PHYSICAL REVIEW E 88, 042924 (2013)
Stability and dynamics of nonautonomous systems with pulsed nonlinearity
Yannis Kominis
1,2
and Tassos Bountis
2
1
School of Electrical and Computer Engineering, National Technical University of Athens, Zographou GR-15773, Greece
2
Department of Mathematics, University of Patras, Patras GR-26500, Greece
(Received 11 July 2013; revised manuscript received 25 September 2013; published 30 October 2013)
We study the dynamics of a class of nonautonomous systems with pulsed nonlinearity that consist of a periodic
sequence of linear and nonlinear autonomous systems, each one acting alone in a different time or space interval.
We focus on the investigation of control capabilities of such systems in terms of altering their fundamental
dynamical properties by appropriate parameter selections. For the case of single oscillators, the stability of the
zero solution as well as the phase space topology is shown to drastically depend on parameters such as the
frequency of the linear oscillations and the durations of the linear and nonlinear intervals. In cases of chain of
coupled oscillators with pulsed onsite nonlinearity, it is shown that appropriate parameter selections can stabilize
an otherwise unstable zero background allowing for the existence of dynamically robust localized excitations,
whose evolution properties can now be explicitly determined and controlled.
DOI: 10.1103/PhysRevE.88.042924 PACS number(s): 05.45.Gg, 05.45.Yv, 45.05.+x
I. INTRODUCTION
Nonautonomous dynamical systems are commonly used as
models for a large variety of natural or man-made systems. In
such systems the independent variable measuring the evolution
appears explicitly in the model equations. The independent
variable can either have the dimensions of time or space. In the
first case, it describes the explicit time variation of one of the
system parameters, which may include a direct external driving
force acting on the system. This time dependence usually
comes from interaction with a different dynamical system,
which is considered as external and practically unaffected from
the system under investigation. A large number of systems fall
into this category, including forced oscillators and oscillators
with time-varying parameters, which are the basic paradigm in
the theory of complex dynamical systems [1,2]. In the second
case, when the independent variable corresponds to a spa-
tial dimension, the nonautonomous systems describe spatial
dynamics in spatially inhomogeneous media. This category
includes the large class of models describing periodic or
solitary wave formation in transversely [3,4] or longitudinally
[5] inhomogeneous structures.
In all cases, the introduction of the explicit dependence
on the independent variable results in an additional degree of
freedom in comparison to the respective autonomous system.
As a result, a symmetry property is lost, the correspond-
ing invariant quantity of the autonomous system becomes
time-varying, and the phase space of the system undergoes
a qualitative modification. In the case of an autonomous
integrable Hamiltonian system, it is well known that even small
nonautonomous perturbations lead, in general, to nonintegra-
bility and irregular dynamics. The utilization of perturbation
methods allows for the investigation of the relation between
the dynamics of the original autonomous integrable system and
the perturbed nonautonomous systems. The Poincare-Birkhoff
theorem as well as Melnikov’s theory for periodic orbits [6,7]
predict that when certain resonance conditions between the
unperturbed system and the time-dependent perturbation are
satisfied, a finite discrete family of periodic solutions persist
under perturbation. Additionally, the Melnikov’s theory for
homoclinic orbits [6,7] relates the existence of an unperturbed
orbit homoclinic to a fixed point with the persistence of a
discrete number of orbits homoclinic to a periodic orbit.
We study the dynamics of a class of nonautonomous
systems for which the explicit dependence on the independent
variable is due to a periodically pulsed nonlinearity. Alterna-
tively, these systems can be considered as nonlinear systems
whose action is interrupted by a periodically interposed linear
system. Such systems consist of a periodic sequence of au-
tonomous linear and nonlinear systems, each one acting alone
in a different time or space interval. We consider a periodic
sequence of an either integrable or nonintegrable nonlinear
and a linear Hamiltonian system. The special conditions
under which the interposition of the linear system does not
drastically modify the dynamics of the respective autonomous
nonlinear system has been previously studied [2,8]. It has
been shown that under such conditions the solutions of the
total nonautonomous systems consists of the solutions of its
nonlinear part interrupted by complete oscillations due to
the linear part. In this work, we consider the general case
where these conditions are not fullfiled, so that the dynamics
of the nonautonomous system can be drastically different from
those of its autonomous nonlinear part, depending crucially
on the parameters. This dependence suggests a control mech-
anism capable of determining some of the most characteristic
features of nonlinear systems, that is their behavior under
the presence of noise and the extent of regions with chaotic
dynamics in the phase space as well as the propagation of
localized excitations in spatially inhomogeneous systems and
chains of oscillators.
In Sec. II we consider single oscillators with pulsed
nonlinearity. The stability type of the zero solution determines
their behavior under the presence of noise and the extent of
the chaotic region of the phase space, which can be therefore
controlled by appropriate choices of the durations of the
intervals at which the nonlinear and linear parts of the system
act. Among several applications of such systems we can refer
to the modeling of particle beam dynamics in storage rings
of high-energy accelerators, where the nonlinear part of the
system corresponds to focusing elements and the linear part to
the beam propagation between them, with the crucial issue
042924-1 1539-3755/2013/88(4)/042924(7) ©2013 American Physical Society