Top-left panel: subhalo mass functions evaluated using sub-to-host halo virial mass ratio at z = 0, stacked over three hosts from the Milky Way suite resimulated with Plummer-equivalent gravitational softening lengths of 4.0 (purple), 2.0 (blue), 0.5 (green), and 0.25 (yellow) times the fiducial value of ϵ _fiducial = 170 pc h⁻¹ (blue-green). Solid lines show the mean cumulative mass function stacked over these three hosts, and transition to dashed at the largest value of M _sub/M _host among these hosts set by our convergence limit of 300m _part. The dotted–dashed line shows the fiducial ﹩{V}_{\max ,\mathrm{sub}}﹩ function, corrected for low-ϵ suppression of centripetal forces using the Mansfield & Avestruz (2021) model. The shaded region shows the 1σ Poisson error on the mean, and the bottom panel shows the ratio of the runs with varying softening length relative to the fiducial setting. These subhalo mass functions are consistent with our fiducial results at the ≈10% level down to the convergence limit. Top-right panel: same as the top-left panel, but for normalized maximum circular velocity functions at z = 0. Note that the resimulations transition from solid to dashed at slightly different values of ﹩{V}_{\max ,\mathrm{sub}}/{V}_{\max ,\mathrm{host}}﹩ because there is not a one-to-one relationship between particle count and ﹩{V}_{\max ,\mathrm{sub}}﹩. The systematic variations in ﹩{V}_{\max ,\mathrm{sub}}﹩ as a function of softening length are discussed in Appendix B.3. Bottom panels: same as the top panels, but for varying time stepping.