def get_structures_from_protosearch(formulas,
source='icsd',
db_interface=None):
"""
Calls protosearch to get the hypothetical structures.
Args:
formulas ([str]): list of chemical formulas from which
to generate candidate structures
source (str): project name in OQMD to be used as source.
Defaults to ICSD.
db_interface (DbInterface): interface to OQMD database
by default uses the one pulled from data.matr.io
Returns:
(pandas.DataFrame) hypothetical pymatgen structures
generated and their unique ids from protosearch
TODO:
- For efficiency, n_max_atoms can be handled within OqmdInterface
"""
if db_interface is None:
cache_matrio_data("oqmd_ver3.db")
oqmd_db_path = os.path.join(CAMD_CACHE, "oqmd_ver3.db")
db_interface = OqmdInterface(oqmd_db_path)
dataframes = [
db_interface.create_proto_data_set(source=source,
chemical_formula=formula)
for formula in formulas
]
_structures = pd.concat(dataframes)
# Drop bad structures
_structures.dropna(axis=0, how='any', inplace=True)
# conversion to pymatgen structures
ase_adap = AseAtomsAdaptor()
pmg_structures = [
ase_adap.get_structure(_structures.iloc[i]['atoms'])
for i in range(len(_structures))
]
_structures['pmg_structures'] = pmg_structures
# The uuid below is probably an overkill. But want the strings
# to be unique. Sometimes in spaces with similar stoichiometries
# they may clash e.g. IrSb2O2 and Ir2SbO2 may end up producing
# the same string, despite different substitutions on same structure.
# We should figure out a way to get the right order from protosearch.
structure_uids = [
_structures.iloc[i]['proto_name'].replace('_', '-') + '-' +
'-'.join(pmg_structures[i].symbol_set) + '-' +
str(uuid.uuid4()).replace('-', '')[:6] for i in range(len(_structures))
]
_structures.index = structure_uids
return _structures
def test_create_proto_dataset(self):
O = OqmdInterface()
atoms_list = O.create_proto_data_set(chemical_formula='FeO6',
max_atoms=7)
assert len(atoms_list) == 5
for atoms in atoms_list["atoms"][:5]:
assert atoms.get_number_of_atoms() == 7
assert atoms.get_chemical_symbols().count('Fe') == 1
assert atoms.get_chemical_symbols().count('O') == 6
def get_structures_from_protosearch(formulas,
source="icsd",
db_interface=None):
"""
Calls protosearch to get the hypothetical structures.
Args:
formulas ([str]): list of chemical formulas from which
to generate candidate structures
source (str): project name in OQMD to be used as source.
Defaults to ICSD.
db_interface (DbInterface): interface to OQMD database
by default uses the one pulled from data.matr.io
Returns:
(pandas.DataFrame) hypothetical pymatgen structures
generated and their unique ids from protosearch
"""
if db_interface is None:
db_interface = OqmdInterface(source)
dataframes = [
db_interface.create_proto_data_set(chemical_formula=formula)
for formula in formulas
]
_structures = pd.concat(dataframes)
# Drop bad structures
_structures.dropna(axis=0, how="any", inplace=True)
# conversion to pymatgen structures
ase_adap = AseAtomsAdaptor()
pmg_structures = [
ase_adap.get_structure(_structures.iloc[i]["atoms"])
for i in range(len(_structures))
]
_structures["structure"] = pmg_structures
# This is for compatibility with Mc1, which doesn't allow
# underscores
structure_uids = [
_structures.iloc[i]["structure_name"].replace('_', '-')
for i in range(len(_structures))
]
_structures.index = structure_uids
return _structures
