[gmx-users] Re: gmx-users Digest, Vol 65, Issue 122

[Mark Abraham]

xiao shijun wrote:
> 
>   Thanks for Mark's fast reply, and It really works.
>   Then, I have another qusetion about the manual: It seems that the 
> exclusion just excludes the nonbond potential of LJ interaction, but 
> Buckingham. If we want to turn off all VDW interaction between some 
> atoms, the energy monitor group is a nice choice. I have no idea if I 
> got it right.
>    And, we can't use LJ and Buckingham in a same system. Unfortunately, 
> I met a force-field that most of the nonbond interactions were described 
> by Buckingham, but LJ only for Ow-Ow among water. I chose nbfunct=2 in 
> [defaults] section,and nbfunct=1 in [nonbond_params] for Ow-Ow. As a 
> result, a error message read:  *Trying to add LJ (SR) while the default 
> nonbond type is Buck.ham (SR).  *I don't know whether it could be 
> settled in gromacs.

AFAIK, you could only do this with tabulated functions. You can supply a 
normal Buckingham table (generated with mdrun -debug?) with vdwtype = 
User, and water-water tables separately with energygrp_table. This will 
be slower than any native implementation, but still decent.

See manual section 6.7 and 7.3.12. I'd strongly suggest starting with 
peptide-in-water test case that was mixing different degrees of LJ 
parameters so that you're confident of your use of the table and 
energygrp_table mechanisms, before trying to mix Buckingham and LJ in 
the way you're describing.

Mark