def frank_wolfe_on_I210():
'''
Frank-Wolfe on I210
'''
graph = np.loadtxt('data/I210_attack_net.csv', delimiter=',', skiprows=1)
demand = np.loadtxt('data/I210_od.csv', delimiter=',', skiprows=1)
demand[:,2] = 1. * demand[:,2] / 4000
# run solver
f = solver_3(graph, demand, max_iter=1000, q=50, display=1, stop=1e-2)
# display cost
for a,b in zip(cost(f, graph), f*4000): print a,b
# visualization
node = np.loadtxt('data/I210_node.csv', delimiter=',', skiprows=1)
# extract features: 'capacity', 'length', 'fftt'
feat = extract_features('data/I210_attack_Sketch_net.csv')
ratio = cost_ratio(f, graph)
# merge features with the cost ratios
features = np.zeros((feat.shape[0],4))
features[:,:3] = feat
features[:,3] = ratio
# join features with (lat1,lon1,lat2,lon2)
links = process_links(graph, node, features)
color = features[:,3] # we choose the costs
names = ['capacity', 'length', 'fftt', 'tt_over_fftt']
geojson_link(links, names, color)
def load_network_data(name):
#import pdb; pdb.set_trace()
#The folder locations of all the input files
graphLocation = 'data/' + name + '_net.csv'
demandLocation = 'data/' + name + '_od.csv'
nodeLocation = 'data/' + name + '_node.csv'
featureLocation = 'data/' + name + '_net.txt'
graph = np.loadtxt(graphLocation, delimiter=',', skiprows=1)
demand = np.loadtxt(demandLocation, delimiter=',', skiprows=1)
features = extract_features(featureLocation)
import pdb
pdb.set_trace()
#LA network has a different way of processing the nodes file
if (name == "LA"):
node = np.loadtxt(nodeLocation, delimiter=',')
features[10787, 0] = features[10787, 0] * 1.5
graph[10787, -1] = graph[10787, -1] / (1.5**4)
features[3348, :] = features[3348, :] * 1.2
graph[3348, -1] = graph[3348, -1] / (1.2**4)
else:
node = np.loadtxt(nodeLocation, delimiter=',', skiprows=1)
return graph, demand, node, features
def reduce_demand():
net, demand, node = load_LA()
features = extract_features('data/LA_net.txt')
f = np.loadtxt('data/LA/LA_output_3.csv', delimiter=',', skiprows=0)
cr = cost_ratio(f, net)
for row in range(net.shape[0]):
if cr[row] >= 10.:
out = []
for i in range(demand.shape[0]):
if int(demand[i,0]) == int(net[row,1]):
out.append(demand[i,2])
demand[i,2] = demand[i,2] / 10.
if len(out) > 0:
out = np.array(out)
print '\nratio:', cr[row]
print 'origin: {}\nflow: {}'.format(int(demand[i,0]), np.sum(out))
print np.sort(out).tolist()
for row in range(net.shape[0]):
if cr[row] >= 10.:
out = []
for i in range(demand.shape[0]):
if int(demand[i,1]) == int(net[row,2]):
out.append(demand[i,2])
demand[i,2] = demand[i,2] / 10.
if len(out) > 0:
out = np.array(out)
print '\nratio:', cr[row]
print 'destination: {}\nflow: {}'.format(int(demand[i,0]), np.sum(out))
print np.sort(out).tolist()
def reduce_demand():
net, demand, node = load_LA()
features = extract_features('data/LA_net.txt')
f = np.loadtxt('data/LA/LA_output_3.csv', delimiter=',', skiprows=0)
cr = cost_ratio(f, net)
for row in range(net.shape[0]):
if cr[row] >= 10.:
out = []
for i in range(demand.shape[0]):
if int(demand[i, 0]) == int(net[row, 1]):
out.append(demand[i, 2])
demand[i, 2] = demand[i, 2] / 10.
if len(out) > 0:
out = np.array(out)
print '\nratio:', cr[row]
print 'origin: {}\nflow: {}'.format(int(demand[i, 0]),
np.sum(out))
print np.sort(out).tolist()
for row in range(net.shape[0]):
if cr[row] >= 10.:
out = []
for i in range(demand.shape[0]):
if int(demand[i, 1]) == int(net[row, 2]):
out.append(demand[i, 2])
demand[i, 2] = demand[i, 2] / 10.
if len(out) > 0:
out = np.array(out)
print '\nratio:', cr[row]
print 'destination: {}\nflow: {}'.format(
int(demand[i, 0]), np.sum(out))
print np.sort(out).tolist()
def load_I210():
net = np.loadtxt('data/I210_attack_net.csv', delimiter=',', skiprows=1)
demand = np.loadtxt('data/I210_od.csv', delimiter=',', skiprows=1)
node = np.loadtxt('data/I210_node.csv', delimiter=',', skiprows=1)
geometry = extract_features('data/I210_attack_Sketch_net.csv')
demand = np.reshape(demand, (1, 3))
return net, demand, node, geometry
def frank_wolfe_on_I210():
'''
Frank-Wolfe on I210
'''
graph = np.loadtxt('data/I210_attack_net.csv', delimiter=',', skiprows=1)
demand = np.loadtxt('data/I210_od.csv', delimiter=',', skiprows=1)
demand[:, 2] = 1. * demand[:, 2] / 4000
# run solver
f = solver_3(graph, demand, max_iter=1000, q=50, display=1, stop=1e-2)
# display cost
for a, b in zip(cost(f, graph), f * 4000):
print a, b
# visualization
node = np.loadtxt('data/I210_node.csv', delimiter=',', skiprows=1)
# extract features: 'capacity', 'length', 'fftt'
feat = extract_features('data/I210_attack_Sketch_net.csv')
ratio = cost_ratio(f, graph)
# merge features with the cost ratios
features = np.zeros((feat.shape[0], 4))
features[:, :3] = feat
features[:, 3] = ratio
# join features with (lat1,lon1,lat2,lon2)
links = process_links(graph, node, features)
color = features[:, 3] # we choose the costs
names = ['capacity', 'length', 'fftt', 'tt_over_fftt']
geojson_link(links, names, color)
def load_LA_3():
# graph is an array of the network
#It is a 28376 by 8 array, where the strucuture is [LINK id, node 1 (of link), node 2 (of link), a0,a1,a2,a3,a4
# where the ai are factors for the link cost functions
graph = np.loadtxt('data/LA_net.csv', delimiter=',', skiprows=1)
#demand is an array with num-OD by 3, where num-OD is the number of OD pairs
#Entry in array is formatted as [origin, destination, demand] (still have to add the units of demands)
demand = np.loadtxt('data/LA_od_3.csv', delimiter=',', skiprows=1)
#node is an num-nod by 3 array, where num-node is the number of nodes
#each entry has format [node id, ...]
node = np.loadtxt('data/LA_node.csv', delimiter=',')
# features = table in the format [[capacity, length, FreeFlowTime]]
features = extract_features('data/LA_net.txt')
# increase capacities of these two links because they have a travel time
# in equilibrium that that is too big
features[10787, 0] = features[10787, 0] * 1.5
graph[10787, -1] = graph[10787, -1] / (1.5**4)
features[3348, :] = features[3348, :] * 1.2
graph[3348, -1] = graph[3348, -1] / (1.2**4)
# divide demand going to node 106 by 10 because too large
for i in range(demand.shape[0]):
if demand[i, 1] == 106.:
demand[i, 2] = demand[i, 2] / 10.
return graph, demand, node, features
def load_I210():
net = np.loadtxt('data/I210_attack_net.csv', delimiter=',', skiprows=1)
demand = np.loadtxt('data/I210_od.csv', delimiter=',', skiprows=1)
node = np.loadtxt('data/I210_node.csv', delimiter=',', skiprows=1)
geometry = extract_features('data/I210_attack_Sketch_net.csv')
demand = np.reshape(demand, (1,3))
return net, demand, node, geometry
def load_LA_2():
#graph = np.loadtxt('data/LA_net.csv', delimiter=',', skiprows=1)
#demand = np.loadtxt('data/LA_od_2.csv', delimiter=',', skiprows=1)
#node = np.loadtxt('data/LA_node.csv', delimiter=',')
#Changing everything to Chicago regional
graph = np.loadtxt('data/ChicagoRegional_net.csv',
delimiter=',',
skiprows=1)
demand = np.loadtxt('data/ChicagoRegional_od.csv',
delimiter=',',
skiprows=1)
node = np.loadtxt('data/ChicagoRegional_node.csv',
delimiter=',',
skiprows=1)
# features = table in the format [[capacity, length, FreeFlowTime]]
#features = extract_features('data/LA_net.txt')
features = extract_features('data/ChicagoRegional_net.txt')
# increase capacities of these two links because they have a travel time
# in equilibrium that that is too big
#features[10787,0] = features[10787,0] * 1.5
#graph[10787,-1] = graph[10787,-1] / (1.5**4)
#features[3348,:] = features[3348,:] * 1.2
#graph[3348,-1] = graph[3348,-1] / (1.2**4)
return graph, demand, node, features
def visualize_result_ratio_study(fileName, ratio, name, mode):
net, demand, node, features = load_network_data(name)
#Loading the flow per link resulting from frank-wolfe
f = np.loadtxt(fileName, delimiter=',', skiprows=0)
#Location of the features
featureLocation = 'data/' + name + '_net.txt'
features = np.zeros((f.shape[0], 4))
features[:, :3] = extract_features(featureLocation)
#Multiply the flow obtained by 4000 since we initially divided by 4000 before frank-wolfs
f = np.divide(f * 4000, features[:, 0])
features[:, 3] = f
links = process_links(net, node, features, in_order=True)
#creates color array used to visulized the links
#values useful in differenciating links based of flow on links
color = 2.0 * f + 1.0
#Keeping track of the percentage of congestion
links_congested = len(color[np.where(color >= 3)])
percentage_of_congestion = float(links_congested) / float(len(color))
print("congestion is at %3f " % percentage_of_congestion)
geojson_link_Scenario_Study(
ratio, links, ['capacity', 'length', 'fftt', 'flow_over_capacity'],
color, name, mode)
def load_I210_modified():
net = np.loadtxt('data/I210_net.csv', delimiter=',', skiprows=1)
demand = np.loadtxt('data/I210_od.csv', delimiter=',', skiprows=1)
node = np.loadtxt('data/I210_node.csv', delimiter=',', skiprows=1)
geometry = extract_features('data/I210Sketch_net.csv')
demand = np.reshape(demand, (1, 3))
# modify link 21
net[21, -1] = net[21, -1] * (4**4)
geometry[21, 0] = geometry[21, 0] / 4.
return net, demand, node, geometry
def visualize_LA_capacity():
graph, demand, node = load_LA()
features = extract_features('data/LA_net.txt')
links = process_links(graph, node, features, in_order=True)
color = features[:, 0] # we choose to color by the capacities
names = ['capacity', 'length', 'fftt']
# color = 2.1 * features[:,0] / 2000.
color = 2. * (features[:, 0] <= 900.) + 5. * (features[:, 0] > 900.)
weight = (features[:, 0] <= 900.) + 3. * (features[:, 0] > 900.)
geojson_link(links, names, color, weight)
def load_I210_modified():
net = np.loadtxt('data/I210_net.csv', delimiter=',', skiprows=1)
demand = np.loadtxt('data/I210_od.csv', delimiter=',', skiprows=1)
node = np.loadtxt('data/I210_node.csv', delimiter=',', skiprows=1)
geometry = extract_features('data/I210Sketch_net.csv')
demand = np.reshape(demand, (1,3))
# modify link 21
net[21,-1] = net[21,-1] * (4**4)
geometry[21,0] = geometry[21,0] / 4.
return net, demand, node, geometry
def visualize_LA_result():
net, demand, node = load_LA()
f = np.loadtxt('data/LA_output.csv', delimiter=',', skiprows=0)
features = np.zeros((f.shape[0], 4))
features[:,:3] = extract_features('data/LA_net.txt')
f = np.divide(f, features[:,0])
features[:,3] = f
links = process_links(net, node, features, in_order=True)
color = 2.0*f + 1.0
geojson_link(links, ['capacity', 'length', 'fftt', 'flow_over_capacity'], color)
def visualize_LA_capacity():
graph, demand, node = load_LA()
features = extract_features('data/LA_net.txt')
links = process_links(graph, node, features, in_order=True)
color = features[:,0] # we choose to color by the capacities
names = ['capacity', 'length', 'fftt']
# color = 2.1 * features[:,0] / 2000.
color = 2.*(features[:,0] <= 900.) + 5.*(features[:,0] > 900.)
weight = (features[:,0] <= 900.) + 3.*(features[:,0] > 900.)
geojson_link(links, names, color, weight)
def visualize_LA_result():
net, demand, node = load_LA()
f = np.loadtxt('data/LA_output.csv', delimiter=',', skiprows=0)
features = np.zeros((f.shape[0], 4))
features[:, :3] = extract_features('data/LA_net.txt')
f = np.divide(f, features[:, 0])
features[:, 3] = f
links = process_links(net, node, features, in_order=True)
color = 2.0 * f + 1.0
geojson_link(links, ['capacity', 'length', 'fftt', 'flow_over_capacity'],
color)
def load_LA_2():
graph = np.loadtxt('data/LA_net.csv', delimiter=',', skiprows=1)
demand = np.loadtxt('data/LA_od_2.csv', delimiter=',', skiprows=1)
node = np.loadtxt('data/LA_node.csv', delimiter=',')
# features = table in the format [[capacity, length, FreeFlowTime]]
features = extract_features('data/LA_net.txt')
# increase capacities of these two links because they have a travel time
# in equilibrium that that is too big
features[10787,0] = features[10787,0] * 1.5
graph[10787,-1] = graph[10787,-1] / (1.5**4)
features[3348,:] = features[3348,:] * 1.2
graph[3348,-1] = graph[3348,-1] / (1.2**4)
return graph, demand, node, features
def visualize_LA():
net, demand, node = load_LA()
f = np.loadtxt('data/la/LA_Cython.csv', delimiter=',', skiprows=0)
features = np.zeros((f.shape[0], 4))
features[:, :3] = extract_features('data/LA_net.txt')
#import pdb; pdb.set_trace()
f = np.divide(f * 4000, features[:, 0])
features[:, 3] = f
links = process_links(net, node, features, in_order=True)
#creates color array used to visulized the links
#values useful in differenciating links based of flow on links
color = 2.0 * f + 1.0
#congestion = f/features[:,0] #determines the congestion levels of links
geojson_link(links, ['capacity', 'length', 'fftt', 'flow_over_capacity'],
color)
def LAfix_heterogeneous(demand_r, capacity): #demand_nr=1-demand_r
'''generate heterogeneous game on LAfix network
demand_r is the ratio of routed users
capacity is the max capacity for which a road is considered low-capacity'''
g2 = np.loadtxt('data/LAfix_net.csv', delimiter=',', skiprows=1)
g1 = np.copy(g2)
features = extract_features('data/LAfix_net.txt')
for i in range(0, len(features)):
if features[i][0] <= capacity:
g1[i, 3:7] *= 3000
d1 = np.loadtxt('data/LA_od.csv', delimiter=',', skiprows=1)
d2 = np.copy(d1)
d1[:, 2] *= (1 - demand_r)
d2[:, 2] *= demand_r
return g1, g2, d1, d2
def I210_ratio_r_total():
'''
study the test_*.csv files generated by I210_parametric_study()
in particular, visualize the ratio each type of users on each link
'''
fs = np.loadtxt('data/test_50.csv', delimiter=',', skiprows=0)
ratio = np.divide(fs[:,1], np.maximum(np.sum(fs, axis=1), 1e-8))
net = np.loadtxt('data/I210_net.csv', delimiter=',', skiprows=1)
node = np.loadtxt('data/I210_node.csv', delimiter=',', skiprows=1)
geometry = extract_features('data/I210Sketch_net.csv')
features = np.zeros((fs.shape[0], 4))
features[:,:3] = geometry
features[:,3] = ratio
links = process_links(net, node, features)
color = 2 * ratio # we choose the ratios of nr over r+nr
geojson_link(links, ['capacity', 'length', 'fftt', 'r_routed'], color)
def I210_ratio_r_total():
'''
study the test_*.csv files generated by I210_parametric_study()
in particular, visualize the ratio each type of users on each link
'''
fs = np.loadtxt('data/test_50.csv', delimiter=',', skiprows=0)
ratio = np.divide(fs[:, 1], np.maximum(np.sum(fs, axis=1), 1e-8))
net = np.loadtxt('data/I210_net.csv', delimiter=',', skiprows=1)
node = np.loadtxt('data/I210_node.csv', delimiter=',', skiprows=1)
geometry = extract_features('data/I210Sketch_net.csv')
features = np.zeros((fs.shape[0], 4))
features[:, :3] = geometry
features[:, 3] = ratio
links = process_links(net, node, features)
color = 2 * ratio # we choose the ratios of nr over r+nr
geojson_link(links, ['capacity', 'length', 'fftt', 'r_routed'], color)
def main():
for alpha in [.75]:
# for alpha in np.linspace(0, .49, 50):
# for alpha in np.linspace(.5, .99, 50):
print "ALPHA:", alpha
start_time2 = timeit.default_timer()
graph = np.loadtxt('data/LA_net.csv', delimiter=',', skiprows=1)
demand = np.loadtxt('data/LA_od_2.csv', delimiter=',', skiprows=1)
graph[10787, -1] = graph[10787, -1] / (1.5**4)
graph[3348, -1] = graph[3348, -1] / (1.2**4)
node = np.loadtxt('data/LA_node.csv', delimiter=',')
features = extract_features('data/LA_net.txt')
# graph = np.loadtxt('data/Chicago_net.csv', delimiter=',', skiprows=1)
# demand = np.loadtxt('data/Chicago_od.csv', delimiter=',', skiprows=1)
# node = np.loadtxt('data/Chicago_node.csv', delimiter=',', skiprows=1)
# features = extract_features('data/ChicagoSketch_net.txt')
# features = table in the format [[capacity, length, FreeFlowTime]]
# alpha = .2 # also known as r
thres = 1000.
cog_cost = 3000.
demand[:, 2] = 0.5 * demand[:, 2] / 4000
g_nr, small_capacity = multiply_cognitive_cost(graph, features, thres,
cog_cost)
d_nr, d_r = heterogeneous_demand(demand, alpha)
fs, hs, n_d = fw_heterogeneous_1([graph, g_nr], [d_r, d_nr],
alpha,
max_iter=30,
display=1)
print n_d
output = {'f': fs, 'h': hs, 'n_d': n_d}
with open('graph_stuff/LA_net_od_2_alpha_{}.txt'.format(alpha),
'w') as outfile:
# with open('graph_stuff/Chicago_net_od_2_alpha_{}.txt'.format(alpha), 'w') as outfile:
outfile.write(pickle.dumps(output))
#end of timer
elapsed2 = timeit.default_timer() - start_time2
print("Execution took %s seconds" % elapsed2)
def load_LA_3():
graph = np.loadtxt('data/LA_net.csv', delimiter=',', skiprows=1)
demand = np.loadtxt('data/LA_od_3.csv', delimiter=',', skiprows=1)
node = np.loadtxt('data/LA_node.csv', delimiter=',')
# features = table in the format [[capacity, length, FreeFlowTime]]
features = extract_features('data/LA_net.txt')
# increase capacities of these two links because they have a travel time
# in equilibrium that that is too big
features[10787, 0] = features[10787, 0] * 1.5
graph[10787, -1] = graph[10787, -1] / (1.5**4)
features[3348, :] = features[3348, :] * 1.2
graph[3348, -1] = graph[3348, -1] / (1.2**4)
# divide demand going to node 106 by 10 because too large
for i in range(demand.shape[0]):
if demand[i, 1] == 106.:
demand[i, 2] = demand[i, 2] / 10.
return graph, demand, node, features
def load_LA_3():
graph = np.loadtxt('data/LA_net.csv', delimiter=',', skiprows=1)
demand = np.loadtxt('data/LA_od_3.csv', delimiter=',', skiprows=1)
node = np.loadtxt('data/LA_node.csv', delimiter=',')
# features = table in the format [[capacity, length, FreeFlowTime]]
features = extract_features('data/LA_net.txt')
# increase capacities of these two links because they have a travel time
# in equilibrium that that is too big
features[10787,0] = features[10787,0] * 1.5
graph[10787,-1] = graph[10787,-1] / (1.5**4)
features[3348,:] = features[3348,:] * 1.2
graph[3348,-1] = graph[3348,-1] / (1.2**4)
# divide demand going to node 106 by 10 because too large
for i in range(demand.shape[0]):
if demand[i,1] == 106.:
demand[i,2] = demand[i,2] / 10.
return graph, demand, node, features
def visualize_LA_result_Scenario_Study(fileName, ratio):
net, demand, node = load_LA()
f = np.loadtxt(fileName, delimiter=',', skiprows=0)
features = np.zeros((f.shape[0], 4))
features[:, :3] = extract_features('data/LA_net.txt')
#features[:,:3] = extract_features('data/ChicagoRegional_net.txt')
f = np.divide(f * 4000, features[:, 0])
features[:, 3] = f
links = process_links(net, node, features, in_order=True)
#creates color array used to visulized the links
#values useful in differenciating links based of flow on links
color = 2.0 * f + 1.0
#Keeping track of the percentage of congestion
links_congested = len(color[np.where(color >= 3)])
percentage_of_congestion = float(links_congested) / float(len(color))
print("congestion is at %3f " % percentage_of_congestion)
geojson_link_Scenario_Study(
ratio, links, ['capacity', 'length', 'fftt', 'flow_over_capacity'],
color)
spnd[i] = nodes[i]
np.savetxt('data/iod_node.csv',
spnd,
delimiter=',',
header='node,lat,lon',
comments='') #Node ID is equal to node index: the file is sparse
np.savetxt('data/iod_node_dense.csv',
ondes,
delimiter=',',
header='node,lat,lon',
comments='') #The file is densified and there are no blank lines
nodes = [el[0] for el in ondes]
#Select Links
links = np.loadtxt('data/LA_net.csv', delimiter=',', skiprows=1)
features = extract_features('data/LA_net.txt')
lniks = []
faetures = []
for i in range(0, len(links)):
if links[i][1] in nodes and links[i][
2] in nodes: #Check if both nodes of the link are in the zone
lniks.append(links[i])
arr = np.concatenate((np.array([0, 0]), features[i]))
faetures.append(arr)
np.savetxt('data/iod_net.csv',
lniks,
delimiter=',',
header='LINK,O,D,a0,a1,a2,a3,a4',
comments='')
np.savetxt('data/iod_net.txt',
faetures,
def load_chicago():
net = np.loadtxt('data/Chicago_net.csv', delimiter=',', skiprows=1)
demand = np.loadtxt('data/Chicago_od.csv', delimiter=',', skiprows=1)
node = np.loadtxt('data/Chicago_node.csv', delimiter=',', skiprows=1)
geometry = extract_features('data/ChicagoSketch_net.txt')
return net, demand, node, geometry
demand = np.loadtxt('data/Chicago_od.csv', delimiter=',', skiprows=1)
demand[:, 2] = 0.5 * demand[:, 2] / 4000
od = construct_od(demand)
# make link to lat/lng coordinate json file
if make_link2coord:
with open(
'{}/{}_net_od_2_alpha_0.{:02d}.txt'.format(dash_data, prefix, 50),
'r') as infile:
run = pickle.loads(infile.read())
fs = run['f']
f = np.array([l[0] + l[1] for l in fs])
if net == 'la':
node = np.loadtxt('data/LA_node.csv', delimiter=',')
features = np.zeros((f.shape[0], 4))
features[:, :3] = extract_features('data/LA_net.txt')
elif net == 'chicago':
node = np.loadtxt('data/Chicago_node.csv', delimiter=',', skiprows=1)
features = np.zeros((f.shape[0], 4))
features[:, :3] = extract_features('data/ChicagoSketch_net.txt')
f = np.divide(f * 4000, features[:, 0])
features[:, 3] = f
links = process_links(graph, node, features, in_order=True)
link2coord = {}
for i in range(len(links)):
link2coord[i] = {
'origin': [links[i, 0], links[i, 1]],
'destination': [links[i, 2], links[i, 3]],
}
with open('{}/link_coord.json'.format(out_data), 'w') as f:
def load_chicago():
net = np.loadtxt("data/Chicago_net.csv", delimiter=",", skiprows=1)
demand = np.loadtxt("data/Chicago_od.csv", delimiter=",", skiprows=1)
node = np.loadtxt("data/Chicago_node.csv", delimiter=",", skiprows=1)
geometry = extract_features("data/ChicagoSketch_net.txt")
return net, demand, node, geometry
