def rotate_wf(wf_list,theta,axis,ref_pt):
"""
This function takes a Wannier Function list, and
rotates the coordinates of the WF centres by an angle
theta about the axis 'axis'.
It returns a new list of Wannier Functions with the
rotated coordinates.
"""
import numpy
from numpy import dot as dot
from numpy import cross as cross
from math import cos as cos
from math import sin as sin
wf_list_final = []
for wf in wf_list:
new_wf = wannier_function()
# first we compute the relative vector
v_rel = numpy.array([wf.x,wf.y,wf.z]) - ref_pt
# then we rotate
v_rot = dot(v_rel,axis)*axis+cos(theta)*(v_rel-dot(v_rel,axis)*axis)+ \
sin(theta)*cross(axis,v_rel)
# then we reintroduce the absolute coordinates
v_final = v_rot + ref_pt
# now we update the wf coordinates
new_wf.setindex(wf.index)
new_wf.setx(v_final[0])
new_wf.sety(v_final[1])
new_wf.setz(v_final[2])
new_wf.setonsite(wf.onsite)
# adding the new wf to the final list
wf_list_final.append(new_wf)
return wf_list_final
def main_routine(wf_total,
wf_per_pl,
cells_per_pl,
cond_dir,
second_dir,
wannier_functions,
atom_number,
atomic_coordinates,
matrix,
angle,
delta):
# first we separate the left-most principal layer, the conductor region
# and the right-most principal layer
PL1 = wannier_functions[:wf_per_pl]
conductor = wannier_functions[wf_per_pl:-wf_per_pl]
PL4 = wannier_functions[-wf_per_pl:]
# now we compute the unit vector for the axis of rotation
point_A = barycenter(PL1)
point_B = barycenter(PL4)
unit_vector = (1.0/numpy.linalg.norm(point_B-point_A))*(point_B-point_A)
if cond_dir=='x':
reference = numpy.array([1.0,0.0,0.0])
elif cond_dir=='y':
reference = numpy.array([0.0,1.0,0.0])
else:
reference = numpy.array([0.0,0.0,1.0])
if numpy.dot(unit_vector,reference)<=0.95:
print ""
print " Warning in rotate_conductor :"
print " The computed unit vector for the axis of"
print " rotation seems to be off from the reference"
print " unit vector : %s" %(cond_dir)
print " The dot product of unit vector with %s is : %3.2f" %(cond_dir,numpy.dot(unit_vector,reference))
print ""
user_answer = ""
while user_answer not in ["yes", "no"]:
user_answer = str(raw_input(" is this OK ? ('yes' or 'no') "))
if user_answer=='no':
print " Exiting the program..."
print ""
sys.exit(1)
# we search the groups of WF with similar centres. To do this
# a group is identified by the index of the first WF in the group
# and the group size
gp_dict = {}
gp_bool = numpy.zeros(len(conductor),dtype='bool')
for i in xrange(len(conductor)):
if gp_bool[i]==False:
gp_bool[i] = True
gp_size = 1
# for each wf we search for similar centres just around that wf
for j in xrange(max(i-10,0),min(i+11,len(conductor))):
if gp_bool[j]==False:
if same_center(conductor[i],conductor[j]):
gp_bool[j] = True
gp_size += 1
# if gp_size is at least 2 we add this group to the dictionnary
if gp_size >= 2:
gp_dict[conductor[i].index] = gp_size
# now we need to remove the redondant WF from the conductor
not_to_add = []
for wf in conductor:
if wf.index in gp_dict.keys():
for j in xrange(1,gp_dict[wf.index]):
not_to_add.append(wf.index+j)
reduced_conductor = []
for wf in conductor:
if wf.index not in not_to_add:
reduced_conductor.append(wf)
# with the reduced_conductor list, we can rotate the centers
reduced_rotated_conductor = rotate_wf(reduced_conductor,angle,unit_vector,point_A)
# we then sort the rotated Wannier Functions
sorted_reduced_conductor = sort(reduced_rotated_conductor,
cond_dir,
second_dir,
delta)
# we re-insert the WF with similar centers
sorted_conductor = []
for wf in sorted_reduced_conductor:
if wf.index not in gp_dict.keys():
sorted_conductor.append(wf)
else:
sorted_conductor.append(wf)
for j in xrange(1,gp_dict[wf.index]):
new_wf = wannier_function()
pos_in_list = wf.index+j-conductor[0].index
new_wf.setindex(conductor[pos_in_list].index)
new_wf.setx(wf.x)
new_wf.sety(wf.y)
new_wf.setz(wf.z)
new_wf.setonsite(conductor[pos_in_list].onsite)
sorted_conductor.append(new_wf)
# we now find the mapping between the original order of WF
# and the final order after sorting
final_order = []
reference = conductor[0].index
for wf in sorted_conductor:
final_order.append(wf.index-reference)
# now we can reconstruct the matrix
final_matrix = numpy.zeros((matrix.shape[0],matrix.shape[1]),dtype='float')
for i in xrange(matrix.shape[0]):
for j in xrange(matrix.shape[1]):
final_matrix[i,j] = matrix[final_order[i],final_order[j]]
write_htC('rotated_matrix_htC.dat',final_matrix)
# at last we re-generate the *_tran_info.dat file
final_wf_list = PL1 + sorted_conductor + PL4
final_atomic_coordinates = rotate_atoms(atomic_coordinates,angle,unit_vector,point_A)
write_tran_info(wf_total,
wf_per_pl,
cells_per_pl,
cond_dir,
second_dir,
final_wf_list,
atom_number,
final_atomic_coordinates)
return
