[gmx-users] About exclusion of non-bonded interaction for pairs of energy groups

[Mark Abraham]

Lee Soin wrote:
> Hello!
> Sorry to bother you again, but I have another question. I have written a
> topology file of two water molecules to test the pair interaction. Here is
> part of my topology file:
> 
> [ moleculetype ]
> SOL  2
> [ atoms ]
>   1  opls_116  1  SOL  OW  1  -0.82
>   2  opls_117  1  SOL  HW1  1  0.41
>   3  opls_117  1  SOL  HW2  1  0.41
>   4  opls_116  2  SOL  OW  2  -0.82
>   5  opls_117  2  SOL  HW1  2  0.41
>   6  opls_117  2  SOL  HW2  2  0.41
> [ settles ]
>   1  1  0.1  0.16330
>   4  1  0.1  0.16330
> [ exclusions ]
>   1  2  3
>   2  1  3
>   3  1  2
>   4  5  6
>   5  4  6
>   6  4  5
> [ exclusions ]
> 1 4
> [ pairs ]
>   1  4  1
> 
> I have also changed the fudgeLJ and fudgeQQ to 1.0.
> Originally, there is no pair interaction in this system. First I did a
> simulation when the last 4 lines were removed, that is, I treated all the
> interactions as non-bonded interactions. 

Since there is no bonded interaction between 1 and 4, your second [ 
exclusions ] directive will not have any effect. Read section 5.4. 
Exclusions are cancellations only of interactions between bonded atoms.

> Then I added the last 4 lines
> to specify the normal non-bonded interaction between atom 1 and 4 as pair
> interaction in order to see whether that will affect the result. It turns
> out that the results for the two cases are different. So did I miss
> something, or are pair interactions and non-bonded interactions treated
> differently by GROMACS?

mdrun is only reproducible if you tell it that you want it to be with 
"mdrun -reprod". So it is possible that your observation of different 
results is not (yet) meaningful.

However, I believe you've merely added a second identical non-bonded 
interaction between 1 and 4, so you will get different results regardless.

Mark