Probabilistic Evolution of Stochastic Dynamical Systems: A Meso-scale Perspective

04/11/2020
by   Chao Yin, et al.
0

Stochastic dynamical systems arise naturally across nearly all areas of science and engineering. Typically, a dynamical system model is based on some prior knowledge about the underlying dynamics of interest in which probabilistic features are used to quantify and propagate uncertainties associated with the initial conditions, external excitations, etc. From a probabilistic modeling standing point, two broad classes of methods exist, i.e. macro-scale methods and micro-scale methods. Classically, macro-scale methods such as statistical moments-based strategies are usually too coarse to capture the multi-mode shape or tails of a non-Gaussian distribution. Micro-scale methods such as random samples-based approaches, on the other hand, become computationally very challenging in dealing with high-dimensional stochastic systems. In view of these potential limitations, a meso-scale scheme is proposed here that utilizes a meso-scale statistical structure to describe the dynamical evolution from a probabilistic perspective. The significance of this statistical structure is two-fold. First, it can be tailored to any arbitrary random space. Second, it not only maintains the probability evolution around sample trajectories but also requires fewer meso-scale components than the micro-scale samples. To demonstrate the efficacy of the proposed meso-scale scheme, a set of examples of increasing complexity are provided. Connections to the benchmark stochastic models as conservative and Markov models along with practical implementation guidelines are presented.

READ FULL TEXT

page 12

page 21

page 24

page 26

research
12/30/2019

Incorporating physical constraints in a deep probabilistic machine learning framework for coarse-graining dynamical systems

Data-based discovery of effective, coarse-grained (CG) models of high-di...
research
07/21/2022

Multi-element flow-driven spectral chaos (ME-FSC) method for uncertainty quantification of dynamical systems

The flow-driven spectral chaos (FSC) is a recently developed method for ...
research
12/29/2021

DeePN^2: A deep learning-based non-Newtonian hydrodynamic model

A long standing problem in the modeling of non-Newtonian hydrodynamics i...
research
12/02/2020

Flow-driven spectral chaos (FSC) method for simulating long-time dynamics of arbitrary-order non-linear stochastic dynamical systems

Uncertainty quantification techniques such as the time-dependent general...
research
05/23/2016

Learning and Policy Search in Stochastic Dynamical Systems with Bayesian Neural Networks

We present an algorithm for model-based reinforcement learning that comb...
research
12/19/2022

Stochastic stability analysis of legged locomotion using unscented transformation

In this manuscript, we present a novel method for estimating the stochas...

Please sign up or login with your details

Forgot password? Click here to reset