The energy of the whole system in a QMMM simulation

Jadzia c.gor... at gmail.com
Wed Sep 13 03:21:16 UTC 2017


​​Glad I could help. By the way, I'm a mathematician specialized in 
computational science as well... :-)

I don't think one can say that QM/MM is generally superior over MM models. 
It will depend on many factors whether QM/MM can be considered as superior, 
e.g. on the molecular system or the properties one is interested in. There 
are MM force fields specialized for certain systems which can outperform 
QM/MM simulations for certain properties. So the question is what 
properties are you interested in? If you generally want to show that QM/MM 
can be superior for certain properties, then I one idea would be to read 
articles about QM/MM methods since they sometimes show that the QM/MM 
method is better for certain properties than MM. The reference value for 
comparison is ideally an experimental value of high quality, but sometimes 
also a pure QM computation of high quality. 





On Monday, September 11, 2017 at 1:28:09 PM UTC+2, jts2... at gmail.com wrote:
>
> Thanks for your patience , Jadzia !
>
> Your answer have solved a problem for me. My major is computational 
> mathematics. 
>
> If we want to prove the superiority of QMMM model over MM model in this 
> simulation, what is the sensible quantity ? The oscillation of the energy ? 
> Or the dos of the system ? What quantity should taken to compare ?
>
>
>
>
>
>
>
> 在 2017年9月11日星期一 UTC+8下午2:53:39,Jadzia写道:
>>
>> The reference point of the absolute potential energy can usually be 
>> chosen arbitrary, and the absolute potential energy itself is therefore 
>> usually meaningless. See for instance:
>>
>>    - 
>>    https://www.quora.com/Why-do-we-need-a-reference-point-to-determine-potential-energy
>>    - http://hyperphysics.phy-astr.gsu.edu/hbase/pegrav.html
>>    - https://en.wikipedia.org/wiki/Potential_energy#Reference_level
>>    
>> What is meaningful and used in most calculations are energy differences. 
>> And here the choice of the reference point does not matter because it 
>> cancels out in the substraction.
>>
>> If you have the same system, and use at at first one layer of QM atoms 
>> and then two layers of QM atoms, then you are using two different molecular 
>> models for your system, even if the MM and the QM models are the same. The 
>> potential energy reference point will change, since some atoms are treated 
>> with another method which has a difference reference point. 
>>
>> Best wishes,
>> Jadzia
>>
>>
>>
>> On Monday, September 11, 2017 at 7:59:06 AM UTC+2, jts2... at gmail.com 
>> wrote:
>>>
>>> Hi Jadzia,
>>>
>>> I don't major in physical or chemistry, so what you have suggested 
>>> confused me for a long time. My example comes from the following page: 
>>> https://www.cp2k.org/exercises:2017_ethz_mmm:qmmm 
>>> <https://www.google.com/url?q=https%3A%2F%2Fwww.cp2k.org%2Fexercises%3A2017_ethz_mmm%3Aqmmm&sa=D&sntz=1&usg=AFQjCNGiF0gJ6mRlaGSe7--tVuWgLftgjA>
>>> . 
>>>
>>> The simulation uses the same molecular model (as far as I know, it's 
>>> true) for the MM region (I use the same potential function through 1-layer 
>>> simulation and 2-layer simulation and more). Does it mean the model can 
>>> compare now ? 
>>>
>>> Besides, the energy of the 1-layer and 2-layer and even more-layer 
>>> simulation is nearly geometric series. If energy of 1-layer is 1, then 
>>> the energy of 2-layer is almost 2, and 3 of 3-layer. It makes me consider 
>>> the energy belongs to the QM region. Hence, I want to know how to calculate 
>>> the whole energy of the system.
>>>
>>> Or the geometric series is just a coincidence ? There is some mistake in 
>>> my understanding ?
>>>
>>>
>>>
>>>
>>>
>>>
>>>
>>>
>>>
>>>
>>>
>>> 在 2017年9月11日星期一 UTC+8上午12:19:07,Jadzia写道:
>>>>
>>>> Hello again, 
>>>>
>>>> in the case that you mean that you are using a different number of QM 
>>>> atoms, a significant energy difference can be expected. The reason is that 
>>>> different molecular models assign different energies to the same 
>>>> atoms/molecules. In particular QM atoms/molecules often have a much lower 
>>>> energy than MM atoms. QM models for instance usually include the 
>>>> electron-nuclei interactions explicitly while MM models do not, therefore 
>>>> you usually cannot directly compare the total energy of different molecular 
>>>> models even if the system is the same. If you shift the scale properly 
>>>> between the different models, then a comparison should be possible in 
>>>> theory, but I have not experience with that. 
>>>>
>>>> Best wishes,
>>>> Jadzia
>>>>
>>>>
>>>>
>>>> On Sunday, September 10, 2017 at 5:24:40 PM UTC+2, jts2... at gmail.com 
>>>> wrote:
>>>>>
>>>>> Hi, Matt, bother you again. 
>>>>>
>>>>> The real problem I want to figure out is, when I set 1-layer QM atoms, 
>>>>> the energy is
>>>>> Total FORCE_EVAL ( QMMM ) energy (a.u.):           -351.508815087281562
>>>>>
>>>>> While I set 2-layer QM-atoms, the energy is 
>>>>> Total FORCE_EVAL ( QMMM ) energy (a.u.):           -696.366434340290539
>>>>>
>>>>> What bother me is that the energy should be almost the same, but the 
>>>>> fact is that they differ nearly twice. 
>>>>> So I want to know if there is some mistake in my understanding, or is 
>>>>> there some mistake in my setting ?
>>>>>
>>>>> 在 2017年9月8日星期五 UTC+8下午5:14:30,Matt W写道:
>>>>>>
>>>>>> The line I suggested _is_ the total energy QM + MM + QM-MM
>>>>>>
>>>>>> If you set print level medium you should see something like
>>>>>>
>>>>>>  QMMM| Evaluating forces on MM atoms due to the:
>>>>>>  QMMM| - QM/MM Coupling computed collocating the Gaussian Potential 
>>>>>> Functions.
>>>>>>  QMMM| QM/MM Nuclear Electrostatic Potential :                     
>>>>>>  -0.210189515
>>>>>>  QMMM| QMMM Total Energy (QM + QMMM electronic + QMMM nuclear):   
>>>>>> -694.056114328
>>>>>>  QMMM| MM energy NOT included in the above term! Check for: 
>>>>>>  FORCE_EVAL ( QMMM )
>>>>>>  QMMM| that includes both QM, QMMM and MM energy terms!
>>>>>>
>>>>>>  ENERGY| Total FORCE_EVAL ( QMMM ) energy (a.u.):           
>>>>>> -696.223035107415399
>>>>>>
>>>>>> Which gives the energy with and without the MM energy - so you can 
>>>>>> figure out the MM energy.
>>>>>>
>>>>>>
>>>>>> On Friday, September 8, 2017 at 6:45:52 AM UTC+1, jts2... at gmail.com 
>>>>>> wrote:
>>>>>>>
>>>>>>> Hi, Matt,
>>>>>>> Thanks for your advice! 
>>>>>>>
>>>>>>> I did as you suggest, then I typed in the command 'grep energy 
>>>>>>> *.out'. But the results said 
>>>>>>>
>>>>>>> QMMM | MM energy NOT included in the above term! Check for: 
>>>>>>> FORCE_EVAL(QMMM) that includes both QM, QMMM and MM energy terms !
>>>>>>>
>>>>>>> So I still didn't get the MM energy message.
>>>>>>>
>>>>>>> As far as my understanding, I should add something else in the &MM 
>>>>>>> section, maybe an additional &FORCE_EVAL to output the information of MM 
>>>>>>> region ?
>>>>>>> Or there is something important ignored ?
>>>>>>>
>>>>>>> 在 2017年9月8日星期五 UTC+8上午4:16:40,Matt W写道:
>>>>>>>>
>>>>>>>> At each step of a geo_opt run you should see something like
>>>>>>>>
>>>>>>>> ENERGY| Total FORCE_EVAL ( QMMM ) energy (a.u.):           
>>>>>>>> -696.237954025524573
>>>>>>>>
>>>>>>>> after all the QS output, that gives you the total energy of the 
>>>>>>>> system (at least if print_level is medium)
>>>>>>>>
>>>>>>>> On Thursday, September 7, 2017 at 6:33:03 PM UTC+1, Jadzia wrote:
>>>>>>>>>
>>>>>>>>> I am not sure if the question is really "where the file is" since 
>>>>>>>>> it will most likely be nowhere. And to be honest I don't know how to write 
>>>>>>>>> a similar file for the energies for geo_opt as for MD simulations since 
>>>>>>>>> only in the MD section there seems to be this print option, which might not 
>>>>>>>>> be used during geo_opt (not sure about it). 
>>>>>>>>>
>>>>>>>>> But a workaround might be as follows: One can run cp2k with the -o 
>>>>>>>>> option to store the output of the geometry optimization in a file, and then 
>>>>>>>>> one could grep the energy out of that file. 
>>>>>>>>>
>>>>>>>>>
>>>>>>>>>
>>>>>>>>> On Thursday, September 7, 2017 at 5:05:04 PM UTC+2, 
>>>>>>>>> jts2... at gmail.com wrote:
>>>>>>>>>>
>>>>>>>>>> Thanks for your suggestion, which works well in a example in 
>>>>>>>>>> official website, but fails in another case.
>>>>>>>>>>
>>>>>>>>>> I run a GEO_OPT simulation using QMMM method,  the original 
>>>>>>>>>> motion.inc is as follows
>>>>>>>>>> &MOTION
>>>>>>>>>>   &GEO_OPT
>>>>>>>>>>      OPTIMIZER LBFGS
>>>>>>>>>>   &END
>>>>>>>>>>   &CONSTRAINT
>>>>>>>>>>      &FIXED_ATOMS
>>>>>>>>>>      LIST 1..16
>>>>>>>>>>      EXCLUDE_MM .TRUE.
>>>>>>>>>>      EXCLUDE_QM .FALSE.
>>>>>>>>>>      &END FIXED_ATOMS
>>>>>>>>>>   &END CONSTRAINT
>>>>>>>>>> &END MOTION
>>>>>>>>>>
>>>>>>>>>> Then I added the MD section, they became
>>>>>>>>>> &MOTION
>>>>>>>>>>   &GEO_OPT
>>>>>>>>>>      OPTIMIZER LBFGS
>>>>>>>>>>   &END
>>>>>>>>>>   &CONSTRAINT
>>>>>>>>>>      &FIXED_ATOMS
>>>>>>>>>>      LIST 1..16
>>>>>>>>>>      EXCLUDE_MM .TRUE.
>>>>>>>>>>      EXCLUDE_QM .FALSE.
>>>>>>>>>>      &END FIXED_ATOMS
>>>>>>>>>>   &END CONSTRAINT
>>>>>>>>>>   &MD
>>>>>>>>>> &PRINT
>>>>>>>>>> &ENERGY
>>>>>>>>>> FILENAME=total.ener
>>>>>>>>>>  &EACH
>>>>>>>>>>    GEO_OPT 1
>>>>>>>>>>  &END EACH
>>>>>>>>>> &END ENERGY
>>>>>>>>>> &END PRINT
>>>>>>>>>>   &END MD
>>>>>>>>>> &END MOTION
>>>>>>>>>>
>>>>>>>>>> The strange thing is that there wasn't any file named 
>>>>>>>>>> *total.ener* in the current directory. Where is the question ?
>>>>>>>>>> The &EACH section is not correct ?
>>>>>>>>>>
>>>>>>>>>> Expect for your kind reply !
>>>>>>>>>>
>>>>>>>>>>
>>>>>>>>>>
>>>>>>>>>>
>>>>>>>>>> 在 2017年9月7日星期四 UTC+8下午12:13:16,Jadzia写道:
>>>>>>>>>>>
>>>>>>>>>>> Hello,
>>>>>>>>>>>
>>>>>>>>>>> To my understanding if you want to write the total energy to a 
>>>>>>>>>>> file for instance for an MD simulation you need to add a section similar to:
>>>>>>>>>>>
>>>>>>>>>>>  &MOTION
>>>>>>>>>>>    &MD
>>>>>>>>>>>     &PRINT
>>>>>>>>>>>       &ENERGY
>>>>>>>>>>>         FILENAME =cp2k.out.energy
>>>>>>>>>>>           &EACH
>>>>>>>>>>>             MD 10
>>>>>>>>>>>           &END EACH
>>>>>>>>>>>       &END ENERGY
>>>>>>>>>>>     &END PRINT
>>>>>>>>>>>   &END MD
>>>>>>>>>>> &END MOTION
>>>>>>>>>>>
>>>>>>>>>>> In this example the energy of every tenth step is written to the 
>>>>>>>>>>> file, and is independent of the type of force evaluation (MM, QM, QM/MM, 
>>>>>>>>>>> ...). 
>>>>>>>>>>>
>>>>>>>>>>> Here is the corresponding section in the manual: 
>>>>>>>>>>> https://manual.cp2k.org/trunk/CP2K_INPUT/MOTION/MD/PRINT/ENERGY/EACH.html
>>>>>>>>>>>
>>>>>>>>>>> Hope this helps,
>>>>>>>>>>> Jadzia
>>>>>>>>>>>
>>>>>>>>>>>
>>>>>>>>>>>
>>>>>>>>>>>
>>>>>>>>>>>
>>>>>>>>>>> On Wednesday, September 6, 2017 at 8:31:10 AM UTC+2, 
>>>>>>>>>>> jts2... at gmail.com wrote:
>>>>>>>>>>>>
>>>>>>>>>>>> Dear researchers and developers,
>>>>>>>>>>>>
>>>>>>>>>>>> I am not so familiar with CP2K. After I run a QMMM simulation 
>>>>>>>>>>>> of KCl just as here: 
>>>>>>>>>>>> https://www.cp2k.org/exercises:2017_ethz_mmm:qmmm, I want to 
>>>>>>>>>>>> know the energy of the whole system not just the energy of the QM region, 
>>>>>>>>>>>> how to achieve it in the input setting?
>>>>>>>>>>>>
>>>>>>>>>>>> I'll appreciate for your apply !
>>>>>>>>>>>>
>>>>>>>>>>>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20170912/e8ea59bb/attachment.htm>


More information about the CP2K-user mailing list