Workflow

Preprocessing

• Create GMSH model.
• Mesh it, generate a .msh file.
• Parse the .msh file within Python using MeshIO.
• Push the mesh representation from MeshIO into a Mesh object (discretedomain.Mesh)
• Optionally partition the mesh into contiguous chunks - this will facilitate faster point location queries. See discretedomain.Mesh.build_partitions().
• Create cell fields and vertex fields on Mesh, see demo_mesh.py:test_3d_tetra().
• Emit mesh data to binary file using discretedomain.Mesh.write().
• Emit field data to binary file using discretedomain.Mesh.write_fields().
• Emit vtu file of Mesh and any cell or vertex fields using Mesh.vtu().

Simulation

• Load binary representation of Mesh using parse.c:parse_mesh().
• Load binary representation of cell/vertex fields using parse.c:parse_field().
• Perform point location queries using point_in_tetra.c:PointLocation_PartitionedBoundingBox(),

Requirements

1. MeshIO (https://pypi.org/project/meshio/1.2.0/)

   pip install meshio

2. METIS (http://glaros.dtc.umn.edu/gkhome/metis/metis/overview)

   Installation instructions

   • Download this file http://glaros.dtc.umn.edu/gkhome/fetch/sw/metis/metis-5.1.0.tar.gz
   • tar zxvf metis-xxx.tar.gz
   • cd metis-5.1.0
   • make config shared=1
     • Note shared=1 is required to ensure the build creates a shared / dynamic library which is essential for interoperability with Python.
   • make all
   • Lastly you will need to set the following environment variable METIS_DLL to indicate the full path to the metis shared/dynamic library. For example
     export METIS_DLL=${PWD}/build/Darwin-arm64/libmetis/libmetis.dylib

3. METIS for Python (https://metis.readthedocs.io/en/latest/)

   pip install metis

4. Compile parse.c, point_in_tetra.c via the following

   • gcc -c -O2 -Wall -std=c99 parse.c
   • gcc -c -O2 -Wall -std=c99 point_in_tetra.c
   • A demo / testbed for the C routines are provided in demo_parse.c. This can be compiled via the following
     gcc -O2 -Wall -std=c99 demo_parse.c -o demo parse.o point_in_tetra.o

Examples

1. First execute the following python parse_regions_from_gmsh.py. This will load a default .msh file and output the mesh and mesh partitions (md.bin) and the region data associated with the cells (region_cell.bin). The user can change the .msh file loaded using the option --filename my_file.msh, e.g.

   python parse_regions_from_gmsh.py --filename my_file.msh
   

2. Now execute ./demo. This will load md.bin and region_cell.bin and perform point locations on particles with randomly generated coordinates.