On Wed, 2009-04-22 at 07:21 -0700, ?lvaro Alves wrote:
> > Hi all. I build a fully relativistic pseudopotencial for cobalt. When > testing it I found the following results for lattice parameters: a = > 2.44 A, c = 4.03 A and c / a = 1.65. The experimental values are 2.51 > A, 4.07 A and 1.62. I want to improve these results. Do you have any > suggestions? My input is below. > alvaro, you can try sacrificing five chickens or a goat, bow five times to mecca or pray 5 ave marias. other than that, i'd say that you have a less than 3% overbinding for the lattice parameter and 1% deviation in c and and since both errors are towards the same direction you have an error of 2% in c/a. to the best of my knowledge, this is pretty much of what people report as typical error margins for a gradient corrected functional, only that i remember seeing underbinding more commonly than overbinding (the latter would be more common in LDA). at this point, the black magic of pseudopotential generation begins. you can twiddle with cutoffs and electron configuration and see if this shifts the results closer to what you want, while not harming transferability (how transferable is this potential anyways? did you test it?). cheers, axel. -- ======================================================================= Axel Kohlmeyer akohlmey at cmm.chem.upenn.edu http://www.cmm.upenn.edu Center for Molecular Modeling -- University of Pennsylvania Department of Chemistry, 231 S.34th Street, Philadelphia, PA 19104-6323 tel: 1-215-898-1582, fax: 1-215-573-6233, office-tel: 1-215-898-5425 ======================================================================= If you make something idiot-proof, the universe creates a better idiot. -------------- next part -------------- An HTML attachment was scrubbed... URL: http://www.democritos.it/pipermail/pw_forum/attachments/20090422/4ee95d57/attachment.htm