void MVSSMS::draw() {
if (!latent_data_initialized_) {
model_->impute_state(rng());
latent_data_initialized_ = true;
impute_nonstate_latent_data();
}
model_->observation_model()->sample_posterior();
for (int s = 0; s < model_->number_of_state_models(); ++s) {
model_->state_model(s)->sample_posterior();
}
impose_identifiability_constraints();
// The complete data sufficient statistics for the observation model and the
// state models are updated when calling impute_state. The non-state latent
// data should be imputed immediately before that, so the complete data
// sufficient statistics reflect all the latent data correctly.
impute_nonstate_latent_data();
model_->impute_state(rng());
// End with a call to impute_state() so that the internal state of
// the Kalman filter matches up with the parameter draws.
}
void SSPS::draw(){
if (!latent_data_initialized_) {
m_->impute_state();
latent_data_initialized_ = true;
}
impute_nonstate_latent_data();
m_->observation_model()->sample_posterior();
for(int s = 0; s < m_->nstate(); ++s) {
m_->state_model(s)->sample_posterior();
}
m_->impute_state();
// End with a call to impute_state() so that the internal state of
// the Kalman filter matches up with the parameter draws.
}