norm topology of L∞(Ω,F,μ;B) for any sequence of finite-valued measurable functions Fn.
The authors declare they have not used Artificial Intelligence (AI) tools in the creation of this article.
This work was supported by National Natural Science Foundation of China (No. 11761074), Project of Jilin Science and Technology Development for Leading Talent of Science and Technology Innovation in Middle and Young and Team Project (No. 20200301053RQ).
The authors declare no conflict of interest.
[1] | W. Rudin, Real and complex analysis, In: The mathematical gazette, New York: McGraw-Hill, 52 (1974), 412. https://doi.org/doi:10.2307/3611894" target="_blank">10.2307/3611894">https://doi.org/doi:10.2307/3611894 |
[2] | J. A. Yan, Lecture notes on measure theory (Chinese), 2 Eds., Beijing: Science Press, 2004. |
[3] | M. Kreuter, Sobolev spaces of vector-valued functions, Ulam university, 2015. |
[4] | G. Q. Zhang, Y. Q. Lin, The lecture of functional analysis, Peking University Press, 1990. |
[5] |
F. Zheng, C. Cui, A theorem on uniform convergence of operator series, J. Bohai Univ. (Nat. Sci. Ed.), 28 (2007), 338–339. https://doi.org/10.3969/j.issn.1673-0569.2007.04.010 doi: 10.3969/j.issn.1673-0569.2007.04.010
![]() |
[6] |
F. León-Saavedra, M. P. R. Rosa, S. Antonio, Orlicz-Pettis theorem through summability methods, Mathematics, 7 (2019), 895. https://doi.org/10.3390/math7100895 doi: 10.3390/math7100895
![]() |
[7] |
F. León-Saavedra, S. Moreno-Pulido, A. Sala, Orlicz-Pettis type theorems via strong ρ-Cesaro convergence, Numer. Funct. Anal. Optim., 40 (2019), 798–802. https://doi.org/10.1080/01630563.2018.1554587 doi: 10.1080/01630563.2018.1554587
![]() |
[8] |
F. León-Saavedra, F. J. Pérez-Fernández, F. P. R. Rosa, A. Sala, Ideal convergence and completeness of a normed space, Mathematics, 7 (2019), 897. https://doi.org/10.3390/math7100897 doi: 10.3390/math7100897
![]() |
[9] |
F. León-Saavedra, S. Moreno-Pulido, A. Sala-Pérez, Completeness of a normed space via strong ρ-Cesàro summability, Filomat, 33 (2019), 3013–3022. https://doi.org/10.2298/FIL1910013L doi: 10.2298/FIL1910013L
![]() |
[10] |
J. Diestel, J. J. Uhl, Jr., Vector measures, American mathematical society, 15 (1977). http://doi.org/10.1090/surv/015 doi: 10.1090/surv/015
![]() |
[11] | D. X. Xia, S. Z. Yan, W. C. Shu, Y. S. Tong, Functional analysis (Second tutorial) (Chinese), Higher education press, 2008. |