[gmx-developers] Question about MM point charges in QM/MM calculation

[Gerrit Groenhof]

Sorry one line should be changed still:
     }
     mm_j_particles[mm_nr].j = j;
     mm_j_particles[mm_nr].shift = is;
     mm_nr++;
   }




Gerrit Groenhof wrote:
> So instead of looking into the QMMMnblists, you could simply scan over 
> all particles. If (!bQMMM[i]), add it to the MM array. This way you 
> get the whole box, but there could be problems with particles being in 
> a different periodic image that cannot be not shifted back correctly 
> prior to the QM call.
>
> I am not sure about this, but you coudl try something like in the 
> update_QMMMrec routine(3.3.x only):
>
>
>
>  if(QMMMlist_sr.nri){
>      qm_i_particles[0].shift = XYZ2IS(0,0,0);
>      for(i=0;i<QMMMlist_sr.nri;i++){
>    qm_i_particles[i].j     = QMMMlist_sr.iinr[i];
>    if(i){
>      qm_i_particles[i].shift = pbc_dx_aiuc(&pbc,x[QMMMlist_sr.iinr[0]],
>                        x[QMMMlist_sr.iinr[i]],dx);
>
>    }
>      crd[0] = IS2X(QMMMlist_sr.shift[i]) + IS2X(qm_i_particles[i].shift);
>    crd[1] = IS2Y(QMMMlist_sr.shift[i]) + IS2Y(qm_i_particles[i].shift);
>    crd[2] = IS2Z(QMMMlist_sr.shift[i]) + IS2Z(qm_i_particles[i].shift);
>    is = XYZ2IS(crd[0],crd[1],crd[2]);
>   for(j=0;j<md->nr;j++){
>     if(!bQMMM[j]){
>      if(mm_nr >= mm_max){
>        mm_max += 1000;
>        srenew(mm_j_particles,mm_max);
>      }
>      mm_j_particles[mm_nr].j = QMMMlist_sr.jjnr[j];
>      mm_j_particles[mm_nr].shift = is;
>      mm_nr++;
>    }
> }
>
> Gerrit
>
> Pia Toelle wrote:
>> Thanks for the comment and your interest.
>>
>>
>> I still would like to get informations about the MM-atom list
>> (mm->xMM[i][..],mm->MMcharges[i]) and how I could change it, so that all
>> MM atoms are included not considering the cut-off, because we would like
>> to print out all (!) MM atoms (and their charges).
>> Please tell me if a change of the (t_MMrec)-variable mm could cause
>> problems for the MM-part of the routine or is  the (t_MMrec)-variable mm
>> only used in the QM part.
>> Or would it be possible to print out all MM atoms + charges at an other
>> point in the program? Where?
>>
>> Thanks in advance.
>>
>> Pia Toelle
>>
>> BCCMS
>> Uni Bremen
>> Am Fallturm 1
>> 28359 Bremen
>> Germany
>>
>>
>>
>> Gerrit Groenhof wrote:
>>  
>>> Hi,
>>>
>>> We did a gmx-dftb interface that used function calls from the linked
>>> dftb librarysome time ago. That was in 3.3.3, but I never found the 
>>> time
>>> to pt it in 4. ALso I remember some compilation issues with the dftb
>>> code on 64 bit machines. I can send you the code if you're interested.
>>>
>>> For QM/MM the MM atoms are indeed set with the cut-off options. 
>>> Although
>>> for dftb it is not very difficult, I have never found the time or need
>>> to complete the ewald implementation.  To avoid duplications, I suggest
>>> we finish the PME implementation for the DFTB code together. Contacting
>>> me off-list  is probably more convenient if you are interested.
>>>
>>> BEst,
>>>
>>> Gerrit
>>>
>>>
>>>
>>>
>>> On 30 Jan 2009, at 18:09, Pia Toelle wrote:
>>>
>>>    
>>>> Hello.
>>>>
>>>> I have got some questions concerning the MM-atom-list in a QM/MM
>>>> calculation.
>>>>
>>>> We do QM-MM calculations using the Gaussian interface (gromacs-3.3.3),
>>>> but modified the code to be able to use DFTB instead of Gaussian
>>>> (www.dftb.org).
>>>> Our (MM-)system is a SiO2-surface (with periodic boundary condition;
>>>> full).
>>>> We use PME and DO NOT (!) define proper charge groups for the surface
>>>> structure, every atom forms its own charge group. Number of MM atoms
>>>> ~1900.
>>>> (The proper definition of charge groups in our system is not 
>>>> obvious or
>>>> even impossible, as it is a solid and a periodic system)
>>>>
>>>> We noticed, that the MM system considered in the QM calculation 
>>>> does not
>>>> consist of the entire MM system. The point charges
>>>> (mm->xMM[i][..],mm->MMcharges[i]) that are written to input.com (by
>>>> qm_gaussian.c) are only those of the MM atoms in the vicinity of 
>>>> the QM
>>>> atoms. The list of MM atoms in the vicinity of the QM-zone is set (as
>>>> far as we can see) using the neighbor-search algorithm depending 
>>>> also on
>>>> the charge groups (in ns5_core.c). Please correct me if I am wrong at
>>>> some point.
>>>> Is it right, that the cut-off for the QM/MM - neighbor search is 
>>>> set by
>>>> 'rcoulomb' in the mdp-file? Or 'rlist'?
>>>>
>>>> Anyhow, we would like to print out all (!) MM atoms (and their 
>>>> charges).
>>>> How to solve this problem?
>>>> Is there a simple way to put all MM atoms in the mm list? And please
>>>> tell me if a change of the (t_MMrec)-variable mm could cause problems
>>>> for the MM-part of the routine or is  the (t_MMrec)-variable mm only
>>>> used in the QM part.
>>>> Or would it be possible to print out all MM atoms + charges at an 
>>>> other
>>>> point in the program? Where?
>>>> I am grateful for any helpful information or idea.
>>>>
>>>>
>>>> Thanks in advance.
>>>>
>>>> Pia Toelle
>>>>
>>>> BCCMS
>>>> Uni Bremen
>>>> Am Fallturm 1
>>>> 28359 Bremen
>>>> Germany
>>>>
>>>> _______________________________________________
>>>> gmx-developers mailing list
>>>> gmx-developers at gromacs.org
>>>> http://www.gromacs.org/mailman/listinfo/gmx-developers
>>>> Please don't post (un)subscribe requests to the list. Use the
>>>> www interface or send it to gmx-developers-request at gromacs.org.
>>>>       
>>> -- 
>>> Gerrit Groenhof
>>> MPI biophysical chemistry
>>> Goettingen
>>> Germany
>>> http://wwwuser.gwdg.de/~ggroenh/
>>>
>>> _______________________________________________
>>> gmx-developers mailing list
>>> gmx-developers at gromacs.org
>>> http://www.gromacs.org/mailman/listinfo/gmx-developers
>>> Please don't post (un)subscribe requests to the list. Use thewww
>>> interface or send it to gmx-developers-request at gromacs.org.
>>>     
>>
>> _______________________________________________
>> gmx-developers mailing list
>> gmx-developers at gromacs.org
>> http://www.gromacs.org/mailman/listinfo/gmx-developers
>> Please don't post (un)subscribe requests to the list. Use the www 
>> interface or send it to gmx-developers-request at gromacs.org.
>>   
>
> _______________________________________________
> gmx-developers mailing list
> gmx-developers at gromacs.org
> http://www.gromacs.org/mailman/listinfo/gmx-developers
> Please don't post (un)subscribe requests to the list. Use the www 
> interface or send it to gmx-developers-request at gromacs.org.