use the empirical correction term of DFT-D3 in the classical MD

Marco synerg... at gmail.com
Mon May 26 03:40:02 UTC 2014


Dear JianBiao,

Sorry I could not be of any help. The DFT-D3 method is implemented in CP2K 
(link below):

http://manual.cp2k.org/trunk/CP2K_INPUT/FORCE_EVAL/DFT/XC/VDW_POTENTIAL/PAIR_POTENTIAL.html#list_TYPE

To perform the type of calculations that you are pursuing, maybe the 
following mixed-Hamiltonian functionality could serve your purpose:

CP2K_INPUT / MULTIPLE_FORCE_EVALS
http://manual.cp2k.org/trunk/CP2K_INPUT/MULTIPLE_FORCE_EVALS.html

CP2K_INPUT / FORCE_EVAL / MIXED /
http://manual.cp2k.org/trunk/CP2K_INPUT/FORCE_EVAL/MIXED.html

The following regtest files have examples of mixed Hamiltonian calculations 
employing DFT and FIST.
/cp2k/tests/QS/regtest-dft-vdw-corr/ch4-ch4-gpw-vdw.inp
/cp2k/tests/QS/regtest-ot-1/dip-mixed.inp

Best regards,
Marco

On Saturday, May 24, 2014 3:27:12 AM UTC-4, JianBiao wrote:
>
> Dear Marco,
>     I mean at each MD step, one can use the DFT-D3 code to calculate the 
> gradient and then add them to the values calculated by CP2K. 
>
> p.s. I am sorry I replied to you by email by mistake.
>
> Best regards
> JianBiao
>
>
>
>
>
> 在 2014年5月24日星期六UTC+8下午12时13分15秒,Marco写道:
>>
>> Hello,
>>
>> In addition to my previous post. Are you trying to add a functional term 
>> to the MM force field potential which is similar to the DFT-D3 expression? 
>> If so, would the GENPOT (
>> http://manual.cp2k.org/trunk/CP2K_INPUT/FORCE_EVAL/MM/FORCEFIELD/NONBONDED/GENPOT.html) 
>> functionality serve your purpose. With GENPOT you can specify a generic 
>> potential type. I have not used GENPOT myself yet but I would be interested 
>> in any results obtained with GENPOT.
>>
>> Regards,
>> Marco
>>
>> On Friday, May 23, 2014 11:37:51 PM UTC-4, Marco wrote:
>>>
>>> Dear JianBiao,
>>>
>>> Yeah that sounds like a reasonable approach. You are using QM data 
>>> derived from DFT-D3 calculations to parameterize a force field of a certain 
>>> functional form (presumably employing the Tang-Toennies damping function as 
>>> the non-bonded component in addition to bond/angle/dihedral etc. terms). 
>>> Your QM data from the DFT-D3 calculations account for dispersion effects 
>>> and this should transfer over into your force field parameterization in the 
>>> case of a good fit. As you know since you work with parameterization, one 
>>> measure of a good fit is whether your parameterized force field 
>>> calculations reproduce your QM reference data set and certain properties of 
>>> your system. Anthony J. Stone has published a lot of good work in this 
>>> field. 
>>> (http://www-stone.ch.cam.ac.uk/personal/bibliography.html)
>>>
>>> What do you mean by add the DFT-D3 to the classical MD parts?
>>>
>>> Best regards,
>>> Marco
>>>
>>> On Friday, May 23, 2014 9:38:32 PM UTC-4, JianBiao wrote:
>>>>
>>>> Dear Marco,
>>>>     Thank you for your reply. I am sorry I didn't say my question 
>>>> clearly. During the force-matching process, one can use the forces 
>>>> calculated from the PBE-AIMD as the references, then one can parameterize 
>>>> some parameters in the force field such as V(A, B, .. C6=0, C8=0). Because 
>>>> PBE fails to describe the dispersion effects, one cannot get the C6 and C8 
>>>> from the force-matching. Then during the MD simulation, one can use the 
>>>> Grimme's damped empirical correction terms to calculate the dispersion 
>>>> parts. So I think this method is reasonable. Could you please give some 
>>>> comments? Could someone please help to add the DFT-D3 to the classical MD 
>>>> parts?
>>>>
>>>> Best, JianBiao
>>>>    
>>>>
>>>>
>>>>
>>>>
>>>> 在 2014年5月24日星期六UTC+8上午7时45分00秒,Marco写道:
>>>>>
>>>>> Hello,
>>>>>
>>>>> DFT and classical MD are two very different model chemistry's. The 
>>>>> parameterizations for BMHFTD (if and when available) and DFT-D3 are 
>>>>> completely different. It would be unsound to transfer Grimme's damped 
>>>>> empirical correction terms into a classical MD force field in such a 
>>>>> straightforward manner. At least I have not seen it done in the literature. 
>>>>>
>>>>> Regards,
>>>>> Marco
>>>>>
>>>>> On Friday, May 23, 2014 8:23:10 AM UTC-4, JianBiao wrote:
>>>>>>
>>>>>> Dear CP2K developers,
>>>>>>     Is it possible to use DFT-D3 to calculate the dispersion energies 
>>>>>> during the classical MD simulations? Because the dispersion term in  
>>>>>> CP2K_INPUT <http://manual.cp2k.org/trunk/CP2K_INPUT.html> / 
>>>>>> FORCE_EVAL <http://manual.cp2k.org/trunk/CP2K_INPUT/FORCE_EVAL.html>
>>>>>>  / MM <http://manual.cp2k.org/trunk/CP2K_INPUT/FORCE_EVAL/MM.html> / 
>>>>>> FORCEFIELD<http://manual.cp2k.org/trunk/CP2K_INPUT/FORCE_EVAL/MM/FORCEFIELD.html>
>>>>>>  / NONBONDED<http://manual.cp2k.org/trunk/CP2K_INPUT/FORCE_EVAL/MM/FORCEFIELD/NONBONDED.html>
>>>>>> /
>>>>>> BMHFTD<http://manual.cp2k.org/trunk/index.html#CP2K_INPUT/FORCE_EVAL/MM/FORCEFIELD/NONBONDED/BMHFTD.html>
>>>>>>  is similar to Grimme's damped empirical correction, one can use the 
>>>>>> dft-d3 to calculate the dispersion energy and corresponding forces. Could 
>>>>>> you please give some suggestions? 
>>>>>>
>>>>>> Best wishes
>>>>>> JianBiao
>>>>>>
>>>>>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20140525/7fc0e051/attachment.htm>


More information about the CP2K-user mailing list