Several screening methods used in HFX calcualtions. More...

Functions/Subroutines
subroutine, public	update_pmax_mat (pmax_set, map_atom_to_kind_atom, set_offset, atomic_block_offset, pmax_atom, full_density_alpha, full_density_beta, natom, kind_of, basis_parameter, nkind, is_assoc_atomic_block)
	updates the maximum of the density matrix in compressed form for screening purposes More...

subroutine, public	calc_screening_functions (qs_env, basis_parameter, lib, potential_parameter, coeffs_set, coeffs_kind, coeffs_pgf, radii_pgf, max_set, max_pgf, n_threads, i_thread, p_work)
	calculates screening functions for schwarz screening More...

subroutine, public	calc_pair_dist_radii (qs_env, basis_parameter, radii_pgf, max_set, max_pgf, eps_schwarz, n_threads, i_thread)
	calculates radius functions for longrange screening More...

Detailed Description

Several screening methods used in HFX calcualtions.

History: 04.2008 created [Manuel Guidon]

Author: Manuel Guidon

Function/Subroutine Documentation

◆ update_pmax_mat()

subroutine, public hfx_screening_methods::update_pmax_mat	(	type(hfx_p_kind), dimension(:), pointer	pmax_set,
		integer, dimension(:), pointer	map_atom_to_kind_atom,
		integer, dimension(:, :, :, :), pointer	set_offset,
		integer, dimension(:, :), pointer	atomic_block_offset,
		real(dp), dimension(:, :), pointer	pmax_atom,
		real(dp), dimension(:, :), pointer	full_density_alpha,
		real(dp), dimension(:, :), pointer	full_density_beta,
		integer, intent(in)	natom,
		integer, dimension(*)	kind_of,
		type(hfx_basis_type), dimension(:), pointer	basis_parameter,
		integer	nkind,
		integer, dimension(:, :)	is_assoc_atomic_block
	)

updates the maximum of the density matrix in compressed form for screening purposes

Parameters

pmax_set	buffer to store matrix
map_atom_to_kind_atom	...
set_offset	...
atomic_block_offset	...
pmax_atom	...
full_density_alpha	...
full_density_beta	...
natom	...
kind_of	...
basis_parameter	...
nkind	...
is_assoc_atomic_block	...

History: 09.2007 created [Manuel Guidon]

Author: Manuel Guidon

Note

updates for each pair of shells the maximum absolute value of p

Definition at line 134 of file hfx_screening_methods.F.

Here is the caller graph for this function:

◆ calc_screening_functions()

subroutine, public hfx_screening_methods::calc_screening_functions	(	type(qs_environment_type), pointer	qs_env,
		type(hfx_basis_type), dimension(:), pointer	basis_parameter,
		type(cp_libint_t)	lib,
		type(hfx_potential_type)	potential_parameter,
		type(hfx_screen_coeff_type), dimension(:, :, :, :), pointer	coeffs_set,
		type(hfx_screen_coeff_type), dimension(:, :), pointer	coeffs_kind,
		type(hfx_screen_coeff_type), dimension(:, :, :, :, :, :), pointer	coeffs_pgf,
		type(hfx_screen_coeff_type), dimension(:, :, :, :, :, :), pointer	radii_pgf,
		integer, intent(in)	max_set,
		integer, intent(in)	max_pgf,
		integer, intent(in)	n_threads,
		integer, intent(in)	i_thread,
		real(dp), dimension(:), pointer	p_work
	)

calculates screening functions for schwarz screening

Parameters

qs_env	qs_env
basis_parameter	...
lib	structure to libint
potential_parameter	contains infos on potential
coeffs_set	set based coefficients
coeffs_kind	kind based coefficients
coeffs_pgf	pgf based coefficients
radii_pgf	coefficients for long-range screening
max_set	Maximum Number of basis set sets in the system
max_pgf	Maximum Number of basis set pgfs in the system
n_threads	...
i_thread	Thread ID of current task
p_work	...

History: 02.2009 created [Manuel Guidon]

Author: Manuel Guidon

Note

This routine calculates (ab|ab) for different distances Rab = |a-b| and uses the powell optimiztion routines in order to fit the results in the following form:

       (ab|ab) = (ab|ab)(Rab) = c2*Rab^2 + c0

The missing linear term assures that the functions is monotonically decaying such that c2 can be used as upper bound when applying the Minimum Image Convention in the periodic case. Furthermore it seems to be a good choice to fit the logarithm of the (ab|ab) The fitting takes place at several levels: kind, set and pgf within the corresponding ranges of the prodiuct charge distributions Doing so, we only need arrays of size nkinds^2*2 instead of big screening matrices

Definition at line 300 of file hfx_screening_methods.F.

Here is the call graph for this function:

Here is the caller graph for this function:

◆ calc_pair_dist_radii()

subroutine, public hfx_screening_methods::calc_pair_dist_radii	(	type(qs_environment_type), pointer	qs_env,
		type(hfx_basis_type), dimension(:), pointer	basis_parameter,
		type(hfx_screen_coeff_type), dimension(:, :, :, :, :, :), pointer	radii_pgf,
		integer, intent(in)	max_set,
		integer, intent(in)	max_pgf,
		real(dp)	eps_schwarz,
		integer, intent(in)	n_threads,
		integer, intent(in)	i_thread
	)

calculates radius functions for longrange screening

Parameters

qs_env	qs_env
basis_parameter	...
radii_pgf	pgf based coefficients
max_set	Maximum Number of basis set sets in the system
max_pgf	Maximum Number of basis set pgfs in the system
eps_schwarz	...
n_threads	...
i_thread	Thread ID of current task

History: 02.2009 created [Manuel Guidon]

Author: Manuel Guidon

Note

This routine calculates the pair-distribution radius of a product Gaussian and uses the powell optimiztion routines in order to fit the results in the following form:

       (ab| = (ab(Rab) = c2*Rab^2 + c0

Definition at line 657 of file hfx_screening_methods.F.

Here is the call graph for this function:

Here is the caller graph for this function:

Functions/Subroutines

Detailed Description

Function/Subroutine Documentation

◆ update_pmax_mat()

◆ calc_screening_functions()

◆ calc_pair_dist_radii()