Maximum Overlap Method and Geometry Optimization

[Sergey Chulkov]

Hi Jen,

sure, simply read the ground state wave function and specify the 
appropriate excitation in the MOM section. Alternatively, you can start the 
calculation using the atomic guess, if you wish, but it is usually less 
stable. 

It is also possible to restart the MOM calculation using an MOM wave 
function for the same excited state. In this case simply read the 
appropriate wave function and provide the empty MOM section (&MOM / &END 
MOM). The MOM procedure will then select those orbitals which overlap most 
with the occupied orbitals from your restart file.

Sergey

On Thursday, September 15, 2016 at 6:35:25 PM UTC+1, Jen E. wrote:
>
> Thanks very much! I am in the process of trying out the recommended steps. 
> I've managed to converge the ground state and determine which orbitals I 
> want to occupy. Is the proper next step to read back in the converged 
> ground state wavefunction and then add the MOM tag to the input and re-run 
> the calc? 
>
> Best, 
> Jen
>
> On Wednesday, September 14, 2016 at 2:10:04 AM UTC-4, Matt W wrote:
>>
>> Hi,
>>
>> the geo_opt bug was fixed shortly after the posts above - so a recent 
>> executable should work fine (the bug was actually for MOM in combination 
>> with the MO_CUBES keyword).
>>
>> We don't really have a definite protocol for running the method beyond: 
>> Converge ground state; Check orbitals; Run MOM calc with appropriate 
>> excitations. You might need to set ALPHA in &MIXING quite small (< 0.1). 
>>
>> For a well separated S1 state, I'd hope it might work quite OK.
>>
>> Matt
>>
>> On Tuesday, September 13, 2016 at 9:59:13 PM UTC+1, Jen E. wrote:
>>>
>>> Hi All, 
>>>
>>> I am a new user of CP2k and I am trying to use the MOM method to 
>>> converge the S1 excited state of a molecule. I have a few related 
>>> questions.. 1.) I was wondering if there is a specific protocol for running 
>>> a MOM calculation  and 2.) Has there been any resolution in getting the MOM 
>>> method to work with a geometry optimization? 
>>>
>>> Thanks very much! 
>>>
>>> Jen 
>>>
>>> On Tuesday, April 19, 2016 at 4:58:16 AM UTC-4, Matt W wrote:
>>>>
>>>> HI Tassem,
>>>>
>>>> it should work, in principle. One think that could cause trouble is if 
>>>> the step size is large during the geometry optimization so the orbitals 
>>>> change too much.
>>>>
>>>> If you send me some output offline, we can have a look.
>>>>
>>>> Matt
>>>>
>>>> On Monday, April 18, 2016 at 4:10:30 PM UTC+1, Al-Moatasem El-Sayed 
>>>> wrote:
>>>>>
>>>>> Hi all,
>>>>>
>>>>> I am trying to run a geometry optimization while using the maximum 
>>>>> overlap method to occupy an excited state. When I try to do this, the first 
>>>>> SCF cycle successfully occupies the higher state I would like to be 
>>>>> occupied. I also manage to occupy the state I would like if I just run a 
>>>>> single point calculation. However, when the geometry optimization begins, 
>>>>> the electron moves back into the lowest unoccupied orbital. So I was 
>>>>> wondering if it is actually possible to use the maximum overlap method with 
>>>>> a geometry optimization.
>>>>>
>>>>> Thanks,
>>>>> Tassem
>>>>>
>>>>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20160916/e779a9ab/attachment.htm>