Installation

Gcmstools requires Python and a number of third-party packages. Below is a complete list of packages and minium versions:

• Python >=3.4 (2.x versions not supported any longer)

• Pip >=6.0.6 (might be part of newer Python releases)

• Setuptools >=11.3.1

• Numpy >=1.9.1

• Matplotlib >= 1.4.2

• Pandas >=0.15.2

• IPython >=2.3.1

• PyTables >=3.1.1

• Scipy >=0.14.0

• Sphinx >=1.2.2 (Optional for documentation.)

  • numfig is a Sphinx extenstion that is needed to autonumber figures references in the docuementation.

IPython is also a very useful advanced interactive Python interpreter. Examples in this documentation assume that you are using this environment. See the IPython section of Appendix A for more details.

Python

The most convenient installation method for third-party Python packages is the all-in-one Anaconda Python distribution. It combines a large number of Python packages for scientific data analysis and a program (conda) for managing package updates, in addition to many other advanced features. The Anaconda developers (Continuum Analytics) provide a lot of useful documentation on installing Anaconda and using conda. There are other ways to install Python and the third-party packages, but for this documentation, it will be assumed that your are using Anaconda.

Note

On Mac/Linux, Python is already part of the operating system. Do not try to install these third-party packages into the builtin Python distribution unless you really know what you are doing. You might overwrite an important file, which can cause problems for your system. Confusion between the system and Anaconda Python installation is a common source of problems for beginners, so make sure that your Anaconda Python is “activated” before running the commands in this document. (See the Anaconda documentation for more information on the activation process.)

Learning the usage of all of these Python packages is far beyond the scope of this document. However, excellent documentation for most of the packages as well as full tutorials are easily discovered.

gcmstools

There are three installation options for gcmstools: 1) install using conda (recommended), 2) install using pip, or 3) install the most recent development version using git.

Option 1 (recommended)

If you are using the Anaconda Python distribution, you can use conda to install the most recent distribution of gcmstools from Binstar, Continuum’s package repository.

home>$ conda install -c https://conda.binstar.org/rnelsonchem gcmstools

To uninstall gcmstools from a conda environment, use the following command:

home>$ conda remove gcmstools

Option 2

gcmstools can also be installed from the official Python packaging site, PyPI, using the standard Python installer script pip. This installation method will work fine if you have the additional dependencies installed; otherwise, you may have some problems depending on your platform.

home>$ pip install gcmstools

To uninstall gcmstools using pip:

home>$ pip uninstall gcmstools

Option 3

The development version of gcmstools is hosted on GitHub. To use this version, you must install the version-control software Git. gcmstools can then be downloaded and installed with one pip command.

home>$ pip install git+https://github.com/rnelsonchem/gcmstools.git

There are advantages and disadvantages with this approach. For example, the GitHub repo will always have the most recent updates and bug fixes; however, these new features might not be as well-tested so you might find new bugs.

To uninstall, use the method described in Option 2