Energy minima vs Pressure minima

Message

rpf · #1 Post by **rpf** » Thu Mar 11, 2010 9:19 am

Hi Friends,

During the geometry optimization in VASP, I see that energy minima and pressure minima DO NOT occurs at same lattice constant (there is a small difference between them). What is the reason for this situation

<span class='smallblacktext'>[ Edited ]</span>

forsdan · #2 Post by **forsdan** » Thu Mar 11, 2010 10:19 am

Well, the pressure is evaluated from the stress tensor and the stress tensor requires a high energy cutoff and prec=accurate in order to be evaluated accurately. So probably your cutoff energy and/or prec is too low. For very accurate total energy evaluation you should also use ISMEAR = -5.

So increase your cutoff at least 30%, set prec=accurate and switch on ISMEAR=-5. Then compare again. If there still is a difference then gradually increase the cutoff energy until the difference disappear.

Cheers,
/Dan
<span class='smallblacktext'>[ Edited Thu Mar 11 2010, 11:27AM ]</span>

rpf · #3 Post by **rpf** » Thu Mar 11, 2010 12:48 pm

it is fixed by increasing the cutoff and setting prec = accurate...

thank you very much

#4 Post by **admin** » Fri Mar 12, 2010 9:49 am

please let me add one note: when optimizing the structure of a cell automatically (ISIF), the FFT-grids are determined at the beginning of the vasp run and not adjusted during the run. Therefore, if the starting geometry (volume, cell shape) is far away from the equilibrium, one should do the optimization step-wise (NSW 3-5) at the beginning (mv CONTCAR to POSCAR, but delete WAVECAR), such that the FFT-grids are adjusted to the cell ceometries during relaxation. Please have a look at accuracy.pdf of the vasp-workshop (http://cms.mpi.univie.ac.at/vasp-workshop tutorials and session talks) for further information on that issue