Page 1 of 1

VASP.6.1.0 OpenACC port reading POSCAR issue

Posted: Sat Feb 27, 2021 7:57 pm
by tj_xie
Dear VASP team,

I have discovered that when I try to use the OpenACC port of VASP (vasp.6.1.0 28Jan20 build). It will concatenate the first two elements together which will cause a mismatch in atom types and numbers from the POSCAR.

I have uploaded my input and output for your reference.
errfolder.zip
In this case, my POSCAR specified the elements to be Ru C H, but the VASP read in as RuC, and H.
If I use the same input for other versions, e.g. regular std, CUDA gpu, etc, the job will run without any issue.

Will there be any patches for 6.1.0 or it is already fixed in later releases such as 6.1.2 or 6.2.0?

Many thanks.
TJ

Re: VASP.6.1.0 OpenACC port reading POSCAR issue

Posted: Tue Mar 09, 2021 1:32 pm
by merzuk.kaltak
Hello,

As a vasp-6 user you have access to vasp-6.2.0 and as such I strongly suggest your update to the newest version,
since we don't plan to release a patch for 6.1.0.

Concerning your problem:
Do you use this POSCAR file for all jobs?
You probably understood already that the error message in the stdout

Code: Select all

 -----------------------------------------------------------------------------
|                                                                             |
|     EEEEEEE  RRRRRR   RRRRRR   OOOOOOO  RRRRRR      ###     ###     ###     |   
|     E        R     R  R     R  O     O  R     R     ###     ###     ###     |   
|     E        R     R  R     R  O     O  R     R     ###     ###     ###     |   
|     EEEEE    RRRRRR   RRRRRR   O     O  RRRRRR       #       #       #      |   
|     E        R   R    R   R    O     O  R   R                               |
|     E        R    R   R    R   O     O  R    R      ###     ###     ###     |   
|     EEEEEEE  R     R  R     R  OOOOOOO  R     R     ###     ###     ###     |   
|                                                                             |
|     ERROR: the type information is not consistent with the number of        |
|     types                                                                   |
|                                                                             |
|       ---->  I REFUSE TO CONTINUE WITH THIS SICK JOB ... BYE!!! <----       |
|                                                                             |
 -----------------------------------------------------------------------------
tells you that the POSCAR file is corrupt.
The section that defines the ion types and number of ions reads:

Code: Select all

Ru  C  H   
64  5   9  
In total vasp expects to read the coordinates of 64+5+9=80 ions.
However the POSCAR contains only 78 coordinates.

Is this POSCAR the CONTCAR after an OpenACC-job? If so, then I would need the input files that create the corresponding CONTCAR with the openacc-port of vasp.

Re: VASP.6.1.0 OpenACC port reading POSCAR issue

Posted: Thu Mar 11, 2021 2:09 am
by tj_xie
Hello Merzuk,

Thank you for the response.

Concerning the problem, I double checked the POSCAR file and I do not see any corruption in the format.

Regarding the sum-up, the 64+5+9=78 is correct, not 80.
Indeed the POSCAR file can be visualized with no problem by VMD, VESTA, and Avogadro, etc.

The issue I'm seeing is that VASP fails to identify the atom types. In my case it combines the first two element: 'Ru' and 'C' into one 'RuC'.: 'POSCAR found type information on POSCAR RuC H'

Please advice. Thank you!

Re: VASP.6.1.0 OpenACC port reading POSCAR issue

Posted: Thu Mar 25, 2021 10:28 am
by merzuk.kaltak
I apologize, this is quite embarrassing :-[ .
You are right, of course. 64+5+9=78.

The problem you have is that the file format of the POSCAR has the "dos" format. I guess you have created it on a Windows machine.
However, if VASP runs on a linux machine, it requires a POSCAR in the "unix" format.
Your POSCAR is written in the "dos" format and shows trailing "^M" when opened with vim on a linux machine.
Take a look at this post and how to convert dos to unix format.