Evolution and properties of self-interacting dark matter subhalos until core collapse

One of the hottest questions in the cosmology of self-interacting dark matter (SIDM) is whether scatterings can induce detectable core-collapse in halos by the present day. Because gravitational tides can accelerate core-collapse, the most promising targets to observe core-collapse are satellite galaxies and subhalo systems. However, simulating small subhalos is computationally intensive, especially when subhalos start to core-collapse. In this work, we present a hierarchical framework for simulating a population of SIDM subhalos, which reduces the computation time to linear order in the total number of subhalos. With this method, we simulate substructure lensing systems with multiple velocity-dependent SIDM models and show how subhalo evolution depends on the SIDM model, subhalo mass and orbits. We find that an SIDM cross section of $\ensuremath{\gtrsim}200\text{ }\text{ }{\mathrm{cm}}^{2}/\mathrm{g}$ at velocity scales relevant for subhalos' internal heat transfer is needed for a significant fraction of subhalos to core-collapse in a typical lens system at redshift $z=0.5$ and that core-collapse has unique observable features in lensing. We show quantitatively that core-collapse in subhalos is typically accelerated compared to field halos, except when the SIDM cross section is non-negligible ($\ensuremath{\gtrsim}\mathcal{O}(1)\text{ }\text{ }{\mathrm{cm}}^{2}/\mathrm{g}$) at subhalos' orbital velocities, in which case evaporation by the host can delay core-collapse. This suggests that substructure lensing can be used to probe velocity-dependent SIDM models, especially if line-of-sight structures (field halos) can be distinguished from lens-plane subhalos. Intriguingly, we find that core-collapse in subhalos can explain the recently reported ultrasteep density profiles of substructures found by lensing with the Hubble Space Telescope.