def test_multivariate_copula_simple(plt, rng):
n = 100000
c = 0.7
dist = MultivariateCopula(
[gaussian_icdf(-1, 1), uniform_icdf(-1, 1)], rho=[[1., c], [c, 1.]])
pts = dist.sample(n, rng=rng)
assert pts.shape == (n, 2)
plt.hist2d(pts[:, 0], pts[:, 1], bins=31)
def test_multivariate_copula_simple(plt, rng):
from nengo_extras.dists import gaussian_icdf, uniform_icdf
n = 100000
c = 0.7
dist = MultivariateCopula([gaussian_icdf(-1, 1),
uniform_icdf(-1, 1)],
rho=[[1., c], [c, 1.]])
pts = dist.sample(n, rng=rng)
assert pts.shape == (n, 2)
plt.hist2d(pts[:, 0], pts[:, 1], bins=31)
def test_icdfs(plt, rng):
p = rng.rand(100000)
rows = 3
plt.subplot(rows, 1, 1)
plt.hist(gaussian_icdf(1, 0.5)(p), bins=51)
plt.subplot(rows, 1, 2)
h, b = np.histogram(loggaussian_icdf(-2, 0.5, base=10)(p),
bins=np.logspace(-4, 0))
plt.semilogx(0.5 * (b[:-1] + b[1:]), h)
plt.subplot(rows, 1, 3)
plt.hist(uniform_icdf(-3, 1)(p), bins=51)
def test_icdfs(plt, rng):
from nengo_extras.dists import (
gaussian_icdf, loggaussian_icdf, uniform_icdf)
p = rng.rand(100000)
rows = 3
plt.subplot(rows, 1, 1)
plt.hist(gaussian_icdf(1, 0.5)(p), bins=51)
plt.subplot(rows, 1, 2)
h, b = np.histogram(loggaussian_icdf(-2, 0.5, base=10)(p),
bins=np.logspace(-4, 0))
plt.semilogx(0.5*(b[:-1] + b[1:]), h)
plt.subplot(rows, 1, 3)
plt.hist(uniform_icdf(-3, 1)(p), bins=51)