1: \begin{abstract}

2: We investigate the impact of using high-order numerical methods to study

3: the merger of magnetised neutron stars with finite-temperature

4: microphysics and neutrino cooling in full general relativity. By

5: implementing a fourth-order accurate conservative finite-difference

6: scheme we model the inspiral together with the early post-merger and

7: highlight the differences to traditional second-order approaches at the

8: various stages of the simulation. We find that even for

9: finite-temperature equations of state, convergence orders higher than

10: second order can be achieved in the inspiral and post-merger for the

11: gravitational-wave phase. We further demonstrate that the second-order

12: scheme overestimates the amount of proton-rich shock-heated ejecta, which

13: can have an impact on the modelling of the dynamical part of the kilonova

14: emission. Finally, we show that already at low resolution the growth rate

15: of the magnetic energy is consistently resolved by using a fourth-order

16: scheme.

17: \end{abstract}

18: