Qualitatively distinct mechanisms of noise-induced escape in diffusively coupled bistable elements

Hidemasa Ishii, Hiroshi Kori · 2025

Read Qualitatively distinct mechanisms of noise-induced escape in diffusively coupled bistable elements by Hidemasa Ishii, Hiroshi Kori (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2512-01388.txt.

Introduction

Hidemasa Ishii and Hiroshi Kori study how noise can push linked units out of a stable state. Each unit has two stable states. Units are linked so each feels its neighbors. Noise, link strength, and the bend of each unit all play a part.

The pair show that mix yields three clear escape routes. Which route fires rests on how strong the links are. Weak, mid, and strong links each give a distinct path out. Knowing the routes helps predict flips in power grids, nerve cells, and other linked bistable sets. The three routes give modelers a map for when a whole set flips at once and when it escapes in parts instead.

Catalog record

CREATORHidemasa Ishii, Hiroshi Kori
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE24 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2512-01388

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Qualitatively distinct mechanisms of noise-induced escape in diffusively coupled bistable elements Hidemasa Ishii, Hiroshi Kori Abstract The analysis of noise-induced escape in ensembles of bistable elements is challenging, because nonlinearity, coupling, and noise all play essential roles. We show that the interplay of these three factors yields three qualitatively distinct escape mechanisms in diffusively coupled bistable elements, depending on the coupling strength. To clarify the relation between coupling strength and mean escape time, we derive effective one-dimensional dynamics: nonlinear mean-field Fokker–Planck equation in the weak-coupling regime, stochastic mean-field dynamics in the strong-coupling regime, and deterministic mean-field dynamics in the intermediate regime. We validate these reduced descriptions by comparing predictions with numerical simulations. We identify a distinct dominant driving factor of escape processes in each regime. Notably, the three escape mechanisms emerge through the interplay of nonlinearity, diffusive coupling, and dynamical noise — rather than bifurcations of the noise-free system. Our approach serves as a framework applicable to other diffusively coupled stochastic nonlinear systems, motivating a further search for similar synergistic phenomena. \usephysicsmodule ab, ab.legacy, op.legacy I Introduction External forcings on bistable systems induce switching between the two typical states 18 ; 6 . This has motivated the use of bistable models in practical research on abrupt changes in systems’s states, from epilepsy 25 ; 13 and climate change 27 to social uprisings 11 . When external forcings are stochastic, the switching behavior is known as noise-induced escape 3 . This phenomenon is theoretically interesting because it necessitates both bistability, i.e. nonlinearity, and noise. Pioneered by Kramers in 1940 21 , the literature on noise-induced escape has extended its scope to ensembles of bistable elements, where the escape of one element affects the dynamics of others through coupling. Previous studies 17 ; 9 ; 8 ; 10 ; 3 ; 14 ; 4 ; 5 have analyzed the influence of changes in the coupling strength on escape dynamics, which are often characterized by the mean escape time: i.e. the expected time until the system escapes. Coupling is often assumed to be diffusive, facilitating synchronization among elements. Despite the accumulated literature on noise-induced escape, we identify a gap in our knowledge of the effect of varying coupling strength. Systems of diffusively coupled bistable elements exhibit bifurcations regarding the coupling strength. Accordingly, many existing studies have distinguished parameter regimes based on bifurcations: the weak-coupling regime, where all the fixed points of the uncoupled system exist; the intermediate “slow-domino” regime, where fixed points disappear through saddle-node bifurcations; and the synchronized or “fast-domino” regime, which is beyond the largest bifurcation point 9 ; 3 . There are two limitations in this bifurcation-based approach. Firstly, it cannot explain the effect of varying coupling strength in the last regime. In our previous study, we numerically showed that the mean escape time first declines, then increases, and then reaches an asymptotic value as the coupling strength is increased 20 . Since this non-monotonic behavior is observed in the fast-domino regime where no bifurcation occurs, one must consider other factors to explain it. Secondly, it is often unfeasible to analyze bifurcations of large or networked systems. Indeed, most previous works have either considered small systems with a few components 3 ; 14 ; 12 ; 5 or conducted numerical and data-driven analyses of large and networked systems 7 ; 13 ; 22 ; 20 . In light of these limitations, more coarse-grained theoretical approaches that do not require detailed knowledge of bifurcations would complement the literature on noise-induced escape, advancing our understanding of the effect of varying coupling strength in a broader parameter regime. In the remainder, we first introduce the model and define the mean escape time. Then, we derive effective one-dimensional dynamics that approximate collective escape processes in three parameter regimes, which is our main results. Each of the reduced descriptions represents a distinct escape mechanism. Our theoretical results are verified by direct numerical simulations of the original model. We conclude this letter by highlighting the synergistic role of nonlinearity, diffusive coupling, and dynamical noise in the collective escape. II Model and mean escape time Figure 1: Typical trajectories in different regimes. Light gray lines depict the evolution of all the elements. Bright red lines show trajectories of the mean field. (a) Elements are non-synchronous in the weak-coupling regime. (b) In the intermediate regime, elements’ states evolve around the mean field while maintaining non-negligible variance. (c) The ensemble behaves as a single unit under strong diffusive coupling. We analyze the model governed by the following stochastic differential equations (SDEs): \displaystyle\dot{x}_{i}=f(x_{i})+\frac{K}{N}\sum_{j=1}^{N}\pab{x_{j}-x_{i}}+\sqrt{2D}\xi_{i}, (1) \displaystyle f(x)\coloneq-x\pab{x-r}\pab{x-1}, (2) where K is the coupling strength, N is the system size, D is the noise strength, and \xi_{i} denotes pairwise independent white Gaussian noise satisfying \aab{\xi_{i}(t)}=0 and \aab{\xi_{i}(t)\xi_{j}(t+\tau)}=\delta_{ij}\delta\pab{\tau} . The local flow f(x) induces bistability of each element in the absence of coupling and noise. The system \dot{x}=f(x) has stable fixed points at x=0 and 1 and an unstable fixed point at x=r . Throughout this research, we assume 0<r\ll 0.5 , in which case x=1 is much more stable than x=0 . In the following, we refer to x=0 and 1 as the “background” and “active” states, respectively. The strong asymmetry in bistability allows us to focus on the escape from the background to active states, ignoring the probability of the opposite transition. In this letter, the diffusive coupling among elements is assumed to be global: i.e. every element interacts with all the others. We generalize our results to networked systems in another article. As illustrated in Fig. 1 , stronger couplings lead to higher degrees of synchronization among elements. In the noise-free case ( D=0 ), the model ( 1 ) exhibits bifurcations regarding K 9 ; 3 . Due to the diffusive nature of coupling, there always exist two stable and one unstable uniform steady states: i.e. \bm{x}_{0}=(0,\dots,0)^{\top} , \bm{x}_{1}=(1,\dots,1)^{\top} , and \bm{x}_{r}=(r,\dots,r)^{\top} , respectively. In the uncoupled limit of K=0 , the system has 3^{N} fixed points. As K increases, non-uniform fixed points disappear through saddle-node bifurcations, leaving only the three uniform steady states. At the largest bifurcation point K_{\mathrm{b}} , a pitchfork bifurcation occurs, where two non-uniform saddles collide the uniform unstable node \bm{x}_{r} , which then becomes a saddle. The linear stability analysis of \bm{x}_{r} reveals K_{\mathrm{b}}=r(1-r) , which is independent of N 1 . We define the mean escape time as follows. First, the system is initialized to the collective background state \bm{x}_{0} . Then, the first escape time of element i is defined as \displaystyle\tau_{i}\coloneq\inf_{t}\Bab{t>0\text{ s.t. }x_{i}(t)\geq\xi\text{ given }x_{i}(0)=0}, (3) where \xi is a fixed threshold between the background and active states. We use \xi=0.5 in the following, but the choice is arbitrary as long as \xi is not close to r or 1 . As \tau_{i} is defined for each element, we summarize them by taking the ensemble average within a system, and define its expectation to be the mean escape time \bar{\tau} : \displaystyle\bar{\tau}\coloneq\aab{\frac{1}{N}\sum_{i=1}^{N}\tau_{i}}. (4) For a one-dimensional system with an associated potential V(x) , the mean first passage time formula ( 19 , Section 5.5) allows us to calculate the mean escape time. To be precise, let us consider a stochastic process governed by \dot{x}=-V^{\prime}(x)+\sqrt{2D}\xi defined in the interval (-\infty,\xi) between a

Read the complete text