This file gives a beginner's introduction to the use of the ABINIT package.
The ABINIT package is written by the ABINIT group. See the files ~ABINIT/Infos/context and ~ABINIT/Infos/planning for more details about the ABINIT group and the ABINIT project.
You will find the welcome message, and basic information about the Web site in the welcome address .
Before reading the present file, you should get the paper ``Iterative minimization techniques for ab initio total-energy calculations: molecular dynamics and conjugate gradients'' M. C. Payne, M. P. Teter, D. C. Allan, T. A. Arias, and J. D. Joannopoulos, Rev. Mod. Phys. 64, 1045-1097 (1992), and read the introductory section.
After having gone through this beginner's introduction, you should follow the tutorial.
ABINIT is a package whose main program allows to find the total energy, charge density and electronic structure of systems made of electrons and nuclei (molecules and periodic solids) within Density Functional Theory, using pseudopotentials and a planewave basis. ABINIT also includes options to optimize the geometry according to the DFT forces and stresses, or to perform molecular dynamics simulation using these forces, or to generate dynamical matrices, Born effective charges, and dielectric tensors. In addition to the main ABINIT code, different utility programs are provided.
We will use the name ~ABINIT to refer to the directory that contains the ABINIT package. In practice, a version number is appended to this name, to give for example : ABINITv1.0.1. ~ABINIT contains different subdirectories. For example, the present file, as well as other descriptive files, should be found in ~ABINIT/Infos . Other subdirectories will be described later.
The main code exists in a sequential version, with the name abinis (ABINIT sequential), and in a parallel version, with the name abinip (ABINIT parallel). In the present new user's help file, we will suppose that the sequential version is used. After installation, it is present in the package as ~ABINIT/abinis .
To run abinis you need four things:
The full list of input variables, all of which are provided in the
single input file, is given in the ABINIT input variables file.
The detailed description of input variables is given in:
Many different sorts of pseudopotentials can be used with ABINIT. Most of them can be found on the ABINIT web site. There is a set of Teter hardness-conserving potentials, a set of Troullier-Martins potentials, a few Goedecker-Teter-Hutter pseudopotentials, and Hartwigsen-Goedecker-Hutter potentials for the whole periodic table. A subset of existing pseudopotentials are used for test cases, and are located in the ~ABINIT/Psps_for_tests directory. Information on pseudopotential files can be found in the ABINIT help file and the ~ABINIT/Infos/Psp_infos directory.
In addition to abinit, there are utility programs.
mrgddb, anaddb, aim, conducti
newsp, and cut3d are present in the package.
Others (presently kptgen) might be found on the
Web site.
mrgddb and anaddb allow to post-process reponses to atomic displacements and/or to homogeneous electric field, as generated by abinit, to produce full phonon band structures, or thermodynamical functions. "mrgddb" is for "Merge of Derivative DataBases", while "anaddb" is for "Analysis of Derivative DataBases".
Another utility is newsp, whose main routine source is called newsp.f. It allows a crude interpolation among the wavefunctions at different k points and is useful in reformatting wavefunction files to restart jobs on either new unit cell geometries, new planewave cutoffs, or new k point grids. Most of its capabilities have been transferred recently inside abinit, however.
cut3d can be used to post-process the three-dimensional density (or potential) files generated by abinit. It allows to deduce charge density in selected planes (for isodensity plots), along selected lines, or at selected points. It allows also to make the Hirshfeld decomposition of the charge density in "atomic" contributions.
aim is also a post-processor of the three-dimensional density files generated by abinit. It performs the Bader Atom-In-Molecule decomposition of the charge density in "atomic" contributions.
conducti allows to compute the frequency-dependent optical conductivity.
A last one is kptgen, that allows to find the symmetries of a set of atoms in a unit cell, and to generate grids of k-points. It is available on the Web site. All of its capabilities are present inside abinit, however, and are even more sophisticated.
At the level of graphics, many commercial or free softwares can be used to visualize ABINIT outputs. Some indications are contained in the ~ABINIT/Infos/Tutorial/lesson_visual file, but this topics has not yet been the subject of a systematic help file.
The ABINIT help file describes the input variables and the output file. As an overview, the most important input variables are listed below:
acell(3) lattice constant of periodic cell in bohr.
ecut planewave kinetic energy cutoff in hartree
ionmov when ionmov = 0 : the ions and cell shape are fixed
= 2 : search for the equilibrium geometry
= 6 : molecular dynamics
iscf either a positive number for defining self-consistent
algorithm (usual), or -2 for band structure in fixed potential
kptopt option for specifying the k-point grid
if kptopt=1, automatic generation, using ngkpt and shiftk.
(for the latter, see abinis_help)
natom total number of atoms in unit cell
ngkpt(4) dimensions of the three-dimensional grid of k-points
nstep maximal number of self-consistent cycles (on the order of 20)
ntime number of molecular dynamics or relaxation steps
ntypat number of types of atoms
occopt set the occupation of electronic levels :
=1 for semiconductors
=3 ... 7 for metals
rfelfd when /= 0 : will do response calculation to electric field
rfphon when = 1 : will do response calculation to atomic displacements
rprim(3,3) dimensionless primitive translations of periodic cell;
each COLUMN of this array is one primitive translation
typat(natom) sequence of integers, specifying the type of each atom.
NOTE: the atomic coordinates (xangst, xcart or xred)
must be specified in the same order
tolmxf force tolerance for structural relaxation in hartree/bohr
tolvrs tolerance on self-consistent convergence
xangst(3,natom) cartesian coordinates (Angstrom) of atoms in unit cell
NOTE: only used when "xred" and "xcart" are absent
xcart(3,natom) cartesian coordinates (bohr) of atoms in unit cell
NOTE: only used when "xred" and "xangst" are absent
xred(3,natom) fractional coordinates for atomic locations;
NOTE: leave out if xangst or xcart is used
znucl(ntypat) Nuclear charge of each type of element; must agree with
nuclear charge found in psp file.
Output from a abinis run shows up in several files and in the standard output. Usually one runs the command with a pipe of standard output to a log file, which can be inspected for warnings or error messages if anything goes wrong or otherwise can be discarded at the end of a run. The more easily readable formatted output goes to the output file whose name is given in the "files" file, i.e. you provide the name of the formatted output file. No error message is reported in the latter file. On the other hand, this is the file that is usually kept for archival purposes.
In addition, wavefunctions can be input (starting point) or output (result
of the calculation),
and possibly, charge density and/or electrostatic potential, if they
have been asked for. These three sets of data are stored in unformatted files.
The Density Of States (DOS) can also be an output as
a formatted (readable) file.
An analysis of geometry can also be provided (GEO file)
The name of these files is constructed from a "root" name, that must be
different for input files and output files, and that is provided
by the user, to which the code will append a descriptor, like
WFK for wavefunctions, DEN for the density, POT for the potential,
DOS for the density of states ...
There are also different temporary files. A "root" name should be provided by the user, from which the code generate a full name. Amongst these files, there is a "status" file, summarizing the current status of advancement of the code, in long jobs. ABINIT abinis_help contains more details.
The simplest sort of job computes an electronic structure for a fixed set of atomic positions within a periodic unit cell. By electronic structure , we mean a set of eigenvalues and wavefunctions which achieve the lowest (DFT) energy possible for that basis set (that number of planewaves). The code takes the description of the unit cell and atomic positions and assembles a crystal potential from the input atomic pseudopotentials, then uses either an input wavefunction or simple gaussians to generate the initial charge density and screening potential, then uses a self-consistent algorithm to iteratively adjust the planewave coefficients until a sufficient convergence is reached in the energy.
Analytic derivatives of the energy with respect to atomic positions and unit cell primitive translations yield atomic forces and the stress tensor. The code can optionally adjust atomic positions to move the forces toward zero and adjust unit cell parameters to move toward zero stress. It can performs molecular dynamics. It can also be used to find responses to atomic displacements and homogeneous electric field, so that the full phonon band structure can be constructed...
In order to know more about ABINIT, please follow the Tutorial