#include <bfgs_update.hpp>
|
typedef Eigen::Matrix< Scalar, DimAtCompile, 1 > | VectorT |
|
typedef Eigen::Matrix< Scalar, DimAtCompile, DimAtCompile > | HessianT |
|
template<typename Scalar = double, int DimAtCompile = Eigen::Dynamic>
class stan::optimization::BFGSUpdate_HInv< Scalar, DimAtCompile >
Definition at line 10 of file bfgs_update.hpp.
template<typename Scalar = double, int DimAtCompile = Eigen::Dynamic>
template<typename Scalar = double, int DimAtCompile = Eigen::Dynamic>
template<typename Scalar = double, int DimAtCompile = Eigen::Dynamic>
Compute the search direction based on the current (inverse) Hessian approximation and given gradient.
- Parameters
-
[out] | pk | The negative product of the inverse Hessian and gradient direction gk. |
[in] | gk | Gradient direction. |
Definition at line 55 of file bfgs_update.hpp.
template<typename Scalar = double, int DimAtCompile = Eigen::Dynamic>
Update the inverse Hessian approximation.
- Parameters
-
yk | Difference between the current and previous gradient vector. |
sk | Difference between the current and previous state vector. |
reset | Whether to reset the approximation, forgetting about previous values. |
- Returns
- In the case of a reset, returns the optimal scaling of the initial Hessian approximation which is useful for predicting step-sizes.
Definition at line 26 of file bfgs_update.hpp.
The documentation for this class was generated from the following file: