prg_genz_mod Module Reference

To produce a matrix \(Z\) which is needed to orthogonalize \(H\). More...

Data Types
type	genzspinp
	Input for the genz driver. This type controlls all the variables that are needed by genz. More...

Functions/Subroutines
subroutine, public	prg_parse_zsp (input, filename)
	The parser for genz solver. More...
subroutine, public	prg_init_zspmat (igenz, zk1_bml, zk2_bml, zk3_bml, zk4_bml, zk5_bml, zk6_bml, norb, bml_type, bml_element_type)
	Initiates the matrices for the Xl integration of Z. More...
subroutine, public	prg_buildzdiag (smat_bml, zmat_bml, threshold, mdimin, bml_type, verbose)
	Usual subroutine involving diagonalization. \( Z=U\sqrt{s}U^{\dagger} \), where \( U \) = eigenvectors and \( s \) = eigenvalues. The purpose of this subroutine is to have an exact way of computing z for comparing with the sparse approach. More...
subroutine, public	prg_buildzsparse (smat_bml, zmat_bml, igenz, mdim, bml_type, zk1_bml, zk2_bml, zk3_bml, zk4_bml, zk5_bml, zk6_bml, nfirst, nrefi, nreff, thresholdi, thresholdf, integration, verbose)
	Inverse factorization using Niklasson's algorithm. More...
subroutine, public	prg_genz_sp_initialz0 (smat_bml, zmat_bml, norb, mdim, bml_type_f, threshold)
	Initial estimation of Z. More...
subroutine, public	prg_genz_sp_initial_zmat (smat_bml, zmat_bml, norb, mdim, bml_type_f, threshold)
	Initial estimation of Z. More...
subroutine	prg_genz_sp_int (zmat_bml, zk1_bml, zk2_bml, zk3_bml, zk4_bml, zk5_bml, zk6_bml, igenz, norb, bml_type, threshold)
	Inverse factorization using Niklasson's algorithm. More...
subroutine, public	prg_genz_sp_ref (smat_bml, zmat_bml, nref, norb, bml_type, threshold)
	Iterative refinement. More...

Variables
integer, parameter	dp = kind(1.0d0)

Detailed Description

To produce a matrix \(Z\) which is needed to orthogonalize \(H\).

\( H_{orth} = Z^{\dagger}HZ \) See Negre 2016 [2]

Function/Subroutine Documentation

prg_buildzdiag()

subroutine, public prg_genz_mod::prg_buildzdiag	(	type(bml_matrix_t), intent(inout)	smat_bml,
		type(bml_matrix_t)	zmat_bml,
		real(dp)	threshold,
		integer, intent(in)	mdimin,
		character(len=*), intent(in)	bml_type,
		integer, intent(in), optional	verbose
	)

Usual subroutine involving diagonalization. \( Z=U\sqrt{s}U^{\dagger} \), where \( U \) = eigenvectors and \( s \) = eigenvalues. The purpose of this subroutine is to have an exact way of computing z for comparing with the sparse approach.

Parameters

smat_bml	Overlap matrix in bml format.
zmat_bml	Congruence transform in bml format.
threshold	Threshold value to use, in this case, only in the backtransformation to ellpack format.
mdim	Maximun nonzero to use, in this case, only in the backtransformation to ellpack format.
bml_type	the bml type we are passing.

Definition at line 174 of file prg_genz_mod.F90.

prg_buildzsparse()

subroutine, public prg_genz_mod::prg_buildzsparse	(	type(bml_matrix_t)	smat_bml,
		type(bml_matrix_t)	zmat_bml,
		integer	igenz,
		integer	mdim,
		character(20)	bml_type,
		type(bml_matrix_t)	zk1_bml,
		type(bml_matrix_t)	zk2_bml,
		type(bml_matrix_t)	zk3_bml,
		type(bml_matrix_t)	zk4_bml,
		type(bml_matrix_t)	zk5_bml,
		type(bml_matrix_t)	zk6_bml,
		integer	nfirst,
		integer	nrefi,
		integer	nreff,
		real(dp)	thresholdi,
		real(dp)	thresholdf,
		logical	integration,
		integer	verbose
	)

Inverse factorization using Niklasson's algorithm.

Parameters

smat_bml	overlap matrix
zmat_bml	congruence transform to be updated or computed. (bml format)
igenz	counter to keep track of the calls to this subroutine.
mdim	dimension of the maxnonzero per row.
zk1_bml-zk6_bml	history of the past congruence transforms.
nfirst	first pre septs with nrefi and thresholdi.
nrefi	number of refinement iterations for the firsts "nfirst" steps.
nreff	number of refinement iterations for the rest of the steps.
integration	if we want to apply xl integration scheme for z (default is always .true.)
verbose	to print extra information.

Definition at line 272 of file prg_genz_mod.F90.

prg_genz_sp_initial_zmat()

subroutine, public prg_genz_mod::prg_genz_sp_initial_zmat	(	type(bml_matrix_t), intent(in)	smat_bml,
		type(bml_matrix_t)	zmat_bml,
		integer	norb,
		integer	mdim,
		character(20)	bml_type_f,
		real(dp), intent(in)	threshold
	)

Initial estimation of Z.

Note

Most of the operations are done in pure dense format. The purpose of this subroutine is to have an exact way of computing z for comparing with the sparse approach.

Parameters

smat_bml	Overlap matrix in bml format.
zmat_bml	Congruence transform in bml format.
norb	Congruence transform in bml format.
mdim	Congruence transform in bml format.
bml_type_f	The bml final type of zmat_bml.
threshold	Threshold value to use, in this case, only in the backtransformation to ellpack format.

Definition at line 478 of file prg_genz_mod.F90.

prg_genz_sp_initialz0()

subroutine, public prg_genz_mod::prg_genz_sp_initialz0	(	type(bml_matrix_t), intent(in)	smat_bml,
		type(bml_matrix_t)	zmat_bml,
		integer	norb,
		integer	mdim,
		character(20)	bml_type_f,
		real(dp)	threshold
	)

Initial estimation of Z.

Note

Most of the operations are done in pure dense format. The purpose of this subroutine is to have an exact way of computing z for comparing with the sparse approach.

Parameters

smat_bml	Overlap matrix in bml format.
zmat_bml	Congruence transform in bml format.
norb	Congruence transform in bml format.
mdim	Congruence transform in bml format.
bml_type_f	The bml final type of zmat_bml.
threshold	Threshold value to use, in this case, only in the backtransformation to ellpack format.

Definition at line 338 of file prg_genz_mod.F90.

prg_genz_sp_int()

subroutine prg_genz_mod::prg_genz_sp_int ( type(bml_matrix_t)  zmat_bml, type(bml_matrix_t)  zk1_bml, type(bml_matrix_t)  zk2_bml, type(bml_matrix_t)  zk3_bml, type(bml_matrix_t)  zk4_bml, type(bml_matrix_t)  zk5_bml, type(bml_matrix_t)  zk6_bml, integer  igenz, integer  norb, character(20)  bml_type, real(dp)  threshold  )

private

Inverse factorization using Niklasson's algorithm.

Parameters

smat_bml	overlap matrix
zmat_bml	congruence transform to be updated or computed. (bml format)
mdim	dimension of the maxnonzero per row.
zk1_bml-zk6_bml	history of the past congruence transforms.
igenz	counter to keep track of the calls to this subroutine.
norb	Congruence transform in bml format.
bml_type_f	The bml final type of zmat_bml.
threshold	Threshold value to use.

Definition at line 633 of file prg_genz_mod.F90.

prg_genz_sp_ref()

subroutine, public prg_genz_mod::prg_genz_sp_ref	(	type(bml_matrix_t), intent(in)	smat_bml,
		type(bml_matrix_t), intent(inout)	zmat_bml,
		integer, intent(in)	nref,
		integer, intent(inout)	norb,
		character(20), intent(in)	bml_type,
		real(dp), intent(in)	threshold
	)

Iterative refinement.

Parameters

smat_bml	overlap matrix
zmat_bml	congruence transform to be updated or computed. (bml format)
nref	Number of refinement iterations.
bml_type_f	The bml final type of zmat_bml.
threshold	Threshold value to use.
verbose	to print extra information.

Definition at line 719 of file prg_genz_mod.F90.

prg_init_zspmat()

subroutine, public prg_genz_mod::prg_init_zspmat	(	integer	igenz,
		type(bml_matrix_t)	zk1_bml,
		type(bml_matrix_t)	zk2_bml,
		type(bml_matrix_t)	zk3_bml,
		type(bml_matrix_t)	zk4_bml,
		type(bml_matrix_t)	zk5_bml,
		type(bml_matrix_t)	zk6_bml,
		integer	norb,
		character(20)	bml_type,
		character(20), optional	bml_element_type
	)

Initiates the matrices for the Xl integration of Z.

Parameters

self	input zsp variables
zk1_bml-zk6_bml	history record of the previous Z matrices.
norb	number of orbitals.
bml_type	the bml format we are passing.

Definition at line 133 of file prg_genz_mod.F90.

prg_parse_zsp()

subroutine, public prg_genz_mod::prg_parse_zsp	(	type(genzspinp), intent(inout)	input,
		character(len=*)	filename
	)

The parser for genz solver.

Definition at line 67 of file prg_genz_mod.F90.

Variable Documentation

dp

integer, parameter prg_genz_mod::dp = kind(1.0d0)

private

Definition at line 18 of file prg_genz_mod.F90.