import numpy as np
import pytest
from beartype.roar import BeartypeCallHintPepParamException
from UQpy.distributions import Normal, MultivariateNormal
from UQpy.sampling import MonteCarloSampling
dist1 = Normal(loc=0., scale=1.)
dist2 = Normal(loc=0., scale=1.)
x = MonteCarloSampling(distributions=dist1, nsamples=5, random_state=np.random.RandomState(123))
x.transform_u01()
y = MonteCarloSampling(distributions=[dist1, dist2])
y.run(nsamples=5, random_state=123)
y.transform_u01()
# Call run method multiple time, to cover lines where samples are append to existing ones
z1 = MonteCarloSampling(distributions=dist1, nsamples=2, random_state=123)
z1.run(nsamples=2)
z2 = MonteCarloSampling(distributions=[dist1, dist2], nsamples=2, random_state=np.random.RandomState(123))
z2.run(nsamples=2)
# Same object as z2, just to cover lines where, random_state is an integer
z3 = MonteCarloSampling(distributions=[dist1, dist2], nsamples=2, random_state=123)
z4 = MonteCarloSampling(distributions=[MultivariateNormal([0, 0])], nsamples=2,
random_state=np.random.RandomState(123))
z4.run(nsamples=2)
z4.transform_u01()
dist3 = Normal(loc=0., scale=1.)
del dist3.rvs
z5 = MonteCarloSampling(distributions=[dist3], random_state=np.random.RandomState(123))
# %%
mc.run(nsamples=2, random_state=RandomState(23))
mc.samples
# %% md
#
# We can transform the samples onto the unit hypercube via applying the probability integral transformation on the
# samples to yield similar samples from the uniform distribution. We call :code:`mc.transform_u01`, from which results
# are stored in the :code:`samplesU01` attribute.
# %%
mc.transform_u01()
mc.samplesU01
# %% md
#
# We can visualize the (untransformed) samples by plotting them on axes of each distribution's range.
# %%
fig, ax = plt.subplots()
plt.title('Samples')
plt.scatter(x=mc.samples[:, 0], y=mc.samples[:, 1], marker='o')
plt.setp(ax, xlim=(-1.7, 1.7), ylim=(-2.6, 2.6))