LCAO_deepks_io Namespace Reference

Functions
template<typename TK >
void	print_dm (const int nks, const int nlocal, const int nrow, const std::vector< std::vector< TK > > &dm)
	print density matrices
void	load_npy_gedm (const int nat, const int des_per_atom, double **gedm, double &e_delta, const int rank)
void	save_npy_d (const int nat, const int des_per_atom, const int inlmax, const std::vector< int > &inl2l, const bool deepks_equiv, const std::vector< torch::Tensor > &descriptor, const std::string &dm_eig_file, const int rank)
	save descriptor
void	save_npy_e (const double &e, const std::string &e_file, const int rank, const double unit_scale=Ry2Hartree)
template<typename TK , typename TH >
void	save_npy_h (const std::vector< TH > &hamilt, const std::string &h_file, const int nlocal, const int nks, const int rank, const double unit_scale=Ry2Hartree)
void	save_matrix2npy (const std::string &file_name, const ModuleBase::matrix &matrix, const int rank, const double &scale=1.0, const char mode='N', const double unit_scale=Ry2Hartree)
template<typename T >
void	save_tensor2npy (const std::string &file_name, const torch::Tensor &tensor, const int rank)

Variables
constexpr double	Ry2Hartree = 0.5
	others print quantities in .npy format

Function Documentation

load_npy_gedm()

void LCAO_deepks_io::load_npy_gedm	(	const int	nat,
		const int	des_per_atom,
		double **	gedm,
		double &	e_delta,
		const int	rank
	)

Here is the call graph for this function:

Here is the caller graph for this function:

print_dm()

template<typename TK >

void LCAO_deepks_io::print_dm	(	const int	nks,
		const int	nlocal,
		const int	nrow,
		const std::vector< std::vector< TK > > &	dm
	)

print density matrices

Here is the caller graph for this function:

save_matrix2npy()

void LCAO_deepks_io::save_matrix2npy	(	const std::string &	file_name,
		const ModuleBase::matrix &	matrix,
		const int	rank,
		const double &	scale = 1.0,
		const char	mode = 'N',
		const double	unit_scale = Ry2Hartree
	)

Here is the call graph for this function:

Here is the caller graph for this function:

save_npy_d()

void LCAO_deepks_io::save_npy_d	(	const int	nat,
		const int	des_per_atom,
		const int	inlmax,
		const std::vector< int > &	inl2l,
		const bool	deepks_equiv,
		const std::vector< torch::Tensor > &	descriptor,
		const std::string &	dm_eig_file,
		const int	rank
	)

save descriptor

Here is the call graph for this function:

Here is the caller graph for this function:

save_npy_e()

void LCAO_deepks_io::save_npy_e	(	const double &	e,
		const std::string &	e_file,
		const int	rank,
		const double	unit_scale = Ry2Hartree
	)

Parameters

[in]

e

or , in Ry

Here is the call graph for this function:

Here is the caller graph for this function:

save_npy_h()

template<typename TK , typename TH >

void LCAO_deepks_io::save_npy_h	(	const std::vector< TH > &	hamilt,
		const std::string &	h_file,
		const int	nlocal,
		const int	nks,
		const int	rank,
		const double	unit_scale = Ry2Hartree
	)

Here is the call graph for this function:

save_tensor2npy()

template<typename T >

template void LCAO_deepks_io::save_tensor2npy< double >	(	const std::string &	file_name,
		const torch::Tensor &	tensor,
		const int	rank
	)

Here is the call graph for this function:

Variable Documentation

Ry2Hartree

constexpr double LCAO_deepks_io::Ry2Hartree = 0.5

constexpr

others print quantities in .npy format

This file contains subroutines that contains interface with libnpy since many arrays must be saved in numpy format It also contains subroutines for printing density matrices which is used in unit tests There are 2 subroutines for printing and loading .npy file:

1. print_dm : print density matrices
2. load_npy_gedm : load gedm from .npy file

1. save_npy_d : descriptor -> deepks_dm_eig.npy
2. save_npy_e : energy
3. save_npy_h : Hamiltonian
4. save_matrix2npy : ModuleBase::matrix -> .npy, for force, stress and orbital
5. save_tensor2npy : torch::Tensor -> .npy, for precalculation variables Ry2Hartree : convert Ry to Hartree, for energy, force, stress, orbital and Hamiltonian, which is consistent with deepks-kit. Used in save_npy_e, save_npy_h and save_matrix2npy by default