Machine learning based rupture generator

This repository contains code for Machine learning based rupture generator (MLRG)

Features

• Generate a slip realization for given dimensions of fault following Mai and Beroza (2002)
• Hypocenter location can be specified manually or using Mai et al., (2005)
• Computes kinematic source parameters such as onset times, rise times, peak slip velocities.
• Allows for source time functions (STFs) to be Yoffe or modified Yoffe using Machine learning.

Usage

• Clone the repository using git clone
• Install pytorch from https://pytorch.org/get-started/locally/
• Install few more dependencies
  • pip install monai
  • pip install scipy
  • pip install scikit-fmm
  • pip install matplotlib
• Use script gen_PD_source.ipynb to list all input parameters and generate source models

Acknowledgements

• Some Python codes have been adopted from the following sources http://equake-rc.info/cers-software/
• The fast marching code is adopted from https://github.com/scikit-fmm/scikit-fmm/tree/master

References

[1] Aquib, T. A., J. C. Vyas, and P. M. Mai (2025). Pseudo-Dynamic Source Characterization for Geometrically Rough Faults Using Machine Learning, Bull. Seismol. Soc. Am. 115, 1570–1590, doi: 10.1785/0120240237

Cite the code