Core Concepts
Fokker-Planck equations can be efficiently solved using the Chang-Cooper method combined with unconditionally positive and conservative Patankar-type time integration schemes, which preserve positivity and steady states.
Abstract
The content discusses efficient numerical methods for solving Fokker-Planck equations, which are partial differential equations that describe the evolution of probability distributions.
The key highlights are:
The Chang-Cooper method is used to discretize the Fokker-Planck equation in space, as it preserves steady states.
The explicit Euler method is not unconditionally positive, leading to severe restrictions on the time step to ensure positivity.
Instead, the authors propose to combine the Chang-Cooper method with unconditionally positive Patankar-type time integration methods, such as the modified Patankar-Euler scheme and the modified Patankar-Runge-Kutta scheme.
These Patankar-type methods are unconditionally positive, robust for stiff problems, only linearly implicit, and also higher-order accurate.
The authors apply the developed schemes to a model on opinion dynamics and compare the schemes in terms of computation time and numerical error.
The results show that the Patankar-type methods are more efficient than classical explicit or fully implicit Runge-Kutta methods, especially for large time step sizes, as they can ensure positivity and preserve steady states.
Stats
The Fokker-Planck equation is determined by:
B[f](w, t) = ∫ℐ (w-v) f(v, t) dv
D(w) = σ2/2 (1-w^2)^2