def test_plot_post_distr(self):
rng = np.random.RandomState(1)
weights_identical = np.ones((100, 1))
params = rng.randn(100, 2, 1, 1)
weights = np.arange(100).reshape(-1, 1)
journal = Journal(1)
journal.add_user_parameters([("par1", params[:, 0]),
("par2", params[:, 1])])
journal.add_user_parameters([("par1", params[:, 0]),
("par2", params[:, 1])])
journal.add_weights(weights=weights_identical)
journal.add_weights(weights=weights)
journal.plot_posterior_distr(single_marginals_only=True, iteration=0)
journal.plot_posterior_distr(true_parameter_values=[0.5, 0.3],
show_samples=True)
journal.plot_posterior_distr(double_marginals_only=True,
show_samples=True,
true_parameter_values=[0.5, 0.3])
journal.plot_posterior_distr(contour_levels=10,
ranges_parameters={"par1": [-1, 1]},
parameters_to_show=["par1"])
with self.assertRaises(KeyError):
journal.plot_posterior_distr(parameters_to_show=["par3"])
with self.assertRaises(RuntimeError):
journal.plot_posterior_distr(single_marginals_only=True,
double_marginals_only=True)
with self.assertRaises(RuntimeError):
journal.plot_posterior_distr(parameters_to_show=["par1"],
double_marginals_only=True)
with self.assertRaises(RuntimeError):
journal.plot_posterior_distr(parameters_to_show=["par1"],
true_parameter_values=[0.5, 0.3])
with self.assertRaises(TypeError):
journal.plot_posterior_distr(ranges_parameters={"par1": [-1]})
with self.assertRaises(TypeError):
journal.plot_posterior_distr(
ranges_parameters={"par1": np.zeros(1)})
param_names = [r"$\theta_1$", r"$\theta_2$", r"$\sigma_e$", r"$\phi$"]
theta1_min = 1.4
theta1_max = 2.2
theta2_min = 0
theta2_max = 1
sigma_e_min = 1.5
sigma_e_max = 2.5
phi_min = 0
phi_max = 1
ranges = dict([("theta1", [theta1_min, theta1_max]), ("theta2", [theta2_min, theta2_max]),
("sigma_e", [sigma_e_min, sigma_e_max]), ("phi", [phi_min, phi_max]), ])
fig, axes = jrnl.plot_posterior_distr(true_parameter_values=theta_obs, show_samples=False, show_density_values=False,
figsize=5, contour_levels=8, write_posterior_mean=False,
ranges_parameters=ranges)
figsize_actual = 2 * len(param_names)
# figsize_actual = 28
label_size = figsize_actual / len(param_names) * 10
title_size = figsize_actual / len(param_names) * 12
fig.suptitle(f"{'ABC' if inference_technique == 'ABC' else 'Exc'}-{technique}", size=title_size)
for j, label in enumerate(param_names):
# remove exponential notation:
axes[j, 0].ticklabel_format(style='plain', axis='y', scilimits=(0, 0))
axes[-1, j].ticklabel_format(style='plain', axis='x') # , scilimits=(0, 0))
axes[j, -1].ticklabel_format(style='plain', axis='y', scilimits=(0, 0))