Top Page > Browsing

PAO and VPS for Carbon Date: 2014/11/26 20:32 Name: Konstantin Khromov   <k.yu.khromov@gmail.com> Dear prof. Ozaki, The DFT_DATA13 set contains C_PBE13.vps and C7.0.pao files for carbon. The value of parameter radial.cutoff.pao is 7.0 in C7.0.pao and 6.0 in C_PBE13.vps. Although openmx runs without errors with C_PBE13.vps and C7.0.pao used together, I wonder does it make any sense using cutoff radius for PAO larger than that for the pseudopotential? Will this result in just harmless increase in running time and memory consumed or will provide totally wrong physical results? Is it a good idea to regenerate this psudopotential using adpack, just changing radial.cutoff.pao from 6.0 to 7.0 and keeping values of all the other parameters intact? Konstantin Khromov

Page: [1]

Re: PAO and VPS for Carbon ( No.1 ) Date: 2014/11/26 21:40 Name: Artem Pulkin  <artem.pulkin@epfl.ch> Hi Kostya, Sounds like a rhetorical question since meaningful results should not depend on the actual basis set used [provided the basis set is full]. However, you might have noticed that there is a choice of PAOs available for most elements: for carbon those are C5.0.pao C6.0.pao C7.0.pao i.e. you have some freedom in selecting a basis set you find more suitable. For the details, you are welcome to visit following page http://www.jaist.ac.jp/~t-ozaki/vps_pao2013/C/index.html From my point of view, 1. It does make sense to to use a larger cutoff for the PAO than the one for pseudopotential. Since outside the pseudopotential cutoff the atomic wavefunctions still exist and non-zero. 2. You are welcome to test whether particular basis sets in a particular system provide "totally wrong physical results". It may be (spin-orbit gap in graphene) or may not be (elastic properties of graphene) the case, depends on what you expect. As for your third question, I do not know. Hope prof. Ozaki will answer you soon. Artem Re: PAO and VPS for Carbon ( No.2 ) Date: 2015/03/17 20:34 Name: T. Ozaki Hi, The keyword "radial.cutoff.pao" is used only for generation of pseudo-atomic orbitals (*.pao), and never referred in the generation of pseudopotential (*.vps). In the generation of pseudopotential (*.vps), an atom is treated as not confined atom but free atom. Thus, the inconsistency of "radial.cutoff.pao" between *.pao and *.vps does not result in any problem. Regards, TO

Page: [1]