I am trying to do Poiseuille flow by applying an external acceleration.
I first ran it with real units ( accn = 8 E-08 A/fs^2) but it gave me a zero velocity in the channel.
Then I ran the same simulation with units lj ( reduced accn =0 .1) and then I got a nice parabolic flow profile.
I am guessing that in real units, it is just treating the acceleration as zero. Does anyone know how many significant digits are read by LAMMPS or how can I solve this problem in real units ?
University of Illinois at Urbana-Champaign
How did you convert your reduced acceleration of 0.1sigma/tau^2 to
real units? I quickly estimated it to be rather on the order of 1e-5
A/fs^2. Therefore with a hundred times smaller acceleration, it can be
understood that you get a hundred times smaller velocities...
More generally, it is possible to run the exact same
simulation in any units you choose. But as Laurent
indicates, all your input params in your input script
have to be matched one by one with the previous
units choice in order to do that.
Following are the values for sigma and tau
tau= 2.16 ps =2160 fs
Therefore, 0.1 sigma/tau^2 = 7.287E-8 A/fs^2
So I think the numbers are correct.
But how do you compute tau, which is not a fundamental parameter of
the simulation? I.e., which values of epsilon and mass do you use?
It's just that your value of tau seems quite large.
Are you using the exact same system (Lennard-Jones), only with real
units? In that case, you have to make sure that every parameter of the
simulation has been scaled correctly (e.g. temperature, pressure,
To check if your acceleration is really taken as zero in real units
(but why?), you can still try to see what happens when using a larger