A study was conducted on the existence of solutions for a class of nonlinear Caputo type higher-order fractional Langevin equations with mixed boundary conditions on a star graph with k+1 nodes and k edges. By applying a variable transformation, a system of fractional differential equations with mixed boundary conditions and different domains was converted into an equivalent system with identical boundary conditions and domains. Subsequently, the existence and uniqueness of solutions were verified using Krasnoselskii's fixed point theorem and Banach's contraction principle. In addition, the stability results of different types of solutions for the system were further discussed. Finally, two examples are illustrated to reinforce the main study outcomes.
Citation: Gang Chen, Jinbo Ni, Xinyu Fu. Existence, and Ulam's types stability of higher-order fractional Langevin equations on a star graph[J]. AIMS Mathematics, 2024, 9(5): 11877-11909. doi: 10.3934/math.2024581
[1] | Guijun Xing, Huatao Chen, Zahra S. Aghayan, Jingfei Jiang, Juan L. G. Guirao . Tracking control for a class of fractional order uncertain systems with time-delay based on composite nonlinear feedback control. AIMS Mathematics, 2024, 9(5): 13058-13076. doi: 10.3934/math.2024637 |
[2] | Jingjing Yang, Jianqiu Lu . Stabilization in distribution of hybrid stochastic differential delay equations with Lévy noise by discrete-time state feedback controls. AIMS Mathematics, 2025, 10(2): 3457-3483. doi: 10.3934/math.2025160 |
[3] | Kareem Alanazi, Omar Naifar, Raouf Fakhfakh, Abdellatif Ben Makhlouf . Innovative observer design for nonlinear systems using Caputo fractional derivative with respect to another function. AIMS Mathematics, 2024, 9(12): 35533-35550. doi: 10.3934/math.20241686 |
[4] | Junsoo Lee, Wassim M. Haddad . On finite-time stability and stabilization of nonlinear hybrid dynamical systems. AIMS Mathematics, 2021, 6(6): 5535-5562. doi: 10.3934/math.2021328 |
[5] | Bin Hang, Weiwei Deng . Finite-time adaptive prescribed performance DSC for pure feedback nonlinear systems with input quantization and unmodeled dynamics. AIMS Mathematics, 2024, 9(3): 6803-6831. doi: 10.3934/math.2024332 |
[6] | Miao Xiao, Zhe Lin, Qian Jiang, Dingcheng Yang, Xiongfeng Deng . Neural network-based adaptive finite-time tracking control for multiple inputs uncertain nonlinear systems with positive odd integer powers and unknown multiple faults. AIMS Mathematics, 2025, 10(3): 4819-4841. doi: 10.3934/math.2025221 |
[7] | Tan Zhang, Pianpian Yan . Asymmetric integral barrier function-based tracking control of constrained robots. AIMS Mathematics, 2024, 9(1): 319-339. doi: 10.3934/math.2024019 |
[8] | Liping Yin, Yangyu Zhu, Yangbo Xu, Tao Li . Dynamic optimal operational control for complex systems with nonlinear external loop disturbances. AIMS Mathematics, 2022, 7(9): 16673-16691. doi: 10.3934/math.2022914 |
[9] | Taewan Kim, Jung Hoon Kim . A new optimal control approach to uncertain Euler-Lagrange equations: H∞ disturbance estimator and generalized H2 tracking controller. AIMS Mathematics, 2024, 9(12): 34466-34487. doi: 10.3934/math.20241642 |
[10] | Zhiqiang Chen, Alexander Yurievich Krasnov . Disturbance observer based fixed time sliding mode control for a class of uncertain second-order nonlinear systems. AIMS Mathematics, 2025, 10(3): 6745-6763. doi: 10.3934/math.2025309 |
A study was conducted on the existence of solutions for a class of nonlinear Caputo type higher-order fractional Langevin equations with mixed boundary conditions on a star graph with k+1 nodes and k edges. By applying a variable transformation, a system of fractional differential equations with mixed boundary conditions and different domains was converted into an equivalent system with identical boundary conditions and domains. Subsequently, the existence and uniqueness of solutions were verified using Krasnoselskii's fixed point theorem and Banach's contraction principle. In addition, the stability results of different types of solutions for the system were further discussed. Finally, two examples are illustrated to reinforce the main study outcomes.
Unless otherwise stated, X will stand for an infinite Tychonoff space. We assume that all linear spaces are over the field R of real numbers and all locally convex spaces are Hausdorff. We denote by Cp(X) the ring C(X) of real-valued continuous functions on X endowed with the pointwise topology τp, and by Ck(X) the space C(X) equipped with the compact-open topology τk. The subspace of C(X) of uniformly bounded functions is represented by Cb(X). The topological dual of Cp(X) is denoted by L(X), or by Lp(X) when equipped with the weak* topology σ(L(X),C(X)). If we provide L(X) with the strong topology β(L(X),C(X)), we shall write Lβ(X), and shall refer to Lβ(X) as the strong dual of Cp(X). A family {Aα:α∈NN} of subsets of a set X is called a resolution for X if it covers X and verifies that Aα⊆Aβ for α≤β. If the sets Aα are compact, we speak of a compact resolution. If E is a locally convex space and {Aα:α∈NN} is a resolution for E consisting of bounded sets (bounded in the locally convex sense [5,1.4.5 Definition]), we say that {Aα:α∈NN} is a bounded resolution. A bounded resolution for E that swallows the bounded sets of E will be referred to as a fundamental bounded resolution. Recall that a Banach disk B in a locally convex space E is an absolutely convex bounded set such that EB, the linear span of B equipped with the Minkowski functional of B as a norm, is a Banach space. A locally convex space E is said to be a quasi-(LB) -space (in the sense of Valdivia) if it admits a resolution {Aα:α∈NN} consisting of Banach disks [33].
If Y is a topological subspace of X, the locally convex space Cp(Y|X) is defined in [2,0.4] as the range (the image) R(T) of Cp(X) in Cp(Y) under the restriction map T:C(X)→C(Y), given by f↦f|Y, equipped with the relative topology of Cp(Y). Note that T is a linear and continuous map. Clearly, C(Y|X)=R(T) consists of those f∈C(Y) that admit a continuous extension to the whole of X. Observe that Cp(Y|X) is always a dense linear subspace of Cp(Y), so the topological dual of Cp(Y|X) coincides algebraically with L(Y). However, since the weak* topology σ(L(Y),C(Y)) on L(Y) is stronger than the weak* topology σ(L(Y),C(Y|X)), we shall denote by Lp(Y|X) the space L(Y) equipped with this second topology to distinguish it from Lp(Y), which as we know represents the space L(Y) with the topology σ(L(Y),C(Y)). In what follows we shall denote by M(X) the bidual of Cp(X), i.e., the dual of Lβ(X). Likewise, to be consistent with our notation, we shall represent by M(Y|X) the bidual of Cp(Y|X).
If Y is dense in X, then the restriction map T:Cp(X)→Cp(Y) is continuous and one-to-one, i.e., it is a so-called condensation map from Cp(X) onto Cp(Y|X). In this case, Cp(Y|X) is linearly homeomorphic to the space (C(X),τp(Y)), where τp(Y) denotes the locally convex topology on C(X) of the pointwise convergence on Y. Moreover, the adjoint map T∗:Lp(Y|X)→Lp(X) carries Lp(Y|X) onto a dense linear subspace of Lp(X). If Y is closed in X, then Cp(Y|X) is an open subspace of Cp(Y) [2,0.4.1 Proposition]; consequently, in this case the restriction map T is in addition a quotient map. If Y is C-embedded in X (which happens in particular if Y is a compact set in X or if X is normal and Y closed), then T is onto and hence Cp(Y|X)=Cp(Y).
We denote by δ(Y) the canonical copy of Y consisting of delta measures, i.e., δ(Y)={δy:y∈Y}. The set δ(Y) looks different when regarded as a subset of Lp(Y) or a subset of Lp(Y|X), since in the former case each δy acts as a continuous linear functional on Cp(Y), and in the latter each δy acts as a continuous linear functional on Cp(Y|X). Of course, in the first case δ(Y) is the standard homeomorphic copy of Y in Lp(Y). When a function f∈C(Y|X) is regarded as a linear functional on w∈Lp(Y|X), we shall write ⟨f,w⟩. So, if w=∑ni=1aiδyi∈L(Y) with ai∈R and yi∈Y for 1≤i≤n, then we have ⟨f,w⟩:=∑ni=1aif(yi).
Note that C(Y|X)⊆C(Y)⊆RY, which implies that M(Y|X)⊆M(Y)⊆RY. However, the following property holds.
Theorem 1.1. [18,Corollary 7] The strong duals of Cp(Y|X) and Cp(Y) coincide, as do their biduals M(Y|X) and M(Y).
In other words, if we denote by Lβ(Y|X) the dual space L(Y) provided with the strong topology β(L(Y),C(Y|X)) of the dual pair ⟨L(Y),C(Y|X)⟩, it turns out that Lβ(Y|X)=Lβ(Y). Hence, M(Y|X)=M(Y) and, consequently, the weak* bidual of Cp(Y), i.e., the dual space M(Y) equipped with the relative product topology of RY coincides with the weak* bidual of Cp(Y|X). So, the complete picture is
C(Y|X)⊆C(Y)⊆M(Y|X)=M(Y)⊆RY. |
This result, along with the next three, will be used along the paper.
Theorem 1.2. [13,Theorem 28] The weak* bidual of Cp(X) admits a bounded resolution if and only if |X|=ℵ0.
Recall that the envelope E={A(α∣n):α∈Σ,n∈N} of a family {Aα:α∈Σ} of subsets of a locally convex space E with Σ⊆NN, where
A(α∣n)=⋃{Aβ:β∈Σ,β(i)=α(i),1≤i≤n} |
is called limited if for each α∈Σ and each absolutely convex neighborhood of the origin U in E there is n∈N with A(α∣n)⊆nU. A covering {Aα:α∈Σ} with Σ⊆NN of a set X is called a Σ-covering of X.
Theorem 1.3. [10,Lemma 2] Let E be a locally convex space. If E admits a Σ -covering {Aα:α∈Σ} with Σ⊆NN of limited envelope, then there exists a Lindelöf Σ-space Z such that E′⊆Z⊆RE.
In the following result, υX stands for the Hewitt realcompactification of X, as usual.
Theorem 1.4. [10,Theorem 3] The space Cp(X) admits a Σ-covering of limited envelope if and only if υX is a Lindelöf Σ -space.
It is worth mentioning that each bounded resolution for E is always a Σ -covering of E of limited envelope [10,Proposition 12]. So, in particular, any compact resolution for E is a Σ-covering of E of limited envelope.
Although Cp(Y|X) is a continuous linear image of Cp(X) and a dense linear subspace of Cp(Y), the topological properties of X and Y induced by certain covering properties of Cp(Y|X) are not always the same as the topological properties of X and Y induced by analogous covering properties of Cp(X) or Cp(Y). For instance, if Cp(X) is σ-countably compact, then X is finite [31]. But if Cp(Y|X) is σ-countably compact and Y is dense in X, then X is pseudocompact and Y is a P-space [2,I.2.2 Theorem]. More generally, if Cp(X) is σ-bounded relatively sequentially complete, then X is finite [17,Corollary 3.2]; but, if Cp(Y|X) is σ-bounded relatively sequentially complete and Y is dense in X, then X is pseudocompact and Y is a P-space [17,Theorem 3.3].
Motivated by the research carried out in [17,18] on countable coverings of Cp(Y|X) and in [13,14] on uncountable coverings of Cp(X), we extend our study of the locally convex space Cp(Y|X) for (i) fundamental bounded resolutions, (ii) compact resolutions swallowing the compact sets, (iii) resolutions consisting of Banach disks, and (iv) resolutions made up of convex compact sets swallowing the local null sequences. We refer the reader to [2,24,32] for topological and locally convex notions not defined here.
First, we prove that if Cp(Y|X) has a fundamental resolution consisting of bounded sets, Y must necessarily be countable, which extends [15,Theorem 3.3 (i)] as well as [16,Theorem 1.9], the latter because Cp(X|βX)=Cbp(X). For the definitions and properties of quasi-Suslin and web-compact spaces that we use below, see [32,I.4.2] and [24,Chapter 4], respectively.
Theorem 2.1. Let Y⊆X. Cp(Y|X) admits a fundamental bounded resolution if and only if Y is countable.
Proof. Assume that {Aα:α∈NN} is a fundamental resolution for Cp(Y|X) consisting of absolutely convex bounded sets. Since each Aα is absolutely convex, the bipolar theorem ensures that A00α=¯Aα, closure in RY. Consequently, the fact that {Aα:α∈NN} swallows all bounded sets in Cp(Y|X) means that
M(Y|X)={¯Aα:α∈NN}. |
Since M(Y|X) coincides with M(Y), according to Theorem 1.1, we see that M(Y) admits a bounded resolution, namely {¯Aα:α∈NN}, where each ¯Aα is a compact set in RY. In this case, Theorem 1.2 tells us that |Y|=ℵ0.
Conversely, if Y is countable then Cp(Y) is metrizable, so does Cp(Y|X). But, as is well-known, every metrizable locally convex space E admits a fundamental bounded resolution, namely the family {Aα:α∈NN} with
Aα=∞⋂i=1α(i)Ui, |
where {Ui:i∈N} is a decreasing base of absolutely convex neighborhoods of the origin in E. So, Cp(Y|X) has a fundamental bounded resolution.
Corollary 2.1. Let X be a P-space and let Y be a dense subspace of X. Then Cp(Y|X) admits a bounded resolution if and only if Y is countable and discrete.
Proof. If X is a P-space and Y is dense in X, according to [18,Theorem 18] the space Cp(Y|X) is sequentially complete, and hence locally complete, i.e., such that every bounded set is contained in a Banach disk [26,39.2]. So, if Cp(Y|X) admits a bounded resolution, then Cp(Y|X) is a quasi-(LB)-space. Hence, by [33,Proposition 22] or [24,Theorem 3.5], there is a resolution for Cp(Y|X) consisting of Banach disks that swallows all Banach disks in Cp(Y|X). Thus, the local completeness of Cp(Y|X) guarantees that Cp(Y|X) has a fundamental bounded resolution, which allows Theorem 2.1 to ensure that Y is countable. As every countable P-space is discrete, we are done. Conversely, if Y is countable, then Cp(Y|X), as a linear subspace of RY, is metrizable. Hence, Cp(Y|X) has a bounded resolution.
Example 2.1. Let X be the one-point Lindelöfication of a discrete space Y with |Y|≥ℵ1. Since Y is uncountable, the previous corollary prevents Cp(Y|X) to admit a bounded resolution.
Example 2.2. If Y is a dense P-space of X and Cp(Y|X) admits a bounded resolution, even a countable one consisting of sequentially complete bounded sets, Y need not be countable or discrete. In fact, if Y is a nondiscrete P-space and X:=βY, we claim that Cp(Y|X) admits a countable resolution consisting of sequentially complete bounded sets.
Proof. Clearly, Cp(Y|X)=Cbp(Y), where Cbp(Y) is the dense linear subspace of Cp(Y) consisting of all bounded functions with relative pointwise topology. If B stands for the closed unit ball of the Banach space (Cb(Y),‖⋅‖∞), then C(Y|X)=Cb(Y)=⋃∞n=1nB. So, the family {Aα:α∈NN} with Aα=α(1)⋅B for each α∈NN is a countable bounded resolution for Cp(Y|X). Since Y is a P-space, Cp(Y) is sequentially complete [6]. Hence, if {fn}∞n=1 is a Cauchy sequence in Cp(Y|X) contained in B, there exists f∈C(Y) such that fn→f in Cp(Y). But, as |fn(y)|≤1 for all (n,y)∈N×Y, we have that |f(y)|≤1 for all y∈Y. Thus, f∈B, which shows that B is sequentially complete in Cp(Y|X).
Theorem 2.2. Let Y be a topological subspace of X. If Cp(Y|X) admits a compact resolution that swallows the compact sets, then the compact sets in Y are finite.
Proof. First note that the topologies σ(L(Y),C(Y)) and σ(L(Y),C(Y|X)) coincide on each compact subset K of Y, or rather on δ(K), regarded as a subset of L(Y). In other words, that each compact set K in Y is embedded both in Lp(Y) and in Lp(Y|X).
In fact, if a net {yd:d∈D} in K and a point y∈K verify that ⟨f,δyd⟩→⟨f,δy⟩ for every f∈C(Y), it is obvious that ⟨f,δyd⟩→⟨f,δy⟩ for every f∈C(Y|X). Thus, δyd→δy under σ(L(Y),C(Y|X)). Conversely, if a net {yd:d∈D} in K and a point y∈K verify that ⟨f,δyd⟩→⟨f,δy⟩ for every f∈C(Y|X), we claim that ⟨g,δyd⟩→⟨g,δy⟩ for every g∈C(Y). Indeed, if g∈C(Y), then g|K∈C(K) and there is h∈C(X) with h|K=g|K. Since h|Y∈C(Y|X), by assumption ⟨h|Y,δyd⟩→⟨h|Y,δy⟩. Thus, ⟨h|K,δyd⟩→⟨h|K,δy⟩, i.e., ⟨g,δyd⟩→⟨g,δy⟩. Consequently, δyd→δy under σ(L(Y),C(Y)).
Let {Aα:α∈NN} be a resolution for Cp(Y|X) consisting of compact sets that swallows the compact sets in Cp(Y|X). To prove that each compact set in Y is finite, we adapt the first part of the argument of [30,3.7 Theorem] (see also [24,Theorem 9.14]).
Assume by contradiction that there exists an infinite compact set K in Y. As Cp(Y|X) has a compact resolution, it is a quasi-Suslin space in the sense of Valdivia by [7,Proposition 1], hence it is web-compact in the sense of Orihuela (cf. [28]). Thus, the space Cp(Cp(Y|X)) is angelic by [28,Theorem 3].
Since K is a compact subspace of Y, it follows from the first part of this proof that K is embedded in Cp(Cp(Y|X)). Hence, K is an infinite Fréchet-Urysohn compact subset of the angelic space Cp(Cp(Y|X)) and consequently there exists a non-trivial sequence {xn}∞n=1 converging to some x∈K. If we set S:={xn:n∈N}∪{x}, then S is a compact countable set, and hence a metrizable compact space. Thus, there is a linear extender map φ:Cp(S)→Cp(Y), i.e., such that φ(f|S)=f for every f∈C(Y), which embeds Cp(S) as a closed linear subspace of Cp(Y), [4,Proposition 4.1]. But, if f:=φ(g) with g∈C(S), then f∈C(Y|X). Indeed, if h∈C(X) extends g to the whole of X, then
h|Y=φ(h|S)=φ(g)=f, |
which means that f∈C(Y|X). Thus, φ can be regarded as a map from Cp(S) into Cp(Y|X). Therefore, Cp(S) embeds in Cp(Y|X) as a closed linear subspace of Cp(Y|X). So, the metrizable space Cp(S) contains a resolution of compact sets, namely the family {Aα∩C(S):α∈NN}, that swallows the compact sets in Cp(S). By Christensen's theorem [8,Theorem 3.3] (see also [12,Theorem 94]) Cp(S) must be a Polish space. Thus, [2,I.3.3 Corollary] guarantees that the compact set S is discrete, and hence finite, a contradiction.
For the next lemma, recall that, according to [19,15.14 Corollary], a Tychonoff space X is realcompact if and only if its homeomorphic copy δ(X) in Lp(X) is a complete set.
Theorem 2.3. Assume that X is realcompact and Y is a closed subspace of X. If Cp(Y|X) has a Σ-covering of limited envelope, then Y is a Lindelöf Σ-space.
Proof. We need the following auxiliary result.
Claim 2.1. Under our assumptions, the copy δ(Y) of Y in Lp(Y|X) is a complete set.
Proof of the claim. Let {δyd:d∈D} be a Cauchy net in the canonical copy δ(Y) of Y in Lp(Y|X), i.e., such that for each f∈C(Y|X) and ϵ>0 there is d(f,ϵ)∈D with
|f(yr)−f(ys)|=|⟨f,δyr−δys⟩|<ϵ |
for r,s≥d(f,ϵ). As for each f∈C(X), one has that f|Y∈C(Y|X), and clearly {δyd:d∈D}, regarded as a net in δ(X), is a Cauchy net in Lp(X). Since X is realcompact there is z∈X such that δyd→δz in Lp(X), i.e., such that ⟨f,δyd⟩→⟨f,δz⟩ for every f∈C(X). But, as Y is closed in X, the canonical copy δ(Y) of Y in Lp(X) is closed in δ(X), which implies that δz∈δ(Y). This means that z∈Y, and thus we have in particular that ⟨f|Y,δyd⟩→⟨f|Y,δz⟩ for every f∈C(X). From this, it follows that ⟨g,δyd⟩→⟨g,δz⟩ for every g∈C(Y|X), which shows that δ(Y) is a complete set in Lp(Y|X).
Now, if {Aα:α∈NN} is a Σ -covering of limited envelope for Cp(Y|X), Theorem 1.3 provides a Lindelöf Σ-space Z such that
Lp(Y|X)⊆Z⊆RC(Y|X). |
By the claim, δ(Y) is a complete subspace of Lp(Y|X), and consequently, a closed set of the complete locally convex space RC(Y|X). Thus, δ(Y), as a closed topological subspace of Z, is a Lindelöf Σ-space.
Observe that if {δyd:d∈D} is a Cauchy net in the copy δ(Y) of Y in Lp(Y|X), then {δyd:d∈D} need not be a Cauchy net in Lp(Y). So, if we require only Y to be realcompact (regardless of whether X is realcompact or not), the argument of the previous claim does not work.
Corollary 2.2. Let X be realcompact. If Cp(X) has a bounded resolution, then X has countable extent.
Proof. If X has uncountable extent, there exists an uncountable closed discrete subspace Y, i.e., consisting of relative isolated points. If {Aα:α∈NN} is a bounded resolution for Cp(X), the restriction map T:Cp(X)→Cp(Y|X) given by T(f)=f|Y maps continuously Cp(X) onto Cp(Y|X). Thus, the family {T(Aα):α∈NN} is a bounded resolution for Cp(Y|X), and hence a Σ-covering of limited envelope. According to Theorem 2.3, the space Y must be Lindelöf, and hence countable, a contradiction.
Corollary 2.3. If M denotes the Michael line and P the subspace of the irrational numbers, the following properties hold.
(1) Cp(M) does not admit a bounded resolution.
(2) Cp(P|M) does not admit a bounded resolution.
Proof. Recall that in the Michael line the set P of irrationals is a discrete open subspace. As the topology of M is stronger than that of the real line R and each subspace of R is realcompact, it turns out that M is realcompact by virtue of [19,8.18 Corollary].
(1) Since there exists an uncountable Euclidean closed set Y in R consisting entirely of irrational numbers, this set Y is closed and discrete in M, so M has uncountable extent and (the contrapositive statement of) Corollary 2.2 applies.
(2) If Y is again an uncountable closed set in M consisting of irrational numbers, the restriction map S:Cp(P|M)→Cp(Y|M) defined as usual by S(f)=f|Y is a continuous linear map. It is also surjective, for if g∈C(Y|M), there exists f∈C(M) such that f|Y=g. So, clearly f|P∈C(P|M) and S(f|P)=f|Y=g. If Cp(P|M) had a bounded resolution, say {Aα:α∈NN}, then {S(Aα):α∈NN} would be a bounded resolution for Cp(Y|M). So, according to the proof of Corollary 2.2, the subspace Y would be countable, a contradiction.
Example 2.3. The converse of Corollary 2.2 does not hold. Let S be the Sorgenfrey line. Since S is a Lindelöf space, it is realcompact and has countable extent. However, Cp(S) does not admit a bounded resolution. Otherwise, by Theorem 1.4, the Sorgenfrey line would be a Lindelöf Σ -space, which is not true since the product of two Lindelöf Σ-spaces is a Lindelöf Σ-space, but S×S is not Lindelöf.
Corollary 2.4. Let X be realcompact and let Y be a closed subspace which is a P -space. Then Cp(Y|X) admits a bounded resolution if and only if Y is countable and discrete.
Proof. Since Y is a P-space, each compact set in Y is finite. If, in addition, Cp(Y|X) admits a Σ-covering of limited envelope, then Theorem 2.3 ensures that Y is a Lindelöf Σ-space. But, every Lindelöf Σ-space with finite compact sets is countable (by [2,IV.6.15 Proposition]), and every countable P-space is discrete. The converse is clear.
Hence, if X is a realcompact P-space and Y a closed subspace, then Cp(Y|X) admits a bounded resolution if and only if Y is countable and discrete.
Theorem 2.4. Let Y be a closed subspace of a realcompact space X. Then Cp(Y|X) admits a compact resolution that swallows the compact sets if and only if Y is countable and discrete.
Proof. According to Theorem 2.2, each compact set K in Y is finite. Moreover, as a compact resolution is a Σ-covering of limited envelope, Theorem 2.3 tells us that Y is a Lindelöf Σ -space. But, as we know, a Lindelöf Σ-space with finite compact sets is countable, so Y is countable and, consequently, both spaces Cp(Y|X) and Cp(Y) are metrizable.
Therefore we have a metrizable topological space Cp(Y|X) with a compact resolution {Aα:α∈NN} that swallows the compact sets in Cp(Y|X). So, again by Christensen's theorem, Cp(Y|X) is a Polish space, and hence a Čech-complete space [9,4.3.26 Theorem]. Since Cp(Y|X) is a Čech-complete dense subspace of Cp(Y), [2,I.3.1 Theorem] asserts that Y is (countable and) discrete.
Example 2.4. In the previous theorem, the requirement that X be realcompact is not necessary. In fact, if X is a non-realcompact P -space [19,Problem 9L] and Y is a countable subspace of X, then Y is C-embedded in X, so Cp(Y|X)=Cp(Y)=RY since Y, a countable subspace, is (closed and) discrete. If Y={yn:n∈N}, then {Aα:α∈NN} with Aα={f∈RN:|f(yn)|≤α(n)} is a compact resolution for Cp(Y|X) that swallows the compact sets.
Remark 2.1. If X is an arbitrary Tychonoff space and Y=X, using the well-known fact that if {Aα:α∈NN} is a compact resolution for Cp(X) that swallows the compact sets of Cp(X) and S:Cp(υX)→Cp(X) is the restriction map f↦f|X, then {S−1(Aα):α∈NN} is a compact resolution for Cp(υX) that swallows the compact sets in Cp(υX), and thus, we can deduce [30,3.7 Theorem] from Theorem 2.4. So, if Y=X, we may exempt the space X from being realcompact in the statement of Theorem 2.4.
Corollary 2.5. Let Y be a closed subspace of a realcompact space X. The following are equivalent
(1) Cp(Y|X) admits a compact resolution that swallows the compact sets.
(2) Cp(Y|X) is a Polish space.
Proof. If Cp(Y|X) has a compact resolution that swallows the compact sets, Y is countable and discrete by Theorem 2.4. Consequently, Cp(Y|X), as a dense linear subspace of Cp(Y)=RY, is a metrizable space with a compact resolution swallowing the compact sets. Hence, a Polish space by Christensen's theorem. Conversely, if Cp(Y|X) is a Polish space, it clearly has a compact resolution swallowing the compact sets.
Remark 2.2. Let Y be a countable dense subspace of X. If Cp(Y|X) admits a compact resolution that swallows the compact sets, then Y is discrete. In fact, if Yis countable, then Cp(Y|X) is metrizable and we may argue as in the final paragraph of the proof of Theorem 2.4.
Example 2.5. If Cp(Y|X) admits a bounded resolution that swallows the compact sets in Cp(Y|X), then Y need not be discrete. If M denotes the Michael line, the set Q of rational numbers is a closed subspace of M. Since, RQ=RN clearly has a compact resolution that swallows the compact sets in RQ, we see that Cp(Q|M)=Cp(Q) has a bounded resolution that swallows the compact sets in Cp(Q|M), but Q is not discrete.
Let us mention that, originally, the definition of quasi-(LB) -spaces emerged as an appropriate range class for the closed graph theorem when strictly barrelled spaces are located at the domain class [33,Corollary 1.5]. Valdivia's class of quasi-(LB) -spaces has proven to be useful both in functional analysis and in topology (see for instance [24,Chapter 3]). Recently, some lifting properties involving quasi-(LB)-spaces at the range class in the closed graph theorem has been relaxed by replacing the strictly barrelled spaces of the domain class by locally convex spaces with a sequential web (see [22,Theorem 1] for details).
Lemma 3.1. The spaces Cp(X) and Ck(X) have the same Banach disks.
Proof. If B is a Banach disk in Ck(X), it is clear that B is a Banach disk in Cp(X). Conversely, we claim that if Q is a Banach disk in Cp(X), then Q is a bounded set in Ck(X). Indeed, if U is a basic τk-closed absolutely convex neighborhood of the origin in Ck(X), there is a compact set K in X and ϵ>0 such that
U={f∈C(X):supx∈K|f(x)|≤ϵ}. |
Clearly, U is a τp-closed set, for if {fd:d∈D} is a net in U such that fd→f in Cp(X), then |f(x)|≤ϵ for every x∈K. As the norm topology of EQ is stronger than the pointwise topology, U∩EQ is also a closed absolutely convex set for the norm topology of EQ. Since E=⋃∞n=1nU, the Baire category theorem provides some m∈N such that Q⊆mU, which shows that Q is a bounded set in Ck(X), as stated. As Q is a τk-bounded set such that EQ is a Banach space, it turns out that Q is a Banach disk in Ck(X).
Theorem 3.1. Let X be metrizable. Then Cp(X) is a quasi-(LB)-space if and only if X is σ-compact.
Proof. If Cp(X) is a quasi-(LB)-space, according to Lemma 3.1 the space Ck(X) is also a quasi-(LB)-space. As X is a kR-space, Ck(X) is complete, and hence locally complete, that is, such that the τk-closed absolutely convex cover of each bounded set in Ck(X) is a Banach disk or, according to [20,Theorem 2.1], such that each convergent sequence in Ck(X) is equicontinuous. Since by [33,Proposition 22] the space Ck(X) admits a resolution consisting of Banach disks that swallows the Banach disks in Ck(X), it follows that Ck(X) has a fundamental bounded resolution. Consequently, by [11,Proposition 3], the space X must be σ-compact.
Conversely, if X is metrizable and σ-compact, again by [11,Proposition 3] the space Ck(X) has a (fundamental) bounded resolution {Aα:α∈NN}. Since the local completeness of Ck(X) guarantees that Bα:=¯abx(Aα)τk, i.e. the τk-closed absolutely convex cover of Aα, is a Banach disk for each α∈NN, and Ck(X) is clearly a quasi-(LB)-space. Thus, Cp(X) is also a quasi-(LB)-space.
Corollary 3.1. Let Y be a topological subspace of a metrizable space X. If X is σ-compact, then Cp(Y|X) is a quasi-(LB)-space.
Proof. If X is metrizable and σ-compact, Cp(X) is a quasi-(LB)-space by Theorem 3.1. If T is the restriction map from Cp(X) onto Cp(Y|X), the image T(B) of each Banach disk of Cp(X) is a Banach disk of Cp(Y|X). Thus, Cp(Y|X) is a quasi-(LB)-space.
Example 3.1. Metrizability cannot be dropped in the 'if' part of Theorem 3.1. If p∈βN∖N, equip X=N∪{p} with the relative topology of βN. Since X is countable, it is σ -compact. However, Cp(X) is not a quasi-(LB)-space. In fact, since Cp(X) is a locally convex Baire space (see [27,Example 7.1]), i.e., such that it cannot be covered by countably many rare, balanced sets [29], if Cp(X) were a quasi-(LB)-space then Cp(X) would be a Fr échet space by virtue of [24,Corollary 3.12]. As Cp(X) is a dense linear subspace of RX, this would imply that Cp(X)=RX. Consequently, the space X should be discrete, which is not true.
Recall that a sequence {xn}∞n=1 in a locally convex space E is local null or Mackey convergent to zero [25,28.3] if there is a bounded, closed, absolutely convex set B in E (a closed disk) such that xn→0 in the normed space EB:=span(B) equipped with the Minkowski functional of B as a norm. Each local null sequence in E is a null sequence. If E is metrizable, each null sequence is local null [25,28.3.(1) c)]. A linear form u defined on a bornological space E (see [5,3.6.2 Definition] or [25,28]) is continuous if and only if u(xn)→0 for each local null sequence {xn}∞n=1 in E, [25,28.3.(4)]. Recall that, according to the Buchwalter-Schmets theorem [6], the space Cp(X) is bornological if and only if X is realcompact.
Theorem 4.1. Let Y be a closed subspace of a realcompact space X. If Cp(Y|X) admits a resolution of convex compact sets that swallows the local null sequences in Cp(Y|X), then Y is countable and discrete.
Proof. As in the proof of [13,Theorem 12], we may assume without loss of generality that Cp(Y|X) admits a resolution {Aα:α∈NN} of absolutely convex compact sets swallowing the local null sequences in Cp(Y|X).
We adapt as possible the argument of the proof of [13,Theorem 12] to the present setting. Let M be the family of all local null sequences in Cp(Y|X). Since the family {Aα:α∈NN} swallows the members of M, the Mackey* topology μ(L(Y),C(Y|X)) of L(Y) is stronger than the topology τc0 on L(Y) of the uniform convergence on the local null sequences of Cp(Y|X). As it is clear that σ(L(Y),C(Y|X))≤τc0, we conclude that (L(Y),τc0)′=C(Y|X). Moreover, as X is realcompact, Cp(X) is bornological, and since Y is a closed subspace of X, the restriction map from Cp(X) onto Cp(Y|X) is a quotient map [2,0.4.1 (2) Proposition], which ensures that Cp(Y|X) is also a bornological space [25,28.4.(2)]. Consequently, the τc0 -dual (L(Y),τc0) of Cp(Y|X) is complete by [25,28.5.(1)]. In fact, it is even μ(L(Y),C(Y|X))-complete by [25,18.4.(4)].
Now we claim that every compact set in Y is finite. Indeed, if K is a compact set in Y, as we have seen in the proof of Theorem 2.2, the canonical copy δ(K) of K in Lp(Y|X) is embedded in Lp(Y|X), i.e., it is a σ(L(Y),C(Y|X))-compact set in L(Y). So, the completeness of (L(Y),τc0), together with Krein's theorem and the fact that τc0 is a locally convex topology of the dual pair ⟨L(Y),C(Y|X)⟩, ensures that the weak* closure Q in Lp(Y|X) of the absolutely convex hull of δ(K) is a compact set in Lp(Y|X), and hence a β(L(Y),C(Y|X))-bounded set. On the other hand, as Cp(Y) is a quasibarrelled space [23,11.7.3 Corollary], the β(L(Y),C(Y))-bounded sets in L(Y) are finite-dimensional. But, according to Theorem 1.1, we have that β(L(Y),C(Y))=β(L(Y),C(Y|X)). So, every β(L(Y),C(Y|X))-bounded set in L(Y) is finite-dimensional. Thus, in particular, Q is finite-dimensional. This means that δ(K), as a linearly independent system of vectors in L(Y) contained in Q, must be finite. Hence, the compact set K is finite, as stated.
Since each compact resolution is a Σ-covering of limited envelope, Theorem 2.3 tells us that Y is a Lindelöf Σ-space. So Y is countable by [2,IV.6.15 Proposition]. Hence, Cp(Y|X) is a metrizable space, which ensures that, in Cp(Y|X), local null sequences and null sequences are the same. Moreover, if M is a compact set in the metrizable space Cp(Y|X), then M lies in the closed absolutely convex cover of a null sequence {fn}∞n=1, [25,21.10.(3)]. So, if {fn}∞n=1⊆Aγ, the fact that Aγ is a closed absolutely convex set guarantees that M⊆Aγ. Thus, {Aα:α∈NN} is a compact resolution for Cp(Y|X) that swallows the compact sets of Cp(Y|X). So, we apply Theorem 2.4 to get that Y is (countable and) discrete.
Remark 4.1. If Y=X, one can get rid of the realcompactness from X by working with Cp(υX) instead of with Cp(X). So, Theorem 4.1 essentially contains [13,Theorem 12].
Let us examine the existence of a bounded resolution for Cp(Y) that swallows the Cauchy sequences in Cp(Y|X). We denote by R(X) the linear subspace of RX consisting of functions with finite support, i.e., those which vanish off a finite set in X.
Theorem 5.1. Assume that X is metrizable and Y hemicompact with Y⊆X. Then Cp(Y|X) has a bounded resolution that swallows the Cauchy sequences in Cp(Y|X) if and only if Y is countable.
Proof. We may assume there is a bounded resolution {Aα:α∈NN} in Cp(Y|X) consisting of absolutely convex sets swallowing the Cauchy sequences in Cp(Y|X). Given ey∈R(Y) defined by ey(y)=1 and ey(z)=0 if z≠y, we extend ey to the whole of X by setting ˆey(z)=ey(z) if z∈Y and ˆey(x)=0 if x∉Y. Since X is first countable, as in the proof of [13,Theorem 33] we find a sequence {fy,n}∞n=1 in C(X) such that fy,n→ˆey in RX. As fy,n|Y→ey in Cp(Y), there is γ∈NN with fy,n|Y∈Aγ for all n∈N, so ey∈¯Aγ. Now, given (n1,α)∈N×NN, let β∈NN be such that β(1)=n1 and β(i+1)=α(i) for each i∈N and define Bβ:=n1¯Aα, closure in RY. Since {ey:y∈Y} is a Hamel basis of R(Y), it follows that {Bβ:β∈NN} is a bounded family in RY such that R(Y)⊆⋃{Bβ:β∈NN}.
On the other hand, note that Ck(Y) is a metrizable space due to Arens' theorem [1] (see also [34,Theorem 13.2.4]). As C(Y|X) is clearly a dense linear subspace of Ck(Y), each g∈Ck(Y) is the limit of a sequence {gn}∞n=1 in C(Y|X) under the compact-open topology τk, and hence under the pointwise topology τp. Thus, there exists δ∈NN such that gn∈Aδ for every n∈N so that g∈¯Aδ, which shows that C(Y)⊆⋃{Bβ:β∈NN}. Thus, we conclude that C(Y)+R(Y)⊆⋃{Bβ:β∈NN}, which means that the linear subspace C(Y)+R(Y) of M(Y) admits a bounded resolution under the weak* topology of M(Y). According to [13,Remarks 20,27], Y must be countable.
For the converse, note that if Y is countable, RY has a compact resolution {Qα:α∈NN} that swallows the compact sets in RY. Hence, {Qα∩C(Y|X):α∈NN} is a bounded resolution for Cp(Y|X) such that if {hn}∞n=1 is a Cauchy sequence in Cp(Y|X) and h∈RY verifies that hn→h in RY, there is γ∈NN with {h,hn:n∈N}∈Qγ. Hence, {hn:n∈N}∈Qγ∩C(Y|X).
Proposition 5.1. Let Cp(Y) be Fréchet-Urysohn. If Cp(Y|X) has a bounded resolution that swallows the Cauchy sequences in Cp(Y|X), then Cp(Y) has a bounded resolution.
Proof. Let {Aα:α∈NN} be a bounded resolution for Cp(Y|X) that swallows the Cauchy sequences in Cp(Y|X). As C(Y)⊆M(Y)=M(Y|X), the latter equality because of Theorem 1.1, given g∈C(Y) there is a bounded set Q in Cp(Y|X) such that g∈¯Q, closure in Cp(Y). But, since Cp(Y) is Fré chet-Urysohn, there exists a sequence {gn}∞n=1 in Q⊆C(Y|X) such that gn→g in Cp(Y). As {gn}∞n=1 is a Cauchy sequence in Cp(Y|X), there is γ∈NN such that gn∈Aγ for every n∈N. Consequently, we have g∈¯Aγ, closure in Cp(Y). Since each bounded set in Cp(Y|X) is a bounded set in Cp(Y) and the closure of a bounded set is bounded, it follows that the family {¯Aα:α∈NN} is a bounded resolution for Cp(Y).
Corollary 5.1. Assume that Y is a Lindelöf P-space, with Y⊆X. Then Cp(Y|X) has a bounded resolution that swallows the Cauchy sequences in Cp(Y|X) if and only if Y is countable and discrete.
Proof. Since Y is a Lindelöf P-space, it turns out that Cp(Y) is Fréchet-Urysohn [3,10.2 Theorem]. Consequently, Cp(Y) has a bounded resolution by the preceding proposition. So, according to [15,Proposition 3.6], the subspace Y must be countable and discrete. The proof of the converse is analogous to that of Theorem 5.1.
If F is a linear subspace of C(X), let Y={x∈X:δx∉F⊥} so that δ(Y)=δ(X)∖F⊥. Since F⊥, the annihilator of F in L(X), is a closed subspace of Lp(X), Y is an open subset of X. Define T:F→Cp(Y|X) by the rule Tf=f|Y.
Theorem 6.1. T is a linear homeomorphism from F onto its range, which embeds F isomorphically into Cp(Y|X).
Proof. Let us first check that T is one-to-one. So, if f∈kerT⇔f|Y=0, we have to show that f=0. Equivalently, if f∈F is such that f≠0, we must prove that Tf≠0. But, if f≠0, there is u∈L(X) with ⟨f,u⟩≠0. If u=v+w with v=∑pi=1aiδxi, where δxi∈F⊥ or, equivalently, xi∈X∖Y for 1≤i≤p, and w=∑qi=1aiδyi with δyi∉F⊥ or, equivalently, yi∈Y for 1≤i≤q, then
0≠⟨f,u⟩=⟨f,w⟩=⟨f|Y,w⟩=⟨Tf,w⟩, |
since w∈L(Y). As L(Y) is the dual of Cp(Y|X), this shows that Tf≠0, as desired.
Let us check that T−1:ImT→F is continuous when ImT is provided with the relative topology of Cp(Y|X) and F is equipped with the relative topology of Cp(X). Indeed, if {fd:d∈D} is a net in F and f∈F are such that Tfd→Tf in ImT, it is clear that fd(y)→f(y) for every y∈Y. Now, if x∈X, either δx∈F⊥ or δx∉F⊥. In the first case, it is obvious that fd(x)=0→0=f(x) since fd,f∈F. In the second case, x∈Y, so fd(x)→f(x). This shows that fd→f in F pointwise on X. Since T is continuous, one-to-one, and a topological homomorphism, it turns out that T is a linear homeomorphism that embeds isomorphically F into Cp(Y|X).
Theorem 6.2. Let Z be a closed set in X. If F={f∈C(X):f(z)=0 ∀z∈Z}, then F is linearly homeomorphic to a dense subspace of Cp(Y|X).
Proof. Since it is clear that δz∈F⊥ for each z∈Z, we have Z⊆X∖Y. But, if x∈X∖Y and x∉Z, there is g∈C(X) with g(x)=1 and g(z)=0 for all z∈Z, which means that g∈F. But ⟨g,δx⟩=1, so δx∉F⊥, i.e., x∈Y, a contradiction. Therefore, X∖Y=Z. Thus, if T:F→Cp(Y|X) is the map considered in the previous theorem, then
T(F)={g∈C(Y):∃f∈C(X),f|Y=g,f|X∖Y=0}. |
By Theorem 6.1 the subspace T(F) of Cp(Y|X) is isomorphic to F. If Δ is a finite subset of Y and h∈C(Y|X), the fact that Z is closed allows us to find f∈C(X) such that f(y)=h(y) for each y∈Δ and f(z)=0 for every z∈Z, i.e., a function f∈C(X) such that g:=f|Y lies in T(F) and verifies that g(y)=h(y) for y∈Δ. This shows that T(F) is dense in Cp(Y|X).
Example 6.1. If X=D∪{ξ} is the one-point Lindelöfication of the discrete space D, the subspace Σ(D) of RD consisting of all countably supported functions on D is linearly homeomorphic to a dense subspace of Cp(D|X). In particular, Cp(D|X) is a Baire space.
Proof. If F:={f∈C(X):f(ξ)=0}, the restriction map S:F→Σ(D) given by Sf=f|D is a linear homeomorphism from F onto Σ(D). If we put Y:={x∈X:δx∉F⊥}, Theorem 6.2 asserts that the subspace F of Cp(X) is isomorphic to a dense subspace of Cp(Y|X). Consequently, Σ(D) is isomorphic to a dense subspace of Cp(Y|X).
But, clearly D=Y, since, if x∈D, there is f∈C(X) with f(x)=1 and f(ξ)=0, such that f∈F and ⟨f,δx⟩=1, i.e., δx∉F⊥. Thus, x∈Y. Whereas if x∈Y, then δx∉F⊥, which means that x≠ξ, so x∈D. The second statement follows from the fact that Σ(D) is a Baire space.
Remark 6.1. If X is uncountable, then Σ(X) does not admit a bounded resolution. This is because Σ(X) is locally complete. Thus, Valdivia's theorem [24,Theorem 3.5] ensures that Σ(X) admits a bounded resolution if and only if it is a quasi-(LB)-space. Since Σ(X) is a Baire space, Σ(X) is necessarily a Fréchet space [24,Corollary 3.12], which implies that Σ(X)=RX with X countable. Note that this implies that if X=D∪{ξ} is the one-point Lindelöfication of the discrete space D with |D|≥ℵ1, then Cp(D|X) does not have a bounded resolution (cf. Example 2.1).
Example 6.2. If X=D∪{ξ} is the one-point compactification of the discrete space D, the subspace R(D) of RD consisting of all finitely supported functions on D is linearly homeomorphic to a dense subspace of the space Cp(D|X).
Proof. If F:={f∈C(X):f(ξ)=0}, the linear map S:F→R(D) defined by Sf=f|D is a linear homeomorphism from the closed one-codimensional linear subspace F of Cp(X) onto R(D). Likewise, the set Y:={x∈X:δx∉F⊥} coincides with D since δ(X)∩F⊥={δξ}. By Theorem 6.2, F is linearly homeomorphic to a dense subspace of Cp(Y|X). So, R(D) is linearly isomorphic to a dense linear subspace of Cp(D|X).
Remark 6.2. R(X) has a fundamental bounded resolution if and only if X is countable.
Proof. Let {Aα:α∈NN} be a fundamental resolution for R(X) consisting of absolutely convex bounded sets. Thus, the bipolar theorem ensures that A00α=¯Aα, closure in RX. Hence, the fact that {Aα:α∈NN} swallows all compact sets in R(X) means that the bidual E of R(X) is given by E={¯Aα:α∈NN}. As each ¯Aα is an absolutely convex compact set in RX, and thus a Banach disk in E, it turns out that E is a quasi-(LB)-space. On the other hand, since each f∈Σ(X) is the limit of a sequence in R(X), we have that Σ(X)⊆E. As Σ(X) is a dense Baire subspace of RX, it follows that E is a Baire space. Since each locally convex space which is both a quasi-(LB)-space and a Baire space is a Fréchet space, necessarily E=RX with X countable. The converse is obvious.
As a consequence, if X=Y∪{ξ} is the one-point compactification of the discrete space Y, then Cp(Y|X) admits a fundamental bounded resolution if and only if Y is countable, which is a particular case of Theorem 2.1.
Juan C. Ferrando: conceptualization, research, methodology, formal analysis, writing-original draft, review & editing, validation; Manuel López-Pellicer: formal analysis, validation, writing review; Santiago Moll-López: validation, writing-review & editing. All authors have read and agreed to the published version of the manuscript.
The authors declare they have not used Artificial Intelligence (AI) tools in the creation of this article.
The first named author is supported by Generalitat Valenciana under project PROMETEO/2021/063.
The authors declare that they have no conflict of interest.
[1] | A. Kilbas, H. Srivastava, J. Trujillo, Theory and applications of fractional differential equations, Amsterdam: Elsevier Science Ltd., 2006. |
[2] |
S. Alizadeh, D. Baleanu, S. Rezapour, Analyzing transient response of the parallel RCL circuit by using the Caputo-Fabrizio fractional derivative, Adv. Differ. Equ., 2020 (2020), 55. https://doi.org/10.1186/s13662-020-2527-0 doi: 10.1186/s13662-020-2527-0
![]() |
[3] |
S. Rezapour, S. Etemad, H. Mohammadi, A mathematical analysis of a system of Caputo-Fabrizio fractional differential equations for the anthrax disease model in animals, Adv. Differ. Equ., 2020 (2020), 481. https://doi.org/10.1186/s13662-020-02937-x doi: 10.1186/s13662-020-02937-x
![]() |
[4] |
C. Li, R. Wu, R. Ma, Existence of solutions for caputo fractional iterative equations under several boundary value conditions, AIMS Mathematics, 8 (2023), 317–339. https://doi.org/10.3934/math.2023015 doi: 10.3934/math.2023015
![]() |
[5] |
B. Ahmad, M. Alghanmi, A. Alsaedi, J. Nieto, Existence and uniqueness results for a nonlinear coupled system involving caputo fractional derivatives with a new kind of coupled boundary conditions, Appl. Math. Lett., 116 (2021), 107018. https://doi.org/10.1016/j.aml.2021.107018 doi: 10.1016/j.aml.2021.107018
![]() |
[6] |
J. Ni, J. Zhang, W. Zhang, Existence of solutions for a coupled system of p-Laplacian Caputo-Hadamard fractional Sturm-Liouville-Langevin equations with antiperiodic boundary conditions, J. Math., 2022 (2022), 3346115. https://doi.org/10.1155/2022/3346115 doi: 10.1155/2022/3346115
![]() |
[7] |
A. Salem, B. Alghamdi, Multi-strip and multi-point boundary conditions for fractional Langevin equation, Fractal Fract., 4 (2020), 18. https://doi.org/10.3390/fractalfract4020018 doi: 10.3390/fractalfract4020018
![]() |
[8] |
Y. Adjabi, M. Samei, M. Matar, J. Alzabut, Langevin differential equation in frame of ordinary and Hadamard fractional derivatives under three point boundary conditions, AIMS Mathematics, 6 (2021), 2796–2843. https://doi.org/10.3934/math.2021171 doi: 10.3934/math.2021171
![]() |
[9] | G. Lumer, Connecting of local operators and evolution equations on networks, In: Potential theory Copenhagen 1979, Berlin: Springer, 2006,219–234. https://doi.org/10.1007/BFb0086338 |
[10] |
J. Graef, L. Kong, M. Wang, Existence and uniqueness of solutions for a fractional boundary value problem on a graph, FCAA, 17 (2014), 499–510. https://doi.org/10.2478/s13540-014-0182-4 doi: 10.2478/s13540-014-0182-4
![]() |
[11] |
V. Mehandiratta, M. Mehra, G. Leugering, Existence and uniqueness results for a nonlinear Caputo fractional boundary value problem on a star graph, J. Math. Anal. Appl., 477 (2019), 1243–1264. https://doi.org/10.1016/j.jmaa.2019.05.011 doi: 10.1016/j.jmaa.2019.05.011
![]() |
[12] |
S. Etemad, S. Rezapour, On the existence of solutions for fractional boundary value problems on the ethane graph, Adv. Differ. Equ., 2020 (2020), 276. https://doi.org/10.1186/s13662-020-02736-4 doi: 10.1186/s13662-020-02736-4
![]() |
[13] |
G. Mophou, G. Leugering, P. Fotsing, Optimal control of a fractional Sturm-Liouville problem on a star graph, Optimization, 70 (2021), 659–687. https://doi.org/10.1080/02331934.2020.1730371 doi: 10.1080/02331934.2020.1730371
![]() |
[14] |
A. Turab, Z. Mitrovicˊ, A. Savi\acute{c}, Existence of solutions for a class of nonlinear boundary value problems on the hexasilinane graph, Adv. Differ. Equ., 2021 (2021), 494. https://doi.org/10.1186/s13662-021-03653-w doi: 10.1186/s13662-021-03653-w
![]() |
[15] | X. Han, H. Cai, H. Yang, Existence and uniqueness of solutions for the boundary value problems of nonlinear fractional differential equations on star graph (Chinese), Acta Math. Sci., 42 (2022), 139–156. |
[16] |
W. Ali, A. Turab, J. Nieto, On the novel existence results of solutions for a class of fractional boundary value problems on the cyclohexane graph, J. Inequal. Appl., 2022 (2022), 5. https://doi.org/10.1186/s13660-021-02742-4 doi: 10.1186/s13660-021-02742-4
![]() |
[17] |
W. Zhang, J. Zhang, J. Ni, Existence and uniqueness results for fractional Langevin equations on a star graph, Math. Biosci. Eng., 19 (2022), 9636–9657. https://doi.org/10.3934/mbe.2022448 doi: 10.3934/mbe.2022448
![]() |
[18] |
D. Baleanu, S. Etemad, H. Mohammadi, S. Rezapour, A novel modeling of boundary value problems on the glucose graph, Commun. Nonlinear Sci., 100 (2021), 105844. https://doi.org/10.1016/j.cnsns.2021.105844 doi: 10.1016/j.cnsns.2021.105844
![]() |
[19] |
V. Mehandiratta, M. Mehra, G. Leugering, Existence results and stability analysis for a nonlinear fractional boundary value problem on a circular ring with an attached edge: A study of fractional calculus on metric graph, Netw. Heterog. Media., 16 (2021), 155–185. https://doi.org/10.3934/nhm.2021003 doi: 10.3934/nhm.2021003
![]() |
[20] |
H. Khan, Y. Li, W. Chen, D. Baleanu, A. Khan, Existence theorems and Hyers-Ulam stability for a coupled system of fractional differential equations with p-Laplacian operator, Bound. Value. Probl., 2017 (2017), 157. https://doi.org/10.1186/s13661-017-0878-6 doi: 10.1186/s13661-017-0878-6
![]() |
[21] |
H. Khan, F. Jarad, T. Abdeljawad, A. Khan, A singular ABC-fractional differential equation with p-Laplacian operator, Chaos Soliton. Fract., 129 (2019), 56–61. https://doi.org/10.1016/j.chaos.2019.08.017 doi: 10.1016/j.chaos.2019.08.017
![]() |
[22] |
A. Devi, A. Kumar, T. Abdeljawad, A. Khan, Stability analysis of solutions and existence theory of fractional Lagevin equation, Alex. Eng. J., 60 (2021), 3641–3647. https://doi.org/10.1016/j.aej.2021.02.011 doi: 10.1016/j.aej.2021.02.011
![]() |
[23] |
W. Zhang, W. Liu, Existence and Ulam's type stability results for a class of fractional boundary value problems on a star graph, Math. Method. Appl. Sci., 43 (2020), 8568–8594. https://doi.org/10.1002/mma.6516 doi: 10.1002/mma.6516
![]() |
[24] |
A. Devi, A. Kumar, Hyers-Ulam stability and existence of solution for hybrid fractional differential equation with p-Laplacian operator, Chaos Soliton. Fract., 156 (2022), 111859. https://doi.org/10.1016/j.chaos.2022.111859 doi: 10.1016/j.chaos.2022.111859
![]() |
[25] |
M. Abbas, Ulam stability and existence results for fractional differential equations with hybrid proportional-Caputo derivatives, J. Interdiscip. Math., 25 (2022), 213–231. https://doi.org/10.1080/09720502.2021.1889156 doi: 10.1080/09720502.2021.1889156
![]() |
[26] |
I. Ahmad, K. Shah, G. Ur Rahman, D. Baleanu, Stability analysis for a nonlinear coupled system of fractional hybrid delay differential equations, Math. Method. Appl. Sci., 43 (2020), 8669–8682. https://doi.org/10.1002/mma.6526 doi: 10.1002/mma.6526
![]() |
[27] | I. Podlubny, Fractional differential equations: an introduction to fractional derivatives, fractional differential equations, to methods of their solution and some of their applications, San Diego: Academic Press, Inc., 1999. |
[28] |
C. Urs, Coupled fixed point theorems and applications to periodic boundary value problems, Miskolc Math. Notes, 14 (2013), 323–333. https://doi.org/10.18514/MMN.2013.598 doi: 10.18514/MMN.2013.598
![]() |
[29] | A. Granas, J. Dugundji, Fixed point theory, New York: Springer, 2003. https://doi.org/10.1007/978-0-387-21593-8 |