def run(self, inputTour):
tour = inputTour.copy()
stop = False
bestCost = math.inf
sameCost = 0
while not stop:
index1 = randint(0, self.numberOfCities - 1)
index2 = randint(0, self.numberOfCities - 1)
city1 = tour[index1]
city2 = tour[index2]
tempTour = tour.copy()
tempTour[index1] = city2
tempTour[index2] = city1
tempCost = getCost(tempTour, self.edges)
if tempCost <= bestCost:
tour = tempTour
bestCost = tempCost
if (tempCost == bestCost):
sameCost += 1
else:
sameCost = 0
stop = sameCost >= self.stopCriterion
return tour, getCost(tour, self.edges)
def runInner(timestamps, max_iter=10):
"""
Implements Inner algorithm
Parameters
----------
timestamps : list of tuples
Sorted list of interactions [(t1, n1, n2), (t2, n3, n4),...], where t is timestamp, n1 and n2 are interactiong nodes.
Nodes in the interactions are sorted lexicographically.
maxiter : int
maximum number of interactions
Returns
-------
tuple of dicts
(dict1, dict2): dict1 of a shape {n1: t1, n2: t2, ..} contains starting point t1 of activity interval of node n1, for every node,
dict2 of a shape {n1: t1, n2: t2, ..} contains ending point t1 of activity interval of node n1, for every node.
"""
nodeEdgeIndex = utils.getIndex(timestamps, 'inner')
m = getInitial(nodeEdgeIndex)
alpha, alphaNodes = getAlphas(timestamps, m)
Xstart, Xend = getSolution(alphaNodes, m)
best = utils.getCost(Xstart, Xend)
bestSol = (Xstart, Xend)
for k in xrange(1, max_iter):
m = getNextInput(nodeEdgeIndex, Xstart, Xend)
alpha, alphaNodes = getAlphas(timestamps, m)
Xstart, Xend = getSolution(alphaNodes, m)
t = utils.getCost(Xstart, Xend)
if t < best:
best = t
bestSol = (Xstart, Xend)
else:
break
return bestSol[0], bestSol[1]
def run(self):
visitedCities = [False] * self.numberOfCities
tour = []
startingCity = randint(0, self.numberOfCities - 1)
visitedCities[startingCity] = True
tour.append(startingCity)
currentIndex = 0
while numberOfCitiesVisited(visitedCities) != self.numberOfCities:
city = self.getNearestCity(tour[currentIndex], visitedCities)
currentIndex += 1
visitedCities[city] = True
tour.append(city)
return tour, getCost(tour, self.edges)
def run(self):
visitedCities = [False] * self.numberOfCities
tour = []
startingCity = randint(0, self.numberOfCities - 1)
tour.append(startingCity)
visitedCities[startingCity] = True
while numberOfCitiesVisited(visitedCities) != self.numberOfCities:
randomIndex = randint(0, self.numberOfCities - 1)
if not visitedCities[randomIndex]:
visitedCities[randomIndex] = True
tour.append(randomIndex)
return tour, getCost(tour, self.edges)
def run(self, inputTour, inputCost, maxTries):
stop = False
bestTour = inputTour.copy()
bestCost = inputCost
oldTour = inputTour.copy()
oldCost = inputCost
sameCost = 0
propabiltyOfAcceptance = 0.9
while True:
for counter in range(0, maxTries):
index1 = randint(0, self.numberOfCities - 1)
index2 = randint(0, self.numberOfCities - 1)
city1 = oldTour[index1]
city2 = oldTour[index2]
tempTour = oldTour.copy()
tempTour[index1] = city2
tempTour[index2] = city1
tempCost = getCost(tempTour, self.edges)
if tempCost < oldCost:
oldTour = tempTour
oldCost = tempCost
if tempCost < bestCost:
bestCost = tempCost
bestTour = tempTour
else:
rnd = (randint(0, 101) / 100)
if (rnd < propabiltyOfAcceptance):
oldCost = tempCost
oldTour = tempTour
propabiltyOfAcceptance = 0.9 * propabiltyOfAcceptance
if propabiltyOfAcceptance < 0.0000001:
break
return bestTour, bestCost
Cost_IP, P_IP, R_IP, F_IP, Costmax_IP = [], [], [], [], []
iterations = range(1, 11)
for iter in iterations:
if iter == 1:
m = inner.getInitial(nodeEdgeIndex)
Xstart, Xend = inner.runInner_iteration(timestamps, nodeEdgeIndex,
m)
else:
m = inner.getNextInput(nodeEdgeIndex, Xstart, Xend)
Xstart, Xend = inner.runInner_iteration(timestamps, nodeEdgeIndex,
m)
Cost_IP.append(
utils.getCost(Xstart, Xend) / ((event_length - 1) * num_nodes))
p, r, f = utils.compareGT(Xstart, Xend, active_truth, timestamps)
P_IP.append(p)
R_IP.append(r)
F_IP.append(f)
Costmax_IP.append(utils.getMax(Xstart, Xend) / (event_length - 1))
name = 'test_convergence'
plt.figure()
plt.plot(iterations, Cost_IP, 'k')
plt.xlabel('iterations', fontsize=25)
plt.ylabel('Relative total length', fontsize=25)
plt.tight_layout()
plt.savefig(name + '_' + 'total.pdf')
plt.figure()
G = utils.generateGraph(n=num_nodes)
print 'number of nodes in the background network:', G.number_of_nodes()
print 'number of edges in the background network:', G.number_of_edges()
timestamps, active_truth = utils.generateIntervals(
G, event_length=event_length, overlap=overlap)
print 'number of timestamps', len(timestamps)
if alg == 'baseline':
Xstart, Xend = baseline.baseline(timestamps)
elif alg == 'inner':
Xstart, Xend = inner_point.runInner(timestamps)
elif alg == 'budget':
Xstart, Xend = budget.runBudget(timestamps)
else:
print('no algorithm specified')
exit()
print
print 'relative total length of solution =', utils.getCost(
Xstart, Xend) / ((event_length - 1) * num_nodes)
print 'relative maximum length of solution =', utils.getMax(
Xstart, Xend) / (event_length - 1)
p, r, f = utils.compareGT(Xstart, Xend, active_truth, timestamps)
print 'precision =', p
print 'recall =', r
print 'f-measure =', f
def isBestTour(self, tour):
newCost = getCost(tour, self.edges)
return newCost < bestCost, newCost