def testfindLCA(self):
root = Node(1)
root.children.append(dag.Node(2))
root.children.append(dag.Node(3))
root.children[0].children.append(dag.Node(4))
root.children[0].children.append(dag.Node(5))
root.children[1].children.append(dag.Node(6))
root.children[1].children.append(dag.Node(7))
# test all nodes are None
self.assertEqual(dag.findLCA(None, None, None), None)
#test when root is euqal to node
self.assertEqual(dag.findLCA(root, 1, 1), 1)
#test if n1 is not present
self.assertEqual(dag.findLCA(root, None, 2), None)
#test if n2 is not present
self.assertEqual(dag.findLCA(root, 2, None), None)
#test when node is value is not in tree
self.assertEqual(dag.findLCA(root, 2, 13), None)
#test dag for different values
self.assertEqual(dag.findLCA(root, 3, 7), 3)
self.assertEqual(dag.findLCA(root, 4, 6), 1)
self.assertEqual(dag.findLCA(root, 2, 5), 2)
self.assertEqual(dag.findLCA(root, 3, 4), 1)
def test_root_node(
self): # Test case where root is a node of interest for lca
root = Node(1)
node1 = Node(2)
node2 = Node(3)
root.succ = [node1, node2]
lca = daglca(root, node1, node1)
assert lca is 2
lca = daglca(root, node2, root)
assert lca is 1
lca = daglca(root, node1, root)
assert lca is 1, "test_root_node failed"
def testNodes(self): #Testing root value in the tree
root = Node(1)
self.assertEqual(1, root.key)
net=_dense_net(),
trainset=splited_trainset[i],
testset=splited_testset[i],
log=True)
else: # Honest nodes
client = Client(args=args,
net=_dense_net(),
trainset=splited_trainset[i],
testset=splited_testset[i],
log=True)
client.set_weights(tmp_client.get_weights()) # same init. weights
clients.append(client)
"""Set DAG
# TODO: DAG connection
"""
genesis = Node(weights=tmp_client.get_weights(), _id=-1)
nodes = []
nodes.append(genesis)
"""Run simulator
# TODO: logging time (train, test)
"""
latest_nodes = deepcopy(nodes) # in DAG
for epoch in range(1, args.nEpochs + 1):
print(">>> Round %5d" % (epoch))
# select activated clients
# At least one honest node
n_activated_byz = random.randint(0, args.nByzs) # in Byz.
n_activated_norm = random.randint(1, args.norm) # in Norm.
"""set nodes"""
clients = []
for i in range(ns):
clients.append(
Client(trainset=splited_trainset[i],
testset=splited_testset[i],
net=Net(),
_id=i))
tmp_client = Client( # tmp
trainset=None, testset=splited_testset[i], net=Net(), _id=-1)
"""set DAG"""
global_id = -1
nodes = []
genesis = Node(r=-1, w=clients[0].get_weights(), _id=global_id)
nodes.append(genesis)
global_id += 1
"""run simulator"""
latest_nodes = deepcopy(nodes) # in DAG
for r in range(rs):
print(">>> Round %5d" % (r))
# select activated clients
n_activated = random.randint(1, ns) # TODO: range
activateds = random.sample([t for t in range(ns)], n_activated)
current_nodes = []
current_accs = []
def test_lowest_node(
self): # Test case with one of the lowest nodes in graph
root = Node(1)
node1 = Node(2)
node2 = Node(3)
node3 = Node(4)
node4 = Node(5)
node5 = Node(6)
root.succ = [node1, node2, node3, node4]
node1.succ = [node3]
node1.pred = [root]
node2.succ = [node3, node4]
node2.pred = [root]
node3.succ = [node4]
node3.pred = [node1, node2, root]
node4.pred = [node2, node3, root]
node5.pred = [node3]
lca = daglca(root, node5, node1)
assert lca is 1, "test_lowest_node failed"
def test_basic(self): #Test for basic case, nodes present in graph
root = Node(1)
node1 = Node(2)
node2 = Node(3)
node3 = Node(4)
node4 = Node(5)
node5 = Node(6)
root.succ = [node1, node2, node3, node4]
node1.succ = [node3]
node1.pred = [root]
node2.succ = [node3, node4]
node2.pred = [root]
node3.succ = [node4]
node3.pred = [node1, node2, root]
node4.pred = [node2, node3, root]
node5.pred = [node3]
lca = daglca(root, node3, node4)
assert lca is 3, "test_basic failed"