This code is able to analyse the files produced by ABINIT, that contain 3-Dimensional real space data, like all types of potential files, density files. Wavefunction files data can also be analysed : first, a k-point number, and the band number must be given, then, the corresponding wavefunction is transformed to real space.
In all these cases, thanks to Cut3D, one can obtain 2-Dimensional data corresponding to a cut by a plane, or 1-Dimensional data along a line. One can also translate the original formatting into many different ones.
Finally, one can also perform angular momenta analysis of wavefunctions with respect to any given atom, computation of the Hirshfeld atomic charge (starting from a density file). Although Wannier functions can also be built, this is a deprecated feature. The use of Wannier90 library is to be preferred.
cut3d
For option 1) you will have the possibility to specify a point in reduced or cartesian coordinates.
For option 2) you will have the possibility to specify a line by its two end-points in reduced or cartesian coordinates, or by it being perpendicular to some plane.
For option 3) and 4) many possibilities are offered, including specifications thanks to points defined in reduced coordinates, cartesian coordinates, or atomic positions.
For a wavefunction file, the two forms of outputs are :
the bare data or a Data Explorer form.
The option 10 also produce input files to draw the molecules in Data
Explorer using modules and macros that JFB and MCote
have developped. Those modules and macros could be made public.
To continue the analysis, simply answer the questions of the code, that
should be sufficiently self-explanatory.
In the formatted case, one must have three different files :
- the formatted 3D function (for example mos.densout)
density or potential file in ASCII format
with nr1 x nr2 x nr3 lines of 1 float
- a file with complementary informations about the cell, the FFT grid ...
(always named cut.in)
Here is an example (6 lines):
mos.xyz
11.502384 6.0558159 16.295674 unit cell parameters (=acell)
1 0 0 0 1 0 -0.215 0 0.976 orientation of the unit cell axes (=rprim)
54 30 75 grid size (= nr1,nr2,nr3, all three integers)
10 number of atoms in the unit cell (= natom, integer)
2 number of atomic types (=ntypat, integer)
The first line gives the name of the file with atomic positions (Xmol format) (=filtau)
No comment can appear on that line, in contrast with the other lines.
- the coordinate file
(for example mos.xyz)
Its format (similar to XMOL) is
On the first line the number of atoms (=natom) is provided.
Blank line
A list of natom lines containing :
Char (1 or 2 characters, = atom label) + 3 float (X, Y, Z cart. coord.)
In order to obtain Wannier type localized orbitals (WLO), you
have to type (or put in your input file) :
* 1st line : Name of the wavefunction file
* 2nd line : Fixed read option "1"
* 3rd line : option "2" to do WLO.
* 4th line : name of an "auxiliary" input file (name should not be more than 40 charater) file
to calculate WLO.
Examples of "auxiliary" input files are given in the automatic tests
Test_v4 cases 48 and 49.
The subroutine localorb_S.F90 referred from cut3d.F90 constructs
Wannier type localized orbitals (WLO) described in cond-mat/0509571 titles
" Localized Orbital Description of Electronic Structure".
Type of calculation
-------------------
These calculations run as a postprocessing of a wavefunction file
having wfs at all kpoints in the Brillouin zone (this may be generated
using kptopt 3 in a nonSCF calculation folloeing an SCF charge density
calculation).
1. Insulator type calculation with "insmet 1". All resultant WLOs would have
normalization 1 in the nkx.nky.nkz supercell where nkx.nky.nkz defines the
size of the k-point mesh.
2. Metal type calculation with "insmet 2". Some of the resultant WLOs might
have normalization less than 1.
3. (Not yet included) Maximally localized Wannier function from the WLOs.
NOTE : The code works for nshiftk = 1 only
OUTPUT
------
1. fort.1***1 : XCRYSDEN plottable WLO fort files.
fields:12345
field 1 : "1" for XCRYSDEN plottable files
field 2-4 : number corresponding to the orbital in the order
in which they are provided in the *.in file under
the variable "orbdetails"
field 5 : spin polarization : isppol 1 or 2
NOTE: To avoide disk space problem these files are not plotted
through out the entire supercell mentioned in the *.in file.
Instead they are plotted in a 2x2x2 supercell (or 1x1x1 if that
is the supercell mentioned in the *.in) with the orbital at the
middle.
2. fort.2***1 : Complex data files of WLO in the plotting region of fort.1***1
if "lofwrite" is set to 1. Only activate if this files are
really required for further quantitative analysis.
1st line : Location of the first grid point in the supercell of WLO
construction.
2nd line : The supercell grid point corresponding to the starting grid
point here.
3rd line : ngx ngy ngz in which the fort files are plotted.
WLO is plotted as : (((WLO(i,j,k),i=1,ngx),j=1,ngy),k=1,ngz)
3. singularvalues.dat : Contains singular values of the overlap matrix
between the energy occupied subspace and the auxiliary
subspace obtained by inverse Fourier transform
of the initial localized template described by
"orbdetails" in the input file.
NOTE: If you find some of the singular values consitently
very close to zero (0.00xxx) you may look at the overlap_abs.dat file
and exclude orbitals (from under the orbdetails variable
in the input file) corresponding to predominantly zero rows of
overlap matrix and start a new calculation.
4. lo_center.dat : position expectation value of the WLOs
column 1 2 3 4 5 6
index spin
5. lo_loclength.dat : variance of WLO
column 1 2 3 4 5 6
index var_x var_y var_z spin
NOTE : Values in the above two data files are meaningfull only
in case of insulator type calculation with sufficiently
large supercell to give WLO normalization close to 1.
6. overlap_abs.dat : Absolute(overlap matrix)
Examples can be found as Test_v4 cases 48 and 49.
Ongoing efforts
---------------
The second part which is yet to be upgraded is the
construction of Maximally localized orbitals from the WLOs.
The code is working but not yet in an efficient presentable
form. I hope to complete that some times in the
coming months.
As a disclaimer, I must admit that while interfacing my codes
with abinit I have really maintained least interference with
the main abinit procedures just to avoid more work for me!
Contact information for Wannier functions.
------------------------------------------
JOYDEEP BHATTACHARJEE
Research fellow
Theoretical Sciences Unit.
Jawaharlal Nehru Centre for
Advanced Scientific Research.
Jakkur, Bangalore -560064, India
Ph :+91-80-22082834-35 (2 lines)
Fax:+91-80-22082766
Cel:+91-9341312074