GROMACS

With the Verlet cut-off scheme molecules are not whole during the force calculation, whereas this is required for the fitting with orientation restraints. This will cause issues when part of a molecule moves over the boundary of the unit cell. Without time and ensemble averaging this would not affects the potential and forces, only the eigenvectors printed in the log file.
Note that this in practice this likely did not cause issues, since orientation restraints do not work correctly with OpenMP, which is used by default with the Verlet cut-off scheme.