Hello Users,
On the telecon today we discussed briefly about plans for implementing MHD in GRHydro of the ET. We did not discuss our science goals for the ET collaboration regarding MHD, but RIT believes this an important topic to consider as we should all be driven by science achievements. Just as we have made plans for our first hydro-related science projects that we intend to publish together as a collaboration, we should also have a "magnetized" set of projects on which we all agree to collaborate.
Now, regarding details of the development: Christian showed interest in implementing the magnetized version of Con2Prim.c. Scott proposed using HARM3D (a three-dimensional MHD code written for "arbitrary" time-independent spacetimes)
Here, I'll list a few papers on my GRMHD code called HARM3D and online, GPL-licensed con2prim code.
2D "HARM" is described by: http://adsabs.harvard.edu/abs/2003ApJ...589..444G
HARM can be downloaded here: http://rainman.astro.illinois.edu/codelib/codes/harm/harm.tgz
and it's "con2prim" routine here which we call "pvs_grmhd": http://rainman.astro.illinois.edu/codelib/codes/pvs_grmhd/pvs_grmhd.tgz http://adsabs.harvard.edu/abs/2006ApJ...641..626N
I would recommend using my 3D version "HARM3D", which is described here: http://adsabs.harvard.edu/abs/2009ApJ...692..411N
HARM3D uses the basic algorithms as HARM, but adds a third spatial dimension, higher-order reconstruction and constrained transport methods (parabolic interpolators), and a more sophisticated recovery procedure when con2prim fails. It uses a slightly different formulation for the conserved variables, but it should be easy translate it into the Valencia formulation.
I have not made HARM3D freely available on the web yet, but I am willing to donate routines from it for open distribution through ET. The procedure we (at RIT) plan to follow is that a routine from HARM3D will be made compatible with ET, we will verify it functions properly, and then we will submit it to the development branch.
Aspects of the MHD code that---I think---can be easily "translated" to the GRHydro framework are:
1) con2prim 2) approximate Riemann solution for the HLL 3) constrained transport (with some labor to be spent adding support in GRHydro for additional flux functions) 4) recovery procedure (may require more effort)
One aspect that the maintainers may wish to consider is that the license for the "pvs_grmhd" asks users to cite two papers when one publishing results derived from it. Would this be agreeable with the collaboration?
And, if we end up using these routines from HARM3D, I, too, would appreciate citation, but acknowledge that I can't demand it as I'm donating it ala GPL.
Cheers,
Scott Noble
&
The RIT Team