[gmx-developers] different nb parameters for inter and intramolecular interactions

[Lubos Vrbka]

> In Gromacs there are user tables, which are described in the manual.
> Normally you would use the two LJ user  tables for dispersion and repulsion
> respectively. But you can (mis)use them by filling each of them with a
> complete interaction function, for instance a Buckingham potential.
> You than still have the C6 and C12 parmeters to make linear combinations
> of these tables.
> For only O-H you could use the first table and set C6 for the 
> intermolecular
> interactions. And use pair interactions for the intramolecular interactions
> where you can use the second table and only set C12.
i went through the manual and i think i got the idea. there is, however, 
one problem left. how do i distinguish bonded and not-bonded pairs in 
the trajectory? recall, that there are O-H intramolecular and O-H 
intermolecular interactions and both O,H have the same atom types in 
both cases... if i am not wrong, gromacs doesn't allow "user" as a valid 
functional form for bonded terms...

this problem could be solved by assigning each water different atom 
types (e.g., OW_1, HW1_1, HW2_1; OW_2, HW1_2, HW2_2; ...) and than 
creating all possible combinations of interaction terms... the question 
is, 1) whether gromacs will be able to handle so many combinations and 
2) whether there will be some performance problem with so many different 
atomtype pairs...

thanks for help. best regards,

-- 
Lubos
_ at _"