[gmx-users] Disable non-bonded interactions between molecules

[Mark Abraham]

On 08/03/11, ruehle at mpip-mainz.mpg.de wrote:
> Dear all,
> 
> I need to perform a simulation in which interactions between molecules (polymers) are calculated only via virtual sites. Within a molecule, non-bonded interactions between atoms should be present. Further, it would be best to be able to select which parts of the molecule explicitly interact (only close units along the chain).
> 
> What is the most efficient way to achieve this? Energy group exclusions are not possible since they would also disable intramolecular interactions.
> 

Not true. Such exclusions act between named *pairs* of groups. See manual 7.3. If there is no named "self-self" pair, then intra-molecular interactions are on (by default).

The real issue is that you are limited to 256 energy groups, i.e. 256 polymer molecules. Also, the work to create the .mdp file grows with the square of the number of polymer molecules.


>  One possibility might be to set all non-bonded parameters (charge + lj) to zero and then explicitly specify the intramolecular ones as pairs:
> 
> I need coulomb (cut-off) as well as Lennard-Jones interactions
> -> use pairs function type 2, just copy the charges + lennard jones parameters
> 

Seems ideal, except that the work creating the topology for an n-mer grows as about n^2. It will compute efficiently.


> What about 1-4 scaling? Can i just scale down charges + lennard jones parameters for those pairs or do I have to use fudge QQ in pairs?
> 

Not sure - might depend what kind of force field you're basing from. Easy to test via mdrun -rerun on a single frame of an isolated small polymer, perhaps with parameters for irrelevant atoms zeroed out. Using -rerun keeps the coordinates constant.


> Would this work or is there a better solution (maybe also with minor code changes)?
> 

In theory, you could hack the neighbour searching code at the point it implements energy group exclusions, and maybe exclude all appropriate intra-molecular interactions at that point. I do not recommend exploring this.

Mark
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://maillist.sys.kth.se/pipermail/gromacs.org_gmx-users/attachments/20110308/aa888dcc/attachment.html>