Re: [ESPResSo-users] Fwd: Re: Scaling of tabulated angle potential

[Vincent Ustach]

OK. I think that script I attached still needs a lot of work. Increasing to larger d\theta values of order 1 gives a large deviation of order 10.

I am particularly unsure of how the force \dot d\theta should be calculated, and I need to check that the rotation is correct.

Anyways thank you for your help, I think I will use the tabulated angle potential without a 1/r scaling.

Best Regards,

Vincent Ustach

--Vincent Ustach
  University of California, Davis

On Thu, Jul 24, 2014 at 2:43 AM, Axel Arnold <address@hidden> wrote:

Just to have it also on the mailing list...

On 07/23/2014 01:59 PM, Vincent Ustach wrote:

Hi Espresso-verse,

I know there are a few threads of this nature already in the archives,
but I have a clarifying question about the tabulated angle potential.

I am building a coarse grained polymer from tabulated potentials
calculated by probability densities. The backbone is composed of beads
of type B and each B bead has one pendant group of type A. Therefore,
I have nonbonded potentials, AB and BB bonded potentials, and BBB and
ABB angular bond potentials.

The UG states that the force must be scaled with the inverse length of
the connecting vectors. For the angular bond potentials, do I need a 3
dimensional map of V(r,phi) in order to calculate this? Right now I
only have V(r) and V(phi)

The part that confuses me is that if one builds the tabulated file
like this

#N_points    0    3.14159
phi    (-dV/dphi)/r     V(phi)
....
....

Then the r to use is not obvious to me.

>From the code it looks like there is no 1/r in angle or dihedral
potentials. You can also test that using
tools/check_potential_and_forces.tcl, which samples force and potentials
and checks by numerical differentiation whether these are likely
correct. I think it can also work with angular potentials.

Axel


--
JP Dr. Axel Arnold
ICP, Universität Stuttgart
Allmandring 3
70569 Stuttgart, Germany
Email: address@hidden
Tel: +49 711 685 67609