Reproducible material for Robust Full Waveform Inversion with deep Hessian deblurring
Alfarhan M., Ravasi M., Chen F., Alkhalifah T.

Project structure

This repository is organized as follows:

• 📂 deepinvhessian: python library containing routines for "DeepFWIInvHessian" Full Waveform Inversion Inverse Hessian with Deep Learning;
• 📂 data: folder containing input data;
• 📂 notebooks: set of jupyter notebooks reproducing the experiments in the paper (see below for more details);
• 📂 asset: folder containing logo;

Supplementary files

The data supporting the findings of this work are available from the corresponding author upon reasonable request.

Notebooks

The following notebooks are provided:

• 📙 Run_Conventional_FWI.ipynb : notebook performing conventional FWI (will be updated later).
• 📙 Run_FWI_Born.ipynb : notebook estimating the inverse Hessian with the migration/demigration approach (will be updated later).
• 📙 Run_FWI_PSF.ipynb : notebook estimating the inverse Hessian with the PSFs approach (will be updated later).
• 📙 FWI-LBFGS-Scipy.ipynb : notebook performing FWI with L-BFGS algorithm from the Scipy implementation (will be updated later).
• 📙 PlottingNotebook.ipynb : notebook reproducing the figures in the paper (for the first report).
• 📙 Marmousi_exp.ipynb : notebook performing FWI with the Barzilai-Borwein method and the proposed approach on Marmousi.
• 📙 Marmousi_LBFGS.ipynb : notebook performing FWI with L-BFGS on Marmousi.
• 📙 Marmousi_create_figures.ipynb : notebook to visualize the results of the Marmousi experiments.
• 📙 Volve_synthetic_exp.ipynb : notebook performing FWI with the Barzilai-Borwein method and the proposed approach on Volve synthetic.
• 📙 Volve_synthetic_LBFGS.ipynb : notebook performing FWI with L-BFGS on Volve synthetic.
• 📙 Volve_synthetic_create_figures.ipynb : notebook to visualize the results of the Volve synthetic experiments.
• 📙 Volve_exp.ipynb : notebook performing FWI with the Barzilai-Borwein method and the proposed approach on Volve.
• 📙 Volve_LBFGS.ipynb : notebook performing FWI with L-BFGS on Volve.
• 📙 Volve_imaging.ipynb : notebook to compute RTM images and extended images for Volve.
• 📙 Volve_create_figures.ipynb : notebook to visualize the results of the Volve experiments.

Getting started

To ensure reproducibility of the results, we suggest using the environment.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. Activate the environment by typing:

conda activate deepinvhessian

After that you can simply install your package:

pip install .

or in developer mode:

pip install -e .

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.

Cite Us

@misc{alfarhan2024robustwaveforminversiondeep, title={Robust Full Waveform Inversion with deep Hessian deblurring}, author={Mustafa Alfarhan and Matteo Ravasi and Fuqiang Chen and Tariq Alkhalifah}, year={2024}, eprint={2403.17518}, archivePrefix={arXiv}, primaryClass={physics.geo-ph}, url={https://arxiv.org/abs/2403.17518}, }