Guessing the Rule for Nature’s Glue in Physics, Chemistry, and Biology

When: Wednesday, February 22, 2012 at 4:00 pm
Where: DA 114
Speaker: John P. Perdew
Organization: Department of Physics and Engineering Physics, Tulane University, New Orleans, LA
Sponsor: Northeastern University College of Science and Department of Physics

The structure of ordinary matter (atoms, molecules, nanostructures, solids) arises from quantum mechanics, with simple electrostatic interactions among electrons and nuclei. But the correlated “bee swarm” of electrons is inefficiently described by a many-electron wavefunction.

The most widely-used computational approach is density functional theory, which invokes simpler objects, the electron density and auxiliary orbitals. This theory is exact in principle for the ground-state energy and density, but in practice it requires an approximation for the exchange-correlation energy (nature’s glue, which binds atoms to other atoms) in terms of the density.

Known constraints on the exact functional enable the construction of approximations without empirical fitting. The local spin density approximation was proposed by Kohn and Sham in 1965.

In 1981, two roads diverged in density functional theory. The road more traveled led to the generalized gradient approximation, etc., and to an accurate description of typical equilibrium bonds (“weak correlation”).

The road less traveled led to the self-interaction correction and derivative discontinuity, and to a realistic description of stretched bonds (“strong correlation”).  Perhaps these roads can converge to make one approach for all ordinary matter.