Input File Description

Program: kcw.x / KCW / Quantum ESPRESSO (version: 7.2)




prefix | outdir | calculation | kcw_iverbosity | kcw_at_ks | read_unitary_matrix | spread_thr | homo_only | l_vcut | assume_isolated | spin_component | mp1 | mp2 | mp3 | lrpa


seedname | num_wann_occ | num_wann_emp | have_empty | has_disentangle | check_ks


niter | nmix | tr2 | i_orb | eps_inf | check_spread


do_bands | use_ws_distance | write_hr | on_site_only



Input data format: { } = optional, [ ] = it depends, # = comment

Structure of the input data:





K_POINTS { tpiba | automatic | crystal | gamma | tpiba_b | crystal_b | tpiba_c | crystal_c }
if (gamma)
   nothing to read
if (automatic)
   nk1, nk2, nk3, k1, k2, k3
if (not automatic)
   xk_x, xk_y, xk_z,  wk
if (tpipa_b or crystal_b in a 'bands' calculation) see Doc/brillouin_zones.pdf

Namelist: &CONTROL

Default: 'pwscf'
Prepended to input/output filenames; must be the same
used in the previous PW calculations.
Default: current directory ('./')
Directory containing input, output, and scratch files;
must be the same as specified in the calculation of
previous PW calculation.
calculation CHARACTER
Default: ' '
Specify the KCW calculation to be done
Possible choices:
'wann2kcw' :
Pre-processing to prepare KCW calculation.
Read previous PWSCF and possibly W90 outputs and prepare the KCW
'screen' :
Perform the calculation of KCW screening coefficient using a
LR approach as described here
and arXiv:2202.08155
'ham' :
Perform the calculation interpolation and diagonalization of the KI hamiltonian
'cc' :
Computes the (estimated) q+G=0 contribution to the bare and screened KC corrections.
A report on this quantities is printed on output and can be used to correct a
posteriori a "screen" calculation performed without any corrective scheme (l_vcut=.false.)
avoiding the need of re-doing a "screen" calculation.
kcw_iverbosity INTEGER
Default: 1
= 0 : minimal output
= 1 : as above + performs additional checks.
> 1 : as above + additional infos on all the steps.
kcw_at_ks LOGICAL
Default: .TRUE.
If true the KS canonical orbitals are used instead of Wannier
functions. It makes sense for isolated system only.
read_unitary_matrix LOGICAL
Default: .FALSE.
If true read the Unitary matrix written by Wannier90.
Implicitely means a previous wannier90 calculation was
performed and a KCW calculation will be performed starting
from MLWF. Requires 'write_hr = .true.' in wannier90.
spread_thr REAL
Default: 0.0001 Ry
HARD-CODED FOR NOW. Two or more Wannier functions are considered
identical if their spread (self-hartree) differ by less than spread_thr.
Requires check_spread = .true.
homo_only LOGICAL
Default: FALSE
If kcw_at_ks = .TRUE. only the screening paramenter for the HOMO is
calculated. Mainly for a perturbative calculation of the first Ionization
Potential in isolated systems.
l_vcut LOGICAL
Default: FALSE
If .TRUE. the Gygi-Baldereschi scheme is used to deal with
the q->0 divergence of the Coulomb integral (bare and screened).
Improves the convergence wrt k/q-point sampling.
Requires to correctly set eps_inf for the calculation of
the screened interaction.

Use it only for periodic system.
For isoleted system use assume_isolated, instead.
assume_isolated CHARACTER
Default: 'none'
Used to perform calculation assuming the system to be
isolated (a molecule or a cluster in a 3D supercell).

Currently available choices:
'none' :
 (default): regular periodic calculation w/o any correction.
'martyna-tuckerman', 'm-t', 'mt' :
Martyna-Tuckerman correction
to both total energy and scf potential. Adapted from:
G.J. Martyna, and M.E. Tuckerman,
"A reciprocal space based method for treating long
range interactions in ab-initio and force-field-based
calculation in clusters", J. Chem. Phys. 110, 2810 (1999),
spin_component INTEGER
Default: 1
Which spin channel to calculate (only collinear calculation).
1 = spin up channel
2 = spin down channel
It has to be consistent with the previous Wannier90
calculation (see 'spin' keyword in Wannier90 documentation)
mp1, mp2, mp3 INTEGER
Default: -1,-1,-1
Parameters of the Monkhorst-Pack grid (no offset).
Same meaning as for nk1, nk2, nk3 in the input of pw.x.
It has to coincide with the regular mesh used for the
wannier90 calculation.
Default: .FALSE.
If .true. the response function is computed neglecting xc
effects both in the kernel and in the response function (RPA).

Namelist: &WANNIER

seedname CHARACTER
Default: wann
The seedname of the previous Wannier90 calculation for occupied states.
NOTA BENE: the code implicitely assumed that the seedname for empty
state is the same as that for occupied state with "_emp" appended.
Keep this in mind when set up the wannier90 inputs.

For example:         is the wannier90 input file for the occupied states.     is the wannier90 input file for the empty states.
num_wann_occ INTEGER
Default: 0
The number of wannier function for the occupied manifold.
It has to coincide with the number of occupied KS orbitals.
The whole KS manifold has to be wannierised (no 'exclude_band'
option for occupied state, at the moment).
num_wann_emp INTEGER
Default: 0
The number of wannier function for the empty manifold.
It has to coincide with the number of empty wannier function
from the previous wannier90 calculation
have_empty LOGICAL
Default: FALSE
If true empty state are computed. Require a previous wannier90
calculation for the empty manifold. The code search for the
unitary matrices in the wannier90 file seedname_emp_u.mat
has_disentangle LOGICAL
Default: FALSE
Specify if a disentangle unitary matrix needs to be read. Requires
a consisten calcuation from the previous wannier90 run.
check_ks LOGICAL
Default: FALSE
Specify if a diagonalization of the KS matrix build using the wannier
function in input has to be performed. This is mainly for debugging purpose.

Namelist: &SCREEN

Default: maxter=100
Maximum number of iterations in a scf step. If you want
more than 100, edit variable "maxter" in PH/phcom.f90
Default: 4
 Number of iterations used in potential mixing.
tr2 REAL
Default: 1e-14
 Threshold for self-consistency.
Default: -1
Perform the screening calculation for a particular orbital.
If i_orb = -1 (default) all the orbitals are computed.
Assumes values between 1 and the total number of wannier
eps_inf REAL
Default: 1.d0
The macroscopic dielectric constant. Needed for the Gygi-Baldereschi
scheme if l_vcut = .TRUE.
Typically from exp or from a ph.x calculation.

NOTA BENE: This would be equivalent to a Makov-Payne correction. It works well
for cubic systems. Less well for anisotropic systems.

ANISOTROPIC SYSTEMS: In this case a generalization of the GB scheme is implemented
based on Nano Lett.,9, 975 (2009). It requires the full dielectric tensor to be provided.
The code searches (in the working dir) for a file named "eps.dat" containing the macrospocic
dielectric tensor. If it does not find it, the value eps_inf provided in input will be
used (isotropic approximation). If not even eps_inf is provided in input no correction
is applied to the screened KC correction.
check_spread LOGICAL
Default: .FALSE.
If .TRUE. the spread (self-hartree) of the Wannier functions is
checked and used to decide whether two or more Wannier functions
can be considered "identical" or not. Two Wannier functions are
considered identical if their spread (self-hartree) differ by less
than 1e-4 Ry (Hard coded for now, see spread_thr).

Namelist: &HAM

do_bands LOGICAL
Default: .FALSE.
If .true. the interpolated band structure is computed along
a path secified with the K_POINTS card ( see PW documentation )
use_ws_distance LOGICAL
Default: .TRUE.
If .true. the position of the Wannier function inside the cell is used
to set the proper distance and to have a smoother interpolation. Requires to be printed by the previous Wannier90 run. If the
file is not found it is automatically switched to .FALSE. and only the
distance between the cells is used (see also Wannier90 documentation)
write_hr LOGICAL
Default: .TRUE.
If .true. the KCW hamiltonain in the Wannier basis and in real spase H(R)_m_n
is printed to file. Usefull for furhter post-processing.
on_site_only LOGICAL
Default: .FALSE.
If .true. only the on-site and diagonal elements of the KCW hamiltonain
are computed (R=0 and n=m).



This file has been created by helpdoc utility on Thu Mar 30 12:13:52 CEST 2023.