How to estimate QMMM box dimensions

[Matt W]

Probably it has just been done by checking that the results don't change 
'too much' in many cases.

You could run an isolated molecule/cluster and use the MULTIPOLE 
<https://manual.cp2k.org/trunk/CP2K_INPUT/FORCE_EVAL/DFT/POISSON/MULTIPOLE.html> 
poisson solver (which is what you get when you use QMMM) and see how that 
compares to say the WAVELET 
<https://manual.cp2k.org/trunk/CP2K_INPUT/FORCE_EVAL/DFT/POISSON/WAVELET.html> 
one, and what cell dimensions are needed.

Happy new year, too.

Matt

On Thursday, January 4, 2018 at 6:50:20 AM UTC, Rahul Hardikar wrote:
>
> Thanks for the reply Jiang,
>
> Maybe I should rephrase my question. I understand that GEEP algorithm 
> implementation takes care of the QMMM electrostatic coupling, however that 
> does not give an
> idea of the box size for QM/MM region. What I am specifically referring to 
> is the following section in QMMM:
>
> CP2K_INPUT <https://manual.cp2k.org/cp2k-2_6-branch/CP2K_INPUT.html> / 
> FORCE_EVAL 
> <https://manual.cp2k.org/cp2k-2_6-branch/CP2K_INPUT/FORCE_EVAL.html> / 
> QMMM 
> <https://manual.cp2k.org/cp2k-2_6-branch/CP2K_INPUT/FORCE_EVAL/QMMM.html> 
> / CELL 
> <https://manual.cp2k.org/cp2k-2_6-branch/CP2K_INPUT/FORCE_EVAL/QMMM/CELL.html>
>
> where the box size of the QM/MM region can be specified with possibly 
> smaller dimensions than the entire cell. So do you or other users know a 
> method to estimate the minimum 
> dimensions of the QM/MM cell by looking at convergence with respect to 
> some observable quantity in QMMM. Any help in this direction will be 
> helpful!
>
> Best Regrads and belated New Year wishes to cp2k users!
>
> Rahul
>
>  
> On Thursday, December 21, 2017 at 4:43:20 PM UTC+5:30, Tianshu Jiang in 
> Beijing wrote:
>>
>> hello Hardikar,
>>
>> The dimension of QM box depends on your system, or the quantity where you 
>> are interested. 
>> You can refer the paper "An Efficient Real Space Multigrid QM/MM 
>> Electrostatic Coupling" written by Teodoro Laino for details.
>> The realization of QMMM in CP2K is based on this paper.
>>
>> Best wishes !
>>
>> 在 2017年12月20日星期三 UTC+8下午6:06:37，Rahul Hardikar写道：
>>>
>>> Dear CP2K Users/Experts,
>>>
>>> Currently, in my QMMM setup I am maintaining box lengths same as the 
>>> overall systems' box lengths. I am aware that one can reduce the box size 
>>> for the QMMM
>>> box thereby reducing the cost of QM calculations. However what is not 
>>> clear to me is the role of electrostatics when QMMM box lengths are less 
>>> the the overall lengths. So my question is (a) how one can estimate the 
>>> QMMM box lengths and maintain the electrostatic interaction between QM and 
>>> MM part consistent? and (b) how can I estimate the amount of vacuum to use 
>>> to avoid any interaction between periodic images? I am using cubecruncher 
>>> to crunch out plane averaged electrostatic potential and hartree energy 
>>> from the corresponding cube files.  
>>>
>>> Any suggestions and/or comments will be appreciated.
>>>
>>> Best Regards,
>>> Rahul Hardikar
>>>
>>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20180104/4c3aafae/attachment.htm>