stan::math::internal::NewtonState Struct Reference

Detailed Description

Holds the state for the Newton-Raphson optimization loop.

This struct centralizes all state needed during the Newton iteration, including the Wolfe line search state, workspace matrices, and convergence flags. It is shared across different solver policies to avoid re-allocation and to maintain progress (e.g., line search history) across solver fallbacks.

Template Parameters

ObjFun	Type of the objective function callable
ThetaGradFun	Type of the theta gradient function callable

Definition at line 347 of file laplace_marginal_density_estimator.hpp.

#include <laplace_marginal_density_estimator.hpp>

Public Member Functions
template<typename ObjFun , typename ThetaGradFun , typename ThetaInitializer >
	NewtonState (int theta_size, ObjFun &&obj_fun, ThetaGradFun &&theta_grad_f, ThetaInitializer &&theta_init)
	Constructs Newton state with given dimensions and functors.
auto &	curr () &
	Access the current step state (mutable).
const auto &	curr () const &
	Access the current step state (const).
auto &&	curr () &&
auto &	prev () &
	Access the previous step state (mutable).
const auto &	prev () const &
	Access the previous step state (const).
auto &&	prev () &&
template<typename Options >
void	update_next_step (const Options &options)

Public Attributes
WolfeInfo	wolfe_info
	Wolfe line search state including current/previous steps.
WolfeStatus	wolfe_status
	Status of the most recent Wolfe line search.
Eigen::VectorXd	b
	Workspace vector: b = W * theta + grad(log_lik)
Eigen::MatrixXd	B
	Workspace matrix: B = I + W_r * Sigma * W_r (or similar)
Eigen::VectorXd	prev_g
	Previous gradient for Barzilai-Borwein step calculation.
bool	final_loop = false
	On the final loop if we found a better wolfe step, but we are going to exit, we want to make sure all of our return values are with the most recent wolfe step that was accepted.

---

The documentation for this struct was generated from the following file:

• stan/math/mix/functor/laplace_marginal_density_estimator.hpp