Installation

You will need Python >= 3.6.

Windows users

Please favor Conda, as Conda will seamlessly install the HDF5 standard library which is a required dependency.

From PyPI

pip install tramway

pip install will install some Python dependencies if missing, but you may still need to install the HDF5 reference library.

Note that the HDF5 library can usually be installed using any OS’ package manager. Only Windows users may have to manually download and install the HDF5 library, if they do not wish to use Conda instead of pip.

Several installation targets are available, including full, that install optional dependencies:

pip install tramway[full]

Most of the functionalities and code examples described in this documentation will run without the optional dependencies this target installs in addition to the required dependencies. It is safe to first install TRamWAy with minimal requirements and then pip install the missing dependencies as you hit ImportError while using TRamWAy.

Using Conda

conda install tramway -c conda-forge

Compared with the bare package pip installs (with no installation targets), the conda package specifies some optional dependencies as required, so that the above command installs them. However, many other optional dependencies are omitted.

Git version

Initial install:

git clone https://github.com/DecBayComp/TRamWAy
cd TRamWAy
pip install -e .

Can be updated with git pull run in the local repository.

Documentation

To compile the documentation and get a local copy, after installing TRamWAy do:

cd doc
make html

The generated documentation will be available at _build/html/index.html from the doc repository.

Building the documentation requires Sphinx.

OS and version specific notes

Some modules require Python>=3.7. These are bayes_factor for force “detection” based on Bayesian statistics, snr that extracts signal-to-noise ratios required by the bayes_factor module, and d.conj_prior which estimates the diffusion similarly to ddrift with no regularization but with additional confidence intervals.