As part of my project I am working with hybrid simulations, combining atomistic and CG representations. In my particular case, I work with a system using CHARMM36 and MARTINI. Searching in the literature, finding the optimal coupling scheme between the AA and the CG regions is still an on going area or research. Some authors uses tabulated potentials, other use the standard combination rules, together with virtual sites to bridge the AA/CG interactions.
So my question is, when to use tabulated potentials and when combination rules are sufficient, to account for the non-bonded interactions?
So my question is, when to use tabulated potentials and when combination rules are sufficient, to account for the non-bonded interactions?
That depends
on the functional form that you use to describe your non-bonded interactions. If the functional form is implemented in GROMACS you do not need to generate tabulated potentials, if the functional form is not implemented you need to use tabulated potential
on the force field philosophy behind the generation of non-bonded parameters. Are combination rules used to generate parameters between different particle types? Then you can use them, in place to generate all the parameter combinations.
That is a general consideration for any type of non-bonded interaction between particles.
I am working on the diffusion of Methanol in MFI zeolite. The interaction of Zeolite-Zeolite is defined by Buckingham Potential and Methanol-Methanol and Methanol-Zeolite is defined by Lennard-Jones. How can I Use these potentials at the same time? How can I describe them with functionality? As I define Buckingham potential in nonbonded.itp file with the functionality of 2 the GMX says you defined Lennard Jones functionality in forcefield.itp while using Buckingham Potential.
i) you cannot mix Lennard-Jones and Buckingham, you have to pick one of the two
ii) since at least GROMACS 2020 the Buckingham potential is not supported anymore (it is not stated in the manual)
One way to go is to fit the potential terms of your Buckingham in terms of a LJ potential and write a “pure” LJ potential. The quality of this approximation largely relies on the quality of the fits, I would say, I have no idea if this is a good approach. The other is to use tabulated potentials. Also this functionality is deprecated, and you will have to rely on older GROMACS versions (2019 being the newest still supporting the tabulated potential). Take a look at this.