monoclinic cell problem (LATTYP) - different lattice vectors different energy

[Greekman]

Hello

i have a question about the "LATTYP" Routine, as included as output in the OUTCAR.
I am doing some research in binary 2D Alloys and i have a POSCAR like this for a monoclinic A4B2-System:
(the Results are without any Relaxation: ISIF=2 IBRION=-1 NSW=0 and with ISMEAR = 1 & SIGMA = 0.1)


##########################
1)
POSCAR:
-Volume
6 0 0
2 1 0
0 0 6.11374298681587706
4 2
C
0 0 0 A
1 0 0 A
2 0 0 A
3 0 0 A
4 0 0 B
5 0 0 B

KPOINTS
Gamma
4 10 1

Result:
F= -.37496896E+02 E0= -.37497636E+02 d E =0.222051E-02 mag= 14.7314
##########################

if i go in the OUTCAR, the LATTYP Routine give me this curious cell:
##########################
2)
-Volume
-4 1 0
0 0 -6.11374298681587706
6 -3 0
4 2
C
0 0 0 A
1 0 0 A
2 0 0 A
3 0 0 A
4 0 0 B
5 0 0 B

KPOINTS
Gamma
8 1 5

if i calculate this i get:
F= -.37600742E+02 E0= -.37600401E+02 d E =-.102088E-02 mag= 14.3858
##########################
A REALLY BIG DIFFERENCE !!!

and NOW the OUTCAR gives me another different unitcell:
##########################
3)
(-4 1 0)
(0 0 6.1137429..)
(6 0 0)
##########################
which i didn't calculate




What i tried was to see what VASP will give me as a Result and as a new LATTYP-cell
so i created cell #4). I shifted the positions of the Atoms by 0.125 in the x & y-direction
because till now i thought maybe VASP doesn't like unitcells where the all of the Atoms
are on the lattice vectors

##########################
4)
POSCAR:
-Volume
2 1 0
-2 2 0
0 0 6.11374298681587706
4 2
C
0.125 0.125 0 A
1.125 0.125 0 A
2.125 0.125 0 A
3.125 0.125 0 A
4.125 0.125 0 B
5.125 0.125 0 B

KPOINTS:
Gamma
7 5 1

Results:
F= -.37522427E+02 E0= -.37521831E+02 d E =-.178692E-02 mag= 14.5918
##########################
again a difference in Energy and magnetization!

here i get in the OUTCAR this totally new Lattice:
##########################
5)
(0 -3 0)
(0 0 -6.11374298681587706)
(2 4 0)
and if I calculate this as
POSCAR:
-Volume
0 -3 0
0 0 -6.11374298681587706
2 4 0
4 2
C
0.5 0 0 A
1.5 0 0 A
2.5 0 0 A
3.5 0 0 A
4.5 0 0 B
5.5 0 0 B

KPOINTS:
Gamma
7 1 5

Result:
F= -.37489317E+02 E0= -.37488782E+02 d E =-.160413E-02 mag= 14.8550
##########################






SO MY QUESTIONS ARE:
1) which Lattice is "the right one" ? or the preffered one to be calculated?
2) IF i put a lattice and positions in the POSCAR and the "LATTYP" Routine
gives another OUTPUT in the OUTCAR, is MY POSCAR-file used or the LATTICE
which LATTYP finds?? And if the LATTICE from LATTYP is used, what happens with the KPOINTS??
3) How can I tell VASP to keep my lattice vector and atom positions?

THANK YOU IN ADVANCE

[admin]

VASP uses the geometry input given in POSCAR for the calculation, the symmetry information given in OUTCAR just informs about the internal vasp symmetry analysis (which re-arranges the lattice vectors, but is not used further on, except for the space group information of the lattice).
Total energies only may be compared if the
-- lattice types, shapes (including the angles between the lattice vectors!) and volumes match, and the corresponding k-point sets are equivalent
-- if all calculations are converged o the same level of accuracy electronically
-- atoms are kept fixed if IBRION=-1, or if you use the selective dynamics option in POSCAR and keep (some of) the atoms fixed

[Greekman]

But what u mean with:
i can compare only the Total Energy if the angles btw lattice vectors are the same?
The POSCARs i included here are all for the SAME Structure and describe the same structure!!!
Why do i get different Energies despite that the Structure are everytime the same and the KPOINT set is equivalent!?

[admin]

I think the angle between the e.g. the lattice vectors a,b and a,c of the Bravais matrices
2 1 0
-2 2 0
0 0 6.11374298681587706

and

0 -3 0
0 0 -6.11374298681587706
2 4 0

differ slightly