l1_prior_pf_1hf_2lf_avg = np.zeros((n_grid,))
for k in range(n_avg):
print('\nRun %d / %d' % (k + 1, n_avg))
# -------------- 1 HF
l1_prior_pf_1hf = []
for idx, n_evals in enumerate(n_evals_mc):
indices = np.random.choice(range(prior_pf_samples_hf.shape[0]), size=n_evals, replace=False)
# Monte Carlo model
prior_pf_samples = prior_pf_samples_hf[indices]
p_prior_pf = Distribution(prior_pf_samples, rv_name='$Q$', label='Prior-PF')
# l1 error between prior push-forward and reference push-forward
l1_prior_pf_1hf.append(ref_p_prior_pf.calculate_l1_error(p_prior_pf))
# -------------- 1 HF, 1 LF
l1_prior_pf_1hf_1lf = []
for idx, n_evals in enumerate(n_evals_mfmc_hf_1lf):
n_evals = [n_evals_mfmc_lf_1lf[idx], n_evals]
indices = np.random.choice(range(prior_samples.shape[0]), size=n_evals[0], replace=False)
# Create a low-fidelity model
lf_model = Model(
eval_fun=lambda x, samples=prior_pf_samples_lf: elliptic_pde.find_xy_pair(x, prior_samples, samples),
rv_samples=prior_samples[indices], rv_samples_pred=prior_samples[indices], n_evals=n_evals[0],
n_qoi=n_qoi, rv_name='$q_0$', label='Low-fidelity')
# Create a high-fidelity model
l1_prior_pf_1hf = []
for idx, n_evals in enumerate(n_evals_mc):
indices = np.random.choice(range(prior_pf_samples_hf.shape[0]),
size=n_evals,
replace=False)
# Monte Carlo model
prior_pf_samples = prior_pf_samples_hf[indices]
p_prior_pf = Distribution(prior_pf_samples,
rv_name='$Q$',
label='Prior-PF')
# l1 error between prior push-forward and reference push-forward
l1_prior_pf_1hf.append(
ref_p_prior_pf.calculate_l1_error(p_prior_pf))
# -------------- 1 HF, 1 LF
l1_prior_pf_1hf_1lf = []
for idx, n_evals in enumerate(n_evals_mfmc_hf_1lf):
n_evals = [n_evals_mfmc_lf_1lf[idx], n_evals]
indices = np.random.choice(range(prior_samples.shape[0]),
size=n_evals[0],
replace=False)
# Create a low-fidelity model
lf_model = Model(
eval_fun=lambda x, samples=prior_pf_samples_lf: elliptic_pde.
find_xy_pair(x, prior_samples, samples),
rv_samples=prior_samples[indices],
