def sample_observation_diagonal_covariance(model, state, observ, hyperparams):
"""
MCMC iteration (Gibbs sampling) for diagonal observation covariance
matrix with inverse-gamma prior
"""
# Calculate sufficient statistics using current state trajectory
suffStats = smp.evaluate_observation_sufficient_statistics(state, observ)
# Update hyperparameters
a, b = smp.hyperparam_update_basic_ig_observation_variance(
suffStats, model.parameters['H'], hyperparams['a0'], hyperparams['b0'])
# Sample new parameter
r = stats.invgamma.rvs(a, scale=b)
model.parameters['R'] = r * np.identity(model.do)
return model
def sample_observation_diagonal_covariance(model, state, observ, hyperparams):
"""
MCMC iteration (Gibbs sampling) for diagonal observation covariance
matrix with inverse-gamma prior
"""
# Calculate sufficient statistics using current state trajectory
suffStats = smp.evaluate_observation_sufficient_statistics(state, observ)
# Update hyperparameters
a,b = smp.hyperparam_update_basic_ig_observation_variance(
suffStats,
model.parameters['H'],
hyperparams['a0'],
hyperparams['b0'])
# Sample new parameter
r = stats.invgamma.rvs(a, scale=b)
model.parameters['R'] = r*np.identity(model.do)
return model
def sample_observation_diagonal_covariance(self):
"""
MCMC iteration (Gibbs sampling) for diagonal observation covariance
matrix with inverse-gamma prior
"""
# Calculate sufficient statistics using current state trajectory
suffStats = smp.evaluate_observation_sufficient_statistics(
self.state, self.observ)
# Update hyperparameters
a, b = smp.hyperparam_update_basic_ig_observation_variance(
suffStats, self.model.parameters['H'], self.hyperparams['a0'],
self.hyperparams['b0'])
# Sample new parameter
r = stats.invgamma.rvs(a, scale=b)
self.model.parameters['R'] = r * np.identity(self.model.do)
# self.flt and self.lhood are no longer up-to-date
self.filter_current = False
def sample_observation_diagonal_covariance(self):
"""
MCMC iteration (Gibbs sampling) for diagonal observation covariance
matrix with inverse-gamma prior
"""
# Calculate sufficient statistics using current state trajectory
suffStats = smp.evaluate_observation_sufficient_statistics(self.state,
self.observ)
# Update hyperparameters
a,b = smp.hyperparam_update_basic_ig_observation_variance(
suffStats,
self.model.parameters['H'],
self.hyperparams['a0'],
self.hyperparams['b0'])
# Sample new parameter
r = stats.invgamma.rvs(a, scale=b)
self.model.parameters['R'] = r*np.identity(self.model.do)
# self.flt and self.lhood are no longer up-to-date
self.filter_current = False