Thread by @8minutesold

Some thoughts on the recent paper looking at the evolution of planes of satellites in the EAGLE simulation. Specifically, they look at the 11 classical satellites around the Milky Way, i.e. the brightest members of the Vast Polar Structure (VPOS). arxiv.org/abs/1904.02719…

I like that they nicely confirm what I've been finding with many other simulations (they just don't speak about that much). The 11 classical satellites (yellow) have a minor-to-major axis ratio of c/a = 0.18, and highly clustered orbital poles (directions of angular momentum).

In EAGLE, the authors find that only 1% of their simulated systems are as flattened as the observed VPOS, and <1% show as well aligned orbital poles.

I've zoomed in and highlighted those parts of the PDF here.

They are still hard to see.

They just don't put the two criteria together,. Since they don't mention any such system, I assume that none of their 1080 systems reproduce both observed parameters simultaneously. That's the satellite planes problem, finding the observed situation is incredibly rare in sims.

Consequently, to be able to study *something* they substantially weaken what it means for a satellite system to be MW-like. An axis ratio of c/a ≤ 0.3 really isn't like the VPOS, and accepting the orbits to be aligned to within 35º vs. 22º gives a much less correlated situation.

Even then, only 40 of their 1080 systems fulfill both of these criteria simultaneously, though they were chosen to be extremely generous.

So, this hydrodynamical cosmological simulation is unable to reproduce the VPOS (that's what I expect and have said before, most recently here: arxiv.org/abs/1903.10513…).

I'd find this a very concerning result for LCDM, but – surprise – they don't at all emphasize this.

Instead, they look at the evolution of their selected systems at earlier snapshots. Ok, you can do that, simulations are great in this respect. I'm just not sure how it tells us anything about the observed system.

They find that their flattened satellite systems were much wider in the very recent past, while systems with aligned orbits tend to have been more stable.

That's of course expected from (a) the fact that most flattened distributions in simulations are simply chance alignments, and (b) basic dynamics, i.e. if you select stuff to orbit in the same plane it'll tend to stay in a common plane for a while.

Some concerns: they say they use luminous satellites, but their initial gas particle mass is 1.8 x 10^6 Msun while the classical satellites have stellar masses as low as 10^5 Msun. So they probably consider satellites with a single star particle, which will be heavily stochastic.

They consider as Milky Way-like any galaxy with a halo mass of 0.3 to 3 x 10^12 Msun. That's a lot wider than the present range in measurements for the actual MW.

They also don't consider the orbital *direction* of satellites. In the observed MW, at least 7 of 11 co-orbit in the same direction, while Sculptor counter-orbits along the same plane.

By allowing both directions equally, they say that something like 4 satellites orbiting in each direction is the same as 7 in one and 1 in the opposite. This choice makes twice the area on the unit sphere available to fit orbital poles in, so it increases the number of "matches".

Finally, they frequently cite Lipnicky & Chakrabarti (2017) as having shown that the VPOS will widen quickly. As I've pointed out before (onlinelibrary.wiley.com/doi/full/10.10…), that is based on a very flawed approach …

… because that study simply integrated the satellite orbits using the then most-likely proper motions. However, PM uncertainties will always introduce additional dispersion, which results in a widening of any stable plane.

However, besides these concerns, on first look I don't think there's anything wrong with this new study. One just can't claim that they find analogs to the VPOS, explain the planes of satellites problem, or can draw conclusions on the past of the MW system from these findings.