The ABINIT Users Mailing List ( CLOSED )

[Masayoshi Mikami <mmikami@rc.m-kagaku.co.jp>]

Dear Stephane,

Please, post your questions to the forum ML...
This is also a sort of Netiquette. (I should have added it ?)
In fact, it is for you to have more chances to get responses.

To reply to your question, I would like to suggest,
you could study a textbook on solid state physics....
In non-spin-polarized calculation for insulators,
one valence band occupies two electrons.
Thus, you need to divide with two to get the number of
valence bands from the total number of electrons in the unit cell,
i.e. Al2O3 (two formula in the unit cell), # of the valence bands
should be, (3*2+6*3)*2/2=24, not 48.

You might want to read your ABINIT output, you will see
"occ" as 2.0... (kindly read also the manual for input variables,
occ & occopt)

Bon courage,
Masayoshi

P.S. If I receive questions on ABINIT, I will relay to  the forum ML,
as I have done before... Thanks for your kind understanding.

On 2006/04/18, at 17:55, Stephane BECU wrote:

> Dear Masayoshi,
>
> I'm doing RF calculations on Al2O3. In my input files there are two  
> unit
> formulas ie 10 atoms. It seems that the true number of valence bands  
> is 24
> but I do not understand exactly why? In fact I'm using  
> Trouiller-Martins
> pseudo-potentials for both Al and O for which the number of valence  
> bands
> seem to be 3 and 6 respectively. The total number of valence bands  
> shouldn't
> be
> (3*2 + 6*3)*2 = 48 instead of 24 since I have 2 formulas?
>
> Thanks for your help!
>
> Stéphane Bécu
> ST Microelectronics
> 38926 Crolles
> France
>
> -----Original Message-----
> From: mmikami@rc.m-kagaku.co.jp [mailto:mmikami@rc.m-kagaku.co.jp]
> Sent: Sunday, April 16, 2006 6:07 AM
> To: forum@abinit.org
> Subject: Re: Re: [abinit-forum] convergence problen in RF calculation
>
>
> Hi,
>
> More exactly speaking, the nband should be set to the number of valence
> bands, which depend on the valences of the pseudopotentials.
>
> The input files attached in the original post (thanks to follow our
> Netiquette), he wants to compute Sr2MgSi2O7. (It would have been  
> mentioned
> clearly in the original post...) The unit cell contains two formula,  
> which
> lead to "natom 24".
>
> If one treats the valence of Sr as two, nband should be
> (2*2 + 2 + 4*2 + 6*7)*2/2 = 56,
> but if one treats the valence of Sr as ten (i.e. shallow core s/p  
> states
> included as valence), as often in oxides, nband should be
> (10*2 + 2 + 4*2 + 6*7)*2/2 = 72.
> In this case, the ecut should be determined by the hardness of Sr  
> pseudo
> rather than O pseudo. "ecut 40" might be barely OK, but higher ecut  
> may be
> preferred for RF calculations.
> (Please check by yourself !)
>
> Good luck,
> Masayoshi
>
> P.S. "=" should be removed in the input file, e.g., "chkprim=0",  
> "brvltt=0".
>
>
> From: "Yanqing Zheng" <yqzheng1@sh163.net>
> To: "forum@abinit.org" <forum@abinit.org>
> Reply-To: forum@abinit.org
> Subject: Re: [abinit-forum] convergence problen in RF calculation
> Date: Sun, 16 Apr 2006 00:28:54 +0800
> X-mailer: Foxmail 5.0 [cn]
>
> Dear Zeming,
> Set nband to be 56.
> Wish it help.
>
> Good luck!
>
> Yanqing Zheng, Ph.D.
> Shanghai Institute of Ceramics,
> Chinese Academy of Sciences
> Jiading, 201800, Shanghai
> P. R. China
> E-mail: yqzheng@sh163.net
> Fax no.: +86 21 59927184
>
> Dear all,
>    I am trying the Response-function calculation of the effect of an
> homogeneous electric field.  For the first step GS calculation, it is  
> easy
> to reach tolvrs=  1.00E-18 and converged. But, for the second step, a
> non-self-consistent response-function computation of the d/dk  
> perturbation,
> it seems that it is very difficult to converge. After 70 steps, the  
> maximum
> residual only reached about 2E-04 (>>tolwfr= 1.0E-22) and oscillated  
> around
> this value. The tutorial and help of Abinit said it should be use a  
> very
> small tolwfr value. So, I'd like to know how to select a suitable  
> tolwfr for
> different system in RF calcualtion and what input variables influence  
> the
> convergence mostly. Anyone have any suggestion to speed convergence?  
> Thanks
> in advance.
>
> regards,
> Zeming
> ---------------------------
> The input file is as follows.
> #Response-function calculation, with q=0
>
>   ndtset  3
>
> #Ground state calculation
>   kptopt1   1             # Automatic generation of k points, taking
>                           # into account the symmetry
>   tolvrs1   1.0d-18       # SCF stopping criterion
>     iscf1   5             # Self-consistent calculation, using  
> algorithm 5
>
> #Response Function calculation : d/dk
>   rfelfd2   2             # Activate the calculation of the d/dk
> perturbation
>    rfdir2   1 1 1
>     nqpt2   1
>      qpt2   0.0 0.0 0.0   # This is a calculation at the Gamma point
>
>   getwfk2   -1            # Uses as input the output wf of the previous
> dataset
>
>   kptopt2   2             # Automatic generation of k points,
>                           # using only the time-reversal symmetry to
> decrease
>                           # the size of the k point set.
>
>     iscf2  -3             # The d/dk perturbation must be treated
>                           # in a non-self-consistent way
>   tolwfr2   1.0d-22      # Must use tolwfr for non-self-consistent
> calculations
>
>
> #Response Function calculation : electric field perturbation and  
> phonons
>   rfphon3   1        #the calculation of the atomic dispacement
> perturbations
>  rfatpol3   1 24           # All the atoms will be displaced
>   rfelfd3   3
>   rfdir3   1 1 1         # All directions
>   nqpt3   1
>   qpt3   0.0 0.0 0.0   # This is a calculation at the Gamma point
>   getwfk3   -2            # Uses as input wfs the output wfs of the  
> dataset
> 1
>   getddk3   -1            # Uses as input ddk wfs the output of the  
> dataset
> 2
>
>   kptopt3   2             # Automatic generation of k points,
>                           # using only the time-reversal symmetry to
> decrease
>                           # the size of the k point set.
>   tolvrs3   1.0d-8
>     iscf3   5             # Self-consistent calculation, using  
> algorithm 5
>
>
>
> #Definition of the unit cell
> acell 15.09060  15.09060  9.76535        # This is lattice constant
> angdeg 90 90 90
>
> #Definition of the atom types
> ntypat 4
> znucl 38 12 14 8
>
>
> #Definition of the atoms
> spgroup 113 #  P-421m
> chkprim=0
>
> natom 24
> brvltt=0
> natrd 6
> nband 80
>
> typat 1 2 3 4 4 4
> xred
> 0.33869   0.16131   0.50958
> 0.00000   0.00000   0.00000
> 0.13924   0.36076   0.95060
> 0.50000   0.00000   0.14329
> 0.14160   0.35840   0.25931
> 0.07613   0.19200   0.80587
>
> ecut 40  # Maximal kinetic energy cut-off, in Hartree
>
>
> #Definition of the k-point grid
>
> ngkpt  4 4 4
> nshiftk 1         # of the reciprocal space (that form a BCC lattice  
> !),
> shiftk 0.50 0.50 0.50
>
> #Definition of the SCF procedure
> nstep 70          # Maximal number of SCF cycles
> diemac 2.0
------------------------------------------------------------------------ 
--
 Masayoshi Mikami, Ph.D
------------------------------------------------------------------------ 
--
Mitsubishi Chemical Group
  Science and Technology Research Center, Inc.
Address: 1000, Kamoshida-cho Aoba-ku,  Yokohama, 227-8502, Japan
 Phone: 045-963-3834                    Fax: 045-963-3835
 E-mail: mikami.masayoshi@mv.m-kagaku.co.jp
------------------------------------------------------------------------ 
--