Restart with magnetic calculation, problem with result variation related to MAGMOM

[lxybenjamin]

I started a restart calculation by setting ISTART=2 (and change CONTCAR to POSCAR) and read in WAVECAR from previous ISPIN=2 calculation.

Now depending on my setup of MAGMOM line in the INCAR file of new run, I got different vasp run behavior:

If I set MAGMOM = 2*2, 2*-2, 8*0 (which is the same as in the previous run), I got,

----------------------------------------------------------------------------------------------
N E dE d eps ncg rms rms(c)
DAV: 1 -0.126334619180E+03 -0.12633E+03 -0.12350E-03 1488 0.167E-01 0.439E-02
DAV: 2 -0.126334869957E+03 -0.25078E-03 -0.12124E-03 1720 0.156E-01 0.244E-02
DAV: 3 -0.126335134549E+03 -0.26459E-03 -0.57455E-04 1184 0.982E-02 0.238E-02
DAV: 4 -0.126335323237E+03 -0.18869E-03 -0.31897E-04 1112 0.684E-02 0.229E-02
DAV: 5 -0.126335608593E+03 -0.28536E-03 -0.57176E-04 1008 0.758E-02 0.225E-02
DAV: 6 -0.126336129557E+03 -0.52096E-03 -0.18873E-03 1640 0.122E-01 0.254E-02
DAV: 7 -0.126335857615E+03 0.27194E-03 -0.57209E-04 1208 0.125E-01 0.343E-02
DAV: 8 -0.126337896107E+03 -0.20385E-02 -0.26263E-02 2048 0.534E-01 0.284E-02
DAV: 9 -0.126340413933E+03 -0.25178E-02 -0.40885E-02 2344 0.619E-01 0.580E-02
DAV: 10 -0.126340710047E+03 -0.29611E-03 -0.14880E-03 1416 0.156E-01 0.621E-02
DAV: 11 -0.126340117352E+03 0.59270E-03 -0.18429E-03 1328 0.157E-01 0.590E-02
DAV: 12 -0.126340176757E+03 -0.59405E-04 -0.13698E-04 1096 0.640E-02
1 F= -.12634018E+03 E0= -.12634018E+03 d E =-.126340E+03 mag= 0.0000
----------------------------------------------------------------------------------------------

This is a nice restart run. However, if I comment out MAGMOM line in the INCAR file in the new run, I got,

----------------------------------------------------------------------------------------------
N E dE d eps ncg rms rms(c)
DAV: 1 0.133460146221E+04 0.13346E+04 -0.38704E+04 1632 0.144E+03
DAV: 2 0.221296806363E+03 -0.11133E+04 -0.10215E+04 1696 0.422E+02
DAV: 3 -0.904380819899E+02 -0.31173E+03 -0.29944E+03 2224 0.192E+02
DAV: 4 -0.124600729116E+03 -0.34163E+02 -0.33294E+02 2640 0.683E+01
......
----------------------------------------------------------------------------------------------

In the OUTCAR, I find "ISTART = 0 job : 0-new 1-cont 2-samecut". This looks like a bug in the code(?).

If I set MAGMOM = 2*1, 2*-1, 8*0, I will get the nice restart run again, but not with other MAGMOM setups such as MAGMOM=4*2, 8*0 or MAGMOM=12*0. I have tried setting also ICHARG=1 (reading in CHGCAR) in addition to reading of WAVCAR, the conclusion is essentially the same.

Now suppose that by reading of WAVCAR and CHGCAR, the initial magnetic information is already given in these two files. Does the MAGMOM line provide the initial symmetry which is required in the continuous run (that is, the code requires that the magnetic symmetry in the new setup be the SAME as in the old run)? Otherwise, I cannot understand the different running behavior depending on the MAGMOM values.

Thanks in advance for answering my question.

Benjamin

[admin]

if you don't set the moments (AFM) explicitely, MAGMOM=+1 is automatically assumed for all atoms if ISPIN=2. This is done before WAVECAR is read, and some parameters (like the symmetry, the number of required bands...) are fixed at this early stage. If the parameters read from WAVECAR (like NBANDS, or the number of -symmetry-reduced- k-points ) do not match with the parameters generated in the startup phase of this run, WAVECAR is not used: please check, most probably NBANDS or the number of symmetry operations differ between the run without MAGMOM setin INCAR and the run whiich generated the WAVECAR file.