def dijkstra_bidirectional(graph, s, t, cancel=False):
graphr = graph.reverse(copy=True)
distances = defaultdict(constant_factory(float("+inf")))
distancesr = defaultdict(constant_factory(float("+inf")))
queue = PriorityQueue()
queue.insert(s, 0)
queuer = PriorityQueue()
queuer.insert(t, 0)
distances[s] = 0
distancesr[t] = 0
mue = float("+inf")
settled = {}
num_settled = 0
num_relaxed = 0
settled_forward = []
settled_backward = []
while not queue.is_empty() and not queuer.is_empty():
u, k = queue.extract_min()
num_settled += 1
for node, edge_details in graph[u].items():
if distances[u] + edge_details['weight'] < distances[node]:
num_relaxed += 1
distances[node] = distances[u] + edge_details['weight']
mue = min(mue, distances[node] + distancesr[node])
if queue.contains(node):
queue.decrease_key(node, distances[node])
else:
queue.insert(node, distances[node])
settled_forward.append(u)
if u in settled:
break
settled[u] = True
ur, kr = queuer.extract_min()
num_settled += 1
for node, edge_details in graphr[ur].items():
if distancesr[ur] + edge_details['weight'] < distancesr[node]:
num_relaxed += 1
distancesr[node] = distancesr[ur] + edge_details['weight']
mue = min(mue, distances[node] + distancesr[node])
if queuer.contains(node):
queuer.decrease_key(node, distancesr[node])
else:
queuer.insert(node, distancesr[node])
settled_backward.append(ur)
if ur in settled:
break
settled[ur] = True
return mue, num_settled, num_relaxed, [settled_forward, settled_backward]
def dijkstra_cancel(graph, s, t):
distances = defaultdict(constant_factory(float("+inf")))
parents = {}
queue = PriorityQueue()
queue.insert(s, 0)
distances[s] = 0
settled = 0
relaxed = 0
settled_nodes = []
while not queue.is_empty():
u, k = queue.extract_min()
settled += 1
settled_nodes.append(u)
if u == t:
break
for node, edge_details in graph[u].items():
if distances[u] + edge_details['weight'] < distances[node]:
relaxed += 1
distances[node] = distances[u] + edge_details['weight']
parents[node] = u
if queue.contains(node):
queue.decrease_key(node, distances[node])
else:
queue.insert(node, distances[node])
return distances[t], settled, relaxed, [settled_nodes]
def a_star(graph, s, t):
xt, yt = graph.node[t]['x'], graph.node[t]['y']
def pi(v):
x, y = graph.node[v]['x'], graph.node[v]['y']
return ((x - xt) **2 + (y - yt) ** 2) ** 0.5
distances = defaultdict(constant_factory(float("+inf")))
parents = {}
queue = PriorityQueue()
queue.insert(s, pi(s))
distances[s] = 0
settled = 0
relaxed = 0
settled_nodes = []
while not queue.is_empty():
u, k = queue.extract_min()
settled += 1
settled_nodes.append(u)
if u == t:
break
for node, edge_details in graph[u].items():
if distances[u] + edge_details['weight'] < distances[node]:
relaxed += 1
distances[node] = distances[u] + edge_details['weight']
parents[node] = u
if queue.contains(node):
queue.decrease_key(node, distances[node] + pi(node))
else:
queue.insert(node, distances[node] + pi(node))
return distances[t], settled, relaxed, [settled_nodes]
def cheater(graph, s, t):
g_reverse = graph.reverse(copy=True)
_, pi = dijkstra(g_reverse, t)
distances = defaultdict(constant_factory(float("+inf")))
parents = {}
queue = PriorityQueue()
queue.insert(s, pi[s])
distances[s] = 0
settled = 0
relaxed = 0
while not queue.is_empty():
u, r = queue.extract_min()
settled += 1
if u == t:
break
for node, edge_details in graph[u].items():
if distances[u] + edge_details['weight'] < distances[node]:
relaxed += 1
distances[node] = distances[u] + edge_details['weight']
parents[node] = u
if queue.contains(node):
queue.decrease_key(node, distances[node] + pi[node])
else:
queue.insert(node, distances[node] + pi[node])
return distances[t], settled, relaxed, [ parents.keys(), [] ]
def dijkstra(graph, s):
distances = defaultdict(constant_factory(float("+inf")))
parents = {}
queue = PriorityQueue()
queue.insert(s, 0)
distances[s] = 0
visited = 0
while not queue.is_empty():
u, k = queue.extract_min()
visited += 1
for node, edge_details in graph[u].items():
if distances[u] + edge_details['weight'] < distances[node]:
distances[node] = distances[u] + edge_details['weight']
parents[node] = u
if queue.contains(node):
queue.decrease_key(node, distances[node])
else:
queue.insert(node, distances[node])
result = nx.DiGraph()
for u, v, data in graph.edges(data=True):
if v in parents and parents[v] == u:
result.add_edge(u, v, **data)
return result, distances
def a_star_bidirectional(graph, s, t):
"""
Bidirectional A* that stops iff one search has found the target
"""
graphr = graph.reverse(copy=True)
xt, yt = graph.node[t]['x'], graph.node[t]['y']
def pi(v):
x, y = graph.node[v]['x'], graph.node[v]['y']
return ((x - xt) **2 + (y - yt) ** 2) ** 0.5
xs, ys = graph.node[s]['x'], graph.node[s]['y']
def pir(v):
x, y = graph.node[v]['x'], graph.node[v]['y']
return ((x - xs) **2 + (y - ys) ** 2) ** 0.5
pf = lambda v : (pi(v) - pir(v))/2.0
pr = lambda v : (pir(v) - pi(v))/2.0
pfs = pf(s)
pft = pf(t)
distances = defaultdict(constant_factory(float("+inf")))
distancesr = defaultdict(constant_factory(float("+inf")))
queue = PriorityQueue()
queue.insert(s, pi(s))
queuer = PriorityQueue()
queuer.insert(t, pir(t))
distances[s] = 0
distancesr[t] = 0
mue = float("+inf")
num_settled = 0
num_relaxed = 0
settled_nodes = []
settled_nodes_r = []
while not queue.is_empty() and not queuer.is_empty():
u, k = queue.extract_min()
num_settled += 1
settled_nodes.append(u)
if u == t:
mue = distances[t]
break
for node, edge_details in graph[u].items():
if distances[u] + edge_details['weight'] < distances[node]:
num_relaxed += 1
distances[node] = distances[u] + edge_details['weight']
mue = min(mue, distances[node] + distancesr[node])
if queue.contains(node):
queue.decrease_key(node, distances[node] + pf(node))
else:
queue.insert(node, distances[node] + pf(node))
ur, kr = queuer.extract_min()
num_settled += 1
settled_nodes_r.append(ur)
if ur == s:
mue = distancesr[s]
break
for node, edge_details in graphr[ur].items():
if distancesr[ur] + edge_details['weight'] < distancesr[node]:
num_relaxed += 1
distancesr[node] = distancesr[ur] + edge_details['weight']
mue = min(mue, distances[node] + distancesr[node])
if queuer.contains(node):
queuer.decrease_key(node, distancesr[node] + pr(node))
else:
queuer.insert(node, distancesr[node] + pr(node))
if k + kr > mue + pfs - pft:
break
return mue, num_settled, num_relaxed, [settled_nodes, settled_nodes_r]
def a_star_bidirectional_onesided(graph, s, t):
"""
Bidirectional A* that stops iff one search has found the target
"""
graphr = graph.reverse(copy=True)
xt, yt = graph.node[t]['x'], graph.node[t]['y']
def pi(v):
x, y = graph.node[v]['x'], graph.node[v]['y']
return ((x - xt) **2 + (y - yt) ** 2) ** 0.5
xs, ys = graph.node[s]['x'], graph.node[s]['y']
def pir(v):
x, y = graph.node[v]['x'], graph.node[v]['y']
return ((x - xs) **2 + (y - ys) ** 2) ** 0.5
distances = defaultdict(constant_factory(float("+inf")))
distancesr = defaultdict(constant_factory(float("+inf")))
queue = PriorityQueue()
queue.insert(s, pi(s))
queuer = PriorityQueue()
queuer.insert(t, pir(t))
distances[s] = 0
distancesr[t] = 0
mue = float("+inf")
settled = {}
num_settled = 0
num_relaxed = 0
while not queue.is_empty() and not queuer.is_empty():
u, k = queue.extract_min()
num_settled += 1
if mue < k:
break
for node, edge_details in graph[u].items():
if distances[u] + edge_details['weight'] < distances[node]:
num_relaxed += 1
distances[node] = distances[u] + edge_details['weight']
mue = min(mue, distances[node] + distancesr[node])
if queue.contains(node):
queue.decrease_key(node, distances[node] + pi(node))
else:
queue.insert(node, distances[node] + pi(node))
ur, kr = queuer.extract_min()
num_settled += 1
if mue < kr:
break
for node, edge_details in graphr[ur].items():
if distancesr[ur] + edge_details['weight'] < distancesr[node]:
num_relaxed += 1
distancesr[node] = distancesr[ur] + edge_details['weight']
mue = min(mue, distances[node] + distancesr[node])
if queuer.contains(node):
queuer.decrease_key(node, distancesr[node] + pir(node))
else:
queuer.insert(node, distancesr[node] + pir(node))
return mue, num_settled, num_relaxed
