[Pw_forum] rdiaghg error # 122 in Phonon Calculation

[Eyvaz Isaev]

Hi, 

> --------
>      iteration #  1     ecut=    30.00 ryd    
> beta=0.70
>      Davidson diagonalization with overlap
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
>      from rdiaghg : error #       122
>      info =/= 0
> 
>
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
> 
>      stopping ...
> --------
This is a very famous error and explaind in the forum
why  it happens. Please have a look at the forum
archive. 
Very briefly, it is due to a fact that an eigenvalue
is not converged  with a machin precision. A solution
is 
to change the diagonalization scheme to 'cg'
(conjugate gradient) method which is believed to be
slower than the Davidson method (by default). 

> 
> What looks interesting for me is that I do not get
> this error message when I use
> the following pseudopotentials in my input file:
> 
> C.pw91-van_ak.UPF
> H.pw91-van_ak.UPF
> 
> C.pz-rrkjus.UPF
> H.pz-rrkjus.UPF
> 
> But these pseudopotentials are not implemented in
> the phcg.x tool as shown in
> the following error message:
> 
> --------
>      nbndx  =    60  nbnd   =    60  natomwfc =   
> 96  npwx   =    3230
>      nelec  =  96.00  nkb   =   160  ngl    =  
> 13407
> 
> 
>
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
>      from cg_readin : error #         1
>      ultrasoft pseudopotential not implemented
> 
>
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
> 
>      stopping ...
> --------
> 
> Is there any reason for the error # 122? What can I
> do? I have also added my  input file:
Plesae see above. Besides, sometime the error served
me as a signal of incorrect structure. 

> 
> &SYSTEM
>                        ibrav = 12 ,
>                    celldm(1) = 15.3030010555 ,
>                    celldm(2) = 1.06841195357 ,
>                    celldm(3) = 0.735119782662 ,
>                    celldm(4) = -0.564967003425 ,
>                          nat = 36 ,
>                         ntyp = 2 ,
>                         nbnd = 60 ,
>                      ecutwfc = 20.0 ,
>                        nosym = .false. ,
ecutwfc=20 (Ry) looks  small for a norm-conserving
pseudopotential, try 30 (but, probably, you did it
according to your message).  
nosym=.false. is taken by default. 

I have also paid attention that you have 20 C and 16 H
atoms which bring total 96 electrons. Thus, 48 bands
should be enough for a semiconductor (insulator)
system. It seems  that empty (conductive) states
(bands) converge somewhat worse compared to valence
bands. So, either you can remove nbnd=60 (it will be
calculated by pw.x) or reduce the number of bands  to
be calculated.

Bests,
Eyvaz.

> /
> &ELECTRONS
>                     conv_thr = 1.0e-6 ,
>                  mixing_beta = 0.6 ,
>                  mixing_mode = 'plain'
> /
> ATOMIC_SPECIES
>  C   12.001  C.pz-vbc.UPF
>  H    1.008  H.vbc.UPF
> ATOMIC_POSITIONS (crystal)
> C        0.082335521   0.328301882   0.982725405
> C        0.112617924   0.222270006   0.837846777
> C        0.048726451   0.037863440   0.894571438
> C        0.077065232   0.925858072   0.748744036
> C        0.987151132   0.253851526   0.191290154
> H        0.132264404   0.472813827   0.933376380
> H        0.186798064   0.279232009   0.672069323
> H        0.153099286   0.985761499   0.584451900
> H        0.962496267   0.339761977   0.307904608
> C        0.417664479   0.671698118   0.482725405
> C        0.917664479   0.671698118   0.017274595
> C        0.582335521   0.328301882   0.517274595
> C        0.387382076   0.777729994   0.337846777
> C        0.887382076   0.777729994   0.162153222
> C        0.612617924   0.222270006   0.662153222
> C        0.451273549   0.962136560   0.394571438
> C        0.951273549   0.962136560   0.105428562
> C        0.548726451   0.037863440   0.605428562
> C        0.422934768   0.074141928   0.248744036
> C        0.922934768   0.074141928   0.251255964
> C        0.577065232   0.925858072   0.751255964
> C        0.512848868   0.746148474   0.691290154
> C        0.012848868   0.746148474   0.808709846
> C        0.487151132   0.253851526   0.308709846
> H        0.367735596   0.527186173   0.433376380
> H        0.867735596   0.527186173   0.066623620
> H        0.632264404   0.472813827   0.566623620
> H        0.313201936   0.720767991   0.172069323
> H        0.813201936   0.720767991   0.327930677
> H        0.686798064   0.279232009   0.827930677
> H        0.346900714   0.014238501   0.084451900
> H        0.846900714   0.014238501   0.415548100
> H        0.653099286   0.985761499   0.915548100
> H        0.537503733   0.660238023   0.807904608
> H        0.037503733   0.660238023   0.692095392
> H        0.462496267   0.339761977   0.192095392
> K_POINTS (gamma)
> 
> Thanks,
> 
> --
> Sigifredo Sanchez-Carrera
> 
> Georgia Institute of Technology
> School of Chemistry and Biochemistry
> Boggs 1-90
> 770 State St. NW
> Atlanta, GA 30332-0400
> 
> Work   (404) 894 6456
> Home   (404) 206 9531
> _______________________________________________
> Pw_forum mailing list
> Pw_forum at pwscf.org
> http://www.democritos.it/mailman/listinfo/pw_forum
> 


__________________________________________________
Do You Yahoo!?
Tired of spam?  Yahoo! Mail has the best spam protection around 
http://mail.yahoo.com