Top Page > Browsing

A few questions about adpack 2.1 Date: 2011/01/25 03:33 Name: G. Faccin Hi, I've got a few questions regarding adpack 2.1, wonder is someone could help. 1 - When choosing "eq.type", according to the manual, one can use "sch", "sdirac" and "dirac". However, reading the source files, there are also "dirac1", "dirac2" and "sdirac2" available. In fact, many pseudos from the 06 database are created using those last two flags. My question: what is the difference between "dirac", "dirac1" and "dirac2", and between "sdirac" and "sdirac2"? 2 - When choosing "vps.type", according to the manual, one can use "tm" or "bhs". On the source files, however, there are also "MR", "EDPP" and "GVPS". What do those other options do? 3 - I've been trying to produce reliable calculations with Openmx for some time already, for a system composed of Fe and Co. To produce good results, one needs good pseudopotentials. So far I've ran a few thousand adpack jobs testing different cutoff radius for the pseudopotentials, in order to generate a good pseudo with the TM scheme for transition metals. I've managed to get pseudos similar in quality to those of the 06 database. However, while googling the subject, I've found a link for a 2011 database: http://www.jaist.ac.jp/~t-ozaki/vps_pao2011/Mn/index.html in this link, there's a very good pseudopotential for Mn: http://www.jaist.ac.jp/~t-ozaki/vps_pao2011/Mn/Mn_PBE11.vps which was created using "vps.type MBK". To my knowledge, this "mbk" flag is not available in the current public version of adpack. So my question here is: what is MBK, and... could it be made publicly available? So far I've been unable to reproduce the results I need for Fe and Co with Openmx. I wonder if perhaps with a better pseudopotential things would work out.In particular, bcc iron and possibly hcp cobalt are hard to reproduce, as explained in PRB 69-195113; however, I cannot stop wondering if it won't work out with pseudopotentials as good as those produced by the MBK flag.

Page: [1]

Re: A few questions about adpack 2.1 ( No.1 ) Date: 2011/02/06 23:54 Name: T.Ozaki Hi, > what is the difference between "dirac", "dirac1" and "dirac2", and between > "sdirac" and "sdirac2"? "1" corresponds to an approximation that the small component of wave functions is neglected, while "2" takes it into account. Since both the treatments give a similar result, the default setting "dirac" and "sdirac" are automatically regarded as "dirac2" and "sdirac2" in the code. > there are also "MR", "EDPP" and "GVPS". What do those other options do? Those parameters are just ruins corresponding to our trials for methodological development. > However, while googling the subject, I've found a link for a 2011 database: We have been trying to develop a reliable database of PAO and VPS which is NOT compatible with OpenMX ver. 3.5. Probably, the database will be officially released around this Summer together with the latest version of OpenMX and ADPACK. Best regards, TO

Page: [1]