def GreedyCycle(distanceMatrix, startingVertex):
path = [startingVertex]
for i in range(math.ceil(len(distanceMatrix) / 2)):
minCost = np.uint64(-1)
minCostIndex = -1
vertexToAdd = -1
for j, item in enumerate(range(len(distanceMatrix))):
if item not in path:
for n in range(len(path)):
distance1 = distanceMatrix[item][path[n - 1]]
distance2 = distanceMatrix[item][path[n]]
cost = distance1 + distance2 - distanceMatrix[path[n - 1]][
path[n]]
if cost < minCost:
minCostIndex = n
vertexToAdd = item
minCost = cost
path.insert(minCostIndex, vertexToAdd)
path = path + [path[0]]
return path
if __name__ == '__main__':
kroA100Data = utils.readTspFile("../data/kroA100.tsp")
kroB100Data = utils.readTspFile("../data/kroB100.tsp")
utils.tester(GreedyCycle, kroA100Data)
loss_list = []
n_neuron_list = [10,20,30]
print("1")
n_neuron = 30
n_neuron_2 = 10
model = MLP_Classifier(input_size,n_neuron,n_neuron_2,output_size)
criterion = nn.BCELoss()
optimizer = optim.SGD(model.parameters(),lr = lr)
for i in range(epochs):
if i % 100 == 0:
print(i)
for batch_idx, samples in enumerate(dataloader):
x_train_data, y_train_data = samples
optimizer.zero_grad()
pred = model(x_train_data)
loss = criterion(pred,y_train_data)
loss.backward()
optimizer.step()
loss_list.append(loss.detach().numpy())
fig, ax = plt.subplots(figsize = (20,20))
ax.plot(loss_list)
print("3")
trained_dict = model.state_dict()
model = MLP_Classifier(input_size,n_neuron,n_neuron_2,output_size).to(device)
tester(x_test_data, y_test_data, model, trained_dict)
print("4")
def Regret(distanceMatrix, startingVertex):
path = [startingVertex]
for i in range(math.ceil(len(distanceMatrix) / 2)):
vertexesCosts = []
for item in range(len(distanceMatrix)):
costList = []
if item not in path:
for n in range(len(path)):
distance1 = distanceMatrix[item][path[n - 1]]
distance2 = distanceMatrix[item][path[n]]
cost = distance1 + distance2 - distanceMatrix[path[n - 1]][
path[n]]
costList.append((cost, n))
costList = sorted(costList, key=lambda e: e[0])[0:3]
vertexesCosts.append((item, costList[0][1],
sum([
costList[i][0] - costList[0][0]
for i in range(len(costList))
])))
vertex, index, _ = max(vertexesCosts, key=lambda e: e[2])
path.insert(index, vertex)
path = path + [path[0]]
return path
if __name__ == '__main__':
kroA100Data = utils.readTspFile("../data/kroA100.tsp")
kroB100Data = utils.readTspFile("../data/kroB100.tsp")
utils.tester(Regret, kroA100Data)