[CFP] Scipy 2018 - Scientific Computing with Python conference


SciPy 2018, the 17th annual Scientific Computing with Python conference, will be held July 9-15 in Austin, Texas. The annual SciPy Conference brings together over 700 participants from industry, academia, and government to showcase their latest projects, learn from skilled users and developers, and collaborate on code development. The Proceedings of the Python in Science Conference, the journal covering SciPy submissions, is also available to publish submissions from academic contributors. The call for abstracts for SciPy 2018 for talks, posters, and tutorials is now open.

Mini Symposia

We would also like to call special attention to a symposium within the conference.

Political & Social Sciences

Python is increasingly used to conduct fundamental research in political science, geography, psychology, sociology, and urban planning. Individuals or teams interested in presenting on their use of scientific tools in Python in projects in these domains (and are interested in honing their skills through training and involvement in tutorias) are especially encouraged to submit to this symposia on computational social science at SciPy 2018.

Talks & Posters (July 11-13, 2018)
In addition to the Political & Social Sciences track and general talks, this year will have specialized tracks focused on:

Data Visualization
Reproducibility and Software Sustainability
Biology and Bioinformatics
Data Science
Earth, Ocean and Geoscience
Image Processing
Language Interoperability
Library Science and Digital Humanities
Machine Learning
Materials Science
SciPy Tools plenary session

Tutorials (July 9-10, 2018)
Tutorials should be focused on covering a well-defined topic in a hands-on manner. We are looking for awesome techniques or packages, helping new or advanced python programmers develop better or faster scientific applications. We encourage submissions to be designed to allow at least 50% of the time for hands-on exercises, even if this means the subject matter needs to be limited. Tutorials will be 4 hours in duration. In your tutorial application, you can indicate what prerequisite skills and knowledge will be needed for your tutorial, and the approximate expected level of knowledge of your students (i.e. beginner, intermediate, advanced). Instructors of accepted tutorials will receive a stipend.
For examples of content and format, you can refer to tutorials from past SciPy tutorial sessions (SciPy2017, SciPy2016)

[Conference Website]
[Submission Website]

Important Dates
February 9, 2018 Submission deadline
March 20, 2018 Tutorial presenters notified of acceptance
April 2, 2018 Conference speakers and poster presenters notified of acceptance
May 14, 2018 First draft of Proceedings Due
July 9-10, 2018 SciPy 2018 Tutorials
July 11-13, 2018 SciPy 2018 Conference
July 14-15, 2018 SciPy 2018 Sprints


Thanks for advertising, @ljwolf!

I think it would be fabulous to showcase PyStan at SciPy 2018 and garner interest from the SciPy community (my favourite).

@ariddell @ahartikainen How about submitting a talk (https://scipy2018.scipy.org/ehome/299527/648140/)?

Or a tutorial (https://scipy2018.scipy.org/ehome/299527/648139/)? The latter is probably more work… I’m happy to team up and contribute my Carpentry teaching expertise (https://carpentries.github.io/instructor-training/). On a note related to @andrewgelman’s point about volunteering to teach tutorials (Stan Meeting 2018-01-18), instructors receive a $1,000 stipend.


I think this is a great idea. Is there an existing RStan talk/tutorial
which would, if translated into PyStan, be a good fit for the venue?

Is there any carpentry module that talks about statistical computing in
Python? That also would be a great thing to have.



I think this is a great idea. Is there an existing RStan talk/tutorial
which would, if translated into PyStan, be a good fit for the venue?

I think that the “Intro to Stan” tutorial which @jonah and @seantalts gave at StanCon 2018 could serve as a basis (materials: http://mc-stan.org/workshops/stancon2018_intro/).

Is there any carpentry module that talks about statistical computing in
Python? That also would be a great thing to have.

This touches on another conversation which I wanted to spark, as I hinted at the latest Stan Meeting (2018-01-18), following this take by Carpentry founder Greg Wilson: https://software-carpentry.org/blog/2018/01/teaching-statistics-in-the-21st-century.html

Contributing a lesson on statistical modeling would be great, I think (process: https://software-carpentry.org/lessons/incubation/). It’s also a great effort and commitment, so I wonder how this could tie in with @Bob_Carpenter’s development of a Coursera course…

But maybe we should start a new thread and keep this one only for things related to the SciPy conference.


Sounds great. I’m biased toward some spatial model. It’s an area with a lot of potential users.

Even more if we acknowledge the fact that mpi / gpu / sparse matrices are on their way.


Great, good to hear. Let me know if I can help advise on the submission. I’m a pystan user myself, & I’m sure there’d be interested attendees for a tutorial (or, at minimum, a talk).


Thanks for the link to Greg Wilson’s post. I’m very interested in a lesson featuring Stan. Let’s definitely start a separate topic on that.

As for Scipy. I’d love to see a PyStan talk there. I’m happy to help someone prepare the talk (if such a person needs help). I’m sure others would be glad to help as well. My summer schedule is a little busy, otherwise I’d volunteer to go.


Could you please point me to an example that you deem a good starting point (script, notebook, blog post, or long-form documentation)?

Even more if we acknowledge the fact that mpi / gpu / sparse matrices are on their way.



I really like the recent car/icar model by @mitzimorris



Oh, yes, I was impressed when @mitzimorris presented these maps at StanCon!
Thanks, @ahartikainen. Is the R code involved readily translatable into Python?


I’ve just updated that case study. Turns out there’s a better way to implement a sum-to-zero constraint on the ICAR prior which speeds up the model fit.

As for visualizations and maps, the visualizations in the new version are seriously slick. I spent the weekend figuring out how to overlay all my plots on Google Maps.

In R all these visualizations are done leaning on ggplot2 and ggmap. What are the Python equivalents?


Either matplotlib + basemap + gmaps

Or even better

(Bokeh + Holoviews) GeoViews


Hi everyone,

I have drafted an outline with @kon: https://docs.google.com/document/d/1Odxcu1Uk1cE8Vjxdu8tyDhFqsWF57Ml8g9KJNLvcCwU/

@ahartikainen @ariddell @jonah @seantalts @mitzimorris Please check it out and comment, should you have anything to share!

We are currently writing up the corresponding abstract and long description (as required for the submission); we’ll add them to the same document shortly.

Thank you,


Looks good.

Ps. Is that build from some template? E.g. comment about parallel stuff is for CmdStan, PyStan has parallel chains implemented.


No, it’s built from my reading, listening, and (approximate) memories… I’ll delete this line then (I thought I had heard this somewhere, but I got mixed up). Thanks for taking a look, @ahartikainen!


Sounds good, but I think the timing might be a bit tight! For example, Ben’s materials took him 15 minutes, and the workflow stuff probably took at least the same (though there were lots of questions). @kon’s case study took him 15 minutes to present as well. I might stick to one illustration with Stan, and maybe something a little easier to understand than either of those examples. If you want to do a hierarchical model to show that off, I might choose the radon case study since there’s a great writeup in Python on it: http://mc-stan.org/users/documentation/case-studies/radon.html


Thank you so much for your feedback, @seantalts! We tend to cram too much into a presentation, so your word of caution makes sense. To clarify, by referring to other materials, I never intended to sum up their entirety into 2 or 3 minutes. For example, we would only cover slide 2 out of your 24 slides; and maybe the equivalent of half a slide from Ben’s materials… Not that it’s purely quantitative either. It takes a clear narrative, so thanks for reminding us to focus on one well-identified, easy to understand example. Thanks for pointing to the Radon case study in Python–studying it right now.

Update: @kon and I have added the ‘short description’ of the talk to the collaborative document.


@kon and I have updated the collaborative document with the title, short and long descriptions we are ready to submit. Please proofread and comment!


Looks good.