Marcos, You are correct that the local potential scheme is different between SIESTA and that typically used in a PW implementation.
I took a look back at Figure 3 of the SIESTA method paper (on the website). It clearly shows that as the basis size increases from DZP, TZP, TZDP, TZTP, etc. the total energy curves approaches the PW result from _above_. Even at the DZP level, the example used here suggests that there is still a difference of ~0.1 eV/atom in the absolute total energy. So, let me be more precise in my formulation of the question. If the NL pseudopotential is decomposed identicaly (local and non-local parts), can one directly compare the total energies between SIESTA and a PW code, e.g. ABINIT? (Or is the some reference energy that needs to be substracte out.) Also, will the PW total energy be in general lower the SIESTA energy barring an overcompleteness of the local basis set. Thanks for pointing out the work of Jean-Christophe Charlier. If there is a specific published work or online presentation, it would be nice if you can send us a link for more information. On 4/23/07, Marcos Verissimo Alves <[EMAIL PROTECTED]> wrote:
Hi Nichols, While this would not be an authoritative answer to your question, I guess it would be possible to compare the total energies directly, even if you do not make the local potential exactly the same - there are articles on the siesta methodology (I do not remember the references now) in which total energies (as a function of the basis set size) are compared to PW calculations. Also, the way siesta determines the local potential seems to be a bit different from that of usual PW codes. Regarding binding energies, I remember having seen a presentation by Jean-Christophe Charlier in October, in which he had adsorbed atoms in the system and used siesta. After BSSE his results agreed very well with the ones obtained with plane waves. But, of course, this is something to be checked. Regarding your last question, this is something that I have always wondered about. Hopefully someone with more expertise on the list will be able to answer it... Best regards, Marcos > Hi, > > Say that one was able to make everything identical (pseudpotential, local > potential, k-poins, T_el, SCF tol, etc.) > > Is it possible to directly compare the total energies between SIESTA & PW > code (e.g. ABINIT)? What should > be used as a reference energy? Would binding energies between the two codes > agree by including the > BSSE? > > -- > Nichols A. Romero, Ph.D. > 1613 Denise Dr. Apt. D > Forest Hill, MD 21050 > 443-567-8328 (C) > 410-306-0709 (O) > -- Dr. Marcos Verissimo Alves Post-Doctoral Fellow Condensed Matter and Statistical Physics Sector International Centre for Theoretical Physics Trieste, Italy -------- I have become so addicted to vi that I try to exit OpenOffice by typing :wq!
-- Nichols A. Romero, Ph.D. 1613 Denise Dr. Apt. D Forest Hill, MD 21050 443-567-8328 (C) 410-306-0709 (O)