Re: [ESPResSo] ESPResSo Digest, Vol 29, Issue 8

[Axel Arnold]

Am Dienstag, 13. Oktober 2009 17:51 schrieb Michael Winokur:
> My C programming skills are minimal and so simplicity of new modifications
> is key.   Olaf suggested I take a closer look at the FENE  interaction but,
> after thinking it over, I believe there is an even simpler implementation. 
> The standard dihedral interaction is "part p_2 bond bondid p_1 p_3 p_4". 
> In this case p_1 and p_4 are needed for ascertaining the torsion angle
> across the bond.  If one made a conditional on the basis of "part p_2 bond
> bondid p_2 p_3 p_3" then whenever the p_1 was set to p_2 or p_4 to p_3 the
> dihedral interaction subroutine could defer to the "dipole orientation" to
> ascertain the needed geometrical constraints.  Is there any reason this
> should/could not be done?

You can do it like this as well, of course. However, since this is after all a 
two-body force (although of bodies with director), I don't see any advantage 
in adding it to the dihedral instead of making it a new two-body bonded 
force. You just copy the FENE code, replace the force and energy computations 
with the ones from the dihedral, but using the directors, and that's it.

> A secondary issue has to do with the subtracted LJ.  My guess is that there
> is no procedure to do the same with the Gay-Berne potential because this is
> the first go at linking GB particles within Espresso.  This all depends on
> whether the subtracted LJ subroutine has been coded to include any GB
> interactions.

No, subtracted LJ only subtracts the LJ interaction. You can however copy the 
code and make a subt_GB analogously.

> I'm not sure I understand how mixtures of GB particles and spherical LJ
> particles are integrated.

For each pair of particle types, you can define whether they interaction via a 
LJ potential or a GB potential, but what you would need is another one that 
describes the potential of sphere and a cigar-shaped object. As such there 
are no GB- or LJ- particles at all in Espresso: all particles have a 
quaternion representation of the rotational state, if you switch on rotation, 
but only the interactions of each pair of particle types determines what 
happens, and they do more than a random rotation. At present, there is no 
potential implemented that could be used to describes the interaction of a 
sphere with a cigar-shaped object, so you cannot really mix GB and LJ 
particles. If you however know how such a potential should look like, you can 
easily implement it by looking at the GB code.

> I also am not sure how the unit vectors u_i and u_j used in the GB
> interaction are established unless they refer back to the dipole
> orientation.  Am I missing something here?

The unit vectors are generated from the quaternion representation of the 
particles orientation, see gb.h . The (magnetostatic) dipole moment is just a 
director and always oriented parallel to the main axis of the particle. 
However, the dipole moment does not represent the minor axis, and therefore, 
you cannot determine the GB axes from the dipole moment, only from the 
quaternions.

Axel

-- 
Dr. Axel Arnold 
Fraunhofer SCAI
Schloss Birlinghoven, 53754 Sankt Augustin, Germany
Tel: +49 2241 14 2575