The week is coming to an end and unfortunately so is #pyastro15. If you were here, you could see that the atmosphere at the Lorentz Centre has changed over the past days. It is kind of obvious that we see the end of the workshop and we want to get things done. That means more hands-on sessions. More coding less talk. That does not mean that we did not have our daily share of talks, lightning talks, tutorials and discussions. There were loads of that, but everybody here wants to go home knowing that we definitively contributed to improve some packages and that we can move forward towards a more powerful and organise community open to astronomer/astrophysicist that uses Python.
The day started with Steve Crawford on reducing optical observations with Python. I would consider his talk challenging from the point of view that it was the first one after a conference dinner. I think we all understand what that means. The talk was fantastic, and two Python packages were introduced to us, ccdproc (https://github.com/astropy/ccdproc) and a #pyastro15 baby package called specreduce (https://github.com/crawfordsm/specreduce). As pointed out by Steve, IRAF is great and many of us have used it on our PhD thesis. We should be critical with IRAF but not attack it. Once that is said, this does not mean that we need to stop with IRAF if we think we can get more using new tools. This is what is intended with these two packages.
The second talk of the day was by Anne Archibald on testing statistical tests. As a user more than a pure coder, Anne talked about the importance of knowing your statistical test if you want to interpret your data right. You can read more about some topics of Anne’s talk on at http://lighthouseinthesky.blogspot.nl. For me it was great to see how relevant PAST (my software for timing analysis), could be to Anne’s analysis in particular, and also the possibility of adding part of it into astropy.stats. So there it goes more motivation to finish it ASAP.
We have two more talks today, one by Nadia Dencheva on GWCS which it is a very relevant work on WCS using Python (sorry, I can’t find the link to the software) and SunPy (http://sunpy.org) by Stuart Mumford. SunPy makes the life of a solar physicist much easier by doing an enormous job on simplifying currently used code and bringing additional tools. I was very impressed on how well optimised is and if Solar Physics is your are I would really encourage you to start working with it right now.
As always, the lightning talks were fantastic, very dynamic and really useful to discover new things. For those not here, the idea is that in one hour, more or less, we have about 10 talks. So you can image how difficult it is to keep track with everything specially after 3 days of lightning talks. However, at the same time, there is no time to get bored or sleepy. Thus, please note I will be missing a few of them in my summary, and they will be biased towards relevance to my own interests.
Matt Craig showed us how to work with ipython widgets (an example can be found at (http://nbviewer.ipython.org/gist/mwcraig/2888d0a3790a81bd3391). That was really interesting and I will definitively start using them for my lectures. Becky Smethrust talked about the starpy package to study star formation history (https://github.com/zooniverse/starpy). She also gave us an update on the next two .Astronomy conferences (http://dotastronomy.com) that will be hosted in Sidney (November 2015) and Oxford (June 2016). I’ll try to go to the Oxford one considering how close it will be to Dublin. We then learned about CosmoABC by Emille Ishida (http://cosmoabc.readthedocs.org/en/latest/), a cosmology tool using Bayesian statistics. We also heard about Toyz by Fred Moolekamp (http://fred3m.github.io/toyz/) a really interesting web tool for “big data” where scientist can interact with their data in many different ways. If you didn’t know about this package, go and play with it, you will love it.
Curtis McCully introduced lacosmicx (https://github.com/cmccully/lacosmicx) a tool to clean optical images from cosmic rays. This software is really fast compared to previous tools available in other languages and as Curtis and I (mostly him actually, but I was there) tested later around 45 times faster than similar tools available in Python. Following on wrappers to create nicer plots from the previous session, Thomas Robitaille, a.k.a. astrofrog) showed us the capabilities of APLpy (https://aplpy.github.io). I will be playing with this package in the next few days so I might actually talk about it in future post. I think I am going to like it a lot. Finally there were a couple of “rants”, one on education by Caroline Villforth and another one in “deluge of data” by Geert Barentsen (https://speakerdeck.com/barentsen/a-3-minute-rant-on-the-deluge-of-data-in-astronomy).
As I said, the lightning talks may be short, but they do give you loads of information.
In the afternoon we had much more unconference sessions than in previous days. People gathered in smaller groups and try to work on different things. There were talks about astropy.model, extragalactic software… Among the things done I would like to mention the tutorials on packaging (continuation from the previous day) and on git branching (http://t.co/ozZnq2Is4t) by Erik Bray, which were fantastic. Also, the great job done by Kelle Cruz and Pauline Barmy putting together resources for astro-computing education (http://www.astrobetter.com/wiki/astro-comp-ed). I would like to share also the tutorial on classes that Erik Bray kindly shared via twitter in case we could not have a session for it http://nbviewer.ipython.org/github/pylunch/tutorials/blob/master/embray/classes.ipynb.
This is likely the longest of the blog posts about the conference and it was probably one of the busiest days too. I go to the last day with mix feelings wanting the workshop to continue a few more days. I am very happy to be here and to have taken part on this amazing workshop with this incredible group of people.