Isaque Dutra
Projects

SIDM core collapse

A dark matter halo whose particles scatter off each other first grows a low-density core, then collapses to a dense, steep centre. Run time forward to follow its density profile and its lensing strength as a subhalo of a cluster at z = 0.348.

Halo
Time

Time is measured from when the halo formed as an NFW halo, and cannot exceed 9.62 Gyr, the age of the Universe at the cluster's redshift.

Lensing
Now
Collapse time tc
–
τ = t / tc
–
Phase
–
Core radius rc
–
Central density
–
Vmax
–
rmax
–
Einstein radius now
–
Einstein radius initially
–
Dls/Ds
–
Density profile: initial NFW (dashed) and the halo now (solid)
Evolution: central density and core radius against time
Mean convergence inside projected radius R, with the threshold 1 − κc

What this shows

Scattering lets dark matter particles exchange energy, so heat flows down the halo's temperature (velocity dispersion) gradient. An NFW halo is colder at its centre than near rs, so heat first flows inward: the central particles are heated, their orbits spread out, and the cusp becomes a low-density core. Once the core is warmer than its surroundings, heat flows outward instead. A self-gravitating core has negative heat capacity: losing energy makes it contract and heat up, which drives more heat out. This runaway, the gravothermal catastrophe, makes the central density climb steeply as t approaches the collapse time tc.

A lens needs a high projected density near its centre. For haloes like the default, whose Einstein radius is comparable to the core, the cored halo is a weaker lens than the NFW halo it started from; the collapsing halo packs more mass into a small radius and becomes a stronger one. Inside a cluster that already supplies a convergence κc, a subhalo produces its own critical curve where its mean convergence reaches 1 − κc. With enough cluster convergence (roughly κc ≳ 0.4 for the default halo) the collapsing halo's Einstein radius grows well beyond the initial one: at κc = 0.6 and zs = 2 it goes from 0.03″ to 0.14″ by τ = 1. With little external convergence, or none for an isolated halo, even the τ = 1 profile of this halo stays subcritical and has no Einstein radius at all. A massive, concentrated halo has an Einstein radius well outside its core, so the core barely weakens it: for M200 = 10¹² M☉ and c200 = 60 its Einstein radius never falls below the initial one. Dutra, Natarajan & Gilman (2025) discuss steeper, core-collapsed subhaloes as a way to raise the rate of galaxy-galaxy strong lensing in clusters.

Model and assumptions

  • The halo follows the parametric SIDM model of Yang, Nadler, Yu & Zhong (2024): a cored NFW form whose scale density, scale radius and core radius are fitted functions of τ = t / tc. The fits come from an isolated N-body SIDM halo followed from τ = 0 to its collapse at τ = 1. The paper reports that the cored NFW form fits the simulated profiles to within 10% even at τ = 1, and that the fitted central density follows the simulation up to τ ≈ 0.87 and is about 15% low at τ = 1. The paper truncates the evolution at tc to avoid extrapolation, and so does this page: past τ = 1 the halo is shown at its τ = 1 profile. The fits do not conserve mass exactly: the mass inside r200 is about 4% low at τ = 0.2 and about 9% low at τ = 1.
  • The collapse time is tc = (150 / C) / [(σ/m) ρs,0 rs,0] · (4π G ρs,0)−1/2 with C = 0.75, from the initial NFW parameters. The cross-section is constant and independent of velocity. The cold (CDM) model has no self-interaction: σ/m is 0, the collapse time is infinite and the NFW halo does not evolve.
  • The halo is isolated and starts as an NFW halo at t = 0, with M200 and c200 defined against the critical density at z = 0.348. There is no tidal stripping, no further accretion and no baryons, all of which matter for real subhaloes. Projections run along the line of sight out to 2 r200.
  • Distances are for flat ΛCDM with H0 = 70 km/s/Mpc and Ωm = 0.3. The cluster is a uniform sheet of convergence κc with no shear; an isolated lens has κc = 0. Einstein radii are only searched above 1 pc. For the cuspy NFW halo κ̄(<R) keeps rising towards the centre, so a very small Einstein radius depends on how far in the grid goes; anything below 0.01″ is reported as unresolved.
  • This is an illustration of the model, not a fit to any observed system.

References

  • Spergel & Steinhardt 2000, arXiv:astro-ph/9909386
  • Balberg, Shapiro & Inagaki 2002, arXiv:astro-ph/0110561
  • Pollack, Spergel & Steinhardt 2015, arXiv:1501.00017
  • Essig, McDermott, Yu & Zhong 2019, arXiv:1809.01144
  • Yang, Nadler, Yu & Zhong 2024, arXiv:2305.16176
  • Navarro, Frenk & White 1997, arXiv:astro-ph/9611107
  • Bartelmann 1996, arXiv:astro-ph/9602053
  • Dutra, Natarajan & Gilman 2025, arXiv:2406.17024