REAL_OPTLAY:internal error

Message

okuno · #1 Post by **okuno** » Mon Mar 02, 2009 1:34 am

Dear Vasp Users and administrators.

I 'm now calculating the SCF calculation with large system with

impurity. Without impurity, the calculation is conversed but

when I replace one atom to the impurity ( in this case Ce )

it soon crash with the message before the SCF calculation.

REAL_OPTLAY: internal error (1) 2980602 2980634

What does this message means ? and how do I treat

this error ?

Sincerely.

#2 Post by **admin** » Mon Mar 02, 2009 12:01 pm

the reason for this error might be the following: please check the cutoffs of the POTCARS you join (ENMAX). If they are very different,
please
1) either use POTCARS with similar cutoffs
2) or combine them such that the element with the hightes ENMAX
is the first one (please don't forget that you will also have to re-arrange POSCAR!)

#3 Post by **admin** » Mon Mar 02, 2009 12:04 pm

please check the cutoffs of the POTCARS you join (ENMAX). If they are very different, please
1) either use different POTCARS, with similar cutoffs
2) or combine them such that the element with the hightest ENMAX is the first one (please don't forget that you will also have to re-arrange POSCAR!)

Chinnappanravi · #4 Post by **Chinnappanravi** » Tue Jul 21, 2020 11:12 am

The discussion suggests that the error is due arrangement of POTCARs with different ENMAX.

But, I encounter similar error during an MD simulation:

"
1 T= ****** E= Infinity F= -.13846926E+04 E0= -.13841577E+04 EK= 0.19753+202 SP= 0.00E+00 SK= Infinity
bond charge predicted
REAL_OPTLAY: internal error (1) 0 4032
REAL_OPTLAY: internal error (1) 0 4032
...
"

For a 4x4x4 bcc supercell with 128 atoms of uranium (U128), VASP works well. No error.

But, it terminates with the above error for U126Vac2, obtained by removing two nearest neighbor U atoms from the U128 POSCAR.

It will be helpful to have a solution to this problem.

#5 Post by **martin.schlipf** » Wed Jul 22, 2020 6:51 am

Without seeing your specific setup it is not possible to tell whether there is an issue with the code or an error in your input. Could you tell me which exact POTCAR and which version of Vasp you are using. Please attach also the input files (INCAR, KPOINTS, POSCAR) that reproduce this issue.

Chinnappanravi · #6 Post by **Chinnappanravi** » Wed Jul 22, 2020 7:29 am

Please find attached the INCAR, KPOINTS and POSCAR files.
POTCAR: PAW_PBE U 06Sep2000.
vasp.4.6.28 (Dr. Alfe Dario version)

#7 Post by **martin.schlipf** » Thu Jul 23, 2020 11:01 am

When does this error occur, during the convergence or right at the start? I could run a electronic self-consistency with the input you gave me and the current version of Vasp. Naturally, I did not run the full 8000 step MD simulation.

If the error does occur for one specific structure during the MD run, please give me exactly that structure to reproduce the issue. I need to point out that you are using an ancient version of Vasp. It may well be that your issue is already fixed in the more recent versions. So please consider upgrading your Vasp license.

Chinnappanravi · #8 Post by **Chinnappanravi** » Fri Jul 24, 2020 9:16 am

The error occurred before proceeding to 2nd MD step after completing 1st MD step.
The error occurred for the POSCAR included in the previous post.
We have upgraded our license(active). This version of vasp is fast for our MD several order of times than the current vasp.

#9 Post by **martin.schlipf** » Sat Jul 25, 2020 4:07 pm

Ok, I traced the issue down to a numerical instability in the Nose update step. I will check if this can be resolved, but because it is just an instability, you can actually overcome this by adjusting the parameters of the MD simulation.

This page gives you some ideas, which parameters to tune. I don't know which of those were available in Vasp 4.6, but what work e.g. is setting SMASS = -3.

If you need to use the old version, I would recommend performing some 10 iterations with the new code and then taking the resulting CONTCAR to start. Hopefully that particular parameter space is then more stable.

#10 Post by **martin.schlipf** » Tue Jul 28, 2020 6:20 am

Okay it turns out the numerical instability comes from your POSCAR file. Because you initialized all velocities to 0, the system starts from a point that is very different from the one you want (T = 1100K).
In the thermostat there is a step, where the next velocities are calculated based on the current velocities and the forces. But it includes higher order corrections that do not work if you start very far from the solution.

I recommend deleting the velocities from the POSCAR file. Then Vasp will automatically generate appropriate initial velocities.

Chinnappanravi · #11 Post by **Chinnappanravi** » Thu Oct 29, 2020 11:39 am

Deleting the velocities from the POSCAR works. Thanks.

SKM · #12 Post by **SKM** » Fri Nov 27, 2020 1:22 pm

coming here from forum/viewtopic.php?f=3&t=3005#p19694 as suggested by Admin.
when this error occurred and as suggested checked POTCAR for ENMAX values and then combined individual POTCARs from max ENMAX to descending order, even then i got the same error.
the sequence of POTCAR ENMAXs...
ENMAX = 269.864; ENMIN = 202.398 eV
ENMAX = 220.318; ENMIN = 165.238 eV
ENMAX = 175.647; ENMIN = 131.735 eV
ENMAX = 105.037; ENMIN = 78.777 eV

the portion from log file is copied here.
POSCAR found : 4 types and 224 ions
scaLAPACK will be used
WARNING: PSMAXN for non-local potential too small
LDA part: xc-table for Pade appr. of Perdew
POSCAR found type information on POSCAR Mn Cs I Bi
POSCAR found : 4 types and 224 ions
POSCAR, INCAR and KPOINTS ok, starting setup
REAL_OPTLAY: internal error (1) 7381103 7381506
REAL_OPTLAY: internal error (1) 7381103 7381506
REAL_OPTLAY: internal error (1) 7381103 7381506
....
....

#13 Post by **martin.schlipf** » Mon Nov 30, 2020 7:43 am

Can you upload a set of input files to reproduce the issue? Will the error occur immediately or do I need to run several steps?

SKM · #14 Post by **SKM** » Mon Nov 30, 2020 10:12 am

hi
please see the attached tar files.
ENCUT 616 eV successful run.
ENCUT 617 eV Not success.
both attached.

is this problem associated with Gcut value assessed from POTCAR file?

Regards

#15 Post by **martin.schlipf** » Tue Dec 01, 2020 2:59 pm

I could not reproduce this with Vasp 6. But I am aware that there was an issue with the Cs PAW in Vasp 5.4.4 which has been fixed in Vasp 6. Does this also happen on a single k-point in Vasp 5.4.4?

My Community