madness::SubspaceK< T, NDIM > Class Template Reference

The SubspaceK class is a container class holding previous orbitals and residuals. More...

#include <solver.h>

Public Member Functions
	SubspaceK (World &world, const ElectronicStructureParams &params, const std::vector< KPoint > &kpoints)
void	update_subspace (vecfuncT &awfs_new, vecfuncT &bwfs_new, const vecfuncT &awfs_old, const vecfuncT &bwfs_old)

Detailed Description

template<typename T, int NDIM>
class madness::SubspaceK< T, NDIM >

The SubspaceK class is a container class holding previous orbitals and residuals.

The Solver class uses the Krylov Accelerated Inexact Newton Solver (KAIN) accelerate the convergence a given calculation. The KAIN solver needs to store a subspace of previous orbitals and residuals. In the case is this implementation, the orbitals are store according to which k-point to which they belong.

Constructor & Destructor Documentation

template<typename T , int NDIM>

madness::SubspaceK< T, NDIM >::SubspaceK ( World &  world, const ElectronicStructureParams &  params, const std::vector< KPoint > &  kpoints  )

inline

Member Function Documentation

template<typename T , int NDIM>

void madness::SubspaceK< T, NDIM >::update_subspace ( vecfuncT &  awfs_new, vecfuncT &  bwfs_new, const vecfuncT &  awfs_old, const vecfuncT &  bwfs_old  )

inline

References madness::abs(), c, madness::compress(), madness::gaxpy(), madness::KAIN(), m, madness::ElectronicStructureParams::maxsub, one(), madness::print(), madness::ElectronicStructureParams::spinpol, and madness::sub().

---

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

• solver.h