Is the Milky Way ringing? The hunt for high velocity streams

We perform numerical simulations of a stellar galactic disk with initial conditions chosen to represent an unrelaxed population which might have been left following a merger. Stars are unevenly distributed in radial action angle, though the disk is axisymmetric. The velocity distribution in the simulated Solar neighborhood exhibits waves traveling in the direction of positive v, where u,v are the radial and tangential velocity components. As the system relaxes and structure wraps in phase space, the features seen in the uv-plane move closer together. We show that these results can be obtained also by a semi-analytical method. We propose that this model could provide an explanation for the high velocity streams seen in the Solar neighborhood at approximate v in km/s, of -60 (HR 1614), -80 (Arifyanto and Fuchs 2006), -100 (Arcturus), and -160 (Klement et al. 2008). In addition, we predict four new features at v ~ -140, -120, 40 and 60 km/s. By matching the number and positions of the observed streams, we estimate that the Milky Way disk was strongly perturbed ~1.9 Gyr ago. This event could have been associated with Galactic bar formation.

💡 Research Summary

The paper investigates the origin of the high‑velocity stellar streams observed in the Solar neighbourhood by modelling the dynamical response of a Galactic disc that has been left in a non‑equilibrium state after a merger event. The authors construct two complementary approaches: (1) a full N‑body simulation of an axisymmetric stellar disc whose stars are initially distributed unevenly in the radial action‑angle (J, θ) space, and (2) a semi‑analytical treatment based on action‑angle variables and Hamiltonian dynamics that reproduces the same phase‑space behaviour. In the simulations the disc is initially axisymmetric, but the uneven distribution of the radial action angle creates a “phase‑space wave” that propagates in the direction of increasing tangential velocity v in the local (u, v) velocity plane (where u is the radial component). As the system evolves, the wave wraps around phase space, causing successive over‑densities (streams) to appear at progressively smaller separations in v. The spacing between the wave crests shrinks roughly as 1/t, reflecting the continual phase‑mixing of the initially coherent angle distribution.

The semi‑analytical model treats the initial angle perturbation as a sinusoidal modulation Δθ cos(k θ) and solves the linearised Hamiltonian equations to obtain the wave’s phase speed and damping rate. By expressing the radial action J and the mean angular frequency Ω, the model predicts a wavelength λ≈2πJ/Δθ and a period T≈2π/Ω, which match the numerical results to within a few per cent. This analytical framework allows rapid exploration of parameter space (e.g., different perturbation amplitudes, disc scale lengths, or pattern speeds) and confirms that the observed velocity structures are a natural consequence of phase‑mixing rather than requiring resonant trapping by a bar or spiral arms.

The authors then compare the simulated (u, v) distribution with the well‑documented high‑velocity streams: HR 1614 (v ≈ ‑60 km s⁻¹), the Arifyanto & Fuchs feature (v ≈ ‑80 km s⁻¹), the Arcturus stream (v ≈ ‑100 km s⁻¹), and the Klement et al. feature (v ≈ ‑160 km s⁻¹). In the simulation these four over‑densities line up with successive wave crests, providing a unified explanation that they are all phases of the same wrapping process. Moreover, the model predicts four additional streams at v ≈ ‑140, ‑120, +40 and +60 km s⁻¹ that have not yet been firmly identified; the authors suggest that forthcoming Gaia data releases should be able to test these predictions.

By measuring the current separation between the simulated streams and matching them to the observed values, the authors infer the time elapsed since the perturbation. The best fit occurs for a perturbation that took place roughly 1.9 Gyr ago. This timescale coincides with independent estimates for the formation of the Galactic bar, leading the authors to speculate that the bar’s rapid growth could have been the agent that excited the disc, leaving the long‑lived phase‑space ripples we now observe.

In summary, the paper demonstrates that a non‑relaxed stellar disc, perturbed by a merger or bar‑formation event, naturally develops a series of velocity over‑densities through phase‑mixing and wrapping in action‑angle space. The combined numerical‑analytical approach not only reproduces the known high‑velocity streams but also makes testable predictions for new streams. If confirmed, these findings would provide a powerful chronometer for past dynamical events in the Milky Way and a framework for interpreting future high‑precision astrometric surveys.