[lammps-users] query on GK

LAMMPS LAMMPS Mailing List Mirror
1

Steve,
I am trying to calculate thermal conductivity using green kubo method using compute heat/flux and fix/correlate. I have a doubt on the example given for LJ Argon.

In this line below, i am not sure about the factor 1.0e25 at the end. I think due to conversion from Ang^3 to m^3 for ($V), there is a factor 10^30, and a factor of 10^-15 due to $dt (from fs to sec) but they both add upto 10^15. But how come it is 1.0e^15??

variable  scale equal ${kCal2J}*${kCal2J}/${kB}/$T/$T/$V*$s*${dt}*1.0e25





Thanks

Ajit
2

I'll let Reese answer this - I don't recall the details
of the unit conversion.

Steve

3

hi Agit,
   The scaling is correct, although it did look odd to me just now.
   The GK formula for thermal conductivity is
k = 1/(k_B T^2 V) \int J.J dt ~ [ (E L/t)^2 t / ( E T L^3) ] = [ E / (t L T) ]
where E are energy units, t time, T temperature, L length. So we get a scale of 10^25 due to the t*L conversion.

hope that helps,
Reese

4

Reese,
I am trying to calculate thermal conductivity for Silicon using stillinger weber potential(metal units) . With reference to the Argon example given in real units, i am making the following unit conversions.

Real Metal

Kcal2J=4186.0/6.02214e23 1.602e-19
kB REMAINS SAME FOR BOTH
The factor 1.0e25 (L*t) in the variable “scale” changes to 1.0e22(since L units is same and t units changes by 1000).

Please let me know if there are any more changes needed.

Thanks
Ajit

5

hi Ajit,
I think that should do it,i.e. change kB and the L*t scale.
Reese