doubled imaginary dielectric function output with ISMEAR=-5 and ISPIN=1

Problems running VASP: crashes, internal errors, "wrong" results.

Moderators: Global Moderator, Moderator

Locked
Message
Author
bonan_zhu
Newbie
Newbie
Posts: 5
Joined: Thu Apr 09, 2020 8:20 am

doubled imaginary dielectric function output with ISMEAR=-5 and ISPIN=1

#1 Post by bonan_zhu » Fri Oct 22, 2021 11:46 am

Using 6.2.0, for LOPTICS=.TRUE. calculations using ISMEAR=-5 and ISPIN=1, the dielectric function outputs are doubled compared to a run with ISPIN=2.

There is an example:

POSCAR:

Code: Select all

Ga As
 1.0000000000000000
     2.8297157547615601    0.0000000000000000    2.8297157547615601
     2.8297157547615601    2.8297157547615601    0.0000000000000000
     0.0000000000000000    2.8297157547615601    2.8297157547615601
   1   1
Cartesian
  0.0000000000000000  0.0000000000000000  0.0000000000000000
  4.2445736321423402  4.2445736321423402  4.2445736321423402
INCAR:

Code: Select all

ALGO = normal
EDIFF = 1e-08
EDIFFG = -0.0001
ENCUT = 500
GGA = pe
IBRION = -1
ISIF = 2
KPAR = 1
LASPH = .TRUE.
LDAU = .TRUE.
LDAUJ = 0.0 0.0 0.0 0.0
LDAUL = -1 -1 -1 2
LDAUTYPE = 2
LDAUU = 0.0 0.0 0.0 4.0
LMAXMIX = 4
LREAL = .FALSE.
LWAVE = .TRUE.
NCORE = 2
NELM = 200
NWRITE = 1
PREC = accurate
SIGMA = 0.05
NBANDS=32
NEDOS=2000

LOPTICS = .TRUE.
ISMEAR = -5
ISPIN = 1
KPOINTS:

Code: Select all

# No comment
0
Gamma
            11           11            11
  0.000000000   0.000000000   0.000000000

Results with ISPIN=2:

Code: Select all

grep -A 10 "IMAG.*density" ispin2/OUTCAR                                                                                   !11030
  frequency dependent IMAGINARY DIELECTRIC FUNCTION (independent particle, no local field effects) density-density
     E(ev)      X         Y         Z        XY        YZ        ZX
  --------------------------------------------------------------------------------------------------------------
    0.000000    0.000000    0.000000    0.000000    0.000000    0.000000    0.000000
    0.030044    0.006671    0.006671    0.006671   -0.000000************************
    0.060088    0.013375    0.013375    0.013375   -0.000000************************
    0.090132    0.020148    0.020148    0.020148   -0.000000************************
    0.120176    0.027026    0.027026    0.027026   -0.000000************************
    0.150220    0.034048    0.034048    0.034048   -0.000000************************
    0.180264    0.041258    0.041258    0.041258   -0.000000************************
    0.210309    0.048703    0.048703    0.048703   -0.000000************************


with ISPIN=1:

Code: Select all

 grep -A 10 "IMAG.*density" ispin1/OUTCAR                                                                                   !11031
  frequency dependent IMAGINARY DIELECTRIC FUNCTION (independent particle, no local field effects) density-density
     E(ev)      X         Y         Z        XY        YZ        ZX
  --------------------------------------------------------------------------------------------------------------
    0.000000    0.000000    0.000000    0.000000    0.000000    0.000000    0.000000
    0.030044    0.013342    0.013342    0.013342   -0.000000   -0.000000    0.000000
    0.060088    0.026751    0.026751    0.026751   -0.000000   -0.000000    0.000000
    0.090132    0.040297    0.040297    0.040297   -0.000000   -0.000000    0.000000
    0.120176    0.054053    0.054053    0.054053   -0.000000   -0.000000    0.000000
    0.150220    0.068097    0.068097    0.068097   -0.000000   -0.000000    0.000000
    0.180264    0.082515    0.082515    0.082515   -0.000000   -0.000000    0.000000
    0.210309    0.097406    0.097406    0.097406   -0.000000   -0.000000    0.000000
which are twice of the values of ISPIN=1.

Inputs and outputs files are attached.

Having looked at the code, I think this is probably caused by the duplicated-introduction of the spin degeneracy factor in the the epsilon routine used for the tetrahedron method in linear_optics.F90.
You do not have the required permissions to view the files attached to this post.

martin.schlipf
Global Moderator
Global Moderator
Posts: 454
Joined: Fri Nov 08, 2019 7:18 am

Re: doubled imaginary dielectric function output with ISMEAR=-5 and ISPIN=1

#2 Post by martin.schlipf » Mon Oct 25, 2021 7:44 am

Thank you for reporting this to us. I could reproduce this issue and will investigate further.

martin.schlipf
Global Moderator
Global Moderator
Posts: 454
Joined: Fri Nov 08, 2019 7:18 am

Re: doubled imaginary dielectric function output with ISMEAR=-5 and ISPIN=1

#3 Post by martin.schlipf » Tue Nov 02, 2021 7:44 am

We have implemented a fix for this, which will be released with VASP 6.3.

For the time being, we suggest using a different ISMEAR or manually removing the duplication of RSPIN from the source file.

bonan_zhu
Newbie
Newbie
Posts: 5
Joined: Thu Apr 09, 2020 8:20 am

Re: doubled imaginary dielectric function output with ISMEAR=-5 and ISPIN=1

#4 Post by bonan_zhu » Wed Nov 03, 2021 5:09 pm

Thanks, glad to know it is fixed!

bonan_zhu
Newbie
Newbie
Posts: 5
Joined: Thu Apr 09, 2020 8:20 am

Re: doubled imaginary dielectric function output with ISMEAR=-5 and ISPIN=1

#5 Post by bonan_zhu » Mon Nov 08, 2021 3:38 pm

Could you please confirm in VASP 6.2.0 (before any patch for this), which ISPIN gives the correct result?

I have done some additonal cross-checking with VASP5 + ISMEAR=-5/0 and VASP6 + ISMEAR=0 , which seems to suggest the ISPIN=1 + ISMEAR=5 in VASP6 is the correct one. (e.g. ISPIN=2+ISMEAR=-5+VASP6 gives halve of the result).

martin.schlipf
Global Moderator
Global Moderator
Posts: 454
Joined: Fri Nov 08, 2019 7:18 am

Re: doubled imaginary dielectric function output with ISMEAR=-5 and ISPIN=1

#6 Post by martin.schlipf » Tue Nov 09, 2021 7:48 am

The results were correct for ISPIN=1 or LNONCOLLINEAR=.TRUE. but wrong for ISPIN=2.

Locked