Installation

This page explains the installation of gravis and its optional dependencies Selenium for static image exports and Jupyter notebooks for embedded visualizations.

gravis

This package is available on the Python Package Index (PyPI) under the name gravis. It can be installed with Python’s default package manager pip:

$ pip install gravis

Further remarks:

  • gravis is compatible with Python 3.5 upwards.

  • Using an environment manager like virtualenv or conda is often a good idea. Such tools allow to create a virtual environment into which pip can install the package in an isolated fashion instead of globally on your system. Thereby it does not interfere with the installation of other projects, which may require a different version of some shared dependency.

Optional: Selenium

gravis requires Selenium to export visualizations as static images in JPG, PNG and SVG format. It can be installed in two steps:

  1. Selenium Python bindings are available on the Python Package Index (PyPI) under the name selenium. It can be installed with Python’s default package manager pip:

    $ pip install selenium

  2. Webbrowser and associated driver

    Selenium connects Python to a webbrowser via a suitable driver. Both the webbrowser and corresponding driver need to be installed manually. Two options are supported by gravis:

    Further remarks:

    • gravis uses Chrome by default, but Firefox can be chosen with an argument of the export functions.

    • The browser version and driver version need to be compatible.

    • The driver executable needs to be in a directory that is listed in the PATH environment variable of the operating system, so that Python can find it. This can be achieved by putting the executable in a directory that is already listed in the PATH environment variable, such as /usr/bin or /usr/local/bin on Linux, or by putting the executable in another directory and adding it to PATH, for example on Linux with an entry in the .bashrc file in the user directory.

Optional: Jupyter notebook

gravis requires Jupyter notebook for displaying graphs inline in notebooks in form of embedded HTML visualizations, accompanied by Python code and Markdown comments. A notebook is a file ending with .ipynb, which can be created, modified and executed with the notebook server and a webbrowser opened by it. Jupyter notebook can be installed according to its recommended installation with the package manager conda from channel conda-forge:

$ conda install -y -c conda-forge notebook

Further remarks:

  • Caution: Some plots may not show up in the notebook with default settings. Instead only a blank area is visible. The reason is a parameter called iopub_data_rate_limit in Jupyter’s config system. Its value is chosen rather low by default. Plots that contain much data can therefore be blocked. This problem can be solved by increasing the value of the parameter, which can be done in two ways:

    1. Permanently change it with a config file in the directory ~/.jupyter or

    2. Temporarily change it when opening a notebook by adding an optional argument to the start-up command:

      $ jupyter notebook --NotebookApp.iopub_data_rate_limit=1.0e12