Bonmin 1.8.9
|
#include <BonCurvatureEstimator.hpp>
Public Member Functions | |
Constructor/Destructor | |
CurvatureEstimator (SmartPtr< Journalist > jnlst, SmartPtr< OptionsList > options, SmartPtr< TNLP > tnlp) | |
Constructor. | |
virtual | ~CurvatureEstimator () |
Destructor. | |
bool | ComputeNullSpaceCurvature (int n, const Number *x, bool new_x, const Number *x_l, const Number *x_u, const Number *g_l, const Number *g_u, bool new_bounds, const Number *z_L, const Number *z_U, int m, const Number *lam, bool new_mults, const Number *orig_d, Number *projected_d, Number &gradLagTd, Number &dTHLagd) |
Method for computing a direction projected_d related to the given direction orig_d and the two-sided product of projected_d with Hessian of Lagrangian. | |
![]() | |
ReferencedObject () | |
virtual | ~ReferencedObject () |
Index | ReferenceCount () const |
void | AddRef (const Referencer *referencer) const |
void | ReleaseRef (const Referencer *referencer) const |
Definition at line 29 of file BonCurvatureEstimator.hpp.
Bonmin::CurvatureEstimator::CurvatureEstimator | ( | SmartPtr< Journalist > | jnlst, |
SmartPtr< OptionsList > | options, | ||
SmartPtr< TNLP > | tnlp ) |
Constructor.
It is given the options list to extract options specifying linear solver options.
|
virtual |
Destructor.
bool Bonmin::CurvatureEstimator::ComputeNullSpaceCurvature | ( | int | n, |
const Number * | x, | ||
bool | new_x, | ||
const Number * | x_l, | ||
const Number * | x_u, | ||
const Number * | g_l, | ||
const Number * | g_u, | ||
bool | new_bounds, | ||
const Number * | z_L, | ||
const Number * | z_U, | ||
int | m, | ||
const Number * | lam, | ||
bool | new_mults, | ||
const Number * | orig_d, | ||
Number * | projected_d, | ||
Number & | gradLagTd, | ||
Number & | dTHLagd ) |
Method for computing a direction projected_d related to the given direction orig_d and the two-sided product of projected_d with Hessian of Lagrangian.
The arrays x, y_c, and y_d constain the primal and dual variables definiting the Lagrangian Hessian. The vectors active_d and active_x contain the indices of active inequality and bound constraints, respectively. A positive index number is interpreted to belong to an upper bound, and a negative number to a lower bound. The return status is false if the computation was not possible, and true otherwise.