After many years of observing the conference from a distance, I was able to travel to the SciPy conference. This year the conference hosted a track on computation tools and education that inspired many ideas for me.


All the keynotes and several of the talks focused on the use of scientific computing in the classroom. I left with several ideas both in terms of tools and pedagogy to experiment with. On the third day, Greg Wilson from Software Carpentry gave a keynote which framed many of the teaching issues beautifully.


It is very easy to promote a single tool I prefer over other tools. Greg Wilson's keynote noted that he uses popular tools as a trojan horse for a more abstract learning objective. For my own teaching, I'll be sure to start with learning objectives rather than focus on my favorite tools.

For example, many of my students will use spreadsheets early in their careers. What skills around structured data, legible models, and output presentation can I teach so that these students are creating spreadsheets that are as accurate and illuminating as possible? While I would rather focus on scripting tools over spreadsheet tools, it is irresponsible to let this bias guide my teaching. I do however want to start with scripting tools to teach the common math syntax and introduce functions so that students have the ability to move between Excel and other computational tools easily in their careers.

Massive Collaboration

My teaching materials from my energy classes are on github since publishing there is quicker than on my campus content management system. Another point Greg Wilson made was that massive scalable collaboration should also be brought to educational materials. This has inspired me not only to make materials available but to cultivate a community of innovation around my energy teaching materials.

Lowering Barriers

In order for modern computing tools to be useful in the classroom, their installation and use must be easy and not create distractions. I have had students use Wakari and SageMathCloud last year and am hoping the new Google Drive implementation of the IPython Notebook will be ready and robust for the fall. Since our students are issued Google accounts for their email, it is a very attractive solution from an ease of use standpoint.

This fall I want to again use Python or Julia REPLs as a stepping stone between calculators and scripting environments. I used the Python Anywhere IPython web REPL in a class last year. Unfortunately it is Python 2.7 and the configuration to perform proper division created confusion and consumed classroom time. There is a nice online REPL for Julia which is very attractive, but I'm reluctant to burden students with a syntax transition later in the semester when I move to python tools.


Interactive software demonstrations for students have been available for a long time. However, the details are usually obscured from the user by a GUI framework. The addition of lightweight interactivity in the IPython Notebook allow me to build an interactive experience for the student that also shows them how the model is constructed. Because the IPython interactivity tools are so simple to add, it is easy to imagine a novice student being able to extend these interactions. By doing this they deepen their understanding of both the model under study and the computational tools to study it. Lorena Barbara described this interaction beautifully as a conversation with the system under study.

Student Libraries

I liked the idea from the lightning talk to create a simple simulation library that is extensible by novices, particularly students. This can provide a stepping stone to their involvement in a more complex and complete model. I've been developing a simple solar energy model for research and teaching that I think will be useful for students.

Looking Forward

It was encouraging to see how much thought and effort is being put into not only the development of these tools but also the way that these tools can positively influence the way science is conducted. I had the pleasure of talking with the IPython team and it is clear that they are as interested in the scientific outcomes as the tools themselves. It is up to the scientific computing community to highlight which tools are so unique and powerful that they deserve resources for development and maintenance. The recent grants for the IPython team and Continuum demonstrate that open source scientific software are being seen as real tools rather than hobbies. I hope the community can build on these successes, cultivate a sustainable funding stream, and create tools that make the discovery, analysis, and communication of science better.