Periodic boundary conditions not working for water dynamics

[FranklinHu1]

Describe the bug
A clear and concise description of what the bug is.

Hello!

I am currently trying to run xtb dynamics for a periodic system of water consisting of 64 water molecules in a periodic box with side lengths of 12.42 angstroms. However, it seems that the trajectories produced by xtb do not take the periodic boundary conditions into account and the structure of water is incorrect. The energy of the system also drifts.

To Reproduce
Steps to reproduce the behaviour:

Attached to this issue is an input file and a pdb geometry file for the 64 water molecules in a zip file. The dynamics run without any errors from xtb, and I ran the following command:

>>> xtb W64.pdb --gfnff --omd -I input.inp

The dynamics terminates normally from here and generates an xtb.trj file and other files expected from a dynamics run (e.g. scoord files). I made sure to use version 6.7.1 of xtb which, at the time of writing this, is the latest release from GitHub.

Expected behaviour

Upon analyzing and visualizing the trajectory obtained from xtb, the water molecules diffuse out of the box and the trajectory is unwrapped. This is not a problem in of itself, but when I wrap the trajectory into the correct unit cell, the structure of the liquid is very unphysical with atoms overlapping each other:

I would expect that the unit cell is more or less uniformly populated and no atoms overlapping or getting too close together. Furthermore, the system energy noticeably drifts over the course of the dynamics:

The energy should fluctuate around some consistent average value after some equilibration period. These dynamics are being done in the NVT ensemble.

Additional context
Ultimately, the system of 64 water molecules is a test system as we want to eventually move on to simulating lattices of transition metals like platinum and gold.

Any help/clarification would be greatly appreciated!

Thank you very much!

files_for_issue.zip