def test_get_model_mfd_noncum(self):
mod_mfd = get_model_mfd(self.bin_gdf)
mod_mfd_res = {
6.0: 0.0814075600000007,
6.2: 0.1160290199999997,
6.4: 0.0732093600000006,
6.6: 0.04619198000000058,
6.8: 0.02914517000000005,
7.0: 0.01838935999999996,
7.2: 0.011602900000000006,
7.4: 0.007320929999999995,
7.6: 0.004619199999999999,
7.8: 0.00162429,
8.0: 0,
8.2: 0,
8.4: 0,
8.6: 0,
8.8: 0,
9.0: 0,
9.2: 0,
}
for k, v in mod_mfd.items():
np.testing.assert_almost_equal(v, mod_mfd_res[k])
def test_get_model_mfd_cum(self):
mod_mfd = get_model_mfd(self.bin_gdf, cumulative=True)
mod_mfd_res = {
6.0: 0.38953977,
6.2: 0.30813221,
6.4: 0.19210319,
6.6: 0.11889383,
6.8: 0.07270185,
7.0: 0.04355668,
7.2: 0.02516732,
7.4: 0.01356442,
7.6: 0.00624349,
7.8: 0.00162429,
8.0: 0,
8.2: 0,
8.4: 0,
8.6: 0,
8.8: 0,
9.0: 0,
9.2: 0,
}
for k, v in mod_mfd.items():
np.testing.assert_almost_equal(v, mod_mfd_res[k])
def m_test_function(
bin_gdf,
t_yrs: float,
n_iters: int,
prospective=False,
not_modeled_likelihood: float = 0.0,
critical_pct: float = 0.25,
):
# get model and observed MFDs
mod_mfd = get_model_mfd(bin_gdf)
obs_mfd = get_obs_mfd(bin_gdf, t_yrs, prospective)
# calculate log-likelihoods
n_bins = len(mod_mfd.keys())
stochastic_eq_counts = {
bc: np.random.poisson((rate * t_yrs), size=n_iters)
for bc, rate in mod_mfd.items()
}
bin_log_likelihoods = {
bc: [
poisson_log_likelihood(
n_stoch,
(mod_mfd[bc] * t_yrs),
not_modeled_val=not_modeled_likelihood,
) for n_stoch in eq_counts
]
for bc, eq_counts in stochastic_eq_counts.items()
}
stoch_geom_mean_likes = np.array([
np.exp(
np.sum([bll_mag[i]
for bll_mag in bin_log_likelihoods.values()]) / n_bins)
for i in range(n_iters)
])
obs_geom_mean_like = np.exp(
np.sum([
poisson_log_likelihood(
int(obs_mfd[bc] * t_yrs),
rate * t_yrs,
) for bc, rate in mod_mfd.items()
]) / n_bins)
pctile = (len(
stoch_geom_mean_likes[stoch_geom_mean_likes <= obs_geom_mean_like]) /
n_iters)
test_pass = True if pctile >= critical_pct else False
test_res = "Pass" if test_pass else "Fail"
test_result = {
"critical_pct": critical_pct,
"percentile": pctile,
"test_pass": test_pass,
"test_res": test_res,
}
return test_result