The ABINIT Users Mailing List ( CLOSED )

["Chun Li " <chun.li@mf.mpg.de>]

Hello, Thanks for your suggestions. 
Could anyone answer me two more questions?

1) I have tried setting iscf1=3 and increasing the k-points mesh density
and the ecut, but the result still not converged with the only warning
message:

scprqt:  WARNING -
  nstep=  120 was not enough SCF cycles to converge;
  maximum energy difference=  5.243E-02 exceeds toldfe=  1.000E-10

Obviously, the result is much far from convergence now. I think that
modifying the mix coefficient (I don't know if it is called correctly. I
mean the percent the results of the former iteration used in the next
iteration. In general, this coefficient should be about 0.2) may work.
But I have not found if there is a parameter like that can be modified.
Could anyone give me some ideas?

2) I think another possibility is because of the PPs. So I want to
modify some parameters in the PP files. Could anyone tell me how to
modify the parameters to obtain softer, medium, and harder PPs,
respectively? Thanks a lot.

I am looking forward to your reply.

Best regards.

Chun Li

-----Original Message-----
From: Masayoshi Mikami [mailto:mmikami@yk.rim.or.jp] 
Sent: 2004年7月25日 17:11
To: forum@abinit.org
Subject: Re: [abinit-forum] about the nstep

Hello,

Please, spare time to check the FAQ, before posting your question:
- Tips to use ABINIT : scfcge : ERROR - Potential-based CG line
minimization not converged
http://www.abinit.org/ABINIT/Infos/FAQ.html#scfcge

Several other things:
- which kind of pseudopotentials for Zn and O ?
  Troullier-Martins ? HGH ? with (or without) Zn-3d as valence ?
- diemac might be set to smaller value, say around 4 to 5.
  But it may be not relevant ...

In passing, I have noticed one more netiquette after reading your
mail...
For Microsoft Outlook* users, in particular:

Please, stick to PLAIN TEXT format, to reduce the size of e-mail as
possible.
(if your e-mail do not have any attachment.)

In this case, your e-mail could have been just 2.4 KB in plain text
format, but it swelled to 30 kB in final form, dueto the HTML format.
In my inbox on my PC, almost 2,000 e-mails for ABINIT, 
and unfortunately not a few e-mails are in HTML format ...
So you can imagine how the plain text format will save our Hard-disk
space.
Further, we can remember that not all of us are Microsoft users.
In fact, I prefer to read e-mails on Unix/Mac systems, to avoid possible
virus for Windows... ;-)

------------------------------------------------------------------------
--
HOW TO SEND YOUR E-MAIL in PLAIN TEXT ?

For MS Outlook Express Users:
kindly see the online help... or, for example
http://email.about.com/cs/oetipstricks/qt/et012401.htm
http://email.about.com/cs/oetipstricks/qt/et082301.htm

For MS Outlook Users:
kindly see the online help... or, for example
http://email.about.com/cs/outlooktips/qt/et052902.htm
http://email.about.com/cs/outlooktips/qt/et070203.htm
http://email.about.com/cs/outlooktips/qt/et060702.htm

For Other Mailer Users:
Kindly consult with online HELP on your mailer..., or visit 
http://email.about.com/od/getthemostoutofemail/
and find a solution to send your e-mail as plain text format.
------------------------------------------------------------------------
--

Good continuation,
Masayoshi

On Sun, 25 Jul 2004 15:48:20 +0200
"Chun Li " <chun.li@mf.mpg.de> wrote:

> Dear all,
>  
> When I tried to calculate the band structure of ZnO, I found that the
> first step (a SCF density computation) did not converge when I set
> nstep=80, but the second step (a non-SCF band structure calculation)
> converged. However, because the density data used in the second step
are
> derived from the first step, so the results could not be reliable. So
I
> tried increasing the value of nstep, but after that, the following
error
> massage lines are appeared in the end of the log file:
>  
> scfcge : ERROR -
>   Potential-based CG line minimization not converged after 13
restarts.
>   Action : read the eventual warnings about lack of convergence.
>   Some might be relevant. Otherwise, raise nband.
>  
>  leave_new : decision taken to exit ...
> FORTRAN STOP  1
>  
> But when I tried to raise the nband, the same error message still
> appeared. I really don't know how to modify the input file then. Can
you
> give me some suggestions? Thanks a lot. (I copied the input file
below)
>  
> Best regards.
>  
> Chun Li
>  
>  
> My input file:
> # Crystalline ZnO
> # Computation of the band structure.
> # First, a SCF density computation, then a non-SCF band structure
> calculation.
>  
> ndtset 2
>  
> #Dataset 1 : usual self-consistent calculation
>  
> ngkpt1  3 3 2
> prtden1  1         # Print the density, for use by dataset 2
> toldfe1  1.0d-10
>  
> #Dataset 2 : the band structure
> iscf2    -2
> getden2  -1
> kptopt2  -4
> nband2   24
> ndivk2   8 16 8 16      # 8, 16, 8 and 16 divisions of the 4 segments,
> delimited
>                         # by 5 points.
> kptbounds2  0.0    0.0     0.0  # Gamma point
>             0.0    0.0     1/2  # A point
>             0.0    1/2     1/2  # H point
>             0.0    1/2     0.0  # K point
>             0.0    0.0     1.0  # Gamma point in another cell.
> prtvol2  10
> tolwfr2  1.0d-20
> enunit2  1         
> acell   6.15956186 6.15956186 9.83980322     
> angdeg  90 90 120
> ntypat 2          
> znucl 30 8        
> natom 4           
> typat 1 2 1 2
> xred                 
>    0.0  0.0  0.0     # Triplet giving the REDUCED coordinate of atom
> Zn1.
>    0.0  0.0  0.3820  # Triplet giving the REDUCED coordinate of atom
O1.
>    1/3  2/3  1/2     # Triplet giving the REDUCED coordinate of atom
> Zn2.
>    1/3  2/3  0.8820  # Triplet giving the REDUCED coordinate of atom
O2.
> #Definition of the planewave basis set
> ecut   32.0         
> #Definition of the SCF procedure
> nstep  100       # Maximal number of SCF cycles
> diemac 9.0