arXiv:2610.28462
Tuning between collective dynamics and scrambling through initial-state symmetry alignment in long-range spin models
Sridevi Kuriyattil, Johannes Kombe, Andrew J. Daley, Pablo M. Poggi
Abstract
Long-range interactions in many-body quantum systems give rise to a wide range of dynamical phenomena, from collective dynamics well described by integrable mean-field models with symmetric all-to-all couplings, to fast scrambling of quantum information from disordered long-range interactions. Here, we investigate scenarios in which both regimes emerge within a single deterministic, translation-invariant XY spin model with long-range couplings. We consider a sparse powers-of-two (PWR2) graph, where each spin interacts with a logarithmic number of neighbors, and a power-law-decaying model. By altering the initial state via local perturbations on a completely symmetric state, we show that the dynamics changes from collective to scrambling as witnessed by bipartite and multipartite entanglement measures. We characterize this collective-to-scrambling crossover using Lyapunov exponents in the semiclassical regime, and show that the spectral properties of the coupling graph adjacency matrix provide a very efficient tool to predict the onset of fast scrambling. Our results offer a unified framework for controlling the switch between collective and scrambling dynamics in a setup accessible in experiments with neutral atoms in cavities or in tweezer arrays.
AI Summary
Can a fixed, deterministic long-range spin Hamiltonian support both approximately collective dynamics and scrambling, selected through initial-state preparation? This paper studies periodic XY chains with either powers-of-two (PWR2) connectivity, giving $O(\log N)$ neighbors per spin, or power-law couplings $\chi_{ij}\propto r_{ij}^{-1}$. Initial states are fully $x$-polarized product states with $k$ spins flipped in contiguous, alternating, or random patterns. Exact diagonalization and matrix-product-state evolution, reaching $N=32$, show that increasing $k$ generally suppresses collective-spin quantum Fisher information and revivals while increasing bipartite entanglement and reducing late-time fluctuations. Large-spin semiclassical calculations find increasing exponential sensitivity for PWR2 under both structured flip patterns, and for power-law interactions under contiguous—but not alternating—flips. For selected configurations, the rescaled scrambling time satisfies $\lambda_L t_*\propto\log N$. Linearization around polarized configurations relates sensitivity growth to adjacency-matrix spectra; for the alternating background, Fourier modes obey $\ddot y_q=\lambda_{N/2}(\lambda_q-\lambda_{N/2})y_q$. A complementary local-spin-flip transformation produces signed coupling graphs whose Laplacian gaps are used to diagnose loss of collective protection. Both spectral-gap and semiclassical threshold analyses yield a crossover scale $k_*\sim\sqrt N$, with fitted exponents approximately $0.50$–$0.53$. Quantum simulations are limited by rapid entanglement growth, leaving sparse $N=32$ data. Scrambling-time evidence comes from the semiclassical limit rather than quantum OTOCs. The contiguous-background spectral treatment is approximate, neglecting higher Fourier harmonics; the distinct late-time behavior of the alternating Néel state remains incompletely explained.
Central claim may not hold: The spin-1/2 entanglement diagnostics and large-S fixed-point instability calculations support a dynamical crossover but do not establish fast quantum scrambling under the same assumptions. The appendix establishes logarithmic scaling of λ_Lt*, which does not establish logarithmic scaling of t* in fixed physical units without controlling the size dependence of λ_L. The signed-Laplacian spectral distance is not generally the collective-to-noncollective many-body gap: the N=4 PWR2 model with one flipped spin has a nonzero Laplacian distance but an exact many-body degeneracy. The claimed early-time power law using weighted graph distance is incorrect for power-law interactions, where every distinct pair has a direct coupling and classical sensitivity starts at order t². The displayed generalized-OAT decomposition sums to the negative of the stated XY Hamiltonian.
Written by gpt-6.1-sol from the paper's full text. Automated: it can be wrong.
- Comments
- 20 pages, 11 figures in the main text, and 6 figures in the appendices