calculating thermal conductivity for liquids

please always respond to the mailing list.

Dear Dr. Kohlmeyer,

Thank you so much, when I used nve on only fluid atoms, walls remain fixed and I
can equilibrate my system for a while.

The next step is to apply different set of temperatures to the walls and look at
the temperature distribution inside the liquid. So I used fix nvt on each wall
with different T, but again walls move away from the domain. I am aware of

sure. you cannot have fixed, i.e. non-moving atoms _and_ have them move,
i.e. contain kinetic energy. if you want _restrain_ atoms to a specific position
in space, you need to use fix spring/self.

Muller-Plathe, green-kubo, or rescaling fluid particles (in the vicinity of the
wall) methods. What I would like to do is what ppl do in experiment and apply
\delta T to the system. If I fix my atoms so they don't move the system how can

you can do whatever you want, but if that will be an efficient way
to give you the result you desire is a different question. experiments
setups are often restricted by technical limitations that don't apply
to computer simulations, whereas in computer simulations you
are often restricted by statistical sampling limitations. thus mimicking
the experiment may be intuitive (but do you also mimic the size?),
but not necessarily the best choice. however, what is the case is
for you to determine.

even they communicate with fluid particles to transport heat. I also tried using
springforce on the wall the same problem happened.

I know you are very busy but I have been away from MD field for a while and
really need some guidance.

sorry, but a mailing list is not a "school for MD".
we are happy to discuss technical issues and
benefits and shortcomings of the implemented
method. if you need a "personal trainer" i suggest
you sit in on relevant classes, if you can, or
hire somebody.

cheers,
    axel.