I have an advection problem with a sever time stepping restriction. I am using explicit RK type time stepping schemes and standard finite difference so everything is matrix free. Other than going for an implicit scheme, are there some purely numerical schemes I can look into to alleviate my time stepping restriction and perhaps accelerate my time to solution?

  • 2
    $\begingroup$ Can you give some more details on your problem? Is it linear/nonlinear, size of the discretized system, etc. For example, if you have a linear problem with very large eigenvalues, there might not be much you can do to improve explicit time-stepping $\endgroup$
    – whpowell96
    Commented May 10 at 15:32
  • $\begingroup$ Yeah I think I got my answer $\endgroup$ Commented May 10 at 15:34
  • 1
    $\begingroup$ For strictly hyperbolic PDEs your options are somewhat limited. The best you usually can do is pick a different RK method which has a higher effective CFL per stage. If you have something mixed hyperbolic/parabolic and its the parabolic part which is causing problems you can try Runge-Kutta-Chebychev methods. $\endgroup$ Commented May 10 at 15:59
  • 1
    $\begingroup$ There are also mixed implicit-explicit time stepping methods where you solve the stiff parts implicitly and everything else explicitly, but this might not be what you're looking for and it's by no means a panacea. $\endgroup$ Commented May 10 at 16:01


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Browse other questions tagged or ask your own question.