| News |
| Perspectives and funding |
| SciDAC-2: on the path to petascale computing |
| Dr Walter M. Polansky |
|
The driving force behind the initiation of the SciDAC program was a growing perception that modeling and simulation can advance the
understanding of scientific challenges. Through SciDAC, perception became reality. Researchers developed state-of-the-art simulation
codes to run on DOE supercomputers and gain substantial insight into the areas of accelerator design, chemistry, combustion, climate, and fusion.
|
|
The key to the overall success of SciDAC has been the unleashing
of the power of multidisciplinary teams; experts in the relevant scientific
discipline, computer scientists, and applied mathematicians have come together. The teams
were able to chart courses for progress that could not be traversed in any other way, and forge
a new paradigm for simulation science. In addition, SciDAC built a unique foundation in applied
mathematics, computer science and high-performance computing technology that could lead to further
scientific advances.
|
|
The research landscape is currently being shaped in two areas, with profound implications. One
area is enabled by the increasing application of modeling and simulation to long-standing scientific
problems. The second area is driven by the rapid adoption and adaptation of advanced technologies
for data storage and for data transfer associated with largescale experimental facilities or
massive data sets.
|
|
The SciDAC program is in a position to help shape that research landscape. Scientific
application partnerships could be enriched in existing areas and introduced to other disciplines.
The need for scientific data storage and data transmission has outstripped current methodologies;
innovative approaches could emerge from research partnerships that marry hardware with the highly
specialized needs of scientific users. The partnership concept could reach out to engage
a broader community of scientists in the activities of scientific discovery through advanced computation.
|
|
Finally, mechanisms could be established to bring together a "critical mass" of experts from
multiple disciplines to tackle problems as they emerge within a discipline.
|
SciDAC-2 has been formulated to capitalize on these opportunities.
As envisioned, SciDAC-2 will be an ensemble of research partnerships
whose activities and results will be coordinated to leverage
advances and optimize overall progress along two major lines:
- The creation of a comprehensive, scientific computing software
infrastructure that fully integrates applied mathematics, computer
science, and computational science in the physical, biological,
and environmental sciences for scientific discovery at the petascale.
- A new generation of data management and knowledge discovery
tools for data-intensive science, including data sets obtained from
experimental facilities and from high-end simulations.
|
|
SciDAC-2 seeks to advance scientific discovery through computing in 13 areas of science:
accelerator science and simulation, astrophysics, climate modeling and simulation, computational biology,
fusion science, groundwater reactive transport modeling and simulation, high-energy physics,
high-energy and nuclear physics with petabytes, materials science and chemistry, nuclear physics,
quantum chromodynamics, radiation transport, and turbulence.
|
|
Advances through simulations in these 13 topical science areas will be provided via enabling computational
technologies, which have been structured into the following three categories:
|
- The Centers for Enabling Technologies, which will address mathematical and computing systems
software issues.
- The Scientific Applications Partnerships, which seek to fund activities that form a partnership
between computational mathematics and computer science with a science application domain.
- SciDAC Institutes, which are university-led centers of excellence intended to complement the
Centers for Enabling Technology and the Scientific Applications Partnerships, as well as centers
that may be formed under specific science application domains. A key feature of a SciDAC Institute
will be a dimension of training and outreach in high-performance computing topics, including
topics for graduate students and post-doctoral staff.
|
|
The Notice inviting grant applications
(www.sc.doe.gov/grants/FAPN06-04.html) and the Program
Announcement to DOE
National Laboratories (www.sc.doe.gov/grants/LAB06_04.html)
were issued on December 23, 2005. The Notice and the Program Announcement describe the scientific
challenges. The research community will provide the ideas, through proposals and grant applications, for meeting those
challenges. SciDAC-2 awards will be made in the summer of 2006 following a thorough peer review.
|
|
Under the SciDAC-2 calls for proposals, letters of intent were due on January 23, 2006. Proposals
and grant applications are due on March 6, 2006. The overall funding available for SciDAC-2 is
approximately $70,000,000 this fiscal year: $66,000,000 from the Office of Science (SC) and
$4,000,000 from the Department's National Nuclear Security Administration.
All SC programs are participating in SciDAC-2: these are Advanced Scientific Computing Research; Basic Energy Sciences,
Biological and Environmental Research; Fusion Energy Sciences, High Energy Physics; and Nuclear Physics. In addition, the National
Science Foundation is a potential funding source for grant applications in the areas of climate modeling, high-energy physics, and nuclear physics.
|
|
Collaborative research proposals are not only encouraged, they are expected. The majority of the
applications and proposals submitted are expected to be for scientific simulation teams with several
institutions. The number, size and duration of each will be determined by the quality and content
of the proposals selected for funding.
However, to meet the goals of SciDAC-2, a relatively small number of awards is expected.
SciDAC-2 represents a commitment to propel simulation science to the petascale and to broaden the
prospects for scientific advancement.
|
|
The scientific challenge problems are there to be solved and petascale computers are on
the horizon. The final terms in the SciDAC-2 "equation for success" represent the dedicated research
scientists, with outstanding credentials, ready to partner with colleagues in other disciplines to
bring about scientific discovery through simulation and modeling.
|
|
Dr Walter M. Polansky is Senior Technical Advisor for Project Management at the Office of Advanced Scientific Computing
Research, Office of Science, DOE.
|
|
|
