arXiv:2610.34615
Higher Order OTOCs in Chaotic Quantum Maps: Ruelle-Pollicott Resonances and Asymptotic Freeness
Nicolás Gómez Brucks, Ignacio García-Mata, Diego A. Wisniacki
Abstract
Out-of-time-ordered correlators (OTOCs) measure the scrambling of information and are a powerful diagnostic of chaos. In chaotic quantum systems with a classical analog, their relaxation towards equilibrium is dictated by the classical Ruelle--Pollicott resonances. We investigate higher-order OTOCs in quantum maps, prototypical one-body systems with a classical limit showing all the essential features of chaotic dynamics. We demonstrate numerically that in the intermediate-time regime they decay exponentially at a rate that depends only on the parity of the order: odd-order OTOCs decay at the same rate as classical correlations, given by the leading Ruelle--Pollicott resonance, and even-order ones at twice that rate. Both are far slower than the naive expectation, under which the rate would grow in proportion to the order. We then mimic the effect of an environment by defining a coarse-grained superoperator whose effect is to introduce decoherence into the system, and find that the expected proportionality between rate and order is thereby restored. Finally, we show that at long times each higher-order OTOC becomes dominated by a single free cumulant, whose saturation value falls off inversely with the dimension of the Hilbert space, signalling the emergence of asymptotic freeness and highlighting the relevance of free probability theory as a framework for the study of quantum chaos.
AI Summary
The paper asks how higher-order out-of-time-ordered correlators relax in chaotic systems with a classical limit, and how this relaxation connects to classical mixing and asymptotic freeness. It studies $F_k(t)=d^{-1}\operatorname{Tr}[(A(t)B)^k]$ numerically in the standard, Harper, and perturbed Arnold cat maps, using traceless sine-like position and momentum observables. Under unitary evolution, the intermediate-time decay after the Ehrenfest time depends on the parity of $k$, rather than growing proportionally with order: odd orders follow $|\lambda_1|^t$, while even orders follow $|\lambda_1|^{2t}$, where $\lambda_1$ is the leading classical Ruelle–Pollicott resonance. Simulations extend through $k=8$ at $d=5000$, with representative resonance moduli approximately $0.57$, $0.48$, and $0.64$ for the three maps, respectively. Resonances are extracted independently from a coarse-grained quantum superoperator using Arnoldi iteration. Replacing unitary evolution by this noisy channel gives order-dependent decay $|\lambda_1|^{kt}$, consistent with its spectral expansion; overlaps with subleading modes explain early-time deviations. Free-cumulant decompositions show that $F_k$ becomes dominated by the alternating cumulant $\kappa_{2k}$, which inherits the parity-dependent decay. Long-time mean absolute cumulants scale numerically as $O(d^{-1})$, consistent with asymptotic freeness and extending previous kicked-top observations. The unitary decay law is established numerically for these maps and observable choices, all odd under a shared parity symmetry. Mixed phase spaces delay or prevent the reported cumulant dominance and scaling. Extension to many-body systems without a classical limit remains open.
Error found: The reported parity-dependent decay is tested only for parity-odd X and P observables in parity-preserving maps, so its extension to general observables is not established. The alternating-cumulant tests support but do not establish asymptotic freeness, which requires vanishing of all mixed cumulants. The printed noise multiplier approaches unity on fixed Fourier modes as dimension increases at fixed ε, so the stated fixed-strength classical coarse-graining argument does not justify the spectral correspondence for that formula. The quoted standard-map chaos threshold uses a normalization inconsistent with the displayed map: its threshold should be approximately 0.154641 rather than 0.971635.
Written by gpt-6.1-sol from the paper's full text. Automated: it can be wrong.
- Comments
- 14 pages, 11 figures