Islands in Simulated Cosmos: Probing the Hubble Flow around Groups and Clusters

TL;DR

The paper investigates whether dark energy leaves detectable imprints on the local Hubble flow around groups and clusters by leveraging the IllustrisTNG simulations and extended Lemaître–Tolman kinematic models. It uses Bayesian inference to fit a velocity-radius relation to simulated data, recovering halo masses $M$ and the Hubble constant $H_0$, and quantitatively assesses biases and model-dependence. The main findings show median mass recovery of $M_{\rm fit}/M_{\rm true} \approx 0.96$ with large scatter, and an average recovered $H_0$ of $63.5$ km s$^{-1}$ Mpc$^{-1}$ versus a true value of $67.74$, indicating roughly a 25% uncertainty in $H_0$ from the local flow method; different model variants (including angular momentum and friction) are statistically indistinguishable, highlighting limitations in using local kinematics as a precision probe of dark energy. The work outlines a practical path to improve robustness—stacking halos, stringent quality cuts, and better environmental screening—while noting that current data do not provide decisive evidence for dark-energy effects on sub-Mpc scales, though they yield meaningful constraints on $M$ and $H_0$ at the population level.

Abstract

The local Hubble flow offers a powerful laboratory to study the interplay between cosmic expansion and gravitational dynamics. On large scales, galaxy velocities follow Hubble's law, but within groups and clusters local gravitational effects introduce significant departures from linearity. Using the IllustrisTNG cosmological simulations, we investigate whether dark energy leaves detectable imprints on the local velocity-radius relation. We model the kinematics with extensions of the Lemaitre-Tolman framework and apply Bayesian inference to recover halo masses and the Hubble constant H0. The fits reveal systematic biases: halo masses are underestimated with a median ratio $M_{fit}/M_{true} = 0.95 \pm 0.28$, while the inferred Hubble constant clusters around $H_0 = 64 \pm 16 km/s/Mpc$, compared to the simulation input of 67.74. This corresponds to an average 25\% uncertainty in H0 recovery from the local flow method. While the mass and expansion rate can be constrained, different model variants whether including angular momentum, friction, or altered radial scaling-remain statistically indistinguishable. Our results highlight both the promise and the limitations of using local kinematics as a precision probe of dark energy.

Figures (4)

• Figure 1: The halo mass function fo the isolated halos from the TNG simulation.
• Figure 2: Comparison between masses recovered from our velocity--radius fits and the true halo masses in the simulation. Points show individual halos; the diagonal marks perfect recovery. The scatter and systematic offsets reflect model dependence, projection effects and dynamical complexity. Different model families (see text) produce distinct biases in the recovered masses.
• Figure 3: Fitting efficiency (as quantified by the goodness-of-fit metric) as a function of the assumed Hubble parameter $H_0$, evaluated with the slope fixed to $n=0.5$. The resulting curve illustrates how the quality of the velocity–radius fits depends on the adopted cosmic expansion rate: a broad maximum identifies the range of $H_0$ values that best reproduce simulated kinematics under this restricted model.
• Figure 5: Scatter of best-fit $H_0$ values obtained for individual systems plotted against the corresponding fit efficiency. The distribution emphasizes the degeneracy between inferred expansion rate and local dynamical state: systems with higher fit efficiency cluster around the simulation’s preferred $H_0$, whereas lower-efficiency fits produce a larger scatter and systematic offsets, underlining the need to combine many systems or apply robust selection criteria for reliable local determinations of $H_0$.