Thermal Conductivity

Dear all,

I have been looking at the thermal conductivity of a Lennard-Jones fluid calculated both via Green-Kubo equilibrium simulation and via NEMD approachs with a home made code.

As a benchmark test I performed some calculations with LAMMPS to confirm the validity of my results. Unfortunately however I get fairly different results with the LAMMPS simulations I have ran.

I have also however read various papers which report the same values that I get from my own code and therefore am presuming that my code is indeed correct and either:

(1) I am running the LAMMPS calculation incorrectly (99.9% likely)


(2) There is an intrinsic problem with how LAMMPS calculates the thermal conductivity.

I attach the input for my LAMMPS simulation in the hope that someone could help me with this issue.

For the record the value I am getting from my simulations and the literature is around k=6.41, whereas LAMMPS is producing a value of k=9.5. All units are that of LJ.

Many thanks in advance, Jeff

in.GK (1.86 KB)

Hi Jeff,

If you equilibrate longer with NVT, instead of NVE + temp/rescale, you should get closer values.