
To address the multi-flexible integrated scheduling problem with setup times, a multi-flexible integrated scheduling algorithm is put forward. First, the operation optimization allocation strategy, based on the principle of the relatively long subsequent path, is proposed to assign the operations to idle machines. Second, the parallel optimization strategy is proposed to adjust the scheduling of the planned operations and machines to make the processing as parallel as possible and reduce the no-load machines. Then, the flexible operation determination strategy is combined with the above two strategies to determine the dynamic selection of the flexible operations as the planned operations. Finally, a potential operation preemptive strategy is proposed to judge whether the planned operations will be interrupted by other operations during their processing. The results show that the proposed algorithm can effectively solve the multi-flexible integrated scheduling with setup times, and it can also better solve the flexible integrated scheduling problem.
Citation: Dan Yang, Zhiqiang Xie, Chunting Zhang. Multi-flexible integrated scheduling algorithm for multi-flexible integrated scheduling problem with setup times[J]. Mathematical Biosciences and Engineering, 2023, 20(6): 9781-9817. doi: 10.3934/mbe.2023429
[1] | Fangfang Zhu, Xinzhu Meng, Tonghua Zhang . Optimal harvesting of a competitive n-species stochastic model with delayed diffusions. Mathematical Biosciences and Engineering, 2019, 16(3): 1554-1574. doi: 10.3934/mbe.2019074 |
[2] | Martin Bohner, Sabrina Streipert . Optimal harvesting policy for the Beverton--Holt model. Mathematical Biosciences and Engineering, 2016, 13(4): 673-695. doi: 10.3934/mbe.2016014 |
[3] | Eduardo Liz, Alfonso Ruiz-Herrera . Delayed population models with Allee effects and exploitation. Mathematical Biosciences and Engineering, 2015, 12(1): 83-97. doi: 10.3934/mbe.2015.12.83 |
[4] | Peng Feng . On a diffusive predator-prey model with nonlinear harvesting. Mathematical Biosciences and Engineering, 2014, 11(4): 807-821. doi: 10.3934/mbe.2014.11.807 |
[5] | Mingwei Wang, Kaisheng Shi, Zhao Wang, Rui Yan, Tao Lei, Jiaqin Xiao . Multi-slot energy harvesting wireless communication in interference environment. Mathematical Biosciences and Engineering, 2021, 18(4): 4127-4145. doi: 10.3934/mbe.2021207 |
[6] | Tainian Zhang, Zhixue Luo, Hao Zhang . Optimal harvesting for a periodic n-dimensional food chain model with size structure in a polluted environment. Mathematical Biosciences and Engineering, 2022, 19(8): 7481-7503. doi: 10.3934/mbe.2022352 |
[7] | Sheng Wang, Lijuan Dong, Zeyan Yue . Optimal harvesting strategy for stochastic hybrid delay Lotka-Volterra systems with Lévy noise in a polluted environment. Mathematical Biosciences and Engineering, 2023, 20(4): 6084-6109. doi: 10.3934/mbe.2023263 |
[8] | Yuanpei Xia, Weisong Zhou, Zhichun Yang . Global analysis and optimal harvesting for a hybrid stochastic phytoplankton-zooplankton-fish model with distributed delays. Mathematical Biosciences and Engineering, 2020, 17(5): 6149-6180. doi: 10.3934/mbe.2020326 |
[9] | Kunquan Lan, Wei Lin . Population models with quasi-constant-yield harvest rates. Mathematical Biosciences and Engineering, 2017, 14(2): 467-490. doi: 10.3934/mbe.2017029 |
[10] | Rajalakshmi Manoharan, Reenu Rani, Ali Moussaoui . Predator-prey dynamics with refuge, alternate food, and harvesting strategies in a patchy habitat. Mathematical Biosciences and Engineering, 2025, 22(4): 810-845. doi: 10.3934/mbe.2025029 |
To address the multi-flexible integrated scheduling problem with setup times, a multi-flexible integrated scheduling algorithm is put forward. First, the operation optimization allocation strategy, based on the principle of the relatively long subsequent path, is proposed to assign the operations to idle machines. Second, the parallel optimization strategy is proposed to adjust the scheduling of the planned operations and machines to make the processing as parallel as possible and reduce the no-load machines. Then, the flexible operation determination strategy is combined with the above two strategies to determine the dynamic selection of the flexible operations as the planned operations. Finally, a potential operation preemptive strategy is proposed to judge whether the planned operations will be interrupted by other operations during their processing. The results show that the proposed algorithm can effectively solve the multi-flexible integrated scheduling with setup times, and it can also better solve the flexible integrated scheduling problem.
In population dynamics, harvesting is quite common and always visible in ecology. In the natural or human haphazardness, harvesting reduces species due to hunting, fishing, disease, war, environmental effects like natural disasters, competition among the species for the same resources, limited living space and limited food supply. For the optimal use of limited natural resources, and to maintain the balance of the ecological system, it is crucial to know the threshold of harvesting. The study of harvesting is very effective both in ecology and economics. Population models with a harvesting effect greatly impact various industries that drive the economy, like fisheries, forestry, plants and poultry.
In [1], the authors delineated two species harvesting strategies where the populations are harvested independently at constant rates. The largest amount of secure harvesting may be much less than what would be considered from the perspective of a local analysis for the equilibrium point. The global behavior of predator-prey systems was investigated in the presence of constant harvesting and the preservation of either or both species [2,3,4,5]. The result is analogous to the characteristics of an unharvested system with several parameters. The combined impacts of harvesting and discrete-time delay on the predator-prey system were studied in [6], where a comparative analysis of stability behavior has been offered without time delay.
Notably, the regular diffusion strategy could be more challenging to analyze; see [7,8,9,10,11,12] and references therein.
Several scenarios can happen when harvesting is applied to a single or interacting species with different diffusive strategies [13,14,15]. Previously, the single species with spatio temporal harvesting was considered to investigate the persistence and extinction of a population [13]. In contrast, the harvesting effort was focused on by using a time-independent cropped function for two competitive populations in [15]. The model is a modified reaction-diffusion model, known as the directed diffusion model, where species diffuse according to the smooth spatial function. Also, the resource distribution of the model is described as per capita available resources. A non-homogeneous Gilpin–Ayala diffusive equation was studied for single species by considering a spatially distributed harvesting map [16,17]. The study emphasized the crucial concept in the ecological system that a perfect mathematical model cannot be gained since we cannot include all of the effective parameters in the model [18]. Moreover, the model will never be able to forecast ecological catastrophes.
This study gives a transparent idea about real-life frameworks of the competitive population in the population ecology. To design the mathematical model, we are taking into account that the harvesting rate is proportional to the intrinsic growth rate such that the harvesting functions are E1(x)∝r(x) and E2(x)∝r(x) which implies that E1(x)=μr(x) and E2(x)=νr(x), where μ and ν are coefficients of proportionality which are non-negative. The model equations are
{∂u∂t=d1Δu(t,x)+r(x)u(t,x)(1−u(t,x)+v(t,x)K(x))−μr(x)u(t,x),∂v∂t=d2Δv(t,x)+r(x)v(t,x)(1−u(t,x)+v(t,x)K(x))−νr(x)v(t,x),t>0,x∈Ω,∂u∂η=∂v∂η=0,x∈∂Ω,u(0,x)=u0(x),v(0,x)=v0(x),x∈Ω, | (1.1) |
where u(t,x) and v(t,x) represent the population densities of two competing species which are non-negative, with corresponding dispersal rates d1 and d2, respectively. The harvesting coefficients of populations u,v are μ,ν, respectively. Hence the harvesting functions are E1(x)=μr(x) and E2(x)=νr(x), respectively. To analyze the model, we have simplified the model into a more compact profile in a later section.
Let us now consider the initial conditions u0(x)≥0 and v0(x)≥0, x∈¯Ω, and these initial conditions are positive in an open nonempty subdomain of Ω. The carrying capacity and intrinsic growth rate are denoted by K(x) and r(x), respectively. The function K(x) is continuous as well as positive on ¯Ω and r(x)⩾0, where x∈¯Ω; moreover, r(x) is positive in an open nonempty subdomain of Ω. The notation Ω is a bounded region in Rn, i.e., typically, n={1,2,3}, with the smooth boundary ∂Ω∈C2+α,0<α<1, and where η represents the unit normal vector on ∂Ω. The zero Neumann boundary condition indicates that no individual crosses the boundary of the habitat, or that individuals going in and out at any location from the boundary stay equal at all times. The Laplace operator Δ:=∑ni=1∂2/∂x2i in Rn implies the random motion of the species.
Now, we modify the system (1.1) in such a technique that the harvesting rate is expedited and easier to complete the analysis. The first equation of the model (1.1) can be written in the following way:
∂u∂t=d1Δu(t,x)+r(x)u(t,x)(1−u(t,x)+v(t,x)K(x))−μr(x)u(t,x)=d1Δu(t,x)+r(x)u(t,x)(1−μ)(1−u(t,x)+v(t,x)(1−μ)K(x)). |
Let K1(x)=(1−μ)K(x) and r1=1−μ. Then, we obtain
∂u∂t=d1Δu(t,x)+r1r(x)u(t,x)(1−u(t,x)+v(t,x)K1(x)). |
After applying same framework for the second equation of the system (1.1), we finally obtain the modified version of the main model:
{∂u∂t=d1Δu(t,x)+r1r(x)u(t,x)(1−u(t,x)+v(t,x)K1(x)),∂v∂t=d2Δv(t,x)+r2r(x)v(t,x)(1−u(t,x)+v(t,x)K2(x)),t>0,x∈Ω,∂u∂η=∂v∂η=0,x∈ ∂Ω,u(0,x)=u0(x),v(0,x)=v0(x),x∈Ω, | (1.2) |
where K1(x)=(1−μ)K(x),K2(x)=(1−ν)K(x),r1=1−μ and r2=1−ν.
In the present study, the main objectives and findings are as follows:
● We study the impact of the harvesting of two species in a spatially non-homogeneous environment on their interaction, where the harvesting rate varies for each population. The competition arises for the same resources, limited food supply and limited living space; predators make the prey species their food.
● We prove that the solution to the initial value problem for positive initial conditions approaches (us(x),vs(x)) globally in time.
● This study aims to illustrate the persistence-extinction story-line due to the harvesting rates between two competitive species.
● We have established the result to describe when they can coexist or when one species will drive the other to extinction.
● If the harvesting rate does not exceed the intrinsic growth rate and imposes conditions on μ and ν, coexistence is possible.
● We also present some numerical results both in one and two dimensional cases. We find the evolution of system energy for species densities with a periodic space and time dependent intrinsic growth rate and carrying capacity. Also, the system energy versus time is plotted for both short-time and long-time evolution with the harvesting coefficients.
We solve (1.1) numerically by using a stable backward Euler, decoupled and linearized fully discrete time-stepping algorithm in a finite element setting and examine whether the theoretical results are supported by giving several numerical experiments. To the best of our knowledge, the modeling, theoretical and numerical results presented in this manuscript are novel.
The rest of the paper is organized as below. In Section 2, the existence and uniqueness of the solution of (1.1) are proven. In Section 3, stability analysis of the equilibrium points is given for the conditions when the intrinsic growth rate exceeds the harvesting rates, one harvesting rate exceeds the intrinsic growth rate and both harvesting rates exceed the intrinsic growth rate. To support the theoretical findings, several numerical experiments are given in Section 4. Finally, a concluding summary and future research directions are discussed in Section 5.
Now we detach each equation to delineate the existence and uniqueness of the paired system. Consider the following system:
{∂u∂t=d1Δu(t,x)+r1r(x)u(t,x)(1−u(t,x)K1(x)),t>0,x∈Ω,u(0,x)=u0(x),x∈Ω,∂u∂η=0,x∈∂Ω. | (2.1) |
The results of single species are also discussed in [19,20,21,22]. Note that the proofs of Lemmas 7 and 8 are analogous to the proofs in [19]. The statements and proofs of Lemmas 7 and 8 can be found in Appendix A. It is also noted that all single-species results are available in Appendix A, and we will recall the results where necessary.
The following result demonstrates the existence and the uniqueness of solutions to a paired system (1.2).
Theorem 1. Let K1(x),K2(x)>0, which implies that μ,ν∈[0,1), and r(x)>0 on x∈¯Ω. When u0(x),v0(x)∈C(Ω), the model (1.2) has a unique solution (u,v). Further, if both initial functions u0 and v0 are non-negative and nontrivial, u(t,x)>0 and v(t,x)>0 for t>0.
Proof. Take into account the following system with μ,ν∈[0,1):
{∂u∂t=d1Δu(t,x)+r1r(x)u(t,x)(1−u(t,x)+v(t,x)K1(x)),∂v∂t=d2Δv(t,x)+r2r(x)v(t,x)(1−u(t,x)+v(t,x)K2(x)),t>0,x∈Ω,∂u∂η=∂v∂η=0,x∈∂Ω,u(0,x)=u0(x),v(0,x)=v0(x),x∈Ω, | (2.2) |
where K1(x)=(1−μ)K(x)>0,K2(x)=(1−ν)K(x)>0,r1=1−μ>0 and r2=1−ν>0 since μ,ν∈[0,1).
We utilize the existence-uniqueness theorem for parabolic paired, systems which is discussed in [23], and methods which are similar to the proof of [21], to show the existence of nontrivial time-dependent solutions. We choose the following constants:
ρu>supx∈Ωu0(x)>0,andρv>supx∈Ωv0(x)>0, |
and use the notations of proven results as discussed in [23] to denote
f1(t,x,u,v)=r1r(x)u(t,x)(1−u(t,x)+v(t,x)K1(x)),f2(t,x,u,v)=r2r(x)v(t,x)(1−u(t,x)+v(t,x)K2(x)). |
Then, it is simple to establish that the following conditions of the theorem are satisfied:
{f1(t,x,ρu,0)≤0≤f1(t,x,0,ρv),f2(t,x,0,ρv)≤0≤f2(t,x,ρu,0). | (2.3) |
The conditions of (2.3) satisfy the conditions of the parabolic coupled systems in [23] for the functions f1 and f2 defined above. As a result, we come to the theorem conclusion that for any nontrivial (u0(x),v0(x)) such that
(u0,v0)∈Sρ≡{(u1,v1)∈C([0,∞)ׯΩ)×C([0,∞)ׯΩ);0≤u1≤ρu,0≤v1≤ρv}, | (2.4) |
where C([0,∞)ׯΩ) is the class of continuous functions on [0,∞)ׯΩ; a unique solution (u(t,x),v(t,x)) for the system (2.2) exists and remains in Sρ for all (t,x)∈[0,∞)ׯΩ. Thus, (u(t,x),v(t,x)) is a unique and positive solution.
Let us establish the existence result for (1.1) for 0≤ν<1≤μ.
Theorem 2. Assume that 0≤ν<1≤μ, and let the initial conditions be u0≥0 and v0≥0; therefore, the model (1.1) has a unique, positive, time-dependent nontrivial solution.
Proof. The system (1.1) can be written as
{∂u∂t=d1Δu(t,x)+r(x)u(t,x)(1−μ−u(t,x)+v(t,x)K(x)),∂v∂t=d2Δv(t,x)+r(x)v(t,x)(1−ν−u(t,x)+v(t,x)K(x)),t>0,x∈Ω,∂u∂η=∂v∂η=0,x∈∂Ω,u(0,x)=u0(x),v(0,x)=v0(x),x∈Ω. | (2.5) |
Recalling the theoretical results as vindicated in [23] and the methods parallel to the proof in [22], to show the existence of nontrivial time-dependent solutions, let us choose the following constants:
ρu=max{supx∈Ωu0(x),1},ρv=max{supx∈Ωv0(x),supx∈ΩK(x)}. |
Note that the definitions of ρu and ρv are analogous to those in [22]. Let us denote
f1(t,x,u,v)=r(x)u(t,x)(1−μ−u(t,x)+v(t,x)K(x)),f2(t,x,u,v)=r(x)v(t,x)(1−ν−u(t,x)+v(t,x)K(x)). |
Then, it is simple to establish that the following conditions of the theorem are satisfied:
{f1(t,x,ρu,0)≤0≤f1(t,x,0,ρv),f2(t,x,0,ρv)≤0≤f2(t,x,ρu,0). | (2.6) |
The conditions (2.6) are valid based on the analysis presented in [23] for the functions f1 and f2. Therefore, we come to the theorem conclusion that for any nontrivial (u0(x),v0(x)) such that
(u0,v0)∈Sρ≡{(u1,v1)∈C([0,∞)ׯΩ)×C([0,∞)ׯΩ);0≤u1≤ρu,0≤v1≤ρv}, | (2.7) |
where C([0,∞)ׯΩ) is the class of continuous functions on [0,∞)ׯΩ; a unique solution (u(t,x),v(t,x)) for the system (2.5) exists and remains in Sρ for all (t,x)∈[0,∞)ׯΩ. Thus, (u(t,x);v(t,x)) is a unique and positive solution.
Let us establish the existence result for (1.1) for 0≤μ<1≤ν.
Theorem 3. Assume that 0≤μ<1≤ν and the initial conditions are u0≥ and v0≥0; therefore, the model (1.1) has a unique, positive, time-dependent nontrivial solution.
Proof. The proof is analogous to that of Theorem 2.
Remark 1. In the next section, the residuum of the main results is presented. We need several preliminary and auxiliary results to prove the theoretical results for the two species populations, which are presented in Appendix A. It is also noted that the main body of the paper contains only the leading and significant findings.
When investigating the consequences of competition between two competitive species, it is a vital task to check the stability analysis of the semi-trivial equilibrium, namely, one should evaluate (u∗(x),0),(0,v∗(x)), the trivial solution (0,0) and the nontrivial stationary solution, which implies the coexistence (us,vs).
The following section describes the case of the intrinsic growth rate transcending harvesting rate such that μ,ν∈[0,1). Given that E1(x)=μr(x),E2(x)=νr(x), where μ,ν belongs to [0,1) then obviously 0≤E1(x)<r(x) and 0≤E2(x)<r(x). We investigate two possible cases: μ≤ν and μ≥ν.
Lemma 1. Assume that μ,ν∈[0,1), which implies that r1>0, r2>0 and K1(x) and K2(x) are positive on ¯Ω. Therefore, the trivial steady state (0,0) of the model (1.2) is an unstable repelling equilibrium.
Proof. Let the linearized system (1.2) near the trivial equilibrium be given as
{∂u∂t=d1Δu(t,x)+r1r(x)u(t,x),t>0,x∈Ω,∂v∂t=d2Δv(t,x)+r2r(x)v(t,x),t>0,x∈Ω,∂u∂η=∂v∂η=0,x∈∂Ω,u(0,x)=u0(x),v(0,x)=v0(x),x∈Ω. | (3.1) |
The corresponding eigenvalue problems are given below:
{γψ=d1Δψ+r1r(x)ψ,x∈Ω,σϕ=d2Δϕ+r2r(x)ϕ,x∈Ω,∂ψ∂η=∂ϕ∂η=0,x∈∂Ω. | (3.2) |
Using the variational characterization of eigenvalues according to [19], we obtain the principal eigenvalue by choosing the eigenfunction ψ=1:
γ1≥1|Ω|∫Ωr1r(x)dx=1|Ω|∫Ω(1−μ)r(x)dx>0,sinceμ∈[0,1). |
Analogously, utilizing the variational characterization of eigenvalues according to [19], we obtain the principal eigenvalue by the choosing the eigenfunction ϕ=1:
σ1≥1|Ω|∫Ω(1−ν)r(x)dx>0,whileν∈[0,1). |
Thus, the trivial equilibrium (0,0) is unstable. Now, we prove that the trivial steady state (0,0) is a repeller. The proof can be found in [20] and [24].
The following case demonstrates the result for the outcome of the competition when the intrinsic growth rate exceeds the harvesting rate for 0≤μ≤ν<1.
The semi-trivial steady state (u∗,0) is unstable whenever μ≤ν, as shown in the following lemma.
Lemma 2. Let μ,ν∈[0,1), where μ≤ν. Thus, there exists ν1 for a certain μ such that, for all ν∈[μ,ν1), the equilibrium (u∗,0) is unstable for the system (1.2).
Proof. The analogous case μ=ν was discussed in [25], where the species have a common carrying capacity. Thus, we discuss the case that μ<ν. Linearization of the second equation from (1.2) near the stationary solution (u∗(x),0) gives
∂v(t,x)∂t=d2Δv(t,x)+r2r(x)v(t,x)(1−u∗(x)K2(x)),t>0,x∈Ω,v(0,x)=v0(x),x∈Ω,∂v∂η=0,x∈∂Ω. |
The corresponding eigenvalue problem is represented as follows:
σψ=d2Δψ+r2r(x)ψ(1−u∗(x)K2(x)),x∈Ω,∂ψ∂η=0,x∈∂Ω. | (3.3) |
The principal eigenvalue of this system is given by [19]:
σ1=supψ≠0,ψ∈W1,2{−d2∫Ω|∇ψ|2dx+∫Ωr2r(x)ψ2(1−u∗(x)K2(x))dx∫Ωψ2dx}. | (3.4) |
We assume that Ψ is the principal eigenfunction for the problem (3.3) with the principal eigenvalue σ1. This value is positive whenever the numerator of (3.4) is positive, leading to
σ1=−d2∫Ω|∇Ψ|2dx+∫Ωr2r(x)Ψ2(1−u∗(x)K2(x))dx∫ΩΨ2dx. | (3.5) |
Since the denominator on the right-hand side of (3.5) is positive, to have a positive eigenvalue, we assume that
−d2∫Ω|∇Ψ|2dx+∫Ωr2r(x)Ψ2(1−u∗(x)K2(x))dx>0⇒∫Ωr2r(x)Ψ2(1−u∗(x)K2(x))dx>d2∫Ω|∇(Ψ)|2dx⇒∫Ωr2r(x)Ψ2dx>d2∫Ω|∇(Ψ)|2dx+∫Ωr2r(x)Ψ2u∗(x)K2(x)dx. |
Multiplying both sides by (1−ν) gives
(1−ν)∫Ωr2r(x)Ψ2dx>(1−ν)d2∫Ω|∇(Ψ)|2dx+(1−ν)∫Ωr2r(x)Ψ2u∗(x)K2(x)dx⇒(1−ν)∫Ωr2r(x)Ψ2dx>(1−ν)d2∫Ω|∇(Ψ)|2dx+(1−ν)∫Ωr2r(x)Ψ2u∗(x)(1−ν)K(x)dx⇒1−ν>r2d2∫Ω|∇(Ψ)|2dx+r2∫Ωr(x)Ψ2u∗(x)K(x)dxr2∫Ωr(x)Ψ2dx. |
Therefore, we have
1−ν>d2∫Ω|∇(Ψ)|2dx+∫Ωr(x)Ψ2u∗(x)K(x)dx∫Ωr(x)Ψ2dx. |
Here, in the above inequality, the left-hand-side (1−ν)∈(0,1] whenever ν∈[0,1) and the right-hand side is positive since the numerator and denominator have a square term and there is no negative term because the parameter d2 is positive. Therefore, we can say that the right-hand side of the above inequality belongs to (0,1) since the right-hand side is less than the left-hand side and positive, where the left-hand side of the above inequality is (1−ν)∈(0,1].
Rearrange the above inequality to obtain
1−d2∫Ω|∇Ψ|2dx+∫Ωr(x)Ψ2u∗(x)K(x)dx∫Ωr(x)Ψ2dx>ν. |
Note that and left-hand side of the above inequality belongs to (0,1) as explained above. We define
ν1:=1−d2∫Ω|∇Ψ|2dx+∫Ωr(x)Ψ2u∗(x)K(x)dx∫Ωr(x)Ψ2dx, |
which implies that ν1>ν since the right-hand side is greater than ν. Hence, we obtain that μ≤ν<ν1. Therefore, there exists ν1 for a fixed μ such that, for all ν∈[μ,ν1), the equilibrium (u∗,0) is unstable.
In the following lemma, we prove that the steady state (0,v∗) is unstable whenever μ≤ν.
Lemma 3. Let μ,ν∈[0,1), where μ≤ν, and let there exist ν1 for a certain μ as long as ν∈[μ,ν1). Thus, the steady state (0,v∗) is unstable in the system (1.2).
Proof. The case that μ=ν was discussed in [25]. Thus, here we only discuss the case that μ<ν. Linearizing the first equation of (1.2) in the neighborhood of (0,v∗), we obtain
∂u(t,x)∂t=d1Δu(t,x)+r1r(x)u(t,x)(1−v∗(x)K1(x)),x∈Ω,u(0,x)=u0(x),x∈Ω,∂u∂η=0,x∈∂Ω. |
The corresponding eigenvalue problem is
σψ=d1Δψ+r1r(x)ψ(1−v∗(x)K1(x)),x∈Ω,∂ψ∂η=0,x∈∂Ω. | (3.6) |
The principal eigenvalue of this problem is given by [19]:
σ1=supψ≠0,ψ∈W1,2{−d1∫Ω|∇ψ|2dx+∫Ωr1r(x)ψ2(1−v∗(x)K1(x))dx∫Ωψ2dx}. | (3.7) |
For ν∈[0,1), take into account the eigenfunction ψ(x)=√(1−ν)K(x)=√K2(x). Then, the principle eigenvalue becomes
σ1≥∫Ωr1r(x)K2(x)(1−v∗(x)K1(x))dx∫ΩK2(x)dx. |
Note that ν∈[μ,ν1), ν1 is defined in Lemma 2. We introduce a constant c:=1−μ1−ν>1 as long as μ<ν; further, it is true for every 0≤μ≤ν<1, and it is definitely true for 0≤μ≤ν<ν1<1 because ν1∈[0,1), which means that μ,ν,ν1∈[0,1); it implies that c>1 for any values in [0,1).
Let us now estimate the principal eigenvalue in the following way:
σ1≥∫Ωr1r(x)K2(x)(1−v∗(x)(1−μ)K(x))dx∫ΩK2(x)dx, |
which can be rewritten as
σ1≥∫Ωr1r(x)K2(x)(1−v∗(x)(1−μ)(1−ν)(1−ν)K)dx∫ΩK2(x)dx. |
Introducing the constant c, we have
σ1≥r1∫Ωr(x)K2(x)(1−v∗(x)cK2(x))dx∫ΩK2(x)dx. | (3.8) |
We want to show that the numerator of the right-hand-side fraction in (3.8) is positive. From Lemma 13, we have
∫Ωr(x)K2(x)(1−v∗(x)K2(x))dx>0. |
Note that, since c>1, we obtain
∫Ωr(x)K2(x)(1−v∗(x)cK2(x))dx>∫Ωr(x)K2(x)(1−v∗(x)K2(x))dx>0. |
Therefore, the numerator of the inequality (3.8) is positive. Thus, the principal eigenvalue is positive:
σ1≥r1∫Ωr(x)K2(x)(1−v∗(x)cK2(x))dx∫ΩK2(x)dx>0, |
which completes the proof.
In the following theorem, we prove that the equilibrium (us,vs) is globally stable for the system (1.2) whenever μ≤ν by using Lemmas 1, 2 and 3.
Theorem 4. Suppose that μ,ν∈[0,1), where μ≤ν. There exists ν1 for a certain μ such that, for all ν∈[μ,ν1), the equilibrium (us,vs) of the system (1.2) is globally stable.
Proof. We consider that 0≤μ≤ν<1. Lemma 2 demonstrates that there is a number ν1∈[0,1) such that, whenever ν∈[μ,ν1), the steady state (u∗,0) is unstable. Lemma 3 illustrates that the steady state (0,v∗) is unstable. Lemma 1 demonstrates that the trivial steady state (0,0) is unstable and a repeller. We extract two results established in [24]. Hence, there exists a globally stable coexistence solution.
We get symmetrical results by exchanging μ and ν, which means that 0≤ν≤μ<1. The following results can be obtained:
● When μ≥ν, where μ,ν∈[0,1). There exists a value μ1 for a certain ν such that, for all μ∈[ν,μ1), the steady state (0,v∗(x)) of (1.2) is unstable.
● When μ≥ν, where μ,ν∈[0,1), there exists a value μ1 for a certain ν for all μ∈[ν,μ1). Thus, the steady state (u∗(x),0) of the system (1.2) is unstable.
● Global coexistence solution: When μ≥ν, where μ,ν∈[0,1). Thus, there exists a value μ1 for a certain ν for all μ∈[ν,μ1); the coexistence steady state of the system (1.2) is globally stable.
In the present section, we examine the outcomes of two competitive species when one harvesting rate in the system (1.1) surpasses the corresponding intrinsic growth rate, which means that there are two possible scenarios that can arise, namely, E1(x)≥r(x) or E2(x)≥r(x) such that 0≤ν<1≤μ or 0≤μ<1≤ν, respectively.
First, we depict the result on the impact of competition when one harvesting function exceeds the respective intrinsic growth function for the case that 0≤ν<1≤μ.
The following lemma shows that there is no coexistence state when 0≤ν<1≤μ.
Lemma 4. Suppose that 0≤ν<1≤μ; thus, there is no nontrivial stationary solution (us,vs) for the model (2.5) or (1.1).
Proof. Take into account that there is a nontrivial stationary solution (us(x),vs(x)), where us≥0 and vs≥0 for all x∈Ω. The coexistence solution is to satisfy the following system of equations:
{0=d1Δus(x)+r(x)us(x)(1−μ−us(x)+vs(x)K(x)),x∈Ω,0=d2Δvs(x)+r(x)vs(x)(1−ν−us(x)+vs(x)K(x)),x∈Ω,∂us∂η=∂vs∂η=0,x∈∂Ω. | (3.9) |
Now, integrating the first equation over Ω and utilizing the boundary conditions yields
∫Ωrus(1−μ−us+vsK)dx=0. |
The integrand is non-positive for all x∈Ω whenever μ≥1 and us≢ (which holds by our assumption of u_s being a nontrivial coexistence solution). Hence, there is no coexistence state, (u_s, v_s).
Next, we delineate that (0, v^*) is the only possible nontrivial stationary solution for the system (2.5) for any nontrivial non-negative initial conditions.
Lemma 5. Suppose that \mu \geq 1 then (0, v^*(\boldsymbol{x})) is the only nontrivial stationary solution of (2.5) and (1.1).
Proof. We assume that there exists a nontrivial equilibrium other than (0, v^*(\boldsymbol{x})) . Since there is no coexistence in the system according to Lemma 4, the other possible solution of such type is (u^*(\boldsymbol{x}), 0) , where u^*(\boldsymbol{x}) \geq 0 on \Omega and satisfies the following boundary value problem for \mu = 1 :
\begin{align*} & \begin{cases} & d_{1}\Delta u^*( \boldsymbol{x})-r( \boldsymbol{x})u^*( \boldsymbol{x})\dfrac{u^*( \boldsymbol{x})}{K( \boldsymbol{x})} = 0, \;\; \boldsymbol{x}\in\Omega, \\ & \dfrac{\partial u^*}{\partial \eta} = 0, \;\; \boldsymbol{x}\in \partial\Omega. \end{cases} \end{align*} |
Now, integrating and applying the boundary conditions yields
\begin{align*} \int_\Omega r( \boldsymbol{x})\dfrac{(u^*( \boldsymbol{x}))^2}{K( \boldsymbol{x})}\;d \boldsymbol{x} = 0, \end{align*} |
which is not true for a nontrivial u^*(\boldsymbol{x}) \geq 0 . Therefore, we arrive at a contradiction, and the only nontrivial stationary solution is (0, v^*(\boldsymbol{x})) , where the function v^*(\boldsymbol{x}) satisfies the second equation given Lemma 13. The same procedure is applicable for \mu > 1 . Thus, (0, v^*(\boldsymbol{x})) is the only nontrivial stationary solution of (1.1) for \mu\geq 1 .
The following lemma proves that (0, 0) of the system (2.5) (equivalent to (1.1)) is unstable but not a repeller when the harvesting rate E_1(\boldsymbol{x}) surpasses or is equal to the intrinsic growth rate r(\boldsymbol{x}) . Note that the proof is parallel with [21].
Lemma 6. Consider the case that 0\leq \nu < 1\leq\mu . Thus, the trivial steady state (0, 0) of the model (1.1) is unstable, but it is not a repeller.
Proof. First, we assume that \mu > 1 and the system (1.2) is linearized near the trivial equilibrium:
\begin{align} \begin{cases} & \dfrac{\partial u}{\partial t} = d_{1}\Delta u+r_1r( \boldsymbol{x})u, \;\; t > 0, \; \boldsymbol{x}\in\Omega, \\\\ & \dfrac{\partial v}{\partial t} = d_{2}\Delta v+r_2r( \boldsymbol{x})v, \;\; t > 0, \; \boldsymbol{x}\in\Omega, \\\\ & \dfrac{\partial u}{\partial \eta} = \dfrac{\partial v}{\partial \eta} = 0, \;\;\; \boldsymbol{x}\in \partial\Omega, \\ & u(0, \boldsymbol{x}) = u_0( \boldsymbol{x}), \;v(0, \boldsymbol{x}) = v_0( \boldsymbol{x}), \;\;\; \boldsymbol{x}\in\Omega. \end{cases} \end{align} | (3.10) |
The corresponding eigenvalue problems are
\begin{equation} \begin{cases} & \gamma \psi = d_{1}\Delta \psi+r_1r( \boldsymbol{x})\psi, \;\; \boldsymbol{x}\in\Omega, \\ & \sigma \phi = d_{2}\Delta \phi+r_2r( \boldsymbol{x})\phi, \;\; \boldsymbol{x}\in\Omega, \\ &\dfrac{\partial \psi}{\partial \eta} = \dfrac{\partial \phi}{\partial \eta} = 0, \;\; \boldsymbol{x}\in \partial\Omega. \end{cases} \end{equation} | (3.11) |
Consider \psi_1 and \phi_1 as two eigenfunctions (that can be chosen positive) with corresponding principal eigenvalues of (3.11): \gamma_1 and \sigma_1 , respectively [19]. Integrating (3.11) by using the boundary conditions yields
\begin{align*} \gamma_1 = \dfrac{\int_\Omega r_1r( \boldsymbol{x})\psi_1\;d \boldsymbol{x}}{\int_\Omega \psi_1\;d \boldsymbol{x}} = \dfrac{\int_\Omega (1-\mu)r( \boldsymbol{x})\psi_1\;d \boldsymbol{x}}{\int_\Omega \psi_1\;d \boldsymbol{x}} < 0, \;\;\text{since}\;\mu > 1, \end{align*} |
and
\begin{align} \sigma_1 = \dfrac{\int_\Omega r_2r( \boldsymbol{x})\phi_1\;d \boldsymbol{x}}{\int_\Omega \phi_1\;d \boldsymbol{x}} = \dfrac{\int_\Omega (1-\nu)r( \boldsymbol{x})\phi_1\;d \boldsymbol{x}}{\int_\Omega \phi_1\;d \boldsymbol{x}} > 0, \;\;\text{given}\;\nu < 1, \end{align} | (3.12) |
respectively. Thus, the steady state (0, 0) is unstable.
Therefore, for 0\leq \nu < 1\leq\mu, the steady state (0, 0) is unstable but not a repeller.
The next result shows the global stability for the solution (0, v^*(\boldsymbol{x})) of the system (1.1) when the harvesting coefficient satisfies 0\leq\nu < 1\leq \mu , as obtained by using Lemmas 4, 5 and 6.
Theorem 5. Let 0\leq\nu < 1\leq \mu . Thus, the stationary solution (0, v^*(\boldsymbol{x})) of the system (1.1) is globally asymptotically stable.
Proof. From Lemma 6, the solution (0, 0) is unstable and there is no coexistence solution according to Lemma 4. According to Lemma 5, the remaining non-negative steady state is (0, v^*(\boldsymbol{x})) . This solution is unique by the positivity and uniqueness proofs; see Lemmas 9 and 10. Recall the definition of {\bf S}_\rho from (2.7):
\begin{align*} {\bf S}_\rho\equiv \left\{\left(u_1, v_1\right)\in C \left([0, \infty) \times\overline{\Omega}\right)\times C \left([0, \infty) \times\overline{\Omega}\right) ; 0\leq u_1\leq \rho_u, \; 0\leq v_1\leq \rho_v\right\}, \end{align*} |
where
\begin{align*} \rho_u = \max \left\{\sup\limits_{ \boldsymbol{x}\in\Omega} u_0( \boldsymbol{x}), 1 \right\}, \;\; \rho_v = \max \left\{ \sup\limits_{ \boldsymbol{x}\in\Omega} v_0( \boldsymbol{x}), \; \sup\limits_{ \boldsymbol{x}\in\Omega} K( \boldsymbol{x})\right\}. \end{align*} |
Using the result in [23], we obtain the time-dependent solution (u(t, \boldsymbol{x}), v(t, \boldsymbol{x})) of (1.1), and the solution will converge to the equilibrium (0, v^*(\boldsymbol{x})) as t\rightarrow \infty for any initial conditions from {\bf S}_\rho , which completes the proof.
Symmetrically, the steady state (u^*(\boldsymbol{x}), 0) of the model (1.1) is globally asymptotically stable when \mu < 1\leq\nu .
Finally, in this section, we examine the case when both harvesting rates exceed the intrinsic growth rate; symbolically, \mu, \nu\geq 1 .
Theorem 6. Let \mu, \nu\geq 1 . Thus the trivial solution (0, 0) of the model (1.1) is globally asymptotically stable.
Proof. The reasoning from the proof of Lemma 6 applies here directly, and it shows that the solution (u(t, \boldsymbol{x}), v(t, \boldsymbol{x})) gradually converges to the trivial solution, which completes the proof.
Biologically, the result of Theorem 6 is valid since more harvesting than growth significantly reduces the density of both populations, and, as a continuation, populations will tend to extinction.
In this section, we represent numerical experiments by using the finite element method to support the theoretical results. Here, \|u\|_{L^2}: = \sqrt{\int_\Omega u^2(t, \textbf{x})d\textbf{x}} is the L^2 norm, whereas \|u\|_{L^1}: = \int_\Omega |u(t, \textbf{x})|d\textbf{x} is the L^1 norm. In finite element simulations, we deal with finite-dimensional space, where all norms are equivalent. Therefore, using either L^1 or L^2 norms, the outcomes and decisions will be similar. But, we used L^2 -norm square with a scaling factor of \frac{1}{2} so that it looks like the kinetic energy. The usual L^2(\Omega) inner product is denoted by (., .) . We define the Hilbert space for our problem as
X: = H^1(\Omega) = \big\{u\in L^2(\Omega):\nabla u\in L^2(\Omega)^{n}\big\}. |
The conforming finite element space is denoted by X_h\subset X , and we assume a regular triangulation \tau_h(\Omega) , where h is the maximum triangle diameter. We consider the following fully discrete, decoupled and linearized scheme of the system (1.1):
Algorithm 1: Fully discrete and decoupled ensemble scheme |
Given a time step \Delta t > 0 , end time T > 0 , initial conditions u^0 , v^0\in X_h , set M = T/\Delta t , and, for n = 1, \cdots, M-1 , compute the following: Find u_{h}^{n+1}\in X_h satisfying, for all \chi_h\in X_h : Find v_{h}^{n+1}\in X_h satisfying, for all l_h\in X_h : |
For all experiments, we consider the diffusion coefficients d_{1} = d_{2} = 1 , a unit square domain \Omega = (0, 1)\times(0, 1) and P_2 finite element and structured triangular meshes. We define the energy of the system at time t for the species densities u and v as
\frac12\int_\Omega u^2(t, \boldsymbol{x})d \boldsymbol{x}, \quad \text{and} \quad \frac12\int_\Omega v^2(t, \boldsymbol{x})d \boldsymbol{x}, |
respectively. The 2D code is written in Freefem++ [26].
In this section, we will consider the stationary carrying capacity together with both constant and space-dependent intrinsic growth rates.
In this experiment, we consider the carrying capacity of the system
K( \boldsymbol{x})\equiv 2.1+\cos(\pi x)\cos(\pi y), |
with a constant intrinsic growth rate r(\boldsymbol{x})\equiv 1.2 . We ran several simulations for various values of the harvesting coefficients \mu and \nu . In Figures 1–4, we considered the initial population densities u_0 = v_0 = 1.8 with time step size \Delta t = 0.1 .
In Figure 1, we present the contour plots for the species densities u and v at time t = 1.6 with fixed harvesting coefficients \mu = 1.5 and \nu = 0.08 . A co existence is observed at the moment.
We also plotted the energy of the system for the species densities u and v versus time for three different combinations of the harvesting coefficient pairs (\mu, \nu) in Figure 2. We consider the harvesting parameter \mu = 1.5 > \nu = 0.08 in Figure 2(a), and thus observe that the species u will die away shortly while the species v survives. The opposite scenario is observed in Figure 2(b), where \mu = 0.08 < \nu = 1.15 is considered. This is because one harvesting coefficient is significantly bigger than the other and exceeds the intrinsic growth rate; that is why one species becomes extinct in a short period of time. The results in Figure 2(a) and (b) support Theorems 2 and 3, respectively. In Figure 2(c), though the harvesting coefficients are the same ( \mu = \nu = 1.5 ), both exceed the intrinsic growth rate; thus, extinction in both species is observed in short-time evolution.
In Figure 3, we plot the energy of the system versus time for the species density u , and v with the coefficients of harvesting (a) \mu = 0.0006 , and \nu = 0.0 , and (b) \mu = 0.0 , and \nu = 0.0006 . We observe the harvesting impact as an extinction of the species u in (a), and the species v in (b).
In Figure 4, we plot the energy of the system for both species versus time, keeping fixed the harvesting parameter \mu = 0.0009 while varying \nu . We ran the simulation until t = 2000 for each case. In Figure 4(a), since \mu > \nu , as time grows, the species density for v remains bigger than that for u , whereas, the scenario is opposite in Figures 4(b)–(f) because \mu < \nu . A possible coexistence is exhibited in Figure 4(a), (c) and (d), which supports Theorem 4. There is a possible threshold from extinction to the persistence of both populations. In Figure 4(b), (e) and (f), the long-run results show that, when \mu < \nu , a possible extinction and evolution of multiple thresholds can exist.
In Figure 5, we show the population density v(t, \boldsymbol{x}) versus time for various values of the initial condition v_0 . In all cases, we considered the initial densities for both species to be the same and omitted the results for u . We observe a unique solution as time grows if the initial conditions are positive.
From Figure 6(a), we observe that, when the harvesting parameters do not exceed the intrinsic growth rate, a nontrivial solution exists, i.e., there is the co-existence of the two species. In Figure 6(b), we observe that the population density of the first species remains bigger than the second species because \mu < \nu . Ultimately, the second species will die out because of the competition between them.
In this experiment, we consider the carrying capacity and the intrinsic growth rate as
K( \boldsymbol{x})\equiv 2.5+\sin(x)\sin(y) \quad \text{and} \quad r( \boldsymbol{x})\equiv 1.5+\cos(x)\cos(y), |
respectively, along with the equal initial population densities u_0 = v_0 = 1.2 . The system energy versus time is plotted until t = 3000 in Figure 6(a) and (b) for two different harvesting parameter pairs.
In this section, we consider a time-dependent periodic system carrying capacity together with constant and time-dependent intrinsic growth rates.
In this experiment, we consider the time-dependent carrying capacity
K(t, \boldsymbol{x})\equiv (2.1+\cos(\pi x)\cos(\pi y))(1.1+\cos(t)), |
harvesting coefficients \mu = 0.0009 , and \nu = 0.0025 , intrinsic growth rate r(\boldsymbol{x})\equiv 1.0 and initial conditions u_0 = 0.5 and v_0 = 1.5 for the species u and v , respectively. We have a fixed time t = T = 13.74 and draw the contour plots at t = T, T+\pi/2, T+\pi, T+3\pi/2 and T+2\pi for the species densities u and v in Figures 7 and 8, respectively. From Figures 7 and 8, we observe quasiperiodic behavior in both species and their coexistence.
We have also plotted the energy of the system for species densities u and v versus time in Figure 9. We observe a clear co-existence of the two populations and change their density quasiperiodically over time. Since, in this case, \mu < \nu , the amplitude of the species density u increases, while it decreases for v , consuming more resources by population u and enjoying the best possible habitat.
In this experiment, we consider the carrying capacity
K(t, \boldsymbol{x})\equiv\big(1.2+2.5\pi^2e^{-(x-0.5)^2-(y-0.5)^2}\big)\big(1.0+0.3\cos(t)\big), |
together with the constant intrinsic growth rate r(\boldsymbol{x})\equiv 1 , initial population density u_0 = v_0 = 1.6 and harvesting coefficients \mu = 0.0009 and \nu = 0.0025 .
In Figure 10, the system energy versus time is plotted for both short-time and long-time evolution with the harvesting coefficients \mu = 0.0009 , and \nu = 0.0025 . We observe the periodic population densities for both species, and that, eventually, the species v dies out while the species u continues to exist.
In Figure 11, we present the contour plots for both species at times t = 80 and t = 1600 . It is observed that the highest population density is at (0.5, 0.5) and there is a coexistence of both species though the population density of the species u remains bigger than the species v at every place. This is the effect of the different harvesting parameters.
In this paper, we have studied two competing species in spatially heterogeneous environments. We observed various scenarios for several harvesting rates. When the harvesting rate does not surpass the intrinsic growth rate, which means that \mu, \nu \in [0, 1) and imposes conditions on \mu and \nu , so coexistence is possible. For small values of \mu and \nu , prey and predator populations coexist, which is observed analytically and numerically. We estimated the threshold of the harvesting coefficient when coexistence is possible; consider the following:
({\rm{i}}) When we fix \mu = 0.0009, the threshold of \nu is 0.5, so species v tends to extinction. Notably, we can get another entry of \nu by choosing any other values of \mu\in [0, 1) .
({\rm{ii}}) Analogously, we can determine the threshold of \mu when \nu is fixed, which has been demonstrated in Theorem 5.
Further, only one species becomes extinct when their harvesting rate is greater than their growth rate and other species persist when the harvesting rate is less than their growth rate. Both species become extinct when their harvesting rate exceeds the growth rate and the systems (1.1) and (1.2) converge to the trivial solution. From these analytical and numerical observations, we can conclude that both species coexist when the harvesting rate is less than their growth rate. Both species die out if the harvesting rate exceeds their intrinsic growth rate.
The model (1.1) is a well-known Lotka-Volterra competition model. Also, it is a predatory system as a special case; in some scenarios, two species fight for the same resources and limited space, and in another scenario or environment, one species attacks another species because of the limitation of food. Moreover, we have the harvesting term depending on the intrinsic growth rate, which is more realistic and biologically feasible. The study can be extended for experimental research on intra- or inter-specific competitive populations. This study is appropriate for application to various farms, wild animal persistence and river-ocean ecologies with seasonal change; in particular, the model is applicable to two-species phytoplankton populations [27]. Finally, there are two open problems for future studies. Consider and analyze the problem (1.2) for the following cases [28]:
● Predator-prey dynamics while \mu < 0 and \nu > 0 . Various lines of evidence attest to the importance of predation, such as the evidence of research on diversity, the ubiquity of anti-predator adaptations, and the impact of interacting with predators and prey in a population circle.
● Symbiotic (mutualism) model when \mu < 0 and \nu < 0 . Mutualism refers to a relationship between two species of organisms that benefit from the association.
The authors thank the anonymous reviewers for their suggestions, which have significantly improved the quality of the manuscript. The author M. Kamrujjaman was partially supported by the University Grants Commission and the Bose Center for Advanced Study and Research in Natural Sciences, University of Dhaka. Also, the author Muhammad Mohebujjaman was supported by the National Science Foundation grant DMS-2213274 and University Research grant Texas A & M International University.
The authors declare no conflict of interest.
In the first appendix, we have a few preliminary and auxiliary results for single-species populations.
Lemma 7. [21] Consider that the parameters are positive on \overline{\Omega} and the initial condition of (2.1) is a nonnegative continuous function u_0(\boldsymbol{x})\in C(\Omega), \; u_0(\boldsymbol{x})\geq 0 in \Omega , and u_0(\boldsymbol{x}) > 0 in some open, bounded, nonempty domain \Omega_1 \subset \Omega . Thus, there exists a unique positive solution of the system (2.1).
Lemma 8. [21] Consider the problem (2.1); then, there exists a function u^*(\boldsymbol{x}) > 0 that is a unique equilibrium solution of (2.1). Further, for any initial condition u_0(\boldsymbol{x})\geq 0, \; u_0(\boldsymbol{x})\not\equiv 0 , the solution u(t, \boldsymbol{x}) satisfies the condition
\begin{align*} \lim\limits_{t\rightarrow \infty} u(t, \boldsymbol{x}) = u^*( \boldsymbol{x}) \end{align*} |
uniformly for \boldsymbol{x}\in \overline{\Omega} .
Now, take into account the next following problem for the population density v = v(t, \boldsymbol{x}) :
\begin{equation} \begin{cases} & \frac{\partial v}{\partial t} = d_{2}\Delta v(t, \boldsymbol{x})+r_{2}r( \boldsymbol{x})v(t, \boldsymbol{x})\left(1-\frac{v(t, \boldsymbol{x})}{K_{2}( \boldsymbol{x})}\right), \;\;t > 0, \;\; \boldsymbol{x}\in\Omega, \\ & v(0, \boldsymbol{x}) = v_0( \boldsymbol{x}), \;\;\; \boldsymbol{x}\in\Omega, \\ & \frac{\partial v}{\partial \eta} = 0, \;\; \boldsymbol{x}\in \partial\Omega. \end{cases} \end{equation} | (A.1) |
Lemma 9. [21] Consider the initial condition of (A.1) as a continuous non-negative function v_0(\boldsymbol{x})\in C(\Omega), \; v_0(\boldsymbol{x})\geq 0 in \Omega , and v_0(\boldsymbol{x}) > 0 in some open, bounded, nonempty domain \Omega_1 \subset \Omega . Therefore, there exists a unique positive solution of the system (A.1).
Lemma 10. [21] Consider the problem (A.1); thus, there exists a function v^*(\boldsymbol{x}) > 0 which is a unique stationary solution of (A.1). Further, for any initial condition v_0(\boldsymbol{x})\geq 0, \; v_0(\boldsymbol{x})\not\equiv 0 , the solution v(t, \boldsymbol{x}) satisfies the condition
\begin{align*} \lim\limits_{t\rightarrow \infty} v(t, \boldsymbol{x}) = v^*( \boldsymbol{x}) \end{align*} |
uniformly for \boldsymbol{x}\in \overline{\Omega} .
Stationary solutions and preliminaries:
Let the following problem have a stationary solution u^*(\boldsymbol{x}) where v is zero in (1.2):
\begin{equation} d_{1}\Delta u^{*}( \boldsymbol{x})+r_{1}r( \boldsymbol{x})u^{*}( \boldsymbol{x})\left(1-\frac{u^{*}( \boldsymbol{x})}{K_{1}( \boldsymbol{x})}\right) = 0, \;\; \boldsymbol{x}\in\Omega, \;\;\; \frac{\partial u^{*}}{\partial \eta} = 0, \;\; \boldsymbol{x}\in\partial\Omega. \end{equation} | (A.2) |
Analogously, the following problem has a stationary solution v^*(\boldsymbol{x}) when u is zero in (1.2)
\begin{equation} d_{2}\Delta v^{*}( \boldsymbol{x})+r_{2}r( \boldsymbol{x})v^{*}( \boldsymbol{x})\left(1-\frac{v^{*}( \boldsymbol{x})}{K_{2}( \boldsymbol{x})}\right) = 0, \;\; \boldsymbol{x}\in\Omega, \;\;\; \frac{\partial v^{*}}{\partial \eta} = 0, \;\; \boldsymbol{x}\in\partial\Omega. \end{equation} | (A.3) |
The following preliminary results are also discussed in [15,25,29,30].
Lemma 11. Let u^{*} be a positive solution of (A.2) and v^{*} be a positive solution of (A.3); let K_{1}(\boldsymbol{x}) satisfy d_{1}\Delta K_{1}(\boldsymbol{x})\not\equiv 0 and K_{2}(\boldsymbol{x}) satisfy d_{2}\Delta K_{2}(\boldsymbol{x})\not\equiv 0 on \Omega . Thus,
\begin{equation} \int_{\Omega}r( \boldsymbol{x})K_{1}( \boldsymbol{x})\;d \boldsymbol{x} > \int_{\Omega}r( \boldsymbol{x})u^{*}( \boldsymbol{x})\;d \boldsymbol{x} \end{equation} | (A.4) |
and
\begin{equation} \int_{\Omega}r( \boldsymbol{x})K_{2}( \boldsymbol{x})\;d \boldsymbol{x} > \int_{\Omega}r( \boldsymbol{x})v^{*}( \boldsymbol{x})\;d \boldsymbol{x}, \end{equation} | (A.5) |
respectively.
Proof. First, we set v = 0 and u = u^{*} in the first equation of (1.2), and, utilizing the boundary conditions, as well as integrating over \Omega , we obtain
\begin{align} & r_{1}\int_{\Omega}ru^{*}\left(1-\dfrac{u^{*}}{K_{1}}\right)d \boldsymbol{x} = 0. \end{align} | (A.6) |
Adding and subtracting K_1 in (A.6), we have
\begin{align} \int_{\Omega}r\left(u^{*}-K_1+K_1\right)\left(1-\dfrac{u^{*}}{K_{1}}\right)d \boldsymbol{x} = 0. \end{align} | (A.7) |
Rewriting
\begin{align} \int_{\Omega}rK_1\left(1-\dfrac{u^{*}}{K_{1}}\right)d \boldsymbol{x} = \int_{\Omega}rK_1\left(1-\dfrac{u^{*}}{K_{1}}\right)^2d \boldsymbol{x} > 0, \end{align} | (A.8) |
we have
\begin{equation} \int_{\Omega}rK_{1}\left(1-\dfrac{u^{*}}{K_{1}}\right)d \boldsymbol{x} > 0. \end{equation} | (A.9) |
Simplifying (A.9), we obtain
\begin{align*} \int_{\Omega}r( \boldsymbol{x})K_{1}( \boldsymbol{x})d \boldsymbol{x} > \int_{\Omega}r( \boldsymbol{x})u^{*}( \boldsymbol{x})d \boldsymbol{x}. \end{align*} |
Analogously, the result (A.5) is justified.
Lemma 12. Assume that u^{*}(\boldsymbol{x}) is a positive solution of (A.2). Moreover, if K_{1}(\boldsymbol{x})\not\equiv const , then
\begin{equation} \int_{\Omega}rK_{1}\left(1-\dfrac{u^{*}}{K_{1}}\right)d \boldsymbol{x} > 0. \end{equation} | (A.10) |
Proof. The proof is analogous to that of Lemma 11.
Lemma 13. Assume that v^{*}(\boldsymbol{x}) is a positive solution of (A.3). Moreover, if K_{2}(\boldsymbol{x})\not\equiv const , then
\begin{equation} \int_{\Omega}rK_{2}\left(1-\dfrac{v^{*}}{K_{2}}\right)d \boldsymbol{x} > 0. \end{equation} | (A.11) |
Proof. The proof is analogous to that of Lemma 11.
[1] |
V. M. Valenzuela-Alcaraz, M. A. Cosio-Leon, A. D. Romero-Ocano, C. A. Brizuela, A cooperative coevolutionary algorithm approach to the no-wait job shop scheduling problem, Expert Syst. Appl., 194 (2022). https://doi.org/10.1016/j.eswa.2022.116498 doi: 10.1016/j.eswa.2022.116498
![]() |
[2] |
M. A. Salido, J. Escamilla, F. Barber, A. Giret, D. Tang, M. Dai, Energy efficiency, robustness, and makespan optimality in job-shop scheduling problems, AI EDAM, 30 (2016), 118–125. https://doi.org/10.1017/S0890060415000335 doi: 10.1017/S0890060415000335
![]() |
[3] |
M. Dai, Z. Zhang, A. Giret, M. A. Salido, An enhanced estimation of distribution algorithm for energy-efficient job-shop scheduling problems with transportation constraints, Sustainability, 11 (2019), 3085. https://doi.org/10.3390/su11113085 doi: 10.3390/su11113085
![]() |
[4] |
X. Hao, M. Gen, L. Lin, G. A. Suer, Effective multiobjective EDA for bi-criteria stochastic job-shop scheduling problem, J. Intell. Manuf., 28 (2017), 833–845. https://doi.org/10.1007/s10845-014-1026-0 doi: 10.1007/s10845-014-1026-0
![]() |
[5] |
Y. Wang, X. Wang, Inventory based two-objective job shop scheduling model and its hybrid genetic algorithm, Appl. Soft Comput., 13 (2013), 1400–1406. http://dx.doi.org/10.1016/j.asoc.2012.03.073 doi: 10.1016/j.asoc.2012.03.073
![]() |
[6] |
S. Nguyen, M. Zhang, M. Johnston, K. C. Tan, Automatic design of scheduling policies for dynamic multi-objective job shop scheduling via cooperative coevolution genetic programming, IEEE Trans. Evol. Comput., 18 (2013), 193–208. http://dx.doi.org/10.1109/TEVC.2013.2248159 doi: 10.1109/TEVC.2013.2248159
![]() |
[7] |
H. Wang, B. R. Sarker, J. Li, J. Li, Adaptive scheduling for assembly job shop with uncertain assembly times based on dual Q-learning, Int. J. Prod. Res., 59(2021), 5867–5883. https://doi.org/10.1080/00207543.2020.1794075 doi: 10.1080/00207543.2020.1794075
![]() |
[8] |
K. C. Ying, P. Pourhejazy, C. Y. Cheng, C. H. Wang, Cyber-physical assembly system-based optimization for robotic assembly sequence planning, J. Manuf. Syst., 58 (2021), 452–466. https://doi.org/10.1016/j.jmsy.2021.01.004 doi: 10.1016/j.jmsy.2021.01.004
![]() |
[9] |
A. De Giorgio, A. Maffei, M. Onori, L. Wang, Towards online reinforced learning of assembly sequence planning with interactive guidance systems for industry 4.0 adaptive manufacturing, J. Manuf. Syst., 60 (2021), 22–34. https://doi.org/10.1016/j.jmsy.2021.05.001 doi: 10.1016/j.jmsy.2021.05.001
![]() |
[10] |
A. Baykasoglu, F. S. Madenoglu, A. Hamzadayi, Greedy randomized adaptive search for dynamic flexible job-shop scheduling, J. Manuf. Syst., 56 (2020), 425–451. https://doi.org/10.1016/j.jmsy.2020.06.005 doi: 10.1016/j.jmsy.2020.06.005
![]() |
[11] |
A. Baykasoglu, F. S. Madenoglu, Greedy randomized adaptive search procedure for simultaneous scheduling of production and preventive maintenance activities in dynamic flexible job shops, Soft Comput., 25 (2021), 14893–14932. https://doi.org/10.1007/s00500-021-06053-0 doi: 10.1007/s00500-021-06053-0
![]() |
[12] |
A. Vital-Soto, A. Azab, M. F. Baki, Mathematical modeling and a hybridized bacterial foraging optimization algorithm for the flexible job-shop scheduling problem with sequencing flexibility, J. Manuf. Syst., 54(2020), 74–93. https://doi.org/10.1016/j.jmsy.2019.11.010 doi: 10.1016/j.jmsy.2019.11.010
![]() |
[13] |
G. Gong, R. Chiong, Q. Deng, X. Gong, A hybrid artificial bee colony algorithm for flexible job shop scheduling with worker flexibility, Int. J. Prod. Res., 58 (2020), 4406–4420. https://doi.org/10.1080/00207543.2019.1653504 doi: 10.1080/00207543.2019.1653504
![]() |
[14] |
R. Li, W. Gong, C. Lu, A reinforcement learning based RMOEA/D for bi-objective fuzzy flexible job shop scheduling, Expert Syst. Appl., 203 (2022), 117380. https://doi.org/10.1016/j.eswa.2022.117380 doi: 10.1016/j.eswa.2022.117380
![]() |
[15] |
R. Li, W. Gong, C. Lu, Self-adaptive multi-objective evolutionary algorithm for flexible job shop scheduling with fuzzy processing time, Comput. Ind. Eng., 168 (2022), 108099. https://doi.org/10.1016/j.cie.2022.108099 doi: 10.1016/j.cie.2022.108099
![]() |
[16] |
R. Liu, R. Piplani, C. Toro, Deep reinforcement learning for dynamic scheduling of a flexible job shop, Int. J. Prod. Res., 60 (2022), 4049–4069. https://doi.org/10.1080/00207543.2022.2058432 doi: 10.1080/00207543.2022.2058432
![]() |
[17] |
K. Lei, P. Guo, W. Zhao, Y. Wang, L. Qian, X. Meng, et al., A multi-action deep reinforcement learning framework for flexible Job-shop scheduling problem, Expert Syst. Appl., 205 (2022), 117796. https://doi.org/10.1016/j.eswa.2022.117796 doi: 10.1016/j.eswa.2022.117796
![]() |
[18] |
Y. Du, J. Li, X. Chen, P. Duan, Q. Pan, Knowledge-based reinforcement learning and estimation of distribution algorithm for flexible job shop scheduling problem, IEEE Trans. Emerging Top. Comput. Intell., 2022 (2022). https://doi.org/10.1109/TETCI.2022.3145706 doi: 10.1109/TETCI.2022.3145706
![]() |
[19] |
X. Li, K. Xing, Iterative widen heuristic beam search algorithm for scheduling problem of flexible assembly systems, IEEE Trans. Ind. Inf., 17 (2021), 7348–7358. https://doi.org/10.1109/TII.2021.3049338 doi: 10.1109/TII.2021.3049338
![]() |
[20] |
C. Finetto, M. Faccio, G. Rosati, A. Rossi, Mixed-model sequencing optimization for an automated single-station fully flexible assembly system (F-FAS), Int. J. Adv. Manuf. Technol., 70 (2014), 797–812. https://doi.org/10.1007/s00170-013-5308-z doi: 10.1007/s00170-013-5308-z
![]() |
[21] |
A. Hottenrott, M. Schiffer, M. Grunow, Flexible assembly layouts in smart manufacturing: An impact assessment for the automotive industry, IISE Trans., 2022 (2022). https://doi.org/10.1080/24725854.2022.2124470 doi: 10.1080/24725854.2022.2124470
![]() |
[22] |
W. Ren, J. Wen, Y. Yan, Y. Hu, Y. Guan, J. Li, Multi-objective optimisation for energy-aware flexible job-shop scheduling problem with assembly operations, Int. J. Prod. Res., 59 (2021), 7216–7231. https://doi.org/10.1080/00207543.2020.1836421 doi: 10.1080/00207543.2020.1836421
![]() |
[23] |
W. Lin, Q. Deng, W. Han, G. Gong, K. Li, An effective algorithm for flexible assembly job-shop scheduling with tight job constraints, Int. Trans. Oper. Res., 1 (2020). https://doi.org/10.1111/itor.12767 doi: 10.1111/itor.12767
![]() |
[24] |
P. Fattahi, N. B. Rad, F. Daneshamooz, S. Ahmadi, A new hybrid particle swarm optimization and parallel variable neighborhood search algorithm for flexible job shop scheduling with assembly process, Assem. Autom., 40 (2020), 419–432. https://doi.org/10.1108/AA-11-2018-0178 doi: 10.1108/AA-11-2018-0178
![]() |
[25] |
X. Wu, X. Liu, N. Zhao, An improved differential evolution algorithm for solving a distributed assembly flexible job shop scheduling problem, Memet. Comput., 11 (2019), 335–355. https://doi.org/10.1007/s12293-018-00278-7 doi: 10.1007/s12293-018-00278-7
![]() |
[26] |
S. Zhang, S. Wang, Flexible assembly job-shop scheduling with sequence-dependent setup times and part sharing in a dynamic environment, constraint programming model, mixed-integer programming model, and dispatching rules, IEEE Trans. Eng. Manage., 65 (2018), 487–504. https://doi.org/10.1109/TEM.2017.2785774 doi: 10.1109/TEM.2017.2785774
![]() |
[27] |
X. Li, J. Lu, C. Yang, J. Wang, Research of flexible assembly job-shop batch-scheduling problem based on improved artificial bee colony, Front. Bioeng. Biotechnol., 10 (2022). https://doi.org/10.3389/fbioe.2022.909548 doi: 10.3389/fbioe.2022.909548
![]() |
[28] |
Z. Xie, X. Zhang, Y. Xia, J. Yang, Y. Xin, A hybrid method of heuristic algorithm and constraint programming for no-wait integrated scheduling problem, J. Int. Technol., 22 (2021), 1085–1092. https://doi.org/10.53106/160792642021092205012 doi: 10.53106/160792642021092205012
![]() |
[29] |
Z. Xie, W. Zhou, Z. Yu, Integrated scheduling algorithm for dynamic adjustment of equipment maintenance start time, J. Mech. Eng., 57 (2021), 240–246. https://doi.org/10.3901/JME.2021.04.240 doi: 10.3901/JME.2021.04.240
![]() |
[30] |
Z. Xie, N. Lv, Integrated scheduling algorithm with pre-start device, J. Mech. Eng., 57 (2021), 217–225. https://doi.org/10.3901/JME.2021.17.217 doi: 10.3901/JME.2021.17.217
![]() |
[31] |
Z. Xie, H. Teng, J. Ming, X. Yue, A two-workshop collaborative, integrated scheduling algorithm considering the prescheduling of the root-subtree processes, Comput. Intell. Neurosci., 2022 (2022). https://doi.org/10.1155/2022/9065638 doi: 10.1155/2022/9065638
![]() |
[32] |
X. Zhan, Z. Xie, D. Yao, Integrated scheduling algorithm of two workshops based on process end time driven and processing area priority, Electronics, 11 (2022), 2594. https://doi.org/10.3390/electronics11162594 doi: 10.3390/electronics11162594
![]() |
[33] |
Y. Gao, Z. Xie, X. Yu, A hybrid algorithm for integrated scheduling problem of complex products with tree structure, Multimedia Tools Appl., 79 (2020), 32285–32304. https://doi.org/doi.org/10.1007/s11042-020-09477-2 doi: 10.1007/s11042-020-09477-2
![]() |
[34] |
Z. Wang, C. Lu, An integrated job shop scheduling and assembly sequence planning approach for discrete manufacturing, J. Manuf. Syst., 61 (2021), 27–44. https://doi.org/10.1016/j.jmsy.2021.08.003 doi: 10.1016/j.jmsy.2021.08.003
![]() |
[35] |
Y. Gao, Z. Xie, D. Yang, X. Yu, Flexible integrated scheduling algorithm based on remaining work probability selection coding, Expert Syst., 38 (2021), e12683. https://doi.org/10.1111/exsy.12683 doi: 10.1111/exsy.12683
![]() |
[36] |
Z. Xie, Z. Gui, J. Yang, Dynamic parallel integrated flexible scheduling algorithm based on device driver and essential path, J. Mech. Eng., 50 (2014), 203–212. https://doi.org/10.3901/JME.2014.18.203 doi: 10.3901/JME.2014.18.203
![]() |
[37] |
Z. Xie, H. Zhou, J. Yu, Z. Gui, Conflict mediation algorithm of integrated flexible scheduling based on device driver, Trans. Beijing Inst. Technol., 34 (2014), 1150–1156. https://doi.org/10.3969/j.issn.1002-137X.2013.04.042 doi: 10.3969/j.issn.1002-137X.2013.04.042
![]() |
[38] |
H. Lu, G. Huang, H. Yang, Integrating order review/release and dispatching rules for assembly job shop scheduling using a simulation approach, Int. J. Prod. Res., 49 (2011), 647–669. https://doi.org/10.1080/00207540903524490 doi: 10.1080/00207540903524490
![]() |
[39] |
Z. Xie, S. Hao, G. Ye, G. Tan, A new algorithm for complex product flexible scheduling with constraint between jobs, Comput. Ind. Eng., 57 (2009), 766–772. https://doi.org/10.1016/j.cie.2009.02.004 doi: 10.1016/j.cie.2009.02.004
![]() |
[40] |
Z. Xie, Q. Wang, Flexible integrated scheduling algorithm based on reverse order layer priority, J. Electron. Inf. Technol., 44 (2022), 1554–1562. https://doi.org/10.11999/JEIT211378 doi: 10.11999/JEIT211378
![]() |
[41] |
R. El-Khalil, Z. Darwish, Flexible manufacturing systems performance in U.S. automotive manufacturing plants, a case study, Prod. Plann. Control, 30 (2019), 48–59. https://doi.org/10.1080/09537287.2018.1520318 doi: 10.1080/09537287.2018.1520318
![]() |
[42] |
X. Yang, J. Liu, Q. Chen, N. Mao, Performance analysis of flexible assembly job shop scheduling under variable distrurbance intensity, Comput. Integr. Manuf. Syst., 27 (2021), 800–814. https://doi.org/10.13196/j.cims.2021.03.013 doi: 10.13196/j.cims.2021.03.013
![]() |
1. | Muhammad Mohebujjaman, Clarisa Buenrostro, Md. Kamrujjaman, Taufiquar Khan, Decoupled algorithms for non-linearly coupled reaction–diffusion competition model with harvesting and stocking, 2024, 436, 03770427, 115421, 10.1016/j.cam.2023.115421 |