Influence of tides and self-gravity on Ultra-Light Dark Matter Bounds from Dwarf Galaxies

Abstract

Dwarf spheroidal galaxies provide some of the most sensitive astrophysical probes of ultra-light dark matter (ULDM), but the inferred constraints can be affected by two important systematics: tidal interactions with the Milky Way, which reduce ULDM-induced dynamical heating, and stellar self-gravity, which can become relevant if the stellar component was more compact at earlier times. In this work, we attempt to estimate both effects by reconstructing dwarf-galaxy orbital histories in a Milky-Way potential, adopting a simple and approximate tidal-susceptibility diagnostic that we argue provides a conservative description of tidal stripping, and explicitly including stellar self-gravity in our numerical simulations. Within our framework, which we apply to five different dwarf galaxies, we find that ULDM with masses $5\times 10^{-22} \lesssim m/{\rm eV} \lesssim 5\times 10^{-21}$ remains in tension with current data.

Figures (12)

• Figure 1: Two examples of halos obtained with different eigenfunction cutoffs $(n_{\rm max},l_{\rm max})$ for the same target soliton+NFW density profile, as described in Sec. \ref{['s:sims']}. Left: case with $m=1e-21eV$, where the density profiles are truncated at $r_{\rm t}\simeq (2,4,6)\,\mathrm{kpc}$ for different cutoff choices. Right: case with $m=2e-21eV$, where the density profiles are truncated at $r_{\rm t}\simeq (1,2,2.5)\,\mathrm{kpc}$.
• Figure 2: Examples of a simulation run, targeting Carina/Leo II-like systems, for $m=5e-21eV$ with star self-gravity, showing the $r_{\rm half}$ evolution over time, compared with the observed one. We show the results of four runs, each one of them with a different mass of the stars. The benchmark is what it is expected for Leo II, $M_* \sim 10^6 M_\odot$, while for Carina everything should be in fact reduced by a factor of $\sim 2$.
• Figure 3: Fornax simulation, with tides, for $m=e-21eV$. Dynamical heating is slowed down with respect to the examples of Ref. Teodori:2025rul, but the inner peak on the dispersion curve is retained.
• Figure 4: Fornax simulations, for $m=5e-22eV$. We compare three runs, one with tides, one without and one both without tides and with zero $M_*$. Although dynamical heating is basically stopped for the tide case, the inner peak on the velocity dispersion curve due to the soliton would put this mass at odds with data nevertheless. Notice also how the non-zero mass of the stars affects the growth of $r_{\rm half}$.
• Figure 5: Leo II/Carina simulation, with tides ($r_{\rm t} = 0.5$ kpc, $r_{\rm t} = 1$ kpc, and $r_{\rm t} = 2$ kpc), for $m=5e-21eV$. Left: $r_{\rm half}$ as a function of time. Right: Normalized eigenmode energies to the ground state $E_0$ for the density profile chosen, sorted from low to high value; horizontal lines show the energy cutoff applied for the different values of $r_{\rm t}$. The half-light radius behavior is compatible with what shown in Fig. \ref{['fig:leoII']} for $r_{\rm t} = 2$ kpc, confirming the relative insensitivity of Carina to tides. For contrast, the $r_{\rm t} = 1$ kpc and $r_{\rm t} = 0.5$ kpc cases have instead a slowed down growth of $r_{\rm half}$.