verlet list does not update the position of the ghost atoms (periodic images)

[felixmusil]

[mastricker]

Is there an example for this?

The neighbour list code update_self resets all position vectors, so at a first glance, this should not happen.

So I guess this is related to the skin parameter? I.e. the skin is a variable, but it is never used. So there basically needs to be a check if none of the original atoms moved more than skin and if so, only update the ghost positions - right?

As for now, I do not see this a bug. It is just not implemented, right?

[felixmusil]

It is a bug in the sense that I expected the feature to be working as of now but it actually does not. We are not really using it for the moment (could have been nice in the tests for the gradients but its a detail).

I.e. the skin is a variable, but it is never used

it is used but as skin2 (skin square). see the update function of ANL.

update the ghost positions

the way I see it can be done:

• have a dx variable that keeps tract of the atoms displacements so that the criteria become: did atoms move more than dx_max since the ANL has been built ?
• when the ANL is not updated because dx is < dx_max, move the position of the periodic images