----- Najmeh Honari <najmeh.honari...@gmail.com> a écrit : Dear Andrei, In terms of converging our simulations for k-mesh values, how do we know > whether the fluctuations of the "total energy" and > > "Cell Vector Mod" > are big or small? Dear Najmeh,do you refer to the recently posted results of Sunetra Das (are they "your simulations"?),or to something else? - Just in order to define what we are talking about.I'd say, the general approach is like this:- you converge the total energy (differences, see my previous comment) vs the k-ponts to the accuracy you need for your particular problem. - once you decide that the convergence is good, you write down the lattice parameters.To look at how do the lattice parameters fluctuate on increasing the k-mesh densitydoes not make much sense, because the lattice parameter is not a variational property. Best regardsAndrei Postnikov
On 14 July 2018 at 03:02, Andrei Postnikov <andrei.postni...@univ-lorraine.fr> wrote: Dear Sunetra, 1) your system is 2D; your "Cell Vector Mod" is presumably the lattice parametouler in (x,y);how about the third parameter? Is there a chance that you have not fix it so that itfluctuates under the effect of stress tensor component which is a numerical noise? 2) "Is the free energy convergence enough?" - the answer fully depends onyour objective; normally this is not the absolute value of energy that you should care about -it has no special meaning anyway - but the energy difference between cases(geometries, magnetic solutions, etc.) which you want to compare in your study. Best regards Andrei Postnikov ----- Sunetra Das <sunetra.das...@gmail.com> a écrit : >Dear experts, I am trying to converge my system unitcell structure for k - mesh values. Even though the Free Energy value has converged for less denser mesh, the unit cell parameter has not converged for even 21x21x1 k mesh. The system is a 2D one with 3 atoms in the unit cell. I am attaching the file containing the k values and the corresponding free energy and cell parameter values along with the plotted graph. please have a look and help me. Is the free energy convergence enough? I have used systems before where I have checked if both the parameters have converged for a given k mesh. Thank you. Regards,Sunetra Das