| News |
| Analysis |
| A strategic view of SciDAC-2: growth ahead |
| Dr Horst Simon and Dr Rick Stevens |
|
During During the past five years, the SciDAC-1 program from the the US Department of Energy (DOE) has shown that important scientific
accomplishments are possible through simulation and modeling with the focused collaboration and active partnership of domain scientists,
applied mathematicians, and computer scientists.
|
Climate modeling is just one of the many areas that SciDAC-2 will focus on.
|
|
Successes have been seen in fields such as accelerator design, chemistry, combustion, climate modeling, and fusion. SciDAC
activities have demonstrated that large-scale simulation offers a cost effective way of answering a number of scientific
questions on the fundamental structure of matter, the production of heavy elements in supernovae, and the functions
of enzymes, for instance.
|
|
The approach to scientific enterprise in the US is in the midst of two fundamental transformations.
One is the increasing application of computational simulation to long-standing challenge problems.
The other is the rapid adoption of advanced infrastructure for the capture, storage, transmission,
sharing, and analysis of large-scale experimental data.
|
|
Together, these two technologies are dramatically improving our nation's ability to solve important
engineering problems and make critical discoveries in many scientific domains. Such advances were
considered intractable in the past because of their extreme complexity, multidisciplinary nature, or
lack of data-analysis capability.
|
|
In spite of tremendous progress, however, much remains to be done, and even more compelling
opportunities for scientific discovery lie ahead. In FY 2004, the DOE's Office of Science (SC) launched an
aggressive program to develop and deploy leadership-class computing facilities and announced a 20-year
scientific facilities roadmap that will provide a rich scientific infrastructure for the next two decades.
|
|
Each of the SC programs - Basic Energy Sciences, Biological and Environment Research, Fusion
Energy Sciences, High-Energy Physics, and Nuclear Physics - has identified a crucial need for sustained
advances in scientific computing.
|
The creation of additional SciDAC-like scientific application partnerships is required for fulfillment
of these programs' missions. To address these needs, it was recommended that the SC's Office
of Advanced Scientific Computing Research (ASCR) make significant investments for FY 2007 to establish
a SciDAC-2 program. SciDAC-2 can build on the success of SciDAC-1 and extend the program in three important ways:
- Strengthen the scientific application partnerships that formed the core of the successful SciDAC-1 program.
The major source of acceleration in simulation-based science has been the strength and depth of partnerships among
application domains, computer science, and applied mathematics.
- Involve experimental science. The use of advanced computing technologies to accelerate scientific
discovery is not limited to modeling and simulation; it can also be used to improve experimental
science. New data-management and analysis tools are required for both small and largescale
experiments such as those that will be carried out in the facilities outlined in the SC's 20-year roadmap.
SciDAC-2 should improve the productivity and capability of experimental science through the development and
application of advanced data and analysis capabilities, and computation and technologies that support
and automate experiments.
- Expand to new scientific communities. Some areas of science do not yet fully benefit from largescale
computing; some communities are not fully aware of successes and methods from other scientific
areas; and in other cases methods, tools, or conceptual approaches are relatively immature or inadequately
developed. SciDAC-2 should improve outreach activities,for instance by creating SciDAC Institutes aimed at the broad
inclusion of new communities, and improving workforce development to ensure a robust supply of
next-generation researchers.
|
|
It was also recommended that these investments should be supported by computational and
experimental science application investments from DOE programs and by additional ASCR infrastructure
investments in leadership-class computing facilities, the National Energy Research Scientific Computing Center,
and the Energy Sciences Network. The resulting research portfolio will accelerate progress in advanced
energy systems, biotechnology, nanotechnology, and environmental modeling.
|
Dr Horst Simon is Associate Laboratory Director for Computing Sciences, LBNL.
Dr Rick Stevens is Acting Associate Laboratory Director for Physical, Biological
and Computing Sciences, Argonne.
|
|
|
