Pseudopotentials, OPIUM, and lmax

Moderators: torrent, jzwanzig

Post Reply
Posts: 3
Joined: Wed Jun 09, 2010 4:18 pm

Pseudopotentials, OPIUM, and lmax

Post by dcellucci » Mon Jul 26, 2010 11:41 pm

Recently I tried my hand at generating my own pseudopotentials for a tin oxide (SnO2) calculation. I started with trying ionized states, but quickly realized that I first needed to at least match the accuracy of the pseudopotentials offered on the website with neutral atoms before graduating to more complex calculations.
In the process of generating these pseudopotentials with the OPIUM code, I came across two things that puzzled me.

One, is there a need to specify if a potential was generated using OPIUM in abinit? What if I use partial core correction? Does Abinit require a special flag if the pseudopotential is using a specific type of PCC (in this case Louie, Froyen, and Cohen) like it requires its IXC to match the pseudopotential's IXC?

Two, I keep seeing in the sample pseudopotentials (the GGA_FHI pspots) that the lmax is defined as being equal to 3, even if the f-orbital isn't and definitely shouldn't be used. This occurs in the tin, the oxygen, even the hydrogen. Is lmax = 3 a flag for something else, something more significant?

Thank you for your time,

User avatar
Posts: 504
Joined: Mon Aug 17, 2009 9:25 am

Re: Pseudopotentials, OPIUM, and lmax

Post by jzwanzig » Tue Jul 27, 2010 3:39 pm

if you use the "fhi" keyword in opium you will get a file readable by abinit, no extra input switches in abinit need to be used (you don't need to specify ixc as it is read from the pseudopotential file, this variable is only in abinit if you want to (dangerously) over-ride the xc choice that the pseudopotentials are using). There is no "pcc" switch in abinit, all the pcc stuff has already been done in making the pseudopotential and its effect will be contained in that file.

Next, for lmax from the FHI code, it does mean that the pseudopotentials were made with for example an L=3 channel. It is not necessary to resitrict yourself to occupied, bound states in making pseudopotentials, so you can try to increase flexibility by adding more channels. In the NCPP world you typically have just a single projector per angular momentum, so in order to have more flexibility you need more L states. In PAW and USPP you typically have multiple projectors per angular momentum, so you would in the case of H add flexibility by using 2 or more s projectors, and then maybe also p projectors and so forth.

hope it helps,
Josef W. Zwanziger
Professor, Department of Chemistry
Canada Research Chair in NMR Studies of Materials
Dalhousie University
Halifax, NS B3H 4J3 Canada

Post Reply