bkmi / non-bonded-periodic Goto Github PK

be able to switch energy from coulomb versus lenard jones energy calc

Don't use a quad deep look in k space. Build a grid of k vectors ahead of time and apply it. In addition, use symmetry to reduce the problem to only real space k vectors

And a more obtuse suggestion:
Take the equations from long range ewald, substitute structure factors, use Euler or symmetry... This process will yield coupled interactions between two particles (but still manage to avoid incorporating any images.) this notion of energy contributions from pairs of particles can be used for the forces. (Another small hint, it involves the complex square of the structure factor)

we need a max sigma and a sigma for every interaction.

Check lennard-jones defs

1. Sigma should be added as a public variable or a method is needed to access it.

2. Calling the particle number is somewhat complicated and it would be easier if it was directly an accessible variable.
   This is how the class neighbour calls it right now:
   __particle_number = System.state().positions().shape[0]

I think the title is self-explanatory

Calculation of the neighbour list needs following variables from SystemInfo:

• sigma

• cutoff radius

How is the box dimension represented?

Proposel for cutoff radius:

`
def set_cutoff_radius(self, sigma):

        # sigma * 2.5 is a standard approximation
        self.__cutoff_radius = sigma * 2.5

        return self.__cutoff_radius

`