[CPMD-list] DAVIDSON| NOT ALL ROOTS ARE CONVERGED

[Ari P Seitsonen]

Dear Sergey,

  The roots are indeed not converged within the maximum number of
iterations in the DIIS routine; this was called because the Davidson
algorithm had a suspicious "jump" in the residual at the 13th iteration
(this might be e.g. due to a very small density or gradient when
evaluating the XC kernel; which functional do you use? LDA or LSD?). The
number of DIIS steps can be changed with the input option "DAVIDSON RDIIS"
in the section '&TDDFT'.

  Are you interested in the excitation energies only or do you want to
relax the geometry in the excited state? It seems that the first two
excitations were converged, so if you tried to relax the system in the
first or the second state, you're probably doing fine, otherwise I'd
suggest you to stop the calculation and try to converge all the roots.

    Greetings,

       apsi

-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-
 Ari Paavo Seitsonen / Ari.P.Seitsonen at iki.fi / http://www.iki.fi/~apsi/
 Tel +41 1 635 44 97 / Fax +41 1 635 68 38 / GSM +41 79 719 09 35      
 Anschrift: Physikalisch Chemisches Institut (PCI), Universität Zürich (UniZh)
 Indirizzo: Winterthurerstrasse 190, CH-8057 Zürich
 Address:   Schweiz / Svizzera / Suisse / Svizra / Switzerland

On Fri, 3 Sep 2004, Sergei Lisenkov wrote:

> 
> Dear CPMD authors and users,
> 
> I am performing geometry relaxation in TDDFT mode. In output file I have seen the next message:
> ....
>     5     0.27560             4       0.18967253E-05       79.35
>      6     0.27558             3       0.14962968E-05       40.73
>      7     0.27552             3       0.19436983E-05       41.04
>      8     0.27558             3       0.14962983E-05       78.91
>      9     0.27556             3       0.12534833E-05       39.08
>     10     0.27615             3       0.87845446E-03       42.09
>     11     0.27556             3       0.12531615E-05       78.80
>     12     0.27555             3       0.10936755E-05       40.27
>     13     0.36236             3       0.13448259           42.31
>  >>> RESIDUAL DIIS OPTIMIZATION OF ROOT                         3
>   ITER      ENERGY      SUBSPACE             RESIDUAL        TCPU
>      1     0.34136             1       0.73140030E-04       38.82
>      2     0.34086             2       0.18153372E-04       39.70
>      3     0.34067             3       0.13018395E-04       41.31
>      4     0.34012             4       0.99779496E-05       40.60
>      5     0.34013             4       0.72485720E-05       42.35
>      6     0.33919             4       0.36165369E-04       38.52
>      7     0.34013             4       0.72449107E-05       73.03
>      8     0.34047             4       0.30520350E-03       42.24
>      9     0.33935             4       0.23402906E-04       41.71
>     10     0.34047             4       0.30504691E-03       79.44
>     11     0.34006             4       0.65478783E-05       41.23
>     12     0.33994             4       0.61508352E-05       38.78
>     13     0.33974             4       0.54714752E-05       38.92
>     14     0.33945             4       0.50588145E-05       40.67
>     15     0.33902             4       0.48082073E-05       42.39
>     16     0.33879             4       0.42508821E-05       39.94
>     17     0.33849             4       0.42512206E-05       39.08
>     18     0.33787             4       0.49817643E-05       41.54
>     19     0.33849             4       0.42724747E-05       79.10
>     20     0.33793             4       0.44100927E-05       41.06
>  !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
>  !! DAVIDSON| NOT ALL ROOTS ARE CONVERGED                      !!
>  !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
>                                                                                                                                                              
>  CALCULATION OF TDDFT FORCES
> ....
> 
> What this message foes mean? Should I stop calculations and make any changes?
> 
> Thanks,
>  Best wishes,
>    Sergey
> _______________________________________________
> CPMD-list mailing list
> CPMD-list at cpmd.org
> http://cpmd.org/mailman/listinfo/cpmd-list
>