GGA+U+SOC

[vrinsom]

Hello,
I am observing some strange behavior when I try to run spin orbit coupling calculations. I am running test calculations on Ni-doped CeO2 and Ni-doped CeO2 with an oxygen vacancy at the nearest-neighbor site to the Ni dopant. The POSCARs for both the runs are identical, except that the latter run has an oxygen vacancy next to the Ni.

My settings for both the calculations are exactly the same. I generate the WAVECAR and CHGCAR from a collinear run using 1 k-point at Gamma (I use the vasp_std binary file). I do not relax the structure. I then run the non-collinear calculation reading the WAVECAR and CHGCAR from the collinear run and at 1 k-point at Gamma.

However, for the non-collinear calculation in the Ni-doped case (without oxygen vacancy), I receive the error
ERROR: while reading WAVECAR, plane wave coefficients changed 31318
15659
But for the non-collinear calculation in the Ni-doped case with the oxygen vacancy, VASP is able to read the WAVECAR correctly and successfully completes the calculation.

This is strange, since the only difference in the calculations is the presence of an oxygen vacancy ! I was wondering why this occurs, and how do I get VASP to read the WAVECAR for the runs without an oxygen vacancy?

[alexey.tal]

Hi,

Could you please provide all relevant input and output files for the steps you have described according to the forum guidelines.

[vrinsom]

Sure, here are the required input and output files. Sorry for the delay, thank you for looking into this!

[alexey.tal]

This problem occurs because of the number of NBANDS. If NBANDS is not set, it will be calculated based on the initial magnetic moments, number of electrons, number of atoms and CPU cores.

In order to be able to read WAVECAR produced in a collinear calculation, the non-collinear calculation should include exactly double the number of bands.

[vrinsom]

Got it ! Thank you so much !