def run(nodes,edges):
"""
execute the Floyd Warshall algorithm
"""
dist = init_dist(nodes)
# set matrix identity elements to 0
for v in nodes:
dist[v][v] = 0
# fill matrix with intial edge weights
for e in edges:
# shorter in our case means with stronger connections, hence:
w = gi.get_weight(e)
if w > 0:
d = 1.0 / w
else:
d = 99999 # "Inf"
# inizialization of v->u and u->v to same value because the graph is undirected
dist[e.source][e.target] = d
dist[e.target][e.source] = d
# the core of the algorithm
for v1 in nodes:
for v2 in nodes:
for v3 in nodes:
if dist[v2][v1] + dist[v1][v3] < dist[v2][v3]:
dist[v2][v3] = dist[v2][v1] + dist[v1][v3]
return dist
def compute_edges():
""" actions performed by the environment
Alzeihmer desease:
select randomly a certain percentage of edges.
The weights of the edges will be reduced
of a certain percentage
"""
num_edges_selected = int(math.ceil(gi.num_edges() * config.frac_edges ))
selected=gi.random_edges(num_edges_selected)
damaged = set()
thisdamage=[]
for e in selected: # each of these edges will be assigned a damage
prev_weight = gi.get_weight(e)
gi.mult_weight(e, config.frac_damage)
diff = prev_weight - e.weight
e.damage = diff # if the edge is too light, destroy it
thisdamage.append((e.source, e.target, e.damage))
if gi.get_weight(e) < 1.0: #del e # FIXME!! careful!!!
pass
damaged.add(e)
return damaged
def total_weight():
ret = 0.0
for e in gi.edges():
ret += gi.get_weight(e)
return ret
