Diagnosing chaos with projected ensembles of process tensors
Peter O’Donovan, Neil Dowling, Kavan Modi, Mark T. Mitchison · 2025
Read Diagnosing chaos with projected ensembles of process tensors by Peter O’Donovan, Neil Dowling, Kavan Modi, Mark T. Mitchison (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2502-13930.txt.
Introduction
Peter O'Donovan and co-authors present a 2025 arXiv paper on sharper tests for quantum chaos. A process tensor captures a quantum system under repeated probing, and the authors build from it a projected ensemble: the set of pure output states across probing choices. Its statistical moments become chaos probes of rising fineness.
The first moment already bundles several known chaos measures. Higher moments reveal entanglement patterns that separate chaotic motion from orderly motion cleanly, where older entropy measures blur. Simulations across spin-chain models, from free to chaotic to localized, back the claims. The authors pitch the scheme as one frame for unitary and monitored many-body dynamics.
Catalog record
| CREATOR | Peter O’Donovan, Neil Dowling, Kavan Modi, Mark T. Mitchison |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 112 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2502-13930
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Diagnosing chaos with projected ensembles of process tensors Peter O’Donovan, Neil Dowling, Kavan Modi, Mark T. Mitchison Abstract The process tensor provides a general representation of a quantum system evolving under repeated interventions and is fundamental for numerical simulations of local many-body dynamics. In this work, we introduce the projected process ensemble, an ensemble of pure output states of a process tensor in a given basis of local interventions, and use it to define increasingly more fine-grained probes of quantum chaos. The first moment of this ensemble encapsulates numerous previously studied chaos quantifiers, including the Alicki-Fannes quantum dynamical entropy, butterfly flutter fidelity, and spatiotemporal entanglement. We discover characteristic entanglement structures within the ensemble’s higher moments that can sharply distinguish chaotic from integrable dynamics, overcoming deficiencies of the quantum dynamical and spatiotemporal entropies. These conclusions are supported by extensive numerical simulations of many-body dynamics for a range of spin-chain models, including non-interacting, interacting-integrable, chaotic, and many-body localized regimes. Our work elucidates the fingerprints of chaos on spatiotemporal correlations in quantum stochastic processes, and provides a unified framework for analyzing the complexity of unitary and monitored many-body dynamics. I Introduction Quantum chaos has become synonymous with randomness. Probes such as level spacing statistics [ 1 , 2 ] , eigenstate properties [ 3 , 4 , 5 , 6 , 7 , 8 ] , tripartite mutual information [ 9 , 10 , 11 ] , and deep thermalization [ 12 , 13 , 14 , 15 , 16 , 17 ] characterize chaos via the onset of universal random behavior by comparing with random matrices and Haar ensembles, which provide a benchmark for randomness [ 18 ] . Yet, there is no unique way of probing the randomness of a quantum system. As such, quantum chaos does not have an agreed-upon definition and its quantifiers are varied and inconsistent. Dynamical probes of chaos like the Loschmidt echo and out-of-time ordered correlator (OTOC) have also been studied extensively without relying on comparisons with randomness, but have been shown to fail in a variety of cases [ 19 , 20 , 21 , 22 , 23 ] . In contrast, classical chaos is a well-understood, deterministic phenomenon defined by the extreme sensitivity of a dynamical system to small perturbations. Intuitively, this results in complex, unpredictable time evolution whenever the phase space is coarse-grained, as it is in practice for any realistic observer. This temporal complexity is captured by the (Kolmogorov-Sinai) dynamical entropy—the Shannon entropy of a multi-time probability distribution over coarse-grained phase space—which quantifies the loss of information about a dynamical system over time [ 24 , 25 , 26 ] . The growth rate of classical dynamical entropy unambiguously distinguishes random, chaotic, and non-chaotic dynamics [ 26 ] and has even been used to study chaos in semi-classical descriptions of quantum systems [ 27 , 28 ] . Alicki and Fannes have defined a quantum version of the dynamical entropy, which reduces to the classical dynamical entropy in an appropriate limit [ 29 , 30 , 31 , 25 ] . Instead of trajectories in a coarse-grained phase space, this quantum dynamical entropy (QDE) quantifies the irreducible complexity of a sequence of measurement outcomes, minimized over all measurement bases. This sequential measurement defines a quantum stochastic process that is described mathematically by a process tensor [ 32 , 33 , 34 ] : the most general description of a quantum system under repeated interventions. The process tensor also underpins important recent results in the theory of open quantum systems, including state-of-the-art simulation methods [ 35 , 36 , 37 , 38 , 39 ] and an operational definition of quantum non-Markovianity [ 32 ] . QDE has been studied for single-particle systems [ 40 , 24 , 41 ] and, more recently, quantum many-body models [ 42 , 43 ] , with a similar quantity being studied under the guise of decoherent histories [ 44 , 45 ] . Unlike its classical counterpart, however, QDE may not unambiguously distinguish chaotic from non-chaotic dynamics in a many-body setting. Scaling of dynamical entropy with the number of measurements has been shown to have a maximal growth rate, not only for highly chaotic systems such as the Sachdev-Ye-Kitaev model but also regular dynamics including free fermions in the thermodynamic limit [ 42 ] and a class of models called Lindblad-Bernoulli shifts [ 46 , 43 ] . This can be explained as resulting from a choice of interventions that act locally on a subsystem: any information lost to the rest of the system—a feature that is not unique to chaotic dynamics—will cause dynamical entropy to grow. Motivated by these deficiencies of the QDE, we develop a framework to analyze chaos in quantum stochastic processes that captures both the unpredictability of the observed dynamics and the complexity of the conditional quantum states. Our approach is based on the pure process tensor [ 43 ] , which describes all temporal and spatial correlations of a sequentially measured quantum many-body system, as detailed in Sec. II . Because bipartite entanglement in a pure state is equivalent to the entropy of the subsystems, the QDE is nothing but the entanglement entropy between the temporal and spatial parts of this pure process tensor, as discussed in Refs. [ 40 , 42 , 43 ] . In order to capture the complexity of the output state, a natural next step is to consider entanglement across other bipartitions of the process that mix its spatial and temporal aspects: this spatiotemporal entanglement (STE) was introduced in Ref. [ 43 ] . It was argued there, that by keeping some of the final (spatial) state together with the temporal parts while tracing the rest, STE should also be sensitive to how correlations spread throughout the final state, i.e. the information scrambling of the dynamics. However, neither QDE nor STE have been studied in detail for short-ranged lattice models, where rich phenomenology can emerge from the competition between inertia, interactions, and disorder. Our primary contributions in this work are twofold. First, we perform an extensive numerical study of the pure process tensor generated by local interventions on paradigmatic spin-chain models, spanning non-interacting, interacting-integrable, chaotic, and many-body localized regimes. In all cases, we use exact diagonalization to perform time evolution; see Sec. III for details of the models considered. Our work is the first in silico experiment to systematically investigate how a quantum stochastic process is impacted by the presence or absence of ergodicity in the underlying Hamiltonian dynamics. As an auxiliary technical contribution, we show how to exploit symmetries to reach large system sizes in exact diagonalization studies of quantum stochastic processes. In particular, we focus on models that conserve particle number and use a restricted basis of interventions to construct a process tensor within a single \mathrm{U}(1) symmetry sector. Our construction follows similar principles to those presented in Ref. [ 47 ] , and may be of use in pushing the boundaries of other numerical techniques based on the process tensor framework [ 35 , 36 , 37 , 38 , 39 ] . Very recently, Ref. [ 48 ] also considered symmetry restrictions on the QDE, albeit for single-body quantum dynamics. Our results