lattice/pyerrors
1
0
Fork 0
mirror of https://github.com/fjosw/pyerrors.git synced 2025-05-14 11:33:42 +02:00
Code Issues Projects Releases Packages Wiki Activity

feat: positive semi-definite estimator for the covariance implemented,

Browse source 
fits.covariance matrix deprecated, covariance can now handle lists of
observables.
This commit is contained in:
Fabian Joswig 2022-03-01 09:45:25 +00:00
parent 8e3e34bbea
commit 82419b7a88
5 changed files with 65 additions and 79 deletions
Download patch file Download diff file Expand all files Collapse all files

6
tests/covobs_test.py
View file

@ -39,8 +39,8 @@ def test_covobs():
assert(np.isclose(oc.value, op.value, rtol=1e-14, atol=1e-14))
[o.gamma_method() for o in cl]
assert(pe.covariance(cl[0], cl[1]) == cov[0][1])
assert(pe.covariance(cl[0], cl[1]) == cov[1][0])
assert(pe.covariance([cl[0], cl[1]])[0, 1] == cov[0][1])
assert(pe.covariance([cl[0], cl[1]])[0, 1] == cov[1][0])
do = cl[0] * cl[1]
assert(np.array_equal(do.covobs['rAP'].grad, np.transpose([pi[1], pi[0]]).reshape(2, 1)))
@ -89,7 +89,7 @@ def test_covobs_covariance():
x = [a + b, a - b]
[o.gamma_method() for o in x]
covariance = pe.fits.covariance_matrix(x)
covariance = pe.covariance(x)
assert covariance[0, 0] == covariance[1, 1]
assert covariance[0, 1] == a.dvalue ** 2 - b.dvalue ** 2

10
tests/fits_test.py
View file

@ -60,7 +60,7 @@ def test_least_squares():
beta[i].gamma_method(S=1.0)
assert math.isclose(beta[i].value, popt[i], abs_tol=1e-5)
assert math.isclose(pcov[i, i], beta[i].dvalue ** 2, abs_tol=1e-3)
assert math.isclose(pe.covariance(beta[0], beta[1]), pcov[0, 1], abs_tol=1e-3)
assert math.isclose(pe.covariance([beta[0], beta[1]])[0, 1], pcov[0, 1], abs_tol=1e-3)
chi2_pyerrors = np.sum(((f(x, *[o.value for o in beta]) - y) / yerr) ** 2) / (len(x) - 2)
chi2_scipy = np.sum(((f(x, *popt) - y) / yerr) ** 2) / (len(x) - 2)
@ -81,7 +81,7 @@ def test_least_squares():
betac[i].gamma_method(S=1.0)
assert math.isclose(betac[i].value, popt[i], abs_tol=1e-5)
assert math.isclose(pcov[i, i], betac[i].dvalue ** 2, abs_tol=1e-3)
assert math.isclose(pe.covariance(betac[0], betac[1]), pcov[0, 1], abs_tol=1e-3)
assert math.isclose(pe.covariance([betac[0], betac[1]])[0, 1], pcov[0, 1], abs_tol=1e-3)
def test_alternative_solvers():
@ -195,7 +195,7 @@ def test_total_least_squares():
beta[i].gamma_method(S=1.0)
assert math.isclose(beta[i].value, output.beta[i], rel_tol=1e-5)
assert math.isclose(output.cov_beta[i, i], beta[i].dvalue ** 2, rel_tol=2.5e-1), str(output.cov_beta[i, i]) + ' ' + str(beta[i].dvalue ** 2)
assert math.isclose(pe.covariance(beta[0], beta[1]), output.cov_beta[0, 1], rel_tol=2.5e-1)
assert math.isclose(pe.covariance([beta[0], beta[1]])[0, 1], output.cov_beta[0, 1], rel_tol=2.5e-1)
out = pe.total_least_squares(ox, oy, func, const_par=[beta[1]])
@ -218,7 +218,7 @@ def test_total_least_squares():
betac[i].gamma_method(S=1.0)
assert math.isclose(betac[i].value, output.beta[i], rel_tol=1e-3)
assert math.isclose(output.cov_beta[i, i], betac[i].dvalue ** 2, rel_tol=2.5e-1), str(output.cov_beta[i, i]) + ' ' + str(betac[i].dvalue ** 2)
assert math.isclose(pe.covariance(betac[0], betac[1]), output.cov_beta[0, 1], rel_tol=2.5e-1)
assert math.isclose(pe.covariance([betac[0], betac[1]])[0, 1], output.cov_beta[0, 1], rel_tol=2.5e-1)
outc = pe.total_least_squares(oxc, oyc, func, const_par=[betac[1]])
@ -233,7 +233,7 @@ def test_total_least_squares():
betac[i].gamma_method(S=1.0)
assert math.isclose(betac[i].value, output.beta[i], rel_tol=1e-3)
assert math.isclose(output.cov_beta[i, i], betac[i].dvalue ** 2, rel_tol=2.5e-1), str(output.cov_beta[i, i]) + ' ' + str(betac[i].dvalue ** 2)
assert math.isclose(pe.covariance(betac[0], betac[1]), output.cov_beta[0, 1], rel_tol=2.5e-1)
assert math.isclose(pe.covariance([betac[0], betac[1]])[0, 1], output.cov_beta[0, 1], rel_tol=2.5e-1)
outc = pe.total_least_squares(oxc, oy, func, const_par=[betac[1]])

31
tests/obs_test.py
View file

@ -251,10 +251,10 @@ def test_covariance_is_variance():
dvalue = np.abs(np.random.normal(0, 1))
test_obs = pe.pseudo_Obs(value, dvalue, 't')
test_obs.gamma_method()
assert np.abs(test_obs.dvalue ** 2 - pe.covariance(test_obs, test_obs)) <= 10 * np.finfo(np.float64).eps
assert np.abs(test_obs.dvalue ** 2 - pe.covariance([test_obs, test_obs])[0, 1]) <= 10 * np.finfo(np.float64).eps
test_obs = test_obs + pe.pseudo_Obs(value, dvalue, 'q', 200)
test_obs.gamma_method()
assert np.abs(test_obs.dvalue ** 2 - pe.covariance(test_obs, test_obs)) <= 10 * np.finfo(np.float64).eps
assert np.abs(test_obs.dvalue ** 2 - pe.covariance([test_obs, test_obs])[0, 1]) <= 10 * np.finfo(np.float64).eps
def test_fft():
@ -268,21 +268,6 @@ def test_fft():
assert np.abs(test_obs1.dvalue - test_obs2.dvalue) <= 10 * max(test_obs1.dvalue, test_obs2.dvalue) * np.finfo(np.float64).eps
def test_covariance_symmetry():
value1 = np.random.normal(5, 10)
dvalue1 = np.abs(np.random.normal(0, 1))
test_obs1 = pe.pseudo_Obs(value1, dvalue1, 't')
test_obs1.gamma_method()
value2 = np.random.normal(5, 10)
dvalue2 = np.abs(np.random.normal(0, 1))
test_obs2 = pe.pseudo_Obs(value2, dvalue2, 't')
test_obs2.gamma_method()
cov_ab = pe.covariance(test_obs1, test_obs2)
cov_ba = pe.covariance(test_obs2, test_obs1)
assert np.abs(cov_ab - cov_ba) <= 10 * np.finfo(np.float64).eps
assert np.abs(cov_ab) < test_obs1.dvalue * test_obs2.dvalue * (1 + 10 * np.finfo(np.float64).eps)
def test_gamma_method_uncorrelated():
# Construct pseudo Obs with random shape
value = np.random.normal(5, 10)
@ -629,8 +614,8 @@ def test_covariance_symmetry():
dvalue2 = np.abs(np.random.normal(0, 1))
test_obs2 = pe.pseudo_Obs(value2, dvalue2, 't')
test_obs2.gamma_method()
cov_ab = pe.covariance(test_obs1, test_obs2)
cov_ba = pe.covariance(test_obs2, test_obs1)
cov_ab = pe.covariance([test_obs1, test_obs2])[0, 1]
cov_ba = pe.covariance([test_obs2, test_obs1])[0, 1]
assert np.abs(cov_ab - cov_ba) <= 10 * np.finfo(np.float64).eps
assert np.abs(cov_ab) < test_obs1.dvalue * test_obs2.dvalue * (1 + 10 * np.finfo(np.float64).eps)
@ -643,10 +628,10 @@ def test_covariance_symmetry():
idx = [i + 1 for i in range(len(configs)) if configs[i] == 1]
a = pe.Obs([zero_arr], ['t'], idl=[idx])
a.gamma_method()
assert np.isclose(a.dvalue**2, pe.covariance(a, a), atol=100, rtol=1e-4)
assert np.isclose(a.dvalue ** 2, pe.covariance([a, a])[0, 1], atol=100, rtol=1e-4)
cov_ab = pe.covariance(test_obs1, a)
cov_ba = pe.covariance(a, test_obs1)
cov_ab = pe.covariance([test_obs1, a])[0, 1]
cov_ba = pe.covariance([a, test_obs1])[0, 1]
assert np.abs(cov_ab - cov_ba) <= 10 * np.finfo(np.float64).eps
assert np.abs(cov_ab) < test_obs1.dvalue * a.dvalue * (1 + 10 * np.finfo(np.float64).eps)
@ -704,4 +689,4 @@ def test_merge_idx():
new_idx = pe.obs._merge_idx(idl)
assert(new_idx[-1] > new_idx[0])
for i in range(1, len(new_idx)):
assert(new_idx[i - 1] < new_idx[i])
assert(new_idx[i - 1] < new_idx[i])