For extended loss structure (i.e., upwards from 50-100 eV), the real space Path Expansion method is so well tested and efficient that I do not see the point in trying to reformulate it in reciprocal space. Therefore, it is always done in real space. If the RECIPROCAL card is active (and the crystal is specified using LATTICE and unit cell ATOMS card), then the rdinp module will generate a real-space cluster based on the information of the LATTICE and ATOMS cards, and the value of the RPATH value. This cluster will be written to the atoms.dat file. The spectrum will then be calculated using the path and genfmt modules, which always work in real space. IT IS GOOD PRACTICE to look at this atoms.dat file when one is still getting used to entering unit cell and atom positions - it's a good check to make sure no mistakes have been entered.

developer's resources