Installation¶

Note that pygeostat requires Python 3.6+. Pygeostat will likely require modification to work with any other version of python. Pygeostat is also strongly dependent on the suite of curated python packages provided with the Anaconda Python distribution. It is highly recommended to install this prior to pygeostat installation.

Full Installation Guide and Test Notebook¶

A full installation guide and test case is provided in the pygeostat_installation.zip file, which is available on the Pygeostat Knowledge Base page at https://ccgsrv.geostats.ualberta.ca/ccgkb/doku.php?id=software:indexes:pygeostat

It is strongly recommended that new users follow the step-by-step instructions that are available in the PDF within this zip file, before applying the test notebook. Though similar instructions are provided below, the instructions in this PDF are more more explicit.

Python Installation¶

The Anaconda distribution by Continuum Analytics: https://www.continuum.io/downloads is recommended as it includes all dependencies for pygeostat. If unfamiliar with Python and virtual environments, then accepting the defaults of registering with the system and installing to the path are recommended. All distributed .whl pygeostat versions are compiled targeting a 64-bit machine with an Anaconda Python 3.6+.

pygeostat Installation¶

The easiest way to install pygeostat is to download the .whl file satisfying the specifications of your machine and python installataion (e.g., at present this is only a 64-bit machine with Python 3.6+). At this point it is assumed that a Anaconda Python distribution with Python 3.6+ is installed. After downloading the .whl, pygeostat is installed by opening a command prompt in the folder containing the .whl, and executing the following command:

> pip install pygeostat-<ver>-py36-none-any.whl

where ver refers to the current version of pygeostat under consideration.

From Source¶

For Windows¶

The easiest way to install pygeostat is to follow the .whl method provided above. To build from source, download the desired version of pygeostat from the CCG Knowledge Base at https://ccgsrv.geostats.ualberta.ca/ccgkb/doku.php

Extract the zip file, then from a command prompt:

> cd pygeostat
> python setup.py install -buildfortran

The -buildfortran flag requires a relatively recent (>0.503) version of pygeostat. This flag ensures that all required components are compiled. The setup script installs pygeostat into your Python 3.6+ site-packages directory. You are ready to start using pygeostat!

Warning

This assumes that an Anaconda Python distribution is installed. It should work without one, but upon first import of pygeostat a number of errors will be thrown for missing packages. These dependencies can be resolved by conda update <package>, but it is highly recommended to use the Anaconda Installer referenced above.

For Linux¶

Warning

Compiling pygeostat for linux is not currently tested or supported.

Compile Fortran under a linux environment, such as cygwin with:

$ make

Alternatively from the Windows command prompt:

$ cd fortran
$ compile_fortransubs.bat
$ cd ..

Then install pygeostat with:

$ python setup.py install

You are ready to start using pygeostat!

Dependencies¶

Anaconda Packages¶

Anaconda provides the major dependencies of pygeostat:

NumPy+SciPy - Python math libraries - http://www.scipy.org/

Pandas - a Python data analysis library - http://pandas.pydata.org/

Matplotlib - a Python plotting library - http://matplotlib.org/

Pyevtk¶

A package to facilitate writing of vtk files:

> pip install pyevtk

Pyperclip¶

A package that facilitates the copying of python-created strings to the clipboard:

> pip install pyperclip

Optional Dependencies¶

Seaborn¶

A Python visualization library - http://stanford.edu/~mwaskom/software/seaborn/. Seaborn isn’t packaged with the Anaconda installer but is available by using the command:

> conda install seaborn

Using R from Python¶

Consider:

rpy2 - a Python library providing a low-level interface to R from Python - http://rpy.sourceforge.net/

Matrix - a R library - https://stat.ethz.ch/R-manual/R-devel/library/base/html/matrix.html

To allow interfacing between Python and R, rpy2 is required. Currently, only the utility gs.nearpd() uses rpy2 which also requires the R library “Matrix” to be installed through R. Installing rpy2 can be temperamental. The following installation instructions for rpy2 should do the trick:

Download the rpy2 wheel file (e.g., rpy2-2.6.0-cp34-none-win_amd64) from: http://www.lfd.uci.edu/~gohlke/pythonlibs/#rpy2

From the command prompt, install the wheel file by using the command
>>> pip install rpy2-2.6.0-cp34-none-win_amd64.whl
Add the path to the folder containing R.dll to the environment variable PATH (e.g., C:Program FilesRR-3.1.2/bini386)

Add an environment variable R_HOME pointing to R (e.g., C:Program FilesRR-3.1.2)

Add an environment variable R_USER that is your windows user name

For Developers¶

Compiling Documentation¶

Requires:

Sphinx - Python documentation generator:
```
> conda install sphinx
```
Sphinx readthedocs theme - theme for Sphinx:
```
> conda install sphinx_rtd_theme
```

See other packages for documentation guidelines and build instructions.

F2PY¶

Compiling the Fortran module source requires:

mingw - Fortran and C compilers, from either:

conda install mingw

http://mingw-w64.org/, and add bin to path

f2py - included with numpy - https://docs.scipy.org/doc/numpy-dev/f2py/

See the fortran documentation for instructions.