[lammps-users] 1D simulation & thermal conductivity

LAMMPS LAMMPS Mailing List Mirror
1

Dear lammps users,

I have three questions to ask you.

a) Do you know which command can perform a 1D simulation in lammps, such as the command fix enforce2d does?

b) In the low dimensional systems(1D and 2D), the thermal conductivity diverges inherently with the increasing system size. Due to this reason, the flux autocorrelation function will have long tail which results in the divergence of the thermal conductivity kappa when we use the Green-Kubo method to integrate the flux autocorrelation function. Does the command “calculate heat/flux” consider this singularity of low dimensional systems?

c) In the low dimensional systems(1D and 2D), the temperature gradient is inherently not linear, such as the 1D FPU-beta model. So why do we adjust it to the original nonexistent linearity when we use the fix thermal/conductivity?

Thank you！

2

Unlike 2d, LAMMPS does not support 1d directly. You'd
have to fake it out by building a system with a narrow y dim
and make it periodic. Or put ylo/yhi walls in a way that
makes a non-periodic system effectively 1d (in x).

Compute heat/flux doesn't do anything specifix to GK
or 1d (see above). It just computes a heat flux. How you
use it (e.g. in a GK formula) is up to you.

Fix thermal/conductivity also does not impose linearity
on the temp profile, 1d or not. It simply exchanges KE.
The temp profile responds as it wishes. This I believe
is actually an advantage of the Muller-Plathe method
over GK (at least it is for the visvosity analog). No
linear response is assumed or enforced.

Steve

2009/7/8 unica111 <[email protected]...>: