Gmx dipoles

GROMACS version: 2019.3
GROMACS modification: No
Hi, I am using gmx dipoles to calculate total dipole moment autocorrelation function:
gmx_mpi dipoles -f nve_prod.trr -corr total -s nve_prod.tpr
By definition, autocorrelation function has a maximum at 0, and then drops. But what I’m getting is something else: the ACF goes from 1.0 at t=0 up to ca. 1.001, then down, and then up again. How is this possible? I believe my system is well equilibriated. I am calculating the ACF from an NVE run with the md-vv integrator and 0.5 fs time step.
Thanks in advance for any advice.
Best regards,
Michal Krompiec
Merck KGaA