Linear scaling for the charge density fitting procedure of the linear combination of Gaussian-type orbitals density functional method

A divide-and-conquer approach to the charge density fitting procedure of the linear combination of Gaussian-type orbitals density functional method has been developed. The density is partitioned among subsystems in a fashion similar to a Mulliken population analysis and subsequently fit by auxilliary basis functions centered on the atoms within a subsystem or within a certain cut-off distance of the subsystem. A relatively small cut-off is required to achieve reasonable accuracy in total energies. For large systems, this procedure will scale linearly with system size. This approach may be extended to Hartree-Fock calculations as well.