[Opium-talk] many electrons pseudo

[Eric J. W. <ew...@pa...>]

Dear Anglade, 

I have put together an example of how to address your problem.

I would appreciate if other list members could comment on this,
especially if they have further/better advice.

I took the example Au file which has 6s0,6p0,5d10 as valence and
modified it to have 5s2,5p6,6s0,6p0,5d10 as valence.  Now here is
orbital ordering I used:

[Atom]
Au
15
100 2.00  -
200 2.00  -
210 6.00  -
300 2.00  -
310 6.00  -
400 2.00  -
410 6.00  -
320 10.00 -
430 14.00 -
420 10.00 -  # last core state
500 2.00  -  # first valence state
510 6.00  -
520 10.00 -
600 0.00  -
610 0.00  -

Notice a few things:

1. Notice that all states that I want in the core are placed before
the first valence state.  Take a look at my [Pseudo] keyblock:

[Pseudo]
5 1.2 1.2 1.4 1.2 1.2
opt


The '5' means that the last 5 states in the [Atom] keyblock are valence states.
There are 5 rc's in the same order as the valence states.  These rc's
will give you a reasonable gold psp, but I am sure you could improve upon this.

Note, the last two rc's (1.2 1.2) should be the same as the first two since they
have the same 'l' value.

2. Also notice that the second 's' and 'p' orbital have 0 electrons.  This is 
because 6s and 6p are merely place holders for the pseudopotential generation
step.  The pseudopotential generation step never sees these orbitals. Therefore, 
they should not contribute density to the atom.

Now, here is my [Optinfo] block:

[Optinfo]
9.5 10
9.5 10
9.5 10
9.5 10
9.5 10
new

5 qc's and #bessel fxns instead of 3 

Finally, [Configs]

[Configs]
3
#
500 2.00  -
510 6.00  -
520 10.00 -
600 1.00  -
610 0.00  -
#
500 2.00  -
510 6.00  -
520 9.50 -
600 1.50  -
610 0.00  -
#
500 2.00  -
510 6.00  -
520 9.00  -
600 1.00  -
610 0.00  -

Now, in the test configs, it is possible (and a good idea) to test the
pseudopotential the 6s and 6p populated.

I hope this helps.  If you need more information or help, please feel
free to mail me directly or post to the mailing-list, whichever you
prefer. 

Have a great day,

Eric J. Walter





> Hi all,
> 
> I'm new to pseudo potential generationthen I apologise in advance if
> my question seems stupide. Here is my problem:
> 
> I need to generate a pseudo potential for gold with many electrons. I
> need to include at least the 5p. I've took the example in the test
> directory and modified it in the followin way:
> -added 6s1
> -put the electron shells in the energetic order resulting from all
> electron calculations: 430,500,510,520,600,610 instead of the 500,
> 510, 430, 600, 610, 520
> -make a few trials with different qc, rc ...
> In every cases, it turned out that the non local computation is not
> possible. But I'm not able to understand the messages.
> 
> As I've heard that some code are not able to handle more than one
> electron shell of one kind (eg s, p, d or f) I've tryed to remove the
> 500 and 610 from the pseudised orbitals. It was just possible to get
> an other error message:
>  ======================================================================== 
>  Begin NL calculation
>  ======================================================================== 
>  Using the s potential as the local potential
>     !WARNING! : schsl found too many nodes   n,l=   2   1  e= -.1843E+04
>     !WARNING! : schsl found too many nodes   n,l=   3   2  e= -.4118E+03
>  !WARNING! : classical turning point at r=0  800 times in a row. Solve FAILED!
>  
>  iter       Etot             Ebs             Ehxc       de_max   dv_max
>    1  ***************  -15174.7205446 *************** 0.81E+03 0.90E+17
> ****terminal error in scpot,niter=     1   exceeds maxit or e=NAN
> 
>  ======================================================================== 
>  End NL calculation
>  ======================================================================== 
> 
> 
> If anyone's got an idea about what I might do to correct this, I'll be
> very grateful.
> 
> 
> Now, just a suggestion: I've came through the tutorial and the user's
> guide. The wrtten part of the tutorial is quite fine. But I find the
> user's guide a little short on most subject. E.g. what is a pseudo
> core. Things like that may be didactically explained to new users
> through hyperlink.
> 
> Regards
> 
> PM Anglade
> 
> 
> -------------------------------------------------------
> The SF.Net email is sponsored by: Beat the post-holiday blues
> Get a FREE limited edition SourceForge.net t-shirt from ThinkGeek.
> It's fun and FREE -- well, almost....http://www.thinkgeek.com/sfshirt
> _______________________________________________
> Opium-talk mailing list
> Opi...@li...
> https://lists.sourceforge.net/lists/listinfo/opium-talk