def structure_featurizer(df_input: pd.DataFrame, **kwargs) -> pd.DataFrame:
"""Return a Pandas DataFrame with all structural features"""
# generate the "structure_oxid" column
df_struct = StructureToOxidStructure(**kwargs).featurize_dataframe(
df_input, col_id="structure", ignore_errors=True)
# generate features based on EwaldEnergy
ee_featurizer = EwaldEnergy()
ee_featurizer.featurize_dataframe(df_struct,
col_id="structure_oxid",
ignore_errors=True,
inplace=True)
# generate features based on the variance in bond lengths and atomic volumes (slow to run)
sh_featurizer = StructuralHeterogeneity()
sh_featurizer.featurize_dataframe(df_struct,
col_id="structure",
ignore_errors=True,
inplace=True)
# calculate global instability index
gii_featurizer = GlobalInstabilityIndex()
gii_featurizer.featurize_dataframe(df_struct,
col_id="structure_oxid",
ignore_errors=True,
inplace=True)
# rename the column from "global instability index" to "gii"
df_struct.rename(columns={"global instability index": "gii"}, inplace=True)
return df_struct
def test_GlobalInstabilityIndex(self):
# Test diamond and ni3al fail precheck
gii = GlobalInstabilityIndex(r_cut=4.0, disordered_pymatgen=False)
self.assertFalse(gii.precheck(self.diamond))
self.assertFalse(gii.precheck(self.ni3al))
# Test they raise errors when featurizing
with self.assertRaises(AttributeError):
gii.featurize(self.ni3al)
with self.assertRaises(ValueError):
gii.featurize(self.diamond)
# Ordinary case of nacl
self.assertTrue(gii.precheck(self.nacl))
self.assertAlmostEqual(gii.featurize(self.nacl)[0], 0.08491655709)
# Test bond valence sums are accurate for NaCl.
# Values are closer to 0.915 than 1.0 due to structure specified here.
# Using CollCode181148 from the ICSD, I see bond valence sums of 0.979
site1, site2 = (self.nacl[0], self.nacl[1])
neighs1 = self.nacl.get_neighbors(site1, 4)
neighs2 = self.nacl.get_neighbors(site2, 4)
site_val1 = site1.species.elements[0].oxi_state
site_el1 = str(site1.species.element_composition.elements[0])
site_val2 = site2.species.elements[0].oxi_state
site_el2 = str(site2.species.element_composition.elements[0])
self.assertAlmostEqual(gii.calc_bv_sum(site_val1, site_el1, neighs1),
0.9150834429025214)
self.assertAlmostEqual(gii.calc_bv_sum(site_val2, site_el2, neighs2),
-0.915083442902522)
# Behavior when disorder is present
gii_pymat = GlobalInstabilityIndex(r_cut=4.0, disordered_pymatgen=True)
nacl_disordered = copy.deepcopy(self.nacl)
nacl_disordered.replace_species({"Cl1-": "Cl0.5Br0.5"})
nacl_disordered.add_oxidation_state_by_element({
'Na': 1,
'Cl': -1,
'Br': -1
})
self.assertTrue(gii.precheck(nacl_disordered))
with self.assertRaises(ValueError):
gii.featurize(nacl_disordered)
self.assertAlmostEqual(
gii_pymat.featurize(nacl_disordered)[0], 0.39766464)
def test_GlobalInstabilityIndex(self):
# Test diamond and ni3al fail precheck
gii = GlobalInstabilityIndex(r_cut=4.0, disordered_pymatgen=False)
self.assertFalse(gii.precheck(self.diamond))
self.assertFalse(gii.precheck(self.ni3al))
# Test they raise errors when featurizing
with self.assertRaises(AttributeError):
gii.featurize(self.ni3al)
with self.assertRaises(ValueError):
gii.featurize(self.diamond)
# Ordinary case of nacl
self.assertTrue(gii.precheck(self.nacl))
self.assertAlmostEqual(gii.featurize(self.nacl)[0], 0.08491655709)
# Behavior when disorder is present
gii_pymat = GlobalInstabilityIndex(r_cut=4.0, disordered_pymatgen=True)
nacl_disordered = copy.deepcopy(self.nacl)
nacl_disordered.replace_species({"Cl1-": "Cl0.5Br0.5"})
nacl_disordered.add_oxidation_state_by_element({
'Na': 1,
'Cl': -1,
'Br': -1
})
self.assertTrue(gii.precheck(nacl_disordered))
with self.assertRaises(ValueError):
gii.featurize(nacl_disordered)
self.assertAlmostEqual(
gii_pymat.featurize(nacl_disordered)[0], 0.39766464)