Re: [ESPResSo-users] problem with magnetic particle system

[Rudolf Weeber]

Hi Judith,
On Thu, Aug 16, 2012 at 04:14:56PM +0200, address@hidden wrote:
> Thanks, I tried to adapt the accuracy and the Bjerrum length to the order
> of my dimensions.
I'm a bit confused by your units.
As I understood from the script, you use kT=1.
At the same time, you use a lennard-jones cutoff of 0.1 and a dipole moment of 
6 * 10^3.
That would mean that dipoles at contact have an energy of 3 * 10^10 kT per 
particle.
That's very much. For usual magnetic nano-particles, the energy at contact 
should be single-digit in units of kT.
In conjunction with the energies you seem to use, the 10^-2 accuracy is not 
reachable.

Typicl dimensions for a ferrofluid in reduced units would be:
Lennard-Jones:
epsilon=1, sigma=1, 
r_cut=sigma *2^(1/6) if you want purely repulsive LJ (This is called 
Weeks-Chandler-Anderson potential)


The dipole moment would be defined by 
lambda =m^2/(sigma^3 kT), where lambda is maybe between 1 and 7.

For the Langevin thermostat:
kT=1 and gamma=1. If you chose a higher gamma, you need a smaller time step.

Time step:
0.001 to 0.1

P3m Accuracy: 10^-3 to 10^-4

Density:
Volume fractions between 1% and maybe 15%.

I hope, I didn't misread something in your script.
If you have no success with the above parameter set, let me know.

Regards, Rudolf