ParamRegistry owns parameter order and maps the unrestricted work vector to
economic values. It also evaluates priors and builds structural covariance
matrices:
registry.to_econ_dict(theta_work)
registry.from_econ_dict(theta_econ)
registry.log_prior(theta_work)
registry.build_Q(theta_work, n_eps)
registry.build_H(theta_work, n_y)build_matrices converts SymPy residuals or equalities into the canonical linear
system consumed by gensys.
log_like evaluates the Gaussian state-space likelihood. log_posterior adds the
registry prior. Use include_jacobian=True only when the posterior density should
be expressed in the unrestricted work space.
The high-level facade is available from:
from src.dsge import DSGEsrc.dgse remains available for old notebooks.