Q. How do I run a solver in parallel?
In a desktop environment, simply prefix the solver executable with the mpirun helper. For
example, to run the Incompressible Navier-Stokes solver on a 4-core desktop computer, you
would run
mpirun -np 4 IncNavierStokesSolver Cyl.xml
In a cluster environment, using PBS for example, the mpiexec command should be
used.
Q. How can I generate a mesh for use with Nektar++?
Nektar++ supports a number of mesh input formats. These are converted to the Nektar++ native XML format (see Section 3) using the NekMesh utility (see Section 4. Supported formats include:
Gmsh (.msh)
Polygon (.ply)
Nektar (.rea)
Semtex (.sem)
Q. When running my solver, I see files appearing with .bak extensions. What are
these?
When running a Nektar++ solver, if a checkpoint or field file already exists with the same
name as that to be written, by default a backup will be created where that directory
is renamed. For example, if writing a file named output.fld, then the existing
output.fld will be renamed to output.bak0.fld. If the solver were run again, then the
output.fld this would be renamed to output.bak1.fld to avoid overwriting the
output.bak0.fld file. This is a deliberate choice to avoid accidentally overwriting simulation
data.
Both NekMesh and FieldConvert utilities will also prevent mesh and visualisation files being
overwritten by default.
Q. How can I disable backup output?
Backup output is useful but can produce a large number of files, which can be an obstruction when developing with Nektar++ or debugging. There are two ways to disable backups:
Pass the command line option –force-output or -f. This will turn off backup
output for Nektar++ solvers, or the NekMesh and FieldConvert utilities.
Set the environment variable NEKTAR_DISABLE_BACKUPS, for example by including the
line:
export NEKTAR_DISABLE_BACKUPS=1
While this environment variable exists, it will prevent backups from being written.