Results of our numerical study on the early amplification stage (t = 870 Myr) without feedback for different cleaning and numerical dissipation. Top panel shows a face-on rendering of the magnetic field strength in the galactic disk. Bottom panel shows the corresponding phase space of divergence errors for each of the four cases. We can see that increasing the cleaning speed efficiently reduces the divergence error, while still leading to amplification within the filament and fragment. In the higher numerical dissipation case (α_B = 2) we can see that we still have quite high divergence error, but the amplification is reduced much more than in the f_clean = 100 case. Indicating that the reduction in amplification is mainly driven by increased numerical dissipation and not divergence errors.