@property
def mean(self):
return self.alpha / self.lamb
@property
def var(self):
return self.alpha / self.lamb**2
def _sample(self):
raise NotImplementedError
def setup_distribution():
ugen = RngStream()
dist = Gamma(alpha=1.5, lamb=6, ugen=ugen)
return dist
if __name__ == "__main__":
parser = get_parser()
args = parser.parse_args()
num_samples = args.num_samples
dist = setup_distribution()
samples = dist.sample(num_samples)
DensityPlotter(dist=dist).make_plots(samples.flatten())
test_mean(dist, samples)
return self._var
def _sample(self):
u1, u2 = self.ugen1.RandU01(), self.ugen2.RandU01()
return self.low + (u1 + u2) * (self.width / 2)
def setup_distribution():
ugen1 = RngStream()
ugen2 = RngStream()
# symm_tri_dist = SymmetricTriangular(ugen1=ugen1, ugen2=ugen2)
symm_tri_dist = SymmetricTriangular(low=1,
high=10,
ugen1=ugen1,
ugen2=ugen2)
# symm_tri_dist = Gaussian(mu=10, sigma=1, ugen1=ugen1, ugen2=ugen2)
return symm_tri_dist
if __name__ == "__main__":
parser = get_parser()
args = parser.parse_args()
num_samples = args.num_samples
symm_tri_dist = setup_distribution()
samples = symm_tri_dist.sample(num_samples)
DensityPlotter(dist=symm_tri_dist).make_plots(samples.flatten())
test_mean(symm_tri_dist, samples)
class StdNormal(Gaussian):
def __init__(self, ugen1: RngStream, ugen2: RngStream):
super().__init__(0, 1, ugen1, ugen2)
def _sample(self):
try:
return self.cache.pop()
except IndexError:
self.fill_cache()
return self.cache.pop()
if __name__ == "__main__":
parser = get_parser()
args = parser.parse_args()
num_samples = args.num_samples
ugen1 = RngStream()
ugen2 = RngStream()
normal_dist = StdNormal(ugen1=ugen1, ugen2=ugen2)
# normal_dist = Gaussian(mu=10, sigma=1, ugen1=ugen1, ugen2=ugen2)
samples = normal_dist.sample(num_samples)
DensityPlotter(dist=normal_dist).make_plots(samples.flatten())
test_mean(normal_dist, samples)
from distributions.make_plots import DensityPlotter
from distributions.utils import get_parser, test_mean
class Exponential(Distribution):
def __init__(self, lamb: float, ugen: RngStream):
self.lamb = lamb
self.mean = 1 / lamb
self.var = 1 / lamb**2
self.ugen = ugen
def _sample(self):
u = self.ugen.RandU01()
return -log(u) / self.lamb
if __name__ == "__main__":
parser = get_parser()
args = parser.parse_args()
num_samples = args.num_samples
ugen = RngStream()
exp_dist = Exponential(1, ugen=ugen)
samples = exp_dist.sample(num_samples)
DensityPlotter(dist=exp_dist).make_plots(samples.flatten())
test_mean(exp_dist, samples)
def sample(self, *shape):
num_samples = np.prod(shape, dtype=int)
samples = [self._sample() for _ in range(num_samples)]
samples = np.reshape(samples, shape)
return samples
class Uniform01(Uniform):
def __init__(self, ugen: RngStream):
super().__init__(0, 1, ugen)
def _sample(self):
return self.ugen.RandU01()
if __name__ == "__main__":
parser = get_parser()
args = parser.parse_args()
num_samples = args.num_samples
ugen = RngStream()
# unif_dist = Uniform01(ugen=ugen)
unif_dist = Uniform(100, 500, ugen=ugen)
samples = unif_dist.sample(num_samples, 100)
DensityPlotter(dist=unif_dist).make_plots(samples.flatten())
test_mean(unif_dist, samples)