Preconditioned conjugate gradient method for the sparse generalized eigenvalue problem in electronic structure calculations (Q5934196)

The paper is inspired by the needs of solid state physics (the lowest few energies and wave functions of the chlorine molecule and of a 64-atomic silicon crystal are calculated for illustration). Its physical background is the density-functional theory represented in localized (non-orthogonal) basis. The algorithm offered is an iterative one, making an ample use of the sparsity of the matrices in question (first of all, via a one-parametric family of the kinetic-energy preconditionings). Empirically, the convergence appears to be linear.