Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Time-Spectral Efficiency

Jan Scheffel

TL;DR

This paper evaluates time-spectral methods, emphasizing the Generalized Weighted Residual Method (GWRM), as efficient alternatives for solving stiff and chaotic differential equations. By representing solutions with truncated Chebyshev expansions in time (and space) and solving a global algebraic system via the SIR solver, GWRM demonstrates implicit robustness that mitigates stiffness and chaotic amplification, enabling long interval solution with high accuracy. However, non-smooth or partially steep solutions pose convergence challenges, and previously proposed smoothing strategies (Time Averaging, Long Time Averaging, and Time Integration) offer limited or problem-dependent gains. The findings highlight both the potential and limits of time-spectral approaches, motivating further development of methods to address steep or non-smooth regions while preserving the gains in convergence and efficiency for stiff/chaotic problems.

Abstract

This study concerns the efficiency of time-spectral methods for numerical solution of differential equations. It is found that the time-spectral method GWRM demonstrates insensitivity to stiffness and chaoticity due to the implicit nature of the solution algorithm. Accuracy is thus determined primarily by numerical resolution of the solution shape. Examples of efficient solution of stiff and chaotic problems, where explicit methods fail or are significantly slower, are given. Non-smooth and partially steep solutions, however, remain challenging for convergence and accuracy. Some, earlier suggested, smoothing algorithms are shown to be ineffective in addressing this issue. Our findings underscore the need for further exploration of time-spectral approaches to enhance convergence and accuracy for steep or non-smooth solutions.

Time-Spectral Efficiency

TL;DR

This paper evaluates time-spectral methods, emphasizing the Generalized Weighted Residual Method (GWRM), as efficient alternatives for solving stiff and chaotic differential equations. By representing solutions with truncated Chebyshev expansions in time (and space) and solving a global algebraic system via the SIR solver, GWRM demonstrates implicit robustness that mitigates stiffness and chaotic amplification, enabling long interval solution with high accuracy. However, non-smooth or partially steep solutions pose convergence challenges, and previously proposed smoothing strategies (Time Averaging, Long Time Averaging, and Time Integration) offer limited or problem-dependent gains. The findings highlight both the potential and limits of time-spectral approaches, motivating further development of methods to address steep or non-smooth regions while preserving the gains in convergence and efficiency for stiff/chaotic problems.

Abstract

This study concerns the efficiency of time-spectral methods for numerical solution of differential equations. It is found that the time-spectral method GWRM demonstrates insensitivity to stiffness and chaoticity due to the implicit nature of the solution algorithm. Accuracy is thus determined primarily by numerical resolution of the solution shape. Examples of efficient solution of stiff and chaotic problems, where explicit methods fail or are significantly slower, are given. Non-smooth and partially steep solutions, however, remain challenging for convergence and accuracy. Some, earlier suggested, smoothing algorithms are shown to be ineffective in addressing this issue. Our findings underscore the need for further exploration of time-spectral approaches to enhance convergence and accuracy for steep or non-smooth solutions.
Paper Structure (23 sections, 31 equations, 2 figures)

This paper contains 23 sections, 31 equations, 2 figures.

Table of Contents

  1. Introduction
  2. The Generalized Weighted Residual Method (GWRM)
  3. Steepness and oscillations - temporal smoothing
  4. The Time Averaging (TA) Method
  5. Long Time Averaging (LTA) and Time Integration (TI) Methods
  6. Resolving oscillations
  7. Stiff and chaotic nonlinear problems
  8. Autocatalytical chemical reactions - a stiff problem
  9. The Lorenz 1984 equations - a chaotic problem
  10. Achievable gain from time-spectral methods
  11. Convergence and accuracy
  12. Efficiency
  13. GWRM solution of non-smooth problems
  14. The Time-Integration (TI) Method
  15. The Time-Average (TA) Method
  16. ...and 8 more sections

Figures (2)

  • Figure 1: SGWRM time-adaptive solution of Robertson equations, using 49 time intervals for $K=6$, with accuracy $\epsilon=0.001$. Note that $y(t)$ has been magnified a factor $10^4$.
  • Figure 2: SGWRM solution of the Lorenz equations with parameters $a=0.25$, $b=4.0$, $F=8.0$ and $G=1.0$. The initial conditions are $(X,Y,Z)=(0.96,-1.1,0.5)$. From top to bottom; $X(t)+8$, $Y(t)+4$, $Z(t)$. The time-adaptive GWRM uses 61 time intervals and $K=8$ to obtain an accuracy $\epsilon=0.001$.