Research highlights and collaborations and interrelations of participants in the "Electron Excitations and Correlations" subgroup within the "Excited-State Electronic Structure and Response Functions" group are summarized as follows:

1. New physics and methods (Lawrence Livermore National Labs and LBNL/UC Berkeley):
   • A first-principles method for forces in excited states has been fully developed within the GW-BSE approach. This allows for atomic and lattice relaxation of optically excited systems and calculation of the photoluminescence and other related properties.
   • The nature and form of coupling of electromagnetic fields to nonlocal Hamiltonians have been elucidated, particularly for the case of nonlocal pseudopotentials. Having the correct coupling is of great importance in being able to compute the quantitative response of materials to external electromagnetic fields and probes.
   • A first stage study of the physics of non-equilibrium optical excitations in solids is complete. The behavior of the electronic gap as a function of the density of excited electrons is studied and it is found that band gap renormalization (the reduction of the energy gap) due to purely electronic effects is not sufficient to create a metal in the case of GaAs. The study requires implementation of the Keldysh formalism for non-equilibrium Green's functions.
   • The lifetime of excited electrons at finite temperature has been studied using the GW approximation and the Matsubara formalism. The results explain the experimental lifetimes in excited aluminum.
   • The proper role of screening in GW-BSE calculations is now much better understood, particularly when calculations are performed with a limited set of basis functions. The role and effect of the omitted functions has been theoretically studied. It is shown that an unconverged calculation requires an unphysical screening in the exchange term to mimic the correct converged results in practice.

2. Comparison of time dependent density functional theory (TDLDA) and GW-BSE methods for optical excitations (U. of Minnesota and LBNL/UC Berkeley):
   • The study of the optical properties of dimer molecules (BeO, MgO, SiO, CO, C₂, BN, and LiF) is near completion. This work on simple and experimentally well-studied molecules allows for detailed comparison of the TDLDA and GW-BSE methods to each other and to available experiments, allowing us to understand, in detail, the accuracy of the methods and differences between them, leading hopefully to future improvements for both methods.
   • A comparison of the two approaches for quartz (SiO₂) are complete and shows that the TDLDA has problems reproducing certain qualitative optical properties of quartz whereas the GW-BSE method produces results in good agreement with experiment.

3. Applications to complex and novel materials (Lawrence Livermore National Lab, Georgia Tech., Ohio State U., LBNL/UC Berkeley, and PARC research center):
   • The role of defects in semiconductors and their effects on band offsets are under study using first-principles GW calculations. The calculations on grain boundaries and defects are used to identify and distinguish the electronic signature of different defects on the boundaries.
   • The optical properties of carbon nanotubes are under study. Preliminary results show very strong excitonic effects in semiconducting nanotubes and weaker effects in metallic ones. Surprisingly, exciton binding is found in metallic tubes contrary to the conventional wisdom for bulk solids.
   • The physics and electronic structure of metal hydrides are studied. The size and nature of the electronic energy gaps in these materials were not well understood. Using the first principles GW method allows for direct calculation of these properties.
   • Molecular crystals have novel and rich electronic properties with future technological applications. Studies are underway regarding the optical properties of solid pentacene and naphthalene.

4. Code development (Ohio State U. and LBNL/UC Berkeley):
   • We are working on reducing the scaling of GW-BSE calculations from O(N⁴) to O(N³) (N is the number of atoms) in the computationally intensive parts of the codes. This will help us calculate much larger physical systems and the nature of more complex defects.
   • The full and complete parallelization of GW and BSE codes are nearly complete and this will help conserve both memory and computation time for GW-BSE calculations.