1: \begin{abstract}

2: 

3:   We perform cosmological simulations of the intergalactic medium

4:   (IGM) at redshift $z\sim 3$ using the numerical

5:   gravity-hydrodynamics codes \texttt{GADGET-3} and \texttt{Enzo} for

6:   the purpose of modelling the gaseous environments of galaxies. We

7:   identify haloes in the simulations using three different

8:   algorithms. Different rank orderings of the haloes by mass result,

9:   introducing a limiting factor in identifying haloes with observed

10:   galaxies. We also compare the physical properties of the gas between

11:   the two codes, focussing primarily on the gas outside the virial

12:   radius, motivated by recent \HI\ absorption measurements of the gas

13:   around $z\sim2$--$3$ galaxies. The internal dispersion velocities of

14:   the gas in the haloes have converged for a box size of 30 comoving

15:   Mpc, but the centre-of-mass peculiar velocities of the haloes have

16:   not up to a box size of 60 comoving Mpc. The density and temperature

17:   of the gas within the instantaneous turn-around radii of the haloes

18:   are adequately captured for box sizes 30~Mpc on a side, but the

19:   results are highly sensitive to the treatment of unresolved, rapidly

20:   cooling gas, with the gas mass fraction within the virial radius

21:   severely depleted by star formation in the \texttt{GADGET-3}

22:   simulations. Convergence of the gas peculiar velocity field on large

23:   scales requires a box size of at least 60~Mpc. Outside the

24:   turn-around radius, the physical state of the gas agrees to 30

25:   percent or better both with box size and between simulation

26:   methods. We conclude that generic IGM simulations make accurate

27:   predictions for the intergalactic gas properties beyond the halo

28:   turn-around radii, but the gas properties on smaller scales are

29:   highly dependent on star formation and feedback implementations.

30: 

31: \end{abstract}

32: