FDES, a GPU-based multislice algorithm with increased efficiency of the computation of the projected potential

While the computational complexity of calculation of the projected potential in the multislice algorithm through reciprocal space scales quadratically with the number of atoms A per slice, a pure real-space calculation scales linearly with A. A hybrid strategy is introduced that has a theoretical complexity of O(AlogA), but that, when measured, outperforms both the reciprocal-space and the real-space approach by approximately an order in A and a large factor, respectively.