def combine_array(file1, file2):
f1 = np.array(loadF(file1))
f2 = np.array(loadF(file2))
f = np.concatenate((f1, f2), axis=1)
print(len(f), len(f[0]))
saveF(f, "400r_CI_result_MMR.pkl")
return
def calculate_variance_h1(f_fname="F_0_1_MMR.pkl"):
pi_l_list, pi_t_list = dp.get_pil_pit()
pi_l_list = np.array(pi_l_list).astype(np.float)
pi_t_list = np.array(pi_t_list).astype(np.float)
k = len(pi_l_list[0])
p_list = dp.loadF("h1.pkl")[:-1]
p_list = np.array(p_list).astype(np.float)
print("p_list", p_list)
f0_list, f1_list = dp.loadF(f_fname)
#print("0,1",f0_list[0],f1_list[0])
Var_list = []
for l in range(len(pi_l_list)):
Var_list.append([])
for t in range(len(pi_t_list)):
Var_list[l].append([])
F = change_f01_to_F(f0_list[l][t], f1_list[l][t])
for p in range(k):
var = calculate_variance(pi_l_list[l], pi_t_list[t], p_list[p],
F)
Var_list[l][t].append(var)
dp.saveF(Var_list, "var_ltp_h1_" + f_fname[6:-4])
print("var_list", Var_list)
def compare_var(first="var_ltp_h1_optReg", second="var_ltp_h1_MMR"):
first_var_list = dp.loadF(first)
second_var_list = dp.loadF(second)
first_vat_list = np.array(first_var_list)
second_var_list = np.array(second_var_list)
L = len(first_var_list)
T = len(first_var_list[0])
P = len(first_var_list[0][0])
total_len = L * T * P
print(total_len)
# all 10 p
rho10 = []
rho25 = []
rho50 = []
rho75 = []
rho_10 = []
rho_25 = []
rho_50 = []
rho_75 = []
pos = []
neg = []
all_rho = []
L15_T610_rho = []
L15_T610_diff = []
L15_T610_pos = 0
L610_T15_rho = []
L610_T15_diff = []
L610_T15_pos = 0
L15_T15_rho = []
L15_T15_diff = []
L15_T15_pos = 0
L610_T610_rho = []
L610_T610_diff = []
L610_T610_pos = 0
diag_rho = []
diag_diff = []
diag_pos = 0
for l in range(L):
for t in range(T):
for p in range(P):
rho = calculate_rho(first_var_list[l][t][p],
second_var_list[l][t][p])
all_rho.append(rho)
diff = first_var_list[l][t][p] - second_var_list[l][t][p]
if rho >= 0:
pos.append(diff)
if rho >= 0.75:
rho75.append(diff)
if rho >= 0.5:
rho50.append(diff)
if rho >= 0.25:
rho25.append(diff)
if rho >= 0.1:
rho10.append(diff)
else:
neg.append(diff)
if rho <= -0.75:
rho_75.append(diff)
if rho <= -0.5:
rho_50.append(diff)
if rho <= -0.25:
rho_25.append(diff)
if rho <= -0.1:
rho_10.append(diff)
if l < 5:
if t < 5:
L15_T15_rho.append(rho)
L15_T15_diff.append(diff)
if rho >= 0:
L15_T15_pos += 1
else:
L15_T610_rho.append(rho)
L15_T610_diff.append(diff)
if rho >= 0:
L15_T610_pos += 1
else:
if t < 5:
L610_T15_rho.append(rho)
L610_T15_diff.append(diff)
if rho >= 0:
L610_T15_pos += 1
else:
L610_T610_rho.append(rho)
L610_T610_diff.append(diff)
if rho >= 0:
L610_T610_pos += 1
if l == t:
diag_rho.append(rho)
diag_diff.append(diff)
if rho >= 0:
diag_pos += 1
print("10", len(rho10) / total_len, sum(rho10) / len(rho10))
print("25", len(rho25) / total_len, sum(rho25) / len(rho25))
print("50", len(rho50) / total_len, sum(rho50) / len(rho50))
if len(rho75) != 0:
print("75", len(rho75) / total_len, sum(rho75) / len(rho75))
print("-10", len(rho_10) / total_len, sum(rho_10) / len(rho_10))
print("-25", len(rho_25) / total_len, sum(rho_25) / len(rho_25))
print("-50", len(rho_50) / total_len, sum(rho_50) / len(rho_50))
print("-75", len(rho_75) / total_len, sum(rho_75) / len(rho_75))
print("pos",
len(pos) / total_len, L15_T610_pos / len(L15_T610_rho),
L610_T15_pos / len(L610_T15_rho), L15_T15_pos / len(L15_T15_rho),
L610_T610_pos / len(L610_T610_rho), diag_pos / len(diag_rho))
print("meandiff", (sum(pos) + sum(neg)) / total_len,
sum(L15_T610_diff) / len(L15_T610_rho),
sum(L610_T15_diff) / len(L610_T15_rho),
sum(L15_T15_diff) / len(L15_T15_rho),
sum(L610_T610_diff) / len(L610_T610_rho),
sum(diag_diff) / len(diag_rho))
print("meanrho",
sum(all_rho) / total_len,
sum(L15_T610_rho) / len(L15_T610_rho),
sum(L610_T15_rho) / len(L610_T15_rho),
sum(L15_T15_rho) / len(L15_T15_rho),
sum(L610_T610_rho) / len(L610_T610_rho),
sum(diag_rho) / len(diag_rho))
# p 1-5
total_len /= 2
rho10 = []
rho25 = []
rho50 = []
rho75 = []
rho_10 = []
rho_25 = []
rho_50 = []
rho_75 = []
pos = []
neg = []
all_rho = []
L15_T610_rho = []
L15_T610_diff = []
L15_T610_pos = 0
L610_T15_rho = []
L610_T15_diff = []
L610_T15_pos = 0
L15_T15_rho = []
L15_T15_diff = []
L15_T15_pos = 0
L610_T610_rho = []
L610_T610_diff = []
L610_T610_pos = 0
diag_rho = []
diag_diff = []
diag_pos = 0
for l in range(L):
for t in range(T):
for p in [0, 1, 2, 3, 4]:
rho = calculate_rho(first_var_list[l][t][p],
second_var_list[l][t][p])
all_rho.append(rho)
diff = first_var_list[l][t][p] - second_var_list[l][t][p]
if rho >= 0:
pos.append(diff)
if rho >= 0.75:
rho75.append(diff)
if rho >= 0.5:
rho50.append(diff)
if rho >= 0.25:
rho25.append(diff)
if rho >= 0.1:
rho10.append(diff)
else:
neg.append(diff)
if rho <= -0.75:
rho_75.append(diff)
if rho <= -0.5:
rho_50.append(diff)
if rho <= -0.25:
rho_25.append(diff)
if rho <= -0.1:
rho_10.append(diff)
if l < 5:
if t < 5:
L15_T15_rho.append(rho)
L15_T15_diff.append(diff)
if rho >= 0:
L15_T15_pos += 1
else:
L15_T610_rho.append(rho)
L15_T610_diff.append(diff)
if rho >= 0:
L15_T610_pos += 1
else:
if t < 5:
L610_T15_rho.append(rho)
L610_T15_diff.append(diff)
if rho >= 0:
L610_T15_pos += 1
else:
L610_T610_rho.append(rho)
L610_T610_diff.append(diff)
if rho >= 0:
L610_T610_pos += 1
if l == t:
diag_rho.append(rho)
diag_diff.append(diff)
if rho >= 0:
diag_pos += 1
print("10", len(rho10) / total_len, sum(rho10) / len(rho10))
print("25", len(rho25) / total_len, sum(rho25) / len(rho25))
print("50", len(rho50) / total_len, sum(rho50) / len(rho50))
if len(rho75) != 0:
print("75", len(rho75) / total_len, sum(rho75) / len(rho75))
print("-10", len(rho_10) / total_len, sum(rho_10) / len(rho_10))
print("-25", len(rho_25) / total_len, sum(rho_25) / len(rho_25))
print("-50", len(rho_50) / total_len, sum(rho_50) / len(rho_50))
print("-75", len(rho_75) / total_len, sum(rho_75) / len(rho_75))
print("pos",
len(pos) / total_len, L15_T610_pos / len(L15_T610_rho),
L610_T15_pos / len(L610_T15_rho), L15_T15_pos / len(L15_T15_rho),
L610_T610_pos / len(L610_T610_rho), diag_pos / len(diag_rho))
print("meandiff", (sum(pos) + sum(neg)) / total_len,
sum(L15_T610_diff) / len(L15_T610_rho),
sum(L610_T15_diff) / len(L610_T15_rho),
sum(L15_T15_diff) / len(L15_T15_rho),
sum(L610_T610_diff) / len(L610_T610_rho),
sum(diag_diff) / len(diag_rho))
print("meanrho",
sum(all_rho) / total_len,
sum(L15_T610_rho) / len(L15_T610_rho),
sum(L610_T15_rho) / len(L610_T15_rho),
sum(L15_T15_rho) / len(L15_T15_rho),
sum(L610_T610_rho) / len(L610_T610_rho),
sum(diag_rho) / len(diag_rho))
# p 5-10
rho10 = []
rho25 = []
rho50 = []
rho75 = []
rho_10 = []
rho_25 = []
rho_50 = []
rho_75 = []
pos = []
neg = []
all_rho = []
L15_T610_rho = []
L15_T610_diff = []
L15_T610_pos = 0
L610_T15_rho = []
L610_T15_diff = []
L610_T15_pos = 0
L15_T15_rho = []
L15_T15_diff = []
L15_T15_pos = 0
L610_T610_rho = []
L610_T610_diff = []
L610_T610_pos = 0
diag_rho = []
diag_diff = []
diag_pos = 0
for l in range(L):
for t in range(T):
for p in [5, 6, 7, 8, 9]:
rho = calculate_rho(first_var_list[l][t][p],
second_var_list[l][t][p])
all_rho.append(rho)
diff = first_var_list[l][t][p] - second_var_list[l][t][p]
if rho >= 0:
pos.append(diff)
if rho >= 0.75:
rho75.append(diff)
if rho >= 0.5:
rho50.append(diff)
if rho >= 0.25:
rho25.append(diff)
if rho >= 0.1:
rho10.append(diff)
else:
neg.append(diff)
if rho <= -0.75:
rho_75.append(diff)
if rho <= -0.5:
rho_50.append(diff)
if rho <= -0.25:
rho_25.append(diff)
if rho <= -0.1:
rho_10.append(diff)
if l < 5:
if t < 5:
L15_T15_rho.append(rho)
L15_T15_diff.append(diff)
if rho >= 0:
L15_T15_pos += 1
else:
L15_T610_rho.append(rho)
L15_T610_diff.append(diff)
if rho >= 0:
L15_T610_pos += 1
else:
if t < 5:
L610_T15_rho.append(rho)
L610_T15_diff.append(diff)
if rho >= 0:
L610_T15_pos += 1
else:
L610_T610_rho.append(rho)
L610_T610_diff.append(diff)
if rho >= 0:
L610_T610_pos += 1
if l == t:
diag_rho.append(rho)
diag_diff.append(diff)
if rho >= 0:
diag_pos += 1
print("10", len(rho10) / total_len, sum(rho10) / len(rho10))
print("25", len(rho25) / total_len, sum(rho25) / len(rho25))
print("50", len(rho50) / total_len, sum(rho50) / len(rho50))
if len(rho75) != 0:
print("75", len(rho75) / total_len, sum(rho75) / len(rho75))
print("-10", len(rho_10) / total_len, sum(rho_10) / len(rho_10))
print("-25", len(rho_25) / total_len, sum(rho_25) / len(rho_25))
print("-50", len(rho_50) / total_len, sum(rho_50) / len(rho_50))
print("-75", len(rho_75) / total_len, sum(rho_75) / len(rho_75))
print("pos",
len(pos) / total_len, L15_T610_pos / len(L15_T610_rho),
L610_T15_pos / len(L610_T15_rho), L15_T15_pos / len(L15_T15_rho),
L610_T610_pos / len(L610_T610_rho), diag_pos / len(diag_rho))
print("meandiff", (sum(pos) + sum(neg)) / total_len,
sum(L15_T610_diff) / len(L15_T610_rho),
sum(L610_T15_diff) / len(L610_T15_rho),
sum(L15_T15_diff) / len(L15_T15_rho),
sum(L610_T610_diff) / len(L610_T610_rho),
sum(diag_diff) / len(diag_rho))
print("meanrho",
sum(all_rho) / total_len,
sum(L15_T610_rho) / len(L15_T610_rho),
sum(L610_T15_rho) / len(L610_T15_rho),
sum(L15_T15_rho) / len(L15_T15_rho),
sum(L610_T610_rho) / len(L610_T610_rho),
sum(diag_rho) / len(diag_rho))
return
def percentage_success(filename):
true_value_list, target_vals_list, clt_inrange, t_inrange, chi_inrange, f_inrange = loadF(
filename)
method_list = [clt_inrange, t_inrange, chi_inrange, f_inrange]
target_data_list = []
for m in range(len(method_list)):
analyze_method = method_list[m]
target_data = []
opposite_data = []
for i in range(len(analyze_method)):
if analyze_method[i] == 1:
target_data.append(true_value_list[i])
else:
opposite_data.append(true_value_list[i])
ten_value_list = seperate_to_10(target_data)
opposite_ten_value_list = seperate_to_10(opposite_data)
target_data_list.append(ten_value_list /
(ten_value_list + opposite_ten_value_list))
plot_hist("Histalgram of percentage success (clt, t, chi, f)",
len(method_list), target_data_list)
def get_variance_info(l, t, p):
var_list = loadF("process_data\\var_f_optVar" + str(p) + ".pkl")
return var_list[l][t]
def analyze_success(filename):
true_value_list, target_vals_list, clt_inrange, t_inrange, chi_inrange, f_inrange = loadF(
filename)
method_list = [clt_inrange, t_inrange, chi_inrange, f_inrange]
target_data_list = []
for m in range(len(method_list)):
analyze_method = method_list[m]
target_data = []
for i in range(len(analyze_method)):
if analyze_method[i] == 1:
target_data.append(true_value_list[i])
ten_value_list = seperate_to_10(target_data)
ten_value_list /= len(target_data)
target_data_list.append(ten_value_list)
plot_hist("Histalgram of true value in CI (clt, t, chi, f)",
len(method_list), target_data_list, 0, 10, 0, 0.25)
def analyze_value(filename, index=0):
true_value_list, target_vals_list, clt_inrange, t_inrange, chi_inrange, f_inrange = loadF(
filename)
if index == 0:
ten_value_list = seperate_to_10(true_value_list)
ten_value_list /= len(true_value_list)
plot_hist("Histalgram of true value", 1, [ten_value_list])
return
def scatter_plot(coverage_file, seperate_variance=False, variance_file=None):
coverage = loadF(coverage_file)
fig = plt.figure()
if seperate_variance == False:
# get x and y
x = []
y = coverage
for l in range(11):
for p in range(10):
for t in range(11):
variance = get_variance_info(l, t, p)
x.append(variance)
# calculate covariance
covariance = stats.pearsonr(x, y)
print(covariance)
# plot
plt.scatter(x, y, s=0.6, color='blue')
plt.xlabel('variance by formula')
plt.ylabel('coverage')
plt.show()
fig.savefig('1' + '.pdf', bbox_inches='tight')
elif seperate_variance == "mean":
variance = loadF(variance_file)
# calculate covariance
covariance = stats.pearsonr(variance, coverage)
print(covariance)
plt.scatter(variance, coverage, s=0.6, color='blue')
plt.xlabel('variance by 400 instance')
plt.ylabel('coverage')
plt.show()
fig.savefig('2' + '.pdf', bbox_inches='tight')
elif seperate_variance == "VvsV":
variance_ins = loadF(variance_file)
x = []
for l in range(11):
for p in range(10):
for t in range(11):
variance = get_variance_info(l, t, p)
x.append(variance)
variance_formula = x
# calculate covariance
covariance = stats.pearsonr(variance_ins, variance_formula)
print(covariance)
plt.scatter(variance_ins, variance_formula, s=0.6, color='blue')
plt.xlabel('variance by 400 instance')
plt.ylabel('variance by formula')
plt.show()
fig.savefig('3' + '.pdf', bbox_inches='tight')
else:
x1 = []
x2 = []
y1 = []
y2 = []
i = 0
for l in range(11):
for p in range(10):
for t in range(11):
variance = get_variance_info(l, t, p)
if variance <= seperate_variance:
x1.append(variance)
y1.append(coverage[i])
else:
x2.append(variance)
y2.append(coverage[i])
i += 1
print("len for <= " + str(seperate_variance) + ": " + str(len(x1)))
print("len for > " + str(seperate_variance) + ": " + str(len(x2)))
# calculate covariance
covariance1 = stats.pearsonr(x1, y1)
covariance2 = stats.pearsonr(x2, y2)
print("covarance <= " + str(seperate_variance), covariance1)
print("covarance > " + str(seperate_variance), covariance2)
# plot
plt.scatter(x1, y1, s=0.7, color='blue')
plt.xlabel('variance by formula <= ' + str(seperate_variance))
plt.ylabel('coverage')
plt.show()
fig.savefig('4' + '.pdf', bbox_inches='tight')
plt.scatter(x2, y2, s=0.7, color='blue')
plt.xlabel('variance by formula > ' + str(seperate_variance))
plt.ylabel('coverage')
plt.show()
fig.savefig('5' + '.pdf', bbox_inches='tight')
def analyze_mean(mean_file, ind_file, ind=None):
mean_list = loadF(mean_file)
ind_list_clt, ind_list_t, ind_list_chi, ind_list_f = loadF(ind_file)
total_list = ind_list_clt + ind_list_t + ind_list_chi + ind_list_f
total_list = list(dict.fromkeys(total_list))
mean_list = np.array(mean_list).astype(np.float)
if ind != None:
if ind == "inf":
for i in total_list:
print("=====================================================")
target_mean_list = mean_list[i]
# normality
print(stats.normaltest(target_mean_list))
min_x = min(target_mean_list)
max_x = max(target_mean_list)
print("min and max:", min_x, max_x)
target_mean_list -= min_x
target_mean_list /= max_x - min_x
[ind_name], l, t, p = get_name_from_index([i])
print(ind_name)
# get variance
variance = get_variance_info(l, t, p)
print("variance:", variance)
ten_val_list = seperate_to_10(target_mean_list) / len(
target_mean_list)
plot_hist(
str(len(mean_list[0])) +
"r Histalgram of mean value with small coverage (" +
ind_name[:3] + ")", 1, [ten_val_list], min_x, max_x, 0, 1,
"Estimates")
else:
target_mean_list = mean_list[ind]
print(stats.normaltest(target_mean_list))
min_x = min(target_mean_list)
max_x = max(target_mean_list)
print("min and max:", min_x, max_x)
target_mean_list -= min_x
target_mean_list /= max_x - min_x
[ind_name], l, t, p = get_name_from_index([ind])
print(ind_name)
# get variance
variance = get_variance_info(l, t, p)
print("variance:", variance)
ten_val_list = seperate_to_10(target_mean_list) / len(
target_mean_list)
plot_hist(
str(len(mean_list[0])) +
"r Histalgram of mean value with small coverage (" +
ind_name[:3] + ")", 1, [ten_val_list], min_x, max_x, 0, 1,
"Estimates")
else:
target_mean_list = np.concatenate(mean_list[total_list])
print(get_name_from_index(total_list))
ten_val_list = seperate_to_10(target_mean_list) / len(target_mean_list)
plot_hist(
str(len(mean_list[0])) +
"r Histalgram of mean value with small coverage", 1,
[ten_val_list])
return
def get_info_coverage(filename, info_type, estimator="MMR"):
true_value_list, target_vals_list, clt_inrange, t_inrange, chi_inrange, f_inrange = loadF(
filename)
coverage_list_clt = np.zeros(1210)
coverage_list_t = np.zeros(1210)
coverage_list_chi = np.zeros(1210)
coverage_list_f = np.zeros(1210)
case_target_val = [[] for _ in range(1210)]
for i in range(len(target_vals_list)):
coverage_list_clt[i % 1210] += clt_inrange[i]
coverage_list_t[i % 1210] += t_inrange[i]
coverage_list_chi[i % 1210] += chi_inrange[i]
coverage_list_f[i % 1210] += f_inrange[i]
case_target_val[i % 1210].append(target_vals_list[i])
saveF(
case_target_val, "mean_" + str(int(len(target_vals_list) / 1210)) +
"r_" + estimator + ".pkl")
# save variance
var_mean_list = []
for i in range(len(case_target_val)):
var_mean_list.append(np.var(case_target_val[i]))
saveF(
var_mean_list, "var_mean_" + str(int(len(target_vals_list) / 1210)) +
"r_" + estimator + ".pkl")
coverage_list_clt /= (len(clt_inrange) / 1210)
coverage_list_t /= (len(clt_inrange) / 1210)
coverage_list_chi /= (len(clt_inrange) / 1210)
coverage_list_f /= (len(clt_inrange) / 1210)
saveF(coverage_list_clt, "coverage_clt_" + estimator + ".pkl")
saveF(coverage_list_t, "coverage_t_" + estimator + ".pkl")
saveF(coverage_list_chi, "coverage_chi_" + estimator + ".pkl")
saveF(coverage_list_f, "coverage_f_" + estimator + ".pkl")
overall_cov_clt = np.mean(coverage_list_clt)
overall_cov_t = np.mean(coverage_list_t)
overall_cov_chi = np.mean(coverage_list_chi)
overall_cov_f = np.mean(coverage_list_f)
print("overall coverage (clt,t,chi,f):", overall_cov_clt, overall_cov_t,
overall_cov_chi, overall_cov_f)
if info_type == "min":
cut_point_clt = np.percentile(coverage_list_clt, 0.49)
cut_point_t = np.percentile(coverage_list_t, 0.49)
cut_point_chi = np.percentile(coverage_list_chi, 0.49)
cut_point_f = np.percentile(coverage_list_f, 0.49)
print("min of clt,t,chi,f ; 0.5 percentile:")
print(min(coverage_list_clt), cut_point_clt)
print(min(coverage_list_t), cut_point_t)
print(min(coverage_list_chi), cut_point_chi)
print(min(coverage_list_f), cut_point_f)
ind_list_clt = []
ind_list_t = []
ind_list_chi = []
ind_list_f = []
for i in range(1210):
if coverage_list_clt[i] <= cut_point_clt:
ind_list_clt.append(i)
if coverage_list_t[i] <= cut_point_t:
ind_list_t.append(i)
if coverage_list_chi[i] <= cut_point_chi:
ind_list_chi.append(i)
if coverage_list_f[i] <= cut_point_f:
ind_list_f.append(i)
print("indlist:", ind_list_clt, ind_list_t, ind_list_chi, ind_list_f)
saveF([ind_list_clt, ind_list_t, ind_list_chi, ind_list_f],
"ind_" + str(int(len(target_vals_list) / 1210)) + "r_" +
estimator + ".pkl")
print("ind_" + str(int(len(target_vals_list) / 1210)) + "r_" +
estimator + ".pkl")
else:
cut_point_clt = np.percentile(coverage_list_clt, 99.5)
cut_point_t = np.percentile(coverage_list_t, 99.5)
cut_point_chi = np.percentile(coverage_list_chi, 99.5)
cut_point_f = np.percentile(coverage_list_f, 99.5)
print("max of clt,t,chi,f ; 99.5 percentile:")
print(max(coverage_list_clt), cut_point_clt)
print(max(coverage_list_t), cut_point_t)
print(max(coverage_list_chi), cut_point_chi)
print(max(coverage_list_f), cut_point_f)
ind_list_clt = []
ind_list_t = []
ind_list_chi = []
ind_list_f = []
for i in range(1210):
if coverage_list_clt[i] >= cut_point_clt:
ind_list_clt.append(i)
if coverage_list_t[i] >= cut_point_t:
ind_list_t.append(i)
if coverage_list_chi[i] >= cut_point_chi:
ind_list_chi.append(i)
if coverage_list_f[i] >= cut_point_f:
ind_list_f.append(i)
print("indlist:", ind_list_clt, ind_list_t, ind_list_chi, ind_list_f)
saveF([ind_list_clt, ind_list_t, ind_list_chi, ind_list_f],
"ind_" + str(int(len(target_vals_list) / 1210)) + "r_max_" +
estimator + ".pkl")
print("ind_" + str(int(len(target_vals_list) / 1210)) + "r_max_" +
estimator + ".pkl")
return
def check_data(filename):
true_value_list, target_vals_list, clt_inrange, t_inrange, chi_inrange, f_inrange = loadF(
filename)
print(np.mean(target_vals_list), len(target_vals_list))
def get_data(filename, index):
# if the data is [data1, data2] and index is 0, we will only get data1
data = dp.loadF(filename)
if index != None:
data = data[index]
return data
def off_policy_target_value(a, delta, pi_l, pi_t, p_astar):
data = dp.loadF("process_data\F_0_1_optVar" + str(p_astar) + ".pkl")
data = data[delta][pi_l][pi_t][a]
return data
