I’ve noticed that when I dereference the x array for the position of a neighbor with index j from some ith atom’s neighbor list it’ll produce a position outside of the periodic box limits. There’s no fix operating on the box (just a simple nve simulation) and I’m checking values in gdb. Are the ghost atoms adjusted when they’re outside the box? What could lead to neighbor positions outside the box for this simple simulation?
Is the position just barely outside of the box? I might remember incorrectly but I thought that LAMMPS rewraps the positions only when building a neighbour list, so for a short while, positions outside of the box are OK.