Cell shape relaxation: k-points and symmetry
Posted: Wed Nov 20, 2013 5:49 pm
Hello,
I am currently doing some structural optimization on ZnO, but my questions are may be more general. I apologize if similar questions have been already asked, but I haven't found them.
- Imagine I am performing a series of fixed volume relaxations (ISIF=4) in order to fit the E(V) curve with the Murnaghan equation of state. Since I allow for cell shape changes, the symmetry of the system may change. I found a lot of explications about the implications of this on the basis set, but what about k-points? Changes in the symmetry of the system may occur, so how is the k-point grid handled?
- Is it always a good idea to use EDIFFG<0 to perform relaxations? Imagine that you have a cubic system; because of symmetry reasons you don't have forces acting on the atoms for any value of the cell-paramter. So, how does the code find the equilibrium parameter in this case?
Thanks for your help.
I am currently doing some structural optimization on ZnO, but my questions are may be more general. I apologize if similar questions have been already asked, but I haven't found them.
- Imagine I am performing a series of fixed volume relaxations (ISIF=4) in order to fit the E(V) curve with the Murnaghan equation of state. Since I allow for cell shape changes, the symmetry of the system may change. I found a lot of explications about the implications of this on the basis set, but what about k-points? Changes in the symmetry of the system may occur, so how is the k-point grid handled?
- Is it always a good idea to use EDIFFG<0 to perform relaxations? Imagine that you have a cubic system; because of symmetry reasons you don't have forces acting on the atoms for any value of the cell-paramter. So, how does the code find the equilibrium parameter in this case?
Thanks for your help.