This sounds attractive, but unfortunately the Harten-Hyman fix
generally needs to know more about the system than just the waves and
speeds that come out of something like a Roe solver. The point of the
fix is to effectively spread a jump propagating at a single speed s
into corrections that affect the cells on both sides of the interface
when the wave should really be a transonic rarefaction. But to detect
that and use the H-H formulas, we need an estimate of the wave speed
for this $k$th family on each side of the rarefaction wave, i.e. an
estimate of the characteristic speed in each intermediate state
bounding this wave (the $\lambda^k_{l,r}$ in formula (15.49)). The
intermediate states are easily compute from the waves, but then
formulas are needed for the characteristic speed in these states.
These are easily computed for systems like Euler or shallow water but
of course are different than the speed coming out of the Roe solver.
Perhaps there's still a nice way to turn it into more of a black box
if it's assumed there is also another function available that returns
the eigenvalues of the Jacobian f'(q) for any intermediate state q,
for example.
- Randy
> --
> You received this message because you are subscribed to the Google Groups
> "claw-dev" group.
> To unsubscribe from this group and stop receiving emails from it, send an
> email to
claw-dev+u...@googlegroups.com.
> To post to this group, send email to
claw...@googlegroups.com.
> Visit this group at
http://groups.google.com/group/claw-dev?hl=en.
> For more options, visit
https://groups.google.com/groups/opt_out.
>
>