Predicting catastrophes in nonlinear dynamical systems by compressive sensing

An extremely challenging problem of significant interest is to predict catastrophes in advance of their occurrences. We present a general approach to predicting catastrophes in nonlinear dynamical systems under the assumption that the system equations are completely unknown and only time series reflecting the evolution of the dynamical variables of the system are available. Our idea is to expand the vector field or map of the underlying system into a suitable function series and then to use the compressive-sensing technique to accurately estimate the various terms in the expansion. Examples using paradigmatic chaotic systems are provided to demonstrate our idea.     

Collisions and umbilic catastrophes

Two procedures to evaluate the hyperbolic umbilic canonical diffraction integral, a function that occurs in many applications of semiclassical collision theory and optics, are described. The first is the integration of the various differential equations satisfied by this function. We present schemes to perform the numerical integration and investigate their range of validity. A second algorithm, which evaluates the integral directly by quadrature, is given. Here, attention is focused on the very rapidly oscillating regions of the integrand. The direct quadrature method is shown to be more widely applicable and better suited to numerical computation.     

Phonon focusing catastrophes

We present an elegant application of formal catastrope theory to phonon focusing phenomena in crystals. The standard analysis of phonon focusing breaks down near singular directions known as caustics. The theory we present, in addition to being numerically accurate, provides a qualitative picture of caustic surfaces which yields important new insights into the results of phonon propagation experiments. Supporting experimental evidence is also presented.     

Simple iron-free strong-focusing systems

In this paper variations are proposed for simple iron-free strong-focusing charged particle accelerators. Two methods for constructing such systems are considered. The first strong-focusing method is based on a spatial variation of the magnetic field along the orbit. The second method is based on a time variation of the focusing magnetic field. In both cases we find the equations for betatron oscillations, the conditions for confining the particles, and the conditions determining the position of the working current in the center of the stability region. Certain modifications of the systems are discussed.Translated from Izvestiya VUZ. Fizika, No. 12, pp. 19–23, December, 1971.In conclusion the author thanks M. S. Rabinovich for consultation and valuable instructions.     

Instabilities,bifurcations and catastrophes

Two independently developed bifurcational theories for the instability of a slowly evolving system, such as a stellar mass, are correlated. Applied to the fracture of a mechanically stressed perfect crystal, they predict a sharp cusp on the failure stress locus.     

Strongly nonlinear shear perturbations in discrete and continuous cubic nonlinear systems

Systems with constraints, the masses in which move only along guides, can execute strongly nonlinear vibrations. This means that nonlinear phenomena manifest themselves at arbitrary small deviations from equilibrium. The form of vibrations of a single mass is described by elliptical Jacobi functions. The spectrum of these vibrations is found. With an increase in amplitude, the period of vibrations decreases. We deduce equations of strongly nonlinear vibrations of a chain of connected masses. In the continuum limit, we obtain a new nonlinear equation in partial derivatives. We devise transformation of variables leading to linearization of this equation. We implemented a factorization procedure that decreases the order of the equation in partial derivatives from second to first. Exact solutions to the first-order equation describe the slow evolution of the displacement profile in a distributed system. In the absence of preliminary tension of elastic elements in the continued model, traveling waves cannot be achieved; however, time-oscillating solutions like standing waves are possible. We obtain an equation for a field of strongly nonlinear deformations. Its exact solution describes periodic movement in time and space. As well, the period of time oscillations decreases with an increase in amplitude, and the spatial period, in contrast, increases. The product of the vibration frequency multiplied by the spatial period is a constant that depends on the deformation energy. We propose a scheme of the mechanical system producing strongly nonlinear torsional vibrations. We experimentally measured the period of torsional vibrations of a single disc. We show that with an increase in amplitude in the process of vibration attenuation, an increase in the period occurs, which agrees with calculations. We measure the shapes of nonlinear vibrations of a chain of connected discs. A strongly nonlinear behavior of the chain is observed.     

Strong perturbations in nonlinear systems

A novel case of probabilistic coupling for hybrid stochastic systems with chaotic components via Markovian switching is presented. We study its stability in the norm, in the sense of Lyapunov and present a quantitative scheme for detection of stochastic stability in the mean. In particular we examine the stability of chaotic dynamical systems in which a representative parameter undergoes a Markovian switching between two values corresponding to two qualitatively different attractors. To this end we employ, as case studies, the behaviour of two representative chaotic systems (the classic Rössler and the Thomas-Rössler models) under the influence of a probabilistic switch which modifies stochastically their parameters. A quantitative measure, based on a Lyapunov function, is proposed which detects regular or irregular motion and regimes of stability. In connection to biologically inspired models (Thomas-Rössler models), where strong fluctuations represent qualitative structural changes, we observe the appearance of stochastic resonance-like phenomena i.e. transitions that lead to orderly behavior when the noise increases. These are attributed to the nonlinear response of the system.     

Similaritons in nonlinear optical systems

By using the lens-type transformation, we find the self-similar evolution of four kinds of optical similaritons in (1+1)-dimensional nonlinear Schrödinger equations. They are the exact solitonic similaritons, quasi solitonic similaritons, asymptotic compact parabolic similaritons and asymptotic Hermite-Gaussian similaritons. Moreover, the interaction of multiple solitonic similaritons are investigated. Approximate but highly accurate analytical methods are developed to describe their center-of-mass motion.     

Adiabatic condition for nonlinear systems

The adiabatic approximation is an important concept in quantum mechanics. In linear systems, the adiabatic condition is derived with the help of the instantaneous eigenvalues and eigenstates of the Hamiltonian, a procedure that breaks down in the presence of nonlinearity. Using an explicit example relevant to photoassociation of atoms into diatomic molecules, we demonstrate that the proper way to derive the adiabatic condition for nonlinear mean-field (or classical) systems is through a linearization procedure, using which an analytic adiabatic condition is obtained for the nonlinear model under study.     

Nonlocal feedback in nonlinear systems

A shifted or misaligned feedback loop gives rise to a two-point nonlocality that is the spatial analog of a temporal delay. Important consequences of this nonlocal coupling have been found both in diffusive and in diffractive systems, and include convective instabilities, independent tuning of phase and group velocities, as well as amplification, chirping and even splitting of localized perturbations. Analytical predictions about these nonlocal systems as well as their spatio-temporal dynamics are discussed in one and two transverse dimensions and in presence of noise.     

Mechanical systems with nonlinear constraints

A geometrical formalism for nonlinear nonholonomic Lagrangian systems is developed. The solution of the constrained problem is discussed by using almost product structures along the constraint submanifold. Constrained systems with ideal constraints are also considered, and Chetaev conditions are given in geometrical terms. A Noether theorem is also proved.     

Fluctuations in nonlinear Markovian systems

We study nonlinear irreversible processes by statistical mechanical methods from a general point of view. Assuming that the macroscopic variables behave approximatively Markovian we derive evolution equations for the mean values as well as for the fluctuations about the mean. The mean values obey nonlinear transport equations and the fluctuations obey linear nonstationary Langevin equations. The equations of motion are completely specified by the entropy and the transport coefficients as functions of the macroscopic state. The theory provides a statistical mechanical basis for some phenomenological approaches put forward recently.