The Visualization Laboratory and the Science Exploratorium at CCS enable researchers to understand and convey the significance of a supercomputer's calculation results.
Atoms
are so close and stars are so far away. In both cases, we cannot
really "see" how they behave when various forces act on them.
Fortunately, technologies exist to transform supercomputers into microscopes
and telescopes that possess ultrahigh resolution. Using experimental
data, scientists can make calculations on supercomputers to predict how
atoms and stars change under different conditions.

Graphical representation of a molecular orbital of the benzene dimer
with the underlying adaptive computational mesh.
|
|
Thanks to techniques
for visualizing data, suddenly "pictures" worth a thousand
gigabytes can be produced. Scientists can "see" in vivid
colors images of an alloy's electrons pointing in different directions— like
a compass needle—as a metallic alloy's magnetic field shifts
because of changes in alloy composition. They can "observe" explosions
in a star as it synthesizes and disperses elements that make life possible
on our planet. These images give researchers insights into scientific
processes and phenomena.
Computational
science generates, processes and interprets nearly
unfathomable quantities of data from scientific simulations of,
for example, galactic supernovae, proteins, nanostructures, and fusion
energy devices. Because researchers have vast amounts of data to
be analyzed, they often prefer to examine data in a compact form that
is accurate, reliable, and readily understandable. Viewing the results
as pictures is immensely more efficient. This is where the science
of visualization demonstrates its value, and why visualization must
keep pace with advances in scientific computing.
Visualization
is a discipline that develops computer graphics,
interaction devices, three-dimensional (3D) imaging, and realistic
rendering techniques. The tools use simulation data as input and create
a scientifically accurate representation.
The
Visualization Laboratory and the Science Exploratorium in the Department
of Energy's Center for Computational Sciences
(CCS) at ORNL offer Laboratory and visiting researchers the ability
to visually analyze and compare results of computer simulations. The
ability to see an image of the results may reveal intricacies and flaws
otherwise concealed in a haystack of data.
For
example, in an image produced computationally from huge amounts of
data, researchers can see up close the dimensions and structure
of an engineered molecule. They can observe pulses of currents and
winds in a simulated weather event or climate map. They can also witness
the events leading to a supernova explosion and understand how it may
create chemical elements.
At
CCS researchers working with high-performance computers and increasing
volumes of data will be able to view quickly the results of their
work in a lab, office or conference room setting. With virtual-lab
capabilities, these collaborations can involve researchers in the
same room or in offices throughout the world. Researchers can compare
their latest work with previous data through applications that are
intuitively designed and customized to the user. The
Science Exploratorium of CCS is a large-scale, immersive venue for
data exploration and analysis. The new research tool is a 30-ft wide
by 8-ft high display comparable in size to 150 standard computer
displays with a resolution of more than 11,000 by 3,000 pixels, creating
a total space of 35 million pixels. The immersive qualities of this environment
create a powerful discovery tool for research groups and collaborations.
The Science Exploratorium provides a premier data analysis and visualization
capability and facility in DOE's Office of Science.
The
Visualization Laboratory houses display venues of different sizes
and capabilities. The reCAVE offers a seamless integration
of multiple projections into a large, reconfigurable, and inexpensive
virtual environment. Because screens are mounted on movable frames
with their respective projectors and mirrors, scientific images can be
projected and viewed in different geometries, from a long wall,
to "L"-shapes, boxes, or other shapes that best present the
data.
An
array of liquid-crystal display, flat-screen monitors, currently
6 tiles in a 3 x 2 grid, can fit in an office—either hung
from the ceiling or on a wall. This portable array can display 12
megapixels of information. Another display, the IBM T221 monitor,
can display images in photographic resolution of 9 megapixels, much higher
than high-definition television, which has a 2-megapixel capability.
The
visualization facilities are integrated with CCS through a high-band-width
network between large-scale, high-performance computers and large-scale
data visualization computers and displays. Images displayed on
each venue are rendered
using a group of commodity
PCs with high-end
graphics capabilities—a
low-cost, high-performance Beowulf-type visualization
cluster. Workload
is distributed among the
PCs in a parallel fashion for high-performance
and real-time interactions.
Viewing Changes in
Climate, Stars, and
Materials
The Science Exploratorium is used to
investigate the intricacies of global climate
change simulations. Through projected climate
simulations portrayed as maps of the
globe, color-coding helps deliver an understanding
of parameters such as surface
temperature and carbon dioxide concentrations.
Such simulations may involve
hundreds of parameters. Because of highly
interconnected and complex interrelationships
among variables, subtle changes in
one variable can have significant long-term
effects. Large, tiled displays provide an
ideal environment for observing such sensitivities
between multiple simulations and
variables and can be experienced by as
many as 20 participants.
Researchers
can examine virtual stellar explosions—complex hydrodynamics
from near the stellar core to the turbulent
shock front. As the models are refined, scientists
can observe parameters such as
temperature, density, and velocity in the
supernovae simulations and compare results
from both theoretical and observational
data. Eventually, it may be possible
to understand how the elements that make
life possible are synthesized and dispersed
by such explosions.
Materials
science research is conducted at the atomic level as researchers
build and study the dynamics of custom engineered
materials. ORNL's visualization
facilities offer scientists the ability to zoom in to the atomistic scale
and observe spin dynamics and vibrational modes or zoom back out to determine
bulk material properties, say, in a magnetic alloy whose composition
is being changed. Visualization venues such as those in the Science Exploratorium
and Visualization Lab enable researchers to learn what works, as well
as what fails—filling in the missing pieces of scientific understanding.
Visual
Mural for Visitors
In
addition to providing researchers with an indispensable visual tool
for comprehending and communicating their research, ORNL's
visualization facilities have repeatedly proven their effectiveness at
communicating and demonstrating the value of ORNL research to the public.
Collectively, the facilities comprise a visual mural for Laboratory
visitors ranging from the nation's highest elected officials to
local school groups. The Exploratorium was a highlight of the recent
Laboratory tour by science writers attending the New Horizons in
Science Briefing sponsored by the Council for the Advancement of Science
Writing.
Scientific
visualization is a rapidly moving discipline as research and technology evolves. As computers advance in power and speed and scientific
models become more complex, the art and science of visualization and its
underlying analytical process must advance in step. As scientific
computing closes in on its 100teraflop goal, visualization researchers
will be pressed to keep pace with the progress. But the resulting discoveries
from images on the screen and the understanding they provide will be
worth the investment.

At the CCS Visualization Laboratory, three researchers view a threedimensional simulation of a supernova explosion.
|
|