Stable black hole solutions with cosmological hair

This paper investigates the existence and stability of black hole solutions in scalar-tensor theories, specifically focusing on the cubic Galileon as a representative model for dynamical dark energy. It addresses a critical gap in gravitational physics: while time-dependent scalar fields naturally produce cosmological hair, previously known hairy solutions have been found to be unstable. Establishing stable, regular black hole solutions is essential for directly probing cosmological dynamics with black hole observations.

The challenge of cosmological hair

In theories involving dynamical dark energy, the time-dependent field introduces cosmological hair, meaning local black hole physics is impacted by the dark energy field rather than just mass and spin. The authors focus on the cubic Galileon, which captures essential features like self-accelerating solutions and screening mechanisms. To analyze this, they utilize a significant hierarchy of scales where cosmological and black hole scales are separated by approximately 20 orders of magnitude. This is expressed through dimensionless parameters:

• alpha 1 about (H 0 r s)-1

• alpha 2 about (H 0 r s) squared

By leveraging these hierarchies, the authors attempt to find solutions where the scalar field is screened locally, making the solution appear similar to a Schwarzschild solution near the horizon.

Failure of static black hole models

The researchers analyzed the branching structure of the scalar solution to determine how short-range and long-range solutions connect. They identified two distinct branches:

• The+ branch, which is necessary to recover the desired cosmological long-range behaviour.

• The- branch, which cannot be connected to cosmological asymptotes without a branch switch that requires an unphysical discontinuity.

Through detailed perturbation analysis in the Regge-Wheeler gauge, the study found that neither static branch is viable. The+ branch exhibits a ghost instability in the scalar modes, while the- branch displays a ghost instability behind the black hole horizon. Consequently, no stable and static black hole solution can be identified that connects to a cosmologically relevant background.

The quasi-stationary approach

To overcome these instabilities, the authors relax the initial assumption of purely static black holes and instead consider quasi-stationary solutions with a suppressed time dependence. This is achieved by introducing a non-zero shift-symmetry current J r, which allows for the accretion of energy associated with the scalar field. The researchers establish several conditions for these solutions:

1. The time dependence must be suppressed to allow for the existence of quasi-stationary solutions.

2. The black hole must evolve on long cosmological time scales of the order of a Hubble time.

3. The accretion parameter alpha 4 must be large enough to compensate for the non-vanishing part of the (tr) equation near the horizon.

Stable accreting solutions

By allowing for this mild time dependence, the authors explicitly derive novel, stable and regular solutions that successfully bridge local and cosmological scales. They demonstrate that for specific values of the accretion parameter—specifically when alpha 4 beta / alpha 1 is close to-1 —the+ branch becomes stable in the short-range regime. In these cases, the kinetic part of the Hamiltonian is bounded from below up to a maximum radius x max. These stable accreting black hole solutions maintain well-behaved dynamics near the horizon while successfully connecting to the desired cosmological long-range behaviour, providing a candidate dark energy theory with a well-behaved black hole solution.

1. Hierarchical Multi-Scale Neural Solvers

• Improvement: Implement a neural architecture that utilizes Hierarchical Parameter Decomposition, inspired by the paper's treatment of extreme scale separations (e.g., alpha 1 about 10 20 and alpha 2 about 10-40). This involves training sub-networks on localized short-range regimes and long-range asymptotes, then using a specialized bridging layer to match them via shooting methods.

• Capability: The improved AI can accurately simulate complex physical or financial systems where local dynamics (micro-scale) and global constraints (macro-scale) operate across dozens of orders of magnitude without the numerical divergence or stiffness that causes standard Physics-Informed Neural Networks (PINNs) to fail.

2. Hamiltonian-Boundedness Verification Modules for Control Systems

• Improvement: Integrate a formal Effective Metric Stability Check into the loss function of Reinforcement Learning (RL) agents. This module would analyze the quadratic action of the agent's learned policy to ensure that the effective kinetic energy remains positive-definite across all perturbation modes, specifically checking for ghost instabilities (non-physical energy growth).

• Capability: The system can provide formal mathematical guarantees that a learned control policy (e.g., for autonomous power grids or chemical reactors) is robust against perturbations and will not enter unstable, divergent oscillations that traditional stability checks might overlook in non-linear environments.

3. Quasi-Stationary State Estimators for Non-Linear Dynamics

• Improvement: Develop a predictive model that moves beyond static equilibrium assumptions by adopting the paper's Quasi-Stationary Ansatz. Instead of searching for =0, the AI would solve for solutions where time dependence is suppressed by a specific scale factor (the accretion rate), allowing for slow, stable evolution.

• Capability: The improved AI can predict long-term trajectories in highly non-linear, time-varying systems (such as climate modeling or market volatility) where the system appears to be in equilibrium but is actually undergoing a continuous, slow-scale evolution that would be missed by static modeling.

4. Branching Structure Discovery Engines

• Improvement: Implement a Topological Branch-Switching Detector within generative models for differential equations. This uses the mathematical concept of odd multiplicity branch points to identify regions in parameter space where a system might undergo a sudden, discontinuous transition from one stable state to another.

• Capability: The AI can predict tipping points in complex systems—identifying when a small change in input parameters will force the system to jump from one stable branch (e.g., an efficient operating state) to a different, potentially catastrophic branch (e.g., a failure state).

Dynamical dark energy theories generically introduce a time-dependent field that causes the accelerated expansion of the Universe on large scales. When embedding black hole solutions in such a cosmological spacetime, this time dependence naturally gives rise to cosmological hair, i.e. the local black hole physics is no longer controlled by just the mass and spin of the black hole, but also impacted by the dark energy field. However, known such solutions are unstable. Focusing on the cubic Galileon as a concrete and illustrative example, we discuss the restrictions imposed on physical solutions by their regularity and stability in detail. We explicitly derive solutions that both recover the desired cosmological long-range behaviour and give rise to well-behaved, regular and stable, short-range dynamics around black holes. We show how the nature of the scalar hair around these local black hole solutions encodes cosmological information, highlighting novel and tantalising prospects of directly probing cosmological dynamics with black hole observations.