An improved Green's function algorithm applied to quantum transport in carbon nanotubes

The renormalization-decimation algorithm (RDA) of López Sancho et al. is used in quantum transport theory to calculate bulk and surface Green's functions. We derive an improved version of the RDA for the case of very long quasi one-dimensional unit cells (in transport direction). This covers not only long unit cells but also supercell-like calculations for structures with disorder or defects. In such large systems, short-range interactions lead to sparse real-space Hamiltonian matrices. We show how this and a corresponding subdivision of the unit cell in combination with the decimation technique can be used to reduce the calculation time. Within the resulting algorithm, separate RDA calculations of much smaller effective Hamiltonian matrices must be done for each Green's function, which enables the treatment of systems too large for the common RDA. Finally, we discuss the performance properties of our improved algorithm as well as some exemplary results for chiral carbon nanotubes.