The principal objectives of the condensed matter theory group at NCSU are
to predict and/or explain the properties of materials from
first-principles, parameter-free calculations and to develop new methods
for such calculations. The work spans a substantial range of theoretical
re-search, from theory and simulations of the newly discovered nanotubes,
to semiconductor growth, interface formation, and interdiffusion in
superlattices. From a techni-cal point of view, it mostly involves the use
of the so-called density functional theory, which describes complex
quantum-mechanical interactions between nuclei and electrons in terms of
only the electron density and the nuclear potentials. This theory, which
led to a 1998 Nobel Prize in Chemistry, is variational and very accurate.
For example, atomic geometries and elastic properties can be predicted to
an accuracy of a few percent. Recent advances in methodology allow for
simulating the dynamics of atomic motion while propagating the
quantum-mechanical solutions. Typical calculations include more than 300
non-equivalent atoms and require a supercomputer.
They have recently developed an advanced multigrid method for such
computations, which solves iteratively very large non-sparse eigenvalue and
Poisson problems on a 3D grid with up to 3 million gridpoints. This method
has been fully parallelized and executes at up to 42 GFLOPS on a
256-processor SGI/Cray T3E supercomputer. Currently, additional multilevel
algorithms for this problem are being developed in collaboration with Prof.
Achi Brandt from the Applied Mathematics Department at the Weizmann
Institute in Israel.
The work of this group is well recognized in the physics, materials
science, and computational physics communities. Snapshots from their
simulations have appeared on the covers of Science, Science News, Physics
World, and Science and Engineering Indicators.
CRSC researchers on this project include J. Bernholc
and his colleagues in the Physics Department at NCSU.
