def generate_fingerprints(self,
feature_methods=['voronoi'],
completed=False):
self.Workflow._connect()
ase_db = self.Workflow.ase_db
atoms_list = []
target_list = {}
ids = self.Workflow.get_new_fingerprint_ids(completed=completed)
if not ids:
print('Fingerprints up do date!')
return
targets_df = pd.DataFrame(columns=['id', 'Ef'])
for id in ids:
row = ase_db.get(id=id)
atoms_list += [row.toatoms()]
if row.get('Ef', None) is not None:
targets_df = targets_df.append({
'id': id,
'Ef': row.Ef
},
ignore_index=True)
targets_df = targets_df.astype({'id': int, 'Ef': float})
# Catlearn Magpie interface
VF = VoronoiFingerprintGenerator(atoms_list)
fingerprints_df = VF.generate().assign(id=ids)
fingerprints_df = fingerprints_df.astype({'id': int})
self.Workflow.write_dataframe(table='fingerprint', df=fingerprints_df)
self.Workflow.write_dataframe(table='target', df=targets_df)
def test_voronoi_fp_gen(self):
"""Test the feature generation."""
alist = self.get_data()
voro = VoronoiFingerprintGenerator(alist)
features = voro.generate()
self.assertEqual(np.shape(features)[0], 10)
def __init__(self, atoms_list):
"""Voronoi fingerprinting setup.
Parameters
----------
atoms_list : list
A list of ase atoms objects to be featurized.
Development Notes:
------------------
* This featurizing method is specific to atoms objects, for the sake
of generality lets just assume that inputs can be anything, therefore
check that inputs are atoms objects
"""
self.check_inputs(atoms_list)
self.Voro_inst = VoronoiFingerprintGenerator(atoms_list,
delete_temp=False)
class VoronoiFingerprint:
"""
Uses the CatLearn interface to MagPie Voronoi Fingerprint
"""
from catlearn.fingerprint.voro import VoronoiFingerprintGenerator
def __init__(self, atoms_list):
"""Voronoi fingerprinting setup.
Parameters
----------
atoms_list : list
A list of ase atoms objects to be featurized.
Development Notes:
------------------
* This featurizing method is specific to atoms objects, for the sake
of generality lets just assume that inputs can be anything, therefore
check that inputs are atoms objects
"""
self.check_inputs(atoms_list)
self.Voro_inst = VoronoiFingerprintGenerator(atoms_list,
delete_temp=False)
def check_inputs(self, atoms_list):
# check atoms list
from ase import Atoms
type_check_list = [
isinstance(atoms_i, Atoms) for atoms_i in atoms_list
]
err_mess = "Inputs to Voronoi must be atom objects"
assert all(type_check_list), err_mess
def generate_fingerprints(self):
self.features = self.Voro_inst.generate()
def get_fingerprints(self):
self.generate_features()
return self.features
# Check the size of the atomic strucures.
size = []
for a in alist:
size.append(len(a))
print('min: {0}, mean: {1:.0f}, max: {2} atoms size'.format(
min(size), sum(size)/len(size), max(size)))
# Now we generate the Voronoi fingerprints for our atoms objects. `voro.generate()` returns a Pandas dataframe.
# In[4]:
voro = VoronoiFingerprintGenerator(alist)
data_frame = voro.generate()
# In cases, where the generated featues does not apply to the input data, `NaN`s are returned.
# There are various ways of filling in this kind of data and the simplest is simply to remove features containing infinite values.
#
# The conventional data format in CatLearn is a matrix, so we first convert the Pandas dataframe into a numpy array.
# In[5]:
matrix = data_frame.to_numpy()
finite_numeric_data = clean_infinite(matrix)
print(np.shape(finite_numeric_data['train']))