[petsc-users] error vector for TSARKIMEX

[Jed Brown]

It is intended that you can implement TSAdapt to change scheme within a family.

It'd be typical to use TSEvaluateWLTE or TSEvaluateStep/TSErrorWeightedNorm.  Would those do what you want?

Mark Lohry <mlohry at gmail.com> writes:

> Is there a mechanism to get the error estimate with a fully implicit IMEX
> method without using adaptive time stepping? TSGetTimeError always returns
> 0, but I guess that's only implemented for GLEE methods.
>
> I'm trying to assess time accuracy for constant steps but varying the SDIRK
> type.