It is a challenging issue to analyze complex dynamics from observed and simulated data. An advantage of extracting dynamic behaviors from data is that this approach enables the investigation of nonlinear phenomena whose mathematical models are unavailable. The purpose of this present work is to extract information about transition phenomena (e.g., mean exit time and escape probability) from data of stochastic differential equations with non-Gaussian Lévy noise. As a tool in describing dynamical systems, the Koopman semigroup transforms a nonlinear system into a linear system, but at the cost of elevating a finite dimensional problem into an infinite dimensional one. In spite of this, using the relation between the stochastic Koopman semigroup and the infinitesimal generator of a stochastic differential equation, we learn the mean exit time and escape probability from data. Specifically, we first obtain a finite dimensional approximation of the infinitesimal generator by an extended dynamic mode decomposition algorithm. Then, we identify the drift coefficient, diffusion coefficient, and anomalous diffusion coefficient for the stochastic differential equation. Finally, we compute the mean exit time and escape probability by finite difference discretization of the associated nonlocal partial differential equations. This approach is applicable in extracting transition information from data of stochastic differential equations with either (Gaussian) Brownian motion or (non-Gaussian) Lévy motion. We present one- and two-dimensional examples to demonstrate the effectiveness of our approach.

In recent years, data-driven methods for discovering complex dynamical systems in various fields have attracted widespread attention. These methods make full use of data and have become powerful tools to study complex phenomena. In this work, we propose a framework for detecting dynamical behaviors, such as the maximum likelihood transition path, of stochastic dynamical systems from data. For a stochastic dynamical system, we use the Kramers–Moyal formula to link the sample path data with coefficients in the system, then use the extended sparse identification of nonlinear dynamics method to obtain these coefficients, and finally calculate the maximum likelihood transition path. With two examples of stochastic dynamical systems with additive or multiplicative Gaussian noise, we demonstrate the validity of our framework by reproducing the known dynamical system behavior.

With the rapid development of computational techniques and scientific tools, great progress of data-driven analysis has been made to extract governing laws of dynamical systems from data. Despite the wide occurrences of non-Gaussian fluctuations, the effective data-driven methods to identify stochastic differential equations with non-Gaussian Lévy noise are relatively few so far. In this work, we propose a data-driven approach to extract stochastic governing laws with both (Gaussian) Brownian motion and (non-Gaussian) Lévy motion, from short bursts of simulation data. Specifically, we use the normalizing flows technology to estimate the transition probability density function (solution of non-local Fokker–Planck equations) from data, and then substitute it into the recently proposed non-local Kramers–Moyal formulae to approximate Lévy jump measure, drift coefficient and diffusion coefficient. We demonstrate that this approach can learn the stochastic differential equation with Lévy motion. We present examples with one- and two-dimensional decoupled and coupled systems to illustrate our method. This approach will become an effective tool for discovering stochastic governing laws and understanding complex dynamical behaviours. This article is part of the theme issue ‘Data-driven prediction in dynamical systems’.

With the rapid development of computational techniques and scientific tools, great progress of data-driven analysis has been made to extract governing laws of dynamical systems from data. Despite the wide occurrences of non-Gaussian fluctuations, the effective data-driven methods to identify stochastic differential equations with non-Gaussian Lévy noise are relatively few so far. In this work, we propose a data-driven approach to extract stochastic governing laws with both (Gaussian) Brownian motion and (non-Gaussian) Lévy motion, from short bursts of simulation data. Specifically, we use the normalizing flows technology to estimate the transition probability density function (solution of nonlocal Fokker-Planck equation) from data, and then substitute it into the recently proposed nonlocal Kramers-Moyal formulas to approximate Lévy jump measure, drift coefficient and diffusion coefficient. We demonstrate that this approach can learn the stochastic differential equation with Lévy motion. We present examples with one-and two-dimensional, decoupled and coupled systems to illustrate our method. This approach will become an effective tool for discovering stochastic governing laws and understanding complex dynamical behaviors.

No abstract