Gas Supply, Inflow, and Feedback in Nuclei

Gas observed near a galactic centre is not automatically fuel for the black hole. It must lose angular momentum across many decades in radius while cooling, forming stars, exchanging torque with non-axisymmetric structures, and possibly being expelled. The central result is a budget, not a universal accretion prescription: gravitational torques can produce an inflow of order $\lvert a\rvert\Sigma_gR^2\Omega$, but star-formation and outflow sinks can remove most of that supply before it reaches a subparsec disk. Every numerical claim must therefore name its outer supply, inner sink, flux surfaces, thermodynamic closure, and resolved scale.

Required background. Classical Field Theory supplies local conservation laws, stress tensors, and flux integrals used to close the budgets.

Helpful background. Black-Hole Geometry and Mechanics identifies the central boundary and the distinction between nuclear and horizon scales; Relativistic Matter and Compact Objects supplies fluid, cooling, and stellar-feedback closures; Black-Hole Astrophysics places fueling in the accretion and feedback chain; and Relativistic Two-Body Dynamics distinguishes gas torques from relativistic compact-body orbital dynamics.

Mass and angular-momentum budgets

For an azimuthally averaged gas surface density $\Sigma_g(R,t)$, write

$$\frac{\partial\Sigma_g}{\partial t} +\frac{1}{R}\frac{\partial}{\partial R} \left(R\Sigma_g v_R\right) =\dot\Sigma_{\rm sup} -\dot\Sigma_\star -\dot\Sigma_{\rm out}.$$

The inward mass flux is $\dot M_{\rm in}=-2\pi R\Sigma_gv_R>0$ for $v_R<0$. Supply at a kiloparsec, inflow through one hundred parsecs, and accretion through an inner sink are different quantities. In a time-dependent system their differences accumulate in the intervening gas reservoir or are consumed by the sink terms.

Let $j=R^2\Omega$ be the circular specific angular momentum and $\tau_\Sigma$ the net torque per unit area, with its sign positive when it adds angular momentum to the gas. Let supplied and expelled gas carry specific angular momenta $j_{\rm sup}$ and $j_{\rm out}$, and assume newly formed stars remove the local gas value $j$. The corresponding equation is

$$\frac{\partial(\Sigma_gj)}{\partial t} +\frac{1}{R}\frac{\partial}{\partial R} \left(R\Sigma_gv_Rj\right) =\tau_\Sigma+j_{\rm sup}\dot\Sigma_{\rm sup} -j\dot\Sigma_\star -j_{\rm out}\dot\Sigma_{\rm out}.$$

Subtracting $j$ times the mass equation makes the angular-momentum changes explicit:

$$\begin{aligned} \Sigma_g\left(\frac{\partial j}{\partial t} +v_R\frac{\partial j}{\partial R}\right) ={}&\tau_\Sigma +(j_{\rm sup}-j)\dot\Sigma_{\rm sup}\\ &-(j_{\rm out}-j)\dot\Sigma_{\rm out}. \end{aligned}$$

In a stationary, source-free disk, this reduces to $v_R=\tau_\Sigma/(\Sigma_g\,dj/dR)$. A negative torque therefore drives inward drift when $dj/dR\gt0$, as in a Keplerian disk. Low-angular-momentum supply or a wind carrying more than the local $j$ can also drive drift, even without an additional torque. These terms must be distinguished from spurious numerical angular-momentum loss at a nested boundary.

For a non-axisymmetric potential perturbation of fractional amplitude $\lvert a\rvert$, orbit crossing and shocks give an order-of-magnitude strong-torque scaling

$$\dot M_{\rm torque}\sim C\,\lvert a\rvert\Sigma_gR^2\Omega,$$

where $\Omega$ is orbital angular frequency and the order-unity coefficient $C$ contains the mode, gas phase shift, and radial structure. The scaling says that a mass of order $\Sigma_gR^2$ loses an appreciable part of its angular momentum in a time of order $(|a|\Omega)^{-1}$ in this strong-shock regime. This is not a local viscosity law. The analytic shock and torque regimes are derived and compared in Hopkins and Quataert 2011, §§2.3–3. Their simulations support the scaling statistically, with substantial instantaneous scatter; a single snapshot should not be inverted for a precise accretion rate.

The bottleneck hierarchy

Large-scale tidal disturbances, bars, spirals, clumps, and eccentric nuclear disks can each dominate over a limited radial range. The potential becomes nearly Keplerian close to the black hole, so a large-scale bar does not simply continue unchanged to the horizon. A lopsided mode, local magnetic stress, or another transport process must take over. “Bars within bars” is therefore a qualitative route, not a boundary-free algorithm.

Self-gravity is organized by the Toomre parameter

$$Q=\frac{\kappa c_{\rm eff}}{\pi G\Sigma_g},$$

where $\kappa$ is the epicyclic frequency and $c_{\rm eff}$ includes the declared thermal or turbulent support. When cooling keeps $Q$ near or below unity, fragmentation and star formation compete with inflow. Raising a pressure floor may prevent unresolved fragmentation numerically while also changing the physical phase balance. The assumed star-formation law and feedback coupling must therefore be varied independently, not hidden inside one “effective sound speed.”

The Milky Way’s ionized, neutral, dusty, and hot gas components illustrate that morphology and phase are selection dependent Genzel et al. 2010, §§3.1–3.4. Nuclear-cluster growth also admits both cluster infall and in-situ star formation, so present stellar light is an integrated record rather than a direct measure of current gas supply Neumayer et al. 2020, §§7.2–7.3, pp. 25–28.

A finite-volume transport benchmark

Before adding self-gravity, cooling, or feedback, a transport solver should reproduce a conservation problem with a known solution. Define $q=R\Sigma_g$ on a dimensionless periodic annulus. For constant radial speed $v_R=-1$ and depletion time $t_\star=2$, solve

$$\frac{\partial q}{\partial t} +v_R\frac{\partial q}{\partial R} =-\frac{q}{t_\star}, \qquad q(R,0)=1+0.2\sin(2\pi R).$$

The exact solution is an advected sinusoid multiplied by $e^{-t/t_\star}$. The following first-order upwind finite-volume method uses exact exponential depletion after each transport step.

Use Python 3.12.5 with NumPy 2.1.0. Save the program as nuclear_inflow_advection_demo.py. The script emits resolution, the mean absolute error at $t=0.2$, and the global mass-budget residual.

import numpy as np

def run(resolution, t_end=0.2, velocity=-1.0, t_star=2.0):
    dx = 1.0 / resolution
    radius = (np.arange(resolution) + 0.5) * dx
    q = 1.0 + 0.2 * np.sin(2.0 * np.pi * radius)
    time = 0.0
    while time < t_end - 1.0e-15:
        dt = min(0.4 * dx / abs(velocity), t_end - time)
        upwind = np.roll(q, -1) if velocity < 0.0 else q
        flux = velocity * upwind
        q -= (dt / dx) * (flux - np.roll(flux, 1))
        q *= np.exp(-dt / t_star)
        time += dt
    exact = np.exp(-t_end / t_star) * (
        1.0 + 0.2 * np.sin(2.0 * np.pi * ((radius - velocity * t_end) % 1.0))
    )
    error = np.mean(np.abs(q - exact))
    budget = abs(np.mean(q) - np.exp(-t_end / t_star))
    return error, budget

for n_cells in (64, 128, 256):
    print(n_cells, *run(n_cells))

Create the pinned environment and rerun with:

python3 -m venv inflow-demo
source inflow-demo/bin/activate
python3 -m pip install numpy==2.1.0
python3 nuclear_inflow_advection_demo.py

The expected refinement record is:

Resolution N	Mean absolute error	Mass-budget residual
64	4.18886e-3	2.22e-16
128	2.11249e-3	8.88e-16
256	1.06110e-3	7.77e-16

The error halves when resolution doubles, consistent with first-order upwinding, while the periodic global mass changes only by the analytically imposed sink to roundoff. This benchmark checks transport and mass accounting. It does not validate a gravitational-torque closure, angular-momentum exchange, cooling, or feedback.

From benchmark to a physical calculation

A physical hydrodynamic or MHD calculation must pin the solver and commit, gravity implementation, equation of state, cooling and chemistry or opacity tables, star-formation and feedback closures, magnetic-field initialization, and random seeds. Initial data include the galaxy and nuclear potential, gas surface density and angular-momentum distribution, outer supply history, and every nested boundary. Diagnostics should report mass, angular momentum, and energy crossing identical flux surfaces.

The minimum independent controls are a zero-torque angular-momentum-conserving disk and a cited non-axisymmetric torque benchmark before star formation and feedback are enabled. Refinement must vary mass, spatial, and temporal resolution separately. It should also move the inner sink, outer supply surface, and refinement interfaces. Hopkins and Quataert explicitly compare mode structure and inflow across resolution and feedback choices in Hopkins and Quataert 2011, §4 and Fig. 4, while their gravitational-torque estimator and its simplifying assumptions are collected in Hopkins and Quataert 2011, §5.2.

Major failure modes are numerical angular-momentum loss, inflow launched by a boundary transient, an unresolved multiphase medium, cooling-floor artifacts, stochastic feedback mistaken for convergence, and an inner sink larger than the physical transport transition. Closing only the mass budget is insufficient: a scheme can conserve mass while diffusing angular momentum. Agreement in a time-averaged inflow rate can also hide wrong variability or phase structure.

The executable calculation on this page is not a claimed production simulation of nuclear fueling. It is a teaching recipe for one operator and one budget. Nor does a resolved flux at parsec scale equal horizon accretion; the intervening disk, winds, and magnetic stresses remain separate physics.

Interpretation and scope

The main scientific inference is conditional: strong gravitational asymmetries can move gas inward rapidly when shocks create a phase offset, but the delivered black-hole fuel is the remainder after storage, star formation, and outflows. Feedback can both remove gas and alter the diffuse phase needed for torque coupling, so “feedback suppresses accretion” is not a universal local sign rule.

Generic galactic dynamics unrelated to the central black hole is outside this page. Large-scale bars, mergers, and halo structure enter only as boundary conditions or torque sources for the nuclear budget. The calculation determines the supply to the inner accretion flow; following that material to the horizon requires a separate flow model.

Exercises

The finite-volume table shows nearly machine-precision mass balance at all three resolutions, while the profile error decreases by a factor of about two. What order of convergence is demonstrated, and why would it be wrong to call the solution converged solely from the mass residual?

Solution

Doubling $N$ halves the mean absolute error, so the resolved profile shows first-order convergence, as expected for the upwind method. The mass residual tests only the global discrete budget; periodic fluxes cancel exactly and the sink is applied analytically. A severely diffused sinusoid can still conserve its mean to roundoff. One must therefore refine the spatial profile, compare with the exact solution, and separately check angular momentum and boundary fluxes before judging a physical inflow calculation.

References

• Genzel, Reinhard, Frank Eisenhauer, and Stefan Gillessen. 2010. “The Galactic Center Massive Black Hole and Nuclear Star Cluster.” Reviews of Modern Physics 82: 3121–3195. DOI.
• Hopkins, Philip F., and Eliot Quataert. 2011. “An Analytic Model of Angular Momentum Transport by Gravitational Torques: From Galaxies to Massive Black Holes.” Monthly Notices of the Royal Astronomical Society 415: 1027–1050. DOI.
• Neumayer, Nadine, Anil Seth, and Torsten Böker. 2020. “Nuclear Star Clusters.” Astronomy and Astrophysics Review 28: 4. DOI.