Prof. Postnikov, Thanks a lot for the note.
To find the equilibrium lattice constant I satisfied the conditions: *MD.MaxStressTol 0.01 GPa **MD.MaxForceTol 0.0001 eV/Ang* *MeshCutoff 300.0 Ry * On Sun, Mar 22, 2009 at 6:56 AM, <apost...@uni-osnabrueck.de> wrote: > Dear Zubaer, > these differences (which are in fact not that large!) are numerical, > and might come from whatever can be different in different supercells, > namely > i) mesh cutoff > ii) k-points. > Probably you thought already about both factors. However, > > The max force on any atom in the supercells is < 0.0001 eV/Ang. > is not any useful information, because this is simply a formal criterion > when your relaxation stops. More interesting is, how do the forces behave > when the atoms are moved off-center. > Make usuall "eggbox effect" tests and you'll see. > > Also I checked the k-mesh convergence in all the cases. > And what does this mean, exactly? The crucial point is, again, - > how do forces off-equilibrium depend on the k-mesh. > > Best regards > > Andrei Postnikov > > > > Dear SIESTA users, > > I have a question regarding the convergence of FC calculations. I have > > done > > FC calculations with 2 atoms, 8 atoms and 64 atoms (for Gamma point > > frequency) with > > SuperCell_1 0 > > SuperCell_2 0 > > SuperCell_3 0 > > The highest *optical frequencies* I get 520.0 in 2-atom, 517.0 for > 8-atom, > > and 513.0 for 64-atom cases. This is for c-Si. The max force on any atom > > in > > the supercells is < 0.0001 eV/Ang. Also I checked the k-mesh convergence > > in > > all the cases. Why the frequency is changing with the supercell size? > > Also, I did 2-atom basis FC calculations with > > SuperCell_1 0 > > SuperCell_2 0 > > SuperCell_3 0 > > and > > SuperCell_1 1 > > SuperCell_2 1 > > SuperCell_3 1 > > First one produces 2-atom supercell, and 2nd one 54. Here, also I find > > different answers. If I compare the *highest* values in the *.FC files > > (which give* force constants* matrices) in these two calculations, I > find: > > 14.22 and 13.88, respectively. > > Could anyone please suggest what's wrong or what things I would need to > > check? > > Any hint would be really helpful for me. Thanks in advance. > > Zubaer >
