The Limits of Recurrence Suppression: Symmetry-Breaking and Participation in Discrete Baths
Description
The companion paper fixed the cost of an exactly localized continuous pointer: under the finite-energy postulate, an exact open spatial void either leaves the form domain or fails to persist under continuous dynamics, leaving a discrete substrate (Path D) as the route to exact spatial structure. A discrete substrate must still suppress coherence recurrences. A finite graph bath has a pure point spectrum, so the reduced coherence is almost-periodic and limsup_t |rho(t)| = |rho(0)| for every finite harmonic graph bath; suppression is delay within a finite window, not erasure. By moving to a discrete space, we have simply traded the continuous spatial tails of Paper I for discrete temporal Poincare recurrences. This paper fixes how much suppression a finite graph provides, and how much of it a local observer can read off the bath's geometry.
We work in one regime throughout: single-site, Gaussian, linear-coupling pure dephasing, with a stipulated sigma_z pointer basis. Two results organize the analysis. First, a mechanical baseline: |rho(t)| = e^{-Gamma(t)} with Gamma a sum over distinct grouped frequencies, and the recurrence resource is extensive coupling-weighted participation N_eff^dyn, which sets the typical log-coherence variance but not the recurrence tails. Second, the observational law: by Wick's theorem the coherence is a functional of the coupling-weighted local spectral density J(omega) and temperature alone. Two baths with matched grouped J(omega) produce identical coherence, and a single-qubit observer is blind to any graph structure not carried by J(omega). This law establishes the exact mechanism behind the complete-graph/star-hub coincidence evaluated here.
Operating under that law, we evaluate the candidates. The spectral gap (and, via Cheeger's inequality, the Cheeger constant) are dissociated from the resource and at the extremes inversely ordered: the complete graph (lambda_2 = N) gives N_eff^dyn ~ 1, the gapless cycle reaches 68-75. Exact cubic symmetry collapses the 7^3 torus to N_dist = 20 active frequencies (N_eff^dyn = 12.30); breaking exact homogeneity by weak bond disorder restores N_eff^dyn ~ 49.3, approaching a degree-matched random regular graph (52.7) up to a small but reproducible residual. The expander is a mathematical asymptote that a disordered Euclidean grid effectively approaches, not a physically separate resource.
Three problems remain open, and they form the program's frontier. Because single-site dephasing depends only on J(omega), whether geometry constrains local decoherence reduces to a realizability question: can a physically admissible 3D Euclidean grid reproduce an arbitrary expander's grouped J(omega) exactly? And reading topology off the bath at all requires leaving the single-site regime, through a multi-site cross-coherence sensitive to the spectral matrix J_ij(omega) or a multi-site transport diagnostic. And securing a robust macroscopic pointer demands a dephasing depth that single-site point coupling does not reach, since the extended form factors that would deepen it suppress the participation that disperses revivals. We state all three precisely and invite their resolution.
Data and Code Availability: The deterministic Python scripts generating all numerical claims and gap-dissociation metrics are permanently hosted in this project's Open Science computational repository (Zenodo, DOI: 10.5281/zenodo.21007211).
Files
NCW_Paper2_v79.12.pdf
Additional details
Related works
- Cites
- Preprint: 10.5281/zenodo.20836306 (DOI)
- Is supplemented by
- Software: 10.5281/zenodo.21007211 (DOI)