Resonant Relaxation Near a Galactic-Center Black Hole

What Resonant Relaxation Is

Very close to a massive black hole, gravity is dominated by the point mass, so stars move on nearly closed Keplerian ellipses that return to almost the same orientation every orbit. Over many periods each fixed ellipse behaves like a smeared-out elliptical wire of mass. Because these wires stay put for a long time, the gravitational torques one star exerts on another do not average to zero — they persist in a fixed direction over a coherence time.

Resonant relaxation is the enhanced diffusion of orbital angular momentum L that results from these persistent, correlated torques. Key features:

• It changes angular momentum (eccentricity and orientation) but essentially not energy (semi-major axis), because the torques do no net work on the closed orbit.
• It is collective: the relevant torque comes from the whole surrounding star cluster, not one close encounter.
• It operates only where orbits are nearly closed — a near-Keplerian potential — i.e. inside the black hole's radius of influence (~2–3 pc for Sgr A*).

Rauch and Tremaine named and quantified the effect in 1996, showing it can dominate over ordinary two-body relaxation in this inner region.

The Mechanism: Coherent Torques and a Random Walk

Consider a test star among N field stars sharing its orbital neighborhood. If the field ellipses held perfectly still, the residual (non-cancelling) torque scales like the Poisson fluctuation of the mass distribution: τ ≈ √N · (G m / a), where m is the field-star mass and a the semi-major axis. During the coherence time t_coh, angular momentum builds up coherently, growing linearly in time: ΔL ≈ τ · t.

But the orbits do not hold still forever. In-plane precession — from the distributed stellar mass and from general relativity (the same 1st-post-Newtonian precession that shifts S2's pericenter) — slowly rotates each ellipse, scrambling the torque direction after t_coh. Beyond that, the accumulated ΔL(t_coh) becomes the effective step of an incoherent random walk, so ⟨ΔL²⟩ grows only linearly in time again, ∝ (t/t_coh).

• Coherent phase (t < t_coh): ΔL ∝ t.
• Random-walk phase (t > t_coh): ΔL ∝ √t, with a much larger effective diffusion coefficient than two-body scattering.

The net result: L randomizes on a timescale T_RR that can be 10–1000× shorter than the classical relaxation time at the same radius.

Scalar and Vector Modes, and Characteristic Numbers

RR splits into two modes distinguished by what randomizes the torque:

• Scalar RR changes the magnitude of L (the eccentricity). Its coherence time is set by in-plane precession — mass precession plus relativistic (GR) precession — typically a few thousand orbits. Characteristic timescales at Sgr A* are ~10⁷–10¹⁰ yr.
• Vector RR changes only the direction of L (the orbital plane), leaving eccentricity fixed. In-plane precession does not affect the plane, so its coherence time is longer, set by the slower reorientation of the background itself. It is the fastest process, ~10⁵–10⁷ yr.

Worked numbers for S2 around Sgr A* (M ≈ 4×10⁶ M_sun, a ≈ 1000 AU, e ≈ 0.88):

• Orbital period: ~16 yr.
• Pericenter (in-plane) precession period: ~30,000 yr — dominated near pericenter by GR (Schwarzschild precession, measured by GRAVITY in 2020).
• Orbital-plane reorientation by vector RR: ~10⁶ yr.

The ordering orbit ≪ scalar-RR coherence ≪ vector-RR timescale is what makes the plane the most quickly randomized orbital element.

How It's Observed and Where It Appears

Resonant relaxation is not seen directly in a single orbit — the timescales dwarf a human lifetime — but its statistical fingerprints appear in the distribution of orbits around Sgr A*:

• The S-star cluster: the young B-type S-stars within ~0.04 pc have a nearly thermal, isotropic distribution of orbital orientations and a specific eccentricity distribution. Vector RR is a leading explanation for how their planes became randomized within their ~6 Myr lifetimes.
• The clockwise stellar disc: at ~0.05–0.5 pc, roughly half the young massive stars orbit in a coherent disc; its observed warp and thickness are attributed to vector RR partially scrambling an initially thin disc.
• Loss-cone feeding: scalar RR drives stars to high eccentricity, boosting rates of tidal disruption events and of extreme-mass-ratio inspirals (EMRIs) that future space detectors like LISA may hear.

Observationally the constraints come from decades of infrared astrometry (Keck, VLT/NACO, and VLTI/GRAVITY) tracking dozens of stars — the work that earned Genzel and Ghez a share of the 2020 Nobel Prize.

Comparison to Related Relaxation Regimes

Resonant relaxation is one of several processes that redistribute orbits near a black hole; distinguishing them matters:

• Two-body (non-resonant) relaxation: uncorrelated pair encounters that change both energy and angular momentum. It sets mass segregation and the density cusp but is slow (~Gyr at 0.1 pc). RR is faster specifically for angular momentum.
• Vector RR vs. the Kozai–Lidov mechanism: both alter orbital planes, but Kozai–Lidov is a coherent secular oscillation driven by a single dominant perturber (e.g. a disc or companion), whereas vector RR is a stochastic torque from many stars.
• Relativistic quenching: for very eccentric or tight orbits, fast GR precession shortens the coherence time so much that scalar RR is suppressed — the "Schwarzschild barrier" (Merritt et al. 2011) that limits how deep stars can be pushed toward the black hole.

In short: energy diffusion → two-body relaxation; coherent plane flips from one perturber → Kozai–Lidov; stochastic, collective angular-momentum diffusion → resonant relaxation.

Significance and Open Questions

Resonant relaxation is central to modern galactic-nucleus dynamics because it controls how quickly stars can be delivered to a black hole's loss cone, setting rates for two of the most sought-after phenomena in the field:

• Tidal disruption events (TDEs): stars driven to extreme eccentricity by scalar RR that graze within the tidal radius and are torn apart.
• Extreme-mass-ratio inspirals (EMRIs): compact remnants whose slow angular-momentum evolution — shaped by RR and the Schwarzschild barrier — determines whether they inspiral quietly or plunge, a key input for LISA event-rate predictions.

Open and debated questions include: the true numerical efficiency of RR (different N-body and Monte-Carlo studies disagree by factors of a few); how the "dark cusp" of stellar-mass black holes contributes to the torque, since it is unseen; whether vector RR fully accounts for the isotropy of the S-stars given their youth; and how strongly the Schwarzschild barrier throttles EMRI production. Resolving these hinges on continued GRAVITY-class astrometry and on ever-larger direct N-body simulations of the innermost parsec.

Scalar vs. vector resonant relaxation vs. classical two-body relaxation near Sgr A*

Property	Scalar RR	Vector RR	Classical (non-resonant) relaxation
What it changes	Magnitude of L (eccentricity)	Direction of L (orbital plane)	Both energy and L (random scattering)
Driving torque	Non-axisymmetric, in-plane residual torque	Non-spherical background mass distribution	Uncorrelated pair encounters
Coherence limited by	In-plane precession (mass + GR), ~10³–10⁴ orbits	Reorientation of orbital planes	None — always incoherent
Typical timescale	~10⁷–10¹⁰ yr	~10⁵–10⁷ yr	~10⁹–10¹⁰ yr at ~0.1 pc
Affects semi-major axis?	No	No	Yes
Physical consequence	Feeds stars to the loss cone / disruption	Randomizes / warps stellar discs	Slow energy redistribution, mass segregation