def test_filtrations_d5(flag_file, filtration):
"""Test all filtrations available for dataset d5.flag, see conftest.py"""
adjacency_matrix = load_weighted_flag(flag_file, fmt='coo')
res = flagser_weighted(adjacency_matrix, max_dimension=1,
directed=False, filtration=filtration)
for filt, tests in filtrations_results.items():
if filtration == filt:
tmp = res["dgms"]
tmp2 = tests["dgms"]
assert are_matrices_equal(tmp, tmp2), \
"Diagrams {} \n and {} \n are not equal"\
.format(tmp, tmp2)
def _flagser_diagram(self, X):
Xdgms = flagser_weighted(X,
max_edge_weight=self.max_edge_weight,
min_dimension=self._min_homology_dimension,
max_dimension=self._max_homology_dimension,
directed=self.directed,
filtration=self.filtration,
coeff=self.coeff,
approximation=self.max_entries)['dgms']
if 0 in self._homology_dimensions:
Xdgms[0] = Xdgms[0][:-1, :] # Remove final death at np.inf
return Xdgms
def top_summary(graphs,dim = 7,approx = None):
'''
graphs: list of networkx directed graphs
dim: maximum dimension to calculate Betti numbers and simplex counts
approx: integer that determines the approximation used in pyflagser
This function returns a list containing the Betti numbers up to dimension dim
and another list containing the simplex counts up to dimension dim
'''
list_Betti = []
list_simplex = []
i = 0
print('Generating Topological Summary')
for g in graphs:
resbetti = np.zeros((dim,))
rescell = np.zeros((dim,))
G_matrix = nx.to_scipy_sparse_matrix(g) #returns scipy sparse matrix
out = pf.flagser_weighted(G_matrix + sp.sparse.identity(nx.number_of_nodes(g)),
min_dimension=0,
max_dimension=dim,
directed=True, approximation= approx)
betti = out['betti']
cell_count = out['cell_count']
for i in range(dim):
if i < len(betti):
resbetti[i] = betti[i]
else:
resbetti[i] = 0
for i in range(dim):
if i < len(cell_count):
rescell[i] = cell_count[i]
else:
rescell[i] = 0
list_Betti.append(resbetti)
list_simplex.append(rescell)
return(np.array(list_Betti), np.array(list_simplex))
def test_higher_coefficients():
"""Regression test for issue #45"""
x = np.random.random((5, 5))
np.fill_diagonal(x, 0.)
flagser_weighted(x, coeff=3)