Efficient semi-implicit schemes for stiff systems

  • Authors:
  • Qing Nie;Yong-Tao Zhang;Rui Zhao

  • Affiliations:
  • Department of Mathematics, Center for Mathematical and Computational Biology, University of California, Irvine, 275 MSTB, Irvine, CA 92697-3875, United States;Department of Mathematics, Center for Mathematical and Computational Biology, University of California, Irvine, 275 MSTB, Irvine, CA 92697-3875, United States;Department of Mathematics, Center for Mathematical and Computational Biology, University of California, Irvine, 275 MSTB, Irvine, CA 92697-3875, United States

  • Venue:
  • Journal of Computational Physics
  • Year:
  • 2006

Quantified Score

Hi-index 31.48

Visualization

Abstract

When explicit time discretization schemes are applied to stiff reaction-diffusion equations, the stability constraint on the time step depends on two terms: the diffusion and the reaction. The part of the stability constraint due to diffusion can be totally removed if the linear diffusions are treated exactly using integration factor (IF) or exponential time differencing (ETD) methods. For systems with severely stiff reactions, those methods are not efficient because the reaction terms in IF or ETD are still approximated with explicit schemes. In this paper, we introduce a new class of semi-implicit schemes, which treats the linear diffusions exactly and explicitly, and the nonlinear reactions implicitly. A distinctive feature of the scheme is the decoupling between the exact evaluation of the diffusion terms and implicit treatment of the nonlinear reaction terms. As a result, the size of the nonlinear system arising from the implicit treatment of the reactions is independent of the number of spatial grid points; it only depends on the number of original equations, unlike the case in which standard implicit temporal schemes are directly applied to the reaction-diffusion system. The stability region for this class of methods is much larger than existing methods using an explicit treatment of reaction terms. In particular, the one with second order accuracy is unconditionally linearly stable with respect to both diffusion and reaction. Direct numerical simulations on test equations, as well as morphogen systems from developmental biology, show the new semi-implicit schemes are efficient, robust and accurate.