[gmx-users] partial exclusion of interactions within 1-4

[Alex]

It is my understanding that either [pairs] or [exclusions] affect LJ and 
electrostatics simultaneously, please correct me if I'm wrong. Maybe 
this just isn't meant to be, being a highly artificial setup 
(nearest-neighbor electrostatics on, LJ off)... The idea for this comes 
from authors who used LAMMPS, and that thing allows such tricks with ease.

The setup is almost guaranteed to work for me without including these 
interactions (just set nrexcl=3 and go home), but I am needlessly 
hesitating. In addition, I am absolutely against LAMMPS in this 
particular case -- my postdoc is using it heavily on the same machine 
with the same GPUs and it's massively slower (I estimate 5-7 times for 
identical system size) than Gromacs. I just can't have 20-30 ns as my 
simulated time.

Alex


On 5/22/2018 3:17 AM, Mark Abraham wrote:
> Hi,
>
> That sounds right, given the wording (which I just didn't remember right
> then). So now that we understand that nrexcl applies to all moleculetypes
> in the .top then we pick whichever of 1 or 3 is most convenient (probably
> 1) and add either manual [pairs] or manual [exclusions] to the appropriate
> moleculetypes until it works how we want. In particular, the manual [pairs]
> should be able to have LJ parameters specified that are zero when you want
> that. nrexcl and fudge values are just sugar for common cases of these
> patterns.
>
> Mark
>
> On Tue, May 22, 2018, 09:57 Alex <nedomacho at gmail.com> wrote:
>
>> I feel like I am being tested, but it says what I thought I knew:
>> "defines the name of your molecule in this *.top and nrexcl = 3 stands
>> for excluding non-bonded interactions between atoms that are no further
>> than 3 bonds away."
>>
>> This to me suggests that all coulombic+LJ interactions will be removed
>> for neighbors closer than third. By setting nrexcl = 1, one would get
>> nearest-neighbor coulombic back, along with LJ. However, I still want
>> the LJ off. Does this make sense?
>>
>> Alex
>>
>>
>> On 5/22/2018 1:48 AM, Mark Abraham wrote:
>>> Hi,
>>>
>>> What does the documentation of nrexcl in the reference manual say?
>>>
>>> Mark
>>>
>>> On Tue, May 22, 2018, 08:51 Alex <nedomacho at gmail.com> wrote:
>>>
>>>> Okay, let me try to rephrase the question, because it seems that I am
>>>> definitely unable to make it work and noone's replied so far.
>>>>
>>>> I have a "molecule" (a periodic boron nitride sheet) and a stand-alone
>>>> itp file for it -- everything works, i.e. it's a stable, well-behaved
>>>> structure the way I set it up.
>>>>
>>>> Within that molecule (and nowhere else) I need Coulombic interactions
>>>> between closest neighbors, while all other non-bonded interactions
>>>> should obey an effective nrexcl = 3. Is this something that can be done
>>>> in Gromacs?
>>>>
>>>> Thank you.
>>>>
>>>> Alex
>>>>
>>>>
>>>> On 5/16/2018 7:12 PM, Alex wrote:
>>>>> Hi all,
>>>>>
>>>>> I think I understand how it works, but to double-check... There's a
>>>>> paper describing a setup of intramolecular interactions for boron
>>>>> nitride (https://pubs.acs.org/doi/10.1021/acs.jpclett.7b03443), in
>>>>> which non-bonded interactions *within* 1-4 were set up in the
>>>>> following way: LJ is off, but all electrostatic interactions are on.
>>>>> This is to achieve a better representation of the phonon spectrum and
>>>>> in fact it appears physically correct.
>>>>>
>>>>> Does this mean setting nrexcl to 1 for the crystal topology and
>>>>> creating a table of nonbond_param between B and N with sigma and
>>>>> epsilon equal to 0?
>>>>>
>>>>> Thanks,
>>>>>
>>>>> Alex
>>>> --
>>>> Gromacs Users mailing list
>>>>
>>>> * Please search the archive at
>>>> http://www.gromacs.org/Support/Mailing_Lists/GMX-Users_List before
>>>> posting!
>>>>
>>>> * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
>>>>
>>>> * For (un)subscribe requests visit
>>>> https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-users or
>>>> send a mail to gmx-users-request at gromacs.org.
>>>>
>> --
>> Gromacs Users mailing list
>>
>> * Please search the archive at
>> http://www.gromacs.org/Support/Mailing_Lists/GMX-Users_List before
>> posting!
>>
>> * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
>>
>> * For (un)subscribe requests visit
>> https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-users or
>> send a mail to gmx-users-request at gromacs.org.
>>