A question about relaxation with ISIF

Message

anin · #1 Post by **anin** » Wed Oct 25, 2006 6:23 am

I have a question about the relaxation of low symmetry lattice(monoclinic or orthorhombic..).

Since it has various freedom(a,b,c...), it will be a burden to optimize by fitting the E~V equations.

If I don't want to change its shape or symmetry. Can I do like this?

first use ISIF=7 to perform a calculation of volume change to determinde the lattice constant then use ISIF=2 to relax the positions until the POSCAR and CONTAR matches. (I will set a high PREC and ENCUT)

My questions is: Is it an valid method? Can I use the last CONTAR as the geometry of the structure? I appreciate any suggestions.

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

#2 Post by **admin** » Wed Oct 25, 2006 12:26 pm

yes, but in any case, I would check the final result by doing another run with ISIF=7 if the symmetry is low, especially if the ions relax significantly in the ISIF=2-run

anin · #3 Post by **anin** » Thu Oct 26, 2006 2:22 pm

Actually the lattice constant change a lot when I check with ISIF=7, does it mean I need to perform relaxtion again with ISIF=2?

#4 Post by **admin** » Tue Nov 14, 2006 9:57 am

yes, if it does you have to iterate this process until the changes in V / ionic positions (or forces) become small (set EDIFFG < 0 as a convergence criterium)