combine ab initio with implicit solvent

[polla... at gmail.com]

Each iteration with SCCS active has additional overhead compared to the 
standard SCF iteration. Your explicit solvent calculation just happens to 
cost about the same as the smaller system with SCCS (by the end of your log 
files, both calculations are taking about 20 seconds per MD step). I don't 
see anything wrong with your calculation otherwise. You may see improved 
performance by using the MOLOPT-SR basis sets which are better suited to 
condensed phase simulations, but it's not necessary.

-Travis



On Wednesday, April 25, 2018 at 2:49:34 PM UTC-4, Xiaoliu wrote:

> Dear Vladimir,
>
> I tested SCCS with a smaller box, and it still ran slower than full HF. 
> Within 70 minutes, the *ab initio* calculation with 1 LiPF6 and  23 DMC 
> went through 172 steps, while the one with 1 LiPF6, 4 DMC, and implicit 
> solvent went through only 96 steps. 
>
> Also, in &SCCS subsection, I set:
>
>    EPS_SCCS 5.0E-7
>    EPS_SCF 2.5E-6,
>
> based on the parameter in SCF subsection:     
>   
> EPS_SCF     4.9999999999999998E-07.
>
> Is there something wrong in those SCCS parameter settings?
>
> Really appreciated!
>
> Regards,
> Xiaoliu
>
>
> On Wednesday, April 25, 2018 at 5:27:12 AM UTC-5, Vladimir Rybkin wrote:
>>
>> Dear Xiaolui,
>>
>> one obvious thing is that in your SCCS calculation you use a larger unit 
>> cell than in you explicit calculation.
>>
>> Yours,
>>
>> Vladimir
>>
>> среда, 25 апреля 2018 г., 1:13:22 UTC+2 пользователь Xiaoliu написал:
>>>
>>> Dear Vladimir,
>>>
>>> Thank you for your advice. I applied SCCS model on ab initio 
>>> calculation, and found out the computational efficiency of the SCCS is 
>>> about the same as treating the whole solution with ab initio. Is it what 
>>> it's supposed to be? I think that the semi-empirical model should run much 
>>> faster compared with full HF method.
>>>
>>> Here're my input files for the implicit solvent and the explicit one.
>>>
>>> Thanks,
>>> Xiaoliu
>>>
>>>
>>>
>>> On Thursday, April 19, 2018 at 3:32:07 AM UTC-5, Vladimir Rybkin wrote:
>>>>
>>>> Dear Xiaoliu,
>>>>
>>>> you need to give explicit coordinates of you LiPF6 with your DMC and 
>>>> use 
>>>> SCCS:
>>>> https://manual.cp2k.org/trunk/CP2K_INPUT/FORCE_EVAL/DFT/SCCS.html
>>>>
>>>> or SCRF:
>>>> https://manual.cp2k.org/trunk/CP2K_INPUT/FORCE_EVAL/DFT/SCRF.html
>>>>
>>>> Those are implicit solvent models in CP2K.
>>>>
>>>> Yours,
>>>>
>>>> Vladimir
>>>>
>>>> четверг, 19 апреля 2018 г., 1:03:26 UTC+2 пользователь Xiaoliu написал:
>>>>>
>>>>> Dear Cp2K users,
>>>>>
>>>>> I'm working with a system of LiPF6 dissolved in DMC. I was wondering 
>>>>> if there is a way to apply ab initio on the Li ion and the DMC molecules in 
>>>>> the first solvation shell, while treat the rest of the solvent DMC with 
>>>>> polarizable continuum model (PCM). Is there any subsection in cp2k 
>>>>> complying this?
>>>>>
>>>>> Thanks in advance.
>>>>>
>>>>> Xiaoliu
>>>>>
>>>>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20180425/bf19417d/attachment.htm>