September 2014

SciDAC-3 investigators gather to share research advances and highlights

Sudip Dosanjh’s talk about the National Energy Research Scientific Computing (NERSC) Center’s next-generation supercomputer was a working-lunch presentation at the SciDAC-3 Principal Investigators meeting held recently in Washington.

SciDAC-3 is the third cycle of the Scientific Discovery through Advanced Computing program. It’s administered by the Advanced Scientific Computing Research (ASCR) program in the DOE Office of Science in partnership with the office’s Basic Energy Sciences, Biological and Environmental Research, Fusion Energy Sciences, High Energy Physics, and Nuclear Physics programs.

Starting in 2011, four SciDAC Institutes – FASTMath, QUEST, SUPER and SDAV – were established, comprised of 24 participating organizations from DOE national laboratories, universities and industry. Institutes disseminate research advances in applied mathematics and computer science and develop high-performance computing tools and resources for scientific discovery through modeling and simulation. Each of 18 science partnership projects collaborate with one or more SciDAC Institutes to exploit high-performance computing and advance scientific frontiers in an area of strategic importance to the Office of Science.

The presentations at the SciDAC-3 PI meeting covered a range of topics from ion-modeling in biology and multiscale Earth-system modeling to first-principle simulations in nuclear physics and nanomaterials to uncertainty quantification in extreme-scale scientific computing.

Other meeting highlights included a session featuring 68 poster presentations from SciDAC Institutes and projects. The meeting concluded with a panel discussion on “Computer Science, Math, and Science: Harnessing the Troika.” The dialogue provided an opportunity for the SciDAC-3 community to share and learn from their personal experiences on effective interdisciplinary collaborations, accelerating technology transfer among various projects, and other matters at the heart of successful high-performance computational science research.