t8_time_partition

Overview

The t8_time_partition program creates a 2D or 3D structured mesh (a "brick"), partitions the mesh among MPI processes, and can export the partitioned mesh for visualization.

It is designed for:

• Testing partitioning strategies
• Timing partitioning performance
• Visualizing partitions in parallel environments

---

Command-Line Options

You can specify options to control mesh size, dimension, and output.

Short	Long	Argument	Default	Description
-x	--x-dim	int	1	Number of mesh cells in x-direction
-y	--y-dim	int	1	Number of mesh cells in y-direction
-z	--z-dim	int	0	Number of mesh cells in z-direction (3D only)
-d	--dim	int (2/3)	2	Mesh dimension: 2 (2D) or 3 (3D)
-o	--no-vtk	switch	off	Disable VTK output (no files written)
-h	--help	switch	off	Display help message

---

Program Workflow

1. MPI and Library Initialization: Initializes MPI and necessary t8code libraries.
2. Option Parsing: Parses command-line options for mesh setup.
3. Mesh Creation: Creates a structured "brick" mesh distributed across processes.
4. Optional Coordinate Translation: Offsets each process’s part in the x-direction for clearer visualization.
5. Partitioning: Partitions the mesh, moving 43% of each process's trees to the next process.
6. Profiling and Timing: Collects and prints timing statistics.
7. VTK Output (Optional): Writes the mesh partition to VTK (.vtu) files for visualization.
8. Cleanup and Finalization: Cleans up memory and finalizes MPI.

---

Example

2D Mesh, 4x4 cells, 2 processes:

mpirun -np 2 ./t8_time_partition -x 4 -y 4

---

Output

• Timing and profiling information for partitioning.
• VTK output files if enabled for visualization of partitions.

---

Troubleshooting

• If the program does not start or reports missing libraries, check your t8code build and MPI setup.
• Use -h to see the usage message with full options.