Stability of hybrid dynamic systems containing singularly perturbed random processes

Finite time stability and stabilization of hybrid dynamic systems

The analysis of incremental stability typically involves measuring the distance between any two solutions of a given dynamical system at the same time instant, which is problematic when studying hybrid dynamical systems. Indeed, hybrid systems generate solutions defined with respect to hybrid time instances (that consists of both the continuous time elapsed and the discrete time, which is the number of jumps experienced so far), and two solutions of the same hybrid system may not be defined at the same hybrid time instant. To overcome this issue, we present novel definitions of incremental stability for hybrid systems based on graphical closeness of solutions. As we will show, defining incremental asymptotic stability with respect to the hybrid time yields a restrictive notion, such that we also investigate incremental asymptotic stability notions with respect to the continuous time only or the discrete time only, respectively. In this manner, two (effectively dual) incremental stability notions are attained, called jump- and flow incremental asymptotic stability. To present Lyapunov conditions for these two notions, in both cases, we resort to an extended hybrid system and we prove that the stability of a well-defined set for this extended system implies incremental stability of the original system. We can then use available Lyapunov conditions to infer the set stability of the extended system. Various examples are provided throughout this paper, including an event-triggered control application and a bouncing ball system with Zeno behavior, that illustrate incremental stability with respect to continuous time or discrete time, respectively.

Polynomial chaos is a powerful technique for propagating uncertainty through ordinary and partial differential equations. Random variables are expanded in terms of orthogonal polynomials and differential equations are derived for the coefficients. Here we study the structure and dynamics of these differential equations when the original system has Hamiltonian structure or displays chaotic dynamics. In particular, we prove that the differential equations for the expansion coefficients in generalized polynomial chaos expansions of Hamiltonian systems retain the Hamiltonian structure relative to the ensemble average Hamiltonian. Additionally, using the forced Duffing oscillator as an example, we demonstrate that when the original dynamical system displays chaotic dynamics, the resulting dynamical system from polynomial chaos also displays chaotic dynamics, limiting its applicability.

Probabilistic ordinary differential equation (ODE) solvers have been introduced over the past decade as uncertainty-aware numerical integrators. They typically proceed by assuming a functional prior to the ODE solution, which is then queried on a grid to form a posterior distribution over the ODE solution. As the queries span the integration interval, the approximate posterior solution then converges to the true deterministic one. Gaussian ODE filters, in particular, have enjoyed a lot of attention due to their computational efficiency, the simplicity of their implementation, and their provable fast convergence rates. In this article, we extend the methodology to stochastic differential equations (SDEs) and propose a probabilistic simulator for SDEs. Our approach involves transforming the SDE into a sequence of random ODEs using piecewise differentiable approximations of the Brownian motion. We then apply probabilistic ODE solvers to the individual ODEs, resulting in a pathwise probabilistic solution to the SDE. We establish worst-case strong 1.5 local and 1.0 global convergence orders for a specific instance of our method. We further show how we can marginalise the Brownian approximations, by incorporating its coefficients as part of the prior ODE model, allowing for computing exact transition densities under our model. Finally, we numerically validate the theoretical findings, showcasing reasonable weak convergence properties in the marginalised version.

Optimal Control of Stochastic Systems

First, we prove a necessary and sufficient condition for global in time existence of all solutions of an ordinary differential equation (ODE). It is a condition of one‐sided estimate type that is formulated in terms of so‐called proper functions on extended phase space. A generalization of this idea to stochastic differential equations (SDE) and parabolic equations (PE) allows us to prove similar necessary and sufficient conditions for global in time existence of solutions of special sorts: L1‐complete solutions of SDE (this means that they belong to a certain functional space of L1 type) and the so‐called complete Feller evolution families giving solutions of PE. The general case of equations on noncompact smooth manifolds is under consideration.

A system of recurrent neural networks for modularising, parameterising and dynamic analysis of cell signalling networks

Extensive research has been conducted on models of ordinary differential equations (ODEs), yet these deterministic models often fail to capture the intricate complexities of real-world systems adequately. Thus, many studies have proposed the integration of Markov chains into nonlinear dynamical systems to account for perturbations arising from environmental changes and random variations. Notably, the field of parameter estimation for ODEs incorporating Markov chains still needs to be explored, creating a significant research gap. Therefore, the objective of this study is to investigate a comprehensive model capable of encompassing real-life scenarios. This model combines a system of ODEs with a continuous-time Markov chain, enabling the representation of a continuous system with discrete parameter switching. We present a machine discovery framework for parameter estimation in nonlinear dynamical systems with Markovian switching, effectively addressing this research gap. By incorporating Markov chains into the model, we adeptly capture the time-varying dynamics of real-life systems influenced by environmental factors. This approach enhances the applicability and realism of the research, enabling more precise representations of dynamical systems with Markovian switching in complex scenarios.

Hybrid dynamical systems are those with interaction between continuous and discrete dynamics. For the analysis and control of such systems concepts and theories from either the continuous or the discrete domain are typically readapted. In this thesis the ideas from perturbation theory are readapted for approximating a hybrid system using a continuous one. To this purpose, hybrid systems that possess a two-time scale property, i.e. discrete states evolving in a fast time-scale and continuous states in a slow time-scale, are considered. Then, as in singular perturbation or averaging methods, the system is approximated by a slow continuous time system. Since the hybrid nature of the process is removed by averaging, such a procedure is referred to as dehybridization in this thesis. It is seen that fast transitions required for dehybridization correspond to fast switching in all but one of the discrete states (modes). Here, the notion of dominant mode is defined and the maximum time interval spent in the non-dominant modes is considered as the 'small' parameter which determines the quality of approximation. It is shown that in a finite time interval, the solutions of the hybrid model and the continuous averaged one stay 'close' such that the error between them goes to zero as the 'small' parameter goes to zero. To utilize the ideas of dehybridization for control purposes, a cascade control design scheme is proposed, where the inner-loop artificially creates the two-time scale behavior, while the outer-loop exponentially stabilizes the approximate continuous system. It is shown that if the origin is a common equilibrium point for all modes, then for sufficiently small values of the 'small' parameter, exponential stability of the hybrid model can be guaranteed. However, it is shown that if the origin is not an equilibrium point for some modes, then the trajectories of the hybrid model are ultimately bounded, the bound being a function of the 'small' parameter. The analysis approach used here defines the hybrid system as a perturbation of the averaged one and works along the lines of robust stability. The key technical diffierence is that though the norm of the perturbation is not small, the norm of its time integral is small. This thesis was motivated by the stick-slip drive, a friction-based micropositioning setup, which operates in two distinct modes 'stick' and 'slip'. It consists of two masses which stick together when the interfacial force is less than the Coulomb frictional force, and slips otherwise. The proposed methodology is illustrated through simulation and experimental results on the stick-slip drive.

Integrating the heterogeneity of contact structures and stochastic factors, we establish a continuous-time Markov SIRS epidemic model with vaccination on networks. Based on this framework, we derive both ordinary differential equations and stochastic differential equations, serving as first- and second-order approximations for the Markov chain, respectively. Additionally, we investigate the quasi-stationary distribution and the time to extinction to characterize the long-term dynamical behavior of the Markov chain. The results reveal that the time to extinction from the quasi-stationary distribution follows an exponential distribution. By using the diffusion approximation, we further obtain approximations for both the quasi-stationary distribution and the expected time to extinction, which are given by a multivariate normal distribution and an explicit formula, respectively. Finally, numerical simulations illustrate that vaccination can accelerate disease extinction and that a greater proportion of individuals with higher degrees leads to a longer time to extinction.

Event Abstract Back to Event Stochastic modeling of neuronal signaling Jukka Intosalmi1*, Tiina Manninen1, Keijo Ruohonen1 and Marja-Leena Linne1 1 Tampere University of Technology, Finland The time evolution of biochemical systems in neurons is traditionally modeled using deterministic ordinary differential equations (ODEs). Chemical reactions, however, are random in nature, and the deterministic approach is valid only for a restricted class of systems. Stochastic models take random fluctuations into account and are thus more realistic. Biochemical reactions can be modeled stochastically using numerous different methods. An ideal model would have the following three important properties. First, the model should be as realistic as possible, second, the mathematical method should be easily implementable as a computer algorithm, and third, the algorithm should be computationally effective. Naturally, all these three conditions cannot be fulfilled at the same time. Some realistic modeling approaches can be derived directly from chemical kinetics without making any approximations. Such approaches are called exact. A good example of an exact modeling approach is the stochastic simulation algorithm (SSA) developed by Gillespie. The SSA is applicable when the molecular populations in the system are small, but it becomes computationally inefficient when the numbers of molecules increase. In order to construct stochastic models that can be effectively simulated, new mathematical approaches have to be explored. As an approximate method also stochastic differential equations (SDEs) have been considered a promising way to model biochemical reactions stochastically. The SDE approach is attractive especially if we consider a system for which the SSA is computationally inefficient and the traditional deterministic ODE approach cannot be used as a good approximation. SDE models treat the chemical populations as real numbers and the construction of the model is based on the law of mass action, similarly as the construction of the traditional deterministic modeling approach. The SDE model, however, takes the random fluctuations into account by describing the time evolution of the system with the stochastic Itô process instead of the deterministic set of ODEs. The Itô process is basically a set of coupled stochastic differential equations. Although SDE modeling usually leads to equations that cannot be solved analytically, the solutions can be approximated numerically using different numerical integration methods. Thus, SDEs provide a mathematically rigorous way to construct models that can be simulated effectively. In order to investigate how well the SDE approach models biochemical systems, the results have to be compared with experimental data or with the simulation results from some exact simulation procedure. In this study, we use the methodology of spectral analysis to compare the simulation results of the SDE model and the SSA. As case studies, we consider e.g. calcium binding and protein kinase C (PKC) signal transduction pathway. We investigate the nature of noise in different modeling approaches and try to interpret the meaning of the noise in real biological systems. The main goal of our study is to find the most essential components and parameters in the SDE model and to adjust them so that the model is capable of giving similar results as the SSA. The simulations are carried out using the distributed computing resources (GRID) provided by Techila Technologies Ltd. Conference: Neuroinformatics 2008, Stockholm, Sweden, 7 Sep - 9 Sep, 2008. Presentation Type: Poster Presentation Topic: Computational Neuroscience Citation: Intosalmi J, Manninen T, Ruohonen K and Linne M (2008). Stochastic modeling of neuronal signaling. Front. Neuroinform. Conference Abstract: Neuroinformatics 2008. doi: 10.3389/conf.neuro.11.2008.01.023 Copyright: The abstracts in this collection have not been subject to any Frontiers peer review or checks, and are not endorsed by Frontiers. They are made available through the Frontiers publishing platform as a service to conference organizers and presenters. The copyright in the individual abstracts is owned by the author of each abstract or his/her employer unless otherwise stated. Each abstract, as well as the collection of abstracts, are published under a Creative Commons CC-BY 4.0 (attribution) licence (https://creativecommons.org/licenses/by/4.0/) and may thus be reproduced, translated, adapted and be the subject of derivative works provided the authors and Frontiers are attributed. For Frontiers’ terms and conditions please see https://www.frontiersin.org/legal/terms-and-conditions. Received: 28 Jul 2008; Published Online: 28 Jul 2008. * Correspondence: Jukka Intosalmi, Tampere University of Technology, Tampere, Finland, jukka.intosalmi@tut.fi Login Required This action requires you to be registered with Frontiers and logged in. To register or login click here. Abstract Info Abstract The Authors in Frontiers Jukka Intosalmi Tiina Manninen Keijo Ruohonen Marja-Leena Linne Google Jukka Intosalmi Tiina Manninen Keijo Ruohonen Marja-Leena Linne Google Scholar Jukka Intosalmi Tiina Manninen Keijo Ruohonen Marja-Leena Linne PubMed Jukka Intosalmi Tiina Manninen Keijo Ruohonen Marja-Leena Linne Related Article in Frontiers Google Scholar PubMed Abstract Close Back to top Javascript is disabled. Please enable Javascript in your browser settings in order to see all the content on this page.

Stochastic physics-informed neural ordinary differential equations

A Study on Uncertainty of Discharge in River Channel Using Stochastic Partial Differential Equation

The ability to conduct fast and reliable simulations of dynamic systems is of special interest to many fields of operations. Such simulations can be very complex and, to be thorough, involve millions of variables, making it prohibitive in CPU time to run repeatedly for many different configurations. Reduced-Order Modeling (ROM) provides a concrete way to handle such complex simulations using a realistic amount of resources. However, when the original dynamical system is very large, the resulting reduced-order model, although much “thinner”, is still as tall as the original system, i.e., it has the same number of equations. In some extreme cases, the number of equations is prohibitive and cannot be loaded in memory. In this work, we combine traditional interval constraint solving techniques with a strategy to reduce the number of equations to consider. We describe our approach and report preliminary promising results.

The understanding and modeling of complex physical phenomena through dynamical systems has historically driven scientific progress, providing essential tools for predicting system behavior under diverse conditions over time. In engineering, the discovery of dynamical systems is indispensable for computational modeling, diagnostics, prognostics, and control of engineered systems. Joining recent efforts that harness the power of symbolic regression in this domain, we propose a novel framework for the end-to-end discovery of ordinary differential equations (ODEs), termed Grammar-based ODE Discovery Engine (GODE). The proposed methodology combines formal grammars with dimensionality reduction and stochastic search for efficiently navigating high-dimensional combinatorial spaces. Grammars serve to inject domain knowledge and provide structure, both constraining and guiding the search for candidate expressions. GODE proves to be more sample- and parameter-efficient than state-of-the-art transformer-based models and to discover more accurate and parsimonious ODE expressions than both genetic programming- and other grammar-based methods, particularly for complex inference tasks, such as the discovery of structural dynamics. Thus, we introduce a tool that could play a catalytic role in dynamics discovery tasks, including modeling, system identification, and monitoring applications. • Proposal of a generative grammar-based method for end-to-end discovery of ODEs. • Efficient generation of syntactically valid libraries with grammars. • Embedding of discrete equations into continuous latent spaces for efficient search. • Grammars allow to incorporate domain biases into the discovery process. • Accurate and parsimonious predictions of noisy ODEs for simulated dynamical systems.