Navigation

Fall 2016 NANOGrav Meeting

Hackathon

You can be part of this exciting era of discovery by participating in one of the hackathons that will take place at NCSA on October 17-18. This event is part of the NANOGrav 2016 Meeting that the Relativity Group at NCSA is hosting during the week of October 17-21, 2016. The theme of these hackathons is related to the development and optimization of software used to find gravitational wave sources. No background in Physics or Astronomy is needed. The only selection criterion for you to participate is to have experience coding with Python, and interest on playing with CPUs and GPUs. If you are interested in participating in the Hackathon, please email Eliu Huerta (elihu@illinois.edu) and include your C.V. or resume.

Pulsar Timing Arrays (PTAs) search for temporal correlations across decades worth of data, and between pulsars widely separated across the sky. The correlation between pulsars is the smoking-gun for Gravitational Waves (GWs). Without approximations or linear-algebra speedups this is an intractable problem. Current codes rely on a rank-reduced signal description to accelerate the analysis. However, detection of nanohertz GWs will likely require arrays of between 50 and 100 pulsars, requiring further likelihood optimization.

Since the bottleneck is linear algebra, we want to investigate whether GPUs can cut down the likelihood evaluation time. Workshop attendees will investigate any performance gains possible with tuned linear algebra routines either on CPUs or GPUs.

Start playing with the codes now!

This git repo currently has the International Pulsar Timing Array Mock Data Challenge datasets, and the 5-year NANOGrav set for something realistic. There is a README file with installation instructions and an Ipython notebook demo.