[gmx-users] Problem while equilibrating system with frozen groups

[Peter Kroon]

Hey Arnab,


Why do you freeze atoms instead of putting position restraints? Why not
let the loose entirely? Freezing creates large artefacts.

Second, 4 ns equilibration is *way* too short.

Lastly, see http://www.gromacs.org/Documentation/Terminology/Pressure


Peter


On 10-09-18 19:01, ARNAB MUKHERJEE wrote:
> Hi,
>
> I am simulating DNA (infinite) protein Martini system with electric field.
> I have 2 systems, for 1 complete DNA is frozen and for the 2nd one, only 4
> atoms of the DNA is frozen. When I perform NPT equilibration for the
> complete DNA frozen system, it shows me this error :
>
>  Fatal error:
> 4442 particles communicated to PME rank 24 are more than 2/3 times the
> cut-off out of the domain decomposition cell of their charge group in
> dimension y.
> This usually means that your system is not well equilibrated.
>
> But it runs fine for the other system with 4 DNA atoms frozen. For the
> completely frozen DNA, if I freeze only 2 dimensions, letting 1 dimension
> free it runs without complaining. I do not understand why it shows problem
> for the completely frozen DNA with 3 dimensions frozen? Since the DNA is
> infinite I use semi-isotropic pressure coupling keeping the compressibility
> along Z 0. So the pressure coupling options in my .mdp file looks like this
> :
>
> freezegrps = DNA
> freezedim = Y N Y
>
> ; Pressure coupling is off
> ;pcoupl     = no        ; no pressure coupling in NVT
> Pcoupl     = parrinello-rahman
> Pcoupltype  = semiisotropic
> tau_p       = 5.0
> compressibility = 3e-4 0
> ref_p       = 1.0 1.0
> refcoord_scaling = all
>
> I tried it for finite DNA too and it shows same problem for completely
> frozen DNA.
>
> I ran the completely Frozen DNA system keeping 1 dimension free, and
> freezing the other 2 dimensions. I ran the equilibration run for 500 ps. I
> plotted the pressure and the average pressure is 9.15 bar, while the target
> pressure that I had put in the .mdp file is 1 bar. So there is a large
> difference. I was searching regarding this problem, and I found that I need
> to use "energygrp-excl" for frozen groups. So I ran the completely frozen
> DNA system with  energygrp-excl = DNA, using cutoff-scheme = group (since
> energygrp-excl is not supported in Verlet in my gromacs version 5.0.6), and
> the average pressure I get is again 9.28 bar. For the system with 4 DNA
> atoms frozen, the average pressure shows 5.17 bar, which is still different
> from the target pressure.
>
> I am now running these equilibration runs for longer (4 ns), to see if that
> helps to get the average pressure closer to 1 bar. Normally when all atoms
> are free what I found was 0.5 ns was enough to get average pressure 1 bar,
> and equilibrate the system. But I don't understand what is the problem I am
> facing here with frozen atoms?
>
> I turn off the electric field during the equilibration runs.
>
> I would highly appreciate any help!
>
> Thank you very much,
>
> Regards,
>
> Arnab