The parallelization strategy to be used to calculate the PES spectrum
using the resources available in the domain parallelization pool.
This option is not available without domain parallelization.
Parallelization over k-points and states is always enabled when available.
Options:
pf_none:
No parallelization.
pf_time:
Parallelize time integration. This requires to store some quantities over a
number of time steps equal to the number of cores available.
pf_momentum:
Parallelize over the final momentum grid. This strategy has a much lower
memory footprint than the one above (time) but seems to provide a smaller
speedup.
pf_surface:
Parallelize over surface points.
Option pf_time and pf_surface can be combined: pf_time + pf_surface.