def generate_displacements(self,
distance=0.03,
cutoff_pair_distance=None,
is_plusminus='auto',
is_diagonal=True):
direction_dataset = get_third_order_displacements(
self._supercell,
self._symmetry,
is_plusminus=is_plusminus,
is_diagonal=is_diagonal)
self._displacement_dataset = direction_to_displacement(
direction_dataset,
distance,
self._supercell,
cutoff_distance=cutoff_pair_distance)
if self._phonon_supercell_matrix is not None:
phonon_displacement_directions = get_least_displacements(
self._phonon_supercell_symmetry,
is_plusminus=is_plusminus,
is_diagonal=False)
self._phonon_displacement_dataset = direction_to_displacement_fc2(
phonon_displacement_directions,
distance,
self._phonon_supercell)
def generate_displacements(self,
distance=0.03,
cutoff_pair_distance=None,
is_plusminus='auto',
is_diagonal=True):
direction_dataset = get_third_order_displacements(
self._supercell,
self._symmetry,
is_plusminus=is_plusminus,
is_diagonal=is_diagonal)
self._displacement_dataset = direction_to_displacement(
direction_dataset,
distance,
self._supercell,
cutoff_distance=cutoff_pair_distance)
if self._phonon_supercell_matrix is not None:
# 'is_diagonal=False' below is made intentionally. For
# third-order force constants, we need better accuracy,
# and I expect this choice is better for it, but not very
# sure.
# In phono3py, two atoms are displaced for each
# configuration and the displacements are chosen, first
# displacement from the perfect supercell, then second
# displacement, considering symmetry. If I choose
# is_diagonal=False for the first displacement, the
# symmetry is less broken and the number of second
# displacements can be smaller than in the case of
# is_diagonal=True for the first displacement. This is
# done in the call get_least_displacements() in
# phonon3.displacement_fc3.get_third_order_displacements().
#
# The call get_least_displacements() is only for the
# second order force constants, but 'is_diagonal=False' to
# be consistent with the above function call, and also for
# the accuracy when calculating ph-ph interaction
# strength because displacement directions are better to be
# close to perpendicular each other to fit force constants.
#
# On the discussion of the accuracy, these are just my
# expectation when I designed phono3py in the early time,
# and in fact now I guess not very different. If these are
# little different, then I should not surprise users to
# change the default behaviour. At this moment, this is
# open question and we will have more advance and should
# have better specificy external software on this.
phonon_displacement_directions = get_least_displacements(
self._phonon_supercell_symmetry,
is_plusminus=is_plusminus,
is_diagonal=False)
self._phonon_displacement_dataset = directions_to_displacement_dataset(
phonon_displacement_directions,
distance,
self._phonon_supercell)