Symmetry problems

[wbutler]

We have observed an unusual symmetry issue that may reflect a subtle bug. The problem was observed when we were running VASP for collinear PtMn which is antiferromagnetic. We expected by symmetry that the Pt atoms would not have moments, so we put zero moments for Pt on the MAGMOM line in INCAR. The code analyzed the symmetry and OUTCAR reported D_4h. Surprisingly, the code gave small moments on the Pt atoms in the output charge density and as reported in OUTCAR (using RWIGS). Conversely when we started the calculation with small moments on the Pt atoms, the code found a lower symmetry (D2_h), but the output generated results with zero moments on the Pt atoms. It seems that high symmetry in yields low symmetry out and low symmetry in yields high symmetry out.

Are we missing something simple or could this be an indicator of a bug? We can send INCAR, OUTCAR, etc. if someone is willing to help us understand.

Thanks - Bill Butler and Zhihong Lu

[admin]

please, first of all let me note that ALL local quantities (local moments, partial charges) written in OUTCAR do not refer to symmetrized quantities in VASP, therefore one should do some symmetrisation (k-points!) to check them with respect to symmetry in case of doubt.
ad 1) zero moment input: It may well be that there are some small moments at self-consistency, depending on the charge flow during the scf-process.
ad 2) depending on the arrangement of the magnetic moments in INCAR (probably AFM?) the found symmetry of the cell may be lower beacuse the orientation of the MAGMOMS enters the symmetry analysis.
As the IBZ of the D2_h cell is twice as large as the IBZ of the D4_h cell, a different set of k-points in the IBZ is generated.
---> without having a look at your actual results, I think the difference in the local output moments may very well be caused by the fact that the charges (and hence the moments) are not symmetrized before they are written to OUTCAR.