This is a dissertation.

The output can be found here: http://moorepants.github.com/dissertation

Copyright (c) Jason K. Moore 2012

The text and figures are licensed under the Creative Commons Attribution 3.0 Unported License and the source code is licensed under the BSD license (see LICENSE.txt).

First clone the dissertation repository:

$ git clone git://github.com/moorepants/dissertation.git

Install the Sphinx 1.1.3:

$ pip install sphinx==1.1.3

Install the latest pybtex development version and merge in two branches:

$ bzr branch lp:pybtex
$ cd pybtex
$ bzr merge lp:~matthias-troffaes/pybtex/label-alpha
$ bzr commit -m "Merged label-alpha."
$ bzr merge lp:~matthias-troffaes/pybtex/sorting-bugfix
$ bzr commit -m "Merged sorting-bugfix."
$ python setup.py install
$ cd ..

Install the development version of Matthias Troffaes's bibtex Sphinx extension:

$ git clone git://github.com/mcmtroffaes/sphinxcontrib-bibtex.git
$ cd sphinxcontrib-bibtex
$ python setup.py install
$ cd ..

The easy method to obtain all of the figures and tables is to download the data file with the pre-generated figures and tables (see below for instructions on generating the figures and tables from source). This should be extracted into the dissertation directory.:

$ cd dissertation
$ wget http://mae.ucdavis.edu/~biosport/jkm/dissertation/dissertation-data.tar.bz2 # 73 mb
$ tar -xjf dissertation-data.tar.bz2

Now build the website:

$ make html

And the pdf documents:

$ make webpdf # colored hyperlinks

or:

$ make printpdf # this one gives black text and inline urls

Open the website in your browser and the pdf in your pdf viewer:

$ firefox _build/html/index.html
$ evince _build/latex/HumanControlofaBicycle.pdf

• Sphinx 1.1.3 [web] [src] [docs]
• sphinxcontrib.bibtex dev [web] [src] [docs]
• pybtex dev [web] [src] [docs]

I've attempted to generate all of the figures and tables[1] in the dissertation from raw data and scripts. They can be generated by running all of the scripts in the src directory. These scripts require several Python, Matlab, and R based software.

• Python 2.7.3 [web] [src] [docs]
• setuptools
• NumPy 1.6.2 [web] [src] [docs]
• SciPy 0.10.1 [web] [src] [docs]
• PyTables 2.4.0 [web] [src] [docs]
• Matplotlib 1.1.1 [web] [src] [docs]
• IPython 0.13 [web] [src] [docs]
• Pandas 0.8.1 [web] [src] [docs]
• uncertainties 1.8 [web] [src] [docs]
• SymPy 0.7.1 [web] [src] [docs]

• Matlab 7.10.0.499 (R2010a)
• Matlab Simulink 7.5
• Matlab Control System Toolbox 8.5
• Matlab System Identification Toolbox 7.4

• R 2.14.1

These are all of the Python packages we've developed for the analysis in the dissertation. They are required to run many of the Python scripts in the src directory.

• AutolevToolKit [src]
• DynamicistToolKit [src]
• BicycleParameters [src]
• BicycleDataProcessor [src]
• CanonicalID [src]
• BicycleID [src]
• Yeadon [src]

These are the Matlab packages we've developed. They are required to run many of the m-files in the src directory.

• HumanControl [src]
• BicycleSystemID [src]

My workflow also includes these helpful tools but they are not necessary to build the dissertation.

• git (http://git-scm.com/)
• pip (http://pypi.python.org/pypi/pip): Useful for pulling releases from PyPi.
• virtualenv (http://pypi.python.org/pypi/virtualenv)
• Autolev 4.1 (http://www.autolev.com/): This software is no longer available, but can be used to process the .al scripts.

Install main Python dependencies to the system. It is easiest to simply use your system's package manager (e.g. apt-get) to install all of the dependencies. For example:

apt-get install git python-setuptools python-virtualenv python-numpy python-scipy ipython python-matplotlib python-pandas python-sphinx python-tables python-uncertainties

To install packages from source use pip or easy_install or download the source and use python setup.py install.

Now create a directory to house all of the development software.

$ mkdir bicycle-dissertation

Clone all of the python packages developed by us (if you use Github, you should fork the code in the web interface and then clone from your fork, for easy pull requests).

$ cd bicycle-dissertation
$ git clone git://github.com/moorepants/AutolevToolKit.git
$ git clone git://github.com/moorepants/DynamicistToolKit.git
$ git clone git://github.com/moorepants/BicycleParameters.git
$ git clone git://github.com/moorepants/BicycleDataProcessor.git
$ git clone git://github.com/moorepants/CanonicalID.git
$ git clone git://github.com/moorepants/BicycleID.git
$ git clone git://github.com/fitze/yeadon.git

I typically set up a virtual environment for the Python development workflow. I usually install the main Python dependencies to the system files as they are typically used by other virtual environments and outside virtual environments, but it is also possible to install them in the virtual environment which is a good idea if you need to pin the versions. The virtual environment is not required but is recommended if you are hacking on the development dependencies.

$ # --system-site-packages allows use of packages installed to the system
$ # (i.e. access to NumPy, SciPy, IPython, etc
$ virtualenv --system-site-packages bicycle

Activate the virtual environment.

$ source bicycle/bin/activate

Checkout the dissertation tag in each of the repositories. The dissertation tag pins the software used when version 1.0 of the dissertation was built, but future versions of each software package may work too (as long as I keep things backward compatible).

For example:

$ cd AutolevToolKit
$ git checkout dissertation

Then install:

python setup.py install

or:

python setup.py develop # do this if hacking on the code

$ cd ..

Repeat for all six Python repositories.

Clone the two repositories into the bicycle-dissertation directory.

$ git clone git://github.com/moorepants/HumanControl.git
$ git clone git://github.com/moorepants/BicycleSystemID.git

The scripts access several sets of data:

• Physical Parameters: The data directory in the BicycleParameters repository has the necessary data to load in the bicycles and riders.
• The Davis bicycle run database can be downloaded here http://mae.ucdavis.edu/~biosport/InstrumentedBicycleData/InstrumentedBicycleData.h5.bz2 [310mb]. To build from raw data files see the BicycleDataProcessor README.

This is an absolute mess so far. A lot of the scripts have explicit paths to the data files which are referenced to my file system. They will have to be manually changed to reflect the locations on the system you install to.

At this point the figures and tables can be generated by running all of the scripts in the src directory. The figures and tables are all stored in the figures and tables directory and most are auto-generated with the source code in the src directory and data stored in the data directory.

Footnotes