Increasing number of symmetry issues with VASP 6

[aaron_kaplan2]

I'm noticing that more of my jobs fail with VASP 6 than in VASP 5 for various symmetry related issues. The attached files do a direct comparison between VASP 5.4.4 and 6.4.1 for the same structure, which is a CONTCAR taken from the Materials Project entry mp-510536:
https://next-gen.materialsproject.org/m ... /mp-510536
In principle, any VASP output file (e.g., CONTCAR) should be re-usable as a valid input file, but this seemingly isn't the case in VASP 6.

While the VASP 5.4.4 calculation proceeds without issue, the VASP 6.4.1 calculation could not converge because of this symmetry-determination issue:

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     ###     ###     ###     |
|                                                                             |
|     Inconsistent Bravais lattice types found for crystalline and            |
|     reciprocal lattice:                                                     |
|                                                                             |
|        Crystalline: triclinic                                               |
|        Reciprocal : simple monoclinic                                       |
|                    (instead of triclinic)                                   |
|                                                                             |
|     In most cases this is due to inaccuracies in the specification of       |
|     the crytalline lattice vectors.                                         |
|                                                                             |
|     Suggested SOLUTIONS:                                                    |
|      ) Refine the lattice parameters of your structure,                     |
|      ) and/or try changing SYMPREC.                                         |
|                                                                             |
|       ---->  I REFUSE TO CONTINUE WITH THIS SICK JOB ... BYE!!! <----       |
|                                                                             | 
 -----------------------------------------------------------------------------

The obvious fixes are to symmetrize the structures with spglib / pymatgen, however it is worrying to me that VASP 6 has symmetry-determination issues not present in VASP 5, and that its output files throw errors when re-inititializing VASP.

[jonathan_lahnsteiner2]

Dear aaron_kaplan2,

You can tune the accuracy with which the symmetry is determined with the SYMPREC tag.
For more information on the SYMPREC tag, you can check the vasp wiki
https://www.vasp.at/wiki/index.php/SYMPREC
I checked your calculation, and when using SYMPREC=2E-5 it will work fine.

All the best Jonathan

[aaron_kaplan2]

Thanks Jonathan.

I'm well aware: we typically automate calculations with handler that increases SYMPREC when this error is printed by VASP. Indeed, the error handler tried SYMPREC in {1e-6, 1e-5, 1e-4}, of which only 1e-4 succeeded.

My concern is that these fixes were not needed as frequently in VASP 5. Since I'm often running > 1,000 calculations in parallel, this is problematic for our compute usage (calculations have to run, fail, and then restart until an optimal SYMPREC is attained).

If these changes in symmetry determination are more rigorous, I will probably increase SYMPREC by default for all our calculations.