An issue in VCA: the old and the new charge density differ

[tonghua_yu]

Greetings!

I am dealing with a crystal using virtual crystal approximation (VCA, https://www.vasp.at/wiki/index.php/VCA). The static calculations (scf, band structures, etc.) work well. However, when I perform an ionic relaxation, the error

BRMIX: very serious problems
the old and the new charge density differ
old charge density: 159.99460 new 160.00000

appears before and after the first DAV, from the second ionic step (It does not occur in the first ionic step), as shown in the attached stdout file.

I have tried many possible solutions to this error which can be found from the internet, but none of them works. Suprisingly It disappears if I turn off atom mixture (e.g., "VCA = 1 0.66666667 0.33333333 1 1" --> "VCA = 1 1 0 1 1"). So I assume in my case this error is related to VCA in ionic relaxation. Could anyone advise to resolve this issue?

Alternatively, I notice the charge density difference is very tiny (old 159.99460, new 160.00000), and moreover the final relaxed structure looks reasonable. So maybe it is safe to simply ignore this error and trust the relaxed structure?

FYI, my vasp versions are 5.4.4 and 6.1.2 (error occurs on both versions). INCAR and standard output file are attached.

Many thanks in advance!

[ferenc_karsai]

Please upload all the neccessary files for this calculation (also POSCAR, POTCAR etc.) according to the forum guidelines.