◆ line_search()

template<typename AVec , typename APrev , typename ThetaVec , typename LLFun , typename LLArgs , typename Covar , typename Msgs >

void stan::math::internal::line_search	(	double &	objective_new,
		AVec &	a,
		ThetaVec &	theta,
		APrev &	a_prev,
		LLFun &&	ll_fun,
		LLArgs &&	ll_args,
		Covar &&	covariance,
		const int	max_steps_line_search,
		const double	objective_old,
		double	tolerance,
		Msgs *	msgs
	)

inline

Performs a simple line search.

Template Parameters

AVec	Type of the parameter update vector (`a`), e.g. Eigen::VectorXd.
APrev	Type of the previous parameter vector (`a_prev`), same shape as AVec.
ThetaVec	Type of the transformed vector (`theta`), e.g. Σ·a.
LLFun	Functor type for computing the log‐likelihood.
LLArgs	Tuple or pack type forwarded to `ll_fun`.
Covar	Matrix type for the covariance Σ, e.g. Eigen::MatrixXd.
Msgs	Diagnostics container type for capturing warnings/errors.

Parameters

[in,out]	objective_new	On entry: objective at the full‐step `a` (must satisfy objective_new < objective_old). On exit: best objective found.
[in,out]	a	On entry: candidate parameter vector. On exit: updated to the step achieving the lowest objective.
[in,out]	theta	On entry: Σ·a for the initial candidate. On exit: Σ·a for the accepted best step.
[in,out]	a_prev	On entry: previous parameter vector, with objective `objective_old`. On exit: rolled forward to each newly accepted step.
[in]	ll_fun	Callable that computes the log‐likelihood given `(theta, ll_args, msgs)`.
[in]	ll_args	Arguments forwarded to `ll_fun` at each evaluation.
[in]	covariance	Covariance matrix Σ used to compute `theta = Σ·a`.
[in]	max_steps_line_search	Maximum number of iterations.
[in]	objective_old	Objective value at the initial `a_prev` (used as f₀ for the first pass).
[in]	tolerance	Minimum tolerance to accept a step
[in,out]	msgs	Pointer to a diagnostics container; may be used by `ll_fun` to record warnings.