Convergence issue from kerker

Top Page > Browsing

Convergence issue from kerker_mixing

Date: 2023/08/21 13:05
Name: Sam <samuel.dechamps@cea.fr>: Dear developers,

I've faced the following issue with OpenMX 3.9 some months ago. It may have been fixed by now, though I didn't see any thread related to it so far.

When using the following SCF parameters:

scf.XcType GGA-PBE # LDA|LSDA-CA|LSDA-PW|GGA-PBE
scf.SpinPolarization off # On|Off|NC
scf.SpinOrbit.Coupling off
scf.ElectronicTemperature 300.0 # default=300 (K)
scf.energycutoff 300.0 # default=150 (Ry)
scf.maxIter 800 # default=40
scf.EigenvalueSolver Band # DC|GDC|Cluster|Band
scf.Kgrid 1 1 1 # means n1 x n2 x n3
scf.Mixing.Type rmm-diisk # Simple|Rmm-Diis|Gr-Pulay|Kerker|Rmm-Diisk
scf.Init.Mixing.Weight 0.001 # default=0.30
scf.Min.Mixing.Weight 0.001 # default=0.001
scf.Max.Mixing.Weight 0.100 # default=0.40
scf.Mixing.History 30 # default=5
scf.Mixing.StartPulay 20 # default=6
scf.Mixing.EveryPulay 1
#scf.Kerker.factor 1
scf.criterion 1.0e-7 # default=1.0e-6 (Hartree)
scf.lapack.dste dstevx # dstevx|dstedc|dstegr,default=dstevx
scf.stress.tensor off # on|off, default=off

I obtain the following error:

=== backtrace (tid: 128753) ====
0 0x0000000000a830db Kerker_Mixing_Rhok_Normal.A() Kerker_Mixing_Rhok.c:0
1 0x0000000000522ccf Mixing_DM() ???:0
2 0x000000000050fabd DFT.A() DFT.c:0
3 0x000000000040dded main.A() openmx.c:0
4 0x0000000000022555 __libc_start_main() ???:0
5 0x0000000000409f69 _start() ???:0
=================================

It also happened that the SCF went through, but then the convergence was erratic:

SCF= 1 NormRD= 1.000000000000 Uele= -3342.132721921273
SCF= 2 NormRD= 22.367911274597 Uele= -3342.132721921278
SCF= 3 NormRD= 22.267924105832 Uele= -3342.757357809105
SCF= 4 NormRD= 14.885854456483 Uele= -3405.337453216196
SCF= 5 NormRD= 11.326412715825 Uele= -3417.431768320585
SCF= 6 NormRD= 8.762190737839 Uele= -3429.049330149574
SCF= 7 NormRD= 7.005111460185 Uele= -3435.979907779719
SCF= 8 NormRD= 6.032249046860 Uele= -3439.377741675632
SCF= 9 NormRD= 6.096590124823 Uele= -3440.953613298794
SCF= 10 NormRD= 5.309079245134 Uele= -3441.297861703613
SCF= 11 NormRD= 2863.530892999428 Uele= -151086.939599176287
SCF= 12 NormRD= 15178.224758521057 Uele= -151625.069244069338
SCF= 13 NormRD= 9948.354516018910 Uele= -151393.362145951804
SCF= 14 NormRD= 16754.994945226168 Uele= -151692.940923026472
SCF= 15 NormRD= 27604.906673903726 Uele= -152077.720412037277
SCF= 16 NormRD= 26503.853745236884 Uele= -152039.358649583650
SCF= 17 NormRD= 26724.793297418611 Uele= -152040.567719414306
SCF= 18 NormRD= 25379.539614843397 Uele= -152004.892991992994
SCF= 19 NormRD= 22021.942841008407 Uele= -151882.911061797349
SCF= 20 NormRD= 22317.845296507010 Uele= -151888.787055976281
SCF= 21 NormRD= 24200.647275747029 Uele= -151956.823730861302
SCF= 22 NormRD= 25715.931693521754 Uele= -152005.605329403596
SCF= 23 NormRD= 27815.976929357177 Uele= -152068.744033536030
SCF= 24 NormRD= 28956.583052670561 Uele= -152104.977342572703
SCF= 25 NormRD= 27827.492251498919 Uele= -152069.699068365037
SCF= 26 NormRD= 27419.378529727994 Uele= -152056.344454987586
SCF= 27 NormRD= 27154.523164323782 Uele= -152054.141309092112
SCF= 28 NormRD= 23354.717769933737 Uele= -151925.318994243775
SCF= 29 NormRD= 23160.259598990047 Uele= -151919.849531330343
SCF= 30 NormRD= 21469.676171773182 Uele= -151860.094856999000
SCF= 31 NormRD= 29884.928799431440 Uele= -142441.671897559369
SCF= 32 NormRD= 53564.074888208466 Uele= -143054.157979849551
SCF= 33 NormRD= 49418.835904393825 Uele= -142899.288904871646
SCF= 34 NormRD= 71501.866478192707 Uele= -143200.368948580697
SCF= 35 NormRD= 70911.335864249952 Uele= -142931.447090229543
SCF= 36 NormRD= 81612.773725810737 Uele= -142835.510652340832
SCF= 37 NormRD= 95769.866521563948 Uele= -142848.836199543759
SCF= 38 NormRD= 114534.616960848143 Uele= -142924.527718019730
SCF= 39 NormRD= 157594.522316717135 Uele= -143373.151716906315
SCF= 40 NormRD= 168027.582120055886 Uele= -143237.477225554641
SCF= 41 NormRD= 200504.713783434243 Uele= -143460.032084750856
SCF= 42 NormRD= 235353.493648702366 Uele= -143603.580458256416
SCF= 43 NormRD= 260301.699024375237 Uele= -143592.515719753806
SCF= 44 NormRD= 295141.237618599378 Uele= -143691.633872486767
SCF= 45 NormRD= 347810.078246433754 Uele= -144024.214638737554
SCF= 46 NormRD= 411603.932857038162 Uele= -144438.421810463333
SCF= 47 NormRD= 440333.282632279501 Uele= -144531.273694411677
...

This is solved by switching the mixing type to:

scf.Mixing.Type rmm-diish # Simple|Rmm-Diis|Gr-Pulay|Kerker|Rmm-Diisk

Note that I ran into this issue only when performing Gamma point calculations.

Best,
S. Dechamps

Page: [1]

Re: Convergence issue from kerker_mixing ( No.1 )

Date: 2023/09/13 12:14
Name: T. Ozaki

Hi,

OpenMX provides four kinds of RMM-DIIS methods which
can be specified by

RMM-DIIS
RMM-DIISK
RMM-DIISH
RMM-DIISV

as explained at
https://www.openmx-square.org/openmx_man3.9/node40.html

In fact, RMM-DIISH is a robust method for the plus U and constraint methods.

Regards,

TO

Re: Convergence issue from kerker_mixing ( No.2 )

Date: 2023/09/15 01:31
Name: Sam <samuel.dechamps@cea.fr>

Dear Pr. Ozaki,

In this case, no constraints (such as Hubbard) were imposed.

So it raises the question as to why such bad convergence behavior (or even crash when mixing the density) would occur.

Regards,
S. Dechamps

Re: Convergence issue from kerker_mixing ( No.3 )

Date: 2023/09/15 12:26
Name: T. Ozaki

Hi,

If you can share your input file, we can trace your
calculation to figure out what's going on.

Regards,

TO

Re: Convergence issue from kerker_mixing ( No.4 )

Date: 2023/09/23 01:34
Name: Sam <samuel.dechamps@cea.fr>

Dear Pr. Ozaki,

The SCF part of the input file was shared above.
Here's the totality of it.

#
# File Name
#

System.CurrrentDirectory ./ # default=./
System.Name ws2_8p_n2
level.of.stdout 1 # default=1 (1-3)
level.of.fileout 0 # default=1 (0-2)

data.path /home/ucl/modl/sdch/files/OpenMX

#
# Definition of Atomic Species
#

Species.Number 2
<Definition.of.Atomic.Species
W W7.0_PBE-s2p1d1 W7.0_PBE
S S7.0_PBE-s2p1d1 S7.0_PBE
Definition.of.Atomic.Species>

#
# SCF or Electronic System
#

scf.XcType GGA-PBE # LDA|LSDA-CA|LSDA-PW|GGA-PBE
scf.SpinPolarization off # On|Off|NC
scf.SpinOrbit.Coupling off
scf.ElectronicTemperature 300.0 # default=300 (K)
scf.energycutoff 300.0 # default=150 (Ry)
scf.maxIter 800 # default=40
scf.EigenvalueSolver Band # DC|GDC|Cluster|Band
scf.Kgrid 1 1 1 # means n1 x n2 x n3
scf.Mixing.Type rmm-diish # Simple|Rmm-Diis|Gr-Pulay|Kerker|Rmm-Diisk
scf.Init.Mixing.Weight 0.001 # default=0.30
scf.Min.Mixing.Weight 0.001 # default=0.001
scf.Max.Mixing.Weight 0.100 # default=0.40
scf.Mixing.History 20 # default=5
scf.Mixing.StartPulay 10 # default=6
scf.Mixing.EveryPulay 1
#scf.Kerker.factor 1
scf.criterion 1.0e-7 # default=1.0e-6 (Hartree)
scf.lapack.dste dstevx # dstevx|dstedc|dstegr,default=dstevx
scf.restart on

scf.stress.tensor off # on|off, default=off

#
# Geometry Optimization
#

MD.Type EF # Opt|DIIS|BFGS|RF|EF
MD.Opt.DIIS.History 3 # default=3
MD.Opt.StartDIIS 5 # default=5
MD.Opt.EveryDIIS 200 # default=200
MD.maxIter 100 # default=1
MD.Opt.criterion 4.0e-4 # default=0.0003 (Hartree/Bohr)

#
# Atoms
#

Atoms.UnitVectors.Unit Ang # Ang|AU
<Atoms.UnitVectors
111.159 0.000 0.0
0.0 22.47 0.0
0.0 0.000 50.0
Atoms.UnitVectors>

Re: Convergence issue from kerker_mixing ( No.5 )

Date: 2023/09/23 01:37
Name: Sam <samuel.dechamps@cea.fr>

Atoms.Number 475
Atoms.SpeciesAndCoordinates.Unit Ang # Ang|AU
<Atoms.SpeciesAndCoordinates
...
Atoms.SpeciesAndCoordinates>

<MD.Fixed.XYZ
...
MD.Fixed.XYZ>

Since the coordinates are too large to be shared here, you can access them at https://drive.google.com/file/d/1BNUYXBQ5Z-XKi9LGwEJlTqQnNNvXQPol/view?usp=sharing

Page: [1]