dive

About

dive is a Python package for Bayesian analysis of dipolar EPR (electron paramagnetic resonance) spectroscopy data through Markov chain Monte Carlo (MCMC) sampling with the Python package PyMC.

Requirements

dive is available for Windows, Mac and Linux systems and requires Python 3.9 or later and PyMC 5.0 or later.

Features

dive's features include:

An output InferenceData object containing many random posterior samples for each parameter
Full uncertainty quantification for all model parameters, including the distance distribution
Visualizations for ensembles of fitted signals and residuals
Visualizations for ensembles of fitted distance distributions
Histograms for margnialized posteriors of other parameters such as modulation depth and background decay rate

Setup

You can install dive using pip. Please note that the PyPI package name is dive-EPR.

pip install dive-EPR

You can also directly clone the dive directory. Please make sure to also import the necessary packages.

pip install pymc deerlab scipy matplotlib numpy pandas mkl-service h5netcdf pytest
git clone https://github.com/StollLab/dive

dive can then be used by importing the package as usual.

import dive

Documentation

See the documentation for a detailed guide on how to use dive. An IPython Notebook guide on using dive can also be found under the examples/ directory.

Citation

When you use dive in your work, please cite the following publication:

Bayesian Probabilistic Analysis of DEER Spectroscopy Data Using Parametric Distance Distribution Models
Sarah R. Sweger, Stephan Pribitzer, and Stefan Stoll
J. Phys. Chem. A 2020, 124, 30, 6193–6202
doi.org/10.1021/acs.jpca.0c05026

License

dive is licensed under the MIT License.