[gmx-users] Re:[gmx-users] box of methanol

[Erik Lindahl]

nicolis at guideo.fr wrote:
> OK here are my density results on the OPLS methanol box. I need your 
> opinion on that:
> I tried two different boxes, both cubic, one 3x3x3 and one 5x5x5 both filled 
> by genbox with the OPLS methanol topology I send a few days ago. The 
> 3x3x3 box equilibrates at a mean density of 759 and the 5x5x5 at a mean 
> density of 807, which is closer to the real 791. Both values are better than 
> the 851 I obtained with the G43a1 force field in a 3x3x3 box.
> 
> Does this mean I have to use a 5x5x5 box for my simulation or do I have to 
> modify something on my cutoffs?
> 
> For both boxes I used PME but with different distance parameters.
> 
> In the 3x3x3 box I use:
> rlist                    = 1.3
> coulombtype              = pme
> rcoulomb-switch          = 0.7
> rcoulomb                 = 1.3
> vdw-type                 = Cut-off
> rvdw-switch              = 0.7
> rvdw                     = 1.3
> 
> In the 5x5x5 box I use:
> rlist                    = 2.2
> coulombtype              = pme
> rcoulomb-switch          = 1.2
> rcoulomb                 = 2.2
> vdw-type                 = Cut-off
> rvdw-switch              = 1.2
> rvdw                     = 2.2
> 
> Thanks for any suggestion.
> 
> Yiannis
> 
> P.S. If you think these are OK I will upload the box and the itp on the 
> gromacs web topologies
> 

Hi Yannis,

That sounds perfect! If you want to try one additional setup I would 
suggest the 5*5*5 box with

rlist = 1.0
coulombtype = pme
rcoulomb = 1.0
rvdw = 1.4
DispCorr = EnerPres

rcoulomb-switch & rvdw-switch aren't used for PME & cut-off LJ,
so that won't matter.

The main difference is that we turn on the dispersion correction, which
should decrease the density slightly, hopefully closer to 791 than 807!
Did you really find it necessary to use cutoffs at 2.2, or was it just
the first one you tried?

Go ahead and upload - we might even include it in the distribution :-)


Cheers,

Erik