On the Fractional Variable Order Thermostat Model: Existence Theory on Cones via Piece-Wise Constant Functions
Abstract
In the current manuscript, we intend to investigate the existence, uniqueness, and the stability of positive solution in relation to a fractional version of variable order thermostat model equipped with nonlocal boundary values in the Caputo sense. In fact, we will get help from the constant piece-wise functions for transforming our variable order model into an auxiliary standard model of thermostat. By Guo-Krasnoselskii’s fixed point theorem on cones, we derive the required conditions ensuring the existence property for positive solutions. An example is illustrated to examine the validity of the observed results.
1. Introduction
One of the most significant subbranches of mathematics in other existing computational and applied disciplines is arbitrary order calculus, notably fractional order calculus. Because of its applicability in a variety of fields including applied science and engineering, as well as its flexibility to model different systems, processes, and phenomena with memory effects, an arbitrary order calculus theory is regarded as an important subject of research for most researchers, mathematicians, scientists, and engineers. Fractional derivatives are powerful tools for describing the memory and heredity qualities of a wide range of materials and processes. The experts have contributed significantly to the study of fractional differential equations in recent years, for example, see some papers about the modeling of the Sturm-Liouville-Langevin problem [1], studying the Langevin boundary value problem (BVP) [2], investigation of the combination synchronization of a Caputo-Hadamard system [3], modeling of the Caputo-conformable pantograph problem [4], modeling of the hybrid thermostat [5], impulsive fractional systems [6], and modeling COVID-19 cases in the fractional settings [7], (for more studies, we refer to [8–13]).
The discussions of BVPs have attracted the focus of many scholars, and in this direction, valuable results have been obtained. Of course, various methods have been utilized to study fractional BVP such as the Banach contraction map principle [14], fixed point theorems [15], monotone iterative method [16], variational method [17], fixed point index theory and coincidence degree theory [18, 19], and numerical methods [20, 21]. On the other hand, numerous researches and review studies have been done in exploring the existence and stability of solutions to constant order fractional problems, while the subject of existence and stability in relation to the variable order problems is discussed seldom in the newly published papers including investigation of the variable order model of alcoholism [22], multiterm variable order BVPs [23], nonsingular variable order problems [24], initial value problems with conformable variable orders [25], approximate solutions on half-axis for a variable order IVP [26], singular variable order BVPs [27], Ulam-Hyers-Rassias stability for an implicit variable order BVP [28], and the generalized variable order Lyapunov-type inequality [29].
On the other side, keeping in view the advancements in studies of fractional calculus, it was observed that the constant order fractional calculus is not the ultimate instrument for the modeling of physical problems. In this regard, Lorenze and Hartley [33], in 1998, proposed the variable order fractional operator to model complex dynamics problems and further study in this field were given by them in [34].
There are no research manuscripts about the application of cones on variable order systems. This gap motivates us to investigate the generalized version of thermostat model in the context of fractional variable orders with respect to constant piece-wise functions. Furthermore, the presented structure is expressed in a unique and simple case, allowing us to generalize several typical specific cases previously addressed, and also this structure makes the novelty of our work. The ultimate goal of this study is to explore some results in relation to existence theory for the solutions of variable order BVP for the Caputo fractional thermostat model (4). Note that in this model, we use the Caputo fractional derivative. In fact, the importance of the fractional derivatives is their memory property which can describe the dynamical behavior of a system during a time interval. This kind of derivatives is not local and that is why we can analyze the solutions of a system with respect to various values of fractional orders. Also, we emphasize that our model is a generalization of two constant order thermostat models given in [31, 32] to variable order version.
The structure of the manuscript is designed as follows: the following section is based on some basic notions related to the current study. In Section 3, to get help of Guo-Krasnoselskii fixed point theorem, a suitable partition of the underlying interval J is proposed. This variable order BVP for the Caputo fractional thermostat model (4) is divided into the finite number of BVPs consisting of the Caputo fractional differential equations of constant orders, and then using the concept of Green’s function, the equivalent solution of the proposed problem is obtained and some properties of Green’s function are discussed. Finally, the existence result for (4) is proved based on the cones in Section 4. In the next section (Section 5), an example is prepared to give the illustrations of the main findings. We end our research by conclusions in Section 6.
2. Preliminaries
The collection C(J, ℝ) of all functions ϰ : J⟶ℝ having the property of continuity will be a Banach space via ‖ϰ‖ = sup{∣ϰ(t)∣:t ∈ J}..
Definition 1 (see [35], [36].)The left Riemann-Liouville fractional integral and left Caputo fractional derivative of variable order u(t) of H are recalled as
Remark 2 (see [37], [38].)In Equations (5) and (6), by choosing the constant order u, the Caputo fractional derivative and the Riemann-Liouville fractional integral of variable order will be the same conventional of the Caputo fractional derivative and Riemann-Liouville fractional integral of constant order, respectively.
Lemma 3 (see [38].)Suppose α1, b1 > 0 and n = [α1] + 1, then
Lemma 4 (see [39].)Let α1, α2 > 0, ψ ∈ L1(b1, b2), and . Then, the differential equation
Moreover,
Remark 5 (see [26], [27].)It is observed that the property of semigroup is not consistent for arbitrary functions u(t), v(t) in the role of variable orders, i.e.,
Example 1. Assume y(t) ≡ 1 for t ∈ J = [0, 3] and u(t) = t/2 and Then,
For t = 2, we write
As a result, the semigroup property is not valid in the generalized case for the Riemann-Liouville fractional integral of variable order.
The next results will be used in the main results concerning the Riemann-Liouville fractional integral given in (5).
Definition 7 (see [26], [27], [40].)Consider a subset of ℝ:
- (a)
A generalized interval is either empty, singleton subset, or an interval
- (b)
The finite set of generalized interval is a partition of whenever every belongs to exactly one of the generalized intervals in
- (c)
is piece-wise constant with respect to of if , h is constant on
Theorem 8 (Guo-Krasnoselskii fixed point theorem [41]). Assume a cone P and bounded subsets B1, B2 of a Banach space E with
- (i)
∥Ty∥≥∥y∥, y ∈ P∩∂B1 and ∥Ty∥≤∥y∥, y ∈ P∩∂B2; or
- (ii)
∥Ty∥≤∥y∥, y ∈ P∩∂B1 and ∥Ty∥≥∥y∥, y ∈ P∩∂B2.
Then, operator T possesses a fixed point in
3. Auxiliary BVP and Green’s Function
Before moving to the main results, we first assume the following assertions:
Let n ∈ ℕ, and the finite sequence of points in a way that 0 = S0 < Sj < Sn = S, j = 1, ⋯, n − 1. Designate Jj≔(Sj−1, Sj], j = 1, 2, ⋯, n. Then, the partition of interval J is .
For all , Ej = C(Jj, ℝ), denotes Banach spaces via .
In the following, relevant Green’s function is obtained in relation to auxiliary BVP of constant order of the Caputo fractional thermostat model (22).
Lemma 9. Let . Then, y ∈ Ej is a solution of the auxiliary BVP of constant order of the Caputo fractional thermostat model
Proof. Using Lemma 3, there are c0, c1 ∈ ℝ such that
In view of Lemma 4,
By y′(0) = 0, we get that c1 = 0.. Moreover,
Using the boundary condition CDα−1y(Sj) + y(Sj−1) = 0, we get
At the end, (26) takes the form
The proof is completed.
Remark10. Notice that for each fixed s ∈ [Sj−1, Sj], ∂Gj/∂t = 0 for t ≤ s and ∂Gj/∂t < 0 for t > s; this turns Gj(t, s) a decreasing function. Therefore
As a result, due to behavior of Gj(t, s) with respect to s, we have
We shall prove that Gj(t, s) meets the following property to ensure the existence of positive solution of the variable order Caputo fractional thermostat model (4) proposed by Lan and Webb in [42]:
(H2): ∃φ : [Sj−1, Sj]⟶[0, ∞) as a measurable mapping, [c, d]⊆[Sj−1, Sj] and ξ ∈ [0, 1] such that
Lemma 11. If , then Gj(t, s) > 0 for all t, s ∈ [Sj−1, Sj], and Gj(t, s) satisfies (H2).
Proof. Taking [c, d] = [Sj−1, Sj], then
The proof is completed.
Lemma 12. If , then Gj(t, s) ≥ 0 for all t, s ∈ [Sj−1, Sj], and Gj(t, s) satisfies (H2).
Proof. Taking [c, d] = [Sj − 1, d] such that Sj−1 ≤ d < Sj, and using the preceding lemma arguments, we obtain
Setting
The proof is completed.
Lemma 13. If , then G(t, s) changes sign on [Sj−1, Sj] × [Sj−1, Sj], and Gj(t, s) satisfies (H2).
Proof. Let [c, d] = [Sj−1, d] be such that Sj−1 ≤ d < Sj and We have
The proof is completed.
4. Existence and Uniqueness
Theorem 14. Let f(s, y(s)) ∈ C([Sj−1, Sj]) × [0, ∞). Assuming one of the below mentioned cases:
- (1)
f0 = ∞ and f∞ = 0 (Sublinear case)
- (2)
and (Superlinear case),
Proof. Define T : C([Sj−1, Sj])⟶C([Sj−1, Sj]) as
We now define the cone
In the start, we prove that T(P) ⊂ P. Since the functions f and Gj are positive and continuous, it follows that if y ∈ P, then Ty ∈ C([Sj−1, Sj]) and Ty(t) ≥ 0 for all t ∈ [Sj−1, Sj]. We reach to the inequality derived below for a fixed y ∈ P for all t ∈ [Sj−1, Sj] and that Gj(t, s) satisfies (H2)
So for all y ∈ B, we have
Tending t1⟶t2 implies that the RHS of the above inequality goes to 0 and thus T(B) is equicontinuous. At the end, the Arzela-Ascoli theorem confirms complete continuity of operator T : P⟶P.
Next, assume that (2) is true. As f0 = ∞, a ρ1 > 0 exists such that f(t, y) ≥ δ1y, ∀0 < y ≤ ρ1, where δ1 satisfies
Taking y ∈ P with ∥y∥ = ρ1, then
Let B1 = {y ∈ C([Sj−1, Sj])\∥y∥<ρ1}. Hence, we have ∥Ty∥≥∥y∥, y ∈ P∩∂B1.
Due the continuity of f(t.) on [0, ∞), a function can be defined which is nondecreasing on (0, ∞) by assumption and
Therefore, there exists ρ2 > ρ1 > 0 such that , where δ2 satisfies
Define Ω2 = {y ∈ C([Sj−1, Sj])\∥y∥<ρ2} and let y ∈ P such that ∥y∥ = ρ2. Then,
Hence, we have ∥Ty∥≤∥y∥, y ∈ P∩∂B2. In conclusion, there is at least one positive solution by (i) of Theorem 8 for the auxiliary BVP of constant order Caputo fractional thermostat model (22).
Now, assume that (3) is true. For δ2 > 0 and by assumption, there is r1 > 0 with f(t, y) ≤ δ2y for 0 ≤ y ≤ r1. Let y ∈ P such that ∥y∥ = r1. Then,
If we let B1 = {y ∈ C[Sj−1, Sj]\∥y∥<r1}, then ∥Ty∥≤∥y∥ for y ∈ P∩∂B1. Again by assumption, there is r > 0 such that f(t, y) ≥ δ1u, ∀y ≥ r. Define B2 = {y ∈ C([Sj−1, Sj])\∥y∥<r2}, where r2 = max(2r1, (r/ξ)). Then, y ∈ P and ∥y∥ = r2 imply that
This shows that ∥Ty∥≥∥y∥ for y ∈ P∩∂B2. In conclusion, it is found at least one positive solution, by (ii) of Theorem 8, for the auxiliary BVP of constant order Caputo fractional thermostat model (22) as
Now, we show the uniqueness of solutions for the auxiliary BVP of constant order Caputo fractional thermostat model (22) based on the Banach contraction principle. We consider the following assumption:
Theorem 15. Let the conditions (H1) and (H3) be satisfied and the inequality
Proof. We shall use the Banach contraction principle to prove that T has unique fixed point. For x(t), y(t) ∈ Ej, by Lemma 11 and Lemma 12, we obtain
Consequently by (57), the operator T is a contraction. Hence, by Banach’s contraction principal, T has a unique fixed point , which is a unique positive solution of the auxiliary BVP of the constant order Caputo fractional thermostat model (22).
In the next result, we generalize the existence criteria for the variable order BVP of the main Caputo fractional thermostat model (4).
Theorem 16. Assuming (H1)-(H2), the variable order BVP of the main Caputo fractional thermostat model (4) admits at least one solution in C(J, ℝ).
Proof. We know that the solution fulfills the auxiliary BVP of constant order Caputo fractional thermostat model (22) by Theorem 14 for any j ∈ {1, 2, ⋯, n}. Now, the continuous function on [0, Sj] is defined as
Therefore, y(t) = yj(t), t ∈ Jj, j ∈ {1, 2, ⋯, n} solves the variable order BVP of the main Caputo fractional thermostat model (4). The proof is completed.
5. Ulam-Hyers Stability
In this section, we are going to investigate the Ulam-Hyers stability for solutions of the given variable model of thermostat.
Definition 17 (see [43].)The variable order Caputo fractional thermostat model (4) is Ulam-Hyers stable if there exists cf > 0 such that for each ε > 0 and for every solution z ∈ C(J, ℝ) of the following inequality
Theorem 18. Assume that the conditions (H1) and (H2) to be held. Then, the Caputo fractional thermostat model (4) is Ulam-Hyers stable.
Proof. Let ε > 0 be an arbitrary number and the function z(t) belonging to C(J, ℝ) satisfies the following inequality
For any j ∈ {1, 2, ⋯, n}, we define the functions z1(t) ≡ z(t), t ∈ [0, S1] and for j = 2, 3, ⋯, n:
For any j ∈ {1, 2, ⋯, n} and according to the equality (6), for t ∈ Jj, we get
Taking on both sides of the inequality (62), we obtain
According to Theorem 16, the Caputo fractional thermostat model (4) has a positive solution y ∈ C(J, ℝ) defined by y(t) = yj(t) for t ∈ Jj, j = 1, 2, ⋯, n, where
and is a positive solution of (22). According to Lemma 9, the integral equation
Then,
We obtain, for each t ∈ Jj
Therefore, by Definition 17, the Caputo fractional thermostat model (4) is Ulam-Hyers stable and the proof is completed.
6. Example
We here simulate the simple form of our variable order BVP of the main Caputo fractional thermostat model (4) by giving an example numerically.
Example 2. Consider the nonlinear function
Then, in consistent with (22) and corresponding to the variable order BVP of fractional differential equation
Clearly, f0 = ∞ and f∞ = 0.
For j = 1, we get Γ(1.4) − 1 ≈ −0.11274 < 0, we take
and define the cone P1 = {y\y ∈ C[0, 1], mint∈[0, 1]y(t) ≥ ξ1∥y∥}. By Theorem 14, it is deduced that that the auxiliary BVP of constant order Caputo fractional thermostat model (75) possesses a positive solution .
For j = 2, we get Γ(1.5) − 1 ≈ −0.11377 < 0, we take
and consider the cone P2 = {y\y ∈ C[1, 2], mint∈[1, 2]y(t) ≥ ξ2∥y∥}.
According to Theorem 14, the auxiliary BVP of constant order Caputo fractional thermostat model (76) admits a positive solution , and by Theorem 16, the BVP for variable order Caputo fractional thermostat model (74) has a solution
7. Conclusion
In science and technology, fractional differential equations are utilized to model and describe a variety of natural processes. In connection with standard fractional models, variable fractional models appear to be more important for complex natural phenomena. The focus of this study was to analyze the BVP of variable order Caputo fractional thermostat model (4) and to explore its solutions’ existence utilizing techniques from fixed point theory on cones. To do such a method, we first noticed the invalidity of semigroup property for the Riemann-Liouville fractional integral of variable order. In order to obtain the solution of problem and solve this issue, we considered a partition of the interval J and the corresponding auxiliary constant order BVP of the Caputo fractional thermostat model (22) was derived from the variable order one. Some properties of relevant Green’s function were reviewed. For the solutions’ existence, utilizing fixed point attributed to Guo-Krasnoselskii’s on cones, the main theorems (Theorem 14 and 16) were deduced. An example to confirm the validity of theoretical findings was provided. This technique can be used to consider various physical models with variable order fractional operators. Moreover, with the help of our results in this research paper, investigations on this open research problem can be also possible and one can extend the proposed BVP to other complicated fractional models. In the future, we want to study these boundary value problems with different conditions involving integral conditions or integroderivative conditions or nonlocal conditions along with infinite delay.
Conflicts of Interest
The authors declare that they have no competing interests.
Authors’ Contributions
The authors declare that the study was realized in collaboration with equal responsibility. All authors read and approved the final manuscript.
Acknowledgments
The first and fifth authors would like to thank Azarbaijan Shahid Madani University.
Open Research
Data Availability
No data were generated or analyzed during the current study.