PSCPrimal is the corresponding positive semidefinite object for PSCOracle like PrimalMatrix for a MatrixFunctionOracle. More...
#include <PSCPrimal.hxx>
Public Member Functions | |
| virtual PrimalData * | clone_primal_data () const =0 |
| returns a newly generated identical Object | |
| virtual int | assign_Gram_matrix (const CH_Matrix_Classes::Matrix &P)=0 |
| intialize the primal corresponding to the (positive semidefinite) Gram matrix P*P^T | |
| virtual int | aggregate_primal_data (const PrimalData &pd, double factor=1.)=0 |
| add factor*pd to this | |
| virtual int | aggregate_Gram_matrix (const CH_Matrix_Classes::Matrix &P, double factor=1.)=0 |
| add factor*P*P^T to this | |
| virtual int | scale_primal_data (double factor)=0 |
| multiply/scale *this with a nonnegative factor | |
| virtual int | primal_ip (CH_Matrix_Classes::Real &value, const SparseCoeffmatMatrix &A, CH_Matrix_Classes::Integer column) const =0 |
| if compatible evaluate value=ip(*this,A.column[i]) | |
Detailed Description
PSCPrimal is the corresponding positive semidefinite object for PSCOracle like PrimalMatrix for a MatrixFunctionOracle.
It allows the system to aggregate eigenvector information obtained in successive evaluations of the maximum eigenvalue function in PSCOralce to a positive semidefinite matrix so that in the Lagrangian relaxation view the primal optimal semidefinite matrix can be approximated, see abstract positive semidefinite cone oracle.
ConicBundle::PSCModel does rely on primal information. It also does not care about whether and in which form primal information is represented. This is all hidden in the Minorant information obtained via PSCOracle::generate_minorant(). If provided, the primal information can be retrieved from the model by the routines PSCModel::get_approximate_primal() and PSCModel::get_center_primal() or via the corresponding routines in the interface.
The documentation for this class was generated from the following file:
