Closed ranocha closed 2 months ago
I cam across this again. It would be nice if we could add additional (production) terms that are not weighted by the (modified) Patankar trick
If you use a PDSProblem
you can put these terms on the diagonal of the production matrix P
. Only the off-diagonal terms of P
have corresponding destruction terms and will be weighted.
I see, thanks. So the off-diagonal terms of the production matrix are used to setup the corresponding conservative production-destruction system, the diagonal terms are used as they are in the underlying Runge-Kutta method (and should be non-negative), and the destruction terms from the additional destruction vector are weighted by the Patankar trick 👍 Do we have some tests for this setup?
Within our problem library prob_pds_stratreac
is such a nonconservative PDS. In runtests.jl
all tests that have the keyword "nonconservative" in the title deal with this kind of problems.
Thanks!
It would be great to support systems that are not necessarily conservative but have a conservative part (treated via the Patankar trick) and an additional part (integrated as usual in RK methods).