The ABINIT Users Mailing List ( CLOSED )

["Anglade Pierre-Matthieu" <anglade@gmail.com>]

Hi,

If you can't perform a converged calculation you can't get trustable
results and using Abinit is meaning less.
I would recommand you the following approach it will at least tell you
to if you can get a result with Abinit+ your machine and to which
extent this result could be trusted :
- use TiCl in a small box with few k-points and determine the cutoff
energy you need
- determine the largest cell you can use on your computer with a
single k-point and the converged cutoff (for Ti I would expect
something between 50 and 90 Ha)
- Use the largest cell with a small cutoff (20-30 Ha) and determine
how many k-point you need to get a converged result (more precisely
determine the usefull kptrlen)
- Perform you calculation with the correct value for kptrlen and ecut
but with various cell size smaller (and up to) the max size you can do
(don't forget to use the various memory saving tricks of Abinit).
- Now you have the evolution of the Ti-Cl distance with respect to
cell size and its you to determine weither it is converged or not.

Also, you should know that Abinit pseudopotentials comes as is. It's
you to determine if they are trustable for your system. Usualy people
look at lattice parameters for elements which must be well reproduced
by any pseudopotential.
You may want to try some of the other pseudopotential species with the
same XC to test...

Regards

PMA

On Fri, Mar 21, 2008 at 11:39 PM, Deskins, Nathaniel A
<nathaniel.deskins@pnl.gov> wrote:
> I'm assuming the main criteria for convergence is the cutoff energy.
>  This is a large box, so a single k-point should suffice. Here's the
>  results of a cutoff energy test.
>
>  Cutoff(hartree)   Erxn(eV)   D-TiCl(angstroms)
>  10                5.11       2.39
>  15                5.12       2.39
>  20                3.80       2.40
>  25                3.79       2.40
>  30                3.79       2.39
>  35                4.64       2.35
>
>  These show some pretty big oscillations. I couldn't go to 40 eV because
>  of memory restraints (2 GB required for 40 hartree).
>
>  I'm guessing I'm doing something wrong, which is why I posted my input
>  files so perhaps someone might notice an obvious error.
>
>  Thank you,
>  Aaron
>
>
>
>  -----Original Message-----
>  From: Anglade Pierre-Matthieu [mailto:anglade@gmail.com]
>  Sent: Friday, March 21, 2008 3:50 AM
>  To: forum@abinit.org
>  Subject: Re: [abinit-forum] Problems with TiCl Reaction
>
>  Hi,
>
>  Did you try to make a converged calculation ? What is the "absolute"
>  precision of your simulation ?
>  For instance usualy light d metals require large cutoff energy.
>
>  regards
>
>  PMA
>
>  On Thu, Mar 20, 2008 at 8:14 PM,  <nathaniel.deskins@pnl.gov> wrote:
>  > Hello all,
>  >
>  >   I'm looking at the following gas-phase reaction. TiCl --> Ti + Cl.
>  >  The ground state for TiCl is a quintet (3 unpaired e-), Ti is a
>  pentet(?) (4 unpaired e-), while Cl is a doublet (1 unpaired e-). I've
>  modeled this reaction with other planewave and gaussian codes. The
>  energy for the reaction varies between 4.1-5.1 eV (depending on basis
>  set and potential) while the Ti-Cl distance is near 2.22-2.23 angstroms.
>  All this was with PBE XC functional.
>  >
>  >   When I've tried Abinit with the FHI GGA-PBE Troullier-Martins
>  pseudopotentials I get strange results. My calculated energy is 6.58 eV
>  while the Ti-Cl distance is 2.40 angstroms. This was for a cutoff of 15
>  hartree. For cutoff of 30 hartree the energy is 3.8 eV(?!) and the
>  distance is 2.39 angstroms.
>  >
>  >   Either there is an issue with the pseudopotentials, or I'm doing
>  something wrong.
>  >
>  >  Below are my input files. Any help is appreciated.
>  >
>  >  Thank you,
>  >  Aaron
>  >
>  >
>  >  ti.in
>  >  acell 25 25 25    # The keyword "acell" refers to the
>  >  ntypat 1          # There is only one type of atom
>  >  znucl 22           # The keyword "znucl" refers to the atomic number
>  of the
>  >  natom 1           # There are two atoms
>  >  typat 1          # They both are of type 1, that is, Hydrogen
>  >  xcart             # This keyword indicates that the location of the
>  atoms
>  >    0.0 0.0 0.0    # Triplet giving the cartesian coordinates of atom
>  1, in Bohr
>  >  ecut 15.0         # Maximal plane-wave kinetic energy cut-off, in
>  Hartree
>  >  nkpt 1            # Only one k point is needed for isolated system,
>  >  nsppol 2
>  >  occopt 2
>  >  nband 8 8
>  >  occ 1 1 1 1 0 0 0 0 0 0 0 0 0 0 0 0
>  >  spinat 0.0 0.0 4.0  # Initialisation of spin
>  >  nstep 50          # Maximal number of SCF cycles
>  >  toldfe 1.0d-6     # Will stop when, twice in a row, the difference
>  >  diemac 2.0        # Although this is not mandatory, it is worth to
>  >
>  >
>  >  cl.in
>  >  acell 25 25 25    # The keyword "acell" refers to the
>  >  ntypat 1          # There is only one type of atom
>  >  znucl 17           # The keyword "znucl" refers to the atomic number
>  of the
>  >  natom 1           # There are two atoms
>  >  typat 1          # They both are of type 1, that is, Hydrogen
>  >  xcart             # This keyword indicates that the location of the
>  atoms
>  >    0.0 0.0 0.0    # Triplet giving the cartesian coordinates of atom
>  1, in Bohr
>  >  ecut 15.0         # Maximal plane-wave kinetic energy cut-off, in
>  Hartree
>  >  nkpt 1            # Only one k point is needed for isolated system,
>  >  nsppol 2
>  >  occopt 2
>  >  nband 8 8
>  >  occ 1 1 1 0 0 0 0 0 1 1 1 1 0 0 0 0
>  >  spinat 0.0 0.0 1.0  # Initialisation of spin
>  >  nstep 50          # Maximal number of SCF cycles
>  >  toldfe 1.0d-6     # Will stop when, twice in a row, the difference
>  >  diemac 2.0        # Although this is not mandatory, it is worth to
>  >
>  >
>  >  ticl.in
>  >  acell 25 25 25    # The keyword "acell" refers to the
>  >  ntypat 2          # There is only one type of atom
>  >  znucl 22 17          # The keyword "znucl" refers to the atomic
>  number of the
>  >  natom 2           # There are two atoms
>  >  typat 1 2         # They both are of type 1, that is, Hydrogen
>  >  xcart             # This keyword indicates that the location of the
>  atoms
>  >    0.0 0.0 0.0    # Triplet giving the cartesian coordinates of atom
>  1, in Bohr
>  >    4.3 0 0
>  >  ecut 30.0         # Maximal plane-wave kinetic energy cut-off, in
>  Hartree
>  >  nkpt 1            # Only one k point is needed for isolated system,
>  >  nsppol 2
>  >  occopt 2
>  >  nband 12 12
>  >  occ 1 1 1 1 1 1 1 0 0 0 0 0 1 1 1 1 0 0 0 0 0 0 0 0  spinat 0.0 0.0
>  > 3.0
>  >        0.0 0.0 1.0
>  >  ionmov 2
>  >  tolmxf 5.0d-4
>  >  ntime 20
>  >  nstep 50          # Maximal number of SCF cycles
>  >  toldfe 1.0d-6     # Will stop when, twice in a row, the difference
>  >  diemac 2.0        # Although this is not mandatory, it is worth to
>  >
>
>
>
>  --
>  Pierre-Matthieu Anglade
>
>



-- 
Pierre-Matthieu Anglade