Tutorial Prerequisites¶
The PyMVPA tutorial assumes some basic knowledge about programming in Python. For a short self-assessment of your Python skills, please read the following questions. If you have an approximate answer to each of them, you can safely proceed to the tutorial. Otherwise, it is recommended that you take a look at the Python documentation resources listed under Recommended Reading and Viewing.
- Are you using spaces or tabs for indentation? Why is that important to know?
- What is the difference between
import numpy
andfrom numpy import *
? - What is the difference between a Python list and a tuple?
- What is the difference between a Python list and a Numpy
ndarray
? - What is the difference between an iterable and a generator in Python?
- What is a list comprehension?
- What is a callable?
- What are
*args
and**kwargs
usually used for? - When would you use
?
or??
in IPython? - What is the difference between a deep copy and a shallow copy?
- What is a derived class?
- Is it always a problem whenever a Python exception is raised?
If you could not answer many questions: Don’t panic! Python is known to be an easy-to-learn language. If you are already proficient in any other programming language, you can expect to be able to write fairly complex Python programs after a weekend of training.
What Do I Need To Get Python Running¶
PyMVPA code is compatible with Python 2.X series (more precisely >= 2.6). Python 3.x is supported as well, but not as widely used (yet), and many 3rd-party Python modules are still lacking Python 3 support. For now, we recommend Python 2.7 for production, but Python 2.6 should work equally well.
Any machine which has Python 2.X available can be used for PyMVPA-based processing (see chap_installation> on how to deploy PyMVPA on your system). Any GNU/Linux distribution already comes with Python by default. The Python website offers installers for Windows and MacOS X.
However, PyMVPA can make use of many additional software packages to enhance its functionality. Therefore it is preferable to use a Python distribution that offers a large variety of scientific Python packages. For Windows, Python(x,y) matches these requirements. For MacOS X, the MacPorts project offers a large variety of Python packages (including PyMVPA).
The ideal environment is, however, the Debian operating system. Debian offers the largest selection of free and open-source software in the world, and runs on almost any machine. Moreover, the NeuroDebian project provides Debian packages for a number of popular neuroscience software package, such as AFNI and FSL.
For those who just want to quickly try PyMVPA, or do not want to deal with installing multiple software package we recommend the NeuroDebian Virtual Machine. This is a virtual Debian installation that can be deployed on Linux, Windows, and MacOS X in a matter of minutes. It includes many Python packages, PyMVPA, and other neuroscience software (including AFNI and FSL).
Recommended Reading and Viewing¶
This section contains a recommended list of useful resources, ranging from basic Python programming to efficient scientific computing with Python.
Tutorial Introductions Into General Python Programming¶
http://en.wikibooks.org/wiki/Non-Programmer’s_Tutorial_for_Python_2.6
Basic from-scratch introduction into Python. This should give you the basics, even if you had no prior programming experience.
http://swaroopch.com/notes/python/
From the author:
The aim is that if all you know about computers is how to save text files, then you can learn Python from this book. If you have previous programming experience, then you can also learn Python from this book.We recommend reading the PDF version that is a lot better formatted.
A famous tutorial that served as the entry-point into Python for many people. However, it has a relatively steep learning curve, and also covers various topics which aren’t in the focus of scientific computing.
http://docs.python.org/tutorial/
Written by the creator of Python itself, this is a more comprehensive, but also more compressed tutorial that can serve as a reference. Recommended as resource for people with basic programming experience in some language.
Scientific Computing In Python¶
Python itself is “just” a generic programming language. To employ Python for scientific computing, where a common analysis deals with vast amounts of numerical data, more specialized tools are needed – and are provided by the NumPy package. PyMVPA makes extensive use of NumPy data structures and functions, therefore we recommend you to get familiar with it.
http://www.scipy.org/Tentative_NumPy_Tutorial
Useful for a first glimpse at NumPy – the basis for scientific computing in Python.
http://mathesaurus.sourceforge.net/
Valuable resource for people coming from other languages and environments, such as Matlab. This pages offers cheat sheets with the equivalents of commands and expressions in various languages, including Matlab, R and Python.
http://www.tramy.us/numpybook.pdf
This is the comprehensive reference manual of the NumPy package. It gives answers to questions, yet to be asked.
Interactive Python Shell¶
To make interactive use of Python more enjoyable and productive, we suggest to explore an enhanced interactive environment for Python – IPython_.
http://showmedo.com/videotutorials/series?name=CnluURUTV
Video tutorials from Jeff Rush walking you through basic and advanced features of IPython. While doing that he also exposes basic constructs of Python, so you might like to watch this video whenever you already have basic programming experience with any programming language.
http://ipython.org/documentation.html
IPython documentation page which references additional materials, such as the main IPython documentation which extensively covers features of IPython.
http://ipython.org/notebook.html
IPython notebook provides an interactive programming environment right in your browser. Our tutorial is available not only as static web-pages but also as IPython notebooks, thus making learning and tinkering with code convenient and fun. Familiarizing with principles of ipython notebooks and keyboard shortcuts would be beneficial.
Multivariate Analysis of Neuroimaging Data¶
There is a constantly growing number of interesting articles related to the field – visit References for an extended but not exhaustive list of related publications. For a quick introduction into the topic read Pereira et. al. 2009. For the generic reference on machine learning methods we would recommend a great text book The Elements of Statistical Learning: Data Mining, Inference, and Prediction by Trevor Hastie, Robert Tibshirani, and Jerome Friedman , PDF of which was generously made available online free of charge. For an overview of recent advances in computational approaches for modeling and decoding of stimulus and cognitive spaces we recommend video recordings from Neural Computation 2011 Workshop at Dartmouth College.