BIG DATA

The SCALE 2009 competition, which took place in Shanghai, China, involved researchers demonstrating real-world problem solving using scalable computing, in which scientists use computer systems that can easily adapt, or scale up, to provide greater performance and computing power and give them greater capability to solve complex problems.

The CCT-led demonstration showcased a scalable, interactive system to simulate and visualize black holes to study the physics of gravitational waves. This complex process involves many challenges that scientists are only now able to address with modern cyberinfrastructure, including scalable computing systems.

“We were honored to receive first prize in this competition, particularly since we had many challenges leading up to our demonstration,” said LSU Department of Computer Science and CCT Professor Gabrielle Allen, who led the demonstration team. “Travel restrictions prevented all but one member of our team from attending the competition in person, so we produced a video of our demonstration that was shown in Shanghai. The rest of our team communicated from LSU during the competition using Skype and live chat to describe the process and answer questions. The fact LSU’s entry won first prize in spite of these obstacles is a good reflection on the quality of scientific research our group was able to demonstrate, and we are grateful for this recognition from our international peers and collaborators.”

The CCT entry addressed the scalable computing challenges of the competition, including automatically generating simulation code, developing programs and software  components to provide fast data transfer across the Louisiana Optical Network Initiative, or LONI, parallelize the rendering process that transforms scientific data into images and building interactive, tangible devices that allow observers to engage directly with the scientific data as it is visualized live.

The CCT demonstration also tested the team’s ability to effectively use high-performance computing machines concurrently, running applications on thousands of computing cores at once while using multiple, distributed resources of different types(computation, storage, networks, graphics) for a single application.

Numerical simulations are the only practical way to study black hole systems, but this requires a complex system of mathematical equations describing effects that span a wide range of length- and time-scales. To address this challenge, the CCT demonstration used Cactus Software Framework, an open-source environment that allows teams of researchers from different fields in different locations to work together at modeling the black hole collisions, solving Einstein’s Equations. Cactus allowed the researchers to automate a process that would be too time-consuming and error-prone for scientists to perform by hand.

The demonstration involved collaboration with LONI and the Texas Advanced Computing Center in Austin, where the black hole simulation was run on 2,048 cores of the Ranger machine.

The CCT team was able to demonstrate live interaction with the simulation using a Web interface for application-level monitoring, debugging, and profiling. The simulation integrated social networking sites into the scientific process, using a new Cactus application an LSU undergraduate student co-developed to announce runtime information to Twitter and provide real-time images of the gravitational field through Flickr.

The demonstration showed live, interactive images of the black hole data using a scientific visualization system distributed across LONI. The CCT group built tangible interaction devices, which they provided on the show floor in Shanghai, allowing observers to interact in real-time with the visualization process.

The team came in first among the five international, short-listed entries for the competition.

“It was a great experience for us to be able to demonstrate University research to our international peers at a venue across the world, so we can show them the groundbreaking processes taking place through work at LSU that are advancing opportunities for the global scientific community,” said CCT Interim Director Stephen David Beck, Ph.D.

To see the video of the CCT demonstration, please visit http://preview.cactuscode.org/media/videos/ .

For more information on this and other CCT research projects, please visit www.cct.lsu.edu.