def execute_C(x):
noisy_counts = real_counts.copy()
# print(noisy_counts)
# print(x)
noisy_counts[x] = noisy_counts[x] + 1 # Add the disease to the count
# print("///////////////")
# print(x)
# print(real_counts)
# print(noisy_counts)
# print("///////////////")
for d in range(real_counts.size):
cnt = noisy_counts[d] - range_start # 0-based count for x
# geom_cnt_row = np.array(Geom[cnt, :])
# print(geom_cnt_row)
noisy_counts[d] = probab.draw(
Geom[cnt, :]) + range_start # draw a noisy count from Geom's cnt-th row, map back to the real range
# rndmstt = create_new_rndm_state()
# obs = np.arange(start=range_start, stop=range_start + range_size, step=1)
# # print(len(obs))
# # print(len(Geom[cnt, :]))
# sample_distr = stats.rv_discrete(name='draw', values=(obs, Geom[cnt, :]), seed=rndmstt)
#
# noisy_counts[d] = sample_distr.rvs(size=1, random_state=rndmstt)
return noisy_counts
def create_single_dataset(size, C):
samples_list = []
for i in range(size):
x = probab.draw(pi)
y = execute_C(x, C)
samples_list.append([y, x])
return np.array(samples_list).reshape((size, len(samples_list[0])))
def draw_x_y():
# pi_ = np.array(pi)
x = probab.draw(pi)
return x, execute_C(x)
def execute_C(
x, C
): # we only have black box access to C. This function runs C under secret x and returns an output y
# C_x = np.array(C[x, :])
return probab.draw(C[x, :])