R: Re: Re: new pair style and current computation

----Messaggio originale----
Da: [email protected]
Data: 19/08/2014 17.07
A: "[email protected]..."<[email protected]...>
Cc: "LAMMPS Users Mailing List"<[email protected]>
Ogg: Re: Re: [lammps-users] new pair style and current computation

Thanks a lot,

[...]

ask if there can be compatibility issues with the heat current

computation

(compute_heat_flux.cpp) and if there is something to be modified that can

be

tricky (I have no runtime errors).

fix heat operates on the kinetic energy,

Sorry, maybe there is a misunderstanding, I'd like to use the heat/flux
command not the fix heat

http://lammps.sandia.gov/doc/compute_heat_flux.html

this command requires correct computation of virial and atomic energies

you are right. this is my mistake.

then, i guess the real question would be, "how can i validate that my
per atom stress tensor is computed correctly?"

unfortunately, i have no experience with that.

This makes me suspect that compatibility is not as obvious as I thought. Maybe
it's worth for me to write a separate program computing the heat current in my
simple two body isotropic case and see if it agrees with a lammps ouput.

the only test that i
can think of right now would be to compare summing the tensor over all
atoms and comparing the result against what you get from computing F
dot r for a simple system without bonds.

actually I didn't understand your suggestion... can you explain it better?

in LAMMPS there are two ways to compute the total stress tensor

1) is the sum of all per atom contributions
2) computing all forces and then computing the stress tensor from the
dot product of the total force on each atoms and the position of that
atom.
  please see the function Pair::virial_fdotr_compute()

for most cases, LAMMPS uses the second options to compute the global
stress (because it is faster).

axel