Dear Gregor, gipaw is based on paw, as long as it stays so it will always require paw description of the atom.. if you are using a traditional norm conserving or ultrasoft pseudopotential that means you need this description to be calculated and written, which will make it a 'gipaw pseudopotential' If you are using a paw pseudopotential however, everything is already calculated. unfortunately it has not been the default to write down this necessary information. in the future i would like to make it default, such that every paw pseudopotential file already is gipaw-suitable. it is just that there are some details to work out,like,you may want the pw.x to use a different set of projectors (possibly cheaper ones) than gipaw.x part..so we should keep this freedom etc.
about not having the paw set for the atoms you are not interested in..what would you propose? the induced magnetic field at a point depends on the induced current in all space..do you suggest a space cut off? or calculating the current response of these atoms we are not interested in without paw constructions? instead there is the 'converse approach' you may want to look at, that sounds more efficient if you are interested in nmr parameters of one atom in huge molecule, for example.. we can hear more from Davide about it. emine kucukbenli, phd student, sissa, italy ---------------------------------------------------------------- SISSA Webmail https://webmail.sissa.it/ Powered by Horde http://www.horde.org/