Switching Diffusion Systems with Past-Dependent Switching and Countable State Space: Successful Couplings and Strong Ergodicity

Fubao Xi, Yafei Zhai, Chao Zhu · 2025

Read Switching Diffusion Systems with Past-Dependent Switching and Countable State Space: Successful Couplings and Strong Ergodicity by Fubao Xi, Yafei Zhai, Chao Zhu (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-2508-04997.txt.

Introduction

Fubao Xi, Yafei Zhai, and Chao Zhu wrote a 2025 math paper on systems that switch between states while they evolve. In their systems, the switch rates depend on the past path of the moving part, not just its present state. The set of states is countable, with no fixed top count. Such past dependence fits biology, ecology, and finance, but it makes the math much harder.

The hard part is joining two copies of the system so they meet and stay together. With past dependent switching, a first meeting is not enough, since the pair can split apart again. The authors build a joining where the pair stay the same for a long stretch after meeting. Under stated conditions, they prove strong stability for the system and show it in a mean field model with many bodies.

Catalog record

CREATORFubao Xi, Yafei Zhai, Chao Zhu
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE82 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2508-04997

Browse the full library · About the Tower of Babel

Full text

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

Switching Diffusion Systems with Past-Dependent Switching and Countable State Space: Successful Couplings and Strong Ergodicity Fubao Xi, Yafei Zhai, Chao Zhu Abstract This work studies a class of switching diffusion systems where the switching component takes values in a countable state space and its transition rates depend on the history of the continuous component. Under suitable conditions, we construct a successful coupling that establishes stability of the underlying process in the total variation norm. The coupling approach also enables us to derive strong ergodicity for the underlying process. Finally, we illustrate the main results with an N N -body mean-field model featuring past-dependent switching and a countable state space. Key Words and Phrases. Switching diffusion system, past-dependent switching, Feller property, successful coupling, stability, total variation norm, strong ergodicity. 2000 MR Subject Classification. 60J25, 60J27, 60J60, 60J75. 1 Introduction Regime-switching stochastic differential equations have become a powerful framework for modeling and analyzing systems subject to random environmental changes, with applications in areas such as biological systems ( Bao and Shao, 2016 , Greenhalgh et al., 2016 , Cai et al., 2021 , Li et al., 2017 ), control systems ( Fleming and Soner, 2006 , Wen et al., 2023 , Savku, 2024 , Song et al., 2011 ), neural networks ( Whitt, 2002 , Huang et al., 2016 ), mathematical finance and risk management ( Zhang, 2001 , Zhou and Yin, 2003 , Elliott and Siu, 2010 ), and population dynamics ( Kuang, 1993 , Luo and Mao, 2009 , Zhu and Yin, 2009 , Settati and Lahrouz, 2014 ). These systems are characterized by abrupt transitions between different regimes, each governed by distinct dynamical laws. Due to their theoretical and practical importance, such systems have attracted growing attention from researchers and practitioners in recent decades. We refer to Mao and Yuan, 2006 and Yin and Zhu, 2010 for comprehensive investigations of such processes as well as their applications in areas such as mathematical biology, finance, and risk management; with the former deals with regime-switching diffusions with Markovian switching and the latter focuses on state-dependent regime-switching diffusions. The recent decades have witnessed extensive developments of the theoretical framework of regime-switching diffusions. To name just a few, Xi and Yin, 2010 established asymptotic properties for nonlinear autoregressive Markov processes with state-dependent switching, while Xi and Yin, 2011 subsequently examined a class of jump-diffusions with state-dependent switching, Zhu et al., 2015 derived Feynman-Kac formulas for regime-switching jump diffusions, and Xi and Yin, 2015 obtained the exponential ergodicity for stochastic Liénard equations with state-dependent switching. Further developments include the work of Shao, 2018 , who provided sufficient conditions for the existence and uniqueness of invariant measures for state-dependent regime-switching diffusion processes. Nguyen and Yin, 2020 investigated the almost sure and L^{p} stability of stochastic functional differential equations with regime switching, while Nguyen et al., 2021 focused on their ergodicity and stability properties. Tran et al., 2022 analyzed stability in distribution of Markovian switching jump diffusions, and Cao et al., 2024 studied the weak convergence of stochastic functional diffusion systems with singularly perturbed regime switching. Several works have also addressed control methodologies. For instance, Li et al., 2020 proposed delay feedback control strategies based on discrete-time observations of both the system and Markovian states, and Li et al., 2022 achieved stabilization in distribution for hybrid stochastic differential equations using feedback control from discrete-time state observations. Applications to biological systems have produced valuable insights: Chen et al., 2023 examined extinction and permanence in a stochastic tumor-immune system with Markovian switching, while Hieu et al., 2024 analyzed a stochastic SIS epidemic model with vaccination in a randomly switching environment. Methodological advancements include Li et al., 2023 on numerical solutions for hybrid systems with infinite delay, and Hu et al., 2019 on stability in distribution for numerical solutions of neutral stochastic functional differential equations with Markovian switching. Recent efforts have also focused on regime-switching (jump) diffusions where the switching component takes values in a countable state space. Shao, 2015 studied the existence of strong solutions and the Feller property for such systems. Xi and Zhu, 2017 established Feller and strong Feller properties and proved exponential ergodicity for regime-switching jump diffusions with countable regimes. Xi et al., 2021 explored stochastic damped Hamiltonian systems with state-dependent switching in a countably infinite regime space. In addition, Shao, 2017 investigated stabilization of regime-switching diffusion processes using feedback control based on discrete-time observations of both the state and switching processes, and Shao and Xi, 2019 extended these results to state-dependent regime-switching processes. An important extension of regime-switching diffusions is to consider past-dependent switching, where the transition rates depend on the past trajectory of the process. This line of research was initiated in Nguyen and Yin, 2016 , which established existence and uniqueness results for a class of stochastic differential equations with past-dependent switching, and investigated their Markov and Feller properties. More recently, Nguyen et al., 2024 examined the stability of stochastic functional differential equations with past-dependent random switching in a countably infinite state space. The past-dependent switching allows for more flexible modeling of systems where the transition rates are influenced by historical states, making it suitable for applications in ecology, biology, and finance. For instance, in ecological systems, the transition rates of species populations may depend on their historical states, such as past population densities or environmental conditions. In financial markets, the transition rates of the general market trend (bull or bear) may depend on the historical prices and/or the volatility patterns of the stocks. The past-dependent switching framework provides a more realistic representation of such systems, capturing the influence of historical states on current dynamics. Building upon the work of Nguyen and Yin, 2016 , we consider the asymptotic properties for regime-switching diffusion process (X(t),\Lambda(t)) with past-dependent switching, in which the switching component \Lambda(t) has a countable state space. Since the switching rates of \Lambda(t) depends on the past states of X , the process (X(t),\Lambda(t)) itself is not Markovian. Consequently, much of our analyses are focused on the Markovian segment process (X_{t},\Lambda(t)) . In particular, we wish to establish strong ergodicity for (X_{t},\Lambda(t)) . Due to the complexity of the past-dependent switching, the analysis of such systems is significantly more challenging than that of standard regime-switching diffusions. The main difficulty arises from the fact that the past-dependent switching introduces inherently nonlinear memory effects, which invalidate the conventional approaches for (jump) diffusions with or without switching. Furthermore, the interaction between regime-switching and past-dependent dynamics complicates stability analysis, making it a non-trivial task to establish strong ergodicity. In this paper, we aim to address these challenges by developing a novel coupling methodology tailored for past-dependent switching diffusion processes. Our approach is inspired by the coupling methods developed in the study of (jump) diffusion processes such as Chen, 2004 , Chen and Li, 1989 , Lindvall and Rogers, 1986 , Priola and Wang, 2006 , and Wang, 2010 . We construct a successful coupling for the past-dependent regime-switching diffusion processes, which allows us to derive strong ergodicity. This extends previous work by Xi, 2013 on Markovian switching jump-diffusion processes and Xi and Shao, 2013 on multidimensional diffusions with state-dependent switching. Specifically, we first introduce a coupling operator \widehat{\mathscr{A}} in ( 4.9 ), which is

Read the complete text