def test_connectivity_array(self):
vc = VoronoiConnectivity(self.get_structure("LiFePO4"))
ca = vc.connectivity_array
expected = np.array([0, 1.96338392, 0, 0.04594495])
self.assertTrue(np.allclose(ca[15, :4, ca.shape[2] // 2], expected))
expected = np.array([0, 0, 0])
self.assertTrue(np.allclose(ca[1, -3:, 51], expected))
site = vc.get_sitej(27, 51)
self.assertEqual(site.specie, Element("O"))
expected = np.array([-0.29158, 0.74889, 0.95684])
self.assertTrue(np.allclose(site.frac_coords, expected))
def test_connectivity_array(self):
vc = VoronoiConnectivity(self.get_structure("LiFePO4"))
ca = vc.connectivity_array
expected = np.array([0, 1.96338392, 0, 0.04594495])
self.assertTrue(np.allclose(ca[15, :4, ca.shape[2] // 2], expected))
expected = np.array([0, 0, 0])
self.assertTrue(np.allclose(ca[1, -3:, 51], expected))
site = vc.get_sitej(27, 51)
self.assertEqual(site.specie, Element('O'))
expected = np.array([-0.29158, 0.74889, 0.95684])
self.assertTrue(np.allclose(site.frac_coords, expected))
def test_connectivity_array(self):
vc = VoronoiConnectivity(self.get_structure("LiFePO4"))
ca = vc.connectivity_array
np.set_printoptions(threshold=np.NAN, linewidth=np.NAN, suppress=np.NAN)
expected = np.array([0, 1.96338392, 0, 0.04594495])
self.assertTrue(np.allclose(ca[15, :4, ca.shape[2] // 2], expected))
expected = np.array([0, 0, 0])
self.assertTrue(np.allclose(ca[1, -3:, 51], expected))
site = vc.get_sitej(27, 51)
self.assertEqual(site.specie, Element('O'))
expected = np.array([-0.29158, 0.74889, 0.95684])
self.assertTrue(np.allclose(site.frac_coords, expected))
def test_connectivity_array(self):
vc = VoronoiConnectivity(self.s)
ca = vc.connectivity_array
np.set_printoptions(threshold = np.NAN, linewidth = np.NAN, suppress = np.NAN)
expected = np.array([0, 1.96338392, 0, 0.04594495])
self.assertTrue(np.allclose(ca[15, :4, ca.shape[2] // 2], expected))
expected = np.array([0, 2.04147512, 2.12118181])
self.assertTrue(np.allclose(ca[1, -3:, 51], expected))
site = vc.get_sitej(27, 51)
self.assertEqual(site.specie, Element('O'))
expected = np.array([-0.04316, 0.25111, 0.29158])
self.assertTrue(np.allclose(site.frac_coords, expected))
def voronoi_Nonperiodic(self, structure):
"""
VoronoiConnectivity().get_connections()
input:
structure - (Structure)
cutoff - (float)
return:
A list of site pairs that are Voronoi Neighbours,
along with their real-space distances.
The shape of a list is N×3, where N is the number of pairs.
e.g. [[0, 1, 3.9710929317605546], [0, 2, 3.971092931760555],
[0, 3, 3.971092931760555], ...]
"""
n_atoms = len(structure)
connections = VoronoiConnectivity(structure, cutoff=self.cutoff).get_connections()
adjacency_matrix = np.zeros((n_atoms, n_atoms))
# compute adjacency matrix
for n in range(len(connections)):
site_i = connections[n][0]
site_j = connections[n][1]
adjacency_matrix[site_i, site_j] += 1
# compute squared distance matrix
squared_distance_matrix = structure.distance_matrix ** 2
warnings.filterwarnings("ignore") # ignore RuntimeWarning (division by zero)
reciprocal_squared_distance_matrix = squared_distance_matrix ** -1
reciprocal_squared_distance_matrix[
reciprocal_squared_distance_matrix == np.inf] = 0
return (adjacency_matrix, reciprocal_squared_distance_matrix)
def test_connectivity_array(self):
vc = VoronoiConnectivity(self.s)
ca = vc.connectivity_array
np.set_printoptions(threshold=np.NAN,
linewidth=np.NAN,
suppress=np.NAN)
expected = np.array([0, 1.96338392, 0, 0.04594495])
self.assertTrue(np.allclose(ca[15, :4, ca.shape[2] // 2], expected))
expected = np.array([0, 2.04147512, 2.12118181])
self.assertTrue(np.allclose(ca[1, -3:, 51], expected))
site = vc.get_sitej(27, 51)
self.assertEqual(site.specie, Element('O'))
expected = np.array([-0.04316, 0.25111, 0.29158])
self.assertTrue(np.allclose(site.frac_coords, expected))
def feas(self, image):
# Returns pymatgen structure from ASE Atoms.
try:
image = AseAtomsAdaptor.get_structure(image)
except:
image = image
# atom feature vector
atom_fea = np.array([self.dict_atom_fea[i]
for i in image.atomic_numbers])
# VoronoiConnectivity for bond feature vector
VC = VoronoiConnectivity(image)
all_nbrs = []
# add connectivity for neighbor atoms distances
conn = VC.connectivity_array
for ii in range(0, conn.shape[0]):
curnbr = []
for jj in range(0, conn.shape[1]):
for kk in range(0, conn.shape[2]):
if conn[ii][jj][kk] != 0:
curnbr.append([ii, conn[ii][jj][kk]
/ np.max(conn[ii]), jj])
else:
curnbr.append([ii, 0.0, jj])
all_nbrs.append(np.array(curnbr))
all_nbrs = [sorted(nbrs, key=lambda x: x[1], reverse=True)
for nbrs in all_nbrs]
nbr_fea_idx = np.array([list(map(lambda x: x[2],
nbr[:self.max_num_nbr]))
for nbr in all_nbrs], dtype=np.int64)
nbr_fea = np.array([list(map(lambda x: x[1], nbr[:self.max_num_nbr]))
for nbr in all_nbrs])
nbr_fea = np.exp(-(nbr_fea[..., np.newaxis] - self.filter)**2
/ self.step**2)
return atom_fea, nbr_fea, nbr_fea_idx
def __getitem__(self, idx):
cif_id, target = self.id_prop_data[idx]
crystal = Structure.from_file(
os.path.join(self.root_dir, cif_id + '.cif'))
atom_fea = np.vstack([
self.ari.get_atom_fea(crystal[i].specie.number)
for i in range(len(crystal))
])
atom_fea = torch.Tensor(atom_fea)
# all_nbrs = crystal.get_all_neighbors(self.radius, include_index=True)
VC = VoronoiConnectivity(crystal)
all_nbrs = []
# add connectivity for neighbor atoms distances
conn = VC.connectivity_array
for ii in range(0, conn.shape[0]):
curnbr = []
for jj in range(0, conn.shape[1]):
for kk in range(0, conn.shape[2]):
if conn[ii][jj][kk] != 0:
curnbr.append(
[ii, conn[ii][jj][kk] / np.max(conn[ii]), jj])
#curnbr.append([ii, conn[ii][jj][kk], jj])
else:
curnbr.append([ii, 0.0, jj])
all_nbrs.append(np.array(curnbr))
all_nbrs = [
sorted(nbrs, key=lambda x: x[1], reverse=True) for nbrs in all_nbrs
]
nbr_fea_idx = np.array([
list(map(lambda x: x[2], nbr[:self.max_num_nbr]))
for nbr in all_nbrs
])
nbr_fea = np.array([
list(map(lambda x: x[1], nbr[:self.max_num_nbr]))
for nbr in all_nbrs
])
nbr_fea = self.gdf.expand(nbr_fea)
atom_fea = torch.Tensor(atom_fea)
nbr_fea = torch.Tensor(nbr_fea)
nbr_fea_idx = torch.LongTensor(nbr_fea_idx)
target = torch.Tensor([float(target)])
return (atom_fea, nbr_fea, nbr_fea_idx), target, cif_id
def __getitem__(self, idx):
atoms = copy.deepcopy(self.atoms_list[idx])
crystal = AseAtomsAdaptor.get_structure(atoms)
atoms_initial_config = copy.deepcopy(
self.atoms_list_initial_config[idx])
crystal_initial_config = AseAtomsAdaptor.get_structure(
atoms_initial_config)
#Stack the features from atom_init
atom_fea = np.vstack([
self.ari.get_atom_fea(crystal[i].specie.number)
for i in range(len(crystal))
])
# If use_tag=True, then add the tag as an atom feature
if self.use_tag:
atom_fea = np.hstack([atom_fea, atoms.get_tags().reshape((-1, 1))])
# If use_fixed_info=True, then add whether the atom is fixed by ASE constraint to the features
if self.use_fixed_info:
fix_loc, = np.where([
type(constraint) == FixAtoms
for constraint in atoms.constraints
])
fix_atoms_indices = set(
atoms.constraints[fix_loc[0]].get_indices())
fixed_atoms = np.array(
[i in fix_atoms_indices for i in range(len(atoms))]).reshape(
(-1, 1))
atom_fea = np.hstack([atom_fea, fixed_atoms])
# If use_voronoi, then use the voronoi connectivity from pymatgen to determine neighbors and distances
if self.use_voronoi:
#Get the connectivity array for the initial and final structure
VC = VoronoiConnectivity(crystal)
VC_initial_config = VoronoiConnectivity(crystal_initial_config)
conn = copy.deepcopy(VC.connectivity_array)
conn_initial_config = copy.deepcopy(
VC_initial_config.connectivity_array)
#Iterate through each atom, find it's neighbors, and add their distances
all_nbrs = []
# Loop over central atom
for ii in range(0, conn.shape[0]):
curnbr = []
#Loop over neighbor atoms
for jj in range(0, conn.shape[1]):
#Loop over each possible PBC image for the chosen image
for kk in range(0, conn.shape[2]):
# Only add as a neighbor if the atom is not the currently selected center one and there is connectivity
# to that image
if jj is not kk and conn[ii][jj][kk] != 0:
#Add the neighbor strength depending on train_geometry base
if self.train_geometry == 'initial':
curnbr.append([
ii, conn_initial_config[ii][jj][kk] /
np.max(conn_initial_config[ii]), jj
])
elif self.train_geometry == 'final':
curnbr.append([
ii, conn[ii][jj][kk] / np.max(conn[ii]), jj
])
elif self.train_geometry == 'final-adsorbate':
#In order for this to work, each adsorbate atom should be set to tag==1 in the atoms object
if (atoms.get_tags()[ii] == 1
or atoms.get_tags()[jj] == 1):
if conn[ii][jj][kk] / np.max(
conn[ii]) > 0.3:
curnbr.append([ii, 1.0, jj])
else:
curnbr.append([ii, 0.0, jj])
else:
curnbr.append([
ii, conn_initial_config[ii][jj][kk] /
np.max(conn_initial_config[ii]), jj
])
else:
curnbr.append([
ii, conn[ii][jj][kk] / np.max(conn[ii]), jj
])
else:
curnbr.append([ii, 0.0, jj])
all_nbrs.append(np.array(curnbr))
# If use_distance=True, then add the distance to an adsorbate (tag=1) as a feature
if self.use_distance:
atom_fea = np.hstack(
[atom_fea,
distance_to_adsorbate_feature(atoms, VC)])
#Find the strongest neighbors for each atom
all_nbrs = np.array(all_nbrs)
all_nbrs = [
sorted(nbrs, key=lambda x: x[1], reverse=True)
for nbrs in all_nbrs
]
nbr_fea_idx = np.array([
list(map(lambda x: x[2], nbr[:self.max_num_nbr]))
for nbr in all_nbrs
])
nbr_fea = np.array([
list(map(lambda x: x[1], nbr[:self.max_num_nbr]))
for nbr in all_nbrs
])
#expand distance one-hot encoding with GDF
nbr_fea = self.gdf.expand(nbr_fea)
else:
all_nbrs = crystal.get_all_neighbors(self.radius,
include_index=True)
all_nbrs = [sorted(nbrs, key=lambda x: x[1]) for nbrs in all_nbrs]
nbr_fea_idx, nbr_fea = [], []
for nbr in all_nbrs:
if len(nbr) < self.max_num_nbr:
nbr_fea_idx.append(
list(map(lambda x: x[2], nbr)) + [0] *
(self.max_num_nbr - len(nbr)))
nbr_fea.append(
list(map(lambda x: x[1], nbr)) + [self.radius + 1.] *
(self.max_num_nbr - len(nbr)))
else:
nbr_fea_idx.append(
list(map(lambda x: x[2], nbr[:self.max_num_nbr])))
nbr_fea.append(
list(map(lambda x: x[1], nbr[:self.max_num_nbr])))
nbr_fea_idx, nbr_fea = np.array(nbr_fea_idx), np.array(nbr_fea)
nbr_fea = self.gdf.expand(nbr_fea)
try:
nbr_fea = torch.Tensor(nbr_fea)
except RuntimeError:
print(nbr_fea)
nbr_fea_idx = torch.LongTensor(nbr_fea_idx)
atom_fea = torch.Tensor(atom_fea)
return (atom_fea, nbr_fea, nbr_fea_idx)
Voronoi_Analyzer = VoronoiAnalyzer()
anay = Voronoi_Analyzer.analyze(s1, n=0)
strucs = [s1, s2]
anays = Voronoi_Analyzer.analyze_structures(strucs)
print("Voronoi_Analyzer.analyze(s1,n=0):", anay)
#plt = Voronoi_Analyzer.plot_vor_analysis(anays)
relax = RelaxationAnalyzer(s1, s2)
volume_change = relax.get_percentage_volume_change()
lattice_parameter_changes = relax.get_percentage_lattice_parameter_changes()
print('initial volume:', s1.volume)
print('final volume:', s2.volume)
print("percent_volume_change:", volume_change)
print("percent_lattice_change:", lattice_parameter_changes)
bond_dist = relax.get_percentage_bond_dist_changes(max_radius=6)
print("percent_bond_distance_change:", bond_dist)
connec = VoronoiConnectivity(s2)
#print("connec.get_connections():",connec.get_connections())
#A_222_dopant = [0]
A_222_dopant_neighbour = [[69, 73], [68, 72], [63, 65], [62, 64], [64, 65],
[62, 63], [72, 73], [68, 69]]
A_222_dopant_neighbour_MA_I_a = [[69, 73], [68, 72], [59, 61], [58, 60],
[60, 61], [58, 59], [68, 69], [72, 73]]
A_222_dopant_neighbour_MA_I_b = [[67, 71], [66, 70], [63, 65], [62, 64],
[70, 71], [66, 67], [62, 63], [64, 65]]
A_222_dopant_neighbour_MA_I_c = [[69, 73], [68, 72], [62, 64], [63, 65],
[72, 73], [68, 69], [64, 65], [62, 63]]
###Calculate dopant bond length
dopant_bond_a = 0
for i in range(2):
def list_maximum_connectivity(list_struc):
return [
VoronoiConnectivity(structure, cutoff=10).max_connectivity
for structure in list_struc
]
def list_Voronoi_Neighbors(list_struc):
return [
VoronoiConnectivity(structure, cutoff=10).get_connections()
for structure in list_struc
]
def list_the_solid_angle_of_polygon_between_atomi_and_imagej_of_atomj(
list_struc):
return [
VoronoiConnectivity(structure, cutoff=10).connectivity_array
for structure in list_struc
]
coordinate_sites = Voronoi.get_coordinated_sites(1)
Voronoi_Analyzer = VoronoiAnalyzer()
anay = Voronoi_Analyzer.analyze(s1,n=0)
strucs = [s1,s2]
anays = Voronoi_Analyzer.analyze_structures(strucs)
print("Voronoi_Analyzer.analyze(s1,n=0):",anay)
#plt = Voronoi_Analyzer.plot_vor_analysis(anays)
relax = RelaxationAnalyzer(s1,s2)
volume_change = relax.get_percentage_volume_change()
lattice_parameter_changes = relax.get_percentage_lattice_parameter_changes()
print('initial volume:',s1.volume)
print('final volume:',s2.volume)
print("percent_volume_change:",volume_change)
print("percent_lattice_change:",lattice_parameter_changes)
bond_dist = relax.get_percentage_bond_dist_changes()
connec = VoronoiConnectivity(s2)
B_233_dopant = [0]
B_233_dopant_neighbour = [144,153,162,165,180,181]
B_233_dopant_neighbour_Pb = [117,111,114,109,110]
B_233_dopant_neighbour_Pb_I =[[144,153,171,174,189,190],[147,156,165,168,183,184],[150,159,162,168,186,187],[145,154,163,166,181,182],[146,155,164,167,180,182]]
B_233 = zip(B_233_dopant_neighbour_Pb,B_233_dopant_neighbour_Pb_I)
### To determine the dopant and its neighbour I pair and bond distance
comb1 = []
for i in B_233_dopant:
for j in B_233_dopant_neighbour:
pair = []
pair.append(i)
pair.append(j)
