internal error in: wave.F at line: 1603

[CieloMist]

Hey All, I've been running some relaxations with the ASE interface lately. On initialization of one relaxation I get the following error:

Code: Select all

|                     _     ____    _    _    _____     _                     |
|                    | |   |  _ \  | |  | |  / ____|   | |                    |
|                    | |   | |_) | | |  | | | |  __    | |                    |
|                    |_|   |  _ <  | |  | | | | |_ |   |_|                    |
|                     _    | |_) | | |__| | | |__| |    _                     |
|                    (_)   |____/   \____/   \_____|   (_)                    |
|                                                                             |
|     internal error in: wave.F  at line: 1603                                |
|                                                                             |
|     internal error in WFINIT_PCG_32: orbitals linearily dependent at        |
|     random-number initialization                                            |
|                                                                             |
|     If you are not a developer, you should not encounter this problem.      |
|     Please submit a bug report.                                             |
|                                                                             |
 -----------------------------------------------------------------------------

Seems like something is going wrong with the orbital initialization. I have played with iniwav, reduced symmetry, and a few algorithms with no luck. I'm using the Nd and I POTCAR files. INCAR, POSCAR, KPOINTS are attached. Can you help me understand what's causing the error?

Inputs.zip

[manuel_engel1]

Hi, thanks for filing a bug report on the VASP forum. This is important information and helps us improve the code.

I have downloaded and run your example and could immediately reproduce the error. I tried to play around with random-number generation as the error suggests. However, that did not lead to any different result. Then I noticed that the k-points specified in your KPOINTS file were given in very large units (they are sitting far outside your cell). You can find this info in your OUTCAR file towards the end before the run crashes:

Code: Select all

 k-point     1 :   0.6973******* 2.5060  plane waves:       0
 k-point     2 :  68.879033.8584 2.5060  plane waves:       0
 k-point     3 :   0.6973******* 7.5180  plane waves:       0
 k-point     4 :  68.879033.8584 7.5180  plane waves:       0
 k-point     5 :   0.6973*******12.5300  plane waves:       0
 k-point     6 :  68.879033.858412.5300  plane waves:       0
 k-point     7 :   0.6973*******17.5419  plane waves:       0
 k-point     8 :  68.879033.858417.5419  plane waves:       0
 k-point     9 :   0.6973*******22.5539  plane waves:       0
 k-point    10 :  68.879033.858422.5539  plane waves:       0

 maximum and minimum number of plane-waves per node :         0        0

 maximum number of plane-waves:         0
 maximum index in each direction:
   IXMAX=    0   IYMAX=    0   IZMAX=    0
   IXMIN=    0   IYMIN=    0   IZMIN=    0

This is where you can also find the problem of this calculation. There are no plane waves setup for any of these k-points, meaning that there are no orbitals, and that VASP will crash with this seemingly random error. I don't fully understand yet what is happening step-by-step with the k-points you supply, but when switching to a regular 4x4x4 mesh via automatic generation, VASP works correctly and moves on with the calculation. I will mark this as a potential bug and investigate a bit further, but for now, you should update your k-points file to either an automatic generation scheme or specify k-points explicitly that fit nicely into your reciprocal-space cell.

[manuel_engel1]

Indeed, it is very likely that the k-points in this system are too large so that the corresponding G-vectors for the plane-wave coefficients lie outside the cutoff radius (ENCUT). That is why no plane waves are generated and why the run fails. I have created an internal issue so that we can improve the error reporting and possibly issue a warning in such a case.

[CieloMist]

Thank you! :)