>> Trung tâm Vật lý tính toán

Công bố khoa học của Nguyễn Văn Liễn

Phys. Stat. Sol. B 175, 319-330 (1993)


Stochastic dynamics - crossover from 1/f(3) to flicker noise

E. Canessa, V. L. Nguyen

Finite time processes within the limits of the Newton equation and zero inertia motion (i.e., road to chaos) are studied by numerically solving the ordinary, stochastic Langevin equation in 1D for a free particle with inertia moving in a medium with viscosity gamma. In these simulations, the scaling behaviour of particle trajectories x(t) and velocities v(t) with time are derived and the inclusion of non-zero particle masses is shown to define the asymptotic time limit tau(c) at which - independently of gamma - the system evolves into the well-known statistically stationary state characterized by [x2(t)] infinity t and flicker noise. The time tau(c) is further analysed from the correlation length given by the two-point autocorrelation function of the particle velocity at each value of gamma. It is found that the noise power spectrum of v(t) is characterized by flicker noise for frequencies f less-than-or-equal-to f(c) almost-equal-to 1/tau(c), whereas for f > f(c), the noise power spectra behaves as 1/f(nu), where nu varies between the limits of Newton's equation (i.e., nu = 3) and road to chaos (i. e., nu = 1). Furthermore, at times tau < tau(c) and 0 < gamma < infinity, an ad-hoc exponent for the scaling of the variance of particle velocities with time is shown to characterize a subset of multifractal dimensions d(f)(gamma) while the single particle trajectories are shown to display a rather different subset of exponents on increasing gamma. Generic features of this transition are nicely given by Poincare maps in the velocity space.