[Wien] equivalent-core approximation

Michael Gurnett michael.gurnett at kau.se
Mon Feb 28 14:46:41 CET 2005


Well, I've been trying all weeken and still no luck (beginning to think it's 
not possible, although just as likely that my input files are wrong). I 
checked :NEC01 as Peter suggested and found that that the last column showed 
approximately 1.002 (I guess this means 0.2 %). The files I have changed are 
now as follows

case.in1c
        All Ge atoms
 0.30    3  0      (GLOBAL E-PARAMETER WITH n OTHER CHOICES, global 
APW/LAPW)
 2   -2.82      0.000 CONT 1 *
 0    0.30      0.000 CONT 1
 1    0.30      0.000 CONT 1

* the standard SCF had -1.82 so I tried to do as the faq and make sure that 
Ge3d wouldn't be found

    All Li atoms

 0.30    2  0      (GLOBAL E-PARAMETER WITH n OTHER CHOICES, global 
APW/LAPW)
 0    0.30      0.000 CONT 1
 0   -3.50      0.005 STOP 1

case.in2c

removed 10 electrons*number of Ge in Ne

case.inc

All Ge atoms
 9 0.00     NUMBER OF ORBITALS (EXCLUDING SPIN), SHIFT
1,-1,2               ( N,KAPPA,OCCUP)
2,-1,2               ( N,KAPPA,OCCUP)
2, 1,2               ( N,KAPPA,OCCUP)
2,-2,4               ( N,KAPPA,OCCUP)
3,-1,2               ( N,KAPPA,OCCUP)
3, 1,2               ( N,KAPPA,OCCUP)
3,-2,4               ( N,KAPPA,OCCUP)
3, 2,4               ( N,KAPPA,OCCUP)
3,-3,6               ( N,KAPPA,OCCUP)

All Li atoms

 1 0.00     NUMBER OF ORBITALS (EXCLUDING SPIN), SHIFT
1,-1,0               ( N,KAPPA,OCCUP)


The main problem that occurs is still the QTL-B errors. I've been using a 
mixing factor of 0.05 and PW also 0.05, but still no luck.

As my lapw0 takes 40 minutes to run It takes along time to make changes and 
test, so I had a quick go at trying to get mpi to work again since it was 
stated that the latest version had fixes. However, I was still not able to 
get this part working.

Michael
----- Original Message ----- 
From: "Peter Blaha" <pblaha at zeus.theochem.tuwien.ac.at>
To: <wien at zeus.theochem.tuwien.ac.at>
Sent: Friday, February 25, 2005 9:29 PM
Subject: Re: [Wien] equivalent-core approximation


Check the amount of leakage and the proper number of electrons (:NEC01)

Eventually try a "shift" in case.inc (see UG, this shift was intended for
putting the 4f states into the core)

> TRouble is at the moment I cannot even place the Ge 3d electrons into core
> follwoing Peters instructions. I get to the fourth scf and it crashes. 
> Usually
> not giving an error. Most reasantly the error was the following
>
> --------------------------------------
>
>
> Input/Output Error 148: Invalid character
>
>   In Procedure: insld
>        At Line: 95
>
>      Statement: Formatted READ
>           Unit: 20
>   Connected To: Li_Ge3x2.struct
>           Form: Formatted
>         Access: Sequential
> Records Read   : 495
> Records Written: 0
>
> Current I/O Buffer:
>
>   2      NUMBER OF SYMMETRY OPERATIONS
>            !
>
>
> End of diagnostics--------------------------------------
>
> What I have done is starting with a normal calculation reaching 
> convergance,
> Than I did the following
>
> 1.    removed the following 2 lines in case.in1c for all Ge atoms
>
>     2   -1.82      0.010 CONT 1
>     2    0.30      0.000 CONT 1
>
> and also changed the 4 to a 2 to reflect the removal of these two lines
>
> 2.    changed NE in case.in2c to reflect that I'm removing 10 electrons in
> every Ge
>
> 3.   Added the follwoing two line in case.inc, placing Ge 3d in core
>
> 3, 2,4               ( N,KAPPA,OCCUP)
> 3,-3,6               ( N,KAPPA,OCCUP)
>
> and also increasing the number of orbitals by 2
>
>
> I've tried decreasing the mixing factor, and the PW mixing to 0.05 and 0.1
> respectively. I've also tried using PRATT.
>
> But I'm still getting crashes
>
> Any help would be great.
>
> Mick
>
> ----- Original Message ----- From: <buero at luitz.at>
> To: <wien at zeus.theochem.tuwien.ac.at>
> Sent: Friday, February 25, 2005 6:19 PM
> Subject: Re: [Wien] equivalent-core approximation
>
>
> On Thu, 24 Feb 2005, Michael Gurnett wrote:
>
> > Thank you Peter for your very indepth reply. This method you have is 
> > what I
> > originally tried. Howerver, once I made all the changes I found that the 
> > SCF
> > would oscilate even with a mixing factor of 0.01 and would eventually 
> > crash.
> > If this is the best method I will try it again, and with any luck I can
> > continue on from there.
>
> When I met such a oscillating condition in my core hole calculations, what
> often helped was a "step by step approach": instead of 1 full core hole, I
> started with 0.5 (half a core hole) ran SCF then increased to 0.8, rerun
> SCF and finally run SCF on a 1.0 corehole. For the systems I was
> interested in this worked amazingly well.
>
> Best
>   Joachim
>
> Mit freundlichen Grüssen
>   Joachim Luitz
>
> --
> luitz.at | interfacing art, science and technology
>
> Dipl.-Ing. Joachim Luitz KEG
> Wohlmuthgasse 18 . A-3003 Gablitz . T +43 2231 61254 | Fax +43 2231 612544
> buero at luitz.at . http://www.luitz.at . callto://jluitz
> _______________________________________________
> Wien mailing list
> Wien at zeus.theochem.tuwien.ac.at
> http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien
>


                                      P.Blaha
--------------------------------------------------------------------------
Peter BLAHA, Inst.f. Materials Chemistry, TU Vienna, A-1060 Vienna
Phone: +43-1-58801-15671             FAX: +43-1-58801-15698
Email: blaha at theochem.tuwien.ac.at    WWW: 
http://info.tuwien.ac.at/theochem/
--------------------------------------------------------------------------

_______________________________________________
Wien mailing list
Wien at zeus.theochem.tuwien.ac.at
http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien






More information about the Wien mailing list