Expand description
§solow-optimize
Unconstrained optimization and numerical differentiation for maximum-likelihood
estimation: newton_stationary (analytic score/Hessian), minimize_bfgs
(gradient only), and finite-difference approx_fprime / approx_hess.
Structs§
- Optimize
Result - Outcome of an optimization run.
Functions§
- approx_
fprime - Central-difference gradient of
fatx. - approx_
hess - Central-difference Hessian of
fatx. - minimize_
bfgs - Minimize
fwith gradientgradusing BFGS and Armijo backtracking. - newton_
stationary - Newton iteration that drives the gradient of an objective to zero.