General Usage

Running a simulation requires two inputs:

1. A velocity model — a Julia array (or file path to .jld2 / .npy / .npz).
2. A configuration — a Julia Dict (or file path to a .yaml file).

The simulation dimension (2D or 3D) is automatically detected from the shape of the velocity model array: a 3-dimensional array (7 × nx × nz) triggers a 2D simulation, and a 4-dimensional array (13 × nx × ny × nz) triggers a 3D simulation.

Running a simulation

using ElasticFDSG
# config and velmod can each be a file path (String) or a Julia object (Dict / Array)
result = runsim(config, velmod)

If the configuration has output_file = nothing, runsim returns the populated FDSG struct directly, which contains all field data. If an output_file path is provided (must end in .h5), the results are written to HDF5 and runsim returns nothing.

Example — passing objects directly

using ElasticFDSG

velmod = zeros(7, 200, 200)   # fill with your data
# ... (see Velocity Models for details)

config = config_template_2d(
    device = "cpu", precision = "Float32",
    # ... (see Configurations for details)
)

fdsg = runsim(config, velmod)

Example — passing file paths

using ElasticFDSG

fdsg = runsim("path/to/config.yaml", "path/to/velmod.jld2")

Loading results from HDF5

When results are saved to disk, they can be loaded back into a nested Julia dictionary:

using ElasticFDSG

data = load_results("path/to/output.h5")

The dictionary mirrors the HDF5 group structure: