DiaBayes documentation#
DiaBayes (pronounced like the mafic rock type diabase) is a collection of Python tools for forward and inverse modelling of rock friction experiments. It is designed to be modular in terms of the physical components that define a forward problem, so that the user can mix and match different types of physics, e.g. classical rate-and-state friction combined with a more exotic state evolution law, or combining the Chen-Niemeijer-Spiers model with an inertial spring-block. These models can be solved in a forward sense (generating a friction time series), or in the inverse sense (estimating the parameters that generated your measured friction). To greatly accelerate the maximum-likelihood and Bayesian inversion, DiaBayes is optimised for GPU execution using JAX and several high-performance libraries (Diffrax, Optax, and Optimistix, to name a few).
See the getting started section for a quick overview of the installation options and basic usage. A self-contained example Jupyter Notebook is included in the GitHub repo (examples/simple_inversion.ipynb). Keep reading for more detailed background on the modelling procedures and implementation.
The Getting Started guide covers the installation options and a few practical example usage cases. Start here if you’re new to DiaBayes.
The User Guide explains a number of core concepts in depth, with detailed descriptions of commonly-used functionalities.
The API Reference Guide is auto-generated based on the docstrings of all the DiaBayes classes and routines.
Found a bug? Want to contribute? How to cite? Go here to find out more.