NPT pressure problem

Dear All

I am trying to calculate 555 nm box of electrolyte solution. The solutes are mixed with solven by considering 1M concentration. However, to gain bulk properties I need to get equilibrium system at 1 atm pressure. Therefore, First I run the NVE with langevine fix. Then I use press/berendsen and temp/berendsen to get the 1 atm pressure. Then I run the npt. But I really got confuse with some issues

  1. When I run press/berendsen and temp/berendsen after NVE I got the my desired temp and pressure with a few steps. After that all values got constant ( e.g Pot_Eng, Tot_Eng everything). I am wondering is this OK for press/berendsen and temp/berendsen? ( Since I could not find any example for this fix I got confused)

  2. After press/berendsen and temp/berendsen calcualtion when I run npt with pressure 1 atm, my pressure changed with a large amount ( e.g from 1.02 to -895.23 ) and then pressure fluctuate ± 600 at every 100 steps.

*** Units is real** and command is like this

fix 1 nve all
fix 2 all langevin 303 303 100 67891

Then for berendsen

fix 1 all press/berendsen iso 1.0 1.0 1000 modulus 1000
fix 2 all temp/berendsen 303 303 100

Then for npt calculation

fix 1 all npt temp 303 303 100 iso 1.0 1.0 1000

I reach at this stage of calculation by using manual and mail lists. And I also tried to find a reasonable solution for present stage. However, I failed to make it, therefore, it will be very helpful if experts show me the way thus I could solve the problem or sort out my mistakes.

Thank in advance

Regard

Anderson

Houston University

Hello Anderson, high fluctuations of the virial stress are common. This is discussed in the literature, textbooks, and you can verify that in example systems that use fix NPT.

For more useful feedback, you need to ask a direct question on a specific problem that has not been discussed already in the literature or LAMMPS documentation.

Giacomo

Hi Giacomo

Thank you for your reply.

However, I know that the fluctuation is common specially for solid and liquid. But, ±600 atm is might be very high range in my sense. Do you think that its OK?

Thanks

Regard

Jeams

Hi Giacomo

Thank you for your reply.

However, I know that the fluctuation is common specially for solid and
liquid. But, +-600 atm is might be very high range in my sense. Do you think
that its OK?

nobody can answer this, because this depends on the compressibility of
your system, which in turn depends on your force field and density.

one general recommendation. before switching to fix npt. you should
run for a while with fix nve *without* thermostat and barostat to see
a) whether you system is really equilibrated (berendsen barostat and
thermostat can easily hide that) and b) whether your set up properly
conserves energy.

axel.

Alex’s hints are awesome!

But, how to realize real equilibration via running with NVE?
and also, what’s the point of assessing a system for energy conservation?

Alex's hints are awesome!

But, how to realize *real equilibration* via running with NVE?

you cannot. due to the immense dimensionality, you cannot prove
whether your sampling is sufficiently representing phase space. you
can only see if you are not.

and also, what's the point of assessing a system for energy conservation?

using a thermostat and barostat, especially of the berendsen type, can
hide errors from a bad choice of simulation parameters. with plain fix
nve, those are exposed. thus, especially when you are simulating a
system where you have no previous experience or have limited
experience in setting up MD simulations, it is best practice to run
for some time with fix nve only to see if energy is sufficiently
conserved before turning on a thermostat/barostat again.

axel.