def test_pval_calibration1():
dof = [1, 1, 1]
w = [0.5, 0.4, 0.1]
loc = [0.0, 0.0, 0.0]
random.seed(1)
samples = _sample(w, dof, loc, n=5000)
pvals = liu_sf(samples, w, dof, loc, kurtosis=False)[0]
assert_allclose(median(pvals), 0.5, rtol=1e-2)
pvals = liu_sf(samples, w, dof, loc, kurtosis=True)[0]
assert_allclose(median(pvals), 0.5, rtol=1e-2)
values = [davies_pvalue(s, diag(w)) for s in samples]
assert_allclose(median(values), 0.5, rtol=1e-2)
def test_liu_sf():
w = [0] * 64 + [4985811.050871694]
t = 906024.952053349
r = liu_sf(t, w, [1] * len(w), [0] * len(w))
assert_allclose(r[:-1], [0.6698986705489816, 1.0, 0], atol=1e-7)
info = r[-1]
assert_allclose(info["mu_q"], 4985811.050871694)
assert_allclose(info["sigma_q"], 7051001.607572404)
r = liu_sf(3.245, [1.02, 3.0], [1] * 2, [0] * 2)
assert_allclose(r[:-1], [0.4054052698281726, 1.2854059895411523, 0])
info = r[-1]
assert_allclose(info["mu_q"], 4.02)
assert_allclose(info["sigma_q"], 4.481160563961081)
assert_allclose(
liu_sf(2, [0.5, 0.4, 0.1], [1, 2, 1], [1, 0.6, 0.8])[0],
0.4577529852208846)
assert_allclose(
liu_sf(6, [0.5, 0.4, 0.1], [1, 2, 1], [1, 0.6, 0.8])[0],
0.0310791861868015)
assert_allclose(
liu_sf(3.5, [0.35, 0.15, 0.35, 0.15], [1, 1, 6, 2], [6, 2, 6, 2])[0],
0.9563148345837034,
)
assert_allclose(
liu_sf(2, [0.7, 0.3], [2, 1], [1.160032, 2])[0], 0.6243581256085478)
q = liu_sf([0.2, 0.5, 13.3], [0.7, 0.3], [2, 1], [1.160032, 2])[0]
assert_allclose(
q, [0.9863204788738281, 0.9475226498132635, 0.0022189212777912193])
def test_liu_sf_kurtosis():
lambs = [0.35, 0.15, 0.35, 0.15]
dofs = [1, 1, 6, 2]
deltas = [6, 2, 6, 2]
(q, dof_x, delta_x, info) = liu_sf(3.5, lambs, dofs, deltas, kurtosis=True)
assert_allclose(
[q, dof_x, delta_x],
[0.9563148345837034, 11.575377563761194, 10.278852084059992],
)
assert_allclose(info["mu_q"], 7.699999999999998)
assert_allclose(info["sigma_q"], 2.844292530665578)
assert_allclose(info["mu_x"], 21.854229647821185)
assert_allclose(info["sigma_x"], 8.016617956704831)
assert_allclose(info["t_star"], -1.4766413632627227)
def _mod_liu(values, weights):
"""Approximates survival function for mixtures of chi-square variables.
Wrapper around liu_sf which assumes each variable is centered with one
degree of freedom.
Args:
values: Points at which the surival function will be evaluated.
weights: Mixture weights.
Returns:
Approximated survival function applied to values as well as
matched mean and variance.
"""
(pv, dof_x, _, info) = liu_sf(
t=values, lambs=weights,
dofs=[1] * len(weights),
deltas=[0] * len(weights),
kurtosis=True)
return (pv, info["mu_q"], info["sigma_q"], dof_x)
def _mod_liu(q, w):
from chiscore import liu_sf
(pv, dof_x, _, info) = liu_sf(q, w, [1] * len(w), [0] * len(w), True)
return (pv, info["mu_q"], info["sigma_q"], dof_x)