A million orbits, the same score. Nine dark matter halos, nine orchestras without a conductor. All musicians play the same notes, but in each hall the music sounds different: sometimes harmonious, sometimes out of sync. Likewise, satellite galaxies falling into a host halo move along orbits that depend not only on the mass but also on the shape of the invisible stage.
Usually, satellite orbits are calculated in a spherical halo, as if the musicians played to a metronome. But a real dark matter halo is not a sphere; it is a triaxial ellipsoid, with three different axes, like a potato. The authors built nine such "potatoes" with the same mass and the same density profile but different elongations. Through each, they ran the same million satellite orbits.
It turned out that the average orbital parameters barely change. But each individual orbit behaves differently: the spread in successive close approaches to the center ranges from 5% to 23% depending on the halo shape. The minimum distance to the center over the age of the Universe decreases by up to 21% in the most flattened halos. And if you try to reconstruct the orbit assuming a spherical halo, the positional error reaches a tenth of the halo size within 2–4 billion years.
For the dwarf galaxy Triangulum II, this means that the uncertainty in the closest point of the orbit due to the unknown halo shape is five times larger than the stated observational error. Now the score has been read anew: the average notes have not changed, but each musician hears them differently. Ignoring the halo shape means underestimating how deeply satellites dive toward the center and making mistakes in the rates of their disruption.
In the most flattened halos, the distribution of minimum pericenters forms a peak near 0.1 virial radius, although no such peak exists in the spherical case. This peak arises from chaotic orbits that dive deeper toward the center.