An ShaSTPrior object encapsulates the prior information for the short-term discrepancies of the shared discrepancy of the ensemble model.
AR_paramThe parameters giving the beta parameters for the main parameter of the AR(1) process.
parametrisation_formThe parametrisation by which the covariance matrix of the noise of the AR process is decomposed.
var_paramsThe parameters characterising the variance of the AR process on the shared short-term discrepancy.
cor_paramsThe parameters characterising the correlations of the AR process on the shared short-term discrepancy.
Shared short-term discrepancies \(\mathbf{\eta}^{(t)}\) are modelled as an AR(1) process so that $$\mathbf{\eta}^{(t+1)} \sim N(R_\eta \mathbf{\eta}, \Lambda_\eta).$$ Accepted parametrisation forms for this discrepancy are lkj, beta, or inv_wishart. See details of the EnsemblePrior() constructor for more details.