[Paper Review] Holistic finite differences ensure fidelity to Burger's equation

AJ Roberts|arXiv (Cornell University)|Jan 13, 1999

Nonlinear Dynamics and Pattern Formation20 references3 citations

TL;DR

This paper introduces a holistic finite difference method based on centre manifold theory to create highly accurate numerical models of Burger's equation. By treating the PDE as a whole system rather than individual terms, the method ensures fidelity to the true dynamics, especially in nonlinear regimes, outperforming conventional discretisations in accuracy and stability.

ABSTRACT

I analyse a generalised Burger's equation to develop an accurate finite difference approximation to its dynamics. The analysis is based upon centre manifold theory so we are assured that the finite difference model accurately models the dynamics and may be constructed systematically. The trick to the application of centre manifold theory is to divide the physical domain into small elements by introducing insulating internal boundaries which are later removed. Burger's equation is used as an example to show how the concepts work in practise. The resulting finite difference models are shown to be significantly more accurate than conventional discretisations, particularly for highly nonlinear dynamics. This centre manifold approach treats the dynamical equations as a whole, not just as the sum of separate terms---it is holistic. The techniques developed here may be used to accurately model the nonlinear evolution of quite general spatio-temporal dynamical systems.

Motivation & Objective

• To develop a systematic, accurate finite difference approximation for the generalized Burger's equation.
• To ensure the numerical model faithfully replicates the true dynamics of the underlying PDE.
• To overcome limitations of conventional finite difference schemes in capturing nonlinear spatio-temporal behavior.
• To demonstrate the effectiveness of a holistic approach—treating the PDE as a whole system—through rigorous dynamical systems theory.
• To provide a generalizable framework for modeling complex spatio-temporal dynamical systems with high accuracy.

Proposed method

• The method employs centre manifold theory to systematically derive the finite difference stencil.
• The physical domain is divided into small elements using insulating internal boundaries, which are later removed to ensure consistency.
• The approach treats the entire dynamical system holistically, avoiding term-by-term discretisation.
• The resulting finite difference model is derived from the underlying dynamics, ensuring it captures slow manifold behavior accurately.
• The technique is applied to Burger's equation as a proof of concept for nonlinear PDEs.
• The method ensures that the discrete model inherits the qualitative and quantitative behavior of the continuous system.

Experimental results

Research questions

• RQ1 How can a finite difference scheme be systematically constructed to preserve the true dynamics of a nonlinear PDE like Burger's equation?
• RQ2 What advantages does a holistic approach—treating the PDE as a whole—offer over conventional term-wise discretisation?
• RQ3 Can centre manifold theory be effectively used to derive accurate and stable finite difference models for nonlinear PDEs?
• RQ4 How does the holistic finite difference method compare in accuracy to conventional schemes, especially in highly nonlinear regimes?
• RQ5 To what extent does the method ensure fidelity to the slow manifold dynamics of the system?

Key findings

• The holistic finite difference method produces significantly more accurate results than conventional finite difference schemes, especially under strong nonlinearity.
• The method ensures that the discrete model faithfully reproduces the dynamics of the continuous system, including slow manifold behavior.
• By using centre manifold theory, the method provides a systematic and rigorous framework for model construction.
• The introduction and subsequent removal of insulating internal boundaries enables consistent and accurate stencil derivation.
• The approach is generalizable and can be applied to a wide class of spatio-temporal dynamical systems beyond Burger's equation.
• The resulting models exhibit improved stability and accuracy, particularly in long-time simulations of nonlinear evolution.