Hi, I am also initiating myself in this theme. I was told (someone with more experience can correct me, please) that the optical k-mesh is independent of kgrid-_Monkhorst_Pack. I was making recently DFT calculations for graphene nanoribbons and one recommended me to use a much finer optical grid along the translational direction of the ribbon. For example, if the ribbon extends along z-direction and I use a kgrid of 1 x 1 x 24, so the optical grid should be approximately the double, around 1 x 1 x 48 to guarantee better convergence for the optical calculations.
best Rocha On 24 July 2013 13:23, FEDERICO IORI <[email protected]> wrote: > Hi all. > > I have a question concerning the use of the optical K mesh in siesta to > calculate the dielectric function: > > is the optical K mesh independent from the number of the usual kgrid mesh > used for the relaxation of the system? > > If I have a system which density matrix has been calculated with 32kpt > let's say, then the way I choose the optilcal k mesh is independent from > this or not? > > The optilcal k mesh define the numeber of k points which enters directly > into the sum to calculate through the "fermi golden rule" the eps2, the > imaginary part of the dielectric function. So since the spectra has to be > converged w.r.t number of bands and number of k points, I presume i can > augment each time the Optical K mesh until the convergence is reached. in > this way I presume the choose of optical K mesh is independent from the > kgrid mesh used before in the scf part. > > Is this correct ? > > Thanks a lot and enjoy the hot summer time! > :) > > -- > Dr. Federico Iori, PhD! > > Dipartimento di Scienze e Metodi per l' Ingegneria (DISMI) > Universita' di Modena e Reggio Emilia > Via Amendola,2 > Pad. Morselli > 42122 Reggio Emilia (Italy)
