A. CASTEP currently supports semi-empirical dispersion corrections for phonons using the finite displacement (including supercell) methods, but not yet with DFPT.
A. Astersisks ("*") instead of numbers is the traditional Fortran response when the number is too large to fit in the field. In this case it is almost certainly caused by a very poorly converged calculation giving numerical nonsense.
A. Use the mode_follow utility which comes with academic source distributions. This reads a <seed>
.phonon file from a previous calculation and generates a sequence of <seed>-nn
mode_follow --help gives some basic help information.
A. No. If you used either the supercell or interpolation methods, the full force constant matrix is saved in the checkpoint file and can be used to recalculate phonons at any set of q. Just set up a "continuation" calculation to start from this file, and change the
PHONON_FINE_XXX keywords in the .cell file
A. Use the new adaptive broadening method in CASTEP 7.0. This is selected by keyword
PHONON_CALCULATE_DOS, and customized by
PHONON_DOS_SPACING, and generates a <seed>
.phonon_dos output file containing the precalculated DOS.