Measuring the irreversibility of numerical schemes for reversible stochastic differential equations (Q2925703)
From MaRDI portal
| This is the item page for this Wikibase entity, intended for internal use and editing purposes. Please use this page instead for the normal view: Measuring the irreversibility of numerical schemes for reversible stochastic differential equations |
scientific article; zbMATH DE number 6357679
| Language | Label | Description | Also known as |
|---|---|---|---|
| English | Measuring the irreversibility of numerical schemes for reversible stochastic differential equations |
scientific article; zbMATH DE number 6357679 |
Statements
17 October 2014
0 references
stochastic differential equations
0 references
detailed balance
0 references
reversibility
0 references
relative entropy
0 references
entropy production
0 references
numerical integration
0 references
(overdamped) Langevin process
0 references
Euler-Maruyama method
0 references
Milstein method
0 references
Measuring the irreversibility of numerical schemes for reversible stochastic differential equations (English)
0 references
The entropy production rate (EPR) is used to estimate the loss of reversibility when approximations of the solution of a reversible stochastic differential equation are generated using a numerical method. This analysis is applied to the overdamped Langevin equation to determine that the explicit Euler-Maruyama method yields order \(\Delta t^2\) decrease of the EPR in the case of additive noise, but the EPR is not reduced by reducing the stepsize \(\Delta t\) in the case of multiplicative noise.NEWLINENEWLINE However, the explicit Milstein method is shown to yield order \(\Delta t\) decrease in the EPR in the case of multiplicative noise. The BBK integrator is shown to yield order \(\Delta t\) decrease in the EPR when applied to a Langevin system with additive noise.
0 references