Planes of satellites, at once transient and persistent

TL;DR

This work resolves a longstanding tension in LCDM predictions for satellite planes by demonstrating that plane lifetimes are frame-dependent: transient within a host halo but potentially persistent when tracking satellite progenitors. Using MW analogues from the TNG-50 simulation and the inertia-tensor metric $c/a$, the authors compare host-centric and satellite-centric evolutions over lookback times up to several Gyr, finding rapid in-halo dissolution but enduring coherence among progenitors outside the halo. The key contribution is showing that both transient and persistent features arise naturally, depending on the reference frame, thereby reconciling conflicting theoretical predictions. The results underscore the importance of progenitor tracing and frame choice for interpreting satellite-plane observations and point to the need for higher-resolution simulations to fully capture the dynamics.

Abstract

The appearance of highly anisotropic planes of satellites around the Milky Way and other galaxies was long considered a challenge to the standard cosmological model. Recent simulations have shown such planes to be common, but they have been described as either "transient", short-lived alignments, or "persistent", long-lived structures. Here we analyse Milky Way analogue systems in the cosmological simulation TNG-50 to resolve this apparent contradiction. We show that, as the satellite populations of individual hosts rapidly change, the observed anisotropies of their satellite systems are invariably short-lived, with lifetimes of no more than a few hundred million years. However, when the progenitors of the same satellites are traced backwards, we find examples where those identified to form a plane at the present day have retained spatial coherence over several Gyr. The two ostensibly conflicting predictions for the lifetimes of satellite planes can be reconciled as two perspectives on the same phenomenon.

Figures (5)

• Figure 1: Position of the 11 brightest satellites, projected edge-on (i.e. in the plane of the minor and major axes of the inertia tensor), in four example systems, at the present day (left), and six lookback times up to 3 Gyr, in the host-centric frame (top, red) and the satellite-centric frame (bottom, blue). In the host-centric frame, filled circles denote satellites that are part of the $t=0$ sample, in the satellite-centric frame, filled circles denote subhalos inside $r_\mathrm{lim}(t)$. On each panel, $r_\mathrm{lim}$ is plotted for scale, and the value of $c/a$ is given. The ID above the $t=0$ panel identifies the host subhalo in the simulation at $t=0$, and the line style below the $t=0$ panel identifies the system in Figure \ref{['fig:ca-evolution']}. In these four systems, the spatial anisotropy varies rapidly in the host-centric frame, but remains stable in the satellite-centric frame.
• Figure 2: Top: evolution of the anisotropy, $c/a$, in the host-centric frame (left) or satellite-centric frame (right). Middle: rate of change of $c/a$ between individual snapshots. Bottom: maximum value of $c/a$ up to lookback time $t$. Individual lines show the 25 systems in the lowest quartile of $c/a$ at $t=0$, with the four systems described in Figure \ref{['fig:positions']} highlighted. Dark line segments indicate the interval in which the system contains the same satellites as at $t=0$ ( host-centric frame), or when all satellites are within $r_{\mathrm{lim}}$ ( satellite-centric frame). Grey bands on the top panels show $\pm 1 \sigma$-equivalent percentiles for all 101 systems, with histograms showing the corresponding distributions at $t=0$ and $t=3$ Gyr. Black lines in the middle and bottom panels show the moving median and the median at $t=0$, respectively. Only in the satellite-centric frame do several systems maintain a persistently high anisotropy.
• Figure 3: Evolution of the distance of the $i_\mathrm{th}$ innermost satellite to the centre, normalised by $r_\mathrm{lim}(z)$ , in the host-centric frame (left), or satellite-centric frame (right). Thick lines show the median ratio, thin lines of corresponding colours show individual systems. In the host-centric frame, by definition, all satellites are always inside $r_\mathrm{lim}(t)$, and normalised by $r_\mathrm{lim}(t)$, the average radial profile shows little evolution. In the satellite-centric frame, the system "expands" with increasing lookback time. Of the 11 satellites at $t=0$, two typically fell in less than one Gyr ago, and four less than two Gyr ago.
• Figure 4: Top: Fraction of systems for which at least $N$ of 11 satellites of the $t=0$ system are part of the system in the host-centric frame. Bottom: Fraction of systems for which at least $N$ of 11 satellites are found within $r_\mathrm{lim}$ in the satellite-centric frame. Grey lines show the corresponding lines of the other panel, vertical lines indicate the timing of snapshots.
• Figure 5: Lifetime of satellite planes in the host-centric frame (top) and the satellite-centric frame (bottom). In the host-centric frame, we distinguish between planes that contain the same satellites, and any satellite plane in the same halo (including those whose membership has changed). Vertical lines indicate the timing of the snapshots.