My Community

When I do a monoclinic-structure optimization, I set ISIF=2 to relax atom position first. Then I get one atom position very different from the previous.(I tried both IBRION=2 and 1.) There are 12 atoms in the cell, others are quite reasonable. That's the one VASP calculated:
0.0128129829852250 0.4439087456513696 0.7668073850653303
That's the one in experiment:
0.00569999 0.4363 0.7292
It seems only X-axis is greatly different. I don't know if the result is reliable or I can vary other calculate parameters to get a better result?
Thx.

please make sure that you used sufficiently large basis sets:
-- PREC=Accurate, ENCUT at least 1.3*max(ENMAX) of the PPs which you used for your POTCAR (the total drift should be significantly smaller than the largest force on the ions)
-- maybe it is necesary to increase the electronic convergence criterium (EDIFF) beyond the default setting and to add ADDGRID=.True.
-- what convergence criterium did you set? (energy difference or forces? Usually the forces (EDIFFG <0) are a safer criterium. (please check the remaining forces on that atom, they should be less than 0.01 eV/Ã…)
-- also, please check if the run stopped because the convergence criterium was matched or whether the maximum relaxation step number (NSW) was reached.
-- was there significant pressure on the cell after all ions were converged? (the external pressure should be about 0.0x for converged cells) If so, please do an intermediate ISIF=3 run (cell vol and shape optimization) and re-otimize the ionic positions after that.