espressomd-users
[Top][All Lists]

## Re: [ESPResSo-users] LJ anisotropic potential

 From: Tristan Bereau Subject: Re: [ESPResSo-users] LJ anisotropic potential Date: Wed, 21 Jan 2015 08:52:45 +0000

Hi Damian,

Please make sure to always cc the ESPResSo mailing list, that way everyone benefits from the discussion.

Here's a number of tips:
- the force is simply calculated by differentiating the potential with respect to all atoms involved. It's the angular dependence in the potential that gives rise to these complicated terms in the angular brackets.
- As you can see in the derivation and read in the ESPResSo documentation, I've reexpressed the force exerted on particles j and n into the other particles. That's because particles j and n don't exist in my system, I just need to compute their geometry from the local structure, and I later propagate their force contribution into the other particles.
- If you have an anisotropic potential, the force will, by definition, not be radial (i.e., along the pairwise vector). Thus you either need to propagate torques that the potential generates, or convert these torques into forces--that's what I've done here.

Best,
Tristan

On Wed Jan 21 2015 at 3:10:17 AM Damian <address@hidden> wrote:

Hello Tristan.

I’ve revised your thesis and it gives me a better undersatnding on the therms the force relies, but I still don’t get how the therms into the brackets were calculated, I suppose that for every angular the therms in brackets is the _expression_:

{_expression_ in brackets} = -sin(theta_jik)*theta_jik_unitary

{_expression_ in brackets} = -sin(theta_ikn)*theta_ikn_unitary

I can’t figure the system reference to obtain these therms.

If you can give me another hint I would be marvelous!

Sent: martes, 20 de enero de 2015 03:37 p.m.
Subject: Re: [ESPResSo-users] LJ anisotropic potential

Hi Damian,

Can you have a look at this document page 26:

and see if that helps?

Best,

Tristan

On Tue Jan 20 2015 at 10:00:32 PM Damian <address@hidden> wrote:

Hello guys!

I’m trying to modify the code for the Lennar-Jones anisotropic potential (LJangle.cpp,hpp) but I can’t get how the gradient of potential is implemented, I mean

The potential is in the form U(r,theta1,theta2)=  dU/dr *  r_unitary + dU/dtheta1 * theta1_unitary + dU/dtheta2 * theta2_unitary

And I can’t figure it out how the vectors theta1_unitary and theta2_unitary were calculated.