Is it possible to study/converge magnetic surfaces? repost..

[Valerio Bellini]

Dear all,
I am trying to simulate with CP2K-Quickstep a magnetic surface, in particular
a slab of 4 layers of Ni(111).
I have not manage to converge the system. Here are my considerations:

-) the two dimensional cell is not large (5x5), and I tried hexagonal and rectangular
two-dimensiional cells. Since no k-points are considered in CP2K, I know that
this cell is rather small, and in order to have converged properties one should
use a larger one.
-) I used the mixing/diagonalization options suggested for metallic surfaces
(recent message from Marcella Iannuzzi on the 4th October).
-) I tried to vary the ALPHA,BETA,NBROYDEN parameters. I also tried to use
direct mixing+DIIS scheme. Things did not change much.
-) I managed to converge the system only with OT scheme, but although I converge
the system up to 10^-7, the results does not seem to me converged to the ground state,
since the magnetic moments of different atoms in the same layer are not the same,
while they should be due to symmetric reasons.

I made a step back and converged an isolated Ni(111) monolayer, and in that
case managed to converge with OT but also with standard diagonalization+broyden
mixing, and with the latter scheme I was able to reach always a converged system
in terms of equal magnetic moments for all the atoms.

I both case (ML and 4 layers slab) I set a multiplicity which is reasonable
considering the total moment in the cell that the system should acquire
(in case of the ML, I varied also the multiplicity and found the one giving the
ground state, while for the 4 layers slab I set it to a reasonable value,
could easily be it is not the ground state one).

My questions are:

1) Are there chances that increasing the two-dimensional cell dimensions,
convergence will be reached also for the slab?
Before trying I would like to have an opinion on that; in other words, could the dimension
of the two-dimensional cell be the responsible of the missing convergence with
diagonalization+broyden techniques?

2) Have anybody ever managed to converged a magnetic surface with this code?

Here is a typical input file I used:

&GLOBAL
ￜ PROJECTￜ ./working
ￜ RUN_TYPE ENERGY_FORCE
ￜ PRINT_LEVEL MEDIUM
&END GLOBAL
&FORCE_EVAL
ￜ METHOD Quickstep
&DFT
ￜￜￜ BASIS_SET_FILE_NAMEￜ ./BASIS_MOLOPT
ￜￜￜ POTENTIAL_FILE_NAMEￜ ./GTH_POTENTIALS
ￜￜￜ RESTART_FILE_NAME ./working-RESTART.wfn
ￜￜￜ LSD T
ￜￜￜ MULTIPLICITY 71
&MGRID
ￜￜￜￜￜ CUTOFF 500
ￜￜￜￜￜ NGRIDS 5
&END MGRID
&QS
ￜￜￜￜￜ EXTRAPOLATION PS
ￜￜￜￜￜ EXTRAPOLATION_ORDER 3
&END QS
&SCF
ￜￜￜￜￜ SCF_GUESS restart
ￜￜￜￜￜ EPS_SCF 1.0E-7
ￜￜￜￜￜ MAX_SCF 500
&OUTER_SCF ON
ￜￜￜￜￜￜￜ MAX_SCF 20
ￜￜￜￜￜￜￜ EPS_SCFￜ 1.0E-7
&END OUTER_SCF
ￜￜￜￜￜ ADDED_MOS 1000
&SMEAR ON
ￜￜￜￜￜￜￜ METHOD FERMI_DIRAC
ￜￜￜￜￜￜￜ ELECTRONIC_TEMPERATURE [K] 300
&END SMEAR
&DIAGONALIZATION ON
ￜￜￜￜￜￜￜ ALGORITHM STANDARD
&END DIAGONALIZATION
&MIXING ON
ￜￜￜￜￜￜￜ METHOD BROYDEN_MIXING
ￜￜￜￜￜￜￜ ALPHAￜￜ 0.05
ￜￜￜￜￜￜￜ BETAￜￜￜ 1.5
ￜￜￜￜￜￜￜ NBROYDENￜ 8
&END MIXING
&END SCF
&XC
&VDW_POTENTIAL
ￜￜￜￜￜￜￜ POTENTIAL_TYPE PAIR_POTENTIAL
&PAIR_POTENTIAL
ￜￜￜￜￜￜￜￜￜ REFERENCE_FUNCTIONAL PBE
ￜￜￜￜￜￜￜￜￜ TYPE DFTD3
ￜￜￜￜￜￜￜￜￜ PARAMETER_FILE_NAME ./dftd3.dat
&END PAIR_POTENTIAL
&END VDW_POTENTIAL
&XC_FUNCTIONAL
&PBE
&END PBE
&END XC_FUNCTIONAL
&XC_GRID
&END XC_GRID
&END XC
&END DFT
&SUBSYS
&CELL
ￜￜￜￜￜ PERIODIC XY
ￜￜￜￜￜ ABC 12.45416482 12.45416482 40
ￜￜￜￜￜ ANGLES 90 90 120
&END CELL
&COORD
Niￜￜￜￜ 0.000000ￜￜￜￜ 0.000000ￜￜￜￜ 7.797460
Niￜￜￜ -1.245416ￜￜￜￜ 2.157125ￜￜￜￜ 7.797460
Niￜￜￜ -2.490833ￜￜￜￜ 4.314249ￜￜￜￜ 7.797460
Niￜￜￜￜ 2.490833ￜￜￜ -4.314249ￜￜￜￜ 7.797460
Niￜￜￜￜ 1.245416ￜￜￜ -2.157125ￜￜￜￜ 7.797460
Niￜￜￜￜ 2.490833ￜￜￜￜ 0.000000ￜￜￜￜ 7.797460
Niￜￜￜￜ 1.245416ￜￜￜￜ 2.157125ￜￜￜￜ 7.797460
Niￜￜￜￜ 0.000000ￜￜￜￜ 4.314249ￜￜￜￜ 7.797460
Niￜￜￜￜ 4.981666ￜￜￜ -4.314249ￜￜￜￜ 7.797460
Niￜￜￜￜ 3.736249ￜￜￜ -2.157125ￜￜￜￜ 7.797460
Niￜￜￜￜ 4.981666ￜￜￜￜ 0.000000ￜￜￜￜ 7.797460
Niￜￜￜￜ 3.736249ￜￜￜￜ 2.157125ￜￜￜￜ 7.797460
Niￜￜￜￜ 2.490833ￜￜￜￜ 4.314249ￜￜￜￜ 7.797460
Niￜￜￜￜ 7.472499ￜￜￜ -4.314249ￜￜￜￜ 7.797460
Niￜￜￜￜ 6.227082ￜￜￜ -2.157125ￜￜￜￜ 7.797460
Niￜￜￜ -4.981666ￜￜￜￜ 0.000000ￜￜￜￜ 7.797460
Niￜￜￜ -6.227082ￜￜￜￜ 2.157125ￜￜￜￜ 7.797460
Niￜￜￜ -7.472499ￜￜￜￜ 4.314249ￜￜￜￜ 7.797460
Niￜￜￜ -2.490833ￜￜￜ -4.314249ￜￜￜￜ 7.797460
Niￜￜￜ -3.736249ￜￜￜ -2.157125ￜￜￜￜ 7.797460
Niￜￜￜ -2.490833ￜￜￜￜ 0.000000ￜￜￜￜ 7.797460
Niￜￜￜ -3.736249ￜￜￜￜ 2.157125ￜￜￜￜ 7.797460
Niￜￜￜ -4.981666ￜￜￜￜ 4.314249ￜￜￜￜ 7.797460
Niￜￜￜￜ 0.000000ￜￜￜ -4.314249ￜￜￜￜ 7.797460
Niￜￜￜ -1.245416ￜￜￜ -2.157125ￜￜￜￜ 7.797460
Niￜￜￜￜ 0.000000ￜￜￜￜ 1.438083ￜￜￜￜ 9.831216
Niￜￜￜ -1.245416ￜￜￜￜ 3.595208ￜￜￜￜ 9.831216
Niￜￜￜￜ 3.736250ￜￜￜ -5.033291ￜￜￜￜ 9.831216
Niￜￜￜￜ 2.490833ￜￜￜ -2.876166ￜￜￜￜ 9.831216
Niￜￜￜￜ 1.245417ￜￜￜ -0.719042ￜￜￜￜ 9.831216
Niￜￜￜￜ 2.490833ￜￜￜￜ 1.438083ￜￜￜￜ 9.831216
Niￜￜￜￜ 1.245417ￜￜￜￜ 3.595208ￜￜￜￜ 9.831216
Niￜￜￜￜ 6.227082ￜￜￜ -5.033291ￜￜￜￜ 9.831216
Niￜￜￜￜ 4.981666ￜￜￜ -2.876166ￜￜￜￜ 9.831216
Niￜￜￜￜ 3.736250ￜￜￜ -0.719042ￜￜￜￜ 9.831216
Niￜￜￜￜ 4.981666ￜￜￜￜ 1.438083ￜￜￜￜ 9.831216
Niￜￜￜￜ 3.736250ￜￜￜￜ 3.595208ￜￜￜￜ 9.831216
Niￜￜￜￜ 8.717915ￜￜￜ -5.033291ￜￜￜￜ 9.831216
Niￜￜￜￜ 7.472499ￜￜￜ -2.876166ￜￜￜￜ 9.831216
Niￜￜￜￜ 6.227082ￜￜￜ -0.719042ￜￜￜￜ 9.831216
Niￜￜￜ -4.981666ￜￜￜￜ 1.438083ￜￜￜￜ 9.831216
Niￜￜￜ -6.227082ￜￜￜￜ 3.595208ￜￜￜￜ 9.831216
Niￜￜￜ -1.245416ￜￜￜ -5.033291ￜￜￜￜ 9.831216
Niￜￜￜ -2.490833ￜￜￜ -2.876166ￜￜￜￜ 9.831216
Niￜￜￜ -3.736249ￜￜￜ -0.719042ￜￜￜￜ 9.831216
Niￜￜￜ -2.490833ￜￜￜￜ 1.438083ￜￜￜￜ 9.831216
Niￜￜￜ -3.736249ￜￜￜￜ 3.595208ￜￜￜￜ 9.831216
Niￜￜￜￜ 1.245417ￜￜￜ -5.033291ￜￜￜￜ 9.831216
Niￜￜￜￜ 0.000000ￜￜￜ -2.876166ￜￜￜￜ 9.831216
Niￜￜￜ -1.245416ￜￜￜ -0.719042ￜￜￜￜ 9.831216
Niￜￜￜￜ 1.245416ￜￜￜￜ 0.719042ￜￜￜ 11.864973
Niￜￜￜ -0.000000ￜￜￜￜ 2.876166ￜￜￜ 11.864973
Niￜￜￜ -1.245417ￜￜￜￜ 5.033291ￜￜￜ 11.864973
Niￜￜￜￜ 3.736249ￜￜￜ -3.595208ￜￜￜ 11.864973
Niￜￜￜￜ 2.490833ￜￜￜ -1.438083ￜￜￜ 11.864973
Niￜￜￜￜ 3.736249ￜￜￜￜ 0.719042ￜￜￜ 11.864973
Niￜￜￜￜ 2.490833ￜￜￜￜ 2.876166ￜￜￜ 11.864973
Niￜￜￜￜ 1.245416ￜￜￜￜ 5.033291ￜￜￜ 11.864973
Niￜￜￜￜ 6.227082ￜￜￜ -3.595208ￜￜￜ 11.864973
Niￜￜￜￜ 4.981666ￜￜￜ -1.438083ￜￜￜ 11.864973
Niￜￜￜ -6.227082ￜￜￜￜ 0.719042ￜￜￜ 11.864973
Niￜￜￜ -7.472499ￜￜￜￜ 2.876166ￜￜￜ 11.864973
Niￜￜￜ -8.717915ￜￜￜￜ 5.033291ￜￜￜ 11.864973
Niￜￜￜ -3.736250ￜￜￜ -3.595208ￜￜￜ 11.864973
Niￜￜￜ -4.981666ￜￜￜ -1.438083ￜￜￜ 11.864973
Niￜￜￜ -3.736250ￜￜￜￜ 0.719042ￜￜￜ 11.864973
Niￜￜￜ -4.981666ￜￜￜￜ 2.876166ￜￜￜ 11.864973
Niￜￜￜ -6.227082ￜￜￜￜ 5.033291ￜￜￜ 11.864973
Niￜￜￜ -1.245417ￜￜￜ -3.595208ￜￜￜ 11.864973
Niￜￜￜ -2.490833ￜￜￜ -1.438083ￜￜￜ 11.864973
Niￜￜￜ -1.245417ￜￜￜￜ 0.719042ￜￜￜ 11.864973
Niￜￜￜ -2.490833ￜￜￜￜ 2.876166ￜￜￜ 11.864973
Niￜￜￜ -3.736250ￜￜￜￜ 5.033291ￜￜￜ 11.864973
Niￜￜￜￜ 1.245416ￜￜￜ -3.595208ￜￜￜ 11.864973
Niￜￜￜ -0.000000ￜￜￜ -1.438083ￜￜￜ 11.864973
Niￜￜￜￜ 0.000000ￜￜￜￜ 0.000000ￜￜￜ 13.898730
Niￜￜￜ -1.245416ￜￜￜￜ 2.157125ￜￜￜ 13.898730
Niￜￜￜ -2.490833ￜￜￜￜ 4.314249ￜￜￜ 13.898730
Niￜￜￜￜ 2.490833ￜￜￜ -4.314249ￜￜￜ 13.898730
Niￜￜￜￜ 1.245416ￜￜￜ -2.157125ￜￜￜ 13.898730
Niￜￜￜￜ 2.490833ￜￜￜￜ 0.000000ￜￜￜ 13.898730
Niￜￜￜￜ 1.245416ￜￜￜￜ 2.157125ￜￜￜ 13.898730
Niￜￜￜￜ 0.000000ￜￜￜￜ 4.314249ￜￜￜ 13.898730
Niￜￜￜￜ 4.981666ￜￜￜ -4.314249ￜￜￜ 13.898730
Niￜￜￜￜ 3.736249ￜￜￜ -2.157125ￜￜￜ 13.898730
Niￜￜￜￜ 4.981666ￜￜￜￜ 0.000000ￜￜￜ 13.898730
Niￜￜￜￜ 3.736249ￜￜￜￜ 2.157125ￜￜￜ 13.898730
Niￜￜￜￜ 2.490833ￜￜￜￜ 4.314249ￜￜￜ 13.898730
Niￜￜￜￜ 7.472499ￜￜￜ -4.314249ￜￜￜ 13.898730
Niￜￜￜￜ 6.227082ￜￜￜ -2.157125ￜￜￜ 13.898730
Niￜￜￜ -4.981666ￜￜￜￜ 0.000000ￜￜￜ 13.898730
Niￜￜￜ -6.227082ￜￜￜￜ 2.157125ￜￜￜ 13.898730
Niￜￜￜ -7.472499ￜￜￜￜ 4.314249ￜￜￜ 13.898730
Niￜￜￜ -2.490833ￜￜￜ -4.314249ￜￜￜ 13.898730
Niￜￜￜ -3.736249ￜￜￜ -2.157125ￜￜￜ 13.898730
Niￜￜￜ -2.490833ￜￜￜￜ 0.000000ￜￜￜ 13.898730
Niￜￜￜ -3.736249ￜￜￜￜ 2.157125ￜￜￜ 13.898730
Niￜￜￜ -4.981666ￜￜￜￜ 4.314249ￜￜￜ 13.898730
Niￜￜￜￜ 0.000000ￜￜￜ -4.314249ￜￜￜ 13.898730
Niￜￜￜ -1.245416ￜￜￜ -2.157125ￜￜￜ 13.898730
&END COORD
&KIND Ni
ￜￜￜￜￜ POTENTIAL GTH-PBE-q18
ￜￜￜￜￜ BASIS_SET DZVP-MOLOPT-SR-GTH
&END KIND
&END SUBSYS
&END FORCE_EVAL

[marci]

Dear Valerio,

I tried your system, 5x5 Ni(111) slab, and I could converge the
electronic structure by using more or less the same settings that were
in your input.
It needs many iterations and the energy keeps oscillating for a long
time before the algorithm can find a good minimum.
However, what is really annoying, is that at the end the electrons are
redistributed between the two spins in such a way that the final
magnetization is zero, in spite of the fact that the initial guess had
a high multiplicity.
It seems that with the present settings and system size, the algorithm
finds a minimum with no magnetization, and this should be also the
reason why starting from a magnetization different from zero it takes
such a long time to converge.
It is possible that one problem is the size of the system. One should
check larger boxes to verify that.
What I can tell for sure is that the optimization of the bulk (216
atoms) electronic structure, gives the expected magnetization (~0.6
magneton per unit cell) by using more or less the same SCF set up.

best
marcella


On Oct 24, 5:37 pm, Valerio Bellini <valerio.bell...@unimore.it>
wrote:

>    PROJECT  ./working
>    RUN_TYPE ENERGY_FORCE

>    PRINT_LEVEL MEDIUM
> &END GLOBAL
> &FORCE_EVAL

>    METHOD Quickstep
> &DFT
>      BASIS_SET_FILE_NAME  ./BASIS_MOLOPT
>      POTENTIAL_FILE_NAME  ./GTH_POTENTIALS
>      RESTART_FILE_NAME ./working-RESTART.wfn
>      LSD T
>      MULTIPLICITY 71
> &MGRID
>        CUTOFF 500

>        NGRIDS 5
> &END MGRID
> &QS

>        EXTRAPOLATION PS

>        EXTRAPOLATION_ORDER 3
> &END QS
> &SCF

>        SCF_GUESS restart
>        EPS_SCF 1.0E-7
>        MAX_SCF 500
> &OUTER_SCF ON
>          MAX_SCF 20
>          EPS_SCF  1.0E-7
> &END OUTER_SCF
>        ADDED_MOS 1000
> &SMEAR ON
>          METHOD FERMI_DIRAC

>          ELECTRONIC_TEMPERATURE [K] 300
> &END SMEAR
> &DIAGONALIZATION ON

>          ALGORITHM STANDARD
> &END DIAGONALIZATION
> &MIXING ON

>          METHOD BROYDEN_MIXING
>          ALPHA   0.05
>          BETA    1.5
>          NBROYDEN  8

> &END MIXING
> &END SCF
> &XC
> &VDW_POTENTIAL

>          POTENTIAL_TYPE PAIR_POTENTIAL
> &PAIR_POTENTIAL
>            REFERENCE_FUNCTIONAL PBE
>            TYPE DFTD3

>            PARAMETER_FILE_NAME ./dftd3.dat
> &END PAIR_POTENTIAL
> &END VDW_POTENTIAL
> &XC_FUNCTIONAL
> &PBE
> &END PBE
> &END XC_FUNCTIONAL
> &XC_GRID
> &END XC_GRID
> &END XC
> &END DFT
> &SUBSYS
> &CELL

>        PERIODIC XY
>        ABC 12.45416482 12.45416482 40

>        ANGLES 90 90 120
> &END CELL
> &COORD

> Ni     0.000000     0.000000     7.797460
> Ni    -1.245416     2.157125     7.797460
> Ni    -2.490833     4.314249     7.797460
> Ni     2.490833    -4.314249     7.797460
> Ni     1.245416    -2.157125     7.797460
> Ni     2.490833     0.000000     7.797460
> Ni     1.245416     2.157125     7.797460
> Ni     0.000000     4.314249     7.797460
> Ni     4.981666    -4.314249     7.797460
> Ni     3.736249    -2.157125     7.797460
> Ni     4.981666     0.000000     7.797460
> Ni     3.736249     2.157125     7.797460
> Ni     2.490833     4.314249     7.797460
> Ni     7.472499    -4.314249     7.797460
> Ni     6.227082    -2.157125     7.797460
> Ni    -4.981666     0.000000     7.797460
> Ni    -6.227082     2.157125     7.797460
> Ni    -7.472499     4.314249     7.797460
> Ni    -2.490833    -4.314249     7.797460
> Ni    -3.736249    -2.157125     7.797460
> Ni    -2.490833     0.000000     7.797460
> Ni    -3.736249     2.157125     7.797460
> Ni    -4.981666     4.314249     7.797460
> Ni     0.000000    -4.314249     7.797460
> Ni    -1.245416    -2.157125     7.797460
> Ni     0.000000     1.438083     9.831216
> Ni    -1.245416     3.595208     9.831216
> Ni     3.736250    -5.033291     9.831216
> Ni     2.490833    -2.876166     9.831216
> Ni     1.245417    -0.719042     9.831216
> Ni     2.490833     1.438083     9.831216
> Ni     1.245417     3.595208     9.831216
> Ni     6.227082    -5.033291     9.831216
> Ni     4.981666    -2.876166     9.831216
> Ni     3.736250    -0.719042     9.831216
> Ni     4.981666     1.438083     9.831216
> Ni     3.736250     3.595208     9.831216
> Ni     8.717915    -5.033291     9.831216
> Ni     7.472499    -2.876166     9.831216
> Ni     6.227082    -0.719042     9.831216
> Ni    -4.981666     1.438083     9.831216
> Ni    -6.227082     3.595208     9.831216
> Ni    -1.245416    -5.033291     9.831216
> Ni    -2.490833    -2.876166     9.831216
> Ni    -3.736249    -0.719042     9.831216
> Ni    -2.490833     1.438083     9.831216
> Ni    -3.736249     3.595208     9.831216
> Ni     1.245417    -5.033291     9.831216
> Ni     0.000000    -2.876166     9.831216
> Ni    -1.245416    -0.719042     9.831216
> Ni     1.245416     0.719042    11.864973
> Ni    -0.000000     2.876166    11.864973
> Ni    -1.245417     5.033291    11.864973
> Ni     3.736249    -3.595208    11.864973
> Ni     2.490833    -1.438083    11.864973
> Ni     3.736249     0.719042    11.864973
> Ni     2.490833     2.876166    11.864973
> Ni     1.245416     5.033291    11.864973
> Ni     6.227082    -3.595208    11.864973
> Ni     4.981666    -1.438083    11.864973
> Ni    -6.227082     0.719042    11.864973
> Ni    -7.472499     2.876166    11.864973
> Ni    -8.717915     5.033291    11.864973
> Ni    -3.736250    -3.595208    11.864973
> Ni    -4.981666    -1.438083    11.864973
> Ni    -3.736250     0.719042    11.864973
> Ni    -4.981666     2.876166    11.864973
> Ni    -6.227082     5.033291    11.864973
> Ni    -1.245417    -3.595208    11.864973
> Ni    -2.490833    -1.438083    11.864973
> Ni    -1.245417     0.719042    11.864973
> Ni    -2.490833     2.876166    11.864973
> Ni    -3.736250     5.033291    11.864973
> Ni     1.245416    -3.595208    11.864973
> Ni    -0.000000    -1.438083    11.864973
> Ni     0.000000     0.000000    13.898730
> Ni    -1.245416     2.157125    13.898730
> Ni    -2.490833     4.314249    13.898730
> Ni     2.490833    -4.314249    13.898730
> Ni     1.245416    -2.157125    13.898730
> Ni     2.490833     0.000000    13.898730
> Ni     1.245416     2.157125    13.898730
> Ni     0.000000     4.314249    13.898730
> Ni     4.981666    -4.314249    13.898730
> Ni     3.736249    -2.157125    13.898730
> Ni     4.981666     0.000000    13.898730
> Ni     3.736249     2.157125    13.898730
> Ni     2.490833     4.314249    13.898730
> Ni     7.472499    -4.314249    13.898730
> Ni     6.227082    -2.157125    13.898730
> Ni    -4.981666     0.000000    13.898730
> Ni    -6.227082     2.157125    13.898730
> Ni    -7.472499     4.314249    13.898730
> Ni    -2.490833    -4.314249    13.898730
> Ni    -3.736249    -2.157125    13.898730
> Ni    -2.490833     0.000000    13.898730
> Ni    -3.736249     2.157125    13.898730
> Ni    -4.981666     4.314249    13.898730
> Ni     0.000000    -4.314249    13.898730
> Ni    -1.245416    -2.157125    13.898730
> &END COORD
> &KIND Ni
>        POTENTIAL GTH-PBE-q18

[Valerio Bellini]

Il 26/10/11 15.30, marci ha scritto:

> Dear Valerio,
>
> I tried your system, 5x5 Ni(111) slab, and I could converge the
> electronic structure by using more or less the same settings that were
> in your input.
> It needs many iterations and the energy keeps oscillating for a long
> time before the algorithm can find a good minimum.
> However, what is really annoying, is that at the end the electrons are
> redistributed between the two spins in such a way that the final
> magnetization is zero, in spite of the fact that the initial guess had
> a high multiplicity.
> It seems that with the present settings and system size, the algorithm
> finds a minimum with no magnetization, and this should be also the
> reason why starting from a magnetization different from zero it takes
> such a long time to converge.
> It is possible that one problem is the size of the system. One should
> check larger boxes to verify that.
> What I can tell for sure is that the optimization of the bulk (216
> atoms) electronic structure, gives the expected magnetization (~0.6
> magneton per unit cell) by using more or less the same SCF set up.
>
> best
> marcella

Dear Marcella,
Thank you for the answer.
Two comments:
1) I did calculation for the same system, using Gamma point only, with
another code (VASP),
and the total magnetic moment of the cell relaxed to around 86 bohr
magneton.
With a better multiplicity guess I thought convergence might be easier,
but it was
not the case.
If I try 87 as multiplicity and I run more than 500 iterations with
Broyden,
the system converges using Diagonalization+Broyden up to 0.005 Hartree,
but as said in the previous e-mail, the magnetic moments are not equal
for different atoms
in the same plane, so the system in reality is far from convergence.
Could I ask you how many iterations did it take to you?
2) I do not understand how you could get a non-magnetic solution.
If you impose the multiplicity to some value, and you don't allow
relaxation of it
(using the keyword, RELAX_MULTIPLICITY) the multiplicity of the system
should remain
constant (like in a fixed spin moment calculation).
So this means that you inserted that flag in the input file, is that
correct?
thanks,
Valerio

[marci]

Dear Valerio

I did a second test with a 6x6 slab and 6 layers, and by initializing
the multiplicity to 131,
I get a reasonable distribution of spins. Atoms belonging to the same
layer have all the same spin moment, the outermost layer have the
largest spin moment 0.66, and the two innermost layer have the lowest
spin moment 0.56.
It takes quite a number of iterations (about 140) to converge with a
convergence criterion of 1E-7.
I used a mixing parameter of 0.08 and an electronic temperature of
2000K.
The RELAX_MULTIPLICITY keyword is not needed when the smearing is
used.
The occupation of the states is attributed according to the Fermi-
Dirac distribution, through the evaluation of the Fermi energy at each
SCF step.
This means that the fractional occupation numbers can change, as well
as the number of electrons per spin channel, only the total number of
electrons is constant.
In the specific case, I monitored the number of electrons per spin
channel, and it remains quite stable from the beginning to the end,
though there are some fluctuations, in particular during the first
iterations.
The final difference between spin up and spin down is of 131.8
electrons.

best
Marcella

On Oct 26, 4:08 pm, Valerio Bellini <valerio.bell...@unimore.it>
wrote:

[marci]

On Oct 26, 4:08 pm, Valerio Bellini <valerio.bell...@unimore.it>
wrote:

[Dr. Roman Leitsmann]

Dear Marcella,

I have some (may be stupid) questions about your calculation:

(1) You are using additional MOs and a certain smearing, right?
However, in the case of fractional charges the computation of the spin
moment is not yet proper working in cp2k (WARNING: S** computation does
not yet treat fractional occupied orbitals). So my question is: How have
you calculated the spin moment.

(2) How have you calculated the difference between spin up and spin down
channel. In my version of cp2k only integer numbers of electrons are
given for each spin channel.

best
Roman

---------------------------------------
AQcomputare GmbH
Annaberger Strasse 240

09125 Chemnitz

Telefon: 0371 5347591
Email: in...@matcalc.de
Internet: http://www.matcalc.de

Geschaeftsfuehrer: Dr. Philipp Plaenitz

Gesellschafter:

Prof. Dr. Christian Radehaus
Dr. Philipp Plaenitz

Amtsgericht Chemnitz: HRB 25386

Diese e-mail kann Betriebs- oder Geschaeftsgeheimnisse oder sonstige
vertrauliche Informationen enthalten. Sollten Sie diese e-mail irrtuemlich
erhalten haben, ist Ihnen eine Kenntnisnahme des Inhalts, eine
Vervielfaeltigung oder Weitergabe der e-mail ausdruecklich untersagt. Bitte
benachrichtigen Sie uns und vernichten Sie die empfangene e-mail. Vielen Dank.

Important Note: This e-mail may contain trade secrets or privileged,
undisclosed or otherwise confidential information. If you have received this
e-mail in error, you are hereby notified that any review, copying or
distribution of it is strictly prohibited. Please inform us immediately and
destroy the original transmittal. Thank you for your cooperation.

[Valerio Bellini]

Il 28/10/11 12.42, Dr. Roman Leitsmann ha scritto:

> Dear Marcella,
>
> I have some (may be stupid) questions about your calculation:
>
> (1) You are using additional MOs and a certain smearing, right?
> However, in the case of fractional charges the computation of the spin
> moment is not yet proper working in cp2k (WARNING: S** computation
> does not yet treat fractional occupied orbitals). So my question is:
> How have you calculated the spin moment.
>
> (2) How have you calculated the difference between spin up and spin
> down channel. In my version of cp2k only integer numbers of electrons
> are given for each spin channel.
>
> best
> Roman

Dear Roman and Marcella,

I had the same (maybe stupid) questions myself! ;-)
in the line '# Total charge and spin" there is always an integer total
spin moment.
maybe that is not where one should look, is it?

Secondly .. 140 iterations.. WOW!!
are there some flag that one should put in the .inp file we should be
aware of?
And more importantly, are the some general rule (apart from experience) one
can rely on, in the choice of the mixing/number of additional MOs/Fermi
Dirac temperature?

Thanks again for the preciuos help,
Valerio

[marci]

Dear Roman,

For a metallic system it is recommended to use additional MOS and
Fermi-Dirac smearing.
By doing so within spin-polarized calculations, the total number of
electrons per spin channel is not constrained to be an integer number,
because the distribution of the occupation numbers is done on the
basis of the energy level of each state with respect to the Fermi
energy, irrespective of its spin.
The total number of electrons per spin is simply the integral of the
charge density associated to that spin.
Estimates of the momentum per atom can be obtained by the population
analysis

Best
Marcella

On Oct 28, 12:42 pm, "Dr. Roman Leitsmann" <leitsm...@matcalc.de>
wrote:

> Email:    i...@matcalc.de

[marci]

Dear Valerio,

If you re speaking of the '# Total charge and spin" given at the end
of the population analysis, it should not give always integer numbers
if the smearing in activated and the calculation is spin polarized.
If this is not the case, it might be a problem of the version of the
code. The correct smearing with LSD was implemented early this year.

The rules to choose the parameters are the same as with other codes
used with metallic systems.
Consider that the default electronic temperature in cp2k is quite low
(300K) if compared to typical values used with other codes.
But you can always increase it from input, and this should help in
speeding the convergence up.

ciao
Marcella




On Oct 28, 12:53 pm, Valerio Bellini <valerio.bell...@unimore.it>
wrote:

[Dr. Roman Leitsmann]

Dear Marcella,

I am a little bit confused.
I understand your point and I would expect the same behavior.
However, does this mean that the following lines in the output are not
reliable in the case of added MOs and applied smearing?

For example:
...

Spin 1

Number of electrons: 602
Number of occupied orbitals: 627
Number of molecular orbitals: 702

Spin 2

Number of electrons: 598
Number of occupied orbitals: 623
Number of molecular orbitals: 698

...

Because at this point always integer numbers are given ...

The other question is how reliable is the population analysis in case of
fractional occupied orbitals? Because the total spin moment given in the
population analysis is sometimes not in agreement with the above output.

best,
Roman

09125 Chemnitz

Telefon: 0371 5347591
Email: in...@matcalc.de

[marcella Iannuzzi]

These are the numbers given before the calculation starts
the number of MOS is obviously unchanged
the total numbe of electrons also,
the number of occupied orbitals and number of electrons per spin are
based on the multiplicity given from input,
This is used for the initial guess (atomic guess and not for a
restart), and no more afterwords

marcella

> --
> You received this message because you are subscribed to the Google
> Groups "cp2k" group.
> To post to this group, send email to cp...@googlegroups.com.
> To unsubscribe from this group, send email to cp2k+uns...@googlegroups.com
> .
> For more options, visit this group at http://groups.google.com/group/cp2k?hl=en
> .
>

[Valerio Bellini]

Il 28/10/11 13.16, marci ha scritto:

> Dear Valerio,
>
> If you re speaking of the '# Total charge and spin" given at the end
> of the population analysis, it should not give always integer numbers
> if the smearing in activated and the calculation is spin polarized.
> If this is not the case, it might be a problem of the version of the
> code. The correct smearing with LSD was implemented early this year.
>
> The rules to choose the parameters are the same as with other codes
> used with metallic systems.
> Consider that the default electronic temperature in cp2k is quite low
> (300K) if compared to typical values used with other codes.
> But you can always increase it from input, and this should help in
> speeding the convergence up.
>
> ciao
> Marcella
>

The version I am using is the one downloaded the 23th of February..
Could be it was prior to the modifications you are pointing to.
I will download the newest version,
Thank you very much,
Valerio