geotecha 0.2.2

geotecha: A software suite for geotechncial engineering

Writen by Dr Rohan Walker

geotecha is a a GPLv3-licensed Python package for geotechnical engineering.

Many components of geotecha derive from geotechnical research conducted by Dr. Rohan Walker, Prof. Buddhima Indraratna, and others at the University of Wollongong, NSW, Australia.

Primarily a repository of programs, tools, and code used by Dr. Rohan Walker, the content reflects his primary interest in soft soil consolidation with and without vertical drains. In particular:

• the speccon programs solve one-dimensional partial differential equations associated with multi-layer problems using the spectral Galerkin method. Material properties are constant with time but piecewise-linear with depth. Loads and boundary conditions are piecewise linear with time (plus a sinusoidal component).

• specbeam models a finite elastic Euler-Bernoulli beam resting on viscoelastic foundation subjected to a moving load(s), with piecewise-linear spatially varying properties. It uses the spectral Galerkin method to solve the dynamic system deflections over time.

• A number of other analytical solutions to soil consolidation problems are available in the consolidation sub-package.

Documentation

geotecha documentation is currently stored at https://rtrwalker.github.io . There you can find more descriptions of speccon and specbeam. Peruse the api docs for a listing of all the modules, classes and code (make sure you scroll up to the top of each api_doc page to see the summary listing of each module - frustratingly a clicked hyperlink doesn’t take you to the top of the page!)

GitHub Repository

The geotecha codebase is on GitHub, https://github.com/rtrwalker/geotecha here you will find the development version of the code set.

Installation

geotecha was developed on a windows platform. While I have tried to use cross platform programming idioms, I have (at the time of writing) made no attempt to build/use geotecha on any platform other than windows. That does not mean that it will not work. I simply do not know if it does or does not.

Requirements

geotecha uses a number of other python packages such as numpy, matplotlib, and scipy. Setting up your python environment to successfully run all these packages can be cumbersome so pre-built python stacks such as the readily available Anaconda are highly recommended . Note it is generally better to uninstall any existing python distributions before installing a new one.

Package requirements:

• numpy

• matplotlib

• scipy

• pandas

• sympy

• brewer2mpl

• testfixtures

• numpydoc

• IPython

• pkg_resources

• mpl_toolkits

• nose

• one of wx (i.e. wxPython) or PyQt (PyQt4 or PyQt5)

• sphinx

• mock

Before worrying about if your system has the required packages just try one of the installation methods below (first try the Windows binaries option). Hopefully you will already have all the packages or else the requirements section of the setup.py file will get them from pypi. However, issues can arise. Due to anomalies in handling dashes in required package names, the required packages pkg_resources and mpl_toolkits will not automatically be installed. Just install these using pip if they are not already present:

pip install pkg_resources
pip install mpl_toolkits

wxPython and PyQt are not always available through pypi so may have to be installed manually (there are usually windows binaries available).

Windows Binaries

The easiest, hassle-free way to install geotecha on a windows machine is to download one of the pre-built binaries available at https://pypi.python.org/pypi/geotecha . Once downloaded double click the .exe file to install. Note that the installer will display the raw text of this file and it may look odd. This does not matter. You may need to install the dependency packages separately. If your setup doesn’t match the binaries then you will have to try Building from source.

Test you installation by opening a command prompt (Windows+R cmd) and enter the following command:

nosetests geotecha -v --with-doctest --doctest-options=+ELLIPSIS --doctest-options=+IGNORE_EXCEPTION_DETAIL

It is common to get an error such as:

ImportError: No module named 'brewer2mpl'

which usually means one of the dependencies is not installed. Simply rerun the tests after installing the missing package with:

pip install brewer2mpl

pip

To install geotecha from the Python Package Index (PyPI) using pip:

pip install geotecha

This will essentially download the source files and build and install the package. geotecha has extension modules written in Fortran which can cause issues if your python environment is not set up to handle them ( I think you need a Fortran and a c compiler). .As such you may have difficultly in building the external extensions (see Building from source below.

See the Windows binaries section above for instructions on how to to test your geotecha installation.

Building from source

You can download the geotecha source files from pypi (release version) or from the Github repository (development version) https://github.com/rtrwalker/geotecha . geotecha uses some external extensions written in Fortran, so you will need to have a Fortran compiler present. Building from source on Windows can be troublesome at the best of times, so see the Issues with building/installing section below if you are trying to build on windows. For other systems it ‘should’ be as easy as:

python setup.py build
python setup.py install --record install.record

The “–record install.record” will make a file containing a list of all the files installed. It is possible to skip the build step (it will be included in the install step). But I find it more informative to use two steps.

See the Windows binaries section above for instructions on how to to test your geotecha installation. When testing you may wish to use the ‘-w’ working directory tag so that nose runs tests on the installed version of geotecha rather than the source code version (the source version will not have the external extensions). Change the working directory to match your python location, for example:

nosetests geotecha -v -w C:\Python36\Lib\site-packages\ --with-doctest --doctest-options=+ELLIPSIS

or if you are using an Anaconda env (like me) use something like:

nosetests geotecha -v -w C:\Anaconda3\envs\py36\Lib\site-packages\ --with-doctest --doctest-options=+ELLIPSIS --verbosity=3

You might get two test failures about importing ext_integrals and ext_epus. This indicates that the fortran extensions are not working. Don’t worry python/numpy (slower) versions of relevant functions will be used instead.

If you have a numpy version less than 1.14 then the tests will probably throw many failures associated with spaces and string representations of numpy arrays. This is due to changes in numpy https://docs.scipy.org/doc/numpy-1.14.0/release.html Don’t worry I’ve just updated things for python3.6, you will eventually upgrade and the test failures will disappear. Check which numpy version you have with :

python -c "import numpy; print(numpy.version.version)"

I have also had some odd behaviour where I run tests and get a couple of test failures. Then run the same tests and they all pass.

Building the docs

The geotecha docs can be built by running the following in the geotecha directory:

python setup.py build_sphinx --source-dir=docs/ --build-dir=docs/_build --all-files

The build requires a symlink to the examples directory. See the README.txt in the docs for instructions.

Issues with building/installing

At times (every time?) I have had issues with building from source on windows. So here are some hints to point you in the right direction. In python2.7 and up to python 3.4 it was relatively easy because there was a Mingwpy package ( https://anaconda.org/carlkl/mingwpy ), however, that very useful project has been abandoned ( https://groups.google.com/forum/#!topic/mingwpy/1k_BLFPLmBI ). So here is what works for me on Windows 10, 64 bit with python3.6

Based on the helpful blog post from Michael Hirsch ( https://www.scivision.co/python-windows-visual-c++-14-required/ ) install the relevant version of Microsoft Build Tools for Visual C++ (2017 for me) from https://visualstudio.microsoft.com/downloads/#build-tools-for-visual-studio-2017 (look in the “Tools for Visual Studio 2017” section). Note it is a large install taking up ~6GB.

Now install m2w64-toolchain hosted by Anaconda:

conda install -c msys2 m2w64-toolchain

Clean up previous builds:

python setup.py clean --all

Now try and build the thing explicitly specifying the compiler:

python setup.py build --compiler=mingw32
python setup.py install --record install.record

Test the install as above. No test failures will indicate that the Fortran extension modules have been successfully built and installed.

Removing geotecha

The cleanest method for removing geotecha is simply to use pip:

pip uninstall geotecha

You can also manually delete all files in the ‘install.record’ file.

Setting up an Anaconda env on Windows

After downloading and installing Anaconda make sure “C:Anaconda3Scripts” is in your PATH environment variable (otherwise conda command will not be found). Open the Anaconda prompt (start menu). Create a full anaconda env named py36 with a specified python version using (note it will download large files):

conda create -n py36 python=3.6 anaconda

If you need to start again remove the env with:

conda env remove --name py36

Close the anaconda prompt and then open the py36 anaconda prompt (start menu). Your py36 env is now ready to install geotecha and other python packages.