Parallel, linear-scaling building-block and embedding method based on localized orbitals and orbital-specific basis sets

摘要

We present a new linear scaling method for the energy minimization step ofsemiempirical and first-principles Hartree-Fock and Kohn-Sham calculations. Itis based on the self-consistent calculation of the optimum localized orbitalsof any localization method of choice and on the use of orbital-specific basissets. The full set of localized orbitals of a large molecule is seen as anorbital mosaic where each tessera is made of only a few of them. The orbitaltesserae are computed out of a set of embedded cluster pseudoeigenvalue coupledequations which are solved in a building-block self-consistent fashion. In eachiteration, the embedded cluster equations are solved independently of eachother and, as a result, the method is parallel at a high level of thecalculation. In addition to full system calculations, the method enables toperform simpler, much less demanding embedded cluster calculations, where onlya fraction of the localized molecular orbitals are variational while the restare frozen, taking advantage of the transferability of the localized orbitalsof a given localization method between similar molecules. Monitoring singlepoint energy calculations of large poly(ethylene oxide) molecules and threedimensional carbon monoxide clusters using an extended Huckel Hamiltonian arepresented.