[gmx-users] Coarse-graining and cut-offs

[Thomas Schlesier]

Dear all,
first of all, sorry to this rather conceptional question, which is not 
totally to GROMOACS related. But probably anyone of you can help.

In my simulations I use mesitylene as a solvent. In future i want to 
coarse-grain the full atomic mesitylene to an effective one-particle. 
For this i did a NPT-simulation to obtain the RDF (with regard to the 
COM of the molecules) from which i want to calculate an effective 
interaction potential by iterative boltzmann. The full-atomistic 
simulations (slovate + solvent) employed a cut-off of 1.4 nm.
At this distance (1.4 nm) the RDF shows it second minimum (RDF=0.97) and 
after 1.8 nm the RDF is around 1.0 +/- 0.15 (I got the RDF only up to a 
distance of 2.3 nm).
Now my question is how long should be the table for the effective 
potential (i.e. maximum distance for an interaction)? If i would use 1.4 
nm, i would have a jump in the interaction potential (which is normal 
due to the truncation of the cut-off). But for coulomb-interaction one 
would have PME or other stuff which would correct the artifects to some 
degree.
Or would it be better to use a long interaction table (interactions at 
longer distances) and truncate the table at a distance after 1.4 nm 
where the RDF is equal to 1?

Hope anybody can give some insight.

Greetings
Thomas