def test_get_mean_std(self):
""" Check calculation of mean and stds """
stds_types = [
const.StdDev.TOTAL, const.StdDev.INTER_EVENT,
const.StdDev.INTRA_EVENT
]
gmpe_name = 'AkkarEtAlRjb2014'
gmm = ModifiableGMPE(gmpe={gmpe_name: {}},
set_between_epsilon={'epsilon_tau': 0.5})
mean, stds = gmm.get_mean_and_stddevs(self.sites, self.rup, self.dists,
self.imt, stds_types)
gmpe = registry[gmpe_name]()
emean, estds = gmpe.get_mean_and_stddevs(self.sites, self.rup,
self.dists, self.imt,
stds_types)
idx = stds_types.index(const.StdDev.INTER_EVENT)
exp_mean = emean + estds[idx] * 0.5
# Check the computed mean + between event variability
np.testing.assert_almost_equal(mean, exp_mean)
# Check that the total std now corresponds to the within event
# standard deviation
np.testing.assert_almost_equal(stds[0], estds[2])
def test_set_between_epsilon_raises_error(self):
""" Check that error is raised for GMPEs with only total std """
stds_types = [const.StdDev.TOTAL]
gmm = ModifiableGMPE(gmpe={'Campbell2003': {}},
set_between_epsilon={'epsilon_tau': 0.5})
with self.assertRaises(ValueError):
_, _ = gmm.get_mean_and_stddevs(self.sites, self.rup, self.dists,
self.imt, stds_types)
def test_set_total_std_as_tau_plus_phi(self):
"""Check set total std as between plus phi SS"""
gmpe_name = 'AkkarEtAlRjb2014'
stddevs = [const.StdDev.TOTAL]
gmm = ModifiableGMPE(gmpe={gmpe_name: {}},
set_total_std_as_tau_plus_delta={"delta": 0.45})
imt = PGA()
[stddev] = gmm.get_mean_and_stddevs(self.sites, self.rup, self.dists,
imt, stddevs)[1]
# Original tau for PGA is 0.6201
np.testing.assert_almost_equal(stddev[0], 0.5701491121, decimal=6)
def test_add_delta(self):
"""Check adding/removing a delta std to the total std"""
gmpe_name = 'AkkarEtAlRjb2014'
stddevs = [const.StdDev.TOTAL]
gmm = ModifiableGMPE(gmpe={gmpe_name: {}},
add_delta_std_to_total_std={"delta": -0.20})
imt = PGA()
[stddev] = gmm.get_mean_and_stddevs(self.sites, self.rup, self.dists,
imt, stddevs)[1]
# Original total std for PGA is 0.7121
np.testing.assert_almost_equal(stddev[0], 0.68344277, decimal=6)
def test_add_between_within_sigma(self):
""" Checks that the modified GMM has between and within std """
stds_types = [
const.StdDev.TOTAL, const.StdDev.INTRA_EVENT,
const.StdDev.INTER_EVENT
]
gmm = ModifiableGMPE(gmpe={'Campbell2003': {}},
add_between_within_stds={'with_betw_ratio': 0.6})
out = gmm.get_mean_and_stddevs(self.ctx, self.ctx, self.ctx, self.imt,
stds_types)
expected_with = np.ones(4) * 0.24381161
expected_betw = np.ones(4) * 0.40635268
aae = np.testing.assert_array_almost_equal
aae(expected_betw, out[1][2])
aae(expected_with, out[1][1])
def test_fixed_total_sigma(self):
"""Check the assignment of total sigma to a fixed value"""
gmpe_name = 'AkkarEtAlRjb2014'
stddevs = [const.StdDev.TOTAL]
gmm = ModifiableGMPE(gmpe={gmpe_name: {}},
set_fixed_total_sigma={
"total_sigma": {
"PGA": 0.6,
"SA(0.2)": 0.75
}
})
for imt, sfact in zip([PGA(), SA(0.2)], [0.6, 0.75]):
[stddev] = gmm.get_mean_and_stddevs(self.sites, self.rup,
self.dists, imt, stddevs)[1]
np.testing.assert_almost_equal(stddev,
sfact * np.ones(stddev.shape))
def test_coefficients_as_dictionary(self):
"""Check the parsing of the coefficients to a dictionary"""
input_coeffs = {"PGA": 1.0, "SA(0.2)": 2.0, "SA(3.0)": 3.0}
output_coeffs = ModifiableGMPE._dict_to_coeffs_table(
input_coeffs, "XYZ")
self.assertListEqual(list(output_coeffs), ["XYZ"])
self.assertIsInstance(output_coeffs["XYZ"], CoeffsTable)
self.assertAlmostEqual(output_coeffs["XYZ"][PGA()]["XYZ"], 1.0)
self.assertAlmostEqual(output_coeffs["XYZ"][SA(0.2)]["XYZ"], 2.0)
self.assertAlmostEqual(output_coeffs["XYZ"][SA(3.0)]["XYZ"], 3.0)
def test_get_mean_std(self):
""" Check calculation of amplified mean"""
for gmpe_name in [
'ECOS2009', 'BindiEtAl2011RepiFixedH',
'BaumontEtAl2018High2210IAVGDC30n7', 'FaccioliCauzzi2006'
]:
stds_types = [const.StdDev.TOTAL]
gmm = ModifiableGMPE(gmpe={gmpe_name: {}},
apply_swiss_amplification={})
mean, stds = gmm.get_mean_and_stddevs(self.ctx, self.ctx, self.ctx,
self.imt, stds_types)
gmpe = registry[gmpe_name]()
emean, estds = gmpe.get_mean_and_stddevs(self.ctx, self.ctx,
self.ctx, self.imt,
stds_types)
exp_mean = emean + np.array([-1.00, 1.50, 0, -1.99])
# Check the computed mean + amplification
np.testing.assert_almost_equal(mean, exp_mean)
def test_scale_median_scalar(self):
"""Check the scaling of the median ground motion - scalar"""
gmpe_name = 'AkkarEtAlRjb2014'
stddevs = [const.StdDev.TOTAL]
gmm_unscaled = ModifiableGMPE(gmpe={gmpe_name: {}})
gmm = ModifiableGMPE(gmpe={gmpe_name: {}},
set_scale_median_scalar={'scaling_factor': 1.2})
mean_unscaled = gmm_unscaled.get_mean_and_stddevs(
self.sites, self.rup, self.dists, self.imt, stddevs)[0]
mean = gmm.get_mean_and_stddevs(self.sites, self.rup, self.dists,
self.imt, stddevs)[0]
np.testing.assert_almost_equal(
np.exp(mean) / np.exp(mean_unscaled), 1.2 * np.ones(mean.shape))
def test_scale_total_sigma_scalar(self):
"""Check the scaling of the total stddev - scalar"""
gmpe_name = 'AkkarEtAlRjb2014'
stddevs = [const.StdDev.TOTAL]
gmm_unscaled = ModifiableGMPE(gmpe={gmpe_name: {}})
gmm = ModifiableGMPE(
gmpe={gmpe_name: {}},
set_scale_total_sigma_scalar={"scaling_factor": 1.2})
[stddev_unscaled
] = gmm_unscaled.get_mean_and_stddevs(self.ctx, self.ctx, self.ctx,
self.imt, stddevs)[1]
[stddev] = gmm.get_mean_and_stddevs(self.ctx, self.ctx, self.ctx,
self.imt, stddevs)[1]
np.testing.assert_array_almost_equal(stddev / stddev_unscaled,
1.2 * np.ones(stddev.shape))
def test_scale_median_vector(self):
"""Check the scaling of the median ground motion - IMT-dependent"""
gmpe_name = 'AkkarEtAlRjb2014'
stddevs = [const.StdDev.TOTAL]
gmm_unscaled = ModifiableGMPE(gmpe={gmpe_name: {}})
gmm = ModifiableGMPE(gmpe={gmpe_name: {}},
set_scale_median_vector={
'scaling_factor': {
"PGA": 0.9,
"SA(0.2)": 1.1
}
})
for imt, sfact in zip([PGA(), SA(0.2)], [0.9, 1.1]):
mean_unscaled = gmm_unscaled.get_mean_and_stddevs(
self.sites, self.rup, self.dists, imt, stddevs)[0]
mean = gmm.get_mean_and_stddevs(self.sites, self.rup, self.dists,
imt, stddevs)[0]
np.testing.assert_almost_equal(
np.exp(mean) / np.exp(mean_unscaled),
sfact * np.ones(mean.shape))
def test_scale_total_sigma_vector(self):
"""Check the scaling of the total stddev - vector"""
gmpe_name = 'AkkarEtAlRjb2014'
stddevs = [const.StdDev.TOTAL]
gmm_unscaled = ModifiableGMPE(gmpe={gmpe_name: {}})
gmm = ModifiableGMPE(gmpe={gmpe_name: {}},
set_scale_total_sigma_vector={
"scaling_factor": {
"PGA": 0.9,
"SA(0.2)": 1.1
}
})
for imt, sfact in zip([PGA(), SA(0.2)], [0.9, 1.1]):
[stddev_unscaled
] = gmm_unscaled.get_mean_and_stddevs(self.sites, self.rup,
self.dists, imt, stddevs)[1]
[stddev] = gmm.get_mean_and_stddevs(self.sites, self.rup,
self.dists, imt, stddevs)[1]
np.testing.assert_almost_equal(stddev / stddev_unscaled,
sfact * np.ones(stddev.shape))
