I agree with Ajing’s advice. In particular, I would advise:
- Unless you know that the x and y dimensions of your wire are well inside and away from the periodic boundary (box length), I would recommend using ( s s p ) as your boundary type. Otherwise, the sides of the wire may start interacting with one another.
- Yes, use nvt, not npt. The “npt” fix will attempt to undo the stretching you’re applying in the z-direction. The “nvt” will not.
- I confess that I’m not up-to-date on lammps’ newest features. If it’s at all possible to do a isobaric energy minimization for zero stress in the z direction, then that would be the preferred first step. If not, you may want to do a zero temperature, zero pressure npt run of dynamics for a very long time.