What do we currently mean by Computational Science?

1
What do we currently mean by Computational
Science?

• Traditionally focuses on the hard sciences and
  engineering
• Physics, Chemistry, Mechanics, Aerospace, etc.
• Mesh-based calculations, direct numerical
  simulations, optimization, etc. modeling the
  scientific phenomenon directly
• Used for verification of experiments,
  verification of theory, and, increasingly,
  predictive capabilities
• Associated research numerical algorithms, big
  hardware (computing, storage, networking)
• Requires teams of experts no single researcher
  can do it all

2
One Example DOE SciDAC Program

• A program that works
• Ive never been so happy as a computational
  scientist, Tony Mezzacappa, TSI, March 11, 2003
• Targets specific Office of Science applications
• Fusion, Accelerator Design, Basic Energy
  Sciences, Climate, Astrophysics, etc.
• Unites these application scientists with ISICS
• 3 Applied Mathematics (TOPS, APDEC, TSTT)
• 4 Computer Science (CCA, PERC, Scalable Systems,
  Data Management)
• ISICs are charged with collaborating with
  application teams
• Whats missing from this program? How can it be
  improved?
• The obvious more applications, more CS, more
  math
• Tighter ties with computer architecture
  development. What else?

3
What new directions can/should be explored?
Research Themes

• CSE digging more deeply into the traditional
  areas
• Multiscale modeling, hybrid solution techniques,
  etc.
• CSE moving into nontraditional areas
• Biological sciences, economics, medicine,
  computational finance, etc.
• Some traditional techniques, but also data
  mining, statistics,
• Its not always about big hardware
• What about
• English, Law, History, Archeology,?
• Almost no exposure in these areas How can this
  be rectified? What can be accomplished if we
  turn our attention in these areas?

4
What has worked? How do we go forward?
Supporting CSE

• Large, goal-based initiatives have worked
• Requires
• Advances in applications, mathematics, and
  computer science
• Integration of the three areas
• An adequate reward system for those interactions
• Examples
• Grand Challenges, DOE SciDAC program
• This is much easier at the labs than
  universities..

5
Is the educational system on track for CSE?
Education

• Many institutions are trying to understand where
  CSE fits within their departmental structures
• Computational science still traditionally done in
  the discipline departments
• Its difficult to find new hires that have been
  trained in all three aspects of computational
  science
• Very few educational programs take a truly
  interdisciplinary tact
• Typically 1 or 2 aspects considered
• Of course, there are exceptions, e.g., TICAM, DOE
  CSGF
• How do we increase the number of programs that
  are truly interdisciplinary? Can we? Should we?
• How can we help stretch departmental budgets to
  accommodate cross-cutting curricula? What
  infrastructure is/can be provided at the national
  level from NSF or SIAM?
• How do we encourage the exploration of new
  disciplines?