SC10: International Conference on High Performance Computing

LSST is one of the premier examples of how high performance computing advances will expand our scientific boundaries. Salman Habib, Jeff Kantor and Stuart Marshall demonstrated these advances at SC10, Super Computing’s 23rd meeting in November 2010 where its theme, The Future of Discovery, showcased how high performance computing, networking, storage and analysis lead to advances in scientific discovery, research, education and commerce. IEEE Computer Society and Association for Computing Machinery (ACM) sponsored the meeting and over 11,000 attended. While much focus was on computing in climate studies, LSST collaborators impressed attendees with the plans for data management and the possibilities for discoveries well beyond our planet.

In 2007, the National Science Foundation pointed out that science and engineering research and education are becoming more and more data-intensive due to expanding digital technologies, instrumentation, and data networks. With the enormous growth of available scientific data, new opportunities arise. Sky surveys have already changed our understanding of our Solar System and the earliest Universe. LSST will scan the sky from Chile, capturing 30 trillion bytes of image data each day – a volume equal to two entire Sloan Digital Sky Surveys each day. Massive computing power is needed to use these data to extend the investigations of our Solar System, our Galaxy and the Universe.

In the “Big Science, Big Data” session, one of the SC10 Masterwork sessions, Jeff Kantor (LSST) and Salman Habib (Los Alamos National Laboratory (LANL)) shared the progress on data management and computational cosmology.

Computational cosmology – cosmological numerical simulation – plays a critical role in incorporating nonlinear aspects of structure and galaxy formation into a precision theoretical framework. Habib demonstrated the need for high-fidelity simulations of the visible Universe to understand the astrophysics underlying interpretations of the cosmological surveys. “The essential role of cosmological modeling and simulation is to extract scientific insight from large, complex observational datasets – such as the detailed maps of the sky to be revealed by LSST. “Petascale computing and beyond will be essential to this endeavor, which will employ some of the world’s largest computers,” Habib told participants.

LSST cyberinfrastructure supports “computing at a distance” to overcome the difficulties in processing all data for all users, Kantor told a large group of listeners. “The data management system (DMS) processes incoming images, produces transient alerts, archives over 50 petabytes of exposures, creates and archives astronomical objects, makes LSST data available without a proprietary period and facilitates analysis and production of user-defined data products with supercomputing resources.”

In addition to these two talks about the computing capabilities of LSST, Stuart Marshall, a member of the LSST camera team, was part of the SLAC exhibition at the conference. More information about the SLAC exhibit can be found at http://today.slac.stanford.edu/feature/2010/sc-2010.asp

For more about Kantor’s , Habib’s and Marshall’s efforts, see the LSST Science Book. Kantor contributed to Chapter 2, LSST System Design, and Habib to section, 15.5, Cosmological Simulations in Chapter 15, Cosmological Physics. Chapter 2 discusses the LSST camera.

Article written by Anna Spitz