This document lists and provides the description of the name (keywords) of "file handling" input variables to be used in the main input file of the abinit code.
Relevant only for non self consistent RF calculations (e.g. to get electron phonon matrix elements)
or for non linear RF calculations (to get mixed higher order derivatives you need several perturbed
densities and wave functions).
Indicate the files from which first-order densities must be obtained,
in multi-dataset mode (in single dataset mode, use ird1den).
NOTE : a negative value of a "get" variable indicates the number of datasets to go backwards;
it is not the number to be subtracted from the current dataset to find the proper dataset.
As an example :
ndtset 3 jdtset 1 2 4 getXXX -1refers to dataset 2 when dataset 4 is initialized.
Eventually used when ndtset>0
(in the multi-dataset mode), to indicate
starting wavefunctions, as an alternative to irdwfk,
irdwfq, ird1wf, irdddk. One should first read the
explanations given for these latter variables.
The
getwfk
,
getwfq
, get1wf and
getddk
variables are typically
used to chain the calculations in the multi-dataset mode,
since they describe from which dataset the OUTPUT
wavefunctions are to be taken, as INPUT wavefunctions
of the present dataset.
We now focus on the
getwfk
input variable (the only
one used in ground-state calculations), but
the rules for
getwfq
and get1wf are similar, with _WFK
replaced by _WFQ or _1WF.
If
getwfk
==0, no use of previously computed output
wavefunction file appended with _DSx_WFK is done.
If
getwfk
is positive, its value gives the index of the dataset
for which the output wavefunction file appended with _WFK
must be used.
If
getwfk
is -1, the output wf file with _WFK
of the previous dataset must be taken,
which is a frequently occurring case.
If
getwfk
is a negative number, it indicates the number
of datasets to go backward to find the needed wavefunction file.
In this case, if one refers to a non existent data set (prior
to the first), the wavefunctions are not initialised from
a disk file, so that it is as if
getwfk
=0 for that
initialisation.
Thanks to this rule, the use of
getwfk
-1 is rather
straightforward : except for the first wavefunctions, that
are not initialized by reading a disk file, the output
wavefunction of one dataset is input of the next one.
In the case of a ddk calculation in a multi dataset
run, in order to compute
correctly the localisation tensor, it is mandatory to
declare give getddk the value of the current dataset
(i.e. getddk3 3 ) - this is a bit strange and
should be changed in the future.
NOTE : a negative value of a "get" variable indicates the number of datasets to go backwards;
it is not the number to be subtracted from the current dataset to find the proper dataset.
As an example :
ndtset 3 jdtset 1 2 4 getXXX -1refers to dataset 2 when dataset 4 is initialized.
Eventually used when ndtset>0
(multi-dataset mode) and, in the case of a Bethe-Salpeter
calculation to indicate that the starting coupling block of the excitonic Hamiltonian will be taken from the output of a previous dataset.
It is used to chain the calculations, since it describes from which dataset the OUTPUT coupling
block is to be taken, as INPUT of the present dataset.
If getbscoup==0, no such use of previously computed coupling block file is done.
If getbscoup is positive, its value gives the index of the dataset to be used as input.
If getbscoup is -1, the output of the previous dataset must be taken, which is a frequently occuring case.
If getbscoup is a negative number, it indicates the number
of datasets to go backward to find the needed file.
In this case, if one refers to a non existent data set (prior to the first), the coupling block is not initialised from
a disk file, so that it is as if getbscoup=0 for that initialisation.
Eventually used when ndtset>0
(multi-dataset mode) and, in the case of a Bethe-Salpeter
calculation to indicate that the starting excitonic eigenstates are to be taken from the output of a previous dataset.
It is used to chain the calculations, since it describes from which dataset the OUTPUT eigenstates
are to be taken, as INPUT eigenstates of the present dataset.
If getbseig==0, no such use of previously computed output eigenstates file is done.
If getbseig is positive, its value gives the index of the dataset
from which the output states is to be used as input.
If getbseig is -1, the output eigenstates of the previous dataset
must be taken, which is a frequently occurring case.
If getbseig is a negative number, it indicates the number
of datasets to go backward to find the needed file.
In this case, if one refers to a non existent data set (prior to the first), the eigenstates are not initialised from
a disk file, so that it is as if getbseig=0 for that initialisation.
Eventually used when ndtset>0
(multi-dataset mode) and, in the case of a Bethe-Salpeter
calculation to indicate that the starting resonant block of the excitonic Hamiltonian will be taken from the output of a previous dataset.
It is used to chain the calculations, since it describes from which dataset the OUTPUT resonant
block is to be taken, as INPUT of the present dataset.
If getbsreso==0, no such use of previously computed resonant block file is done.
If getbsreso is positive, its value gives the index of the dataset to be used as input.
If getbsreso is -1, the output of the previous dataset must be taken, which is a frequently occurring case.
If getbsreso is a negative number, it indicates the number
of datasets to go backward to find the needed file.
In this case, if one refers to a non existent data set (prior to the first), the resonant block is not initialised from
a disk file, so that it is as if getbsreso=0 for that initialisation.
This variable should be used when performing electron-phonon
or temperature-dependence calculations.
The Born effective charge
as well as the dielectric tensor will be read from a previous
DFPT calculations of the electric field at q=Gamma.
The use of this variable will trigger the cancellation of a
residual dipole that leads to an unphysical divergence of the
GKK with vanishing q-points.
The use of this variable greatly improves the k-point convergence
speed as the density of the k-point grid required to obtain the
fulfillment of the charge neutrality sum rule is usually prohibitively
large.
If getddb==0, no such use of previously computed Born effective charge
and dielectric tensor is done.
If getddb is positive, its value gives the index of the dataset
from which the output density is to be used as input.
If getddb is -1, the output density of the previous dataset
must be taken, which is a frequently occurring case.
If getddb is a negative number, it indicates the number
of datasets to go backward to find the needed file.
NOTE : a negative value of a "get" variable indicates the number of datasets to go backwards;
it is not the number to be subtracted from the current dataset to find the proper dataset.
As an example :
ndtset 3 jdtset 1 2 4 getXXX -1refers to dataset 2 when dataset 4 is initialized.
Eventually used when ndtset>0
(in the multi-dataset mode), to indicate
starting wavefunctions, as an alternative to irdwfk,irdwfq,ird1wf,irdddk.
One should first read the
explanations given for these latter variables.
The
getwfk
,
getwfq
,
get1wf
and getddk variables are typically
used to chain the calculations in the multi-dataset mode,
since they describe from which dataset the OUTPUT
wavefunctions are to be taken, as INPUT wavefunctions
of the present dataset.
We now focus on the
getwfk
input variable (the only
one used in ground-state calculations), but
the rules for
getwfq
and
get1wf
are similar, with _WFK
replaced by _WFQ or _1WF.
If
getwfk
==0, no use of previously computed output
wavefunction file appended with _DSx_WFK is done.
If
getwfk
is positive, its value gives the index of the dataset
for which the output wavefunction file appended with _WFK
must be used.
If
getwfk
is -1, the output wf file with _WFK
of the previous dataset must be taken,
which is a frequently occurring case.
If
getwfk
is a negative number, it indicates the number
of datasets to go backward to find the needed wavefunction file.
In this case, if one refers to a non existent data set (prior
to the first), the wavefunctions are not initialised from
a disk file, so that it is as if
getwfk
=0 for that
initialisation.
Thanks to this rule, the use of
getwfk
-1 is rather
straightforward : except for the first wavefunctions, that
are not initialized by reading a disk file, the output
wavefunction of one dataset is input of the next one.
In the case of a ddk calculation in a multi dataset
run, in order to compute
correctly the localisation tensor, it is mandatory to
declare give getddk the value of the current dataset
(i.e. getddk3 3 ) - this is a bit strange and
should be changed in the future.
NOTE : a negative value of a "get" variable indicates the number of datasets to go backwards;
it is not the number to be subtracted from the current dataset to find the proper dataset.
As an example :
ndtset 3 jdtset 1 2 4 getXXX -1refers to dataset 2 when dataset 4 is initialized.
Eventually used when ndtset>0
(multi-dataset mode) and, in the case of a ground-state
calculation, if iscf<0 (non-SCF calculation),
to indicate that the starting density is to be taken
from the output of a previous dataset.
It is used to chain the calculations,
since it describes from which dataset the OUTPUT density
are to be taken, as INPUT density of the present dataset.
If getden==0, no such use of previously computed output
density file is done.
If getden is positive, its value gives the index of the dataset
from which the output density is to be used as input.
If getden is -1, the output density of the previous dataset
must be taken, which is a frequently occurring case.
If getden is a negative number, it indicates the number
of datasets to go backward to find the needed file.
In this case, if one refers to a non existent data set (prior
to the first), the density is not initialised from
a disk file, so that it is as if getden=0 for that
initialisation.
Thanks to this rule, the use of getden -1 is rather
straightforward : except for the first density, that
is not initialized by reading a disk file, the output
density of one dataset is input of the next one.
Be careful : the output density file of a run with
non-zero ionmov does not have the proper name (it has a "TIM"
indication) for use as an input of an iscf<0 calculation.
One should use the output density of a ionmov==0 run.
NOTE : a negative value of a "get" variable indicates the number of datasets to go backwards;
it is not the number to be subtracted from the current dataset to find the proper dataset.
As an example :
ndtset 3 jdtset 1 2 4 getXXX -1refers to dataset 2 when dataset 4 is initialized.
Eventually used when ndtset>0
(multi-dataset mode) and, in the case of a Bethe-Salpeter
calculation to indicate that the Haydock iterative technique will be restarted from the output of a previous dataset.
If gethaydock==0, no such use of previously computed coupling block file is done.
If gethaydock is positive, its value gives the index of the dataset to be used as input.
If gethaydock is -1, the output of the previous dataset must be taken, which is a frequently occuring case.
If gethaydock is a negative number, it indicates the number
of datasets to go backward to find the needed file.
In this case, if one refers to a non existent data set (prior to the first), the coupling block is not initialised from
a disk file, so that it is as if gethaydock=0 for that initialisation.
This variable is typically used to chain the calculations,
in the multi-dataset mode (ndtset>0),
since it describes from which dataset the array occ
is to be taken, as input of the present
dataset. The occupation numbers are EVOLVING variables,
for which such a chain of calculations is useful.
If getocc==0, no such use of previously computed output occupations is done.
If getocc is positive, its value gives the index of the dataset
from which the data are to be used as input data.
It must be the index of a dataset already computed in the
SAME run.
If getocc is -1, the output data of the previous dataset
must be taken, which is a frequently occurring case.
If getocc is a negative number, it indicates the number
of datasets to go backward to find the needed data.
In this case, if one refers to a non existent data set (prior to the first), the date is not initialised from a disk file, so that it is as if getocc==0 for that initialisation.
NOTE that a non-zero getocc MUST be used with occopt==2,
so that the number of bands has to be initialized for
each k point. Of course, these numbers of bands must be
identical to the numbers of bands of the dataset from which
occ will be copied. The same is true for the number of k points.
NOTE : a negative value of a "get" variable indicates the number of datasets to go backwards;
it is not the number to be subtracted from the current dataset to find the proper dataset.
As an example :
ndtset 3 jdtset 1 2 4 getXXX -1refers to dataset 2 when dataset 4 is initialized.
Used when ndtset>0
(multi-dataset mode) and optdriver=3, or 4
(screening or sigma step of a GW calculation),
to indicate that the eigenvalues and possibly the wavefunctions have to be
taken from a previous quasiparticle calculation (instead of the usual LDA starting
point). This is to achieve quasiparticle self-consistency.
See also irdqps
NOTE : a negative value of a "get" variable indicates the number of datasets to go backwards;
it is not the number to be subtracted from the current dataset to find the proper dataset.
As an example :
ndtset 3 jdtset 1 2 4 getXXX -1refers to dataset 2 when dataset 4 is initialized.
Used when ndtset>0
(multi-dataset mode) and optdriver=4
(sigma step of a GW calculation),
to indicate that the dielectric matrix (_SCR file) is to be taken
from the output of a previous dataset.
It is used to chain the calculations,
since it describes from which dataset the OUTPUT dielectric matrix
is to be taken, as INPUT of the present dataset.
If getscr==0, no such use of previously computed output
_SCR file is done.
If getscr is positive, its value gives the index of the dataset
from which the output _SCR file is to be used as input.
If getscr is -1, the output _SCR file of the previous dataset
must be taken, which is a frequently occurring case.
If getscr is a negative number, it indicates the number
of datasets to go backward to find the needed file.
In this case, if one refers to a non existent data set (prior
to the first), the _SCR file is not initialised from
a disk file, so that it is as if getscr=0 for that
initialisation.
NOTE : a negative value of a "get" variable indicates the number of datasets to go backwards;
it is not the number to be subtracted from the current dataset to find the proper dataset.
As an example :
ndtset 3 jdtset 1 2 4 getXXX -1refers to dataset 2 when dataset 4 is initialized.
Used when ndtset>0
(multi-dataset mode) and optdriver=4
(sigma step of a GW calculation),
to indicate that the irreducible polarizability (_SUSC file) is to be taken
from the output of a previous dataset.
It is used to chain the calculations, since it describes from which dataset the OUTPUT susceptibility
is to be taken, as INPUT of the present dataset.
Performing a GW calculations starting from the _SUSC file instead of the _SCR file presents
the advantage that starting from the irreducible polarizability, one can calculate the screened interaction
using different expressions without having to perform a screening calculation from scratch.
For example, it is possible to apply a cutoff to the Coulomb interaction in order
to facilitate the convergence of the GW correction with respect to the size of the supercell
(see vcutgeo and icutcoul)
If getsuscep==0, no such use of previously computed output _SUSC file is done.
If getsuscep is positive, its value gives the index of the dataset
from which the output _SUSC file is to be used as input.
If getsuscep is -1, the output _SUSC file of the previous dataset
must be taken, which is a frequently occurring case.
If getsuscep is a negative number, it indicates the number
of datasets to go backward to find the needed file.
In this case, if one refers to a non existent data set (prior
to the first), the _SUSC file is not initialised from
a disk file, so that it is as if getsuscep=0 for that
initialisation.
NOTE : a negative value of a "get" variable indicates the number of datasets to go backwards;
it is not the number to be subtracted from the current dataset to find the proper dataset.
As an example :
ndtset 3 jdtset 1 2 4 getXXX -1refers to dataset 2 when dataset 4 is initialized.
Eventually used when ndtset>0
(in the multi-dataset mode), to indicate
starting wavefunctions, as an alternative to irdwfk,irdwfq,ird1wf, or irdddk.
One should first read the
explanations given for these latter variables.
The getwfk,
getwfq
,
get1wf
and
getddk
variables are typically
used to chain the calculations in the multi-dataset mode,
since they describe from which dataset the OUTPUT
wavefunctions are to be taken, as INPUT wavefunctions
of the present dataset.
We now focus on the getwfk input variable (the only
one used in ground-state calculations), but
the rules for
getwfq
and
get1wf
are similar, with _WFK
replaced by _WFQ or _1WF.
If getwfk==0, no use of previously computed output
wavefunction file appended with _DSx_WFK is done.
If getwfk is positive, its value gives the index of the dataset
for which the output wavefunction file appended with _WFK
must be used.
If getwfk is -1, the output wf file with _WFK
of the previous dataset must be taken,
which is a frequently occurring case.
If getwfk is a negative number, it indicates the number
of datasets to go backward to find the needed wavefunction file.
In this case, if one refers to a non existent data set (prior
to the first), the wavefunctions are not initialised from
a disk file, so that it is as if getwfk=0 for that
initialisation.
Thanks to this rule, the use of getwfk -1 is rather
straightforward : except for the first wavefunctions, that
are not initialized by reading a disk file, the output
wavefunction of one dataset is input of the next one.
In the case of a ddk calculation in a multi dataset
run, in order to compute
correctly the localisation tensor, it is mandatory to
declare give getddk the value of the current dataset
(i.e. getddk3 3 ) - this is a bit strange and
should be changed in the future.
NOTE : a negative value of a "get" variable indicates the number of datasets to go backwards;
it is not the number to be subtracted from the current dataset to find the proper dataset.
As an example :
ndtset 3 jdtset 1 2 4 getXXX -1refers to dataset 2 when dataset 4 is initialized.
Eventually used when ndtset>0
(in the multi-dataset mode), to indicate
starting wavefunctions, as an alternative to irdwfk,irdwfq,ird1wf or irdddk.
One should first read the
explanations given for these latter variables.
The
getwfk
, getwfq,
get1wf
and
getddk
variables are typically
used to chain the calculations in the multi-dataset mode,
since they describe from which dataset the OUTPUT
wavefunctions are to be taken, as INPUT wavefunctions
of the present dataset.
We now focus on the
getwfk
input variable (the only
one used in ground-state calculations), but
the rules for getwfq and
get1wf
are similar, with _WFK
replaced by _WFQ or _1WF.
If
getwfk
==0, no use of previously computed output
wavefunction file appended with _DSx_WFK is done.
If
getwfk
is positive, its value gives the index of the dataset
for which the output wavefunction file appended with _WFK
must be used.
If
getwfk
is -1, the output wf file with _WFK
of the previous dataset must be taken,
which is a frequently occurring case.
If
getwfk
is a negative number, it indicates the number
of datasets to go backward to find the needed wavefunction file.
In this case, if one refers to a non existent data set (prior
to the first), the wavefunctions are not initialised from
a disk file, so that it is as if
getwfk
=0 for that
initialisation.
Thanks to this rule, the use of
getwfk
-1 is rather
straightforward : except for the first wavefunctions, that
are not initialized by reading a disk file, the output
wavefunction of one dataset is input of the next one.
In the case of a ddk calculation in a multi dataset
run, in order to compute
correctly the localisation tensor, it is mandatory to
declare give getddk the value of the current dataset
(i.e. getddk3 3 ) - this is a bit strange and
should be changed in the future.
NOTE : a negative value of a "get" variable indicates the number of datasets to go backwards;
it is not the number to be subtracted from the current dataset to find the proper dataset.
As an example :
ndtset 3 jdtset 1 2 4 getXXX -1refers to dataset 2 when dataset 4 is initialized.
If first order density is needed in single dataset mode (for example in nonlinear optical response), use ird1den=1 to read first-order densities from _DENx files produced in other calculations. In multi-dataset mode use get1den.
When iscf < 0, the reading of a DEN file is always enforced.
A non-zero value of
ird1den
is treated in the same way
as other "ird" variables,
see the
section 4
of abinit_help.
Indicates eventual starting
wavefunctions. As alternative, one can use the
input variables getwfk,
getwfq, get1wf
or getddk.
Ground-state calculation :
Start the Bethe-Salpeter calculation from the BSC file containing the coupling block produced in a previous run.
Start the Bethe-Salpeter calculation from the BS_EIG contining the exciton eigenvectors produced in a previous run.
Start the Bethe-Salpeter calculation from the BSR file containing the resonat block produced in a previous run.
This variable should be used when performing electron-phonon or temperature-dependence calculations. The Born effective charge as well as the dielectric tensor will be read from a previous DFPT calculations of the electric field at q=Gamma. The use of this variable will trigger the cancellation of a residual dipole that leads to an unphysical divergence of the GKK with vanishing q-points. The use of this variable greatly improves the k-point convergence speed as the density of the k-point grid required to obtain the fulfillment of the charge neutrality sum rule is usually prohibitively large.
A non-zero value of irdddb is treated in the same way
as other "ird" variables,
see the
section 4
of abinit_help.
Indicates eventual starting
wavefunctions. As alternative, one can use the
input variables getwfk,
getwfq, get1wf
or getddk.
Ground-state calculation :
Start the ground-state calculation from the density file of a previous run. When iscf < 0, the reading of a DEN file is always enforced.
A non-zero value of irdden is treated in the same way
as other "ird" variables,
see the
section 4
of abinit_help.
Used to re-start the Haydock iterative technique from the HAYDR_SAVE file produced in a previous run.
Relevant only when optdriver=3 or 4. Indicate the file from which the eigenvalues and possibly the wavefunctions must be obtained, in order to achieve a self-consistent quasiparticle calculations. See also getqps
Relevant only when optdriver=4.
Indicate the file from which the dielectric matrix must be obtained.
As alternative, one can use the input variable
getscr.
When optdriver=4, at least one of
irdscr or getscr
(alternatively, irdsuscep or getsuscep)
must be non-zero.
A non-zero value of irdscr is treated in the same way as other "ird" variables, see the section 4 of abinit_help.
Relevant only when optdriver=4.
Indicate the file from which the irreducible polarizability must be obtained.
As alternative, one can use the input variable
getsuscep.
When optdriver=4, at least one of
irdsuscep or getsuscep
(alternatively, irdscr or getscr)
must be non-zero.
A non-zero value of irdsuscep is treated in the same way as other "ird" variables, see the section 4 of abinit_help.
Indicates eventual starting
wavefunctions. As alternative, one can use the
input variables getwfk,
getwfq, get1wf
or getddk.
Ground-state calculation :
Indicates eventual starting
wavefunctions. As alternative, one can use the
input variables getwfk,
getwfq, get1wf
or getddk.
Ground-state calculation :
Governs the choice of the format for the file that contains the Kohn-Sham electronic structure information, for use in GW calculations, see the input variables optdriver and nbandkss.
Very important : for the time being, istwfk must be 1 for all the k-points.
If set >= 1, provide one-dimensional projection of potential and density, for each of the three axis. This corresponds to averaging the potential or the density on bi-dimensional slices of the FFT grid.
If set to 1 or a larger value , provide output of electron density
in real space rho(r), in units of electrons/Bohr^3.
If ionmov==0, the name of the density file will be
the root output name, followed by _DEN .
If ionmov==1 or 2, density files will be output
at each time step, with the name being made of
Provide output of Density of States if set to 1, 2 or 3. Can either use a smearing technique (prtdos=1), or the tetrahedron method (prtdos=2). If prtdos=3, provide output of Local Density of States inside a sphere centered on an atom, as well as the angular-momentum projected DOS, in the same sphere. The resolution of the linear grid of energies for which the DOS is computed can be tuned thanks to dosdeltae.
If prtdos=1, the smeared density of states is obtained
from the eigenvalues, properly weighted at each k point
using wtk, and smeared according to occopt
and tsmear. All levels that are present in the calculation
are taken into account (occupied and unoccupied).
Note that occopt must be between 3 and 7 .
Also note that the sampling of the Brillouin Zone that is needed to get a converged DOS
is usually much finer than the sampling needed to converge the total energy or the geometry of the system,
unless tsmear is very large (hence the DOS is not obtained properly)..
A separate convergence study is needed.
In order to compute the DOS of an insulator with prtdos=1, compute its density thanks to
a self-consistent calculation (with a non-metallic occopt
value, 0, 1 or 2), then use prtdos=1, together
with iscf=-3, and a metallic occopt,
between 3 and 7, providing the needed smearing.
If prtdos=1, the name of the DOS file is the root name for the output
files, followed by "_DOS" .
If prtdos=2, the DOS is computed using the tetrahedron method.
As in the case of prtdos=1, all levels that are present in the calculation
are taken into account (occupied and unoccupied). In this case, the
k-points must have been defined using the input variable ngkpt
or the input variable kptrlatt. There must be at least
two non-equivalent points in the Irreducible Brillouin Zone to use prtdos=2.
It is strongly advised to use a non-shifted k-point grid
(shiftk 0 0 0): such grids contain naturally
more extremal points (band minima and maxima at Gamma or at the zone-boundaries) than
shifted grids, and lead to more non-equivalent points than shifted grids, for the same grid spacing.
There is no need to take care of
the occopt or tsmear input variables,
and there is no subtlety to be taken into account for insulators. The computation
can be done in the self-consistent case as well as in the non-self-consistent case,
using iscf=-3. This allows to refine the DOS at fixed
starting density.
In that case, if ionmov==0, the name of the potential file will be
the root output name, followed by _DOS (like in the prtdos=1 case).
However, if ionmov==1 or 2, potential files will be output
at each time step, with the name being made of
If prtdos=3, the same tetrahedron method as for prtdos=2 is used, but
the DOS inside a sphere centered on some atom is delivered, as well as the angular-momentum
projected (l=0,1,2,3,4) DOS in the same sphere. The preparation of this
case, the parameters under which the computation is to be done, and the file
denomination is similar
to the prtdos=2 case. However, three additional input variables might be provided,
describing the atoms that are the center of the sphere (input variables
natsph and iatsph), as well as the radius of this
sphere (input variable ratsph).
In case of PAW, ratsph radius has to be greater or equal to
largest PAW radius of the atom types considered (which is read from the PAW atomic data file; see rc_sph or r_paw).
Additional printing and/or approximations in PAW mode can be controlled with pawprtdos keyword
(in particular,pawprtdos=2 can be used to compute quickly a very good approximation of the DOS).
Note 1: when prtdos=3, it is possible to output m-decomposed LDOS in _DOS file; simply use prtdosm keyword.
Note 2: the integrated total DOS in spheres around atoms can be obtained when
prtdensph flag is activated.
It can be compared to the integrated DOS provided in _DOS file when prtdos=3.
prtdos=4 delivers the sphere-projected DOS (like prtdos=3), on the basis of a smearing approach (like prtdos=1)
prtdos=5 delivers the spin-spin DOS in the nspinor==2 case, using the tetrahedron method (as prtdos=2).
Relevant only when prtdos=3.
If set to 1, the m-decomposed LDOS is delivered in DOS file.
Note that prtdosm computes the M-resolved partial dos for complex spherical harmonics,giving e.g.
DOS(L,M) == DOS(L,-M) (without spin-orbit). In the contrary, the LDA+U occupation matrix,
see dmatpawu is in the real spherical harmonics basis.
If set to 2, the m-decomposed LDOS is delivered in DOS file.
In this case, prtdosm computes the M-resolved partial dos for real spherical harmonics
in the same basis as the LDA+U occupation matrix.
If set to 1, a file *_EIG, containing the k-points and one-electron eigenvalues is printed.
If set to 1 or a larger value, provide output of ELF
in real space elf(r). This is a dimensionless quantity bounded between 0 and 1.
The name of the ELF file will be the root output name, followed by _ELF.
Like a _DEN file, it can be analyzed by cut3d. However unlike densities, in case of spin polarized
calculations, the spin down component can not be obtained by subtracting the spin up component to
the total ELF. Hence when spin polarized calculations are performed the code produces also output files with
_ELF_UP and _ELF_DOWN extensions. (For technical reasons these files contain also two components but the second is zero.
So to perform analysis of _ELF_UP and _ELF_DOWN files with cut3d you have to answer "ispden= 0 ==> Total density"
when cut3d ask you which ispden to choose. Also remember that spin down component can not be obtained by using cut3d on the _ELF file.
Sorry for the inconvenience, this will be fixed in the next release.)
ELF is not yet implemented in non collinear spin case.
If prtelf is set to 2, in the case of spin polarized calculation, the total ELF is computed from an alternative approach which should better take into account the existence of spin dependent densities (see the documentation in /doc/theory/ELF of your ABINIT repository)
Please note that ELF is
not
yet implemented in the case of PAW (usepaw=1) calculations.
If set to 1, provide Fermi surface file in the BXSF format (Xcrysden) If prtfsurf=1, a _BXSF file readable by XCrySDen will be produced at the end of the calculation. The file contains information on the band structure of the system and can be used to visualize the Fermi surface or any other energy isosurface. prtfsurf=1 is compatible only with SCF calculations (iscf > 1) or NSCF runs in which the occupation factors and Fermi level are recalculated once convergence is achieved (iscf = -3). The two methods should produce the same Fermi surface provided that the k-meshes are sufficiently dense. The k-mesh used for the sampling of the Fermi surface can be specified using the standard variables ngkpt, (shiftk, and nshiftk. Note, however, that the mesh must be homogeneous and centered on gamma (multiple shifts are not supported by Xcrysden)
If set to 1 or a larger value, provide output of gradient of electron density
in real space grho(r), in units of Bohr^-(5/2).
The names of the gradient of electron density files will be the root output name,
followed by _GDEN1, _GDEN2, GDEN3 for each principal direction (indeed it is a vector).
Like a _DEN file, it can be analyzed by cut3d. The file structure of the unformatted output file is
described below, see section 6).
If set to 1 or a larger value, provide output of geometrical analysis
(bond lengths and bond angles). The value
of prtgeo is taken by the code to be the
maximum coordination number of atoms in the system.
It will deduce a maximum number of "nearest" and "next-nearest"
neighbors accordingly , and compute corresponding bond lengths.
It will compute bond angles for the "nearest" neighbours only.
If ionmov==0, the name of the file will be
the root output name, followed by _GEO .
If ionmov==1 or 2, one file will be output
at each time step, with the name being made of
If set to 1, provide output of electron-phonon "gkk" matrix elements, for further treatment by mrggkk utility or anaddb utility. Note that symmetry will be disabled for the calculation of the perturbation, forcing the inclusion of all k-points and all perturbation directions. Additional information on electron-phonon treatment in ABINIT is given in the tutorial ~abinit/doc/tutorial/lesson_eph.html and in ~abinit/doc/users/elphon_manual.ps
If set to 1, ABINIT will produce a GSR file at the end of the GS
calculation. The GSR file contains the most important GS results
(band structure, forces, stresses, electronic density). The GSR
file can be read by AbiPy and used for futher postprocessing.
Note that, by default, the GSR file contains the electronic density
unless prtden is set to 0.
If set to 1 or a larger value , provide output of kinetic energy density
in real space tau(r), in units of Bohr^-5.
The name of the kinetic energy density file will be the root output name, followed by _KDEN.
Like a _DEN file, it can be analyzed by cut3d. The file structure of the unformatted output file is
described below (see
section 6
).
Note that the computation of the kinetic energy density must be activate, thanks to the input variable usekden.
Please note that kinetic energy density is
not
yet implemented in the case of PAW (usepaw=1) calculations.
If set /= 0 , proceeds to a detailed analysis of different k point grids. Works only if kptopt is positive, and neither kptrlatt nor ngkpt are defined. ABINIT will stop after this analysis.
Different sets of k point grids are defined, with common values of shiftk. In each set, ABINIT increases the length of vectors of the supercell (see kptrlatt) by integer steps. The different sets are labelled by "iset". For each k point grid, kptrlen and nkpt are computed (the latter always invoking kptopt=1, that is, full use of symmetries). A series is finished when the computed kptrlen is twice larger than the input variable kptrlen. After the examination of the different sets, ABINIT summarizes, for each nkpt, the best possible grid, that is, the one with the largest computed kptrlen.
Note that this analysis is also performed when prtkpt=0, as soon as neither kptrlatt nor ngkpt are defined. But, in this case, no analysis report is given, and the code selects the grid with the smaller ngkpt for the desired kptrlen. However, this analysis takes some times (well sometimes, it is only a few seconds - it depends on the value of the input kptrlen), and it is better to examine the full analysis for a given cell and set of symmetries, shiftk for all the production runs.
if set to -2, the code stops in invars1 after the computation of the irreducible set and a file named kpts.nc with the list of the k-points and the corresponding weights is produced
If set to 1 or a larger value, provide output of Laplacian of electron density
in real space grho(r), in units of Bohr^-(7/2).
The name of the Laplacian of electron density file will be the root output name,
followed by _LDEN.
Like a _DEN file, it can be analyzed by cut3d. The file structure of the unformatted output file is
described below (see
section 6
).
If set >=1 , provide output of the total (Kohn-Sham) potential (sum of local pseudo-potential, Hartree potential, and xc potential).
If ionmov==0, the name of the potential file will be
the root output name, followed by _POT.
If ionmov==1 or 2, potential file will be output
at each time step, with the name being made of
If set to 1, the code produces a netcdf file (PSPS.nc) with the internal tables used by Abinit to apply the pseudopotential part of the KS Hamiltonian. The data can be visualized with AbiPy. if prtpsps is set to -1, the code will exit after the output of the PSPS.nc file.
If set to 1 or a larger value, provide output of the current density of different direction spins (x,y,z) in the whole unit cell. Should require spinorial wave functions nspinor = 2. Experimental: this does not work yet.
If set to 1 or a larger value, provide output of the electron density
in real space rho(r), made only from the electrons
close to the Fermi energy, in a range of energy (positive or negative), determined
by the (positive or negative, but non-zero) value of the STM bias stmbias.
This is a very approximate way to obtain STM profiles : one can choose an equidensity surface,
and consider that the STM tip will follow this surface. Such equidensity surface might be determined
with the help of Cut3D, and further post-processing of it (to be implemented). The big approximations
of this technique are : neglect of the finite size of the tip, and
position-independent transfer matrix elements
between the tip and the surface.
The charge density is provided in units of electrons/Bohr^3.
The name of the STM density file will be the root output name, followed by _STM .
Like a _DEN file, it can be analyzed by cut3d.
The file structure of this unformatted output file is
described in
section 6.5
of abinit_help.
For the STM charge density to be generated, one must give, as an input file, the
converged wavefunctions obtained from a previous run, at exactly the same k-points and cut-off energy,
self-consistently determined, using the occupation numbers from occopt=7.
In the run with positive prtstm, one has to use :
If set to 0, no _SUSC file will be produced after the screening calculation, only the _SCR file will be output.
If set >= 0 outputs a file with the Coulomb potential, defined as Hartree + local Pseudopotential.
If prtvclmb=1 and in case of PAW (usepaw > 0), the full core potential is added for the Hartree part, with the on-site corrections vh1 - vht1.
If prtvclmb=2, only the smooth part of the Coulomb potential is output.
If set >=1 , provide output of the Hartree potential.
If ionmov==0, the name of the potential file will be
the root output name, followed by _VHA.
If ionmov==1 or 2, potential files will be output
at each time step, with the name being made of
If set >=1 , provide output of the sum of the Hartree potential and xc potential.
If ionmov==0, the name of the potential file will be
the root output name, followed by _VHXC.
If ionmov==1 or 2, potential files will be output
at each time step, with the name being made of
Control the volume of printed output. In particular, this
concerns the explicit echo of eigenenergies and residuals for all
bands and k points in the main output file.
Also, the analysis
of the value and location of the maximal density (and magnetization).
Standard choice is 0. Positive values print more in the
output and log files, while negative values are
for debugging (or preprocessing only), and cause
the code to stop at some point.
Control the volume of printed output
when an algorithm using images of the cell is used (nimage>1).
When such an algorithm is activated, the printing volume (in output file)
can be large and difficult to read.
Using
prtvolimg=1
, the printing volume, for each image,
is reduced to unit cell, atomic positions, total energy, forces, stresses, velocities
and convergence residuals.
Using
prtvolimg=2
, the printing volume, for each image,
is reduced to total energy and convergence residuals only.
If set >=1 , provide output of the local pseudo potential.
If ionmov==0, the name of the potential file will be
the root output name, followed by _VPSP.
If ionmov==1 or 2, potential files will be output
at each time step, with the name being made of
If set >=1 , provide output of the exchange-correlation potential.
If ionmov==0, the name of the potential file will be
the root output name, followed by _VXC.
If ionmov==1 or 2, potential files will be output
at each time step, with the name being made of
Flag used to indicate that either the Wannier90 or the WanT interfaces will be used.
Provide an output file that can be used by the WanT postprocessing program (see http://www.wannier-transport.org). The value of the prtwant indicates the version of the WanT code that can read it. Currently only the value prtwant=1 is implemented, corresponding to WanT version 1.0.1, available since Oct. 22, 2004.
Notes : Several requirements must be
fulfilled by the wavefunction. Among them, two are mandatory:
As an example of k-point grid in case of systems that have some 3D character (1D systems are easy) :
nkpt 8 kpt 0 0 0 0 0 1/2 0 1/2 0 0 1/2 1/2 1/2 0 0 1/2 0 1/2 1/2 1/2 0 1/2 1/2 1/2 istwfk 8*1
Also, in order to use WanT as a postprocessing program for ABINIT you might have to recompile it with the appropriate flags (see ABINIT makefile). Up to now only the -convert big-endian was found to be mandatory, for machines with little-endian default choice.
ABINIT will produce the input files required by Wannier90 and it will run Wannier90 to produce the Maximally-locallized Wannier functions (see http://www.wannier.org ).
Notes:
ABINIT will produce the input files required by Wannier90 and it will run Wannier90 to produce the Maximally-localized Wannier functions (see http://www.wannier.org ).
Additional Notes:
If prtwf=1 , provide output of wavefunction
and eigenvalue file, as described in
section 6.7
of the main abinit help file.
For a standard ground-state calculation, the name of the wavefunction file will be
the root output name, followed by _WFK. If nqpt=1,
the root name will be followed by _WFQ. For response-function calculations,
the root name will be followed by _1WFx, where x is the number of the perturbation.
The dataset information will be added as well, if relevant.
No wavefunction output is provided by prtwf=0.
If prtwf=-1, the code writes the wavefunction file only if convergence is not
achieved in the self-consistent cycle.
If prtwf=2, a file pwfn.data is produced, to be used as input for the
CASINO QMC code. See more explanation at the end of this section.
If prtwf=3, the file that is created is nearly the same as with prtwf=1,
except that the records that should contain the wavefunction is empty (so, such records exist, but store nothing).
This is useful to generate size-reduced DDK files,
to perform an optic run. Indeed, in the latter case, only matrix elements are needed [so, no wavefunction], but possibly
a large number of conduction bands, so that the DDK file might be huge if it contains the wavefunctions.
Further explanation for the prtwf=2 case. To produce a wave function suitable for use as a CASINO trial wave
function, certain ABINIT parameters must be set correctly. Primarily,
CASINO (and QMC methods generally) can only take advantage of
time-reversal symmetry, and not the full set of symmetries of the crystal
structure. Therefore, ABINIT must be instructed to generate k-points not
just in the Irreducible Brillouin Zone, but in a full half of the
Brillouin Zone (using time-reversal symmetry to generate the other half).
Additionally, unless instructed otherwise, Abinit avoids the need for
internal storage of many of the coefficients of its wave functions for
k-points that have the property 2k=G_latt, where G_latt is a reciprocal
lattice vector, by making use of the property that
c_k(G)=c^*_k(-G-G_latt). Abinit must be instructed not to do this in order
to output the full set of coefficients for use in CASINO. See the ABINIT
theoretical background documents ABINIT/Infos/Theory/geometry.pdf and
ABINIT/Infos/Theory/1WF.pdf for more information.
The first of these requirements is met by setting the ABINIT input
variable kptopt to 2 (see ABINIT/Infos/varbas.html#kptopt) and the second
by setting istwfk to 1 for all the k points (see
ABINIT/Infos/vardev.html#istwfk). Since CASINO is typically run with
relatively small numbers of k-points, this is easily done by defining an
array of "1" in the input file.
For example, for the 8 k-points generated with ngkpt 2 2 2, we add the
following lines to the input file:
# Turn off special storage mode for time-reversal k-points istwfk 1 1 1 1 1 1 1 1 # Use only time reversal symmetry, not full set of symmetries. kptopt 2Other useful input variables of relevance to the plane waves ABINIT will produce include ecut, nshiftk, shiftk, nband, occopt, occ, spinat and nsppol (see relevant input variable documents in ABINIT/Infos/). If ABINIT is run in multiple dataset mode, the different wave functions for the various datasets are exported as pwfn1.data, pwfn2.data, ..., pwfnn.data where the numbers are the contents of the contents of the input array jdtset (defaults to 1,2,...,ndtset).
Other issues related to prtwf=2.
The exporter does not currently work when ABINIT is used in parallel mode
on multiple processors if k-point parallelism is
chosen. ABINIT does not store the full wave function on each processor but
rather splits the k-points between the processors, so no one processor
could write out the whole file. Clearly this could be fixed but we have not
done it yet. The sort of plane wave DFT calculations usually required to
generate QMC trial wave functions execute very rapidly anyway and will
generally not require a parallel machines. The outqmc routine currently
bails out with an error if this combination of modes is selected - this
will hopefully be fixed later.
There has not been very extensive testing of less common situations such
as different numbers of bands for different k-points, and more complicated
spin polarized systems, so care should be taken when using the output in
these circumstances.
If there is any doubt about the output of this routine, the first place to
look is the log file produced by ABINIT: if there are any warnings about
incorrectly normalized orbitals or non-integer occupation numbers there is
probably something set wrong in the input file.
If set to 1 in a ground-state calculation, the code will output another WFK file (with extension FULL_WFK) containing the wavefunctions in the full BZ as well as a text file with the tables used for the tetrahedron method. Note that prtwf_full requires prtwf == 1 and a ground-state calculation done on a homogeneous k-mesh (see ngkpt and shiftk). The tetrahedron table is produced only if the number of k-points in the irreducible zone (nkpt) is greater than 3.
Create an XML output with common values. The corresponding DTD is distributed in sources as extras/post_processing/abinitRun.dtd. All the DTD is not yet implemented and this one is currently restricted to ground-state computations (and derivative such as geometry optimisation).