[gmx-users] Unreasonably high pressure values

Kamps, M. m.kamps at student.rug.nl
Tue May 16 11:03:56 CEST 2017


Dear Mark,

Thanks for your answer. I did some experiments and I think I've
located the problem.
Within my simulation there are two main components: Two surfaces and a
polymer fluid. These two surfaces are created by placing gold atoms in
a FCC lattice (lattice parameters derived from the LJ potential). The
only interactions between these surfaces are the LJ potentials.

When I run a simple NVT equilibrium on a basic system (without
pressure, without accelerations, etc), the pressure is already really
high (again, around 6000 bar, it stabilizes around this value!). When
I run the same NVT equilibrium on only the fluid (so no surfaces), the
pressure remains stable at around 0-1 bar, as should be expected.

Performing an NPT equilibrium after this first one grants the same
results. The system with the surfaces shows unreasonably high pressure
values, while the fluid-only system behaves as expected. (The
pressures are calculated via gmx energy)

My guess, the problem is related to the gold surfaces. However, when
running only a NVT and NPT equilibration on the surfaces, nothing
happens. There are no high stresses within the system, everything runs
fine. The only difference is the pressure, which is extremely high.

Is there a way to exclude the surface from the calculations? I am only
interested in the LJ behaviour of the surface towards the fluid. I
cannot use the GROMACS walls options since I am interested in non-flat
surface.

Any ideas?

___________________________

Message: 1
Date: Mon, 15 May 2017 16:35:56 +0000
From: Mark Abraham <mark.j.abraham at gmail.com>
To: gmx-users at gromacs.org, gromacs.org_gmx-users at maillist.sys.kth.se
Subject: Re: [gmx-users] Unreasonably high pressure values
Message-ID:
     <CAMNuMATE7wmshfJJFpMx0QAUNGBpyNbgoxZ=CXuvN22q4YG=4A at mail.gmail.com>
Content-Type: text/plain; charset="UTF-8"

Hi,

Convergence times depend on system type and size, of course, but an
equilibration on an inhomogeneous system should probably run for more than
a few nanoseconds. That might help expose whether the issue is the system,
or the protocol, or the applied force.

Mark

On Mon, 15 May 2017 18:25 Kamps, M. <m.kamps at student.rug.nl> wrote:

> Dear GMX users,
>
> I am struggling with understanding my systems behaviour.
> I have a semi-isotropic system where my bottom and upper plane are
> composed of gold FCC plates. In between is a fluid, which during
> equilibration is stationary (no acceleration). I equilibrate the
> system with a Berendsen barostat, as the literature suggests. The
> compressibility is that of the fluid used (Hexadecane C16H34), while
> the reference pressure is 1 bar. See the below section of my systems
> MDP:
>
> ; Pressure coupling is on
> pcoupl                    = berendsen        ; Pressure coupling on
> pcoupltype                = semiisotropic            ; uniform scaling
> of box vectors
> tau_p                    = 2.0                ; time constant, in ps
> ref_p                    = 1.0     1.0            ; reference pressure, in
> bar
> compressibility             = 0      8.6e-5            ; isothermal
> compressibility, bar^-1
>
> When I equilibrate the system, everything appears to be normal. The
> volume shrinks a little to compress the fluid, which leads to a
> correct density (correct as in corresponding with known values of the
> real-life density). The fluid behaves as expected and everything looks
> good.
>
> I then want to use the system to accelerate the fluid, however that
> blows up the system. Blowing up as in; fluctuations in the box size,
> shrink, expand, rapid shrink, rapid expand, extreme shrink, blowing
> up.
>
> Upon inspection, during equilibrating the pressure of my system nicely
> converges, however, it converges to -6000 bar, which seems rather
> strange.. How does this happen? What am I doing wrong?
>
> I use a 0.5 fs timestep, equilibriate over 400.000 steps (0.2 ns),
> 300K v-rescale temperature coupling, PME electrostatic calculations.
>
> Kind regards,
> Mark


More information about the gromacs.org_gmx-users mailing list