This example provides both jax-based and torch-based implementations of the acoustic FWI with source encoding.
Please note that the jax-based implementation only support .hdf5 format for the observed data, while the torch-based implementation only support .npy format, at least for now.
# Jax-based implementation
# 1. Generate model geometry
python generate_model_geometry.py
# 2. Generate observed data
sh forward_hdf5.sh
# 3. Run inversion
sh sefwi_jax.sh# Torch-based implementation
# 1. Generate model geometry
python generate_model_geometry.py
# 2. Generate observed data
sh forward_npy.sh
# 3. Run inversion
sh sefwi_torch.sh