Rigorous validation of stochastic transition paths

DOI10.1016/J.MATPUR.2019.04.012MaRDI QIDQ2334873zbMATH OpenOpenAlexFDO

Publication date 8 November 2019

Published in Journal de Mathématiques Pures et Appliquées. Neuvième Série (Search for Journal in Brave)

Full work available at URL https://arxiv.org/abs/1804.10171

fixed-point problem heteroclinic orbit invariant manifolds minimum-energy path rigorous numerics stochastic dynamics

Invariant manifolds for ordinary differential equations (34C45) Homoclinic and heteroclinic orbits for dynamical systems (37C29) Action-minimizing orbits and measures for finite-dimensional Hamiltonian and Lagrangian systems; variational principles; degree-theoretic methods (37J51) Computational methods for stochastic equations (aspects of stochastic analysis) (60H35) Algorithms with automatic result verification (65G20)

Abstract: Global dynamics in nonlinear stochastic systems is often difficult to analyze rigorously. Yet, many excellent numerical methods exist to approximate these systems. In this work, we propose a method to bridge the gap between computation and analysis by introducing rigorous validated computations for stochastic systems. The first step is to use analytic methods to reduce the stochastic problem to one solvable by a deterministic algorithm and to numerically compute a solution. Then one uses fixed-point arguments, including a combination of analytical and validated numerical estimates, to prove that the computed solution has a true solution in a suitable neighbourhood. We demonstrate our approach by computing minimum-energy transition paths via invariant manifolds and heteroclinic connections. We illustrate our method in the context of the classical M"uller-Brown test potential.

Review text

The goal of the paper is an approach for rigorous validated computation for stochastic systems. The analytical part of the method is based on the fixed-point theorems in some Banach space. For the demonstration of the approach, the authors consider the following stochastic ordinary differential equation \[ dX=-\nabla V(X)dt+\varepsilon W(t), \] where \({0<\varepsilon\ll1}\), \(W=(W_1,W_2)\) are independent identically distributed Brownian motions and \(V\) is the Müller-Brown potential, i.e., \[ V(x,y)=\sum_{i=1}^4\alpha_i\exp\left(a^{(i)}(x-x^{(i)}_0)^2+b^{(i)}(x-x^{(i)}_0)(y-y^{(i)}_0)+c^{(i)}(y-y^{(i)}_0)^2\right) \] with some standard parameters \({\alpha,x_0,y_0,a,b,c\in\mathbb{R}^4}.\) The authors validate numerically obtained minimum energy path (MEP) for such a system by rigorously computing a chain of heteroclinic orbits. The following steps are done: localization and validation of all saddles and minima; introduction an equivalent polynomial vector field; computation and validation of the local unstable manifold of the saddles; validation of the trapping regions around the minima; computation and validation of the orbits connecting unstable manifolds and trapping regions. The authors use Matlab with the Intlab package for computing. Some generalizations are discussed at the end of paper.

Recommendations

Cites work

Cited in

(13)

Describes a project that uses

Uses Software

This page was built for publication: Rigorous validation of stochastic transition paths

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q2334873)