Dear Ruslan, I think it is possible to relax 3000 electron system in a reasonable time. Once I managed to relax a structure consisting of about 1500 Si/C atoms on a pretty old and slow parallel computer in a couple of month without any fancy tricks. I don't really know what is the current performance of Siesta in parallel with only one k-point, you might look into this issue to achieve the best possible speedup (eg. linking against scalapack, machine optimized libraries, order-N calculations). I think more experienced users could give useful hints on this. Anyway, I suggest you to use little tricks to ease convergence. For example, you could prerelax the structure by hand or using a smaller basis set. You may gain experience on smaller systems on what is difference between a coarse optimization and a converged optimization so some little scripts could generate a very reasonable starting point for the large system.
Cheers, Marton On Wed, Jun 22, 2011 at 4:03 PM, Руслан Жачук <[email protected]> wrote: > Dear Siesta users, > > can somebody tell the maximum Si system size with DZP basis set which Siesta > can handle? > Relax in less than a month? Only Gamma k-point. Using cluster, parallel > calculations. > > Regards > Ruslan >
