Stochastic processes with finite correlation time: modeling and application to the generalized Langevin equation

Generalized Einstein relations and conditions for anomalous relaxation

The generalized Einstein relation (GER) for nonergodic processes is investigated within the framework of the generalized Langevin equation. The conditions for anomalous relaxation such as long-tail decay and non-vanishing velocity autocorrelation function (VAF) are proposed and distinguished. For the stationary nonergodic process, if the initial preparation of the particle velocity is non-thermal, an asymptotic GER occurs in a departure from the usual result. It is shown that the GER holding is a necessary condition rather than a full condition for the system being close to equilibrium. For the nonergodic process of the second type due to cutoff of high frequencies, the VAF oscillates with time, the GER holds but the equilibrium fails in the long-time limit. Applications to some practical examples confirm the present theoretical findings.

Collective Langevin dynamics of conformational motions in proteins

Functionally relevant slow conformational motions of proteins are, at present, in most cases inaccessible to molecular dynamics (MD) simulations. The main reason is that the major part of the computational effort is spend for the accurate description of a huge number of high frequency motions of the protein and the surrounding solvent. The accumulated influence of these fluctuations is crucial for a correct treatment of the conformational dynamics; however, their details can be considered irrelevant for most purposes. To accurately describe long time protein dynamics we here propose a reduced dimension approach, collective Langevin dynamics (CLD), which evolves the dynamics of the system within a small subspace of relevant collective degrees of freedom. The dynamics within the low-dimensional conformational subspace is evolved via a generalized Langevin equation which accounts for memory effects via memory kernels also extracted from short explicit MD simulations. To determine the memory kernel with differing levels of regularization, we propose and evaluate two methods. As a first test, CLD is applied to describe the conformational motion of the peptide neurotensin. A drastic dimension reduction is achieved by considering one single curved conformational coordinate. CLD yielded accurate thermodynamical and dynamical behaviors. In particular, the rate of transitions between two conformational states agreed well with a rate obtained from a 150 ns reference molecular dynamics simulation, despite the fact that the time scale of the transition (approximately 50 ns) was much longer than the 1 ns molecular dynamics simulation from which the memory kernel was extracted.

Non-Markovian stochastic processes: colored noise

We survey classical non-Markovian processes driven by thermal equilibrium or nonequilibrium (nonthermal) colored noise. Examples of colored noise are presented. For processes driven by thermal equilibrium noise, the fluctuation-dissipation relation holds. In consequence, the system has to be described by a generalized (integro-differential) Langevin equation with a restriction on the damping integral kernel: Its form depends on the correlation function of noise. For processes driven by nonequilibrium noise, there is no such a restriction: They are considered to be described by stochastic differential (Ito- or Langevin-type) equations with an independent noise term. For the latter, we review methods of analysis of one-dimensional systems driven by Ornstein-Uhlenbeck noise.

Stochastic equation for a jumping process with long-time correlations

A jumping process, defined in terms of the jump size and waiting time distributions, is presented. The jumping rate depends on the process value. The process, which is Markovian and stationary, relaxes to an equilibrium and is characterized by a power-law autocorrelation function. Therefore, it can serve as a model of 1/f noise as well as of the stochastic force in the generalized Langevin equation. This equation is solved for noise correlations approximately 1/t ; the resulting velocity distribution has sharply falling tails. The system preserves memory about the initial condition for a very long time.

Functional characterization of linear delay Langevin equations

We present an exact functional characterization of linear delay Langevin equations driven by any noise structure defined through its characteristic functional. This method relies on the possibility of finding an explicitly analytical expression for each realization of the delayed stochastic process in terms of those of the driving noise. General properties of the transient dissipative dynamics are analyzed. The corresponding interplay with a color Gaussian noise is presented. As a full application of our functional method we study a model for population growth with non-Gaussian fluctuations: the Gompertz model driven by multiplicative white shot noise.

Simple jumping process with memory: transport equation and diffusion

We present a stochastic jumping process, defined in terms of jump-size probability density and jumping rate, which is a generalization of the well-known kangaroo process. The definition takes into account two process values: after and before the jump. Therefore, the process is able to preserve memory about its previous values. It possesses a simple stationary limit. Its master equation is interpreted as the kinetic equation with variable collision rate. The process can be easily applied to model systems which relax to distributions other than Maxwellian. The case of a constant jumping rate corresponds to the diffusion process, either normal or ballistic.

Solving the Chapman-Kolmogorov equation for a jumping process

A general solution to the Chapman-Kolmogorov equation for a jumping process called the "kangaroo process" is derived. A special case of algebraic dependences is discussed in detail. In particular, simple asymptotic formulas for probability distribution are presented. It is demonstrated that there are two different classes of limiting stationary distributions. An expression for the covariance is also derived.