DOCSLIB.ORG

Kick Synchronization Versus Diffusive Synchronization

Total Page：16

File Type：pdf, Size：1020Kb

Download full-text PDF Read full-text
Kick Synchronization Versus Diffusive Synchronization 51st IEEE Conference on Decision and Control December 10-13, 2012. Maui, Hawaii, USA Kick synchronization versus diffusive synchronization Alexandre Mauroy, Pierre Sacré, and Rodolphe Sepulchre Abstract— The paper provides an introductory discussion analysis [3], dissipativity analysis [4], [29], [65], [70], [71], about two fundamental models of oscillator synchronization: and—to a growing extent—contraction analysis [59], [60], the (continuous-time) diffusive model, that dominates the math- [64], [68], [69], [80], provide natural system theoretic tools ematical literature on synchronization, and the (hybrid) kick model, that accounts for most popular examples of synchroniza- to study synchronization. The literature of synchronization tion, but for which only few theoretical results exist. The paper (closely related to consensus theory and coordination theory) stresses fundamental differences between the two models, such is growing and the topic has attracted many systems and as the different contraction measures underlying the analysis, control researchers in the recent years. as well as important analogies that can be drawn in the limit The kick model is largely dominant in natural manifes- of weak coupling. tations of synchronization. Popular examples include syn- I. INTRODUCTION chronization of metronomes [84], clocks [5], [33], heart beats [61], flashing fireflies [11], neurons [23], earth- Synchronization is a pervasive concept in science and quakes [56], and in fact most if not all spiking oscillators. engineering. Currently, it is perhaps the most widely studied In addition, kick synchronization is a source of inspira- dynamical concept across systems biology [24], [26], [83], tion for engineering applications (e.g. synchronization in neuroscience [32], [35], chemistry [39], physics [34], [62], wireless sensor networks [31], unsupervised classification astronomy [6], and engineering [53], [73]. Because synchro- problems [63]). Despite the widespread occurrence of the nization involves interconnection at its core, the relevance of phenomenon, the mathematical literature on kick synchro- systems theory to model, understand, and control synchro- nization is rather sparse compared to the literature on diffu- nization is obvious and was recognized early, e.g. [52]. sive synchronization. Two fundamental mathematical models of synchronization Primarily motivated by the recent thesis [44], the present have emerged across the literature: the diffusive model paper aims at comparing and contrasting the diffusive model and the kick model (a nickname throughout the paper for and the kick model for the synchronization of periodic pulse-coupled synchronization model). The diffusive model oscillators. We stress both the differences and the analogies analyzes synchronization as the result of diffusive coupling: between the two models, with a particular emphasis on the interconnection has the input–output interpretation of their global stability properties. The discussion is tutorial in a static diffusive passive map. Owing to the fundamental nature and focuses on simple examples, such as the coupling homogenization nature of diffusion, diffusive interconnec- of van der Pol oscillators, which provides an insightful tions tend to reduce differences between the time-course illustration of diffusive synchronization in the weakly non- of interconnected variables, thereby favoring synchronized linear oscillation regime and of kick synchronization in the behavior if they are strong enough. In contrast, the kick relaxation oscillation regime. The diffusive model is studied model analyzes synchronization as the result of mutual in continuous-time models while the kick model, hybrid rhythmic locking by short and weak pulses, akin to the in nature, is typically studied in discrete time. Ultimately, physical phenomenon of resonance. The impulsive nature of synchronization is always proven by showing that a certain the coupling combined with the continuous-time flow of the distance between trajectories contracts over time, but the model between the pulses results in a hybrid model, see [54] contraction measure is distinctively different in diffusive and for a rigorous description of the kick model as a hybrid kick models. model. While the diffusive and kick models of oscillator syn- The diffusive model is largely dominant in the mathemat- chronization are fundamentally different, they also exhibit ical literature of synchronization. Synchronization between a remarkable analogy in the limit of weak coupling. This trajectories of state-space models is analyzed as an incre- is because arbitrary oscillator models all reduce to one- mental stability property [43] (i.e. the trajectories converge dimensional phase models when the interconnection is suffi- to one another rather than being attracted toward some ciently weak to maintain system trajectories in the neigh- equilibrium position). The leading concepts of Lyapunov borhood of the limit cycle oscillations of the uncoupled oscillators. Rooted in the seminal contributions of Win- A. Mauroy is with the Department of Mechanical Engineering, Uni- versity of California Santa Barbara, Santa Barbara, CA 93106, USA. free [82] and Kuramoto [38], phase models of interconnected [email protected] oscillators have a universal structure entirely characterized P. Sacré and R. Sepulchre are with the Department of Electrical by their coupling function, which is strongly related to Engineering and Computer Science (Montefiore Institute, B28), Univer- sity of Liège, 4000 Liège, Belgium. [email protected], the phase response curve of the oscillators (i.e. a function [email protected] which corresponds to the phase sensitivity of the uncoupled 978-1-4673-2064-1/12/$31.00978-1-4673-2066-5/12/$31.00 ©2012 IEEE 7171 strong coupling section III section IV Fig. 2. The asymptotic phase map Θ: B(γ) → S1 assigns to each point q in the basin B(γ) a single scalar phase θ on the unit circle S1, such that γ limt→+∞ kΦ(t, q, 0) − Φ(t, p, 0)k2 = 0 where p = x (θ/ω). The set of all points q characterized by the same phase θ is the isochron Iθ. weak coupling section V A. State-space models We consider open dynamical systems described by non- linear time-invariant state-space models diffusive coupling kick coupling x˙ = F(x) + G(x)u, x Rn, u R, (1a) ∈ ∈ R Fig. 1. The paper is organized according to the coupling models. y = H(x), y , (1b) Sections III and IV focus on (possibly strong) diffusive and kick synchro- ∈ nization, respectively. Section V deals with the phase models encountered where the vector fields F and G, and the measurement in the limit of weak coupling. map H support all usual smoothness conditions that are necessary for existence and uniqueness of solutions. We write Φ(t, x0, u) for the solution of the initial value problem (1a) oscillators to an external perturbation). As a result, the with x(0) = x0. fundamental difference between diffusive synchronization An oscillator is an open dynamical system whose zero- and kick synchronization is entirely coded in the shape of input steady-state behavior is periodic rather than constant. the coupling function, a phase map defined on the nonlinear Formally, we assume that the zero-input system x˙ = F(x) unit circle. It is typically harmonic in the weak coupling admits a (locally hyperbolic) stable periodic orbit γ with pe- limit of diffusive synchronization and typically monotone riod T (and angular frequency ω = 2π/T ). Picking an initial γ (and hence discontinuous) in the weak coupling limit of kick condition x0 on the periodic orbit γ, this latter is described γ γ synchronization. Again, the synchronization mechanisms and by the (nonconstant) periodic trajectory Φ(t, x0 , 0) = x (t), the contraction measure are distinctively different even in the such that xγ (t) = xγ (t + T ). The basin of attraction of γ weak coupling limit, despite the shared model structure. is the maximal open set (γ) from which the periodic orbit B The paper structure is illustrated in Figure 1. Section II attracts. reviews state-space models of oscillators and their phase re- Since the periodic orbit γ is homeomorphic to the unit duction. The next sections present global stability results for circle S1, it is naturally parametrized by a single scalar phase. the different synchronization models. Section III focuses on Any point p γ is associated with a phase θ S1, such ∈ ∈ (possibly strong) diffusive synchronization while Section IV that focuses (possibly strong) kick synchronization. Section V p = xγ (θ/ω) deals with the phase models encountered in the limit of weak γ coupling. Section VI provides concluding remarks. (where x0 is by convention associated with the phase θ = 0). For hyperbolic periodic orbits, the notion of phase is extended to any point q in the basin of attraction (γ) through the concept of asymptotic phase. The asymptoticB II. OPEN OSCILLATOR MODELS phase map Θ: (γ) S1 assigns to each point q in the basin (γ) its asymptoticB → phase θ S1, such that This section provides a short introduction to oscillators B ∈ viewed as open dynamical systems, that is, as dynamical lim Φ(t, q, 0) Φ(t, xγ (θ/ω), 0) = 0. (2) t→+∞ k − k2 systems that interact with their environment [67]. We first γ recall basic definitions about stable periodic orbits in n- This mapping is constructed such that the image of x0 is dimensional state-space models (see [21], [28] for details). equal to 0 and such that the progression along any orbit We then introduce (finite and infinitesimal) phase response in (γ) (in absence of perturbation) produces a constant B d curves as fundamental mathematical information required for increase in θ, that is, dt Θ(Φ(t, x0, 0)) = ω. the reduction. We finally show how to reduce n-dimensional An isochron is a level set of the asymptotic phase map Θ, state-space models into one-dimensional phase models de- that is, the set of all points in the basin of attraction of γ pending on the nature of the input. characterized by a same asymptotic phase. 7172 7173 7174 7175 7176 A −1 oscillator 1 oscillator 2 x−1.5 −2 0 5 10 t 15 20 25 B 0 5 10 t 15 20 25 Fig.
Load more

Recommended publications
  • Student Version an Interactive Illustration of the Van Der Pol Oscillator
    STUDENT VERSION AN INTERACTIVE ILLUSTRATION OF THE VAN DER POL OSCILLATOR Mark A. Lau Kwan Electrical Engineering University of Turabo Gurabo PUERTO RICO STATEMENT The purpose of this paper is to present an interactive spreadsheet application that allows the reader to investigate the Van der Pol oscillator. Through a number of suggested activities, the reader will gain a better understanding of the qualitative behavior of solutions as the parameters of the governing differential equation are varied. A Brief Description of the Van der Pol Oscillator The Van der Pol oscillator is described by the nonlinear, second-order ordinary differential equation x¨ + µ(x2 − 1)x _ + !2x = 0; (1) where µ and ! are positive parameters, and x (a function of time t) is the quantity of interest (e.g., position or voltage). The quantitiesx _ andx ¨ denote the first- and second-order derivative with respect to time t, respectively. This celebrated equation (1) is attributed to the Dutch electrical engineer Balthasar Van der Pol, who pioneered experimental investigations in nonlinear dynamics in the early days of radio and telecommunications. He proposed his equation in order to describe the nonlinear oscillations produced by self-sustained oscillating triode circuits [5]. His equation has been extensively studied and finds numerous applications in engineering, physics, biology, sociology, and economics. The Van der Pol equation (1) introduces a nonlinear damping term, namely µ(x2 − 1)x _. Unlike the familiar linearly damped harmonic oscillator, the presence of the nonlinear damping term can 2 Van der Pol Oscillator Figure 1. Main user interface of the Microsoft Excel spreadsheet for investigating the behavior of the Van der Pol oscillator.
    [Show full text]
  • Van Der Pol Meets LSTM
    bioRxiv preprint doi: https://doi.org/10.1101/330548; this version posted May 25, 2018. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. Learning Nonlinear Brain Dynamics: van der Pol Meets LSTM Germán Abrevaya Aleksandr Aravkin Departamento de Física, FCEyN Department of Applied Mathematics Universidad de Buenos Aires University of Washington & IFIBA, CONICET Seattle, WA Buenos Aires, Argentina [email protected] [email protected] Guillermo Cecchi Irina Rish Pablo Polosecki IBM Research IBM Research IBM Research Yorktown Heights, NY Yorktown Heights, NY Yorktown Heights, NY [email protected] [email protected] [email protected] Peng Zheng Silvina Ponce Dawson Department of Applied Mathematics Departamento de Física, FCEyN University of Washington Universidad de Buenos Aires Seattle, WA & IFIBA, CONICET [email protected] Buenos Aires, Argentina [email protected] Abstract Many real-world data sets, especially in biology, are produced by highly multivari- ate and nonlinear complex dynamical systems. In this paper, we focus on brain imaging data, including both calcium imaging and functional MRI data. Standard vector-autoregressive models are limited by their linearity assumptions, while nonlinear general-purpose, large-scale temporal models, such as LSTM networks, typically require large amounts of training data, not always readily available in biological applications; furthermore, such models have limited interpretability. We introduce here a novel approach for learning a nonlinear differential equation model aimed at capturing brain dynamics. Specifically, we propose a variable-projection optimization approach to estimate the parameters of the multivariate (coupled) van der Pol oscillator, and demonstrate that such a model can accurately repre- sent nonlinear dynamics of the brain data.
    [Show full text]
  • Arxiv:Cond-Mat/0212203V1
    A nonlinear dynamical model of human gait Bruce J. West1,2,3 and Nicola Scafetta.1,2 1 Pratt School of EE Department, Duke University, P.O. Box 90291, Durham, NC 27701. 2 Physics Department, Duke University, Durham, NC 27701. and 3 Mathematics Division, Army Research Office, Research Triangle Park, NC 27709. (Dated: September 27, 2018) We present a nonlinear stochastic model of the human gait control system in a variety of gait regimes. The stride interval time series in normal human gait is characterized by slightly multifractal fluctuations. The fractal nature of the fluctuations become more pronounced under both an increase and decrease in the average gait. Moreover, the long-range memory in these fluctuations is lost when the gait is keyed on a metronome. The human locomotion is controlled by a network of neurons capable of producing a correlated syncopated output. The central nervous system is coupled to the motocontrol system, and together they control the locomotion of the gait cycle itself. The metronomic gait is simulated by a forced nonlinear oscillator with a periodic external force associated with the conscious act of walking in a particular way. PACS numbers: 05.45.-a,87.18.Sn, 05.45.Df, 87.17.Nn I. INTRODUCTION neural firing activity. The dynamics of human gate may also be voluntarily forced, for example, by following the Walking is a complex process which we have only re- frequency of a metronome. We model the complex gait cently begun to understand through the application of system by assuming that the amplitude of the impulses of nonlinear data processing techniques [1, 2, 3, 4] to study the correlated firing neural centers regulate only the un- interval data.
    [Show full text]
  • Abdus Salam United Nations Educational, Scientific and Cuftura! XA0053813 Organization International Centre International Atomic Energy Agency for Theoretical Physics
    the abdus salam united nations educational, scientific and cuftura! XA0053813 organization international centre international atomic energy agency for theoretical physics HARMONIC OSCILLATIONS, CHAOS AND SYNCHRONIZATION IN SYSTEMS CONSISTING OF VAN DER POL OSCILLATOR COUPLED TO A LINEAR OSCILLATOR P. Woafo i 31-09 J IC/99/188 United Nations Educational Scientific and Cultural Organization and International Atomic Energy Agency THE ABDUS SALAM INTERNATIONAL CENTRE FOR THEORETICAL PHYSICS HARMONIC OSCILLATIONS, CHAOS AND SYNCHRONIZATION IN SYSTEMS CONSISTING OF VAN DER POL OSCILLATOR COUPLED TO A LINEAR OSCILLATOR P. Woafo1 Laboratoire de Mecanique, Faculte des Sciences, Universite de Yaounde I, B.P. 812, Yaounde, Cameroon and The Abdus Salam International Centre for Theoretical Physics, Trieste, Italy. Abstract This paper deals with the dynamics of a model describing systems consisting of the classical Van der Pol oscillator coupled gyroscopically to a linear oscillator. Both the forced and autonomous cases are considered. Harmonic response is investigated along with its stability boundaries. Condition for quenching phenomena in the autonomous case is derived. Neimark bifurcation is observed and it is found that our model shows period doubling and period-m sudden transitions to chaos. Synchronization of two and more systems in their chaotic regime is presented. MIRAMARE - TRIESTE December 1999 'Regular Associate of the Abdus Salam ICTP. I- Introduction Due to their occurrence in various scientific fields, ranging from biology, chemistry, physics to engineering, coupled nonlinear oscillators have been a subject of particular interest in recent years [1,2,3]. Among these coupled systems, a particular class is that containing self-sustained components such as the classical Van der Pol oscillator.
    [Show full text]
  • Hopf Bifurcation and Stability of Periodic Solutions for Van Der Pol
    Nonlinear Analysis 62 (2005) 141–165 www.elsevier.com/locate/na Hopf bifurcation and stability of periodic solutions for van der Pol equation with time delayଁ WenwuYu, Jinde Cao ∗ Department of Mathematics, Southeast University, Nanjing 210096, China Received 5 January 2005; accepted 9 March 2005 Abstract In this paper, the van der Pol equation with a time delay is considered, where the time delay is regarded as a parameter. It is found that Hopf bifurcation occurs when this delay passes through a sequence of critical value. A formula for determining the direction of the Hopf bifurcation and the stability of bifurcating periodic solutions is given by using the normal form method and center manifold theorem. © 2005 Published by Elsevier Ltd. Keywords: Van der Pol equation; Time delay; Hopf bifurcation; Periodic solutions 1. Introduction The well-known van der Pol equation, which describes the oscillations is the second-order nonlinear damped system governed by x(t)˙ = y(t) − f (x(t)), y(t)˙ =−x(t). (1.1) This model is considered as one of the most intensely studied system in nonlinear system [9,12,23] and has served as a basic model in physics, electronics, biology, neurology and ଁ This work was jointly supported by the National Natural Science Foundation of China under Grant 60373067, the 973 Program of China under Grant 2003CB317004, the Natural Science Foundation of Jiangsu Province, China under Grants BK2003053, Qing-Lan Engineering Project of Jiangsu Province, China. ∗ Corresponding author. Tel.: +86 25 83792315; fax: +86 25 83792316. E-mail address: [email protected] (J. Cao).
    [Show full text]
  • 1 Introduction 2 the Van Der Pol Oscillator
    Van der Pol Oscillator Celestial Mechanics Wesley Cao · 1 Introduction The van der Pol oscillator is a non-conservative oscillator with non-linear damping. Energy is dissipated at high amplitudes and generated at low amplitudes. As a result, there exists oscillations around a state at which energy generation and dissipation balance. The state towards which the oscillations converge is known as a limit cycle, which shall be formalized later in this paper. Balthazar van der Pol was a pioneer in the field of radio and telecommunications. While he was working at Phillips, van der Pol discovered these stable oscillations. Van der Pol himself came across the system as he was building electronic circuit models of the human heart. Due to the unique nature of the van der Pol oscillator, it has become the cornerstone for studying systems with limit cycle oscillations. In fact, the van der Pol equation has become a staple model for oscillatory processes in not only physics, but also biology, sociology and even economics. For instance, it has been used to model the electrical potential across the cell membranes of neurons in the gastric mill circuit of lobsters [5, 9]. Additionally, Fitzhugh and Nagumo used the model to describe spike generation in giant squid axons [4, 7]. The equation has also been extended to the Burridge–Knopoff model which characterizes earthquake faults with viscous friction [3]. Thus, it stands to reason that we should develop a deep understanding of the van der Pol oscillator due to its widespread applications. 2 The Van der Pol Oscillator The van der Pol oscillator is described by the equation x¨ ǫ(1 x2)x ˙ + x = 0 − − and equivalently the autonomous system, x3 x˙ = y F (x) := y ǫ x − − µ 3 − ¶ y˙ = x − It differs from the systems we have studied in that it is non-conservative.
    [Show full text]
  • Deterministic Chaos: Applications in Cardiac Electrophysiology Misha Klassen Western Washington University, [email protected]
    Occam's Razor Volume 6 (2016) Article 7 2016 Deterministic Chaos: Applications in Cardiac Electrophysiology Misha Klassen Western Washington University, [email protected] Follow this and additional works at: https://cedar.wwu.edu/orwwu Part of the Medicine and Health Sciences Commons Recommended Citation Klassen, Misha (2016) "Deterministic Chaos: Applications in Cardiac Electrophysiology," Occam's Razor: Vol. 6 , Article 7. Available at: https://cedar.wwu.edu/orwwu/vol6/iss1/7 This Research Paper is brought to you for free and open access by the Western Student Publications at Western CEDAR. It has been accepted for inclusion in Occam's Razor by an authorized editor of Western CEDAR. For more information, please contact [email protected]. Klassen: Deterministic Chaos DETERMINISTIC CHAOS APPLICATIONS IN CARDIAC ELECTROPHYSIOLOGY BY MISHA KLASSEN I. INTRODUCTION Our universe is a complex system. It is made up A system must have at least three dimensions, of many moving parts as a dynamic, multifaceted and nonlinear characteristics, in order to generate machine that works in perfect harmony to create deterministic chaos. When nonlinearity is introduced the natural world that allows us life. e modeling as a term in a deterministic model, chaos becomes of dynamical systems is the key to understanding possible. ese nonlinear dynamical systems are seen the complex workings of our universe. One such in many aspects of nature and human physiology. complexity is chaos: a condition exhibited by an is paper will discuss how the distribution of irregular or aperiodic nonlinear deterministic system. blood throughout the human body, including factors Data that is generated by a chaotic mechanism will a ecting the heart and blood vessels, demonstrate appear scattered and random, yet can be dened by chaotic behavior.
    [Show full text]
  • Dynamics of a 2D Lattice of Van Der Pol Oscillators with Nonlinear Repulsive Coupling
    Dynamics of a 2D lattice of van der Pol oscillators with nonlinear repulsive coupling I.A. Shepelev ,∗ S.S. Muni ,y T.E. Vadivasova∗ September 22, 2020 Abstract We describe spatiotemporal patterns in a network of identical van der Pol oscillators coupled in a two-dimensional geometry. In this study, we show that the system under study demonstrates a plethora of different spatiotemporal structures including chimera states when the coupling pa- rameters are varied. Spiral wave chimeras are formed in the network when the coupling strength is rather large and the coupling range is short enough. Another type of chimeras is a target wave chimera. It is shown that solitary states play a crucial role in forming an incoherence cluster of this chimera state. They can also spread within the coherence cluster. Furthermore, when the coupling range increases, the target wave chimera evolves to the regime of solitary states which are randomly distributed in space. Growing the coupling strength leads to the attraction of solitary states to a certain spatial region, while the synchronous regime is set in the other part of the system. This spatiotemporal pattern represents a soli- tary state chimera, which is firstly found in the system of continuous-time oscillators. We offer the explanation of these phenomena and describe the evolution of the regimes in detail. Introduction The study of the self-organization phenomena in complex multicomponent sys- tems in the form of oscillatory networks and ensembles is one of the most relevant arXiv:2009.09584v1 [nlin.AO] 21 Sep 2020 directions in the nonlinear dynamics and related disciplines [18, 28, 30, 4, 8].
    [Show full text]
  • Invariant Manifolds of the Bonhoeffer-Van Der Pol Oscillator
    Invariant manifolds of the Bonhoeffer-van der Pol oscillator R. Ben´ıtez1, V. J. Bol´os2 1 Dpto. Matem´aticas,Centro Universitario de Plasencia, Universidad de Extremadura. Avda. Virgen del Puerto 2. 10600, Plasencia (C´aceres),Spain. e-mail: [email protected] 2 Dpto. Matem´aticas,Facultad de Ciencias, Universidad de Extremadura. Avda. de Elvas s/n. 06071, Badajoz, Spain. e-mail: [email protected] May 2007 Abstract The stable and unstable manifolds of a saddle fixed point (SFP) of the Bonhoeffer- van der Pol oscillator are numerically studied. A correspondence between the existence of homoclinic tangencies (which are related to the creation or destruction of Smale horseshoes) and the chaos observed in the bifurcation diagram is described. It is observed that in the non-chaotic zones of the bifurcation diagram, there may or may not be Smale horseshoes, but there are no homoclinic tangencies. 1 Introduction The Bonhoeffer van der Pol oscillator (BvP) is the non-autonomous planar system 9 x3 x0 = x − − y + I(t) = 3 ; (1) y0 = c(x + a − by) ; being a, b, c real parameters, and I (t) an external forcement. We shall consider only a periodic forcement I(t) = A cos (2πt) and the specific values for the parameters a = 0:7, b = 0:8, c = 0:1. These values were considered in [1] because of their physical and biological importance (see [2]). In previous works [3], the existence of \horseshoe chaos" in BvP was studied analytically by means of the Melnikov method applied to an equivalent system, the Duffing-van der Pol oscillator (DvP).
    [Show full text]
  • 1 Van Der Pol Oscillator
    1 Van der Pol oscillator The second order non-linear autonomous differential equation d2x dx + µ x2 − 1 + x =0, µ> 0 (1) dt2 dt is called the van der Pol equation. It describes many physical systems collec- tively called van der Pol oscillators. The equation models a non-conservative system in which energy is added to and subtracted from the system, result- ing in a periodic motion called a limit cycle. The parameter mu is a positive scalar indicating the nonlinearity and the strength of the damping. The dx sign of the damping term in Eq. (1), x2 − 1 changes, depending upon dt whether |x| is larger or smaller than unity. 1.1 Numerical integration Let’s write Eq. (1) as a first order system of differential equations, dx = y, dt (2) dy 2 dx = −µ (x − 1) − x, dt dt The results of numerical integration of Eqs. (2) are presented in Figs. 1–4. Numerical integration of Eq. (2) shows that every initial condition (except x = 0,x ˙ = 0) approaches a unique periodic motion. The nature of this limit cycle is dependent on the value of µ. For small values of µ the motion is nearly sinusoidal, whereas for large values of µ it is a relaxation oscillation, meaning that it tends to resemble a series of step functions, jumping between positive and negative values twice per cycle. Numerical integration shows that the limit cycle is a closed curve en- closing the origin in the x-y phase plane. From the fact that Eqs. (2) are invariant under the transformation x → −x, y → −y, we may conclude that the curve representing the limit cycle is point symmetric about the origin.
    [Show full text]
  • Arxiv:1910.05233V1 [Physics.Comp-Ph] 11 Oct 2019 Email Address: [email protected] (Artem Chashchin)
    Predicting dynamical system evolution with residual neural networks Artem Chashchina,b,∗, Mikhail Botcheva, Ivan Oseledetsb,c, George Ovchinnikovb aKeldysh Institute of Applied Mathematics of the Russian Academy of Sciences, Moscow, Russia bSkolkovo Institute of Science and Technology, Moscow, Russia cMarchuk Institute of Numerical Mathematics of the Russian Academy of Sciences, Moscow, Russia Abstract Forecasting time series and time-dependent data is a common problem in many applications. One typical example is solving ordinary differential equation (ODE) systemsx _ = F (x). Oftentimes the right hand side function F (x) is not known explicitly and the ODE system is described by solution samples taken at some time points. Hence, ODE solvers cannot be used. In this paper, a data- driven approach to learning the evolution of dynamical systems is considered. We show how by training neural networks with ResNet-like architecture on the solution samples, models can be developed to predict the ODE system solution further in time. By evaluating the proposed approaches on three test ODE systems, we demonstrate that the neural network models are able to reproduce the main dynamics of the systems qualitatively well. Moreover, the predicted solution remains stable for much longer times than for other currently known models. Keywords: dynamical systems, residual networks, deep learning ∗Corresponding author arXiv:1910.05233v1 [physics.comp-ph] 11 Oct 2019 Email address: [email protected] (Artem Chashchin) Preprint submitted to Journal of Computational Physics October 14, 2019 1. Introduction Neural network techniques are becoming an important tool for analyzing time-dependent data sets and multivariate time series. One typical problem is to reconstruct solution of an ODE (ordinary differential equation) system x_ = F (x) by learning the right hand side function F (x) with a suitable neural network.
    [Show full text]
  • 8 Example 1: the Van Der Pol Oscillator (Strogatz Chapter 7)
    Dynamical systems 2018 [email protected] 8 Example 1: The van der Pol oscillator (Strogatz Chapter 7) So far we have seen some different possibilities of what can happen in two-dimensional systems (local and global attractors and bifurcations) using mainly constructed examples. In this and the next lectures we consider two examples with real-world applications. 8.1 Self-sustained oscillations Self-sustained oscillations are frequent in nature and in technology, some examples being stick-slip oscillations, unwanted mechanical vi- brations, and oscillators in biology. Usually a system with self-sustained oscillations has a feedback mechanism such that small-amplitude os- cillations grow in size. The small oscillations are frequently modeled using an oscillator with negative damping 2 x¨ − γx_ + !0x = 0 : The system for x and y =x _ has an unstable spiral at the origin, blow- ing up small-scale oscillations. As oscillations grow, non-linear terms may form a stable limit cycle and the system shows self-sustained oscillations. One frequently used model is the van der Pol oscillator: x¨ + µ(x2 − 1) x_ + x = 0 (1) | {z } f(x) Nonlinear damping coefficient f(x) damps large oscillations (friction when jxj > 1) and amplifies small oscillations (forcing when jxj < 1) ) We expect self-sustained oscillations to be possible. Indeed the corresponding dynamical system x_ = y y_ = −x − µ(x2 − 1)y 0 p shows ap stable limit cycle if µ > 0 (after a rescaling x = x = µ and y = y0= µ, the system has a supercritical Hopf bifurcation): 1 Dynamical systems 2018 [email protected] Period time and shape of the cycle depends on µ.
    [Show full text]