Building and Running RMHD2D:
1. Edit one of the machine-dependent Makefiles,
e.g. Make.reynolds-linux, to set up the various compilers, compiler
flags, libraries, and Fortran-to-C translation directives.
NOTE: this version of the code is designed to interface with
sundials versions 2.4.0 or 2.5.0, although it may work with
older/newer versions as well. However, for versions <= 2.3.0, when
installing sundials you must ensure that the source code file
is copied to the installation directory. Unfortunately, this is
not the default behavior when building/installing in a different
directory than where it is installed.
2. Link this machine-dependent Makefile to Make.machine: e.g.
ln -fs Make.reynolds-linux Make.machine
3. Edit one of the problem configuration makefiles,
e.g. Make.config-KH, to set up the appropriate dimensionality (2D
vs 2.5D), boundary conditions, flux discretization choices,
parallelism, and preconditioning for the type of problem that you
wish to run.
4. Link this problem-dependent Makefile to Make.config: e.g.
ln -s Make.config-KH Make.config
5. Type 'make <target>', where <target> specifies the type of executable
(problem type and time integration algorithm) that you wish to
build. Typing 'make' alone will bring up a list of possible
6. Copy problem-specific input files to the run-time directory from
the ./inputs directory. You will need the files mhd.inp, prop.inp
and mesh.inp for all problems (remove the problem specifier suffix
from the file name). For some problems, you will also need the
problem-specific input parameter files provided in this directory,
e.g. the Kelvin-Helmholtz problem requires kh_init.inp. If the
time-integration algorithm is KINSOL, you will also need
kincontrol.inp; if it is CVODE, you will need cvcontrol.inp.
7. Edit the input files to your tastes and run your executable.
* To set the problem size and parallelism topology, edit the
mesh.inp file to set the problem size (nx, ny and nz give the
total problem dimensions) along with the parallelism topology
information (xprocs, yprocs and zprocs give the number of
processors in each Cartesian dimension).