def pyscal_steinhardt_parameter(
self,
neighbor_method="cutoff",
cutoff=0,
n_clusters=2,
q=None,
averaged=False,
clustering=None,
):
"""
Calculate Steinhardts parameters
Args:
neighbor_method (str) : can be ['cutoff', 'voronoi']. (Default is 'cutoff'.)
cutoff (float) : Can be 0 for adaptive cutoff or any other value. (Default is 0, adaptive.)
n_clusters (int/None) : Number of clusters for K means clustering or None to not cluster. (Default is 2.)
q (list) : Values can be integers from 2-12, the required q values to be calculated. (Default is None, which
uses (4, 6).)
averaged (bool) : If True, calculates the averaged versions of the parameter. (Default is False.)
Returns:
numpy.ndarray: (number of q's, number of atoms) shaped array of q parameters
numpy.ndarray: If `clustering=True`, an additional per-atom array of cluster ids is also returned
"""
return get_steinhardt_parameter_structure(
self._structure,
neighbor_method=neighbor_method,
cutoff=cutoff,
n_clusters=n_clusters,
q=q,
averaged=averaged,
clustering=clustering,
)
def test_steinhardt_parameters_qs(self):
"""
Test the calculation of Steinhardts parameters
"""
perfect_vals = [0.00, 0.00, 0.190, 0.00, 0.575, 0.00, 0.404, 0.00,
0.013, 0.00, 0.600]
qtest = np.random.randint(2, 13, size=2)
qs, _ = pas.get_steinhardt_parameter_structure(self.structure, cutoff=0, n_clusters=2, q=qtest)
for c, q in enumerate(qs):
self.assertLess(np.abs(np.mean(q) - perfect_vals[qtest[c]-2]), 1E-3)
def pyscal_steinhardt_parameter(self, neighbor_method="cutoff", cutoff=0, n_clusters=2,
q=(4, 6), averaged=False, clustering=True):
"""
Calculate Steinhardts parameters
Args:
neighbor_method (str) : can be ['cutoff', 'voronoi']
cutoff (float) : can be 0 for adaptive cutoff or any other value
n_clusters (int) : number of clusters for K means clustering
q (list) : can be from 2-12, the required q values to be calculated
averaged (bool) : If True, calculates the averaged versions of the parameter
clustering (bool) : If True, cluster based on the q values
Returns:
list: calculated q parameters
"""
return get_steinhardt_parameter_structure(
self._structure, neighbor_method=neighbor_method, cutoff=cutoff, n_clusters=n_clusters,
q=q, averaged=averaged, clustering=clustering
)
def test_steinhardt_parameters_clustering(self):
noisy_structure = self.structure.copy()
noisy_structure.positions += 0.5 * np.random.rand(*noisy_structure.positions.shape)
n_clusters = 3
_, inds = pas.get_steinhardt_parameter_structure(noisy_structure, n_clusters=n_clusters)
self.assertEqual(n_clusters, len(np.unique(inds)), msg='Expected to find one label for each cluster.')
def test_steinhardt_parameters_returns(self):
self.assertEqual(2, len(pas.get_steinhardt_parameter_structure(self.structure)),
msg='Expected default return value to be a tuple of qs and cluster indices.')
self.assertIsInstance(pas.get_steinhardt_parameter_structure(self.structure, n_clusters=None), np.ndarray,
msg='Expected just the qs when no clustering is used.')