Therm/cond - molecules

Hi,

Is it possible for the fix therm/cond command to exchange KE between moelcules rather than atoms?

I’m trying to measure the thermal conductivity of a molecular crystal via rNEMD.

Thanks,

Romain

Hi,

Is it possible for the fix therm/cond command to exchange KE between
moelcules rather than atoms?

​if it is not mentioned in the documentation (and i assume you are asking
because you looked at it and didn't see anything), then it is not available.
LAMMPS comes without a GUI but it is WYSITMIWYG (i.e. any supported feature
is documented).

you may have to consider using an alternate approach to determine thermal
conductivity.
http://lammps.sandia.gov/doc/Section_howto.html#howto-20​

axel.

Thanks for the reply Axel.
Indeed, I didn’t see any mention of molecule’s KE exchange in the documentation, and rNEMD is presented in the literature as very efficient (especially compared to the “direct” thermostated method).

Anyway, I’ll look at the other methods, and maybe also the fix itself to see if it can be easily modified.

Best,
Romain

The hard part will be how to compute and assign KE to an entire molecule.

If the molecule is not entirely on one proc that will require coordination

between procs. In principle a molecule could span several procs. Also,

compute the COM to figure out which rNEMD bin the molecule is “in”

will be non-trivial. There might be no good alternative to doing MPI_Allreduce

operations and using something like compute com/chunk.

Steve

Thanks for the input Steve. I hadn’t thought of these possible complications.
I haven’t tested the potential yet, but if I can get away with the direct method I’ll probably go for that. Unless GK does the job, but I haven’t looked at it yet.

Thanks