Re: [ESPResSo-users] mdds

[Rudolf Weeber]

Hi,
> I have been playing with espresso-3.3.0 for the past couple of days and
> have a small question.
> 
> I put two dipoles with different dipole moments, different orientation and
> distance between them and calculate the magnetic interaction energy of the
> system. The result doesn`t change depending on the l_B parameter of "inter
> magnetic" command. I use the magnetic direct summation with n_cut=0. There
> are no other interactions (lj, thermostat etc)
The mdds method, as far as I know, was only written to verify 
schemes like dipolar p3m.

If you do not need periodic boundary conditions, you're better off with the 
dawaanr method (dipolar all-with-all and no replicas).


> 
> Is it right that the MDDS algorithm does not take into account the magnetic
> Bjerrum length?
> Is this the correct behaviour?
No, it's a bug. I just had a look in the code.
In magnetic_dipolar_directsum_calculations(), the prefactors are messing for
the energy. Forces and torques seem to be ok.

I also (hopefully) removed two other issues (bounds checks for small dipoles). 
The code was probably always used with dimensionless units, before.

The fix is available at my github repository in the master branch 
(www.github.com/RudolfWeeber/espresso).
I need to point out though that there is no coverage for any of the direct 
summation magnetic methods in the testsuite, so you need to do your own testing.

Regards, Rudolf