1: \begin{abstract} To study the resolution required for simulating

2: gravitational fragmentation with newly developed Lagrangian hydrodynamic

3: schemes, Meshless Finite Volume method (MFV) and Meshless Finite Mass method

4: (MFM), we have performed a number of simulations of the Jeans test and

5: compared the results with both the expected analytic solution and results

6: from the more standard Lagrangian approach: Smoothed Particle Hydrodynamics

7: (SPH). We find that the different schemes converge to the analytic solution

8: when the diameter of a fluid element is smaller than a quarter of the Jeans

9: wavelength, $\lambda_\mathrm{J}$. Among the three schemes, SPH/MFV shows the

10: fastest/slowest convergence to the analytic solution. \RED{Unlike the

11: well-known behaviour of Eulerian schemes,} none of the Lagrangian schemes

12: investigated displays artificial fragmentation when the perturbation

13: wavelength, $\lambda$, is shorter than $\lambda_\mathrm{J}$, even at low

14: numerical resolution. \RED{For larger wavelengths ($\lambda >

15: \lambda_\mathrm{J}$) the growth of the perturbation is delayed when it is not

16: well resolved. Furthermore, with poor resolution, the fragmentation seen with

17: the MFV scheme proceeds very differently compared to the converged solution.

18: All these results suggest that, when unresolved, the ratio of the magnitude

19: of hydrodynamic force to that of self-gravity at the sub-resolution scale is

20: the largest/smallest in MFV/SPH, the reasons for which we discussed in

21: detail.} These tests are repeated to investigate the effect of kernels of

22: higher-order than the fiducial cubic spline. Our results indicate that the

23: standard deviation of the kernel is a more appropriate definition of the

24: `size' of a fluid element than its compact support radius.

25: \end{abstract}

26: