Weighing gas-rich starless halos: dark matter parameters inference from their gas distributions

Abstract

Reionization-Limited $H_{I}$ Clouds (RELHICs) are starless dark matter halos retaining a significant neutral hydrogen($H_{I}$ ) reservoir. The gas resides in near hydrostatic equilibrium within the dark matter potential and in thermal equilibrium with the cosmic ultraviolet background. This simplicity allows analytic frameworks to link observable $H_{I}$ column densities directly to fundamental dark matter halo structural parameters. We systematically assess the accuracy of inferring host halo parameters from RELHIC gas distributions on an object-by-object basis, quantifying biases, intrinsic degeneracies, and the limits of parameter recovery. Using RELHICs from a redshift z = 0 high-resolution cosmological hydrodynamical simulation, we employ Bayesian nested sampling to infer dark matter halo mass and concentration. We evaluate this against 3D spherically averaged total gas and $H_{I}$ density profiles, alongside 2D $H_{I}$ column density profiles. We found that while the ensemble inference yields a robust, unbiased recovery of halo virial mass from 3D profiles, individual systems exhibit a mass-concentration degeneracy driven by local environmental density. Overdense environments yield slightly overestimated masses and underestimated concentrations; underdense regions show the inverse. We demonstrate that treating environmental density as a free parameter breaks this degeneracy and completely neutralizes the systematic mass bias. Although concentration recovery remains limited by simulation resolution, the virial mass is exceptionally well constrained, establishing a highly reliable framework for weighing starless halos in upcoming surveys.

Figures (17)

• Figure 1: Top left: gas mass vs. halo mass for the RELHIC sample. Points are coloured by H i mass. The black solid line shows the expected gas-halo mass relation from the Benitez-Llambay2017 model using the RELHICs’ effective equation of state and the median RELHIC mass-concentration relation; the shaded region indicates the expected scatter in gas mass from the scatter in RELHICs' concentrations. Stars mark H i-rich RELHICs; the large orange diamond symbol indicates an individual RELHIC example, displayed by the thick orange line in Figure \ref{['fig:AllGasDensiyProfiles']}. Top right: H i mass vs. halo mass. Points are colored by gas mass; the dotted line shows the threshold H i mass adopted to divide the systems between H i-rich ($M_{\textup{H\,i}} \gtrsim 10^6\,M_{\odot}$) and H i-poor ($M_{\textup{H\,i}} \lesssim 10^6\,M_{\odot}$). The black solid line and shaded region indicates, as before, the expected H i-halo mass relation from the Benitez-Llambay2017 model. Bottom left: H i mass vs. gas mass, colored by halo virial mass; the black solid line and shaded region mark the Benitez-Llambay2017 predictions together with the scatter arising from the scatter in concentration. Bottom right: halo mass-concentration relation for the selected RELHIC sample (blue) compared to the relation for galaxies (green). Individual RELHICs are shown by the small triangles. The black line indicates the expected trend from the Ludlow2016 relation, with the shaded regions representing the 16th–84th percentile intervals. The blue solid line and shaded region indicate the median and the scatter of the mass-concentration relation for our RELHIC sample.
• Figure 2: The temperature–density relation for all gas particles belonging to our RELHIC sample. The heatmap illustrates the particle number density on a logarithmic scale, computed using a $100 \times 100$ grid. The bins are equally spaced in log-space, covering the ranges $-6 \le \log_{10}(n_{\text{H}} / \text{cm}^{-3}) \le 0$ and $4 \le \log_{10}(T/ \text{K}) \lesssim 4.8$. The colormap represents the number of particles per bin, ranging from 10 (blue) to a peak density of $\approx 7 \times 10^{3}$ (yellow). The red solid line shows the running median (our effective EoS), while the shaded red region spans the 16th–84th percentiles. For comparison, the black solid line shows the analytic EoS reported by Benitez-Llambay2017.
• Figure 3: Gas density profiles for our RELHIC sample, colored by halo virial mass. The thick orange line shows the profile of the RELHIC example highlighted in Figure \ref{['Fig:SelectedHalos']} (orange diamond), and discussed in Section \ref{['Subsec:WorkedExample']}. The horizontal red dashed line indicates the expected density at 100 kpc from the Benitez-Llambay2017 model for a halo similar to the example.
• Figure 4: Posterior distributions of mass and concentration of a RELHIC example, obtained by applying nested sampling to its gas density profile. The red lines and shaded regions mark the ground-truth values from the simulation. The top-right panel compares the input gas density profile (black symbols) with the best-fitting RELHIC model (solid line). The recovered nuisance parameter $s$ is $\approx 0.1$.
• Figure 5: Distribution of the recovered halo parameters relative to the ground truth. For the virial mass, we define the logarithmic deviation as $\Delta \log_{10}(M_{200}) = \log_{10}(M_{200}/M_{200,\rm T})$, while for the concentration we show the fractional error, $\Delta c/c_{\rm T} = c/c_{\rm T} - 1$. The top and right panels show the marginalized one-dimensional histograms, whereas the central panel displays the joint two-dimensional distribution. The blue points and black histograms represent the full RELHIC sample, while the orange points and histograms highlight the subset of "well-resolved" systems, i.e., those RELHICs for which the inner density profile can be probed below 1 kpc. Red lines indicate the unbiased reference values.