How would you distinguish a phase-space accident from genuine intermediate resonance structure?

This is a sharp technical question that sits at the boundary of dynamical systems theory and quantum/gravitational physics. Let me work through what the archive says about distinguishing genuine resonance structure from accidents of the phase-space representation.

The Core Distinction

A phase-space accident occurs when resonance signatures appear in measurements or calculations simply because of how you set up the coordinate system, boundary conditions, or observation apparatus—not because the underlying physics has genuine resonant coupling. A genuine intermediate resonance is a persistent feature of the dynamics independent of representation.

The clearest guidance comes from [AG-2024.04-2240], which directly tackles this in a driven quantum system. The authors found that level *attraction* in a two-tone driven bosonic system initially looked like evidence for dissipative coupling, but detailed theoretical analysis revealed it "originates from interferences due to the measurement setup, and not dissipative coupling." This is a textbook phase-space accident: the resonance signature was real in the data, but the mechanism was an artifact of *how* they measured it [AG-2024.04-2240]. The fix was to derive the correct effective Hamiltonian accounting for measurement geometry.

How to Test for Genuine Structure

Several diagnostic strategies emerge:

1. Topological robustness. Genuine resonances near exceptional points (EPs)—where eigenvalues and eigenfunctions coalesce—leave characteristic signatures that survive representation changes. In black hole ringdown, the EP framework shows that near-resonant modes produce "enhanced mode contributions in the time domain, resulting in characteristic departures from exponentially damped oscillations" [AG-2025.12-027]. If you switch observables or coordinates and the signature vanishes, it was accidental.

2. Non-Hermitian structure in scattering. If your system has open boundaries or dissipation, genuine resonances appear as poles of the S-matrix or of the dilated Hamiltonian in the complex energy plane [AG-2025.12-1323]. Accidental phase-space crossings won't show this pole structure—they'll disappear under complexification.

3. Boundary-condition dependence. A system described by a Hamiltonian is deeply sensitive to boundary conditions: different choices can yield "several different physically observable phases, each exhibiting its own characteristic global symmetry" [AG-2024.03-2290]. If a resonance disappears when you change boundary conditions (or measurement geometry), you've likely found an accident. If it persists, it's usually real.

4. Dynamical signature in time domain. Genuine resonances produce characteristic time-domain behavior. In the black hole case, resonant excitation near an EP gives a "linear growth in time" superposed on the usual exponential decay [AG-2025.12-027]; in driven systems, you see beating and energy transfer patterns [AG-2024.06-1593]. Accidents of phase space typically don't produce these persistent dynamical signatures.

5. Parameter continuity. In the oscillon study [AG-2024.04-589], the energy-frequency diagram is fragmented into disjoint segments—a sign of genuine resonant structure—whereas the variational approximation breaks down outside certain frequency windows, revealing the limits of any single representation.

Practical Workflow

When you observe resonance-like behavior:

1. Vary your measurement setup or coordinates. Does the signature persist?

2. Compute the effective Hamiltonian (including non-Hermitian terms if appropriate). Does it predict the resonance from first principles?

3. Look in the complex plane (eigenvalues, S-matrix poles, Riemann surfaces). Are there genuine singularities, or just crossings?

4. Check time-domain evolution. Real resonances leave persistent dynamical marks; accidents typically don't.

The provided preprints don't contain a general algorithm for this distinction, but they converge on the principle: genuine resonances are representation-independent features of the underlying dynamics, often signaled by non-Hermitian structure, topological properties, or characteristic time-domain behavior, whereas accidents vanish when you change how you observe the system.