[DFTB-Plus-User] Excited State calculation: Multiple StateOfInterest and format of TRA.DAT

[Dirk Ziemann]

Dear Ben,

thank you very much for the quick response and your extensive help.

If I follow your explanation for the TRA.DAT file I have no further 
problems with the output.

Concerning the exciton calculation, I am looking into a ZnO nanocrystal 
(>1000 atoms). For this system I need the excition transition charges 
(Mullikan/loewdin) to calculate the relaxation of excitons in the range 
between the lowest excition and around 1 or 2 eV above.
I think I can construct the transition charges from the informations of 
TRA.DAT, but there are only 4 significant digits, which might lead to 
some inaccuracy.
Maybe you have any idea how I can simply extract the information to 
build these transition charges or the wave function?

Thank you again and best wishes         Dirk

On 29/06/17 16:23, Ben Hourahine wrote:
> Hello Dirk,
>
> at the moment calculation of charges for multiple states simultaneously
> is missing (I'll add it to the feature request list).
>
> Could I ask what you are calculating that needs this data? If if is for
> multiple excited states in a spectra (for example plasmonic features),
> the Casida approach might not be the most efficient way to obtain this
> information (there may be something in the 18.1 release in this
> direction, based on coupled perturbed or Sternheimer equations).
>
> The reason for the behaviour in the TRA.DAT is due (probably) to
> symmetry in your system. What is printed is the dominant single particle
> transition contributing to the many particle excited states. But as a
> result of degeneracy in a group of single particle states, the relative
> unitary transformation of the single particle states is undefined. The
> eigensolver has chosen a set of degenerate vectors such that the same
> single particle excitation occurs in multiple degenerate many particle
> transformations.
>
> In principle, it would be possible to choose a unitary transformation
> such that the single particle transition only contributes in one of
> these transitions, but this would not change the energy of the
> transitions, only the relative dipole intensities within the degenerate
> many particle transitions.
>
> If you notice, sum(Weight^2) over the degenerate group = 1. At some
> point there might be a change in the output of DFTB+ to show the sum and
> degeneracy of transitions in these cases, instead of the individual
> transitions.
>
> If you would like me to investigate further, you can send the input off
> list.
>
> Regards
>
> Ben
>
>
> On 28/06/17 15:49, Dirk Ziemann wrote:
>> Dear all,
>>
>> I have two questions concerning the excited state calculation.
>>
>> I am interested in the Mulliken charges of many different excited
>> states. If I specify a single state of interest at StateOfInterest
>> everything works fine, but there is an error for multiple numbers. Is it
>> possible to get the Mulliken charges for many different states at
>> once?            (I need >1000 states)
>>
>> In my TRA.DAT files I see the following lines (see below). Why does a
>> transition appears multiple times?
>>
>> As far as I have checked it, this happens only to the first transition.I
>> have a feeling that it is related to the value of OscillatorWindow. If I
>> choose a high value this does not happen. The smaller the value the
>> higher the number of "multiple first transitions".
>>
>>
>> Thank you for your help and best wishes   Dirk
>>
>>
>> everything the same in dftb_in.hsd except the  ExcitedState-part
>>
>> example 1:
>>
>>    Energy      1          3.625 eV   S
>>
>>    Transition         Weight       KS [eV]
>>   =============================================
>> 13723   -> 13908     0.7361       4.616
>> 13723   -> 13908     0.5047       4.616
>> 13723   -> 13908     0.3897       4.616
>> 13723   -> 13908     0.2272       4.616
>>   1011   -> 20029     0.0000      39.368
>>   1011   -> 20028     0.0000      39.368
>>
>> dftb_in.hsd
>>
>> ExcitedState {
>>      Casida {
>>          NrOfExcitations = 100
>>          StateOfInterest = 0
>>          Symmetry = singlet
>>          OscillatorWindow = 1E-1
>>          WriteStatusArnoldi = Yes
>>          WriteTransitions = Yes
>>     }
>> }
>>
>> //==============================================//
>>
>> example 2
>>
>>    Energy      1          3.754 eV   S
>>
>>    Transition         Weight       KS [eV]
>>   =============================================
>> 13749   -> 13754     0.8301       3.754
>> 13749   -> 13754     0.4022       3.754
>> 13749   -> 13754     0.3398       3.754
>> 13749   -> 13754     0.1559       3.754
>> 13749   -> 13754     0.0687       3.754
>> 13749   -> 13754     0.0550       3.754
>> 13749   -> 13754     0.0417       3.754
>>    215   -> 20033     0.0000      39.768
>>    215   -> 20035     0.0000      39.768
>>
>> dftb_in.hsd
>>
>> ExcitedState {
>>      Casida {
>>          NrOfExcitations = 30
>>          StateOfInterest = 0
>>          Symmetry = singlet
>>          OscillatorWindow = 1E-2
>>          WriteStatusArnoldi = Yes
>>          WriteTransitions = Yes
>>     }
>> }
>>
>>
>>