[CP2K-user] [CP2K:20530] creating a mixed van der Waals damping scheme

[Andrey Poletayev]

Poking around the documentation, it seems I should be able to repeat the 
PAIR_POTENTIAL section twice, one for the DFTD3 zero damping, and the 
second for DFTD3(BJ), each with SCALING of 0.5, to average the two 
dispersions, each with its appropriate parameters set by D3_SCALING and 
D3BJ_SCALING respectively.

Does this seem reasonable / is SCALING just a linear strength coefficient? 
The default of 0 on SCALING is a bit confusing because I would think the 
default SCALING would be 1. 

Andrey

On Friday, August 2, 2024 at 10:52:24 AM UTC+1 Andrey Poletayev wrote:

> Just to clarify the jargon in my orginal post: the chemRxiv argues that 
> water ML potential trained with the revPBE0 functional and van der Waals 
> damping set to average between zero and Becke-Johnson is the most accurate. 
>
> Thank you for the answer, it sounds like I could attempt to halve the 
> relevant coefficients first. 
>
> Andrey
>
> On Friday, August 2, 2024 at 10:41:08 AM UTC+1 Jürg Hutter wrote:
>
>> Hi 
>>
>> I don't know what "50% of Becke-Johnson damping with revPBE0+D3" really 
>> means, but you 
>> can provide the input parameters to the D3(BJ) method with 
>>
>> FORCE_EVAL / DFT / XC / VDW_POTENTIAL / PAIR_POTENTIAL 
>>
>> D3BJ_SCALING (s6,a1,s8,a2) 
>>
>> regards 
>> JH 
>>
>> ________________________________________ 
>> From: cp... at googlegroups.com <cp... at googlegroups.com> on behalf of 
>> Andrey Poletayev <andrey.p... at gmail.com> 
>> Sent: Friday, August 2, 2024 11:23 AM 
>> To: cp2k 
>> Subject: [CP2K:20526] creating a mixed van der Waals damping scheme 
>>
>> Hello! 
>> This may be a dumb question coming from a beginner user. I am looking to 
>> use CP2K for training ML potentials for water. I found this recent work on 
>> chemRxiv combing through functionals (although they used VASP) : 
>> https://chemrxiv.org/engage/chemrxiv/article-details/66616c4e21291e5d1d301165 
>>
>> They argue that the best behavior of simulated water is achieved with 
>> about 50% of Becke-Johnson damping with revPBE0+D3. There are two options - 
>> either compute every frame twice (zero damping and B-J) and average, or 
>> make up a halfway damping. Is it possible / straightforward to create such 
>> a halfway damping? 
>>
>> Thank you, 
>> -Andrey 
>>
>> -- 
>> You received this message because you are subscribed to the Google Groups 
>> "cp2k" group. 
>> To unsubscribe from this group and stop receiving emails from it, send an 
>> email to cp2k+uns... at googlegroups.com<mailto:cp2k+uns... at googlegroups.com>. 
>>
>> To view this discussion on the web visit 
>> https://groups.google.com/d/msgid/cp2k/0fdd3357-0619-4e25-98fc-5a4db44ea837n%40googlegroups.com
>> <
>> https://groups.google.com/d/msgid/cp2k/0fdd3357-0619-4e25-98fc-5a4db44ea837n%40googlegroups.com?utm_medium=email&utm_source=footer>. 
>>
>>
>

-- 
You received this message because you are subscribed to the Google Groups "cp2k" group.
To unsubscribe from this group and stop receiving emails from it, send an email to cp2k+unsubscribe at googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/cp2k/58153cbf-7349-4c9f-8eab-f1eda050340bn%40googlegroups.com.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20240802/e0a6cb9f/attachment.htm>