lammps to calculate thermal conductivity of molecular like water

Dear lammps-users
There is a problem has been bothering me for a long time.I want to use lammps to calculate thermal conductivity of molecular like water.
There are two methods (NEMD and EMD )calculating thermal conductivity.
if we use EMD like Green-Kubo, the thermal conductivity of water overestimates experiment values of water so much(30%-50%). However if we use NEMD like fix Heat or fix thermal/conductivity command(RNEMD), the energy of system is not conserved,the energy is increasing slowly.
So I really want to know if we can use lammps to calculate thermal conductivity of molecular,because lammps was designed for atomic systems not molecular ones.
Looking forward to your reply!
Jack zhuang

Dear lammps-users
    There is a problem has been bothering me for a long time.I want to use
lammps to calculate thermal conductivity of molecular like water.

You sound quite philosophical. Problem bothering you for a long time...
So, what have you done during that long time in addition to allowing the
problem to keep bothering you?

     There are two methods (NEMD and EMD )calculating thermal
conductivity.

if we use EMD like Green-Kubo, the thermal conductivity of water

overestimates experiment values of water so much(30%-50%).

Really? And how do you convince this list that such overestimation is not
the result of you making a ton of mistakes?

However if we use NEMD like fix Heat or fix thermal/conductivity
command(RNEMD), the energy of system is not conserved,the energy is
increasing slowly.

Have you even tried browsing the lammps archives for the answers to your
questions?

      So I really want to know if we can use lammps to calculate thermal
conductivity of molecular,because lammps was designed for atomic systems
not molecular ones.

Sure you can. Yet, is this answer going to make any difference to you? If
you had spend a bit of time understanding how the different methods operate
on the system you would have been able to answer the question yourself.

    Looking forward to your reply!

Carlos