def test_bad_params_t():
with pytest.raises(ValueError) as e:
t_copula(cov,
df=0,
samples=4,
seed=123,
chunks=None,
rng="Mersenne Twister")
assert str(e.value) == "df must always be greater than zero"
with pytest.raises(ValueError) as e:
t_copula(cov,
df=[1, 0, 1],
samples=4,
seed=123,
chunks=None,
rng="Mersenne Twister")
assert str(e.value) == "df must always be greater than zero"
with pytest.raises(ValueError) as e:
t_copula(cov,
df=[1, 1],
samples=4,
seed=123,
chunks=None,
rng="Mersenne Twister")
assert str(e.value) == (
"df must be either a scalar or a 1D vector with as many points as the width "
"of the correlation matrix")
def test_student_t_mersenne_da():
actual = t_copula(cov,
df=3,
samples=4,
seed=123,
chunks=2,
rng="Mersenne Twister")
expect = [
[0.9251499, 0.78838765, 0.52075254],
[0.00733126, 0.58581369, -0.50181461],
[-0.46342698, -0.55862319, -0.25037537],
[-0.92150594, -0.70869994, -0.84036731],
]
assert_allclose(expect, actual, 1e-6, 0)
assert actual.chunks == ((2, 2), (2, 1))
def test_student_t_mersenne_np():
actual = t_copula(cov,
df=3,
samples=4,
seed=123,
chunks=None,
rng="Mersenne Twister")
expect = [
[-1.48810072, -0.86441534, -1.54668921],
[-1.29120595, -1.35445918, 0.04614266],
[-1.69841381, -1.67427423, -0.76713426],
[-0.52427748, -0.64222262, -0.18205599],
]
assert_allclose(expect, actual, 1e-6, 0)
assert isinstance(actual, np.ndarray)
def test_tail_dependence(df, expect_td, rng, chunks):
cov2 = [
[1.0, 0.5, 0.5, 0.5],
[0.5, 1.0, 0.5, 0.5],
[0.5, 0.5, 1.0, 0.5],
[0.5, 0.5, 0.5, 1.0],
]
s = t_copula(cov2, df=df, samples=262143, rng=rng, chunks=chunks)
s = scipy.stats.norm.cdf(s)
actual_td = [
tail_dependence(s[:, 0], s[:, 1], 0.99),
tail_dependence(s[:, 1], s[:, 2], 0.99),
tail_dependence(s[:, 2], s[:, 3], 0.99),
]
assert_allclose(expect_td, actual_td, atol=0.02, rtol=0)
def test_it_mersenne_da():
actual = t_copula(cov,
df=[3, 4, 5],
samples=4,
seed=123,
chunks=2,
rng="Mersenne Twister")
expect = [
[0.92514990, 0.84367196, 0.57859631],
[0.00733126, 0.60760124, -0.53208793],
[-0.46342698, -0.60402557, -0.28417274],
[-0.92150594, -0.75838980, -0.94757843],
]
assert_allclose(expect, actual, 1e-6, 0)
assert actual.chunks == ((2, 2), (2, 1))
def test_it_mersenne_np():
actual = t_copula(cov,
df=[3, 4, 5],
samples=4,
seed=123,
chunks=None,
rng="Mersenne Twister")
expect = [
[-1.48810072, -0.80579141, -1.48680149],
[-1.29120595, -1.40226642, 0.04565698],
[-1.69841381, -1.77537961, -0.79048734],
[-0.52427748, -0.68508399, -0.20095568],
]
assert_allclose(expect, actual, 1e-6, 0)
assert isinstance(actual, np.ndarray)
def test_it_sobol(chunks, expect_chunks):
actual = t_copula(cov,
df=[3, 4, 5],
samples=4,
seed=123,
chunks=chunks,
rng="Sobol")
expect = [
[0.0, 0.0, 0.0],
[-0.90292647, -0.41928686, -1.35361744],
[0.51756147, 0.25248047, 0.91032037],
[0.59093028, 1.20943444, -0.81940488],
]
assert_allclose(expect, actual, 1e-6, 0)
if chunks:
assert actual.chunks == expect_chunks
else:
assert isinstance(actual, np.ndarray)
def test_student_t_sobol(chunks, expect_chunks):
actual = t_copula(cov,
df=3,
samples=4,
seed=123,
chunks=chunks,
rng="Sobol")
expect = [
[0.0, 0.0, 0.0],
[-0.90292647, -0.44513114, -1.38033019],
[0.51756147, 0.24504617, 0.84650386],
[0.59093028, 1.28328308, -0.94456215],
]
assert_allclose(expect, actual, 1e-6, 0)
if chunks:
assert actual.chunks == expect_chunks
else:
assert isinstance(actual, np.ndarray)