[gmx-users] mixing rules for LJ parameters?

[Erik Lindahl]

On Dec 19, 2006, at 8:35 PM, Nicolas Winter wrote:

>
>> Mixing rules only apply when no specific parameters for an  
>> interaction are provided. "Gromacs FF" is mostly GROMOS87 (i.e.,  
>> don't use it for new runs), which has a handful of exceptions from  
>> the standard combination rules.
>
> What do you mean by "new runs"? And why not use it? Parameters are  
> too outdated or something? They don't work?

Don't use it when you start a new project, unless you have a very  
particular reason do to so (compatibility with an old project or  
something).

The later GROMOS96 versions are simply better since it's basically  
the same parameters that have gone through several additional rounds  
of improvements - with faster computers it has also been possible to  
calibrate non-trivial properties such as free energy for more complex  
systems.

Then there's of course also all-atom force fields like OPLS-AA/L or  
Amber, but that's a different decision that involves taste too.

Cheers,

Erik