Reducing external pressure in magnetic relaxation

[vahid_askarpour]

Dear VASP Community,

In the attached files of the relaxation of an antiferromagnet, the magnetization is zero as expected for an antiferromagnet. However, the external pressure (in the OUTCAR) does not go down to zero. I have increased the Ecut to 1000eV but that did not help. I have copied CONTCAR to POSCAR and resubmitted to no avail. I have also used IBRION=1 with no reduction in external pressure.I have two questions:

1. How does one reduce the external pressure?

2. Is the INCAR correct for magnetic relaxation? Should I remove the constraints over the direction of moments and let the moments find their resting place?

Thank you,
Vahid

[manuel_engel1]

Hi Vahid,

There are few things to try.

1. I recommend lowering EDIFF. 1e-4 is very large and will lead to errors. I would suggest running with 1e-7 to 1e-8. In a simple test case that I built on your example, EDIFF=1e-7 already reduced the external stress to zero.
2. I also recommend lowering ENCUT to about 500 eV. This should still be accurate but it runs much faster.
3. Setting PREC=accurate improves the accuracy of forces, among other things, which may help in this scenario.

The parameters related to magnetism are OK. However, the constrained local moments approach modifies the total energy. I am not 100% certain if this also affects the external pressure but it is possible. Just as explained on the wiki page, you will have to slowly increase the LAMBDA parameter from run to run and watch the results.

[vahid_askarpour]

Hi Manuel,

Thank you for your suggestions and also confirming that the input for magnetic relaxation is OK. I have implemented your suggestions and if any issues arise, I will report.

Cheers,
Vahid