Fix the version and the physical comparison
The fixed source is Tree-ring structure of Galactic bar resonance in N-body simulations, by Rimpei Chiba, Michiko Fujii, Junichi Baba, John Dubinski and Ralph Schönrich. Its arXiv stamp is 1 April 2026; the 21-page first-version PDF says draft version April 3. This guide follows that preprint, including its appendices, without assuming it is a final journal version.
Section 2 defines four models. A has a live dark halo and no Sagittarius satellite; B replaces the halo with a static potential. C adds a Sagittarius-like satellite late to A; D deliberately gives the satellite an unrealistically large mass. The A/B comparison tests the role of the halo response. C/D test survival under strong external perturbations.
A live halo can exchange angular momentum with the bar and slow it through dynamical friction. A static halo cannot respond in that way, although the outer stellar disk still can. This is a self-gravitating collisionless simulation, with initial distribution functions; the companion prescribed-field release answers a smaller question.
Figure 1 is an initial model, not a new rotation-curve measurement
On printed page 3, Figure 1 gives circular speed against radius. The solid black total combines a halo, disk and bulge; the coloured component curves explain the model's gravitational support. The velocity unit is kiloparsecs per gigayear. These curves describe the adopted initial galaxy.
The disk and bulge are sampled from distribution functions and the halo is given a specified spin. Gas and ongoing star formation are absent. Remember those choices when later figures show spiral activity fading: an initially mature disk heats itself, and there is no continuing supply of newly born cold stars.
Figure 2 changes coordinates after its top row
The top row is an ordinary projected stellar surface-density map. The next rows are not spatial maps. Their horizontal coordinate is an azimuthal angle variable relative to the bar; their vertical coordinate is azimuthal action, equivalent here to specific angular momentum. For eccentric orbits, the angle variable is not simply the geometric azimuth of the star's current position.
A Hamiltonian is the energy function written in coordinates and their canonical momenta. The rotating form here includes the same pattern-speed times angular-momentum subtraction introduced in the torque chapter.
Equations 5–7 explain the transformation and averaging. The authors use AGAMA and average over the fast radial and vertical angles. The second row shows an averaged rotating Hamiltonian at fixed fast actions. The third and fourth rows additionally marginalize over a selected domain of radial and vertical actions. The maps therefore have a selection as well as a coordinate transformation.
The black horizontal line marks the circular corotation reference. In live-halo model A, the resonant region moves toward larger angular momentum as the bar slows. The fourth row colours mean initial angular momentum. Low initial values remain near the resonant core while more recently captured high values lie nearer the outside.
Equation 8 normalises the Hamiltonian so that the resonant centre has h=0 and the separatrix has h=1. The centre is a maximum of this averaged Hamiltonian. For an illustrative set Hˉres=100, Hˉsep=80, Hˉ=90 in common units, the normalised value is 0.5. These are explanatory numbers, not fitted simulation values.
Figures 3–5 make the ordering test persuasive
Figure 3 repeats the maps for static-halo B. After about two gigayears the bar's resonance moves little, and its initial-angular-momentum contrast is much weaker. Compare this control with A before interpreting a slope in isolation.
Figure 4 plots mean initial angular momentum against libration action Jℓ. Libration means the slow resonant angle swings around a centre rather than circulating. Its action is the enclosed area in the angle-action plane divided by 2π. Appendix B describes contour extraction, area calculation and splitting the two resonant islands. This is not the area of a star's path in the disk.
The colours in Figure 4 label simulation time; open circles locate the separatrix at each time. Earlier captured stars tend to have smaller libration action, while later captured stars enter nearer the growing outside. Slow evolution approximately preserves that action, so some capture ordering survives mixing.
Figure 5 follows a mass-weighted regression slope. A develops a substantially stronger trend than B. B nevertheless has a weak positive trend too. Appendix C and Figure 12 show how an asymmetric Hamiltonian can preferentially sample higher angular momentum near the outside even with steady rotation. A positive slope alone is therefore an incomplete slowdown test.
Figures 6–10 stress the resonance with a satellite
Figure 6 shows the satellite's trajectory and mass evolution. The models intentionally overestimate its present-day mass. Figures 7 and 8 show strong induced two-arm structure and large fluctuations of the resonant maps; Figure 9 shows a weakened, noisy capture-ordering trend that remains positive in these runs.
The authors connect that survival to the tidal disturbance being relatively weak near corotation, coherent across the resonant region and unable to change the orbit-averaged pattern speed greatly. Figure 10 in Appendix A provides that last comparison, alongside bar length and amplitude. These are tests of the adopted satellite orbits and galaxy models, not a proof for every possible encounter.
The same appendix explains why slowdown need not increase corotation radius divided by bar length in every galaxy. In a growing disk the bar can elongate along with its outward-moving corotation. The introduction also notes simulations of accelerating bars in baryon-dominated environments. The exchange of angular momentum depends on the system that can absorb or supply it.
Figures 13–14 calibrate an inference inside the simulation
Appendix D deliberately uses wrong pattern speeds when constructing the resonant coordinates. Figure 13 shows the ordered relation losing coherence away from the true value. Equations D21–D22 turn positive adjacent-bin gradients into an inference score; this requires assumptions about gradient uncertainties and their combination.
Figure 14 then compares recovered and known pattern speeds over the simulation. The insets give fractional residual means near minus one percent and scatter of roughly five to seven percent after the stated early-time cut. This is a simulation calibration with known truth, not a new measured confidence interval for the Milky Way.
For scale, seven percent of a historical 35.5kms−1kpc−1 estimate is about 2.5kms−1kpc−1, larger than its quoted 0.8 statistical error. The paper discusses an additional modelling uncertainty of that order. One should not replace a galaxy's full error model by the scatter of one simulation.
An observational application also substitutes chemistry for the unobservable initial angular-momentum tag. Section 4 says that metallicity-to-birth-radius mapping depends on age, particularly for stars born during early rapid enrichment. Coeval populations and chemical evolution matter. This preprint tests the dynamical interpretation; it does not directly observe each star's birth radius.
What the tree rings teach about evolving patterns
The common mechanism is time-dependent exchange near corotation. A temporary spiral can leave stars on new guiding radii after it fades. A slowing bar can transport a trapped group as the resonance sweeps outward. Diffusion from other structures blurs the ordering, and a selected phase-space population may retain part of it.
Section 4 identifies the next physical ingredients: gas structure, molecular-cloud scattering, ongoing star formation and chemical enrichment. The simulation's old disk may underproduce late spiral activity. Its surviving tree-ring structure is therefore a valuable result within these models and a reason to test richer histories.
When reading another claim of resonant migration, ask which frame defines the slow angle, what quantity tags origin, which selection defines the sample, what time history changes the resonance and what control could create a similar trend. Those questions connect a striking phase-space figure to a dynamical explanation.
Reading: Chiba, Fujii, Baba, Dubinski & Schönrich (2026), Tree-ring structure of Galactic bar resonance in N-body simulations, arXiv:2604.01294v1 · Sellwood & Binney (2002), Radial mixing in galactic discs, astro-ph/0203510v2