def subgraphIsomorphismCheck(self, G1, G2):
"""
Checks whether G1 contains a subgraph isomorphic to G2 by using Whitney isomorphism theorem.
Parameters:
G1 (NetworkX graph): The bigger graph.
G2 (NetworkX graph): The smaller graph.
Returns:
bool: Is graph G2 isomorphic to a subgraph in G1.
isomorphism.GraphMatcher: Graph matcher object with parameters.
"""
# transform graphs into line graphs and check for subgraph isomorphism
# isomorphism.GraphMatcher tries to find an induced subgraph of G1, such that it is isomorphic to G2.
# Consequently, if G2 is non-induced subgraph of G1, the algorithm will return False
GM = isomorphism.GraphMatcher(nx.line_graph(G1), nx.line_graph(G2))
subgraph_is_iso = GM.subgraph_is_isomorphic()
# check for exceptions
# e.g. line graphs of K_3 triangle graph and K_1,3 claw graph are isomorphic, but the original graphs are not
if subgraph_is_iso:
edgeListG1 = []
edgeListG2 = []
for edgeMaping in GM.mapping.items():
edgeListG1.append(edgeMaping[0])
edgeListG2.append(edgeMaping[1])
# let's construct the graphs the algorithm thinks are isomorphic and check them for a quick isomorphism.is_isomorphic
testG1 = nx.Graph(edgeListG1)
testG2 = nx.Graph(edgeListG2)
subgraph_is_iso = isomorphism.is_isomorphic(testG1, testG2)
return subgraph_is_iso, GM
def centrality(file_graphml, road_type, place_country, bc=False, cc=False):
# load grapho
grafo = ox.load_graphml(file_graphml)
#### replace "length" values with "cost" values #####
# for u, v, key, attr in grafo.edges(keys=True, data=True):
# # print(attr)
# print(attr["length"])
# attr['length'] = attr.get("cost")
# # grafo.add_edge(u, v, key, attr_dict=attr)
# grafo.add_edge(u, v, key)
# ox.extended_stats(grafo, bc=True)
if cc:
c_name = "close_centrality"
edge_centrality = nx.closeness_centrality(nx.line_graph(grafo))
if bc:
c_name = "btw_centrality"
edge_centrality = nx.betweenness_centrality(nx.line_graph(grafo), weight='pippo') # not working!!!!!
ev = [edge_centrality[edge + (0,)] for edge in grafo.edges()]
# color scale converted to list of colors for graph edges
norm = colors.Normalize(vmin=min(ev)*0.8, vmax=max(ev))
# cividis, viridis, YlGn (good colormaps
# 'Greys', 'Purples', 'Blues', 'Greens', 'Oranges', 'Reds',
# 'YlOrBr', 'YlOrRd', 'OrRd', 'PuRd', 'RdPu', 'BuPu',
# 'GnBu', 'PuBu', 'YlGnBu', 'PuBuGn', 'BuGn', 'YlGn',
# 'viridis', 'plasma', 'inferno', 'magma', 'cividis']
cmap = cm.ScalarMappable(norm=norm, cmap=cm.YlGn)
ec = [cmap.to_rgba(cl) for cl in ev]
fig, ax = ox.plot_graph(grafo, bgcolor='k', axis_off=True, node_size=0, node_color='w',
node_edgecolor='gray', node_zorder=2,
edge_color=ec, edge_linewidth=1.5, edge_alpha=1)
gdf_edges = ox.graph_to_gdfs(grafo, nodes=False, fill_edge_geometry=True)
gdf_edges['edge_color'] = ec
my_map = plot_graph_folium_FK(gdf_edges, graph_map=None, popup_attribute=None,
zoom=1, fit_bounds=True, edge_width=2, edge_opacity=1) #tiles='cartodbpositron'
name_place_country = re.sub('[/," ",:]', '_', place_country)
road_type = road_type.replace(' ', '')
road_type = list(road_type.split(","))
roadtype = ' '.join([str(elem) for elem in road_type])
roads = re.sub('[/," ",:]', '_', roadtype)
my_map.save(c_name + "_" + roads + "_" + name_place_country + ".html")
#############################################################
#############################################################
# road_type = "motorway, motorway_link"
# # road_type = "secondary"
# # road_type = "motorway, motorway_link, secondary, primary, tertiary"
# place_country = "Catania, Italy"
# # file_graphml = 'Catania__Italy.graphml' # to be used when run cost assignment and shortest path calculation (no saving!!)
# file_graphml = 'Catania__Italy_cost.graphml' # to be used when run folium map classification and centrality
# distance = 20000
# bc=False
# cc=True
def partition_graph(aa,bb,gg):
E=gg.number_of_edges()
gl=nx.line_graph(gg)
lbl={}
for i,g in enumerate(gg.edges()):
lbl[g]=i
gl=nx.relabel_nodes(gl,lbl)
ad=nx.all_pairs_shortest_path_length(gl)
am=np.zeros((E,E))
for ni,n in enumerate(gl.nodes()):
for mi,m in enumerate(gl.nodes()):
if(n in ad.keys()):
if(m in ad[n].keys()):
am[ni,mi]=ad[n][m]
am[mi,ni]=ad[n][m]
M=int(am.max()+0.5)+1
am[am==0]=M
hh=np.zeros((M,2,2))
for e0 in range(E):
for e1 in range(e0):
xa=np.max([(e0 in i)+(e1 in i) for i in aa])
xb=np.max([(e0 in i)+(e1 in i) for i in bb])
di=int(am[e0,e1])-1
if(xa==2 and xb==2):
hh[di,0,0]+=1.0
elif(xa<2 and xb==2):
hh[di,1,0]+=1.0
elif(xa==2 and xb<2):
hh[di,0,1]+=1.0
elif(xa<2 and xb<2):
hh[di,1,1]+=1.0
return hh
def get_contains_cases(_test_suite, _series_map: Dict[Type[VisionsBaseType],
Set[str]], typeset):
"""Parametrize contains tests
Args:
mapping: mapping from type to a set of series' identifiers
Returns:
the args for the generated tests
"""
# Include children's series in parent
reversed_topological_edge_sort = list(
reversed(list(nx.topological_sort(nx.line_graph(typeset.base_graph)))))
for parent, child in reversed_topological_edge_sort:
_series_map[parent] |= _series_map[child]
all_series_included(_test_suite, _series_map)
argsvalues = []
for item in _test_suite:
for type, series_list in _series_map.items():
args = {"id": f"{item.name} x {type}"}
member = item.name in series_list
argsvalues.append(pytest.param(item, type, member, **args))
return {"argnames": "series,type,member", "argvalues": argsvalues}
def max_span_trail(D, G, input_arc, cache, reactions):
"""Given a directed graph D and an undirected graph G built on the same
vertex set and in input arc in D, returns a trail of maximum span in D that
passes through the input arc, such that its vertex set induces a connected
subgraph in G.
The cache dict is modified, storing best partial paths in every SCC of LD
between all possible pairs of entry and exit points from predecessor SCCs to
successor SCCs.
:param D: input directed graph for the HNet algorithm, with the same vertex
set as G and representing a metabolic pathway
:param G: input undirected graph for the HNet algorithm, with the same
vertex set as D and representing gene neighborhood (in terms of
reactions; see Model in the methods section of Zaharia et al., 2018)
:param input_arc: input arc in D on which the HNet algorithm is executed
:param cache: dict storing best partial paths for every strongly connected
component (SCC) of the line graph of D
:param reactions: dict of dicts storing reaction information (obtained by
parsing a KGML file)
:return: a trail of maximum span in D such that its vertex set induces a
connected subgraph in G
"""
LD = nx.line_graph(D)
if nx.number_of_nodes(LD) == 0:
return list()
access = AccessPoints(LD) # access points for every SCC in the line graph
partial_paths(LD, access, cache, reactions)
return get_corresponding_trail(
max_span_path(LD, G, input_arc, cache, reactions))
def _complete_summaries(self):
for from_type, to_type in nx.topological_sort(
nx.line_graph(self.typeset.base_graph)
):
self.summary_map[to_type] = (
self.summary_map[from_type] + self.summary_map[to_type]
)
def vertexes_to_edges_graph(self, curr_path, v_graph):
"""
Description:
Converts vertex to edges in a graph.
Input:
curr_path - path of a current graphml file
v_graph - the vertex graph
Output:
e_graph - edges graph.
"""
xmldoc = minidom.parse(curr_path)
itemlist = xmldoc.getElementsByTagName('edge')
e_graph = nx.line_graph(v_graph)
if itemlist[0].firstChild==None:
ds=nx.degree_centrality(v_graph)
nx.set_node_attributes(v_graph,ds,'label')
return v_graph
for item in itemlist:
value = float(item.firstChild.TEXT_NODE)
source = item.attributes['source'].value
target = item.attributes['target'].value
if (source, target) not in e_graph.nodes.data():
e_graph.add_node((source, target))
e_graph.nodes.data()[(source, target)].update({'label': value})
[x[1].update({'value': 0}) for x in e_graph.nodes.data() if len(x[1]) == 0]
return e_graph
def __init__(self, g, features, n_machines, radius, activation, device):
super().__init__()
adj = nx.adj_matrix(g)
p = my.adj2p(sp.sparse.triu(adj))
adj = tf.cast(my.sparse_sp2tf(adj), tf.float32)
deg = tf.expand_dims(tf.sparse_reduce_sum(adj, 1), 1)
lg = nx.line_graph(g)
adj_lg = tf.cast(my.sparse_sp2tf(nx.adj_matrix(lg)), tf.float32)
deg_lg = tf.expand_dims(tf.sparse_reduce_sum(adj_lg, 1), 1)
for i, (m, n) in enumerate(zip(features[:-1], features[1:])):
setattr(
self, 'layer%d' % i,
GNNLayer(m, n, adj, deg, adj_lg, deg_lg, p, radius,
activation))
self.n_layers = i + 1
self.dense = tf.keras.layers.Dense(input_shape=(n, ), units=n_machines)
self.device = device
with tf.device(device):
x = deg
x -= tf.reduce_mean(x)
x /= tf.sqrt(tf.reduce_mean(tf.square(x)))
y = deg_lg
y -= tf.reduce_mean(y)
y /= tf.sqrt(tf.reduce_mean(tf.square(y)))
self.x, self.y = x, y
def lineDigraph(n, temp): # L(multidigraph)を出力
# multigraph
V = range(1, n + 1)
G = nx.MultiDiGraph()
G.add_nodes_from(V)
for (i, j) in temp:
if (temp[i, j] > 0):
for k in range(0, temp[i, j]):
G.add_edge(i, j)
# LineDigraph of G
L = nx.line_graph(G)
#adjance matrix : L.edges() → m
n = len(L.nodes())
v = {}
atai = 1
for i in L.nodes():
v[i] = atai
atai += 1
m = {}
for i in range(1, n + 1):
for j in range(1, n + 1):
m[i, j] = 0
for (i, j) in L.edges():
m[v[i], v[j]] = 1
return n, m
def line_graph(layer,
target_state,
hidden_nodes=None,
not_traversable_states=[]):
graph, hidden_nodes = _traversability_graph(
layer,
hidden_nodes=hidden_nodes,
not_traversable_states=not_traversable_states)
line_graph = nx.line_graph(graph)
line_graph.edges_in_state = collections.defaultdict(list)
for e in line_graph.edges(data=True):
n1, n2, data = e
(_, so, ti1) = n1
(_, _, ti2) = n2
s = _state_for_line_graph_edge(e)
p1 = layer.transitions[ti1].geometry.centroid
p2 = layer.transitions[ti2].geometry.centroid
data['length'] = p1.distance(p2)
line_graph.edges_in_state[s].append(e[:2])
target = _add_to_line_graph(layer, graph, line_graph, to_node=target_state)
hidden_state = [line_graph.edges_in_state[node] for node in hidden_nodes]
for n, data in line_graph.nodes(data=True):
data['observe'] = []
for i, edges in enumerate(hidden_state):
for x, y in edges:
line_graph[x][y][0]['hidden'] = True
line_graph.node[x]['observe'].append(i)
line_graph.node[y]['observe'].append(i)
return graph, nx.Graph(line_graph), target, hidden_state
def line_graphs():
print("Line graph")
G = nx.star_graph(3)
L = nx.line_graph(G)
print(sorted(map(sorted, L.edges()))) # makes a 3-clique, K3
draw_graph(G)
draw_graph(L)
def test_is_berge(pool : Pool, alt=False):
if alt:
from alternate import is_berge_alt as is_berge
else:
from berge import is_berge
# Note that graphs are perfect iff they are Berge
for i in range(5):
# Bipartite graphs are always Berge
n1, n2 = random.randint(1, 12), random.randint(1, 12)
graph = random_bipartite(n1=n1, n2=n2, p=.4)
assert(is_berge(graph, pool=pool))
for i in range(5):
graph = nx.line_graph(random_bipartite(n1=10, n2=10, p=.15))
# Line graphs of bipartite graphs are perfect by Konig's theorem
assert(is_berge(graph, pool=pool))
for i in range(10, 15):
assert(is_berge(nx.complete_graph(i), pool=pool))
for i in range(5):
# Make sure we work properly on disconnected graphs
graph = nx.disjoint_union_all([
random_bipartite(
random.randint(1, 6),
random.randint(1, 6), .2)
for i in range(3)])
assert(is_berge(graph, pool=pool))
for i in range(5):
m = random.randint(2, 12)
graph = nx.triangular_lattice_graph(m, 2)
assert(is_berge(graph, pool=pool))
for i in range(5):
n = random.randint(4, 20)
graph = random_chordal(n, .2)
assert(is_berge(graph, pool=pool))
for i in range(10):
n = random.randint(4, 20)
graph = nx.cycle_graph(n)
assert(is_berge(graph, pool=pool) == (n % 2 == 0))
def get_contains_cases(
_test_suite: Dict[str, Sequence],
_series_map: Dict[T, Set[str]],
typeset: VisionsTypeset,
):
"""Parametrize contains tests
Args:
_test_suite: mapping from sequence identifiers to sequences
_series_map: mapping from type to a set of sequence identifiers
typeset: A VisionsTypeset
Returns:
the args for the generated tests
"""
# Include children's series in parent
reversed_topological_edge_sort = list(
reversed(list(nx.topological_sort(nx.line_graph(typeset.base_graph))))
)
for parent, child in reversed_topological_edge_sort:
_series_map[parent] |= _series_map[child]
all_series_included(_test_suite, _series_map)
argsvalues = []
for name, item in _test_suite.items():
for type, series_list in _series_map.items():
args = {"id": f"{name} x {type}"}
member = name in series_list
argsvalues.append(pytest.param(name, item, type, member, **args))
return {"argnames": ["name", "series", "type", "member"], "argvalues": argsvalues}
def __init__(self, root, N=100, deg=2, p=0.3, seed=123):
"""
Args:
# G: network topology
root: dir for save G and edge_timestamps
N: original graph, with N nodes
deg: original graph, with average degree=deg.
The original graph should have N nodes, N*deg edges.
For timestamp simulation, the number of 'nodes' should be N*deg. Because we regard each edge of the original graph as one event 'type'.
But we share a same model for all edges.
And the 'adjacency' for tick should be [N*deg, N*deg]
p: rewire probability
seed: random seed
"""
self.root = root
self.N = N
self.deg = deg
self.p = p
self.seed = seed
self.G = self.generate_G(self.N, self.deg, self.p, seed=self.seed)
self.G_e2n = nx.line_graph(self.G)
self.elabel_idx_map = dict(
zip(self.G_e2n.nodes(), range(self.G_e2n.number_of_nodes())))
self.idx_elabel_map = dict(
zip(range(self.G_e2n.number_of_nodes()), self.G_e2n.nodes()))
self.hawkes = []
def to_graph(self, personalization=None):
''' builds document graph from several link types '''
svos = self.svos
g = nx.DiGraph()
for e in self.to_edges():
f, t = e
g.add_edge(f, t)
if self.params.svo_edges:
for s, v, o in svos:
if s == o: continue
if v == 'as_in':
g.add_edge(s, o)
else:
g.add_edge(o, s)
if personalization == None and self.params.pers_idf:
personalization = self.pers_from_freq(get_freqs())
pr = nx.pagerank(g, personalization=personalization)
if self.params.use_line_graph and g.number_of_edges() < 20000:
lg = nx.line_graph(g)
lpr = nx.pagerank(lg)
for xy, r in lpr.items():
x, y = xy
if isinstance(x, str) and isinstance(y, str):
pr[x] = pr[x] + r
pr[y] = pr[y] + r
return g, pr
def to_line_graph(data: Data, directed: bool = True) -> Data:
"""
Convert a graph G to its corresponding line-graph L(G)
Args:
data: a torch_gemoetric Data object representing representing a graph
directed: whether the original graph is directed or undirected
"""
original_edge_attrs = data.edge_attr
original_edge_names = [
(from_.item(), to_.item())
for from_, to_ in zip(data.edge_index[0, :], data.edge_index[1, :])
]
original_edge_to_attr = {
e: attr
for e, attr in zip(original_edge_names, original_edge_attrs)
}
ctor = nx.DiGraph if directed else nx.Graph
G = to_networkx(data,
node_attrs=['x'],
edge_attrs=['edge_attr'],
to_undirected=not directed)
line_graph = nx.line_graph(G, create_using=ctor)
res_data = from_networkx(line_graph)
# Copy original attribtues
res_data.x = torch.stack(
[original_edge_to_attr[e] for e in line_graph.nodes])
res_data.y = data.y
return data
def __init__(self, mol, diameter=8, ignore_hydrogen=True, timeout=5):
mol.require("Valence")
self.diam = diameter
self.ignoreh = ignore_hydrogen
self.timeout = timeout
# Results
self.array = []
self.max_size = 0
self.int_to_node = {}
self.elapsed_time = 0
self.valid = False
if len(mol) < 3:
return
start_time = time.perf_counter()
self.mol = mol
self.preprocess()
# Generate line graph and reindexing
lg = nx.line_graph(self.mol.graph)
node_to_int = {}
for i, ln in enumerate(lg.nodes()):
node_to_int[ln] = i
lg.nodes[ln]["type"] = self.node_desc(ln)
self.int_to_node = {v: k for k, v in node_to_int.items()}
self.graph = nx.relabel_nodes(lg, node_to_int)
# Edges
edges = []
for u, v, attr in self.edge_gen():
edges.append((u, v))
self.array.append((u, v, self.edge_desc(attr)))
# Max fragment size determination
fcres = find_cliques(self.graph.nodes(), edges, timeout=timeout)
self.max_size = len(fcres["max_clique"])
self.elapsed_time = round(time.perf_counter() - start_time, 7)
self.valid = not fcres["timeout"]
def get_qubit_registers_for_adder(qc: QuantumComputer, num_length: int,
qubits: Optional[Sequence[int]] = None)\
-> Tuple[Sequence[int], Sequence[int], int, int]:
"""
Searches for a layout among the given qubits for the two n-bit registers and two additional
ancilla that matches the simple layout given in figure 4 of [CDKM96]_.
This method ignores any considerations of physical characteristics of the qc aside from the
qubit layout. An error is thrown if the appropriate layout is not found.
:param qc: the quantum resource on which an adder program will be executed.
:param num_length: the length of the bitstring representation of one summand
:param qubits: the available qubits on which to run the adder program.
:return: the necessary registers and ancilla labels for implementing an adder
program to add the numbers a and b. The output can be passed directly to :func:`adder`
"""
if qubits is None:
unavailable = [] # assume this means all qubits in qc are available
else:
unavailable = [qubit for qubit in qc.qubits() if qubit not in qubits]
graph = qc.qubit_topology().copy()
for qubit in unavailable:
graph.remove_node(qubit)
# network x only provides subgraph isomorphism, but we want a subgraph monomorphism, i.e. we
# specifically want to match the edges desired_layout with some subgraph of graph. To
# accomplish this, we swap the nodes and edges of graph by making a line graph.
line_graph = nx.line_graph(graph)
# We want a path of n nodes, which has n-1 edges. Since we are matching edges of graph with
# nodes of layout we make a layout of n-1 nodes.
num_desired_nodes = 2 * num_length + 2
desired_layout = nx.path_graph(num_desired_nodes - 1)
g_matcher = nx.algorithms.isomorphism.GraphMatcher(line_graph,
desired_layout)
try:
# pick out a subgraph isomorphic to the desired_layout if one exists
# this is an isomorphic mapping from edges in graph (equivalently nodes of line_graph) to
# nodes in desired_layout (equivalently edges of a path graph with one more node)
edge_iso = next(g_matcher.subgraph_isomorphisms_iter())
except IndexError:
raise Exception(
"An appropriate layout for the qubits could not be found among the "
"provided qubits.")
# pick out the edges of the isomorphism from the original graph
subgraph = nx.Graph(graph.edge_subgraph(edge_iso.keys()))
# pick out an endpoint of our path to start the assignment
start_node = -1
for node in subgraph.nodes:
if subgraph.degree(node) == 1: # found an endpoint
start_node = node
break
return assign_registers_to_line_or_cycle(start_node, subgraph, num_length)
def generateLabeledLineGraph(G): lineGraph=nx.line_graph(G) for vertexIndex in lineGraph: lineGraph.node[vertexIndex]['label']=(G.node[vertexIndex[0]]['label'],G.node[vertexIndex[1]]['label']) for n,nbrsdict in lineGraph.adjacency_iter(): for nbr,eattr in nbrsdict.items(): lineGraph.edge[n][nbr]['label']=G.node[findCommonNode(n,nbr)]['label'] return lineGraph
def test(graph):
if (fetch.is_line_graph(graph, G1, G2, G3, G4, G5, G6, G7, G8, G9)):
line_graph = nx.line_graph(graph)
euler = fetch.euler_cycle(line_graph)
if (euler != False):
return list(euler)
return euler
return (fetch.is_line_graph(graph, G1, G2, G3, G4, G5, G6, G7, G8, G9))
def seed_graph(M, connected=False):
if connected: suffix = "c"
else: suffix = "d1"
for idx, seed in enumerate(
nx.read_graph6(f"data/undirected/{M}{suffix}.g6")):
yield nx.convert_node_labels_to_integers(nx.line_graph(seed))
def edge_distances(self, e):
""" Return vector of shortest path edge distances between e
and all other edges
"""
L = nx.line_graph(self.G)
path_lens = nx.shortest_path_length(L, source=e)
return np.array([path_lens[f] for f in self.G.edges_iter()])
def generateLabeledLineGraph(G): lineGraph=nx.line_graph(G) for vertexIndex in lineGraph: lineGraph.node[vertexIndex]['label']=(G.node[vertexIndex[0]]['label'],G.node[vertexIndex[1]]['label']) for n,nbrsdict in lineGraph.adjacency_iter(): for nbr,eattr in nbrsdict.items(): lineGraph.edge[n][nbr]['label']=G.node[findCommonNode(n,nbr)]['label'] return lineGraph
def centralize_graph(graph, epb='lgth', efb='capa', ndg='capa', nec='capa', npr='capa'):
"""Compute edge centralities.
Parameters
----------
graph : original graph
epb : edge property used for computation of edge path betweenness
efb : " flow betweenness
ndg : " degree centrality
nec : " eigenvector centrality
npr : " page rank
Returns
-------
graphCentralities : graph with computed edge centralities
"""
graphCentralities = graph.copy()
edges = graphCentralities.edges(data=True)
edgeCapacity = 1.0 * np.array([property['capa'] for node1, node2, property in edges])
edgeCapacity /= edgeCapacity.sum()
edgeLength = 1.0 / edgeCapacity
for index, (node1, node2, property) in enumerate(edges):
property['capa'] = edgeCapacity[index]
property['lgth'] = edgeLength[index]
edgeBetweenCentrality = nx.edge_betweenness_centrality(graphCentralities, weight=epb)
edgeFlowBetweennessCentrality = nx.edge_current_flow_betweenness_centrality(graphCentralities, weight=efb)
lineGraph = nx.line_graph(graphCentralities)
degree = graphCentralities.degree(weight=ndg)
for node1, node2, property in lineGraph.edges(data=True):
intersectingNodes = list(set(node1).intersection(node2))[0]
property[ndg] = degree[intersectingNodes]
eigenvectorCentrality = nx.eigenvector_centrality_numpy(lineGraph, weight=ndg)
pageRank = nx.pagerank(lineGraph, weight=ndg)
degreeCentrality = dict(lineGraph.degree(weight=ndg))
for index, (node1, node2, property) in enumerate(edges):
edge = (node1, node2)
if (edge in edgeBetweenCentrality.keys()):
property['epb'] = edgeBetweenCentrality[edge]
else:
property['epb'] = edgeBetweenCentrality[edge[::-1]]
if (edge in edgeFlowBetweennessCentrality.keys()):
property['efb'] = edgeFlowBetweennessCentrality[edge]
else:
property['efb'] = edgeFlowBetweennessCentrality[edge[::-1]]
if (edge in degreeCentrality.keys()):
property['ndg'] = degreeCentrality[edge]
else:
property['ndg'] = degreeCentrality[edge[::-1]]
if (edge in eigenvectorCentrality.keys()):
property['nec'] = eigenvectorCentrality[edge]
else:
property['nec'] = eigenvectorCentrality[edge[::-1]]
if (edge in pageRank.keys()):
property['npr'] = pageRank[edge]
else:
property['npr'] = pageRank[edge[::-1]]
return(graphCentralities)
def networkx_succession_diagram_motif_based(ar,
include_attractors_in_diagram=True
):
"""Label the succesion diagram and (optionally) attractors of the input attractor
repertoire according to the conventions of Zanudo and Albert (2015). Useful
for plotting.
Parameters
----------
ar : AttractorRepertoire
Attractor repertoire object for which to build the diagram.
include_attractors_in_diagram : bool
Whether attractors should be represented as nodes in the diagram (the
default is True).
Returns
-------
networkx.DiGraph
A labeled digraph that represents the succession diagram.
"""
G_reduced_network_based = networkx_succession_diagram_reduced_network_based(
ar, include_attractors_in_diagram=False)
G_motif_based = nx.line_graph(G_reduced_network_based)
for i, j in G_motif_based.nodes():
node_motif = set([
frozenset(k.items()) for k in
ar.succession_diagram.motif_reduction_dict[j].motif_history
]) - set([
frozenset(k.items()) for k in
ar.succession_diagram.motif_reduction_dict[i].motif_history
])
node_label = format_reduction_label(str(dict(list(node_motif)[0])))
G_motif_based.nodes[(i, j)]['label'] = node_label
G_motif_based.nodes[(i,
j)]['virtual_nodes'] = dict(list(node_motif)[0])
if include_attractors_in_diagram:
for a_index, a in enumerate(ar.attractors):
G_motif_based.add_node('A' + str(a_index))
G_motif_based.nodes[
'A' + str(a_index)]['label'] = format_reduction_label(
str(a.attractor_dict))
G_motif_based.nodes[
'A' + str(a_index)]['virtual_nodes'] = a.attractor_dict
for r in a.reductions:
r_key = list(
ar.succession_diagram.motif_reduction_dict.keys())[list(
ar.succession_diagram.motif_reduction_dict.values(
)).index(r)]
for n in G_motif_based.nodes():
if type(n) == tuple:
i, j = n
if r_key == j:
G_motif_based.add_edge((i, j), 'A' + str(a_index))
return G_motif_based
def build_line_graph(people):
"""
Edge coloring and Vizing's theorem solution
can be found from Stack Overflow question below
ref: https://stackoverflow.com/questions/51758406/creating-time-schedule-from-list-of-people-and-who-they-have-to-meet
"""
G = nx.Graph()
G.add_edges_from(((p, q) for p, L in people for q in L))
return nx.line_graph(G)
def test_line(self):
G=nx.star_graph(5)
L=nx.line_graph(G)
assert_true(nx.is_isomorphic(L,nx.complete_graph(5)))
G=nx.path_graph(5)
L=nx.line_graph(G)
assert_true(nx.is_isomorphic(L,nx.path_graph(4)))
G=nx.cycle_graph(5)
L=nx.line_graph(G)
assert_true(nx.is_isomorphic(L,G))
G=nx.DiGraph()
G.add_edges_from([(0,1),(0,2),(0,3)])
L=nx.line_graph(G)
assert_equal(L.adj, {})
G=nx.DiGraph()
G.add_edges_from([(0,1),(1,2),(2,3)])
L=nx.line_graph(G)
assert_equal(sorted(L.edges()), [((0, 1), (1, 2)), ((1, 2), (2, 3))])
def genereate_line_graph(self):
start_time = time.time()
print("start generating line graph...")
self.line_graph = nx.line_graph(self.graph)
nx.write_gpickle(self.line_graph,
os.path.join("cache", "line_graph.gpickle"))
if self.verbose:
print("write line_graph in cache, using time",
time.time() - start_time)
def test_line(self):
G = nx.star_graph(5)
L = nx.line_graph(G)
assert_true(nx.is_isomorphic(L, nx.complete_graph(5)))
G = nx.path_graph(5)
L = nx.line_graph(G)
assert_true(nx.is_isomorphic(L, nx.path_graph(4)))
G = nx.cycle_graph(5)
L = nx.line_graph(G)
assert_true(nx.is_isomorphic(L, G))
G = nx.DiGraph()
G.add_edges_from([(0, 1), (0, 2), (0, 3)])
L = nx.line_graph(G)
assert_equal(L.adj, {})
G = nx.DiGraph()
G.add_edges_from([(0, 1), (1, 2), (2, 3)])
L = nx.line_graph(G)
assert_equal(sorted(L.edges()), [((0, 1), (1, 2)), ((1, 2), (2, 3))])
def load_data_for_city(city, root_dir='.'):
speeds_filename = 'movement-speeds-quarterly-by-hod-%s-2019-Q2.csv.zip' % city
graph_filename = '%s.gpickle.gz' % city
graph = nx.read_gpickle(os.path.join(root_dir, graph_filename))
print('OSM MultiDiGraph has %d nodes, %d edges' %
(nx.number_of_nodes(graph), nx.number_of_edges(graph)))
graph = nx.Graph(graph)
print('OSM Graph has %d nodes, %d edges' %
(nx.number_of_nodes(graph), nx.number_of_edges(graph)))
speeds = pd.read_csv(os.path.join(root_dir, speeds_filename))
# Print basic stats on the data.
speeds_num_rows = len(speeds)
speeds_num_distinct_segment_ids = speeds['segment_id'].nunique()
print('Speeds has %d rows, %d distinct segment IDs' %
(speeds_num_rows, speeds_num_distinct_segment_ids))
# Get p85 before dropping hours
p85 = speeds.groupby('osm_way_id').mean()['speed_mph_p85']
# Drop speeds with hour not in 7-10
speeds_to_drop = (speeds['hour_of_day'] < 7) | (speeds['hour_of_day'] > 10)
speeds.drop(speeds[speeds_to_drop].index, inplace=True)
print('Dropped %d/%d Uber speeds with hour not in 7-10' % (speeds_to_drop.sum(), len(speeds_to_drop)))
# For each OSM way ID, we'll just take the average of all the rows present
speeds = speeds.groupby('osm_way_id').mean()[ ['speed_mph_mean', 'speed_mph_stddev']]
speeds = speeds.join(p85)
print('After processing, %d distinct OSM way IDs in the speeds dataset' % len(speeds))
orig_graph = graph
orig_speeds = speeds
graph, speeds = merge_uber_osm_data(graph, speeds)
print('After merging, graph has %d nodes, %d edges' %
(nx.number_of_nodes(graph), nx.number_of_edges(graph)))
# Do this after merging
# Compute 1 - mean / p85 as the measure of traffic
speeds['traffic'] = (1 - speeds['speed_mph_mean'] / speeds['speed_mph_p85']).clip(0.0, 1.0)
# Group traffic into 5 classes like uber does
speeds['traffic_class'] = speeds['traffic'].floordiv(0.2).astype('int')
# Add traffic as edge attribute in the graph
for v1, v2, edge in graph.edges(data=True):
edge['traffic'] = speeds.loc[edge['osmid'], 'traffic']
edge['traffic_class'] = speeds.loc[edge['osmid'], 'traffic_class']
data = CityData(
orig_speeds = orig_speeds,
speeds = speeds,
orig_graph = orig_graph,
graph = graph,
gmmc = nx.algorithms.community.greedy_modularity_communities(graph),
line_graph = nx.line_graph(graph),
)
print('Greedy modularity maximization produced %d communities' % len(data.gmmc))
print('Line graph has %d nodes, %d edges' %
(nx.number_of_nodes(data.line_graph), nx.number_of_edges(data.line_graph)))
return data
def make_graph(self, edges, edge_costs):
'''Make a graph object'''
graph = nx.Graph()
for edge in edges:
vertex_u, vertex_v = (edge.start,
edge.end) if edge.start < edge.end else (
edge.end, edge.start)
graph.add_edge(vertex_u, vertex_v, weight=edge.cost)
# Graph which has edges as vertices of 'graph' and vertices as edges of 'graph'
alt_graph = nx.line_graph(graph)
# Original vertex pair indexes to original edges
alt_vertix_index_to_original_edges = {}
# Original edges (pairs index) to original vertex pair indexes
original_edge_to_alt_vertex_index = {}
for index, original_edge in enumerate(alt_graph.nodes()):
# The vertex is an original edge and pair of original vertex
vertex_u, vertex_v = original_edge
edge = (vertex_u, vertex_v) if vertex_u < vertex_v else (vertex_v,
vertex_u)
alt_vertix_index_to_original_edges[index] = edge
original_edge_to_alt_vertex_index[(vertex_u, vertex_v)] = index
original_edge_to_alt_vertex_index[(vertex_v, vertex_u)] = index
alt_edges = {}
max_weight = 0
for index, alt_edge in enumerate(alt_graph.edges()):
edge0_index = original_edge_to_alt_vertex_index[alt_edge[0]]
edge1_index = original_edge_to_alt_vertex_index[alt_edge[1]]
weight = (edge_costs[alt_edge[0]] + edge_costs[alt_edge[1]]) / 2.0
alt_edges[(edge0_index, edge1_index)] = weight
alt_edges[(edge1_index, edge0_index)] = weight
max_weight = max(weight, max_weight)
size = len(alt_graph.nodes())
penalty = size * max_weight * 3
edge_distances = []
for from_index in range(0, size):
row = []
for to_index in range(0, size):
edge0 = (from_index, to_index)
edge1 = (to_index, from_index)
if from_index == to_index:
weight = 0.0
elif edge0 in alt_edges:
weight = alt_edges[edge0]
elif edge1 in alt_edges:
weight = alt_edges[edge1]
else:
weight = penalty
row.append(weight)
edge_distances.append(row)
return graph, alt_vertix_index_to_original_edges, CreateDistanceCallback(
edge_distances)
def __init__(self, *args, **kwargs):
super().__init__(*args, **kwargs)
# Weighted dual graph definition taken from Eq. 6, https://arxiv.org/pdf/0912.4389.pdf
self.vertex_dual = nx.line_graph(self.G)
off_diag = 1 - np.eye(len(self))
inc = nx.incidence_matrix(self.G,
oriented=True) # TODO: extract weights
inv_deg = np.diag([(0. if self.G.degree[x] <= 1 else
(1 / self.G.degree[x])) for x in self.G.nodes()])
self.vertex_dual_adj = inc.T @ inv_deg @ inc @ off_diag
def to_line(graph):
'''
:param graph
:return G_line: line/Subgraph network
'''
graph_to_line = nx.line_graph(graph)
graph_line = nx.convert_node_labels_to_integers(graph_to_line,
first_label=0,
ordering='default')
return graph_line
def build_line_graph(people):
"""
Edge coloring and Vizing's theorem solution
can be found from Stack Overflow question below
ref: https://stackoverflow.com/questions/51758406/creating-time-schedule-from-list-of-people-and-who-they-have-to-meet
"""
G = nx.Graph()
G.add_edges_from(((p, q) for p, L in people for q in L))
return nx.line_graph(G)
def cycle_distances(self, e):
""" Return vector of shortest path cycle distances between
e and all other edges
"""
L = nx.line_graph(self.cycle_dual)
# find all nodes of the line graph that e belongs to
line_nodes = [list(combinations(self.edge_in_cycles[f], 2)) for f in self.G.edges_iter()]
e_ind = self.G.edges().index(e)
dists_from_all_sources = [nx.shortest_path_length(L, source=src) for src in line_nodes[e_ind]]
cycle_dists = np.array(
[
[min([d[h] for h in line_nodes[i]]) for i in xrange(self.G.number_of_edges())]
for d in dists_from_all_sources
]
).min(axis=0)
return cycle_dists
def weighted_line_graph(G, average=False):
""" Return a line graph of G where edge attributes are propagated
properly. Node attributes are ignored.
If average is set to True, perform an averaging over
conductivities.
"""
line_graph = nx.line_graph(G)
line_graph.add_nodes_from((tuple(sorted((u, v))), d)
for u, v, d in G.edges_iter(data=True))
# average
if average:
new_node_conds = {}
for n, d in line_graph.nodes_iter(data=True):
neighbor_conds = mean([line_graph.node[m]['conductivity']
for m in line_graph.neighbors(n)])
new_node_conds[n] = 0.5*(d['conductivity'] +
neighbor_conds)
for n, v in new_node_conds.iteritems():
line_graph.node[n]['conductivity'] = v
return line_graph
def demo_line():
"""demo_line"""
g = nx.star_graph(10)
l = nx.line_graph(g)
plot(g)
def test_line_inverse_line_dgm(self):
G = nx.dorogovtsev_goltsev_mendes_graph(4)
H = nx.line_graph(G)
J = nx.inverse_line_graph(H)
assert_true(nx.is_isomorphic(G, J))
def test_create2(self):
G = nx.Graph()
G.add_edges_from([(0, 1), (1, 2), (2, 3)])
L = nx.line_graph(G, create_using=nx.DiGraph())
assert_edges_equal(L.edges(), [((0, 1), (1, 2)), ((1, 2), (2, 3))])
def test_digraph2(self):
G = nx.DiGraph()
G.add_edges_from([(0, 1), (1, 2), (2, 3)])
L = nx.line_graph(G)
assert_edges_equal(L.edges(), [((0, 1), (1, 2)), ((1, 2), (2, 3))])
def test_digraph1(self):
G = nx.DiGraph()
G.add_edges_from([(0, 1), (0, 2), (0, 3)])
L = nx.line_graph(G)
# no edge graph, but with nodes
assert_equal(L.adj, {(0, 1): {}, (0, 2): {}, (0, 3): {}})
def test_cycle(self):
G = nx.cycle_graph(5)
L = nx.line_graph(G)
assert_true(nx.is_isomorphic(L, G))
G19.add_node(1,label="B") G19.add_node(2,label="A") G19.add_node(3,label="B") G19.add_node(4,label="A") G19.add_edge(1,2) G19.add_edge(2,3) G19.add_edge(3,4) G19.add_edge(4,1) generateIndependentEmbeddings(G10,G11) generateIndependentEmbeddings(G14,G15) generateIndependentEmbeddings(G18,G19) print checkSubGraphIsomorphismWithLabels(G16,G17) generateIndependentEmbeddings(G16,G17) print nx.line_graph(G16).nodes() print "20-21" G20=nx.Graph() G20.add_node(1,label="A") G20.add_node(2,label="A") G20.add_edge(1,2) G21=nx.Graph() G21.add_node(1,label="B") G21.add_node(2,label="B") G21.add_node(3,label="B") G21.add_edge(1,2) G21.add_edge(2,3) checkSubGraphIsomorphismWithLabels(G20,G21) print generateIndependentEmbeddings(G16,G17)
def test_star(self):
G = nx.star_graph(5)
L = nx.line_graph(G)
assert_true(nx.is_isomorphic(L, nx.complete_graph(5)))
def test_line_inverse_line_complete(self):
G = nx.complete_graph(10)
H = nx.line_graph(G)
J = nx.inverse_line_graph(H)
assert_true(nx.is_isomorphic(G, J))
def generate_line_graph(edge_labels,graph):
g2=nx.DiGraph()
ln_graph=nx.line_graph(graph)
for edge in ln_graph.edges():
g2.add_edge(edge_labels[edge[0]],edge_labels[edge[1]])
return g2
def test_line_inverse_line_path(self):
G = nx.path_graph(10)
H = nx.line_graph(G)
J = nx.inverse_line_graph(H)
assert_true(nx.is_isomorphic(G, J))
def test_line_inverse_line_hypercube(self):
G = nx.hypercube_graph(5)
H = nx.line_graph(G)
J = nx.inverse_line_graph(H)
assert_true(nx.is_isomorphic(G, J))
def test_path(self):
G = nx.path_graph(5)
L = nx.line_graph(G)
assert_true(nx.is_isomorphic(L, nx.path_graph(4)))
def test_line_inverse_line_star(self):
G = nx.star_graph(20)
H = nx.line_graph(G)
J = nx.inverse_line_graph(H)
assert_true(nx.is_isomorphic(G, J))
def test_line_inverse_line_multipartite(self):
G = nx.complete_multipartite_graph(3, 4, 5)
H = nx.line_graph(G)
J = nx.inverse_line_graph(H)
assert_true(nx.is_isomorphic(G, J))
