Top Page > Browsing
MD convergence bilayer
Date: 2024/11/05 08:24
Name: Lillian

Dear Contributors,
I am having trouble of the MD convergence during geometry optimization. The scf process converged well, and the MD keeps reducing, however, it is still not converged after more than >800 iterations even after I reduce the criterion to 10^-4. Additionally, if I run from the restart file .dat#, it finishes only at the first MD iteration but it doesn't reach the convergence. My system is bilayer, and below I put the controllable parameters.

#
# SCF or Electronic System
#

scf.XcType                GGA-PBE    # LDA|LSDA-CA|LSDA-PW|GGA-PBE
scf.SpinPolarization        off        # On|Off|NC
scf.ElectronicTemperature  300.0      # default=300 (K)
scf.energycutoff          250    # default=150 (Ry)
#scf.Ngrid                8 8 1
scf.maxIter                100        # default=40
scf.EigenvalueSolver      Band      # DC|GDC|Cluster|Band
scf.Kgrid                16 8 1    # means n1 x n2 x n3
scf.ProExpn.VNA              off      # default=on
scf.Mixing.Type          rmm-diisk    # Simple|Rmm-Diis|Gr-Pulay|Kerker|Rmm-Diisk
scf.Init.Mixing.Weight    0.0010
scf.Min.Mixing.Weight    0.0001
scf.Max.Mixing.Weight    0.3000
scf.Mixing.History        60
scf.Mixing.StartPulay    80
scf.Mixing.EveryPulay    1
scf.criterion            1.0e-9      # default=1.0e-6 (Hartree)

MD.Type                    EF        # Nomd|Opt|DIIS|NVE|NVT_VS|NVT_NH
MD.Opt.DIIS.History          15        # default=7
MD.Opt.StartDIIS            20        # default=5
MD.maxIter                  1000        # default=1
MD.Opt.criterion            1.0e-4    # default=1.0e-4(Hartree/Bohr)


<MD.Fixed.Cell.Vectors
    1 1 1
    1 1 1
    1 1 1
MD.Fixed.Cell.Vectors>

<MD.Fixed.XYZ
1  0 0 0
2  0 0 0
3  0 0 0
4  0 0 0
5  0 0 0
6  0 0 0
7  0 0 0
8  0 0 0
9  0 0 0
10  0 0 0
11 0 0 0
12 0 0 0
13 0 0 0
14 0 0 0
15 0 0 0
16 0 0 0
17 0 0 0
18 0 0 0
19 0 0 0
20 0 0 0
21 0 0 0
22 0 0 0
23 0 0 0
24 0 0 0
MD.Fixed.XYZ>

scf.dftD                    on            # on|off, default=off
version.dftD                  3            # 2|3, default=2
DFTD3.damp                  bj            # zero|bj, default=bj
DFTD.Unit                    AU            # Ang|AU
DFTD.rcut_dftD            100.0            # default=100 (DFTD.Unit)
DFTD.cncut_dftD              40            # default=40 (DFTD.Unit)
DFTD.IntDirection        1 1 1            # default=1 1 1 (1:on 0:off)


<DFTD.periodicity
1 1
2 1
3 1
4 1
5 1
6 1
7 1
8 1
9 1
10 1
11 1
12 1
13 1
14 1
15 1
16 1
17 1
18 1
19 1
20 1
21 1
22 1
23 1
24 1
DFTD.periodicity>

I am wondering if you could suggest me about the issue.

Thank you very much for your assistance.

Regards,
Lillian


メンテ
Page: [1]

Re: MD convergence bilayer ( No.1 )
Date: 2024/11/05 12:45
Name: Lillian

Dear Contributors,
I forget to mention that the forces rise up again to 0.1 Hartree/Bohr after reaching 0.002 Hartree/Bohr.

I am looking forward to your response.

Regards,
Lillian
メンテ
Re: MD convergence bilayer ( No.2 )
Date: 2024/11/05 22:38
Name: Naoya Yamagcuhi

Hi,

I can't say for sure since the details of the structure are not clear, but I think you should delete the following line.

```
scf.ProExpn.VNA              off      # default=on
```

>Additionally, if I run from the restart file .dat#, it finishes only at the first MD iteration but it doesn't reach the convergence.

If the number of MD iterations is more than the 1000 you set, you need to set `MD.maxIter` to 2000, for example, because the calculation will end immediately after one calculation.

Regards,
Naoya Yamaguchi
メンテ
Re: MD convergence bilayer ( No.3 )
Date: 2024/11/06 10:50
Name: Lillian

Dear Yamaguchi-sensei,
Thank you very much for your response. I have deleted the line as you suggested and it does faster the calculation. It seems like my system needs larger iteration number as well.

Thank you again.

Regards,
Lillian
メンテ
Re: MD convergence bilayer ( No.4 )
Date: 2024/11/06 14:16
Name: Lillian

Dear Yamaguchi-sensei,
May I know why does the scf.ProExpn.VNA affect the calculation performance even though it is already off ?

Regards,
Lillian
メンテ
Re: MD convergence bilayer ( No.5 )
Date: 2024/11/06 23:15
Name: Naoya Yamaguchi

Dear Lillian,

>May I know why does the scf.ProExpn.VNA affect the calculation performance even though it is already off ?

The default value is `On`, and if it is `On`, Projector expansions are used instead of grids to describe neutral atom potentials.

Regards,
Naoya Yamaguchi
メンテ
Re: MD convergence bilayer ( No.6 )
Date: 2024/12/04 06:27
Name: Lillian

Dear Yamaguchi-sensei,
Thank you for your previous responses. In equilibrium situation, my problem was solved. However,  the same issue appears when the external field is applied. What would be the cause this time possibly?

Below I put my input for your consideration :

# Definition of Atomic Species
#

Species.Number      3
<Definition.of.Atomic.Species
Ge      Ge7.0-s3p3d2f1  Ge_PBE19
Si      Si7.0-s3p3d2f1  Si_PBE19
As      As7.0-s3p3d2f1  As_PBE19
Definition.of.Atomic.Species>

#
# Atoms
#

Atoms.Number        24
Atoms.SpeciesAndCoordinates.Unit  Ang # Ang|AU
<Atoms.SpeciesAndCoordinates
    1    Si    -1.9426694    -4.1786335    21.8048354    2.00000  2.00000
    2    Si    -1.8847293    -9.5018187    19.1025527    2.00000  2.00000
    3    Si    -3.8600965    -1.9197752    20.0792489    2.00000  2.00000
    4    Si    -3.8354637    -7.2664345    20.9116168    2.00000  2.00000
    5    As    -3.8236403    -0.1393362    18.4070280    7.50000  7.50000
    6    As    -3.8832304    -5.4923794    22.5490002    7.50000  7.50000
    7    As    -3.9066058    -1.0136356    22.3438229    7.50000  7.50000
    8    As    -3.8388972    -6.1873678    18.7788517    7.50000  7.50000
    9    As    -1.9176616    -3.3535503    19.5373797    7.50000  7.50000
    10    As    -1.8867079    -8.7209752    21.3715671    7.50000  7.50000
    11    As    -1.9686806    -2.3486999    23.3831504    7.50000  7.50000
    12    As    -1.8828501    -7.4906262    17.7165035    7.50000  7.50000
    13    Ge    0.0226252    4.4587250    14.5594221    2.00000  2.00000
    14    Ge    0.0451486    9.9051419    11.7233435    2.00000  2.00000
    15    Ge    1.9433948    2.3046777    12.4877140    2.00000  2.00000
    16    Ge    1.9915764    7.6225779    13.6955271    2.00000  2.00000
    17    As    1.9781712    0.4856185    10.7812216    7.50000  7.50000
    18    As    1.9626248    5.8048832    15.4520543    7.50000  7.50000
    19    As    1.9767527    1.0380603    14.6449388    7.50000  7.50000
    20    As    2.0238983    6.6242762    11.3775821    7.50000  7.50000
    21    As    -0.0085508    3.8720680    12.1331183    7.50000  7.50000
    22    As    0.0493511    9.1824913    14.1188743    7.50000  7.50000
    23    As    0.0335682    2.2458642    15.7872017    7.50000  7.50000
    24    As    0.0890822    7.8882694    10.2745107    7.50000  7.50000
Atoms.SpeciesAndCoordinates>

Atoms.UnitVectors.Unit            Ang # Ang|AU
<Atoms.UnitVectors
3.860  0    0
0    10.454  0
0    0        25
Atoms.UnitVectors>

#
# SCF or Electronic System
#

scf.XcType                GGA-PBE    # LDA|LSDA-CA|LSDA-PW|GGA-PBE
scf.SpinPolarization        off        # On|Off|NC
scf.ElectronicTemperature  300.0      # default=300 (K)
scf.energycutoff          250      # default=150 (Ry)
#scf.Ngrid                8 8 1
scf.maxIter                200        # default=40
scf.EigenvalueSolver      Band      # DC|GDC|Cluster|Band
scf.Kgrid                16 8 1    # means n1 x n2 x n3
scf.Mixing.Type          rmm-diisk    # Simple|Rmm-Diis|Gr-Pulay|Kerker|Rmm-Diisk
scf.Init.Mixing.Weight    0.0010
scf.Min.Mixing.Weight    0.0001
scf.Max.Mixing.Weight    0.3000
scf.Mixing.History    60
scf.Mixing.StartPulay    80
scf.Mixing.EveryPulay    1
scf.criterion            1.0e-9      # default=1.0e-6 (Hartree)
scf.Electric.Field  0.0 1.5 0.0  # default=0.0 0.0 0.0 (GV/m)

MD.Type                    EF        # Nomd|Opt|DIIS|NVE|NVT_VS|NVT_NH
MD.Opt.DIIS.History          15        # default=7
MD.Opt.StartDIIS            30        # default=5
MD.maxIter                  4000        # default=1
MD.Opt.criterion            1.0e-5    # default=1.0e-4(Hartree/Bohr)


<MD.Fixed.Cell.Vectors
    1 1 1
    1 1 1
    1 1 1
MD.Fixed.Cell.Vectors>

<MD.Fixed.XYZ
1  0 0 1
2  0 0 1
3  0 0 1
4  0 0 1
5  0 0 1
6  0 0 1
7  0 0 1
8  0 0 1
9  0 0 1
10  0 0 1
11 0 0 1
12 0 0 1
13 0 0 1
14 0 0 1
15 0 0 1
16 0 0 1
17 0 0 1
18 0 0 1
19 0 0 1
20 0 0 1
21 0 0 1
22 0 0 1
23 0 0 1
24 0 0 1

scf.dftD                    on            # on|off, default=off
version.dftD                  3            # 2|3, default=2
DFTD3.damp                  bj            # zero|bj, default=bj
DFTD.Unit                    AU            # Ang|AU
DFTD.rcut_dftD            100.0            # default=100 (DFTD.Unit)
DFTD.cncut_dftD              40            # default=40 (DFTD.Unit)
DFTD.IntDirection        1 1 1            # default=1 1 1 (1:on 0:off)


<DFTD.periodicity
1 1
2 1
3 1
4 1
5 1
6 1
7 1
8 1
9 1
10 1
11 1
12 1
13 1
14 1
15 1
16 1
17 1
18 1
19 1
20 1
21 1
22 1
23 1
24 1
DFTD.periodicity>

I wish you can help me with this matter.

Regards,
Lillian
メンテ
Re: MD convergence bilayer ( No.7 )
Date: 2024/12/05 14:22
Name: Naoya Yamaguchi

Dear Lillian,

`scf.Electric.Field` is not compatible with MD or structural optimization. Instead, you can use the ESM for the purpose:
https://www.openmx-square.org/openmx_man3.9/node150.html

Regards,
Naoya Yamaguchi
メンテ

Page: [1]

Thread Title (must) Move the thread to the top
Your Name (must)
E-Mail (must)
URL
Password (used in modification of the submitted text)
Comment (must)

   Save Cookie