Proceedings of the 7th Python in Science Conference SciPy Conference – Pasadena, CA, August 19-24, 2008. Editors: Gaël Varoquaux, Travis Vaught, Jarrod Millman Proceedings of the 7th Python in Science Conference (SciPy 2008) Contents Editorial 3 G. Varoquaux, T. Vaught, J. Millman The State of SciPy 5 J. Millman, T. Vaught Exploring Network Structure, Dynamics, and Function using NetworkX 11 A. Hagberg, D. Schult, P. Swart Interval Arithmetic: Python Implementation and Applications 16 S. Taschini Experiences Using SciPy for Computer Vision Research 22 D. Eads, E. Rosten The SciPy Documentation Project (Technical Overview) 27 S. Van der Walt Matplotlib Solves the Riddle of the Sphinx 29 M. Droettboom The SciPy Documentation Project 33 J. Harrington Pysynphot: A Python Re-Implementation of a Legacy App in Astronomy 36 V. Laidler, P. Greenfield, I. Busko, R. Jedrzejewski How the Large Synoptic Survey Telescope (LSST) is using Python 39 R. Lupton Realtime Astronomical Time-series Classification and Broadcast Pipeline 42 D. Starr, J. Bloom, J. Brewer Analysis and Visualization of Multi-Scale Astrophysical Simulations Using Python and NumPy 46 M. Turk Mayavi: Making 3D Data Visualization Reusable 51 P. Ramachandran, G. Varoquaux Finite Element Modeling of Contact and Impact Problems Using Python 57 R. Krauss Circuitscape: A Tool for Landscape Ecology 62 V. Shah, B. McRae Summarizing Complexity in High Dimensional Spaces 66 K. Young Converting Python Functions to Dynamically Compiled C 70 I. Schnell 1 unPython: Converting Python Numerical Programs into C 73 R. Garg, J. Amaral The content of the articles of the Proceedings of the Python in Science Conference is copyrighted and owned by their original authors. For republication or other use of the material published, please contact the copyright owners to obtain permis- sion. 2 Proceedings of the 7th Python in Science Conference (SciPy 2008) Editorial Gael Varoquaux ([email protected])– Neurospin, CEA Saclay , Bât 145, 91191 Gif-sur-Yvette France Travis Vaught ([email protected])– Enthought, Austin TX USA Jarrod Millman ([email protected])– University of California Berkeley, Berkeley CA USA The Annual Scipy Conference began in 2002 with a of note were the inroads that Python has gained in gathering of some extremely bright (and admittedly many other contexts, ranging from ecology and neu- odd and passionate) folks. They gathered at the Cal- roimaging, to mechanical engineering and computer tech campus in Pasadena, California to discover and vision. As presented at the conference, scientists are share ideas about a compelling approach to scientific attracted to Python by the quality and ease of use computing. These pioneers had been using Python, a of the language itself, and the richness of the exist- dynamic language, to perform and drive their model- ing scientific packages. Another major presence at the ing, data exploration and scientific workflows. At that conference this year was the more abstract-oriented time, Python already had the advantage of a very fast research fields of mathematics or graph theory. The numerics library as well as the flexibility of being in- SAGE project has been very successful at building terpreted, procedural (if you wanted it to be), and, upon Python to give a consistent package for num- in a true academic spirit, permissively open source. ber theory, or computational mathematics in general. That gathering 7 years ago yielded many interesting More component-oriented mathematical projects such ideas, but more importantly, it crystalized a commu- as NetworkX, for graph theory, or pyinterval, for inter- nity and established relationships that persist today. val arithmetic, contribute to the stack from a very dif- From those inauspicious beginnings, the conference has ferent perspective than the usual array-manipulating now grown to a fully international meeting attracting numerical packages. Finally, a sizable fraction of the a variety of interests—yet still inspired by the propo- talks where focused on the current effort to improve sition of a humane language interface to fast compu- the available tools. On the numerical side, the focus tation, data analysis and visualization. is on speed, with some research in compiling part or This year marked the 7th edition of the conference; all of the Python code to fast machine code. We also however, it is the first edition for which proceedings note a strong effort on the documentation of the tools, are to be published. We are thrilled by this new devel- where there has been a tremendous amount of work opment. These proceedings are the sign of a maturing toward filling a historical gap in user docs. community not only of developers, but also of scientific The SciPy Conference has been supported since its users. Python’s use as a tool for producing scientific creation by the Center for Advanced Computing Re- results is, by now, well established. Its status as the search, at CalTech, and Enthought Inc. In addition, subject of research and academic publication is now this year we were delighted to receive funding from being recognized and we hope that the proceedings the Python Software Foundation which allowed us to of the SciPy Conference will help communicate how pay for travel, conference fees and accommodation for scientists and engineers are using and building upon 10 students. We are very grateful to Leah Jones, of various Python tools to solve their problems. Commu- Enthought, and Julie Ponce, of the CACR (Caltech), nication is indeed paramount to both the software and for their invaluable help in organizing the conference. the scientific community. A scientist needs access to The conference, like much of the software presented, the methods of others; he also needs to get academic was truly a global effort and we are hopeful that the credit for his work, which is often measured in publi- impact will continue to grow. cations and citations. A developer needs to be aware of the ongoing software efforts to reduce duplication of effort; he also needs to advertise his work to potential users. Organizers The variety of subjects covered in the talks, and in the Organizing Committee following proceedings, is striking because is shows the extent of the use made of these tools in heterogeneous Conference Chairs communities. This is the power of open source projects and a key factor to their success. Different actors are • Jarrod Millman, UC Berkeley (USA) able to contribute to the quality of the tools in ways unique to their field and interests, resulting in a whole • Travis Vaught, Enthought (USA) stack that is much greater than the sum of its parts. Tutorial Coordinators Among the scientific results presented this year, there is a strong presence of the astronomy community—a • Fernando Pérez, UC Berkeley (USA) community that has a long history in software en- gineering and is an early adopter of Python. Also • Travis Oliphant, Enthought (USA) 3 G. Varoquaux, T. Vaught, J. Millman: Proc. SciPy 2008, G. Varoquaux, T. Vaught, J. Millman (Eds), pp. 3–5 Editorial Web Masters • Charles Harris, Space Dynamics Laboratory (USA) • Gaël Varoquaux, Neurospin, CEA - Saclay (France) • Ryan Krauss, Southern Illinois University Ed- wardsville (USA) • Stéfan van der Walt, University of Stellenbosch (South Africa) • Stéfan van der Walt, University of Stellenbosch (South Africa) Program Committee Proceedings Reviewers Program Chair • Matthew Brett, MRC Cognition and Brain Sciences • Gaël Varoquaux, Neurospin, CEA - Saclay (France) Unit (UK) Program Members • Ryan Krauss, Southern Illinois University Ed- wardsville (USA) • Anne Archibald, McGill University (Canada) • Jarrow Millman, UC Berkeley (USA) • Matthew Brett, MRC Cognition and Brain Sciences Unit (UK) • Gaël Varoquaux, Neurospin, CEA - Saclay (France) • Perry Greenfield, Space Telescope Science Institute • Stéfan van der Walt, University of Stellenbosch (USA) (South Africa) http://conference.scipy.org/proceedings/SciPy2008/paper_0 4 Proceedings of the 7th Python in Science Conference (SciPy 2008) The State of SciPy Jarrod Millman ([email protected])– University of California Berkeley, Berkeley, CA USA Travis Vaught ([email protected])– Enthought, Austin, TX USA The annual SciPy conference provides a unique op- freely distributable for both academic and commercial portunity to reflect on the state of scientific pro- use. It has a simple, expressive, and accessible syntax. gramming in Python. In this paper, we will look It does not impose a single programming paradigm back on where we have been, discuss where we are, on scientists but allows one to code at many levels and ask where we are going as a community. of sophistication, including Matlab style procedural Given the numerous people, projects, packages, and programming familiar to many scientists. Python is mailing lists that make up the growing SciPy commu- available in an easily installable form for almost ev- nity, it can be difficult to keep track of all the disparate ery software platform, including Windows, Macintosh, developments. In fact, the annual SciPy conference is Linux, Solaris, FreeBSD and many others. It is there- one of the few events that brings together the commu- fore well suited to a heterogeneous computing environ- nity in a concerted manner. So it is, perhaps, appro- ment. Python is also powerful enough to manage the priate that we begin the conference proceedings with complexity of large applications, supporting functional a paper titled “The State of SciPy”. Our hope is we programming, object-oriented programming, generic can help provide the context for the many interesting programming, and metaprogramming. In contrast, and more specific papers to follow. We also aim to commercial languages like Matlab and IDL, which also promote a much more detailed discussion of the state support a simple syntax, do not scale well to many of the project, community, and software stack, which complex programming tasks. Lastly, Python offers will continue throughout the year. strong support for parallel computing. Because it is The last year has seen a large number of exciting de- freely available, and installed by default on most Unix velopments in our community.