Recoiling Black Holes I. Burst of Observables

Aims. To investigate the tidal disruption and gravitational wave events shortly after a massive black hole binary merges in the galactic centre and whose remnant black hole is ejected from the galaxy.

Methods. Using computational methods, black holes of mass M• = 105 M⊙ and 4 × 105 M⊙ embedded in a nuclear star cluster are kicked at velocities of vk = 300 and 600 km s−1. Systems are integrated for 0.1 Myr using a 4th-order Hermite scheme.

Results. The kick instantaneously repopulates the loss cone, producing a strong burst of tidal disruption events and gravitational wave mergers. The anisotropy in the apsidal orientation of bound stars prolongs this burst phase. Rates increase for lower vk, larger M• and for steeper nuclear star cluster density profiles at moment of merger.

Conclusions. Assuming binary black holes scour a Bahcall-Wolf density profile during coalescence, ejected remnants with mass between 105 ≤ M• [M⊙] ≤ 4 × 105 generate observable offset events at a forecasted rate of Ṅ ≲ 290 yr−1 up to redshift z = 3. If, at the moment of merger, the milliparsec scales of the nuclear star cluster are described by a shallow density profile (γ = 1), this decreases to Ṅ ≲ 30 yr−1. The strong dependence on the initial density profile and recoil kick makes observables powerful probes of the nuclear star cluster post-massive black hole binary coalescence, and provides test for numerical relativity.

Key results from the paper:

1. The instantaneous repopulation of the loss cone following a SMBHB or IMBHB coalescence, triggered by the recoil kick, produces a transient burst of gravitational-wave (GW) and tidal-disruption-event (TDE) activity. This burst phase lasts for several orbital periods, consistent with Madigan et al. (2018) and Akiba et al. (2024), and temporarily enhances the disruption rate from these systems.

2. Assuming the inner ∼ 10 mpc of post-merger galactic nuclei follow a Bahcall–Wolf cusp, recoiling systems hosting a central IMBH and ejected from their host galaxy will yield up to ≲ 290 offset TDEs and GWs per year up to redshifts z < 3. If γ = 1, forecasted rates decrease to ≲ 30 TDE and GW events per year.

3. With event rates being sensitive to γ, future observations (or lack thereof) can provide insights on the density profile of the inner parsec of nuclear star clusters the moment of IMBHB merger. Observations will also inform models of galaxy–galaxy merger rates, the stellar initial mass function (IMF) within NSCs and the extent of mass segregation prior to massive BH mergers.

Additional details:

• The study focuses on the burst phase of recoiling IMBH systems following dwarf-galaxy mergers. While several assumptions lead to optimistic forecasts, the resulting events should be detectable and may constitute smoking-gun evidence for IMBHB mergers, thereby constraining SMBH seed formation. Conversely, a null detection would imply either that IMBHB mergers are rare or that post-merger density profiles are shallow. The ratio between TDE-to-EMRI events will also provide insights on the mass function of the inner region within the NSC of galaxies post-merger.

• The simulations used direct N-body methods with the 4th-order predictor-corrector Hermite scheme Ph4, with no softening applied. Stars were initialised with masses between 0.5 and 100 M⊙, sampled from a power-law distribution with α = −1.35, and evolved to 100 Myr using SeBa before integration.

• The paper finds that the burst phase can have merger rates exceed that in a static NSC given the right parameters. When considering vk ≈ vesc, M• = 105 M⊙ and γ = 2.0, merging events are between ∼ 30 − 100 times greater than a case where the same BH mass is static in the galactic center. For γ = 1.75, this reduces to 7 − 20. Extrapolating to larger BH masses (M• = 106 M⊙), then for γ = 2, the increase is ∼ 102 − 103.

• The paper also discusses the potential connection to Little Red Dots (LRDs), suggesting that frequent TDEs during the early stages of a recoiling HCSC could mimic the radiative output of young stars and an accretion disk, producing the observed UV-bright continua and broad emission lines. However, the authors posit that HCSC can, at best, explain the faintest LRDs and only constitute a small subset of the population.

Improvements to AI Systems:

1. Transient Event Rate Predictor for Gravitational Wave and Tidal Disruption Events

• Improvement: Integrate the paper’s parameterized dependencies (black hole mass M•, kick velocity vk, nuclear star cluster density slope γ, and time since merger) into a machine-learning model that predicts time-varying TDE and GW merger rates during the post-merger burst phase.

• Capability: The AI can forecast transient event rates for arbitrary galactic nuclei parameters, enabling real-time prioritization of observational targets for surveys like LSST, ZTF, or LISA follow-ups.

2. Bayesian Inference Engine for Nuclear Star Cluster Properties

• Improvement: Train a Bayesian neural network on the simulation outputs (e.g., event rate ratios, burst duration, anisotropy signatures) to invert observables—such as TDE-to-EMRI ratios or offset event spatial distributions—into posterior distributions for γ, M•, and vk.

• Capability: The AI can infer the density profile and recoil kick of a post-merger galactic nucleus from sparse observational data, directly testing numerical relativity predictions and SMBH seed formation models.

3. Simulation Surrogate for N-Body Dynamics

• Improvement: Use the paper’s 4th-order Hermite simulation results as training data for a graph neural network or transformer that emulates the orbital evolution of stars around a recoiling black hole, including loss-cone repopulation and apsidal alignment effects.

• Capability: The AI can rapidly generate synthetic light curves and GW waveforms for thousands of merger scenarios without costly N-body integrations, enabling large-scale parameter sweeps and mock catalog generation for detector sensitivity studies.

4. Anomaly Detector for Offset Transient Events

• Improvement: Train a classifier on the predicted spatial offsets, luminosity evolution, and spectral signatures of recoiling IMBH-induced TDEs (as described in the paper) to distinguish them from other astrophysical transients (e.g., supernovae, AGN flares).

• Capability: The AI can flag candidate offset TDEs and GW host galaxies in real-time survey data, including those potentially linked to Little Red Dots, with quantified false-positive rates based on the paper’s rate forecasts (≤290 yr−1 vs. ≤30 yr−1).

5. Population Synthesis Optimizer for Galaxy Merger Models

• Improvement: Incorporate the paper’s rate dependence on γ and vk into a population synthesis framework (e.g., semi-analytic or Monte Carlo) that models dwarf galaxy merger histories and IMBHB formation channels.

• Capability: The AI can optimize input parameters (e.g., stellar IMF, mass segregation timescales, merger rates) to match observed TDE/GW event rates, providing constraints on galaxy–galaxy merger rates and the prevalence of IMBH seeds.

6. Real-Time Decision Support for Multi-Messenger Follow-Up

• Improvement: Build a reinforcement learning agent that uses the paper’s burst-phase timing (e.g., enhanced rates for several orbital periods post-kick) to schedule electromagnetic and gravitational-wave observations across multiple facilities.

• Capability: The AI can autonomously trigger follow-up observations of recoiling BH candidates, maximizing detection probability for the predicted burst events and distinguishing between shallow (γ=1) and steep (γ=2) density profiles.