Liang-Ding / seishmc

Full moment tensor inversion using Hamiltonian Monte Carlo (HMC) sampling

SEISHMC

SEISHMC is a Python package that is based on MTUQ and utilizes Hamiltonian Monte Carlo (HMC) algorithm to invert the full moment tensor of earthquake sources.

Installation

1. Requirements

   • MTUQ (https://github.com/uafgeotools/mtuq)
   • Seaborn (>= 0.11.2)
   • Numpy
   • Pandas

2. Install seisHMC

• For basic install:

git clone https://github.com/Liang-Ding/seishmc.git
cd seishmc
pip install -e .

• or using pip

pip install seishmc

Why HMC?

We achieve lower waveform misfits in our examples using HMC algorithm.

• DC: Double-Couple solution
• FMT: Full Moment Tensor solution
• GS: Grid Search
• HMC: Hamiltonian Monte Carlo algorithm
• (Grid search is performed using MTUQ and HMC sampling is carried out with SeisHMC.)

Examples

Overview of many examples provided by SeisHMC.

Synthetic examples	Real-data examples
GS (DC), syn.	GS (DC)
GS (FMT), syn.	GS (FMT)
HMC (DC), syn.	HMC (DC)
HMC (FMT), syn.	HMC (FMT)

Check the gallery for more figures and result obtained using HMC algorithm.

Tutorials

• From grid search solver to HMC solver