def test_no_generative_model(self):
D = 6
J = 6
bf_meta = build_meta_dict({'n_params': D, 'n_models': J}, DEFAULT_SETTING_INVARIANT_BAYES_FLOW)
amortizer = ex.amortizers.InvariantBayesFlow(bf_meta)
trainer = MetaTrainer(amortizer,
loss=kl_latent_space,
learning_rate=.0003
)
generative_model = GenerativeModel(
model_prior,
[TPrior(D // 2, 1.0, 5.0)] * J,
[MultivariateTSimulator(df) for df in np.arange(1, J + 1, 1)]
)
model_indices, params, sim_data = generative_model(64, 128)
_losses = trainer.train_offline(2, 16, model_indices, params, sim_data)
with self.assertRaises(OperationNotSupportedError):
_losses = trainer.train_online(epochs=2, iterations_per_epoch=20, batch_size=32, n_obs=110)
with self.assertRaises(OperationNotSupportedError):
_losses = trainer.train_rounds(epochs=1, rounds=5, sim_per_round=100, batch_size=32, n_obs=110)
with self.assertRaises(OperationNotSupportedError):
_losses = trainer.simulate_and_train_offline(n_sim=100, epochs=2, batch_size=16, n_obs=110)
def init_same_param_shapes(cls):
M = 10
D = 8
prior = TPrior(D // 2, mu_scale=1.0, scale_scale=5.0)
priors = [prior] * M
simulators = [
MultivariateTSimulator(df) for df in np.arange(1, 101, M)
]
generative_model = MetaGenerativeModel(model_prior, priors, simulators)
return generative_model
def test_meta_generative_model(self):
M = 10
D = 100
prior = TPrior(D // 2, mu_scale=1.0, scale_scale=5.0)
priors = [prior] * M
simulators = [
MultivariateTSimulator(df) for df in np.arange(1, 101, M)
]
generative_model = GenerativeModel(model_prior, priors, simulators)
_model_indices, _params, _sim_data = generative_model(n_sim=N_SIM,
n_obs=N_OBS)
def test_same_data_transform(self):
def data_transform(x):
noise = 0.001 * np.random.random(x.shape)
return x + noise
M = 10
D = 8
prior = TPrior(D // 2, mu_scale=1.0, scale_scale=5.0)
priors = [prior] * M
simulators = [
MultivariateTSimulator(df) for df in np.arange(1, 101, M)
]
_generative_model = MetaGenerativeModel(model_prior=model_prior,
priors=priors,
simulators=simulators,
data_transforms=data_transform)
def test_same_param_transform(self):
def param_transform(x):
return np.exp(x)
M = 10
D = 8
prior = TPrior(D // 2, mu_scale=1.0, scale_scale=5.0)
priors = [prior] * M
simulators = [
MultivariateTSimulator(df) for df in np.arange(1, 101, M)
]
_generative_model = MetaGenerativeModel(
model_prior=model_prior,
priors=priors,
simulators=simulators,
param_transforms=param_transform)
def setUpClass(cls):
D = 10
J = 10
bf_meta = build_meta_dict({'n_params': D, 'n_models': J}, DEFAULT_SETTING_INVARIANT_BAYES_FLOW)
amortizer = ex.amortizers.InvariantBayesFlow(bf_meta)
generative_model = GenerativeModel(
model_prior,
[TPrior(D // 2, 1.0, 5.0)] * J,
[MultivariateTSimulator(df) for df in np.arange(1, J + 1, 1)]
)
trainer = MetaTrainer(amortizer,
generative_model,
loss=kl_latent_space,
learning_rate=.0003
)
cls.trainer = trainer
def test_individual_param_and_data_transform(self):
param_transforms = [
lambda x: np.exp(x), None, lambda x: np.round(x, 3)
]
data_transforms = [
lambda x: x + np.random.random(x.shape), lambda x: np.exp(x), None
]
M = 3
D = 4
prior = TPrior(D // 2, mu_scale=1.0, scale_scale=5.0)
priors = [prior] * M
simulators = [
MultivariateTSimulator(df)
for df in np.round(np.linspace(1, 101, M))
]
_generative_model = MetaGenerativeModel(
model_prior=model_prior,
priors=priors,
simulators=simulators,
param_transforms=param_transforms,
data_transforms=data_transforms)