[lammps-users] compute heat/flux

No I can't - unless you think LAMMPS is doing something
wrong or there is a bug, then I don't have time to help people
figure out how to do simulations correctly.


2011/1/10 王晓亮 <[email protected]...>:

Thank your review

Hi Xiaoliang,

I assume you use compute heat/flux with Green-Kubo formalism to calculate the thermal conductivity just like the Ar example in the manual. However, Green-Kubo formalism is not suitable for your heterostructures. It is only applicable for homogeneous systems. You should try to use non-equilibrium molecular dynamics method which can be performed with

  1. giving heat flux with fix heat command to obtain stable temperature gradient, then get thermal conductivity,
  2. keep the two ends of your system at two fixed temperatures with fix temp/rescale command, get a stable energy change to calculate heat flux, then you get thermal conductivity. Note: Maybe you need modify the fix_temp_rescale.cpp to output the energy change. I am not sure whether the new version has this function.



Thank you!I try to do it in non-equilibrium molecular dynamics method .But i read some article about thermal conductivity in equilibrium molecular dynamics method by Green-Kubo such as

Hi, Xiaopeng

I am interested in the distinction of heterostructures and homogeneous structures. Do the former always have more than one phases while the latter always have only one phase ? Thanks.


No, both are not true. A heterostructure may also only have one phase but with lots of defects, voids, dislocations, serious anisotropic properties, or even just nonuniform density distribution. If you can treat all the areas which are small enough in the structure as the same one, then it is homogeneous.