Reproducible material for A Deep Learning-based time shift objective function for Full Waveform Inversion - Alfarhan M., Chen F., Turkiyyah G., Keyes D., Vasconcelos I., Ravasi M.

Project structure

This repository is organized as follows:

• 📂 deeptimeshift: python library containing routines for time shift prediction for FWI;
• 📂 asset: folder containing logo;
• 📂 data: folder containing data (or instructions on how to retrieve the data);
• 📂 notebooks: set of jupyter notebooks reproducing the experiments in the paper (see below for more details);
• 📂 scripts: set of python scripts reproducing the experiments in the paper (see below for more details);

Notebooks

The following notebooks are provided:

• 📙 Training_Marmousi.ipynb: notebook training a neural network on time shift estimation using Marmousi.
• 📙 FWI_L2_Marmousi.ipynb: notebook performing FWI with L2 norm loss on Marmousi.
• 📙 FWI_SoftDTW_Marmousi.ipynb: notebook performing FWI with SoftDTW loss on Marmousi.
• 📙 FWI_TimeshiftNN_Marmousi.ipynb: notebook performing FWI with the time shift estimated with the neural network on Marmousi.
• 📙 Training_Chevron.ipynb: notebook training a neural network on time shift estimation using Chevron data.

Scripts

The following scripts are provided:

• 🐍 FWI_Chevron.ipynb: script performing FWI on Chevron data with argparsing for different loss functions.

Getting started 👾 🤖

To ensure reproducibility of the results, we suggest using the environment-cpu.yml or environment-gpu.yml file when creating an environment.

Simply run:

./install_env.sh

It will take some time, if at the end you see the word Done! on your terminal you are ready to go. Note that if cuda is detected, environment-gpu.yml will be used otherwise environment-cpu.yml.

Remember to always activate the environment by typing:

conda activate deeptimeshift

After that you can simply install your package:

pip install .

or in developer mode:

pip install -e .

If you get an error related to Numpy > 2.0, follow these steps:

pip uninstall numpy
pip install numpy==1.26.4

Disclaimer: All experiments have been carried on a Intel(R) Xeon(R) CPU @ 3.90GHz equipped with a single NVIDIA GEForce RTX 3090 GPU. Different environment configurations may be required for different combinations of workstation and GPU.