Quantum Drude Models : A Complete Description of Polarization and Dispersion

Saturday, July 28, 2007 - 9:00am - 9:40am
EE/CS 3-180
Glenn Martyna (IBM Corporation)
Present empirical molecular force fields ignore polarization,
that is the rearrangement of charge distribution due to environment,
and only treat dispersion or vanderWaals forces at the pair level. The
former approximation limits the ability of empirical models to describe
the properties of materials at surface/interfaces. For example, the
shell of a chlorine ion in water can only be predicted by properly
for the polarizability of the water molecules. The latter approximation
is equally problematic at surfaces, yielding large errors in surfaces
which leads to errors in acouning for hydrophobic solvation.
Although ab initio calculations could in principle restore manybody
polarization and manybody dispersion, in fact, the present level of theory
that is sufficiently computational efficient to employ in studies of large
systems, Gradient Corrected Density Functional Theory, does not treat
dispersion even at the pair level. Since the goal of present theoretical
research is to describe the complex interfacial phenomena that drive
processes such as protein folding and transport of materials through
membranes, it is critical to develop novel, linear scale methods, that can
treat both manybody polarization and dispersion. Here, the applied
that underlies the Drude method including a new path integral scheme that
not only models
manybody polarization and dispersion efficient and accurately, but goes
beyond the standard dipole
approximation to include all multipole interactions without requiring an
explicit multipole expansion is presented.