def do_cluster_analysis_DBSCAN(
group, cluster_cut, box, threshold_density=None, molecular=True):
""" Performs a cluster analysis using DBSCAN
:returns [labels,counts]: lists of the id of the cluster to which\
every atom is belonging to, and of the\
number of elements in each cluster.
Uses a slightly modified version of DBSCAN from sklearn.cluster
that takes periodic boundary conditions into account (through
cKDTree's boxsize option) and collects also the sizes of all
clusters. This is on average O(N log N) thanks to the O(log N)
scaling of the kdtree.
"""
if isinstance(threshold_density, type(None)):
min_samples = 2
if isinstance(threshold_density, (float, int)):
min_samples = threshold_density * 4. / 3. * np.pi * cluster_cut**3
if min_samples < 2:
min_samples = 2
# NOTE: extra_cluster_groups are not yet implemented
points = group.atoms.positions[:]
tree = cKDTree(points, boxsize=box[:6])
neighborhoods = np.array([np.array(neighbors)
for neighbors in tree.query_ball_point(
points, cluster_cut, n_jobs=-1)]
)
if len(neighborhoods.shape) != 1:
raise ValueError("Error in do_cluster_analysis_DBSCAN(), the cutoff\
is probably too small")
if molecular == False:
n_neighbors = np.array([len(neighbors)
for neighbors in neighborhoods])
else:
n_neighbors = np.array([len(np.unique(group[neighbors].resids))
for neighbors in neighborhoods])
if isinstance(threshold_density, str):
if not (threshold_density == 'auto'):
raise ValueError("Wrong value of 'threshold_density' passed\
to do_cluster_analysis_DBSCAN() ")
modes = 2
centroid, _ = vq.kmeans2(n_neighbors * 1.0, modes, iter=10,
check_finite=False)
# min_samples = np.mean(centroid)
min_samples = np.max(centroid)
labels = -np.ones(points.shape[0], dtype=np.intp)
counts = np.zeros(points.shape[0], dtype=np.intp)
core_samples = np.asarray(n_neighbors >= min_samples, dtype=np.uint8)
dbscan_inner(core_samples, neighborhoods, labels, counts)
return labels, counts, n_neighbors
def do_cluster_analysis_dbscan(group,
cluster_cut,
threshold_density=None,
molecular=True):
""" Performs a cluster analysis using DBSCAN
:returns [labels,counts,neighbors]: lists of the id of the cluster to
which every atom is belonging to, of the
number of elements in each cluster, and of
the number of neighbors for each atom
according to the specified criterion.
Uses a slightly modified version of DBSCAN from sklearn.cluster
that takes periodic boundary conditions into account (through
cKDTree's boxsize option) and collects also the sizes of all
clusters. This is on average O(N log N) thanks to the O(log N)
scaling of the kdtree.
"""
box = group.universe.dimensions[:3]
# NOTE: extra_cluster_groups are not yet implemented
points = group.atoms.positions[:]
tree = cKDTree(points, boxsize=box[:3])
neighborhoods = np.array([
np.array(neighbors)
for neighbors in tree.query_ball_point(points, cluster_cut, workers=-1)
],
dtype=object)
if len(neighborhoods.shape) != 1:
raise ValueError("Error in do_cluster_analysis_DBSCAN(), the cutoff\
is probably too small")
if molecular is False:
n_neighbors = np.array([len(neighbors) for neighbors in neighborhoods])
else:
n_neighbors = np.array([
len(np.unique(group[neighbors].resids))
for neighbors in neighborhoods
])
min_samples = determine_samples(threshold_density, cluster_cut,
n_neighbors)
labels = -np.ones(points.shape[0], dtype=np.intp)
counts = np.zeros(points.shape[0], dtype=np.intp)
core_samples = np.asarray(n_neighbors >= min_samples, dtype=np.uint8)
dbscan_inner(core_samples, neighborhoods, labels, counts)
return labels, counts, n_neighbors