def run(i):
with open ('data/json/comm' + str(i) + '.json', 'r') as fp:
data = json.load(fp)
network = data['trustMatrix']
opinion_attr = "sex"
n = len(network)
X = [0, 3, 6, 9, 15, 20, 30, 45, 60, 80, 100, 120, 150, 200, 300]
Y = [] # FTPL payoff
BR = [] # BR payoff
file_X = open("FTPL_X.txt", "w")
file_BR = open("FTPL_BR.txt", "w")
file_Y = open("FTPL_Y.txt", "w")
XAs, XBs = [], []
for x in X:
print("BUDGET ", x)
# file_X.write(str(x) + ', ')
A = Candidate("A", x, 1, n)
B = Candidate("B", 30, 0, n)
e = Election(data, [A, B], 10, opinion_attr, rand=False)
e.update_network()
ftpl(e, 3, 0.1)
ftpl_mean = e.calculate_mean()
XAs.append(A.X)
XBs.append(B.X)
Y.append(ftpl_mean)
file_Y.write(str(ftpl_mean)+ ', ')
print("sum", sum(e.A.X))
print("FTPL MEAN", e.advertise())
print("theta", e.theta_T)
print(sum(e.theta_T))
e.A.X = mov_oracle(e, A, np.zeros(n))
e.B.X = mov_oracle(e, B, np.zeros(n))
# print(A.X)
br_mean = e.calculate_mean()
BR.append(br_mean)
print("BR MEAN", br_mean)
file_BR.write(str(br_mean) + ', ')
for file in [file_X, file_BR, file_Y]:
file.flush()
print("means: {}, {}".format(ftpl_mean, br_mean))
file.write("X: {}".format(X))
file.write("Y: {}".format(Y))
file.write("BR: {}".format(BR))
file.write("X_A: {}".format(XAs))
file.write("X_B: {}".format(XBs))
file.close()
return X, np.array(Y)-np.array(BR)
def run(i):
with open ('data/json/comm' + str(i) + '.json', 'r') as fp:
data = json.load(fp)
network = data['trustMatrix']
opinion_attr = "sex"
n = len(network)
A = Candidate("A", 20, 1, n)
B = Candidate("B", 30, 0, n)
e = Election(data, [A, B], 30, opinion_attr, rand=False)
e.update_network()
h = np.array(e.calculate_homophily(e.theta))
if h[0] < h[1]:
mu = e.calculate_mean()/e.n
else:
mu = 1-e.calculate_mean()/e.n
print(h, mu)
return h, mu
def run(n, i):
with open ('data/json/comm' + str(i) + '.json', 'r') as fp:
data = json.load(fp)
N = len(data['trustMatrix'])
opinion_attr = "sex"
A = Candidate("A", 20, 1, n)
B = Candidate("B", 30, 0, n)
e = Election(data, n, [A, B], 5, opinion_attr, rand=False)
i, r, c_times, nc_times = iterated_best_response(e, 1e-2, 1e3)
c, nc = 0, 0
if len(c_times):
c = np.mean(c_times)
if len(nc_times):
nc = np.mean(nc_times)
# print(e.A.X)
# print(e.B.X)
# print('A POV', e.calculate_pov_exact())
# print('iterations:{}, restarts:{}'.format( i, r))
# print('avg convex time:{}, avg nonconvex time{}'.format(np.mean(c_times), np.mean(nc_times)))
return [i, r, c, nc]
def run(i):
with open('json/comm' + str(i) + '.json', 'r') as fp:
data = json.load(fp)
opinion_attr = "sex"
X = [0, 5, 10]
# X = [0, 5, 10, 15, 20, 30, 40, 50, 60, 70, 80, 90, 100, 105, 110, 115, 118, 120]
# X = [0, 3, 6, 9, 15, 20, 30, 45, 60, 80, 100, 120, 150, 200, 300]
# X = np.linspace(0, 120, 13)
n = 32
Y = [] # FTPL payoff
BR = [] # BR payoff
file_BR = open("data_fixed_sum_slurm/{}_br.txt".format(i), "w")
file_Y = open("data_fixed_sum_slurm/{}_Y.txt".format(i), "w")
A = Candidate("A", 0, 1, n)
B = Candidate("B", 0, 0, n)
e = Election(data, n, [A, B], 10, opinion_attr, rand=False)
for x in X:
print("BUDGET ", x)
A.k = x
B.k = 100 - x
e.update_network()
ftpl(e, 5, .1, x)
ftpl_mean = e.calculate_mean()
Y.append(ftpl_mean)
file_Y.write(str(ftpl_mean) + ', ')
A.X = mov_oracle(e, A, np.zeros(n))
B.X = mov_oracle(e, B, np.zeros(n))
br_mean = e.calculate_mean()
BR.append(br_mean)
file_BR.write(str(br_mean) + ', ')
for file in [file_BR, file_Y]:
file.flush()
print("means: {}, {}".format(ftpl_mean, br_mean))
file_BR.close()
file_Y.close()
return X, np.array(Y) - np.array(BR)
def run(i):
blockPrint()
with open('data/json/comm' + str(i) + '.json', 'r') as fp:
data = json.load(fp)
opinion_attr = "sex"
# X = [0, 5, 10]
# X = [0, 5, 10, 15, 20, 30, 40, 50, 60, 70, 80, 90, 100, 105, 110, 115, 117, 118, 119, 119.5, 120]
# X = [0, 3, 6, 9, 15, 20, 30, 45, 60, 80, 100, 120, 150, 200, 300]
X = list(np.linspace(0, 300, 2))
n = 32
# file = open("data_utility/{}.txt".format(i), "w")
A = Candidate("A", 0, 1, n)
B = Candidate("B", 0, 0, n)
e = Election(data, n, [A, B], 10, opinion_attr, rand=False)
# if i == 1:
# X = X[3:]
res = {}
for x in X:
budget_dict = {}
enablePrint()
# print("BUDGET ", x)
blockPrint()
A.k = x
B.k = max(X) - x
# print("BUDGETS 1", A.k, B.k)
e.update_network()
i, r, c, nc = iterated_best_response(e, 1e-2, 20, 1e3)
e.update_network()
alt_pov = e.calculate_pov_exact()
A.X, B.X = [0] * n, [0] * n
A_X_SBR, _, _, _ = pov_oracle(e, A, 0, e.n, 20, [], [])
A.X, B.X = [0] * n, [0] * n
B_X_SBR, _, _, _ = pov_oracle(e, B, 0, e.n, 20, [], [])
enablePrint()
A.X = A_X_SBR
B.X = B_X_SBR
e.update_network()
single_pov = e.calculate_pov_exact()
budget_dict["i"] = i
budget_dict["r"] = r
budget_dict["c"] = np.mean(c)
budget_dict["nc"] = np.mean(nc)
budget_dict["abr"] = alt_pov
budget_dict["sbr"] = single_pov
res[x] = json.dumps(budget_dict)
# file.flush()
# enablePrint()
# print("Completed in {} iters with {} restarts".format(i, r))
# print("ABR POV: {}, SBR POV: {}".format(alt_pov, single_pov))
# blockPrint()
print("{}: ".format(i) + json.dumps(res) + ", ")
# enablePrint()
# file.close()
A.X, B.X = [0] * n, [0] * n
e.update_network()
single_pov = e.calculate_pov_exact()
print("default POV: {}".format(single_pov))
return X
# e2 = Election(network, [A, B], 20, theta)
# e2.update_network()
# print("max spend 1", ftpl_max_spend(e2, A, B, 1))
# print("ad", e2.advertise([.5, 2], p_a, np.zeros(2), p_b))
# print("ad", e2.advertise([2.5, 0], p_a, np.zeros(2), p_b))
n = 3
P = [[1, 0, 0], [0.3, 0.5, 0.2], [0.2, 0.1, 0.7]] #np.identity(n)
network = {'trustMatrix': P}
p_a = [1, 0, 0.8]
p_b = [1, 0, 0.8]
theta = [1, 0, 1]
A = Candidate("A", 3, 1, n, p_a)
B = Candidate("B", 4, 0, n, p_b)
A.X = [1, 0, 0]
B.X = [0.75, 0, 1.25]
e3 = Election(network, [A, B], 3, theta=theta, rand=False)
e3.update_network()
# print(e3.calculate_homophily(e3.theta))
print(e3.theta)
print(A.p)
print(B.p)
# print(e3.theta_0)
# print(e3.theta_T)
ftpl(e3, 5, 1e-3)
# display(e3, A, B)
# A = Candidate("A", .5, 1, n, p_a)
# B = Candidate("B", 0, 0, n, p_b)
# e2 = Election(network, [A, B], 20, theta)
# e2.update_network()
def calc_pov(e):
return (round(e3.calculate_mean(), 3))
# return (round(e3.calculate_pov_exact(), 3))
n = 3
network = {'trustMatrix': np.identity(n)}
p_a = [0.5, 0.1, 1]
p_b = [0.7, 0.4, 1]
theta = [0.3, .2, 1]
A = Candidate("A", 3, 1, n, p_a)
B = Candidate("B", 3, 0, n, p_b)
e3 = Election(network, [A, B], 0, "sex", theta=theta)
# e3.A.X = [1,0,0]
# e3.B.X = [0,0,1]
# e3.update_network()
# print(e3.theta_0)
# print(e3.calculate_mean())
def test2(e):
e.A.X = [2.0, 0.0, 1.0]
e.B.X=[1.4285714285714286, 1.5714285714285714, 0.0]
e.update_network()
# print(e.theta_T)
# print(e.calculate_pov_exact())
# random_allocate(e, e.A)
iterated_best_response(e, 1e-4)
print(e.A.X)