[gmx-users] how to choose rlist and cutoffs

[Mark Abraham]

On 2/04/2010 3:56 AM, Robin C. Underwood wrote:
> Greetings,
>
> My system consists of large non-polar spheres in tip4p water, and the user
> potential option is implemented to define vdw interactions. Because the radius
> of the sphere is on the order of 1 nm, I am using a cutoff of 1.9 nm for the vdw
> portion.

What the sphere contains and how those parameters were developed is a 
much more important indicator of suitable rvdw choice than wishing every 
atom in the sphere to be able to see every other...

> I would like to save on computational expense by using a cutoff for
> electrostatics (non-user defined) and choosing a much shorter electrostatics
> cutoff, around 0.9 nm. Because I must also decrease my rlist value to 0.9, I am
> wondering how this may effect the interaction energy between the sphere and
> water. In particular, if my rlist is not much larger than the radius of the
> sphere, or less than that radius, will I be neglecting important energetic
> information with rlist=0.9 (even though the electrostatic interaction between
> the sphere and water is practically 0)?

I've said this several times already in the last fortnight, but using a 
model physics similar to that for which the parameters were developed is 
the right approach. You need to treat water-water, water-sphere and 
intra-sphere interactions correctly, so you will need parameters that 
have been developed with that in mind, and to use conditions under which 
those parameters have been demonstrated to be useful. Thus, conditions 
not too far different from those under which TIP4P, etc. has been 
parameterized and tested will be needed.

> I did see section 4.6.3 in the Gromacs manual, which leads me to think that it
> is reasonable to decrease rlist because my vdw interactions should still be
> computed, but I want to be sure this is correct.

Your suggested computation looks well-defined. Whether it's a reasonable 
model physics is unknown.

Mark