2024 Volume 33 Issue 6
Article Contents

Ling Xu, Lei Jiang. Effects of asymmetric coupling and boundary on the dynamic behaviors of a random nearest neighbor coupled system[J]. Chinese Physics B, 2024, 33(6): 060503. doi: 10.1088/1674-1056/ad3b86
Citation: Ling Xu, Lei Jiang. Effects of asymmetric coupling and boundary on the dynamic behaviors of a random nearest neighbor coupled system[J]. Chinese Physics B, 2024, 33(6): 060503. doi: 10.1088/1674-1056/ad3b86

Effects of asymmetric coupling and boundary on the dynamic behaviors of a random nearest neighbor coupled system

  • Received Date: 02/02/2024
    Accepted Date: 25/03/2024
    Available Online: 01/06/2024
通讯作者: 陈斌, bchen63@163.com
  • 1. 

    沈阳化工大学材料科学与工程学院 沈阳 110142

  1. 本站搜索
  2. 百度学术搜索
  3. 万方数据库搜索
  4. CNKI搜索

Figures(13)

Article Metrics

Article views(171) PDF downloads(0) Cited by(0)

Access History

Effects of asymmetric coupling and boundary on the dynamic behaviors of a random nearest neighbor coupled system

Abstract: This study investigates the dynamical behaviors of nearest neighbor asymmetric coupled systems in a confined space. First, the study derivative analytical stability and synchronization conditions for the asymmetrically coupled system in an unconfined space, which are then validated through numerical simulations. Simulation results show that asymmetric coupling has a significant impact on synchronization conditions. Moreover, it is observed that irrespective of whether the system is confined, an increase in coupling asymmetry leads to a hastened synchronization pace. Additionally, the study examines the effects of boundaries on the system’s collective behaviors via numerical experiments. The presence of boundaries ensures the system’s stability and synchronization, and reducing these boundaries can expedite the synchronization process and amplify its effects. Finally, the study reveals that the system’s output amplitude exhibits stochastic resonance as the confined boundary size increases.

1.   Introduction
  • It is widely accepted that a system characterized by a multitude of interacting individuals is denominated a complex system. Due to its complex characteristics and rich collective behaviors, it has become a research hotspot[1,2] in multiple interdisciplinary fields and is therefore widely used in biology, engineering, and even social sciences.[3–5] The complexity is not only reflected in the number of individuals contained in the system and the degree of freedom of the system, but also in the form of interaction between individuals, and these forms of interaction are usually modeled as coupling, which can be realized information exchange within complex systems, such as biochemical reactions of positive feedback and negative feedback regulation processes in multicellular biological systems,[6] power distribution, and fault identification networks in power grids.[7] These couplings can be established as regular networks, such as global coupling,[8] nearest neighbor coupling,[9] and star coupling,[10] or they can be established as random networks, small-world networks, etc.[11]

    There are various collective behaviors of complex systems, and synchronicity is one of the most basic manifestations and applications.[12] The mechanisms behind the cooperative behavior of complex systems are directly related to synchronization. For example, clock synchronization in power systems ensures the stability of power frequency and accurate power dispatch,[13] and the phenomenon of synchronous flashing in groups of fireflies is another example.[14] Therefore, the synchronization mechanism behind complex systems has received extensive attention from researchers. Existing research has pointed out that the coupling form and the introduction of noise have an important impact on the synchronization of the system.[15] Initially, it was generally believed that noise would destroy the synchronization of the system, whereas coupling usually maintained the synchronization of the system.[16] However, the researchers later discovered that noise can also promote complete synchronization between subsystems. For example, the introduction of noise enhances the synchronization behavior of nonlinear coupling systems.[17] The introduction of dichotomous noise in stochastic bistable systems significantly enhances the phase synchronization effect.[18] The authors in Ref. [19] found that the impact of noise on the contraction domain is greater than the impact on the expansion domain, and explained the internal mechanism of perfect synchronization caused by noise. However, in most cases, especially for non-deterministic complex systems, it is not easy to obtain precise synchronization conditions. In addition, as the scale and complexity of the system increase, how to ensure the stability of the system becomes increasingly important. Stability is the ability of a system to return to an equilibrium state when subjected to external disturbances or internal noise.[20] In stochastically coupled systems, the stability conditions are more stringent because the system is not only disturbed by external perturbations but also by internal noise. Researchers have proposed some methods to analyze the stability of complex systems. These methods include Lyapunov stability, Hamiltonian stability, impulse stability, etc.[21] These stability conditions provide important theoretical tools for analyzing and predicting system behavior. The asymmetry of complex systems has a crucial impact on their collective behaviors.[22] Asymmetry can arise not only through the system’s special network topology, but also through asymmetric coupling. In practical applications, asymmetric coupling widely exist in complex systems such as signal transmission,[23] brain neural networks,[24] swarm intelligence networks,[25] and biomolecular motors,[26] and produce rich group effects, such as stochastic resonance (SR), cluster synchronization, etc. For example, Zhang et al.[23] studied signal transmission under a one-way coupled bistable system and discovered SR. Hillier et al.[27] found that asymmetric coupling causes chaotic oscillator arrays to exhibit a new form of cluster synchronization. Antonio[28] confirmed the existence of stable synchronization states in asymmetrically coupled homogeneous oscillators. Batra et al.[29] found that driven by quadratic asymmetric bipartite noise, the system still exhibits SR when the potential field parameter is zero. Based on the important role of asymmetric coupling in the system, it is necessary to study the collective behavior of complex systems under asymmetric coupling.

    In addition, traditional research mainly considers macroscopic systems dominated by pure potential energy, that is, the system is in an unrestricted space. However, systems are sometimes restricted, even mesoscopic.[30,31] Research has found that systems in confined spaces may exhibit completely different dynamic properties. Burada[32] considered the free diffusion of Brownian particles in a dumbbell-shaped microcavity and discovered entropic SR. Zhang et al.[33] found that the unrestricted linear overdamped system has no SR, which occurs when the system is restricted. Chen et al.[34] studied the collective dynamic behavior of an adjacent coupling chain in unconfined and confined spaces, and found that the existence of boundaries can promote the synchronization and stability of the system, and exhibit more complex SR. Therefore, the scale and shape of space have an important impact on the dynamic properties of the system.

    This study will investigate the collective behaviors of nearest neighbor asymmetric coupling chains driven by multiplicative dichotomous noise and cosine periodic forces in a one-dimensional (1D) confined space. The precise stability and synchronization conditions of the unconfined system will be solved and the reliability verified. Discuss the effects of asymmetric coupling, confined space, and other system parameters on synchronization region, synchronization speed, and stability conditions, and analyze their mechanisms. In addition, this study will explore the SR of the system.

    The arrangement of this study is as follows: Section 2 mainly introduces the nearest neighbor asymmetric coupling chain model in 1D unconfined and confined spaces. In Section 3, the main results of this article are given through numerical simulation and theoretical derivation. Finally, Section 4 summarizes the conclusions.

2.   Model
  • In the study, we consider a stochastic nearest neighbor asymmetric coupled system. Its dynamics can be described by the following linear over-damped Langevin equation:

    where xi(t) represents the displacement of the i-th particle at time t. It is only connected to its nearest two neighbors by linear coupling. In particular, the first particle is only coupled to the second particle, and the last particle is only coupled to the penultimate particle. ε1 and ε2 are the coupling strengths with the left neighbor and right neighbor, respectively. The coupling coefficients along different directions are not the same (Fig. 1). ε1(xi+1(t)–xi(t)) + ε2(xi–1(t)–xi(t) is the nearest neighbor force for the i-th (i = 2,…,N–1) particle. In particular, the nearest neighbor coupling force for the first particle is ε1(x2(t)–x1(t)). ε2(xN–1(t)–xN(t)) is the nearest neighbor coupling force of the last particle. The nearest neighbor coupling model is widely used in the study of the dynamic behavior of the Josephson structure arrays,[35,36] the neural networks,[37] and the coupled protein motors.[35]

    A0 ≥ 0 and Ω > 0 are the external periodic-driven amplitude and frequency of the system (1), respectively. Fluctuation of the potential field parameter ω > 0 is described by the symmetric dichotomous noise ξ(t).[38–40] It is a random stationary Markovian process that consists of jumps between two values ±σ with σ > 0. The mean value and the autocorrelation of ξ(t) satisfy

    where σ is the noise amplitude, λ is the transition rate, and it is the reciprocal of the noise correlation time τc = 1/λ. Affected by potential field fluctuations, the potential well of particle xi switches randomly between V±(x) = (ω±σ)x2/2.

    (i) When 0 < σ < ω, both potential wells V±(x) open upward. In this case, the particle motion is bounded by the potential field force, driven by external periodic forces and noise to make a bounded motion near the bottom of the potential field. Thus, both V±(x) are in steady states.

    (ii) When 0 < ω < σ, V+(x) opens upwards, but V−(x) opens downwards. Particle xi may slide to infinity in V−(x) under the action of noise. Thus, V+(x) is stable, but V−(x) is unstable. In this case, the potential field switches randomly between stable and unstable.

    Thus, the stability of the output response of the system (1) is not trivial. The physical meaning of this potential energy fluctuation can be found in biological systems, such as the ATP-ADP cyclic potential. In biological systems, ATP releases energy by hydrolyzing the bond, which involves the conversion of energy and ADP. This energy conversion process has similarities with the fluctuations of the model’s potential energy. Moreover, the asymmetric coupling in molecular motors will lead the system to be subjected to periodic force driven in the ATP-ADP potential field, which can be seen as an application scenario of the model (1). In Ref. [41], the study of this type of model (1) implies its potential applications in different scenarios. In Ref. [23], the signal transmission of a similar model under one-way coupling conditions was investigated, and it was pointed out that this type of model has widespread applications in open flow systems and one-way driven dynamic chains. In particular, when N = 1, system (1) degenerates into a single particle model.[42] It is widely used in many fields, such as problems of fluctuating barrier crossing,[43] enzimatic kinetics,[44] nuclear magnetic resonance,[45] gene expression,[46] and electric circuit.[47] In fact, if one considers the single particle model as a limiting case of a nonlinear problem, dozens of different applications may exist.[42]

    Since many real physical systems are always confined, studying confined systems has its practical application interest.[30–34] Thus, in this study, we also consider the asymmetric nearest neighbor coupling chain with N particles in a confined space. Its dynamic model can be described by the following dimensionless linear overdamped Langevin equations:

    where the displacement of the i-th particle xi(t) is confined to the interval [−B,B] (Fig. 2). When the particle hits the boundary, it is bounced back into [−B,B] with the same speed and opposite direction.

3.   Dynamics of asymmetric coupled nearest neighbor systems
  • To analyze the impact of coupling asymmetry on the dynamics nearest neighbor coupling chain. In this section, we give the stability condition and synchronization condition of the system in the unconfined case (system (1)).

  • The stability discussed here is bounded-input–bounded-output stability of the first and second moments of the particles’ displacements. For this purpose, it is necessary to give the complete system of equations satisfied by 〈xi〉. Firstly, we average over all realizations of the trajectories of the stochastic Eqs. (1), which gives

    Then, applying the Shapiro–Loginov (SL) formula

    we can get the complete system of equations for 〈xi〉. Finally, based on the linear system stability criterion,[48] the stability condition of the first moment 〈xi〉 is derived in the following equation:

    Besides, we also need to consider the fluctuation of each particle’s displacement, that is, the degree to which each particle deviates from its average displacement, which is described by the variance 〈Δi〉 of xi

    Since 〈Δi〉=〈xi2〉−〈xi〉2, it can be seen that when the first moment 〈xi〉 is stable, the stability of 〈Δi〉 is equivalent to the stability of the second moment 〈xi2〉. Using the SL formula and equation d〈xixj〉/dt=〈xi.xj〉+〈xixj.〉, the stability condition of the second moment 〈xi2〉 for Eq. (1) is as follows:

    The derivation process of inequalities (5) and (7) is given in Appendix A. Because σsta22<σsta12, the stability of the second moment of particles' displacements implies the stability of the first moments. From Eqs. (5) and (7), it can be seen that the stability of the output responds of system (1) only depends on the noise amplitude σ, noise correlation rate λ, and potential field parameter ω, but has nothing to do with the number of particles N, coupling coefficients ε1, ε2, and the external periodic force A0cos(Ωt). Therefore, it can be considered that the stability of the system (1) is determined by the random potential field V±(xi).

  • To obtain the synchronization condition of system (1), we need to define new variables δi = xi+1–xi,i = 1,…,N–1. Subtracting the previous one from the latter equation in Eqs. (1), the following homogeneous linear equations for δi are obtained

    We call the system (1) statistical synchronization if the first moments 〈δi〉,i = 1,…,N–1 are stable and 〈δi〉 = 0. Similarly, using the SL formula, a complete system of equations including 〈δi〉 can be obtained. By analyzing the distribution of the eigenvalues of its coefficient matrix, the synchronization condition of system (1) is as follows:

    where

    We call inequality (9) the synchronization condition, and its proof is given in Appendix B.

    Since

    so we have σsta12≤σsyn2. Combined with inequalities (5) and (9), it can be seen that when the first moment of the displacement of each particle is stable, the system (1) is statistical synchronization. This is

    According to the synchronization condition (9), in addition to the potential field parameter ω and noise correlation rate λ, the synchronization interval (0,σsyn2) of the system (1) also depends on the number of particles N and the coupling strengths ε1,ε2. This suggests that the system topology affects its synchronization.

  • In this section, we will verify the main results given in Subsections 3.1.1 and 3.1.2 by numerical algorithm based on the stochastic Euler algorithm.[49] For a sufficiently small time step Δt, the approximate representation of the model (1) in discrete time is

    where

    Δt = tn – tn–1 is the time step, ΔXi is the increment of the dichotomous noise ξ(t) during a single time step, which can be simulated through the method proposed in Ref. [50]. Let the number of simulations be K, xi,k(t) denotes the displacement of the i-th particle in the k-th simulation. Thus, when K is large enough, the simulation approximations of the first moment and second moment of the displacement of the i-th particle are

    respectively. Therefore, the system mean field 〈S(t)〉 and the mean value of the second moment 〈S2(t)〉 can be obtained by the following simulation:

  • In order to verify the stability conditions of the system (1) (i.e. inequalities (5) and (7)), figure 3 gives the curves of the system mean field 〈S(t)〉 and the second moment 〈S2(t)〉 changing with time t under different noise intensities σ2. Combined with the parameters given in Fig. 3, the stability region threshold of the system’s first (second) moment is σsta12=2 (σsta22=1.5). It can be seen from Fig. 3 that when the noise intensity σ2 is less than the noise threshold, the amplitudes of the first (second) moment of the system are always bounded, and the first moment of the system output is a cosine function. However, as the noise intensity σ2 exceeds the stability condition threshold, the amplitudes of 〈S(t)〉 and 〈S2(t)〉 gradually diverge to unbounded. This is consistent with inequalities (5) and (7).

  • In order to verify the synchronization condition given in inequality (9), we first give plots of phase diagram of the synchronization region in σ2–ε1–ε2 coordinate system in Fig. 4(a) and in σ2–N plane in Fig. 4(b). Figure 4(a) gives a surface plot of the synchronization region as a function of coupling strengths ε1 and ε2. The gray surface is the synchronization region boundary (σ2=σsyn2), below which is the synchronization region of the system (1). It can be seen that the synchronization region is symmetrical about ε1 = ε2. This indicates that reversing the coupling strength of the two directions does not affect the synchronization. In addition, we can also see that the difference in the coupling strength in the two directions has a significant impact on the synchronization region. In order to reflect this change more intuitively, we set ε1+ε2 = 2, and draw the boundary of the synchronization region under different coupling strength differences as shown in Fig. 4(b). We found that the greater the difference between ε1 and ε2, the larger the synchronization region is. And the easier it is for the stochastic asymmetric nearest neighbor coupled system (1) to realize the synchronization. This is because, in system (1), the particles in the middle of the coupling chain is subjected to the coupling force of the two adjacent particles. The two forces are in opposite directions. When the coupling strengths in the two directions are close, the two forces acting on the intermediate particle will partially cancel, and the particles reciprocate in the equilibrium position, making the system more difficult to synchronize. However, with the increase of the difference in the coupling strength between the two directions, the offset portion of the coupling forces in the two directions decreases, making it easier for the system to achieve the synchronization.

    Then, we introduce the synchronization error Esyn of the first moment of the adjacent particles’ displacements as a measurement index:

    Figure 5 illustrates the influences of noise intensity σ2 on synchronization error Esyn. The synchronization threshold of the system here is σ02=8.0469. As can be seen from Fig. 5, under each noise intensity, the synchronization error Esyn will level off after a period of time. Its stable value increases with the increase of noise intensity σ2. When σ2<σ02, the Esyn is relatively small (approximately less than 10−2) and the system is synchronized. When σ2>σ02, Esyn diverges to 105. Thus the system is out of synchronization. The simulation results in Fig. 5 are consistent with the synchronization condition (9).

  • Since the movement of particles in the confined space is difficult to solve analytically, we use numerical iterative algorithms to analyze their collective dynamic behaviors such as stability, synchronization, and SR. Let yi(tn) represent the displacement of the i-th particle at time tn in the confined space. Firstly, generate a set of random initial values xi(t0), let yi(t0) = xi(t0). Then, equation (16) is used to iterate. When |xi(tn)| ≤ B, set yi(tn) = xi(tn) and continue to use Eq. (16) to generate xi(tn+1). When |xi(tn)| > B, we assume that the particle hits the boundary within [tn–1, tn] and bounces back to the interval, with the speed unchanged and in the opposite direction. Then we obtain yi(tn), set xi(tn) = yi(tn). Use Eq. (16) to iteratively generate xi(tn+1). Therefore, the iterative formula for the displacement of the i-th particle in the confined space is as follows:

    Figure 6 shows the particle displacement curve in a single simulation in a confined space. As shown in Fig. 6(a), when the boundary value B = 0.5, the time–domain plot of the confined system exhibits an approximate square wave. As the boundary gradually increases, as seen in Fig. 6(b) for the boundary value B = 10, the time–domain plot of the confined system (3) becomes almost identical to that of the corresponding unconfined system. This means that, in this dataset, the boundary weakens its restriction on the system. Therefore, we can observe that the boundary may have a significant impact on the dynamic properties of the system, and it also suggests that we might be able to optimize the dynamic properties of the system by adjusting the boundary value.

  • In order to analyze the impact of the boundary on the stability of the system, we use Eq. (16) to perform numerical simulation on model (3). Figure 7 show the curves of the first-moment 〈S(t)〉 and the second-moment 〈S2(t)〉 changing with time t at different boundaries B and different noise intensities σ2. Here, we use different line styles and markers to distinguish each curve. In particular, we have made a partial enlargement in Fig. 7 so that readers can better distinguish curves in different categories. Under the parameters given in the figure, the stability region boundary of the first (second) moment of the unconfined model is σsta12=2 (σsta22=1.5). The average output amplitudes when σ2 = 1 and σ2 = 2 are 0.7670 and 0.7071, respectively.

    From the four subfigures in Fig. 7, we can see that no matter what values B and σ2 take, the first-moment 〈S(t)〉 and the second-moment 〈S2(t)〉 of the system are stable. In Figs. 7(a) and 7(b), the boundary width is smaller than the original output amplitude when the system is unconfined. The first-order moment amplitude decreases as the noise increases. This is because regardless of the noise increases, the particle displacement can reach the boundary position. The increase in noise will increase the disorder of the particle’s single displacement. Therefore, the peaks and troughs of the two particle trajectories do not overlap, so that the average displacement decreases as the noise intensity increases. In Figs. 7(c) and 7(d), the boundary width is greater than the original output amplitude when the system is unconfined. The first-order moment amplitude increases as the noise increases. This is because when the noise intensity is small, the system is stable and the particle displacement does not diverge to the boundary. As the noise intensity increases, the particle motion becomes more disordered and the particle displacement diverges to the boundary more frequently. Therefore, the amplitude of 〈S(t)〉 increases with the increase of noise. Comparing the uncondined model (1), it can be seen that the boundary plays a decisive role in the stability of system (3). In fact, by performing Fourier transform on the two curves σ2 = 10 in Figs. 7(a) and 7(c) (see Fig. 8), it can be seen that its spectrum is mainly composed of fundamental harmonic with frequency Ω and high-order harmonics with frequency (2k + 1)Ω, k = 1,2,…, as shown in Fig. 8.[33] This also confirms that no matter what values B and σ2 take, the confined system is always stable.

  • In this section, we analyze the synchronicity of the confined coupled system (3). Figure 9 depicts the curves of the average displacement difference Esyn of the system as a function of time t under different noise intensities σ2 and different boundaries B. From this we can see that the system (3) is always synchronized regardless of the noise intensity and boundary values. In Fig. 9(a), it can be seen that as the noise intensity increases, the synchronization speed slows down and the synchronization error gradually increases. Figure 9(b) shows that the larger the boundary of the confined system, the weaker the synchronization control effect, and the slower the synchronization speed. This is primarily due to the fact that noise disrupts the system’s orderly output, and the presence of a boundary constraints the movement of particles, leading to a more organized state. Consequently, an increase in noise intensity and boundary width contributes to greater disorder in the system’s output, which in turn makes particle synchronization more challenging. This results in a longer time required for synchronization and larger synchronization errors.

    Figure 10 depicts the curves of Esyn at different coupling strengths in confined and unconfined spaces. Comparing Figs. 10(a) and 10(b), it can be seen that whether in the confined space or an unconfined space, the increase in coupling asymmetry will speed up the synchronization speed of the system, making the system enter the synchronization state faster. This is because an increase in the coupling asymmetry in the two directions will make the forces on the intermediate particles in the coupling chain more unbalanced in the two directions, reducing the loss of the coupling force within the system, making it easier for the movements of each particle to become consistent, thereby speeding up system synchronization. It can be seen from Figs. 10(b) and 10(c) that the confined system (3) is always synchronized regardless of the coupling strengths. Furthermore, figure 10(c) shows that increasing coupling strength or coupling asymmetry not only speeds up synchronization but also reduces synchronization errors. This is mainly due to the fact that the increase in coupling strength or coupling asymmetry suppresses the randomness of the noise, making the system output more orderly.

    Figure 11 depicts the curves of Esyn at different particle number N in confined and unconfined spaces. It can be seen from the two sub-figures in Fig. 11 that the increase in the number of particles will slow down the synchronization speed of the system, whether in a confinted space or an unconfined space. In addition, figure 11(b) indicates that the convergence value of Esyn will increase with the number of particles. This is mainly because, in the nearest neighbor coupled system, the effect of the coupling force is transmitted one by one through adjacent particles. Increasing the number of particles will increase the propagation distance of the coupling force. Therefore, the more particles there are, the longer the synchronization time will be, and the greater the synchronization error will be.

    From Figs. 9–11, we can conclude that, unlike the unconfined system (1), the average displacement of particles in the confined system (3) is always synchronous, regardless of the noise intensity σ2, the coupling strengthes ε1 and ε2, and the number of particles N. In other words, boundary B plays a decisive role in the synchronization of the system. In addition, the interaction of the four elements B, N, σ2, and ε1, ε2 will also affect the synchronization speed and synchronization effect of the system. In order to more intuitively analyze the impact of each system parameter on synchronization speed, we introduce the synchronization time tsyn. Set the threshold Δ* = 10−8, when Esyn < Δ*, we assume that the average displacement of each particle is synchronized, and define the time tsyn as

    Figure 12(a) shows the curve of tsyn as a function of the coupling difference Δε, where Δε = ε1–ε2. It can be seen from the figure that the increase in coupling difference in the two directions can shorten the synchronization time. In addition, we also found that as ε1 + ε2 increases, the curves decrease more and more slowly. This indicates that increasing the coupling strength in both directions at the same time can also speed up the synchronization speed of the system.

    Figure 12(b) depicts the curve of tsyn as a function of the particle number N under different boundary B. It can be seen that the synchronization time increases as the number of particles increases. That is, an increase in the number of particles will hinder the realization of system synchronization. This confirms the conclusion of Fig. 11. Therefore, the more particles there are, the longer the synchronization time will be. In addition, comparing the three curves in Fig. 12(b), we can see that increasing the boundary will increase the synchronization time. The generation mechanism is similar to Fig. 9(b). That is, noise disrupts the average displacement of particles. When B increases, the boundary’s restriction on particle movement weakens, making the system output more disordered and the synchronization time increases.

    Figure 12(c) depicts the curve of tsyn as a function of the noise intensity σ2 under different boundary B. It can be seen that the synchronization time increases with the increase of noise intensity regardless the boundary. This is mainly because the increase in noise intensity intensifies the disorder of the average particle displacement, thereby inhibiting the synchronization of the confined system.

  • From the previous analysis, it can be seen that the confined system (3) is always stable and synchronous. The system output is the superposition of several sinusoidal functions. In this section, we use the system output amplitude gain G = A/A0 as the indicator to analyze the SR phenomenon of the system (3), where A is the system output amplitude. For unconfined system (1), due to the synchronicity of the system, the study of SR can be attributed to the study of single particle behavior. The corresponding SR phenomenon, SR condition and mechanism analysis can be found in Ref. [8].

    Figure 13 illustrates the influences of the boundary B on the output amplitude gain G. We can see that the parameter stochastic resonance (PSR)[51] to B occurs under different values of σ2. Its physical mechanisms is as follows: The generation of PSR requires the transfer of part of the noise energy to the system output signal. That is, PSR is the result of the competition between disorder and order in the system. For a given noise intensity, when the boundary is too small, the system output is orderly under the boundary limit. As the boundary increases, the boundary’s restriction on particle motion gradually weakens, and the system output becomes disordered. When the boundary is too large, the system output is too disordered. Therefore, a boundary that is too small or too large cannot achieve an optimal match between the noise and the output signal. PSR can only occur in intermediate states. Additionally, in Fig. 13, when σ2 = 4.5, as the boundary increases, the system (3) exhibits PSR, and under suitable boundaries, the output amplitude gain G at the peak of the PSR is G > 1. This means that with an appropriate boundary, the system’s output amplitude can exceed the amplitude of the periodic driving force, leading to amplitude amplification. This phenomenon may occur in nonlinear complex systems, and both the mode of coupling and the topology of complex networks can lead to amplitude amplification, as referenced in Refs. [52,53]. As depicted in Fig. 13 with σ2 = 4.5, the combined effect of asymmetric coupling and boundaries in this study causes the linearly coupled system to exhibit nonlinear effects, thereby amplifying the system’s output amplitude. Of course, this amplitude amplification can be beneficial, such as efficient conversion of input energy and signal amplification, but it can also be detrimental, as excessively large amplitudes may lead to energy dissipation. When σ2 increases, the resonance intensity increases, and the position of the resonance peak shifts to the right. This is because an increase in noise intensity σ2 transfers more noise energy to the signal energy of the system. But B limits the movement range of the particles, thereby limiting the output amplitude of the system. At this time, increasing B can increase G, thereby shifting the resonance peak to the right. Under the parameters given in Fig. 13, the unconfined system (1) is unstable, so there is no SR. This shows that the existence of the boundary weakens the impact of noise on the orderliness of the system and expands the resonance area of the system.

4.   Conclusion
  • To summarize, we mainly explored the effects of coupling asymmetry and boundary on the dynamic behaviors of an nearest neighbor coupled system with random dichotomous fluctuation potential under an external periodic sine force. Firstly, the precise conditions of stability, synchronization of the asymmetric nearest neighbor coupled unconfined system (1) are derived analytically and verified by numerical simulation. Subsequently, we incorporate the system boundary and explore the impact of the boundary on the system’s dynamic behaviors, analyzing its effects on stability, synchronization, and SR.

    Asymmetric coupling has an important impact on system synchronization. It is observed that increased coupling asymmetry leads to wider synchronization regions and accelerated synchronization speeds regardless of whether the system is constrained or not. This is mainly caused by the topology of the system. In a nearest neighbor coupling system, a particle is subject to the coupling force of two adjacent particles in the coupling chain, and the two coupling forces are in opposite directions. When the coupling strengths in the two directions are close, the two forces acting on the intermediate particle will partially cancel out, and the particle will reciprocate in the equilibrium position, making the system more difficult to synchronize. However, as the difference in coupling strength in the two directions increases, the resultant force acting on the intermediate particles tends to be in a unified direction, which makes the particles more inclined to move in the same direction, making it easier for the system to achieve synchronization.

    Boundary plays the most important role in the collective behaviors of the asymmetric coupled system. Unlike the unconfined system, the existence of boundary makes the system always stable and synchronized. Reducing the boundary can shorten the synchronization time and enable the system to achieve better synchronization effects with the fixed parameters σ2, N, ε1, ε2. In addition, we also observed the PSR of the confined system with respect to the boundary. Compared with the unconfined system, this shows that the boundary can enrich the resonance form of the system and expand the resonance region of the system. Particularly, appropriate boundary values may cause the system’s output amplitude can exceed the amplitude of the periodic driving force, leading to amplitude amplification. That is, the combined effect of asymmetric coupling and boundaries in this study causes the linearly coupled system to exhibit nonlinear effects, thereby amplifying the system’s output amplitude.

    Although the results in this study are carried out around a specific coupling model, its method can be extended to other regular coupling networks such as star coupling networks. Besides, we believe that these conclusions in this study are highly generalizable given that asymmetric coupled confined systems are widely present in various biological and physical mesoscopic systems. Through these studies, we can better understand the dynamical behavior of the system and design more stable and efficient systems for practical problems, and expected to be very useful for further investigation of synchronous control and SR of other complex coupled systems.

Appendix A: Appendix for Subsection 3.1.1
  • In this section, the corresponding stability conditions both for the first and second moments of the particles' displacements are derived. To analyze the stability of the first moment of the system (1), we first give the complete system of equations satisfied by 〈xi〉. Averaging over all realizations of the trajectories of the stochastic Eqs. (1), we get Eqs. (4). A new coupling term 〈ξxi〉 appears in Eqs. (4), in order to solve it, we multiply each equation in system (1) by ξ and take the average. Applying the Shapiro–Loginov formula 〈ξ⋅dxi/dt〉 = d〈ξxi〉/dt+λ〈ξxi〉, we have

    Equations (4) and (A1) can be written in the following matrix form:

    where X = (〈x1〉,…,〈xn〉,〈ξx1〉,…,〈ξxN〉), F(t) = (A0cos(Ωt),…,A0cos(Ωt),0,…,0),

    where E is the identity matrix. The first moment of particles’ displacement of the system (1) is stable if and only if the roots of the characteristic equation r(s)=|sE+U∼| of Eq. (A2) are all distributed in the left-half complex plane, that is, the eigenvalues of matrix U∼ are all distributed in the right-half complex plane.

    Theorem A1 The first moment of particles displacements of the asymmetric nearest neighbor coupled system (1) is stable if and only if σ2<σsta12=ω2+λω.

    Proof Let f(k)=|kE−U∼|, the roots of f(k) are the eigenvalues of the matrix U∼. Simplifying f(k), we have

    Let the eigenvalues of N order square matrix U be r1,…,rN, set rsta1=min1≤i≤N{ri}. The Jordan standard form of U is J = P−1UP. Thus,

    According to Eq. (A4), the N roots of f(k) = 0 are ki=ri+(λ±λ2+4σ2)/2, i = 1,…,N. The first moment of particles displacements of the system (1) is stable if and only if the smallest root of f(k) is greater than 0. This is, rsta1+(λ−λ2+4σ2)/2>0. By direct calculations, we have σ2<rsta12+λrsta1. Since the sum of each row of U is ω, ω is one of the eigenvalues of U. From Gerschgorin circle theorem,[54] we have rsta1 = ω. In summary, the first moment of particles’ displacements of the system (1) is stable if and only if σ2 < ω2 + λω.

    To obtain the stability condition of the second moment of the system (1), we multiply Eqs. (1) by xi, i = 1, 2, …, N, respectively, and average them. Using d〈xixj〉/dt=〈xi.xj〉+〈xixj.〉 to simplify these formulas, we have

    To solve the new coupling term 〈ξxixj〉, we multiply Eqs. (1) by ξxi, respectively, and average them. By Shapiro–Loginov formula, we obtain

    Equations (A5) and (A6) can be written in the following matrix form:

    where, Y=(〈x12〉,…, 〈xixj〉,…, 〈xN2〉, 〈ξx12〉,…, 〈ξxixj〉,…, 〈ξxN2〉), it contains 2N2 components.

    where, U1 = U + (ω+ε1)E, U2 = U + (ω+ε1+ε2)E, U3 = U + (ω + ε2)E.

    The characteristic equation of Eq. (A7) is |sE+U^|. The stability of the second moment of the system (1) is equivalent to the fact that all roots of the characteristic equation |sE+U^| are distributed in the right-half complex plane, that is, the eigenvalues of matrix U^ are all distributed in the right-half complex plane.

    Theorem A2 The second moment of particles displacements of the asymmetric nearest neighbor coupled system (1) is stable if and only if σ2<σsta22=ω2+12λω.

    Proof Because matrix U^ has the same form as matrix U∼ in Eq. (A2). It can be derived by analogy that the eigenvalues of U^ are distributed in the right half complex plane if and only if σ2<14(rsta22+λrsta2). Where rsta2 is the minimum eigenvalue of the matrix Γ. The sum of each row in Γ is 2ω, that is, 2ω is a factor of the determinant of Γ. Thus, 2ω is an eigenvalue of Γ. From Gerschgorin circle theorem, 2ω is the smallest eigenvalue. Thus, rsta2 = 2ω, and 14(rsta22+λrsta2)=ω2+12λω. The stability condition of the second moment of the system (1) becomes σ2<ω2+12λω.

Appendix B: Appendix for Subsection 3.1.2
  • To solve the synchronization condition of the system (1), take the average of both sides of Eq. (8), we have

    A new variable 〈ξδi〉 appeared in Eq. (4). Multiplying (4) by ξ and then averaging, we get the following N–1 linear differential equations

    Equations (B1) and (B2) can be written in the following matrix form:

    where, δ∼=(〈δ∼1〉,…,〈δ∼N−1〉,〈ξδ∼1〉,…,〈ξδ∼N−1〉),

    where

    δ∼=0 is obviously a solution of Eq. (B3) under zero initial value conditions. This means that the system (1) is synchronous. When the equation system (B3) is stable, the initial value does not affect the motion state of the particles after a period of time. So when equation (B3) is stable, the system (1) is statistics synchronization.

    Theorem B1 The system (1) is statistics synchronization if and only if σ2<σsyn2=rsyn2+λrsyn, where

    Proof The statistical synchronization of the system (1) is equivalent to the fact that all eigenvalues of the matrix V∼ are distributed in the right-half complex plane. And V∼ has the same form as matrix U∼ in Eq. (A2). It can be deduced by analogy that the eigenvalues of matrix V∼ are distributed in the right-half complex plane if and only if σ2<rsyn2+λrsyn. rsyn is the minimum eigenvalue of the matrix V. According to Ref. [55], we obtain

Figure (13)  Reference (55)

Catalog

    /

    DownLoad:  Full-Size Img  PowerPoint
    Return
    Return