[gmx-users] opls combination rule and pairs info

[Erik Lindahl]

Hi,


> 1) why is it that in ffoplsaa.itp combination rule 2 is the default 
> when I believe oplsaa
> has been parameterized with geometric mean not arithmetic mean?

That's a known bug, I've already changed it for the next version. The 
difference isn't huge, but you can go ahead and change it to "3" in 
ffoplsaa.itp if you want the right behavior already now.

>
> 2)I made my own itp file for the sugars with stretch bend torsions 
> charges and lj parameters from a paper from tirado rives and > jorgensen.
> The question is the following: if I don't put a [ pairs ] entry in the 
> itp file and I just specify "generate pairs yes" and the fudge factors 
> to 0.5 at the beginning of the file, I do not get the same energy as 
> if I put the [ pairs ] entry with all the corresponding 1,4 pairs 
> (always keeping the fudge factors to 0.5).
> If I run g_energy I do not even see the entry for 1-4 interactions in 
> the lj and coulomb section. I thought that if you specified generate 
> pairs yes, that would automatically check for the 1-4 pairs and apply 
> the corresponding fudge factor.
> Could you please shed light on this issue??
>

Sure. The defaults setting and fudge parameters refer to the generation 
of the 1,4 _parameters_. A lot of force fields (like Gromos96) have 
special values (not scaled) for these, so some of the force fields 
specify all 1,4 parameters explicitly. If you generate them by scaling 
you don't need the [pairtypes] section.

The [pairs] section specifies the actual atoms involved in 1,4 
interactions for this molecule.

All topologies should have the [pairs] section, but if you implement 
your new molecule as a building block in the rtp file the generation of 
the actual topology file (including the pairs section) is completely 
automatic.

Cheers,

Erik