It's been quite a while since I updated the MESA SDK, and even longer since I made any significant new additions to mad star. Hopefully, today marks the end of this extended hiatus — I've release a new version (20120727) of the SDK with extensive improvements, and my plan is to get back to (at least) weekly updates of the website.
After a fair bit of feedback from users, I've updated the MESA SDK to support 32-bit Linux. This means that the most commonly-used platforms are all now covered.
Also, mad star is now the primary distribution site for the SDK. This is because the update cycles of MESA and the SDK are driven by different considerations, and it doesn't make too much sense to bundle the two together.
Last month, I was out in California participating in the Asteroseismology in the Space Age program at the Kavli Institute for Theoretical Physics (KITP). It was a relaxing yet productive three weeks for me; I presented a blackboard lunch talk, finished the adiabatic and nonadibatic versions of my new GYRE pulsation code (more of that later); and made many new friends.
I especially enjoyed my first face-to-face meeting with Bill Paxton, the author of the EZ stellar evolution code which forms the backend to EZ-Web. As a follow-on to EZ, Bill has been working on MESA, a new stellar evolution code which takes a modular approach to the governing physics. MESA avoids the many simplistic approximations adopted in EZ and EZ-Web (users, please take note!), and is therefore suitable for doing 'real' science.
MESA is proving to be very popular. However, it can be difficult to get up and running, due to incompatibilities and bugs in compilers and libraries. To help overcome these obstacles, I've put together a unified software development kit (SDK) which contains compilers and libraries known to compile MESA correctly. The SDK is now distributed as a standard part of MESA; instructions for its use can be found here (link now fixed). If you're already a MESA user, give the SDK a whirl; if not, download MESA already!
Inspired by a word cloud posted on Jay Gallagher's door, I've been playing around with IBM's Java-based word-cloud generator. The image above comes from feeding the titles and abstracts of all of my publications (harvested from ADS) into the generator. I'm ultimately hoping to automate the process, so that as my publications evolve, so will the cloud.
![Slurm: the delicious beverage made from [censored]](http://user.astro.wisc.edu/~townsend/resource/news/slurm.jpg)
Over the Winter break, I rebuilt tedesco — the server machine that, amongst other duties, handles all of the calculation requests associated with EZ-Web and Poly-Web. This was needed to get rid of a lot of cruft that had accumulated over the years, and to make the system more fault tolerant by storing important data on a RAID1 array.
However, during this process the law of unintended consequences bit me hard. Previously, I had been using the excellent and free Sun Grid Engine (SGE) software to handle the queuing and scheduling of calculation requests. However, when I came to reinstall SGE, I discovered that Oracle (who now own Sun) have turned it into a commercial product, and to continue using it I would have to pay. Kudos to Oracle for alienating your user base so quickly (and this isn't the only dumb thing they've done recently; see here and here). You utter plonkers.
Fortunately, I found a simple SGE replacement in SLURM (Simple LinUx Resource Management), an open-source queue/scheduling system created by the clever folks at Lawrence Livermore. After a few hours spent installing SLURM on tedesco, I'm pleased to announce that EZ-Web is back up, and Poly-Web should follow soon.