[gmx-users] Npt equilibration of the membrane-mimicking CCl4 layer

[Justin A. Lemkul]

James Starlight wrote:
> Justin,
> 
> 
>     Larger cutoffs would not only make the problem worse (read the error
>     message carefully and consider the minimum image convention), but it
>     would also potentially break the validity of the force field model.
>      The vdW cutoff for Gromos96 should be 1.4 nm, not 0.9 nm, in any case.
> 
> 
> This sounds controversially alitle :)
> 
> Today I've tried such simulation with the 1.2 cutoffs and obtain the 
> same error on the 3rd ns ( density was 1.35 )

Still wrong.  Set rvdw = 1.4 and rcoulomb = 0.9 for Gromos96.  Don't play 
haphazardly with cutoffs.  Use what the force field dictates unless you can 
demonstrate that what you are doing is superior.  The original value of rcoulomb 
was 0.8 with reaction field, but 0.9 is fine in the case of PME.

>  From the error message I've understood that the problem was due to the 
> Z- box vector wich was decreased up to the 3 nm ( this is smaller rhan 
> 2x 1.2 of R_list cutoff ). So in accordance to that message I understood 
> that cutof must be > 1.5 in the case of my system didnt it?
> 

No, absolutely not.  You don't *want* the cutoffs to be equal to twice the box 
vector, they must always be *less* than this amount to avoid spurious PBC 
interactions due to minimum image violations.  Are you familiar with this 
concept and its implications?

http://www.gromacs.org/Documentation/Terminology/Minimum_Image_Convention

You said in your first message:

"2) Initially I've tried to use membrane mimicking dimensions 8.6   6.5   3.0 
but during equilibration my box was shrinked to the 8.0 6.0 2.0"

So your problem comes from the box shrinking to 2.0 nm in the z-dimension, which 
is less than twice your longest cutoff.  Your box is compressing in all 
directions, which means your initial configuration is too diffuse and the 
application of NPT conditions is squishing it together to arrive at the density 
the force field model predicts.  Build a system with an initial density closer 
to the target value.

-Justin

-- 
========================================

Justin A. Lemkul
Ph.D. Candidate
ICTAS Doctoral Scholar
MILES-IGERT Trainee
Department of Biochemistry
Virginia Tech
Blacksburg, VA
jalemkul[at]vt.edu | (540) 231-9080
http://www.bevanlab.biochem.vt.edu/Pages/Personal/justin

========================================