def dummy():
# running_time()
pl_mat, cms, c = topo.read_data('../data/topologies/waxman_30_2_870.txt')
ufuncs = utility.read_ufuncs("../data/ufuncs/uf_870.txt")
scaled_ufuncs = utility.scale_ufuncs(ufuncs)
new_c = [10000.0 for i in range(len(c))]
c = new_c
pl_mat = np.matrix(pl_mat)
start_time = time.time()
cm_alloc, paths_alloc, useful_when_error = UMMP.max_min_program(
pl_mat, cms, c, scaled_ufuncs, len(cms))
tot_time_1 = time.time() - start_time
#print("UMMP total time is "+str(tot_time))
start_time = time.time()
initial_splits = UIEWF.exp_congestion_decay_splits(cms, pl_mat)
cm_alloc, diffs = UIEWF.Util_IEWF(pl_mat,
cms,
c,
initial_splits,
scaled_ufuncs,
it=10,
th=0.03)
tot_time_2 = time.time() - start_time
#print("UIEWF total time is "+str(tot_time))
start_time = time.time()
cm_alloc, diffs = Hybrid.hybrid_alloc(pl_mat,
cms,
c,
scaled_ufuncs,
it=10,
th=0.03,
k=500)
tot_time_3 = time.time() - start_time
#print("Hybrid total time is "+str(tot_time))
print("UMMP total time is " + str(tot_time_1))
print("UIEWF total time is " + str(tot_time_2))
print("Hybrid total time is " + str(tot_time_3))
def UIEWF_time():
size = [20, 30, 40]
times = []
for s in size:
time_sum = 0
for i in range(1, 11):
ufuncs = utility.read_ufuncs("../data/ufuncs/uf_" + str(s *
(s - 1)) +
"_v" + str(i) + ".txt")
scaled_ufuncs = utility.scale_ufuncs(ufuncs)
pl_mat, cms, c = topo.read_data("../data/topologies/waxman_" +
str(s) + "_3_" + str(s * (s - 1)) +
".txt")
new_c = [10000.0 for i in range(len(c))]
c = new_c
pl_mat = np.matrix(pl_mat)
start_time = time.time()
initial_splits = UIEWF.exp_congestion_decay_splits(cms, pl_mat)
cm_alloc, diffs = UIEWF.Util_IEWF(pl_mat,
cms,
c,
initial_splits,
scaled_ufuncs,
it=100,
th=0.03)
tot_time = time.time() - start_time
time_sum += tot_time
avg_time = time_sum / 10
times.append(avg_time)
print(times)
def hybrid_alloc(pl_mat, cms, c, ufuncs, it, th, k):
path_num = pl_mat.shape[0]
link_num = pl_mat.shape[1]
cms_alloc, paths_alloc, active_cms = UMMP.max_min_program(
pl_mat, cms, c, ufuncs, k)
splits = np.zeros(path_num)
paths_alloc = list(paths_alloc)
for cm in cms:
flow_sum = 0
for p in cm:
flow_sum += paths_alloc[p]
if flow_sum == 0:
for p in cm:
splits[p] = 1 / len(cm)
else:
for p in cm:
splits[p] = paths_alloc[p] / flow_sum
cms_alloc, diffs = UIEWF.Util_IEWF(pl_mat, cms, c, splits, ufuncs, it, th)
return cms_alloc, diffs
def running_time():
plt.rcParams['font.family'] = 'serif'
plt.rcParams['font.size'] = 18
plt.rcParams['font.weight'] = 'bold'
plt.rcParams['axes.labelweight'] = 'bold'
plt.rcParams['axes.labelsize'] = 18
fig, ax = plt.subplots()
fig.subplots_adjust(left=0.12, top=0.99, bottom=0.13, right=0.995)
pl_mat, cms, c = topo.read_data(
'D:/utility-mmf/scripts/dataset/convergence/waxman_20_4_380_1.txt')
new_c = [10000.0 for i in range(len(c))]
c = new_c
cm_nums = np.arange(100, 381, 40)
UIEWF_times = [0 for i in range(len(cm_nums))]
UIEWF_bin_times = [0 for i in range(len(cm_nums))]
for i in range(3, 4):
ufuncs = utility.read_ufuncs(
"D:/utility-mmf/scripts/dataset/convergence/uf_380_time_" +
str(i) + ".txt")
for i in range(len(cm_nums)):
cm_num = cm_nums[i]
comms = cms[:cm_num]
util_functions = ufuncs[:cm_num]
start_time = time.time()
initial_splits = UIEWF.exp_decay_splits(comms, pl_mat)
cm_alloc, diffs = UIEWF.Util_IEWF(pl_mat, comms, c, initial_splits,
util_functions, 5)
ummf_time = time.time() - start_time
UIEWF_times[i] += ummf_time
for i in range(len(cm_nums)):
cm_num = cm_nums[i]
comms = cms[:cm_num]
util_functions = ufuncs[:cm_num]
start_time = time.time()
initial_splits = UIEWF.exp_decay_splits(comms, pl_mat)
cm_alloc, diffs = UIEWF_bin.Util_IEWF(pl_mat, comms, c,
initial_splits,
util_functions, 5)
ummf_time = time.time() - start_time
UIEWF_bin_times[i] += ummf_time
ax.set_ylabel('Time(s)')
ax.set_xlabel('Number of commodities')
times = [t / 1 for t in UIEWF_times]
bin_times = [t / 1 for t in UIEWF_bin_times]
ax.plot(cm_nums, times, label='UIEWF', marker='.')
ax.plot(cm_nums, bin_times, label='UIEWF-bin', marker='v')
plt.xticks(np.arange(100, 381, 40))
plt.grid(ls='--')
plt.legend()
plt.show()
def test():
ufuncs=utility.read_ufuncs("D:/github/UtiliyUpwardMMF/data/ufuncs/uf_110.txt")
scaled_ufuncs=scale_utility_functions(ufuncs,100)
pl_mat,cms,c=topo.read_data("D:/github/UtiliyUpwardMMF/data/topologies/abilene_2_110.txt")
initial_splits=UIEWF.exp_decay_splits(cms,pl_mat)
model={'commodities':cms,'utility_functions':scaled_ufuncs,'splits':initial_splits,'pl_matrix':pl_mat,'capacities':c}
def performance_abilene():
plt.rcParams['font.family'] = 'serif'
plt.rcParams['font.size'] = 18
plt.rcParams['font.weight'] = 'bold'
plt.rcParams['axes.labelweight'] = 'bold'
plt.rcParams['axes.labelsize'] = 18
fig, ax = plt.subplots()
fig.subplots_adjust(left=0.16, top=0.995, bottom=0.13, right=0.995)
ufuncs = utility.read_ufuncs("../data/ufuncs/uf_110.txt")
pl_mat, cms, c = topo.read_data('../data/topologies/abilene_2_110.txt')
scaled_ufuncs = utility.scale_ufuncs(ufuncs)
new_c = [15000.0 for i in range(len(c))]
c = new_c
pl_mat = np.matrix(pl_mat)
x = [i for i in range(1, 111)]
#initial_splits=UIEWF.uniform_splits(cms)
#cm_alloc,diffs=UIEWF.Util_IEWF(pl_mat,cms,c,initial_splits,scaled_ufuncs,20)
#util_alloc=utility.flow2util(ufuncs,cm_alloc)
#ax.plot(x,sorted(util_alloc),label='uniform',linestyle='solid')
#initial_splits=UIEWF.random_splits(cms)
#cm_alloc,diffs=UIEWF.Util_IEWF(pl_mat,cms,c,initial_splits,scaled_ufuncs,20)
#util_alloc=utility.flow2util(ufuncs,cm_alloc)
#ax.plot(x,sorted(util_alloc),label='random',linestyle='solid')
#initial_splits=UIEWF.exp_decay_splits(cms,pl_mat)
#cm_alloc,diffs=UIEWF.Util_IEWF(pl_mat,cms,c,initial_splits,scaled_ufuncs,20)
#util_alloc=utility.flow2util(ufuncs,cm_alloc)
#ax.plot(x,sorted(util_alloc),label='len_decay',linestyle='solid')
initial_splits = UIEWF.exp_congestion_decay_splits(cms, pl_mat)
cm_alloc, diffs = UIEWF.Util_IEWF(pl_mat,
cms,
c,
initial_splits,
scaled_ufuncs,
it=20,
th=0.00001)
util_alloc = utility.flow2util(ufuncs, cm_alloc)
ax.plot(x, sorted(util_alloc), label='U-IEWF', linestyle='dotted')
cm_alloc, paths_alloc, useful_when_error = UMMP.max_min_program(
pl_mat, cms, c, scaled_ufuncs, len(cms))
util_alloc = utility.flow2util(ufuncs, cm_alloc)
ax.plot(x, sorted(util_alloc), label='Optimal', linestyle='solid')
cm_alloc, diffs = Hybrid.hybrid_alloc(pl_mat,
cms,
c,
scaled_ufuncs,
it=20,
th=0.00001,
k=20)
util_alloc = utility.flow2util(ufuncs, cm_alloc)
ax.plot(x, sorted(util_alloc), label='Hybrid-20', linestyle='dashed')
ax.set_xticks(np.arange(0, 101, step=10))
ax.set_ylabel('Utility')
ax.set_xlabel('Commodity Index')
#ax.grid(ls='--')
ax.legend()
plt.show()