This document lists and provides the description of the name (keywords) of miscellaneous (ground state-related) input variables to be used in the main input file of the abinit code.
The new user is advised to read first the new user's guide, before reading the present file. It will be easier to discover the present file with the help of the tutorial.
When the user is sufficiently familiarized with ABINIT, the reading of the ~abinit/doc/users/tuning file might be useful. For response-function calculations using abinit, please read the response function help file
Used for the generation of alchemical pseudopotentials, that is, when ntypalch is non-zero.
Give the algorithm to be used to generate the ntypalch alchemical potentials from the different npspalch pseudopotentials dedicated to this use.
Presently, algalch can only have the value 1, that is :
Note that alchemical mixing cannot be used with PAW.
Go to the top
| Complete list of input variables
Defines the center of the box, in reduced coordinates.
At present, this information is only used in the case of
Time-Dependent DFT computation of the oscillator strength.
One must take boxcenter such as to be roughly the center of
the cluster or molecule. The default is sensible when
the vacuum surrounding the cluster or molecule has xred 0 or 1.
On the contrary, when the cluster or molecule is close to
the origin, it is better to take boxcenter=(0 0 0).
Go to the top
| Complete list of input variables
The box cut-off ratio is the ratio between the wavefunction plane wave sphere
radius, and the radius of the sphere that can be inserted in the
FFT box, in reciprocal space. In order for the density to be exact
(in the case of plane wave, not PAW), this ratio should be at least two.
If one uses a smaller ratio, one will gain speed, at the expense of accuracy.
In case of pure ground state calculation (e.g. for the determination
of geometries), this is sensible. However,
the wavefunctions that are obtained CANNOT be used for starting response function
calculation.
Go to the top
| Complete list of input variables
Used to establish charge balance between
the number of electrons filling the bands and the
nominal charge associated with the atomic cores.
The code adds up the number of valence electrons
provided by the pseudopotentials of each type
(call this "zval"), then add charge, to get the
number of electrons per unit cell,
nelect.
Then, if iscf is positive,
the code adds up the band occupancies (given in
array occ) for all bands at each k point,
then multiplies
by the k point weight wtk at each k point.
Call this sum "nelect_occ" (for the number of electrons
from occupation numbers). It is then
required that:
nelect_occ = nelect
To treat a neutral
system, which is desired in nearly all cases, one must
use charge=0. To treat a system missing one electron
per unit cell, set charge=+1.
Go to the top
| Complete list of input variables
If chkexit is 1 or 2, ABINIT will check whether the user wants to interrupt the run (using the keyword "exit" on the top of the input file or creating a file named "abinit.exit": see the end of section 3.2 of abinit_help).
If chkexit=0, the check is not performed at all
If chkexit=1, the check is not performed frequently (after each SCF step)
If chkexit=2, the check is performed frequently (after a few bands, at each k point)
In all cases, the check is performed at most every 2 seconds of CPU time.
Go to the top
| Complete list of input variables
If the symmetry finder is used (see nsym), a non-zero value of chkprim will make the code stop if a non-primitive cell is used. If chkprim=0, a warning is issued, but the run does not stop.
If you are generating the atomic and cell geometry using
spgroup, you might
generate a PRIMITIVE cell using
brvltt=-1 .
Go to the top
| Complete list of input variables
This variable governs the behaviour of the code when there are potential source of symmetry breaking, related e.g. to the k point grid or the presence of non-symmorphic translations which might not be coherent with the exchange-correlation grid.
When chksymbreak=1, the code stops if :
Explanation :
In the ground-state calculation, such breaking of the symmetry is usually harmless. However, if the user is doing a
calculation of phonons using DFPT (rfphon=1), the convergence with respect to the number of k points will
be much worse with a non-symmetric grid than with a symmetric one. Also, if the user is doing a GW calculation, the
presence of non-symmorphic translations that are not coherent with the FFT grid might cause
problems.
In the GW part, indeed, one needs to reconstruct the wavefunctions in the full Brillouin zone for
calculating both the polarizability and the self-energy.
The wavefunctions in the full Brillouin zone are obtained from the irreducible wedge by applying the symmetry
operations of the space group of the crystal.
In the present implementation, the symmetrization of the wavefunctions is done in real space on the FFT mesh
that, therefore, has to be coherent both with the rotational part as well as with the fractional translation
of each symmetry operation.
If the condition (2) is met, the GW code won't be able to find a symmetry-preserving FFT mesh.
So, it was decided to warn the user about these possible problems already at the level of the ground state calculations,
although such warning might be irrelevant.
If you encounter a problem outlined above, you have two choices : change your atomic positions (translate them) such that the origin
appears as the most symmetric point ; or ignore the problem, and set chksymbreak=0 .
Go to the top
| Complete list of input variables
One of these three real parameters can be
defined in the input file, to set up a CPU time limit.
When the job reaches that limit, it will try to end smoothly.
However, note that this might still take some time.
If the user want a firm CPU time limit, the present
parameter must be reduced sufficiently. Intuition
about the actual margin to be taken into account
should come with experience ...
Note that only one of these three parameters can be defined
in a single input file.
A zero value has no action of the job.
Internally, only cpus is used in the dtset array: adequate
conversion factors are used to generate it from cpum or
cpuh.
Go to the top
| Complete list of input variables
Kinetic energy cutoff that controls the number
of planewaves used to represent the dielectric matrix:
(1/2)[(2 Pi)*(Gmax)]^{2}=ecut for Gmax.
Can be specified in Ha (the default), Ry, eV or Kelvin, since
diecut has the
'ENERGY' characteristics.
(1 Ha=27.2113845 eV)
All planewaves inside this "basis sphere" centered
at G=0 are included in the basis.
This is useful only when iprcel>=21, which means that
a preconditioning scheme based on the dielectric matrix
is used.
NOTE : a negative diecut will define the same dielectric
basis sphere as the corresponding positive value,
but the FFT grid will be identical to the one used
for the wavefunctions.
The much smaller FFT grid, used when diecut is positive,
gives exactly the same results.
No meaning for RF calculations yet.
Go to the top
| Complete list of input variables
Gives a rough estimation of the dielectric gap
between the highest energy level computed in the run,
and the set of bands not represented.
Used to extrapolate dielectric matrix when iprcel >= 21.
Can be specified in Ha (the default), Ry, eV or Kelvin, since
diegap has the
'ENERGY' characteristics.
(1 Ha=27.2113845 eV)
No meaning for RF calculations yet.
Go to the top
| Complete list of input variables
Gives the amount of occupied states with mean energy given by the
highest level computed in the run, included
in the extrapolation of the dielectric matrix.
Used when iprcel >= 21.
No meaning for RF calculations yet.
Go to the top
| Complete list of input variables
Used for screening length (in Bohr) of the model
dielectric function, diagonal in reciprocal space.
By default, given in Bohr atomic units
(1 Bohr=0.5291772108 Angstrom), although Angstrom can be specified,
if preferred, since dielng has the
'LENGTH' characteristics.
This model dielectric function is as follows :
( 1 + dielng^{2} * K^{2} ) diel(K)= -------------------------------------------- ( 1/diemac + dielng^{2} * K^{2} ) * diemixThe inverse of this model dielectric function will be applied to the residual, to give the preconditioned change of potential. Right at K=0, diel(K) is imposed to be 1.
If the preconditioning were perfect,
the change of potential would lead to an exceedingly fast solution
of the self-consistency problem (two or three steps).
The present model dielectric function is excellent for
rather homogeneous unit cells.
When K->0 , it tends to the macroscopic dielectric
constant, eventually divided by the mixing factor diemix
(or diemixmag for magnetization).
For metals, simply put diemac to a very large value (10^6 is OK)
The screening length dielng governs the length scale
to go from the macroscopic regime to the microscopic
regime, where it is known that the dielectric function
should tend to 1. It is on the order of 1 Bohr for
metals with medium density of states at the Fermi level,
like Molybdenum, and for Silicon. For metals with a
larger DOS at the Fermi level (like Iron),
the screening will be more effective, so that dielng
has to be decreased by a factor of 2-4.
This works for GS and RF calculations.
TO BE IMPROVED : it is not clear what the variable K is and how to deal it, David Waroquiers, 090831.
Go to the top
| Complete list of input variables
A rough knowledge of the macroscopic dielectric constant diemac of the system is a useful help to speed-up the SCF procedure: a model dielectric function, see the keyword dielng, is used for that purpose. It is especially useful for speeding up the treatment of rather homogeneous unit cells.
Some hint :
The value of diemac should usually be bigger than 1.0d0,
on physical grounds.
For metals, simply put diemac to a very large value (the default 10^{6} is OK)
For silicon, use 12.0 . A similar value is likely to work well for
other semiconductors
For wider gap insulators, use 2.0 ... 4.0
For molecules in an otherwise empty big box, try 1.5 ... 3.0
Systems that combine a highly polarisable part and some vacuum are rather
badly treated by the model dielectric function. One has to use the
"extrapolar" technique, activated by the input variable
iprcel.
In sufficiently homogeneous systems, you might have to experiment
a bit to find the best diemac. If you let diemac
to its default value, you might even never obtain the self-consistent convergence !
For response function calculations, use the same
values as for GS. The improvement in speed can be considerable
for small (but non-zero) values of the wavevector.
Go to the top
| Complete list of input variables
Gives overall factor of the preconditioned
residual density/potential to be transferred in the SCF cycle.
It should be between 0.0 and 1.0 .
If the model dielectric function were perfect, diemix
should be 1.0 . By contrast, if the model dielectric function
does nothing (when diemac=1.0d0 or dielng
is larger than the
size of the cell), diemix can be used
to damp the amplifying factor inherent to the SCF loop.
For molecules, a value on the order 0.5 or 0.33 is rather usual.
When mod(iscf,10)=3, 4 ,5 or 7, diemix
is only important at the
few first iterations when anharmonic effects are important,
since these schemes compute their own mixing factor
for self-consistency.
Also note that a different value of diemix can be used for the magnetization
(see diemixmag).
Go to the top
| Complete list of input variables
Gives overall factor of the preconditioned
residual magnetization/magnetic field to be transferred in the SCF cycle (see diemix for further information).
For the time being, apply only when the SCF mixing is done on the density
(iscf>=10).
A negative value of diemixmag means that magnetization is only preconditionned by ABS(diemixmag),
without the use of any preconditionner.
When SCF cycle has some difficulties to converge, changing the value of diemixmag
can have a positive effect.
In particular diemixmag=-4 is a good choice
(i.e. diemixmag=4, no other preconditionner on magnetization).
Go to the top
| Complete list of input variables
Defines the linear grid resolution (energy increment) to be used for the
computation of the Density-Of-States, when prtdos
is non-zero.
If dosdeltae is set to zero (the default value), the actual
increment is 0.001 Ha if prtdos=1, and
the much smaller value 0.00005 Ha if prtdos=2.
This different default value arises because the prtdos=1 case,
based on a smearing technique, gives a quite smooth DOS, while the DOS from the
tetrahedron method, prtdos=2, is rapidly varying.
Go to the top
| Complete list of input variables
Governs the units to be used for output of eigenvalues (and eventual phonon frequencies)
Governs the number of bands to be used in the code in the case the parameter nband is not defined in the input file (which means that occopt is not equal to 0 or 2).
In case fband is 0.0d0, the code computes from
the pseudopotential files and the geometry data
contained in the input file, the number of electrons
present in the system. Then, it computes the minimum
number of bands that can accomodate them, and use
that value for nband.
In case fband differs from
zero, other bands will be added, just
larger than fband times the number of atoms.
This parameter is not echoed in the top of the main
output file, but only the parameter nband that it allowed
to compute. It is also not present in the dtset array (no internal).
The default values are chosen such as to give naturally some
conduction bands. This improves the robustness of the code,
since this allows to identify lack of convergence coming from
(near-)degeneracies at the Fermi level. In the metallic
case, the number of bands generated might be too small
if the smearing factor is large. The occupation numbers
of the higher bands should be small enough such as to
neglect higher bands. It is difficult to automate
this, so a fixed default value has been chosen.
Go to the top
| Complete list of input variables
This input variable is active only in the
prtdos=3 case or if
pawfatbnd=1 or 2.
It gives the number of the natsph atoms around which the sphere
for atom-projected density-of-states will be build,
in the prtdos=3 case.
The radius of these spheres is given by ratsph.
If pawfatbnd=1 or 2, it gives the number of the natsph atoms around which atom-projected band structure will be built.
Go to the top
| Complete list of input variables
Defines the type of computation used for Hartree potential, local part of pseudo-potential and ion-ion interaction:
Used when iscf>0, to define the SCF preconditioning scheme.
Potential-based preconditioning schemes for the SCF loop
(electronic part) are still a subject of active research.
The present parameter (electronic part) describes the way the
change of potential is derived from the residual.
The possible values of iprcel correspond to :
Used when iscf>0, to use the TFvW preconditioner. This is still in an early DEVELOPMENT stage and is not usable as is.
References:
Only used if nqpt=1, and qptopt=1 to 4.
Defines the index of the Q point to be selected in the list of q points generated by ngqpt, qptrlatt, nshiftq, and shiftq.
If iqpt=0, then the q point is Gamma (0 0 0).
The usual working mode is to define a series of values for iqpt, starting with iqpt=0 or 1 (so through the definition of iqpt:), and increasing it by one for each dataset (thanks to iqpt+).
Relevant only when positron/=0.
Define the type of electron-positron correlation that is used in case
of a electron-positron two-component DFT calculation.
Define also the analytical formula of the enhancement factor used to compute the electron-positron annhilation rate:
Electron-positron correlation functional:
If set to 1, a slab of uniform positive background charge density,
that is, a jellium slab, is included in the calculation cell.
A portion of the unit cell is filled with such positive charge density distribution
which is equal to a bulk-mean value n_{bulk} between two edges
and zero in the vacuum region if present.
For the sake of convenience the unit cell is supposed
to have the third crystal primitive lattice vector orthogonal
to the other ones so that the portion of the cell filled by the jellium slab can be defined through its edges along z.
The bulk-mean positive charge density is fixed by the input variable slabwsrad,
while the position of the slab edges along z is defined through
the input variables slabzbeg and slabzend.
Go to the top
| Complete list of input variables
It is used to generate the circuit to be followed by the band structure, when kptopt is negative (it is not read if kptopt is zero or positive).
There are abs(kptopt) segments to be defined, each of which starting from the end point of the preceeding one. Thus, the number of points to be input is abs(kptopt)+1. They form a circuit starting at kptbounds(1:3,1)/kptnrm and ending at kptbounds(1:3,abs(kptopt)+1)/kptnrm. The number of divisions of each segment can be defined either using the array ndivk or the variable ndivsm that just defines the number of divisions for the smallest segment
As for kpt, kptbounds is specified using the primitive vectors in reciprocal space. If your Bravais lattice is simple, then it should be quite easy to find the coordinates of the end points. On the other hand, for centered, body-centered, face-centered, hexagonal, and rhombohedral Bravais lattice, the conversion might be more difficult. See the description of kpt for an explanation of how to convert data from the "conventional" cartesian coordinates to the primitive vectors in the reciprocal space. In order to help a bit, we list below a series of typical values, for the FCC, BCC, hexagonal and rhombohedral Bravais lattices. Note : all the data below are given in dimensionless units ; they have to be rescaled by the actual lengths defined by the acell values. However, kptbounds values can be used as such, if the values of rprim given below are adopted.
A. FCC lattice
Suppose the primitive vectors in real space are given by
rprim 0 1 1 1 0 1 1 1 0or
rprim 0 1/2 1/2 1/2 0 1/2 1/2 1/2 0(these two possibilities only differ by a scaling factor, irrelevant for the definition of the k points in the primitive vectors in reciprocal space). Then, the reciprocal primitive vectors (in conventional cartesian coordinates) are
(-1/2 1/2 1/2), (1/2 -1/2 1/2), (1/2 1/2 -1/2)or
(-1 1 1), (1 -1 1), (1 1 -1)and, in both cases, the coordinates of several special points with respect to primitive vectors in reciprocal space are
X (0 1/2 1/2) (conventional cartesian coordinate 1/2 0 0) X'(1/2 1/2 1 ) (conventional cartesian coordinate 1/2 1/2 0) (an other instance of X, in another Brillouin zone) L (1/2 1/2 1/2) (conventional cartesian coordinate 1/4 1/4 1/4) L'(1/2 0 0 ) (conventional cartesian coordinate -1/4 1/4 1/4) (an other instance of L, on another face of the BZ) W (1/4 1/2 3/4) (conventional cartesian coordinate 1/2 1/4 0) U (1/4 5/8 5/8) (conventional cartesian coordinate 1/2 1/8 1/8) K (3/8 3/8 3/4) (conventional cartesian coordinate 3/8 3/8 0)Note that K is actually equivalent to U, by spatial and translational symmetry. So, if you want to specify a typical circuit, the following might do the work : L-Gamma-X-W-K,U-L-W-X-K,U-Gamma with
kptbounds 1/2 0 0 0 0 0 0 1/2 1/2 1/4 1/2 3/4 3/8 3/8 3/4 1/2 1/2 1/2 1/4 1/2 3/4 1/2 1/2 1 3/8 3/8 3/4 0 0 0
The lengths of segments (this information is useful to draw the band structure, with the correct relative scale between special points) can be found using the conventional cartesian coordinates : l(L-Gamma)=sqrt(3)/4=0.433... ; l(Gamma-X)=1/2=0.5 ; l(X-W)=1/4=0.25 ; l(W-K)=sqrt(2)/8=0.177... ; l(K-L)=sqrt(6)/8=0.306... ; l(L-W)=sqrt(2)/4=0.354... ; l(W-X)=1/4=0.25 ; l(X-K)=sqrt(2)/8=0.177... ; l(K-Gamma)=sqrt(2).3/8=0.530...
B. BCC lattice
Suppose the primitive vectors in real space are given by
rprim -1 1 1 1 -1 1 1 1 -1(as for the FCC lattice, there is a scale invariance). Then, the reciprocal primitive vectors (in conventional cartesian coordinates) are (0 1/2 1/2), (1/2 0 1/2), and (1/2 1/2 0) and the coordinates of several special points with respect to primitive vectors in reciprocal space are
H (-1/2 1/2 1/2) (conventional cartesian coordinate 1/2 0 0) N ( 0 0 1/2) (conventional cartesian coordinate 1/4 1/4 0) P ( 1/4 1/4 1/4) (conventional cartesian coordinate 1/4 1/4 1/4)So, if you want to specify a typical circuit, the following might do the work : Gamma-H-N-Gamma-P-N-P-H
kptbounds 0 0 0 -1/2 1/2 1/2 0 0 1/2 0 0 0 1/4 1/4 1/4 0 0 1/2 1/4 1/4 1/4 -1/2 1/2 1/2
The lengths of segments (this information is useful to draw the band structure, with the correct relative scale between special points) can be found using the conventional cartesian coordinates : l(Gamma-H)=1/2=0.5 ; l(H-N)=sqrt(2)/4=0.354... ; l(N-Gamma)=sqrt(2)/4=0.354... ; l(Gamma-P)=sqrt(3)/4=0.433... ; l(P-N)=1/4=0.25 ; l(N-P)=1/4=0.25 ; l(P-H)=sqrt(3)/4=0.433...
C. Hexagonal lattices
Suppose the primitive vectors in real space are given by
rprim 1 0 0 -1/2 sqrt(0.75) 0 0 0 1The coordinates of several special points with respect to primitive vectors in reciprocal space are
M (1/2 0 0) or (0 1/2 0) or (-1/2 1/2 0) L (1/2 0 1/2) or (0 1/2 1/2) or (-1/2 1/2 1/2) K (1/3 1/3 0) or (2/3 -1/3 0) or (-1/3 2/3 0) H (1/3 1/3 1/2) or (2/3 -1/3 1/2) or (-1/3 2/3 1/2) A (0 0 1/2)So, if you want to specify a typical circuit, the following might do the work : K-Gamma-M-K-H-A-L-H-L-M-Gamma-A
kptbounds 1/3 1/3 0 0 0 0 1/2 0 0 1/3 1/3 0 1/3 1/3 1/2 0 0 1/2 1/2 0 1/2 1/3 1/3 1/2 1/2 0 1/2 1/2 0 0 0 0 0 0 0 1/2
In order to find the lengths of segments (this information is useful to draw the band structure, with the correct relative scale between special points) one needs to know the a and c lattice parameters. Also, in what follows, we omit the 2*pi factor sometimes present in the definition of the reciprocal space vectors. The reciprocal vectors are (1/a 1/(sqrt(3)*a) 0) , (0 2/(sqrt(3)*a) 0), (0 0 1/c). The lengths of the above-mentioned segments can be computed as : l(K-Gamma)=2/(3*a)=0.666.../a ; l(Gamma-M)=1/(sqrt(3)*a)=0.577.../a ; l(M-K)=1/(3*a)=0.333.../a ; l(K-H)=1/(2*c)=0.5.../c ; l(H-A)=2/(3*a)=0.666.../a ; l(A-L)=1/(sqrt(3)*a)=0.577.../a ; l(L-H)=1/(3*a)=0.333.../a ; l(H-L)=1/(3*a)=0.333.../a ; l(L-M)=1/(2*c)=0.5.../c ; l(M-Gamma)=-1/(sqrt(3)*a)=0.577.../a ; l(Gamma-A)=1/(2*c)=0.5.../c
D. Rhombohedral lattices
Rhombohedral lattices are characterised by two parameters, the length of the primitive vectors, that we will denote a0, and the angle they form, alpha. These can be directly input of ABINIT, as acell and angdeg
This will generate the primitive vectors in real space , with
acell a0 a0 a0 and rprim a 0 c -a/2 a*sqrt(0.75) c -a/2 -a*sqrt(0.75) cwith a^2+c^2=1, a^2=(1-cos(alpha))*2/3, c^2=(1+2*cos(alpha))*1/3, (a/c)^2=2*(1-cos(alpha))/(1+2*cos(alpha)) and also cos(alpha)=(1-(a/c)^2/2)/(1+(a/c)^2). Alternatively, these values of rprim might directly be the input of ABINIT (then, the balance of the scaling factor might be adjusted between acell and rprim).
Unlike for the simple cubic, FCC, BCC, hexagonal (and some other) Bravais lattice, the topology of the Brillouin zone will depend on the alpha (or a/c) value. We give below information concerning the case when cos(alpha) is positive, that is, (a/c)^2 lower than 2.
The coordinates of several special points with respect to primitive vectors in reciprocal space will not depend on the a/c ratio, but some others will depend on it. So, some care has to be exercised. Notations for the Brillouin Zone special points are the same as in Phys. Rev. B 41, 11827 (1990).
L (1/2 0 0) or (0 1/2 0) or (0 0 1/2) (or with negative signs) T (1/2 1/2 1/2) X (1/2 1/2 0) or (1/2 0 1/2) or (0 1/2 1/2) (or with separate negative signs) W (5/6 - (a/c)^2/6 , 1/2 , 1/6 + (a/c)^2/6 ) = (1 0 -1)*(1-(a/c)^2/2)/3 + (1 1 1)/2 U ( (1+(a/c)^2)/6 , (8-(a/c)^2)/12 , (8-(a/c)^2)/12 ) = (-1 1/2 1/2)*(1-(a/c)^2/2)/3 + (1 1 1)/2 K (1 0 -1)*(1+(a/c)^2/4)/3So, if you want to specify a typical circuit, the following might do the work (the representative points on lines of symmetry are indicated - there are sometimes more than one way to go from one point to another) : X-V-K-Sigma-Gamma-Lambda-T-Q-W-Y-L-sigma-Gamma-sigma-X . The suggestion is to sample this path with the following coordinates for the special points X, Gamma, T, L, Gamma, X :
kptbounds 1/2 0 -1/2 0 0 0 1/2 1/2 1/2 1 1/2 0 1 0 0 1 1/2 1/2
In order to find the lengths of segments (this information is useful to draw the band structure, with the correct relative scale between special points) one needs to know the a and c lattice parameters. Also, in what follows, we omit the 2*pi factor sometimes present in the definition of the reciprocal space vectors. The reciprocal vectors are (2/(3*a) 0 1/(3*c)) , -(1/(3*a) 1/(sqrt(3)*a) 1/(3*c), -(1/(3*a) -1/(sqrt(3)*a) 1/(3*c) ). The lengths of the above-mentioned segments can be computed as : l(X-Gamma)=2/(sqrt(3)*a)=1.155.../a , with l(K-Gamma)=(1+(a/c)^2/4)*4/(3*sqrt(3)*a); l(Gamma-T)=1/(2*c) ; l(T-L)=2/(sqrt(3)*a)=1.155.../a , with l(T-W)=(1-(a/c)^2/2)*4/(3*sqrt(3)*a); l(L-Gamma)=sqrt(4/(a^2)+1/(c^2))/3 l(Gamma-X)=sqrt(1/(a^2)+1/(c^2))*2/3
Go to the top
| Complete list of input variables
This input variable is used only when kptopt is positive. It partially defines the k point grid. The other piece of information is contained in shiftk. kptrlatt cannot be used together with ngkpt.
The values kptrlatt(1:3,1), kptrlatt(1:3,2), kptrlatt(1:3,3) are the coordinates of three vectors in real space, expressed in the rprimd coordinate system (reduced coordinates). They defines a super-lattice in real space. The k point lattice is the reciprocal of this super-lattice, possibly shifted (see shiftk).
If neither ngkpt nor kptrlatt
are defined, ABINIT will automatically generate a set
of k point grids, and select the best combination
of kptrlatt and shiftk
that allows to reach a sufficient value of kptrlen.
See this latter variable for a complete description of this
procedure.
Go to the top
| Complete list of input variables
This input variable is used only when kptopt is positive and non-zero.
Preliminary explanation :
The k point lattice defined by ngkpt
or kptrlatt is used to perform integrations
of periodic quantities in the Brillouin Zone, like
the density or the kinetic energy. One can relate the
error made by replacing the continuous integral by a sum
over k point lattice to the Fourier transform of the
periodic quantity. Erroneous contributions will appear
only for the vectors in real space that belong to the reciprocal
of the k point lattice, except the origin.
Moreover, the expected size of these
contributions usually decreases exponentially with the distance.
So, the length of the smallest of these real space vectors
is a measure of the accuracy of the k point grid.
When either ngkpt or kptrlatt is defined, kptrlen is not used as an input variable, but the length of the smallest vector will be placed in this variable, and echoed in the output file.
On the other hand, when neither ngkpt nor kptrlatt are defined, ABINIT will automatically generate a large set of possible k point grids, and select among this set, the grids that give a length of smallest vector LARGER than kptrlen, and among these grids, the one that, when used with kptopt=1, reduces to the smallest number of k points. Note that this procedure can be time-consuming. It is worth doing it once for a given unit cell and set of symmetries, but not use this procedure by default. The best is then to set prtkpt=1, in order to get a detailed analysis of the set of grids.
If some layer of vacuum is detected in the unit cell
(see the input variable vacuum), the
computation of kptrlen will ignore the
dimension related to the direction perpendicular
to the vacuum layer, and generate a bi-dimensional k point grid.
If the system is confined in a tube,
a one-dimensional k point grid will be generated.
For a cluster, this procedure will only generate the Gamma point.
Go to the top
| Complete list of input variables
Used for the generation of alchemical pseudoatoms, that is, when ntypalch is non-zero.
This array gives, for each type of alchemical pseudatom (there are ntypalch such pseudoatoms), the mixing coefficients of the basic npspalch pseudopotentials for alchemical use. For each type of alchemical pseudoatom, the sum of the mixing coefficients must equal 1.
The actual use of the mixing coefficients is defined by the input variable algalch. Note that the masses of the atoms, amu are also mixed according to the value of mixalch, by default.
Example 1. Suppose that we want to describe Ba(0.25) Sr(0.75) Ti O3.
The input variables related to the construction of the alchemical Ba(0.25) Sr(0.75)
potential will be :
npsp 4 ! 4 pseudopotentials should be read. znucl 8 40 56 38 ! The nuclear charges. Note that the two ! atoms whose pseudopotentials are to be mixed ! are mentioned at the end of the series. ntypat 3 ! There will be three types of atoms. ntypalch 1 ! One pseudoatom will be alchemical. ! Hence, there will be ntyppure=2 pure pseudoatoms, ! with znucl 8 (O) and 40 (Ti), corresponding to ! the two first pseudopotentials. Out of the ! four pseudopotentials, npspalch=2 are left ! for alchemical purposes, with znucl 56 (Ba) ! and 38 (Sr). mixalch 0.25 0.75 ! For that unique pseudoatom to be ! generated, here are the mixing coeeficients, ! to be used to combine the Ba and Sr pseudopotentials.
Example 2. More complicated, and illustrate some minor drawback of the
design of input variables.
Suppose that one wants to generate Al(0.25)Ga(0.75) As(0.10)Sb(0.90).
The input variables will be :
npsp 4 ! 4 pseudopotentials should be read znucl 13 31 33 51 ! The atomic numbers. All pseudopotentials ! will be used for some alchemical purpose ntypat 2 ! There will be two types of atoms. ntypalch 2 ! None of the atoms will be "pure". ! Hence, there will be npspalch=4 pseudopotentials ! to be used for alchemical purposes. mixalch 0.25 0.75 0.0 0.0 ! This array is a (4,2) array, arranged in the 0.0 0.0 0.1 0.9 ! usual Fortran order.Minor drawback : one should not forget to fill mixalch with the needed zero's, in this later case.
In most cases, the use of mixalch
will be as a static (non-evolving) variable. However, the possibility to have
different values of mixalch for different images has been coded. A population of
cells with different atomic characteristics can thus be considered,
and can be made to evolve, e.g. with a genetic algorithm (not coded in v7.0.0 though).
There is one restriction to this possibility : the value of ziontypat for the atoms that are mixed should be
identical.
Go to the top
| Complete list of input variables
This input variable is active only in the
prtdos=3 case or if
pawfatbnd=1 or 2.
It gives the number of atoms around which the sphere
for atom-projected density-of-states will be built,
in the prtdos=3 case.
The indices of these atoms are given by iatsph.
The radius of these spheres is given by ratsph.
If pawfatbnd=1 or 2, it gives the number of atoms around which atom-projected band structure will be
built (the indices of these atoms are given by iatsph).
Go to the top
| Complete list of input variables
nbdbuf gives the number of bands, the highest in energy, that, among the nband bands, are to be considered as part of a buffer. This concept is useful in three situations: in non-self-consistent calculations, for the determination of the convergence tolerance ; for response functions of metals, to avoid instabilities, and also when finite electric fields or non-linear responses (with electric field perturbations) are considered. For the two first, the need of a buffer is a natural requirement of the problem, so that the default value is changed to 2 automatically, as explained in the following. The third case is only for implementation convenience.
In non-self-consistent GS calculations (iscf<0),
the highest levels might be
difficult to converge, if they are degenerate with another level,
that does not belong to the set of bands treated. Then, it might
take extremely long to reach tolwfr, although
the other bands are already extremely well-converged, and the energy
of the highest bands (whose residual are not yet good enough), is
also rather well converged.
In response to this problem, for non-zero nbdbuf, the
largest residual (residm), to be later compared with tolwfr,
will be computed only in the set of non-buffer bands (this modification
applies for non-self-consistent as well as self-consistent calculation,
for GS as well as RF calculations).
For a GS calculation, with iscf<0, supposing
nbdbuf is not initialized in the input file,
then ABINIT will overcome the default nbdbuf value,
and automatically set nbdbuf to 2.
In metallic RF calculations, in the conjugate gradient optimisation
of first-order wavefunctions, there is an instability situation
when the q wavevector of the perturbation brings the eigenenergy of the
highest treated band at some k point higher than the lowest
untreated eigenenergy at some k+q point.
If one accepts a buffer of frozen states, this instability can be made to
disappear. Frozen states receive automatically a residual value of -0.1d0.
For a RF calculation, with 3<=occopt<=7,
supposing
nbdbuf is not initialized in the input file, then
ABINIT will overcome the default nbdbuf value,
and automatically set nbdbuf to 2. This value might be too low
in some cases.
Also, the number of active bands, in all cases, is imposed
to be at least 1, irrespective of the value of nbdbuf.
Go to the top
| Complete list of input variables
Gives the number of divisions of each of the segments of the band structure, whose path is determined by kptopt and kptbounds. This is only needed when kptopt is negative. In this case, the absolute value of kptopt is the number of such segments.
For example, suppose that the number of segment is just one (kptopt=-1), a value ndivk=4 will lead to the computation of points with relative coordinates 0.0, 0.25, 0.5, 0.75 and 1.0 , along the segment in consideration.
Now, suppose that there are two segments (kptopt=-2), with ndivk(1)=4 and ndivk(2)=2, the computation of the eigenvalues will be done at 7 points, 5 belonging to the first segment, with relative coordinates 0.0, 0.25, 0.5, 0.75 and 1.0, the last one being also the starting point of the next segment, for which two other points must be computed, with relative coordinates 0.5 and 1.0 .
It is easy to compute disconnected circuits (non-chained segments), by separating the circuits with the value ndivk=1 for the intermediate segment connecting the end of one circuit with the beginning of the next one (in which case no intermediate point is computed along this segment).
Alternatively it is possible to generate automatically the array ndivk
by just specifying the number of divisions for the smallest segment.
See the related input variable ndivsm.
Go to the top
| Complete list of input variables
Gives the number of divisions used to sample the smallest segment
of the circuit employed during a band structure calculation
(see related input variables
kptopt
and
kptbounds).
If ndivsm is given in the input file, there is no need to specify the number of divisions
to be used for the other segments.
Indeed ndivk is automatically calculated inside the
code in order to generate a path where the number of divisions in each segment is proportial
to the length of the segment itself.
This option is activated only when kptopt is negative.
In this case, the absolute value of
kptopt is the number of such segments.
Go to the top
| Complete list of input variables
gives the size of fast Fourier transform
(fft) grid in three dimensions. Each number must be
composed of the factors 2, 3, and 5 to be consistent with
the radices available in our fft. If no ngfft is provided or
if ngfft is set to 0 0 0, the code will automatically provide
an optimal set of ngfft values, based on acell,
rprim and ecut
(see also boxcutmin for speed/accuracy concerns).
This is the recommended procedure, of course.
The total number of FFT points
is the product:
When ngfft is made smaller
than recommended values (e.g. by setting boxcutmin
to a value smaller than 2.0 or by setting ngfft manually), the code runs faster and the
equations in effect are approximated by a low pass Fourier
filter. The code reports to standard output (unit 06) a
parameter "boxcut" which is the smallest ratio of the fft
box side to the G vector basis sphere diameter. When
boxcut is less than 2 the Fourier filter approximation is being
used. When boxcut gets less than about 1.5 the
approximation may be too severe for realistic results
and should be tested against larger values of ngfft.
When boxcut is larger than 2, ngfft could be reduced without
loss of accuracy. In this case, the small variations
that are observed are solely due to the
xc quadrature, that may be handled with intxc=1
to even reduce this effect.
Internally, ngfft is an array of size 18. The present components are stored in ngfft(1:3), while
Used when nqpt=1, with kptopt>=0,
if qptrlatt
has not been defined (qptrlatt
and ngqpt are exclusive of each other).
At variance with ngkpt, note that only one q point
is selected per dataset (see iqpt).
Its three positive components
give the number of q points of Monkhorst-Pack grids
(defined with respect to primitive axis in reciprocal space)
in each of the three dimensions.
The use of nshiftq
and shiftq, allows to generate
shifted grids, or Monkhorst-Pack grids defined
with respect to conventional unit cells.
For more information on Monkhorst-Pack grids, see ngkpt.
Go to the top
| Complete list of input variables
Gives maximum number of line minimizations
allowed in preconditioned conjugate gradient minimization
for each band. The Default, 4, is fine.
Special cases, with degeneracies or near-degeneracies
of levels at the Fermi energy may require a larger value of
nline (5 or 6 ?)
Line minimizations will be stopped anyway when improvement
gets small. With the input variable nnsclo,
governs the convergence of the wavefunctions
for fixed potential.
Note that nline=0 can be used to diagonalize the Hamiltonian
matrix in the subspace spanned by the input wavefunctions.
Go to the top
| Complete list of input variables
Usually, the number of pseudopotentials to be read is equal to the number of type of atoms. However, in the case an alchemical mixing of pseudopotential is to be used, often the number of pseudopotentials to be read will not equal the number of types of atoms.
Alchemical pseudopotentials will be present
when ntypalch is non-zero.
See ntypalch
to understand how
to use alchemical potentials in ABINIT.
The input variables
(ntypalch,
algalch,mixalch)
are active, and generate alchemical potentials from the available
pseudopotentials. Also, the inner variables
(ntyppure,npspalch)
become active. See these input variables, especially
mixalch, to understand how
to use alchemical potentials in ABINIT.
Go to the top
| Complete list of input variables
Gives the number of pseudopotentials that are used for alchemical mixing (when ntypalch is non-zero) :
npspalch=npsp-ntyppure
Go to the top
| Complete list of input variables
Determines whether one q point
must be read (See the variable qptn).
Can be either 0 or 1.
If 1 and used in ground-state calculation,
a global shift of all the k-points is applied, to give
calculation at k+q.
In this case, the output wavefunction will be appended
by _WFQ instead of _WFK (see the section 4 of abinit_help)
Also, if 1 and a RF calculation is done, defines the
wavevector of the perturbation.
Go to the top
| Complete list of input variables
This parameter
gives the number of shifted grids
to be used concurrently to generate the full grid of q points.
It can be used with primitive grids defined either from
ngqpt
or
qptrlatt.
The maximum allowed value of nshiftq is 8.
The values of the shifts are given by shiftq.
Go to the top
| Complete list of input variables
If nspden=1, no spin-magnetisation : the density matrix is diagonal, with same values spin-up and spin-down (compatible with nsppol=1 only, for both nspinor=1 or 2)
If nspden=2, scalar magnetization (the axis is arbitrarily fixed in the z direction) : the density matrix is diagonal, with different values for spin-up and spin-down (compatible with nspinor=1, either with nsppol=2 -general collinear magnetisation- or nsppol=1 -antiferromagnetism)
If nspden=4, vector magnetization : the density matrix is full, with allowed x, y and z magnetisation (useful only with nspinor=2 and nsppol=1, either because there is spin-orbit without time-reversal symmetry - and thus spontaneous magnetization, or with spin-orbit, if one allows for spontaneous non-collinear magnetism). Not yet available for response functions. Also note that, with nspden=4, time-reversal symmetry is not taken into account (at present ; this has to be checked) and thus kptopt has to be different from 1 or 2.
The default (nspden=nsppol)
does not suit the case of vector magnetization.
Go to the top
| Complete list of input variables
If nspinor=1, usual case : scalar wavefunction (compatible with (nsppol=1, nspden=1) as well as (nsppol=2, nspden=2) )
If nspinor=2, the wavefunction is a spinor (compatible with nsppol=1, with nspden=1 or 4, but not with nsppol=2)
When nspinor is 2, the values of istwfk
are automatically set to 1. Also, the number of bands, for each k-point,
should be even.
Go to the top
| Complete list of input variables
Used for the generation of alchemical pseudopotentials : when ntypalch is non-zero, alchemical mixing will be used.
Among the ntypat types of atoms, the last ntypalch will be "alchemical" pseudoatoms, while only the first ntyppure will be uniquely associated with a pseudopotential (the ntyppure first of these, actually). The ntypalch types of alchemical pseudoatoms are to be made from the remaining npspalch pseudopotentials.
In this case,
the input variables
algalch,mixalch
are active, and generate alchemical potentials from the available
pseudopotentials. See these input variables, especially
mixalch, to understand how
to use alchemical potentials in ABINIT.
Go to the top
| Complete list of input variables
Gives the number of type of atoms that are "pure" when alchemical mixing is used (ntypalch /= 0) :
ntyppure=ntypat-ntypalch
Go to the top
| Complete list of input variables
In the wavelet basis set, the ground state is found by direct minimisation. The algorithm used can be either the steepest descent or the DIIS (Direct Inversion of Iteration Space). When nwfshist = 0, the steepest descent is used (i.e. there is no history storage of the previous iterations). If nwfshist is strictly positive, a DIIS is used. A typical value is 6. Using a DIIS increases the memory required by the program since N previous wavefunctions are stored during the electronic minimisation.
This input parameter can be positive or negative.
Negative values for positron are only relevant for PAW calculations.
Electron-positron correlation functional is defined by ixcpositron.
Other relevant input parameter: posocc (occupation number for the positron).
Positive values for positron:
For positron=1 or 2, will perform the calculation of positron
lifetime (and annihilation rate).
Relevant only when positron<0.
Sets the maximum number of electronic/positronic iterations that, when reached, will cause the two-component DFT SCF cycle to stop.
The code will first compute the electronic ground-state, then the positronic ground state in the electronic density, then the electronic ground-state in the positronic density, ...
...until diff_Etotal<postoldfe or diff_Forces<postoldff
or the number of electronic/positronic steps is posnstep.
Go to the top
| Complete list of input variables
Relevant only when positron/=0.
Sets the occupation number for the positron. Has to be <=1.
Changing posocc is only usefull for bulk calculation when one wants to perform lifetime computations using a small simulation cell (can avoid the use of a supercell). It simulates the dispersion of the positron in the whole crystal.
Go to the top
| Complete list of input variables
Relevant only when positron<0.
Sets a tolerance for absolute difference of total energy (of ions+electrons+positron system)
that, when reached, will cause the SCF cycle to stop before the number of SCF
steps is nstep or the number of electronic/positronic steps is posnstep.
Can be specified in Ha (the default), Ry, eV or Kelvin, since
toldfe has the 'ENERGY' characteristics.
Only one and only one of postoldfe or postoldff can be set.
Go to the top
| Complete list of input variables
Relevant only when positron<0.
Sets a tolerance for absolute difference of maximum force acting on ions (due to ions+electrons+positron system)
that, when reached, will cause the SCF cycle to stop before the number of SCF
steps is nstep or the number of electronic/positronic steps is posnstep.
Only one and only one of postoldfe or postoldff can be set.
Go to the top
| Complete list of input variables
Gives occupation numbers for all
bands in the problem. Needed if occopt==0 or
occopt==2.
Ignored otherwise. Also ignored when iscf=-2.
Typical band occupancy is either
2 or 0, but can be 1 for half-occupied band or other
choices in special circumstances.
If occopt is not 2,
then the occupancies must be the same for each k point.
If occopt=2, then the band occupancies must be
provided explicitly for each band, EACH k POINT,
and EACH SPIN-POLARIZATION, in an
array which runs over all bands, k points,
and spin-polarizations.
The order of entries in the array would correspond to
all bands at the first k point (spin up), then all bands at the
second k point (spin up), etc, then all k-points spin down.
The total number of array elements
which must be provided is
( nband(1)+nband(2)+...+
nband(nkpt) ) *
nsppol .
The occupation numbers evolve only for metallic occupations,
that is, occopt ≥ 3 .
Go to the top
| Complete list of input variables
For each dataset, choose the task to be done, at the level of the "driver" routine.
The choice is among :
optdriver=0 : ground-state calculation (GS), routine "gstate"
optdriver=1 : response-function calculation (RF), routine "respfn"
optdriver=2 : susceptibility calculation (SUS), routine "suscep"
optdriver=3 : susceptibility and dielectric matrix calculation (SCR), routine "screening"
(see the input variables ecutwfn,
ecuteps,
ppmfrq,
getkss,
as well as nbandkss and nband)
optdriver=4 : self-energy calculation (SIG), routine "sigma"
optdriver=5 : non-linear response functions, using the 2n+1 theorem, routine "nonlinear"
optdriver=99 : Bethe-Salpeter calculation, routine "bethe_salpeter"
If one of rfphon, rfddk,
rfelfd,
or rfstrs is non-zero, while optdriver
is not defined in the input file, ABINIT will set optdriver to 1
automatically. These input variables (rfphon,
rfddk, rfelfd,
and rfstrs) must be
zero if optdriver is not set to 1.
Go to the top
| Complete list of input variables
If set to 1, the computation of stresses is done,
in the SCF case
(under the conditions iscf > 0 , prtstm==0 ,
positron==0,
and either nstep >0 , or
usepaw==0 or irdwfk==1).
Otherwise, to save CPU time, if no optimization of the cell is required,
one can skip the computation of stresses. The CPU time saving might be interesting
for some PAW calculations.
Go to the top
| Complete list of input variables
This input parameter can be positive or negative.
Negative positron are only relevant for PAW calculation.
Electron-positron correlation functional is defined by ixcpositron.
Positive values for positron:
For positron=1 or 2, will perform the calculation of positron
lifetime (and annihilation rate).
Relevant only when positron<0.
Sets a tolerance for absolute difference of total energy (of ions+electrons+positron system)
that, when reached, will cause the SCF cycle to stop before the number of
steps is nstep.
Can be specified in Ha (the default), Ry, eV or Kelvin, since
toldfe has the 'ENERGY' characteristics.
Go to the top
| Complete list of input variables
When this flag is activated, values of integral(s) of total density inside sphere(s) around each atom are printed in output file (for each spin component).
Spheres around atoms are defined by a radius given by ratsph keyword.
Note: integral of density inside a sphere around an atom
can be used to determine a rough approximation of the local magnetic moment;
this is particulary useful for antiferromagnetic systems.
The algorithm to compute this integral is particularly primitive : the points on the FFT grids, belonging
to the interior of the sphere are determined, and the value of the functions on these points are summed,
taking into account a fixed volume attributed to each point.
In particular, the integral as a function of the radius will be a constant, except when
a new point enters the sphere, in which case a sudden jump occurs.
However, since the purpose of this output is to get a rough idea of the repartition of the density,
this is not a real problem. If you are interested in a more accurate estimation
of the density within a sphere, you should use the cut3d postprocessor.
Go to the top
| Complete list of input variables
Only used if nqpt=1.
Combined with qptnrm, define the q vector qptn(1:3) in the case qptopt=0.
This input variable is not internal (qptn(1:3) is used
instead), but is used to echo the value of qptn(1:3),
with renormalisation factor one.
Go to the top
| Complete list of input variables
Only used if nqpt=1.
Controls the set up to generate the Q point qptn(1:3) to be used for the specific dataset, either as a shift of k-point grid in ground-state calculations, or as a stand-alone phonon wavevector.
There are two basic techniques to generate the Q point : either by specifying it directly, possibly with a renormalisation factor (qptopt=0), or extracting it from a grid a Q points (qptopt=1 to 4), using the index iqpt. At variance with the similar generation of k points, only ONE q point can be used per dataset.
With qptopt=1 to 4, rely on ngqpt or qptrlatt, as well as on nshiftq and shiftq to set up a q point grid, from which the q point with number iqpt will be selected. The values qptopt=1 to 4 differ by the treatment of symmetries. Note that the symmetries are recomputed starting from the values of rprimd xred and spinat. So, the explicit value of symrel are not used. This is to allow doing calculations with nsym=1, sometimes needed for T-dependent electronic structure, still decreasing the number of q points in the case qptopt=1 or qptopt=3.
Only used if nqpt=1 and qptopt=0
Provides re-normalization
of qpt.
Must be positive, non-zero.
The actual q vector (renormalized) is
qptn(1:3)=
qpt(1:3)/qptnrm.
Go to the top
| Complete list of input variables
This input variable is used only when qptopt is positive. It partially defines the q point grid. The other piece of information is contained in shiftq. qptrlatt cannot be used together with ngqpt.
The values qptrlatt(1:3,1), qptrlatt(1:3,2), qptrlatt(1:3,3) are the coordinates of three vectors in real space, expressed in the rprimd coordinate system (reduced coordinates). They defines a super-lattice in real space. The k point lattice is the reciprocal of this super-lattice, possibly shifted (see shiftq).
If neither ngqpt nor qptrlatt
are defined, ABINIT will automatically generate a set
of k point grids, and select the best combination
of qptrlatt and shiftq
that allows to reach a sufficient value of kptrlen.
See this latter variable for a complete description of this
procedure.
Go to the top
| Complete list of input variables
Relevant only when
prtdos=3 or prtdensph=1.
When prtdos=3:
Provides the radius of the spheres around the natsph atoms
of indices iatsph, in which the local
DOS and its angular-momentum projections will be analysed.
The choice of this radius is quite arbitrary. In a plane-wave basis set,
there is no natural definition of an atomic sphere. However, it might be wise
to use the following well-defined and physically motivated procedure
(in version 4.2, this procedure is NOT implemented, unfortunately) :
from the Bader analysis, one can define the radius of the sphere
that contains the same charge as the Bader volume. This
"Equivalent Bader charge atomic radius" might then be used to perform
the present analysis.
See the AIM (Bader) help file for more explanations.
Another physically motivated choice would be to rely on another
charge partitioning, like the Hirshfeld one (see the cut3d utility).
The advantage of using charge partitioning schemes comes from the fact that the
sum of atomic DOS, for all angular momenta and atoms, integrated on the
energy range of the occupied states,
gives back the total charge.
If this is not an issue, one could rely on the half of the nearest-neighbour distances, or
any scheme that allows to define an atomic radius. Note that the choice of this
radius is however critical for the balance between the s, p and d components. Indeed,
the integrated charge within a given radius, behave as a different power of the
radius, for the different channels s, p, d. At the limit of very small radii, the s component
dominates the charge contained in the sphere ...
When prtdensph=1:
Provides the radius of the spheres around (all) atoms in which the total charge density will be integrated.
In case of PAW, ratsph radius has to be greater or equal to
PAW radius of considered atom type (which is read from the PAW dataset file; see rc_sph or r_paw).
Go to the top
| Complete list of input variables
Give extent, in number of primitive unit cells, of the supercell being used for
a self-consistent phonon calculation. Presumes the phonon frequencies and eigenvectors
have been calculated in the original primitive unit cell, on a grid of q-points which
corresponds to the supercell in the present calculation.
TO BE IMPROVED : should contain a tutorial on how to do self-consistent phonon calculations, David Waroquiers 090831
Go to the top
| Complete list of input variables
Temperature which is imposed on phonon distribution, in the self-consistent scheme of
Souvatzis et al. PRL 100, 095901. Determines the extent of the finite displacements
used, and consequent anharmonic effects. Experimental.
Go to the top
| Complete list of input variables
It is used only when qptopt>=0,
and must be defined if nshiftq is larger than 1.
shiftq(1:3,1:nshiftq) defines
nshiftq shifts
of the homogeneous grid of q points
based on ngqpt or
qptrlatt.
See shiftk for more information on the definition,
use, and suitable values for these shifts.
Go to the top
| Complete list of input variables
Fix the bulk-mean positive charge density n_{bulk} of a jellium slab (if the latter is employed, e.g. jellslab &ne 0). Often called "r_{s}" [see for example N. D. Lang and W. Kohn PRB 1, 4555 (1970)], slabwsrad is the radius of a sphere which has the same volume as the average volume per particle in a homogeneous electron gas with density n_{bulk}, so:
1/n_{bulk} = 4/3 Pi * slabwsrad^{3}For example, the bulk aluminum fcc lattice constant is a=4.0495 Angstroms (webelements.com), each cubic centered cell includes 4 Al atoms and each atom has 3 valence electrons, so the average volume per electron is a^{3}/12=37.34 Bohr^{3} which has to be equal to 4/3 Pi*r_{s}^{3}. Consequently Al has approximately r_{s}=2.07 Bohr, while for example magnesium has r_{s}=2.65 Bohr, sodium 3.99 Bohr.
slabzend Mnemonics: jellium SLAB ENDing edge along the Z direction Characteristic: Variable type: real parameter Default is 0.0d0, 0.0d0.
Define the edges of the jellium slab (if used, so if jellslab &ne 0) along z, namely the slab starts at a point along z which is expressed in Bohr by slabzbeg and it ends at a point expressed in Bohr by slabzend. The z direction is parallel to the third crystal primitive lattice vector which has to be orthogonal to the other ones, so the length of the cell along z is rprimd(3,3). In addition slabzbeg and slabzend have to be such that:
0 &le slabzbeg < slabzend &le rprimd(3,3)Together with slabwsrad they define the jellium positive charge density distribution n_{+}(x,y,z) in this way:
n_{+}(x,y,z) = n_{bulk} if slabzbeg &le z &le slabzend = 0 otherwise,so the positive charge density is invariant along the xy plane as well as the electrostatic potential generated by it.
For each type of atom (each pseudopotential), specify
the treatment of spin-orbit interaction (if nspinor==2).
If 0 : no spin-orbit interaction, even if nspinor=2
If 2 : treat spin-orbit in the HGH form
(not allowed for all pseudopotentials)
If 3 : treat spin-orbit in the HFN form
(not allowed for all pseudopotentials)
Also, so_psp=1 lead automatically to treatments 0, 2, or 3 according to the data contained in the pseudopotential file (0= there is no spin-orbit information in the psp file; 2= the spin-orbit information is of the HGH form; 3= the spin-orbit information is of the HFN form ) So, for typical usage, one will need only the values 0 or 1
Note that if nspinor==1, the spin-orbit cannot be treated anyhow, so the value of so_psp is irrelevant.
Prior to v5.4, the input variable so_typat
was used, in place of so_psp. Because the values 0 and 1 have been switched
between so_psp and so_typat, it was dangerous to continue to allow the use of so_typat.
Go to the top
| Complete list of input variables
Gives the initial electronic spin-magnetisation for each atom, in unit of h-bar/2.
Note that if nspden=2,
the z-component must be given
for each atom, in triplets (0 0 z-component).
For example, the electron of an hydrogen atom
can be spin up (0 0 1.0) or spin down (0 0 -1.0).
This value is only used to create
the first exchange and correlation potential,
and is not used anymore afterwards.
It is not checked against the initial occupation numbers
occ for each spin channel.
It is meant to give an easy way to break
the spin symmetry, and to allow
to find stable local spin fluctuations, for example :
antiferromagnetism, or the spontaneous spatial
spin separation of elongated H2 molecule.
Gives, in Hartree, the
bias of the STM tip, with respect to the sample, in order to generate
the STM density map.
Used with positive iscf,
occopt=7 (metallic, gaussian),
nstep=1 ,
and positive prtstm, this
value is used to generate a charge density map from electrons
close to the Fermi energy, in a (positive or negative) energy range.
Positive stmbias will lead to the inclusion of occupied (valence) states only, while
negative stmbias will lead to the inclusion of unoccupied (conduction) states only.
Can be specified in Ha (the default), Ry, eV or Kelvin, since
stmbias has the
'ENERGY' characteristics
(0.001 Ha = 27.2113845 meV = 315.773 Kelvin).
With occopt=7,
one has also to specify an independent broadening tsmear.
Go to the top
| Complete list of input variables
In case the material is magnetic (well, this is only interesting in the
case of antiferromagnetism, collinear or not), additional symmetries might appear, that
change the sign of the magnetisation.
They have been introduced by Shubnikov (1951). They can be used by ABINIT
to decrease the CPU time, by using them to decrease the number of k-points.
symafm should be set to +1 for all the usual symmetry operations,
that do not change the sign of the magnetisation, while it should be
set to -1 for the magnetisation-changing symmetries.
If the symmetry operations are not specified by the user
in the input file, that is, if nsym=0,
then ABINIT will use the values of spinat
to determine the content of symafm.
The symmetries found as "antiferro magnetic" (symafm=-1) are used to symmetrize density and magnetization in the following cases:
- antiferromagnetism (nsppol=1,
nspinor=1,
nspden=2)
- non-collinear magnetism (nsppol=1,
nspinor=1,
nspden=4)
In other cases they are not used.
Go to the top
| Complete list of input variables
This input variable allows to modulate the use of the timing routines.
If 0 => as soon as possible, suppresses all calls to timing routines
If 1 => usual timing behaviour, with short analysis, appropriate
for sequential execution
If 2 => close to timopt=1, except that the analysis routine
does not time the timer, appropriate for parallel execution.
If 3 => close to timopt=1, except that the different parts of the lobpcg routine are timed in detail.
If 4 => close to timopt=1, except that the different parts of the lobpcg routine are timed in detail.
A different splitting of lobpcg than for timopt=-3 is provided.
If -1 => a full analysis of timings is delivered
If -2 => a full analysis of timings is delivered,
except timing the timer
If -3 => a full analysis of timings is delivered, including the detailed timing of the different parts of the lobpcg routine.
(this takes time, and is discouraged for too small runs - the timing would take more time than the run !). The timer is timed.
If -4 => a full analysis of timings is delivered, including the detailed timing of the different parts of the lobpcg routine.
A different splitting of lobpcg than for timopt=-3 is provided
(this takes time, and is discouraged for too small runs - the timing would take more time than the run !). The timer is timed.
The sum of the independent parts is closer to 100% than for timopt=-3.
Go to the top
| Complete list of input variables
This variable is the same than wvl_nprccg
but for the preconditionner iterations during the tail
corrections (see tl_radius).
TO BE IMPROVED : all tl_* and wvl_* variables should contain a link to a tutorial, David Waroquiers 090831.
Go to the top
| Complete list of input variables
In the wavelet computation case, the linkage between the grid and the
free boundary conditions can be smoothed using an exponential
decay. This means a correction on the energy at the end on each
wavefunction optimisation run. If this parameter is set to zero,
no tail computation is done. On the contrary, put it to a
positive value makes the tail correction available. The value
correspond to a length in atomic units being the spacial expansion
with the exponential decay around the grid.
Go to the top
| Complete list of input variables
Gives, in Hartree, the physical temperature of the
system, in case occopt=4, 5, 6, or 7.
Can be specified in Ha (the default), Ry, eV or Kelvin, since
ecut has the
'ENERGY' characteristics
(0.001 Ha = 27.2113845 meV = 315.773 Kelvin).
One has to specify an independent broadening tsmear.
The combination of the two parameters
tphysel and tsmear is described
in a paper by M. Verstraete and X. Gonze, Phys. Rev. B 65, 035111 (2002).
Note that the signification of the entropy is modified with respect
to the usual entropy. The choice has been made to use
tsmear as a prefactor of the entropy,
to define the entropy contribution to the free energy.
Go to the top
| Complete list of input variables
Gives the broadening of occupation
numbers occ, in the metallic cases
(occopt=3, 4, 5, 6 and 7).
Can be specified in Ha (the default), eV, Ry, or Kelvin, since
tsmear has the
'ENERGY' characteristics
(0.001 Ha = 27.2113845 meV = 315.773 Kelvin).
Default is 0.04 Ha. This should be OK for a free-electron
metal like Al. For d-band metals, use 0.01 Ha.
Always check the convergence of the calculation
with respect to this parameter, and simultaneously,
with respect to the sampling of k-points (see nkpt)
If occopt=3, tsmear is the
physical temperature, as the broadening is based on Fermi-Dirac statistics.
However,
if occopt=4, 5, 6, or 7,
the broadening is not based on Fermi-Dirac statistics, and
tsmear is only a convergence parameter. It is still possible
to define a physical temperature, thanks to the input variable
tphysel. See the paper
by M. Verstraete and X. Gonze, Phys. Rev. B (2002).
Go to the top
| Complete list of input variables
If usekden=1 the kinetic energy density will be computed during the self-consistency loop, in a way similar to the computation of the density. This is needed if a meta-GGA is to be used as XC functional. Otherwise (usekden=0), the kinetic energy density is not computed during the self-consistency loop.
Establishes the presence (if 1) or absence (if 0) of a vacuum layer, along the three possible directions normal to the primitive axes.
This information might be used to generate k-point grids,
if kptopt=0 and neither
ngkpt nor kptrlatt
are defined (see explanations with the input variable
prtkpt).
It will allow to select
a zero-, one-, two- or three-dimensional
grid of k points. The coordinate of the k points
along vacuum directions is automatically set to zero.
If vacuum is not defined, the input variable
vacwidth
will be used to determine automatically whether the
distance between atoms is sufficient to have the
presence or absence of vacuum.
Go to the top
| Complete list of input variables
Give a minimum "projected" distance between
atoms to be found in order to declare that there
is some vacuum present for each of the three
directions.
By default, given in Bohr atomic units
(1 Bohr=0.5291772108 Angstroms), although Angstrom can be specified,
if preferred, since vacwidth has the
'LENGTH' characteristics.
The precise requirement is that a slab
of width vacwidth, delimited by two
planes of constant reduced coordinates in the
investigated direction, must be empty of atoms.
Go to the top
| Complete list of input variables
This factor is used to defined the expansion of the coarse resolution
grid in the case of wavelets (see usewvl). The grid is made of
points inside spheres centered on atoms. The radius of these
spheres are the product between this factor and the covalent
radius of element (read from the pseudo-potential file).
This factor is responsible for the amount of used memory (see also wvl_hgrid).
Go to the top
| Complete list of input variables
This factor is used to defined the expansion of the fine resolution
grid in the case of wavelets (see usewvl). This fine resolution
grid has the same grid step than the coarse one (see wvl_crmult), but on each
point, 8 coefficients are stored instead of one, increasing the
precision of the calculation in this area. The grid is made of
points inside spheres centered on atoms. The radius of these
spheres are the product between this factor and a value read from the pseudo-potential file.
This factor is responsible for the amount of used memory (see also wvl_hgrid).
Go to the top
| Complete list of input variables
In the wavelet computation case, the wavefunctions are directly
minimised using a real-space preconditionner. This preconditionner
has internally some conjugate gradient iterations. This value
defines a boundary for the number of conjugate gradient
iterations on each wavefunction convergence step.
Go to the top
| Complete list of input variables
Automatically determined from the value of ixc.