This thread is locked.Only browsing is available.
Top Page > Browsing
Date: 2012/08/04 16:09
Name: Vikas Sharma   <>

In most of the example the scf.Kgrid is 11 11 11 (n1,n2,n3) for example in case of Febcc-EvsLC.dat and many more on the examples of vps and pao page.
The k-points in OpenMX are genereated according to the Monkhorst-Pack method [PRB 13 (1976) 5188.]
However, in their paper the authors suggested to use the even numbers especially for cubic systems and hence, I feel safer always to use the even number while limiting the kgrid size in my calculation using this scheme. I think, instead of 11 11 11 I would be more optimistic about my calculations results if I would use 10 10 10 or 12 12 12 for (n1, n2, n3).

Vikas Sharma
Page: [1]

Re: scf.Kgrid ( No.1 )
Date: 2012/08/28 16:20
Name: T.Ozaki


Thank you for your notification.
We also checked the odd case (11x11x11) and the even case (12x12x12)
occasionally, but didn't see large difference in the band structure and
the structural parameters. But anyway, thank you again.



Page: [1]