From Black Hole to Galaxy: Neural Operator Framework for Accretion and Feedback Dynamics

Nihaal Bhojwani, Chuwei Wang, Hai-Yang Wang, Chang Sun, Elias R. Most, Anima Anandkumar, [3pt] Department of Computer, Mathematical, Natural Sciences, University of Maryland · 2025

Read From Black Hole to Galaxy: Neural Operator Framework for Accretion and Feedback Dynamics by Nihaal Bhojwani, Chuwei Wang, Hai-Yang Wang, Chang Sun, Elias R. Most, Anima Anandkumar, [3pt] Department of Computer, Mathematical, Natural Sciences, University of Maryland (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2512-01576.txt.

Introduction

Nihaal Bhojwani and co-authors model how black holes and galaxies grow together with neural operators. The physics spans a huge range of scales. Gas near the hole acts at tiny scales. Galaxy-wide feedback acts at vast scales. Full simulations of both ends are too costly.

Their neural operator frame learns the coupled dynamics fast. It bridges accretion at the center with feedback across the galaxy. Fast surrogates let teams scan many what-if cases. Those scans test how black hole growth steers star birth over cosmic time. Surrogate speed also lets observers fit live merger data faster, since galaxy feedback priors can be scanned while wave data is still fresh.

Catalog record

CREATORNihaal Bhojwani, Chuwei Wang, Hai-Yang Wang, Chang Sun, Elias R. Most, Anima Anandkumar, [3pt] Department of Computer, Mathematical, Natural Sciences, University of Maryland
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE45 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2512-01576

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

From Black Hole to Galaxy: Neural Operator Framework for Accretion and Feedback Dynamics Nihaal Bhojwani, Chuwei Wang, Hai-Yang Wang, Chang Sun, Elias R. Most, Anima Anandkumar, [3pt] Department of Computer, Mathematical, Natural Sciences, University of Maryland Abstract Modeling how supermassive black holes co-evolve with their host galaxies is notoriously hard because the relevant physics spans nine orders of magnitude in scale—from milliparsecs to megaparsecs—making end-to-end first-principles simulation infeasible. To characterize the feedback from the small scales, existing methods employ a static subgrid scheme or one based on theoretical guesses, which usually struggle to capture the time variability and derive physically faithful results. Neural operators are a class of machine learning models that achieve significant speed-up in simulating complex dynamics. We introduce a neural-operator–based “subgrid black hole” that learns the small-scale local dynamics and embeds it within the direct multi-level simulations. Trained on small-domain (general relativistic) magnetohydrodynamic data, the model predicts the unresolved dynamics needed to supply boundary conditions and fluxes at coarser levels across timesteps, enabling stable long-horizon rollouts without hand-crafted closures. Thanks to the great speedup in fine-scale evolution, our approach for the first time captures intrinsic variability in accretion-driven feedback, allowing dynamic coupling between the central black hole and galaxy-scale gas. This work reframes subgrid modeling in computational astrophysics with scale separation and provides a scalable path toward data-driven closures for a broad class of systems with central accretors. 1 1 footnotetext: ∗ Equal contribution. In alphabetical order. 1 Introduction Background. Supermassive black holes (SMBHs) co-evolve with their host galaxies through a two-way feeding-feedback loop that operates across nine orders of magnitude in scale [ 26 ] . Accretion flows near the event horizon (milliparsec scales) are governed by general relativistic magnetohydrodynamics (GRMHD) and launch relativistic jets and winds. These outflows propagate from AU to megaparsec scales, well beyond the host galaxy, depositing energy and momentum into the interstellar and intergalactic media [ 1 , 35 ] . The resulting feedback regulates star formation and galaxy growth, while galaxy-scale gas dynamics in turn modulate the inflow feeding the central black hole. Figure 1: Illustration of our method in a two-level setting. The neural operator efficiently replaces the evolution on small scales (zoom in), enabling proper feedback on large scale (zoom out). Challenge. This multiscale coupling makes end-to-end simulation computationally intractable: accurately resolving accretion flows demands timesteps set by the gravitational radius ( r_{g}\sim\mathrm{mpc} for M87), while capturing galaxy-scale feedback requires following dynamics over \sim 10^{9} times larger spatial and temporal scales. Bridging these domains requires passing information across many orders of magnitude in space and time without washing out variability or violating conservation. A considerable amount of efforts adopting various techniques has been made to bridge different scales, including direct simulations under reduced scale separation [ 29 , 28 , 41 , 30 ] , nested mesh “zoom-in” approaches in Eulerian codes [ 15 , 16 ] , “Lagrangian hyper-refinement” in Larangian codes [ 20 , 21 ] , remapping between domains [ 24 ] , and iterative “multi-zone” [ 4 , 6 , 5 ] or “cyclic-zoom” [ 17 ] schemes that repeatedly refine/de-refine around the SMBH. Among these, multi-zone and cyclic-zoom methods propagate small-scale feedback to larger domains and evolve to consistency, but can still struggle with temporal variability and with specifying a physically faithful inner boundary (here we name this by subgrid black hole ) across levels. The inner boundary adopted this way could feed erroneous information (e.g., in both methods, the relativistic jet cannot be turned on/off when crossing the boundary) to the active simulation domain, consequently alter the final results. Neural Operator. Neural operators (NOs) represent a recent advance in machine learning for partial differential equations, learning mappings between infinite-dimensional function spaces rather than approximating individual functions [ 32 , 27 ] . This approach has demonstrated substantial computational acceleration and shown promise as surrogate models across various applications [ 14 , 39 , 49 , 45 , 47 , 48 ] . Recent studies have begun exploring NOs for coarse graining or closure modeling problems [ 47 ] . Astrophysical systems present unique challenges. Realistic galaxy formation involves large-scale, strongly coupled dynamics with central singularities at supermassive black holes, requiring stable long-term integration over cosmological timescales. The extreme dynamic range, combined with the inherent scarcity of training data in astrophysical simulations, creates a particularly demanding test case for NO or other machine learning methods. To the best of our knowledge, despite some preliminary attempts in simplified systems or transient-state predictions [ 38 , 8 , 40 , 25 ] , no previous work has demonstrated NO performance in such complex, multi-scale systems with limited training data. Contribution. We introduce a neural-operator-based “subgrid black hole” coupled to a direct multi-level solver, replacing hand-crafted closure rules with a data-driven model that captures realistic variability in accretion and feedback (see fig. 1 for a schematic illustration of our method): (1) Operator-learning subgrid. We train a neural operator to approximate the small-scale (GR)MHD time-evolution semigroup. The learned model supplies boundary conditions and fluxes to the next coarser level, enabling long-horizon rollouts without assuming steady or time-averaged injection. (2) Two-way multiscale coupling. Embedded within a multi-level framework, the learned subgrid responds to evolving large-scale conditions and, in turn, drives feedback that propagates outward—preserving variability critical to galaxy–SMBH co-evolution. (3) General applicability. The demonstrated approach is broadly applicable as a subgrid model for systems with a central accretor (e.g., SMBHs and neutron stars), independent of problem setups and generalizable to different codes. This work reframes subgrid modeling in computational astrophysics as operator learning : rather than prescribing fixed closures, we learn the small-scale dynamics to provide dynamically-updated boundary conditions for larger domains. This advance in subgrid modeling fills up the missing piece for the black hole accretion and feedback problem. More importantly, this could revolutionize the modeling of black hole feedback in cosmological simulations such as FIRE [ 23 ] and IllustrisTNG [ 37 ] . Related Works. Analytical subgrid prescriptions remain the standard tool in large-scale such as cosmological simulations (e.g., supernova feedback, cold-gas processes, and black-hole growth [ 22 , 3 ] ). Machine-learning–based subgrid/surrogate models are only beginning to appear, with early results for supernova feedback [ 19 , 18 ] and for 2D black-hole accretion surrogates [ 9 ] trained on Newtonian hydrodynamical simulation of steadily accreting torus. Our work advances this line by learning an operator-level “subgrid black hole” compatible with multi-level MHD/GRMHD. Figure 2: MHD outputs. Simulated (‘DNS’), CNN-predicted, and NO-predicted magnetic field component B_{z} , density \rho , internal energy E_{int} , and velocity component v_{z} from the MHD simulation on a horizontal slice through the domain center (after 50 steps). Unphysical artifacts are marked. (2) Comparison of the mass density distributions on a 2D sphere, between analytically prescribed, directed simulated, and NO-rollout (after 10 steps). (3) The radial profiles of predicted quantities: mass density \rho , temperature T , and total mass flux \dot{M} ( Left to Right ). Profile from simulation after 50 steps is also shown. The grey-shaded region shows the part that will be coupled to the direct simulation (see Appendix E ). 2 Method Method Framework. We consider the (general relativistic) magnetohydrodynamics (GRMHD) equations on a

Read the complete text