Homec4science
Diffusion lammps (efficient_neuronet) Browse

lammps/examples/gaussFlow/README
95755ba8b2e9efficient_neuronet

Last Change
Raw File

README
View Options

The input script in.GD is an example simulation using Gaussian dynamics (GD).
The simulation is of a simple 2d Lennard-Jones fluid flowing through a pipe.
For details see online LAMMPS documentation and
Strong and Eaves, J. Phys. Chem. Lett. 7(10) 2016, p. 1907.
Note that the run times and box size are chosen to allow a fast example run.
They are not adequate for a real simulation.
The script has the following parts:
1) initialize variables
These can be modified to customize the simulation. Note that if the
pipe dimensions L or d are changed, the geometry should be checked
by visualizing the coordinates in all.init.lammpstrj.
2) create box
3) set up potential
4) create atoms
5) set up profile-unbiased thermostat (PUT)
see Evans and Morriss, Phys. Rev. Lett. 56(20) 1986, p. 2172
By default, this uses boxes which contain on average 8 molecules.
6) equilibrate without GD
7) initialize the center-of-mass velocity and run to achieve steady-state
The system is initialized with a uniform velocity profile, which
relaxes over the course of the simulation.
8) collect data
The data is output in several files:
GD.out contains the force that GD applies, and the flux in the x- and
y- directions. The output Jx should be equal to the value of
J set in section 1, which is 0.1 by default.
x_profiles contains the velocity, density, and pressure profiles in
the x-direction. The pressure profile is given by
(-1/2V)*(c_spa[1] + c_spa[2]), where V is the volume of a
slice. The pressure profile is computed with IK1, see
Todd, Evans, and Davis, Phys. Rev. E 52(2) 1995, p. 1627.
Note that to compare with the pump method, or to
compute a pressure drop, you must correct this pressure
profile as described in Strong 2016 above.
Vy_profile is the velocity profile inside the pipe along the
y-direction, u_x(y).
c4science · Help