Skip to content

README.md

Latest commit

 

History

History
86 lines (57 loc) · 2.93 KB

README.md

File metadata and controls

86 lines (57 loc) · 2.93 KB

multiFabConverter Offline Tools

This directory provides offline tools to read AMReX plot files (.plt) and convert them to NumPy arrays or 2D slices, using pyAMReX (MPI-enabled).

main.py — parallel visualization

Purpose

Reads an AMReX plot file in a parallel MPI run. AMReX's DistributionMapping automatically distributes boxes across ranks. The script:

  1. Prints how many boxes each MPI rank owns.
  2. Gathers the full domain into a NumPy array via input_mf[:] (collective operation across all ranks).
  3. On rank 0, extracts a 2D slice of x_velocity at a given physical z-coordinate and saves it as slice_z.png.

Dependencies

  • Python 3.8+
  • numpy, matplotlib, mpi4py
  • pyAMReX (MPI-enabled build): conda install -c conda-forge "pyamrex=*=*mpi_openmpi*"

Usage

mpirun -np <N> python main.py <pltfile> [lev] [z_real]
Argument Description Default
pltfile Path to the AMReX plot file directory (e.g. plt00002) required
lev AMR level to read (0 = coarsest) 0
z_real Physical z-coordinate for the 2D slice 5.0

Examples:

mpirun -np 4 python main.py plt00002
mpirun -np 4 python main.py ../../../Tutorials/FlowPastSphere/plt00002 0 5.0

Output

  • Terminal: AMReX/MPI initialization info; per-rank box count.
  • File: slice_z.png written by rank 0 in the current directory.

Note: AMReX is initialized with amr.initialize([]). Do not pass sys.argv to it — AMReX's ParmParse will misinterpret the arguments and abort.

script_plt_to_npz.py — batch conversion to NPY

Purpose

Single-process script (no mpirun needed) that reads one or more AMReX plot files, selects specified field components, and stacks them into a NumPy array of shape (n_samples, ncomp, nx, ny) saved as a .npy file. Intended for machine learning workflows or offline analysis.

Usage

python script_plt_to_npz.py <pltfile> [pltfile2 ...] [-o output.npy] [-l lev] [-c comp1 comp2 ...]
Argument Description Default
pltfile One or more plot file paths, or folders containing plt* subdirectories (uses the latest one) required
-o Output .npy file path dataset.npy
-l AMR level to export 0
-c Space-separated list of component names to export x_velocity y_velocity avg_pressure

Examples:

# Single plot file with default components
python script_plt_to_npz.py plt00002

# Multiple plot files, export x_velocity only
python script_plt_to_npz.py plt00001 plt00002 plt00003 -c x_velocity -o uvel.npy

# Folder containing multiple plt* subdirectories (uses the latest)
python script_plt_to_npz.py ../../../Tutorials/LidDrivenCavity/Re_100/

Output

  • Terminal: path of each plot file read; final array shape and component names.
  • File: .npy array of shape (n_samples, ncomp, nx, ny).