POSCAR: Import-Export Issues

[markmick]

I'm experiencing a really odd problem with importing/exporting poscar files with vasp.  Attached are two files: a good, working poscar before importing into VNL, and another one that is exported from VNL.  I made no changes during the export.

The exported version is unusable.  First of all, the atoms are set to fixed positions (which I didn't specify), and if delete the "T" option so they can relax, the vasp run crashes with errors that don't make any sense.  The "before" version works perfectly fine.  But the "after" version does not run on vasp.  I'm using the same INCAR and KPOINTS for both vasp runs.

I don't understand how to fix this behavior.  Any suggestions would be appreciated.

Thanks,
Mark

[Daniele Stradi]

As far as I can see the "POSCAR-after" file is correctly set up for atomic relaxation. You might want to have a look at the description of POSCAR file in the VASP manual.

- The "T" option means that you allow for relaxation, the corresponding option for fixed atoms is "F".
- If you delete the "T" options, you should also delete the "Selective dynamics" line.

[Anders Blom]

As you can see, the cell in the two files are quite different, but the coordinates are similar. So something is odd, and precisely what depends on which work flow you used. But if you imported OUTCAR to VNL and then re-exported as POSCAR, I think I know what the issue is.

We found that when VNL imports a POSCAR, it will try to identify the lattice type, rather than just represent it as the generic UnitCell class. However, it forgot to also transform the coordinates while doing so, so you ended up with the wrong structure.

The bug is solved and we are pushing an updated plugin to the server any day now.