Excited state geometric optimization and maximum overlap method

[Matt W]

Hi Yifei,

everything you say is correct. We believe that the MOM method is 
implemented correctly, but it does not seem to reliably converge to excited 
states if other states are nearby. Sometimes changing the guess / alpha / 
other settings helps, but nothing systematic.

We are starting to work on TDDFT gradients, but there is no facility at the 
moment. 

In some cases you may be able to use ROKS, but I know very little about 
that method.

Matt

On Wednesday, March 7, 2018 at 6:57:56 PM UTC, Yifei Shi wrote:
>
> Hi,
>
> I'm trying to do an excited state geo opt, but it seems that CP2K doesn't 
> do force evaluation with TDDFT, and the only viable way is to use the 
> maximum overlap method.
>
> However when I try to calculate the excited state energy with MOM for a 
> naphthalene molecule it always goes back to the ground state. I have 
> attached the input file. From TDDFT calculation the first excited singlet 
> state is mostly MO 64 to 65, I tried different ways to reoccupy the alpha 
> MOs, like taking out 64 putting in 65, taking out 63,64, putting in 65,66, 
> but it eventually goes back to the ground state. This did not happen with a 
> simple test on water molecule. I'm not sure what is the problem.
>
> Thanks for any suggestions!
>
> Best,
> Yifei
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20180307/3ce4fd59/attachment.htm>