[gmx-users] combination rules -- the part aboutthecombinationrules

[Mark Abraham]

Shuangxing Dai wrote:
> No. I only changed 3 lines in ffgmxnb.itp. But there are hundreds of 
> lines of error informaion like this:
> ERROR 77 [file ffgmxnb.itp, line 144]:
>   Trying to add LJ (SR) while the default nonbond type is Buck.ham (SR)

Sure, this is predictable. This force field's files come filled with LJ 
functions, and the default pertains to all of them. So you can either 
have one form or the other.

The issue is that if the force field generates the functions for 
interactions based on the atom types, you can't mix the two types, since 
the result is not defined. Sorry I didn't pick this up the first time, 
but if you'd had a descriptive header "I'm trying to add parameters for 
(?) crystalline zinc oxide, which require Buckingham nonbonded 
interactions" then your strategy problem would have been obvious.

I'm guessing here, but you may not be constrained by the default type if 
you just add your functions in the right form in your [ molecule ] 
section. This means you don't need to modify the force field files at 
all! However you will need to supply a non-bonded interaction manually 
for any new atom type with every atom type in your system. Those also 
seems likely to be undefined.

You also shouldn't be using ffgmx for a new simulation, because it has 
been deprecated for years now.

Mark