[Pw_forum] ld1.x & US requires...

Paul M. Grant w2agz at pacbell.net
Thu Mar 12 05:29:37 CET 2009

Something bad did happen...

The post-processing dos.x tool got the d-s (or s-d) ordering correctly,
The LDA+U "starting_ns_eigenvalues" coaxing input does not lead to the
"correct" spin configuration.  Have a look at the PP header change to
Fe.pz.nd.UPF below.  Try it out on Example 25 and get back to me.

Designing PPs is a black art, as many have pointed out.  It would be great
if there were a resource that contained some examples of ld1.x input one
could play around with.  There are, of course, considerable
databases...espresso, berlin, opium, Vanderbilt...whatever, but, as far as I
could find, none provided a large number of input examples one could play
around with.

Oh...very strange...the d-s ordering in all Cu PPs I've found, including
those I've "designed," is the reverse of all those other TMs in the QE
database.  I use Cu a lot, including tweaking the "sne's."  Is there an
additional issue I should I should be aware of?

Generated using Andrea Dal Corso code (rrkj3)
Author: Andrea Dal Corso   Generation date: unknown  (reversed s & d order)
Info:   Fe LDA 3d7 4s1 RRKJ3 US
    1        The Pseudo was generated with a Scalar-Relativistic Calculation
  1.70000000000E+00    Local Potential cutoff radius
nl pn  l   occ               Rcut            Rcut US             E pseu
4S  1  0  1.00      2.00000000000      2.20000000000      0.00000000000
4S  1  0  0.00      2.00000000000      2.20000000000      0.00000000000
4P  2  1  0.00      2.20000000000      2.30000000000      0.00000000000
4P  2  1  0.00      2.20000000000      2.30000000000      0.00000000000
3D  3  2  7.00      1.60000000000      2.20000000000      0.00000000000
3D  3  2  0.00      1.60000000000      2.20000000000      0.00000000000

   0                   Version Number
  Fe                   Element
   US                  Ultrasoft pseudopotential
    T                  Nonlinear Core Correction
 SLA  PZ   NOGX NOGC   PZ   Exchange-Correlation functional
    8.00000000000      Z valence
  -41.49093103800      Total energy
  0.0000000  0.0000000 Suggested cutoff for wfc and rho
    2                  Max angular momentum component
  957                  Number of points in mesh
    2    6             Number of Wavefunctions, Number of Projectors
 Wavefunctions         nl  l   occ
                       3D  2  7.00
                       4S  0  1.00

Paul M. Grant, PhD
Principal, W2AGZ Technologies
Visiting Scholar, Applied Physics, Stanford (2005-2008)
EPRI Science Fellow (Retired)
IBM Research Staff Member Emeritus
w2agz at pacbell.net

-----Original Message-----
From: pw_forum-bounces at pwscf.org [mailto:pw_forum-bounces at pwscf.org] On
Behalf Of Paolo Giannozzi
Sent: Monday, March 09, 2009 7:45 AM
To: PWSCF Forum
Subject: Re: [Pw_forum] ld1.x & US requires...

On Mar 9, 2009, at 6:34 , Paul M. Grant wrote:

> Why did I bother doing this exercise?  Simply because dos.x  
> apparently arranges the notation for the projected pdos output  
> according to the sequence of states listed in the PP.  For Cu, it  
> is 3d-4s, whereas for Ni, Co, Fe, and Mn, it is the reverse.

if this is the problem, you can reverse the positions of 3d and 4s  
atomic states in
the pseudopotential file. Nothing bad should happen.

Paolo Giannozzi, Democritos and University of Udine, Italy

Pw_forum mailing list
Pw_forum at pwscf.org

More information about the users mailing list