[CPMD-list] Excited geometry optimisation : buggy behaviour when using the SAOP XC potential [CPMD 3.11.1]

[Latévi Max LAWSON DAKU]

Dear Juerg Hutter,

Greetings, and many thanks for your kind answer.
I do understand that addressing this issue will
require time.
I have started looking into the code, gently.. This indeed
takes time :-)

All the best,
Max

Juerg Hutter wrote:
> Hi
>
> I'm very sorry that I myself am not able to help you.
> Due to a large teaching load and other administrative
> duties I don't have time to look into this problem.
> Maybe somebody else will step forward and will commit
> some time to this problem.
>
> best regards
>
> Juerg Hutter
>
> ----------------------------------------------------------
> Juerg Hutter                   Phone : ++41 44 635 4491
> Physical Chemistry Institute   FAX   : ++41 44 635 6838
> University of Zurich           E-mail: hutter at pci.unizh.ch
> Winterthurerstrasse 190
> CH-8057 Zurich, Switzerland
> ----------------------------------------------------------
>
>
> On Fri, 24 Nov 2006, Latévi Max LAWSON DAKU wrote:
>
>> Hi,
>>
>> I've noticed that excited geometry optimisations do not
>> work if the SAOP potential is used: at a given step of the
>> optimisation process, the  part of the code dealing with
>> the solution of the TDA-TDDFT eigenvalue problem stop
>> working correctly:
>>
>>     ================================================================
>>     ==            DAVIDSON DIAGONALISATION OF TDDFT MATRIX        ==
>>     ================================================================
>>      ITER      STATES      SUBSPACE             RESIDUAL        TCPU
>>         1           0             2        .00000000E+00        2.59
>>         2           0             2        .00000000E+00         .01
>>         3           0             2        .00000000E+00         .01
>>         4           0             2        .00000000E+00         .01
>>         5           0             2        .00000000E+00         .01
>>         6           0             2        .00000000E+00         .01
>>         7           0             2        .00000000E+00         .01
>>    ....
>>    ....
>>       150           0             2        .00000000E+00         .01
>>     >>> RESIDUAL DIIS OPTIMIZATION OF ROOT                         1
>>      ITER      ENERGY      SUBSPACE             RESIDUAL        TCPU
>>         1         NAN             1                  NAN        1.23
>>         2         NAN             2                  NAN        1.24
>>    ....
>>    ....
>>       100        -NAN             1                 -NAN        1.24
>>     !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
>>     !! DAVIDSON| NOT ALL ROOTS ARE CONVERGED                      !!
>>     !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
>>     ***     TD_OS| SIZE OF THE PROGRAM IS  135612/ 869600 kBYTES ***
>>     LB:-------------------------------------------------------------
>>     TD_OS_BERRY|            dE=       NAN eV            f=       NAN
>>     TD_OS_BERRY|            x:      -NAN  y:      -NAN  z:      -NAN
>>     TD_OS_BERRY|            dE=      -NAN eV            f=      -NAN
>>     TD_OS_BERRY|            x:       NAN  y:       NAN  z:       NAN
>>     LB:-------------------------------------------------------------
>>
>>
>> I would say that a stack get corrupted, but I do not know the
>> internal of the program.
>>
>> Of course, the nuclear gradients can no more be computed
>> and the program stops.
>>
>> I've set up a simple test-case that helps illustrate what is going
>> on. Please see the attached files.
>>
>>    - "ch3cn.s1_spec.in": input of the excitation calculations
>>      that  have to be carried first.
>>    - "ch3cn.s1_opt.in": input for performing the excited states
>>      calculations.
>>
>> The files "ch3cn.s1_{spec,opt}.out" are the outputs which I
>> have obtained.
>>
>> For the larger system which I am studying I do not observe this
>> behaviour if I remove from the &TDDFT..&END section the statement
>> "TD_METHOD_A SAOP" and thus let the program use the PBE
>> XC potential. I do need the SAOP potential. Is there a workaround ?
>> Are other asymptotically corrected potential implemented in CPMD ?
>>
>> Thanks in advance.
>>
>> Regards,
>> Max
>>
>> -- 
>> ***********************************************
>> Latevi Max LAWSON DAKU
>> Departement de chimie physique
>> Universite de Geneve - Sciences II
>> 30, quai Ernest-Ansermet
>> CH-1211 Geneve 4
>> Switzerland
>>
>> Tel: (41) 22/379 6548 ++ Fax: (41) 22/379 6103
>> ***********************************************
>>
>>

-- 
***********************************************
Latevi Max LAWSON DAKU
Departement de chimie physique 
Universite de Geneve - Sciences II
30, quai Ernest-Ansermet  
CH-1211 Geneve 4 
Switzerland 

Tel: (41) 22/379 6548 ++ Fax: (41) 22/379 6103
***********************************************