Please read section 7.6 and 7.6.2.
Slightly more involved are bulk calculations with internal degrees of freedom. The non ideal hcp phase (i.e. c/a non ideal) is a simple example for this case. To avoid problems due to Pulay stress, it is safest to relax at a set of constant volumes. Add the lines
ENCUT = 200.00 eV # energy cut-off for the calculation PREC = Normal # Normal precision LREAL = .FALSE # real space projection .FALSE. or Auto EDIFFG = -0.01 # threshold for forces and stress tensor EDIFF = 1E-5 NSW = 10 # 10 ionic steps IBRION = 2 # CG algorithm ISIF = 4 # forces and stress are optimized # optional parameters not required POTIM = size of trial step for ions (try the default 1.0)to the INCAR file and use a UNIX batch file to calculate the equilibrium cell shape for different volumes. The batch file might look similar to
rm WAVECAR for i in 3.7 3.8 3.9 4.0 4.1 do cat >POSCAR <<! C: hcp $i 1.00000 0.00000000000000 0.00000 -0.50000 0.86602540378444 0.00000 0.00000 0.00000000000000 1.63299 2 direct 0.00000000000000 0.00000000000000 0.000000 0.33333333333333 0.66666666666667 0.500000 ! echo "a= $i" ; vamp E=`tail -1 OSZICAR` ; echo $i $E >>SUMMARY.hcp done cat SUMMARY.fcc
Exercise 5: If you want to relax the volume as well, always use a large cutoff. Usually 1.3 times the default cutoff is sufficient. Start a relaxation allowing all degrees of freedom to relax simultaneously. Is the volume identical to the manual search. Repeat the calculations at the default cutoff. How large is the error in the volume.