def one_shot_distribution(self):
if self.cache_dist_3 is None:
t = {-1: 1 / 3, 0: 1 / 3, 1: 1 / 3}
Dfm = dist_iter_convolution(t, self.n - 1)
Dgr = dist_scale(Dfm, 3)
if self.improper:
Dfm = dist_scale(Dfm, 3)
self.cache_dist_3 = dist_convolution(Dfm, Dgr)
return self.cache_dist_3
def secret_l2_distribution(self):
if self.cache_dist_1 is None:
t = {0: 1 / 3, 1: 2 / 3}
Df = dist_iter_convolution(t, self.n - 1)
Dg = dist_scale(Df, 3**2)
if self.improper:
Df = dist_scale(Df, 3**2)
self.cache_dist_1 = (Df, Dg)
return self.cache_dist_1
def one_shot_s_quantile(self, lgu):
t = {0: 1 / 3, 1: 2 / 3}
D = dist_iter_convolution(t, self.n - 1)
pnz = expectation(top_quantile(D, 2**(lgu / 2))) / (self.n - 1)
t = {-1: pnz / 2, 0: 1 - pnz, 1: pnz / 2}
Dfm = dist_iter_convolution(t, self.n - 1)
Dgr = dist_scale(Dfm, 3)
if self.improper:
Dfm = dist_scale(Dfm, 3)
return dist_convolution(Dfm, Dgr)