arXiv:2609.28441
Analog neutral-atom for in-memory processing in quantum reservoir computing
Luca Nigro, Gian Luca Giorgi, Enrico Prati, Roberta Zambrini
Abstract
Quantum Reservoir Computing (QRC) exploits the rich dynamics of quantum many-body systems to process time-dependent information with high-dimensional state spaces. While neutral atom arrays offer a scalable platform for this paradigm, realizing intrinsic temporal memory without relying on external classical buffering requires precise control over the relaxation dynamics of the system. We address the necessity of specific non-unitary dynamics for enforcing fading memory while overcoming the lack of separability. This study simulates and empirically identifies the specific dynamical regimes required to maximize the computational capacity of the reservoir, addressing the amount of dissipation, the effectiveness of lying at the edge of quantum chaos, and that of time-multiplexing compared to the scaling of the physical size of the system. Here we show that controlled dissipation is strictly necessary to induce echo state property, fading memory, and separability, in neutral atom arrays. Computational performance increases at the edge of quantum chaos. The counter-intuitive residual memory provided by the unital phase damping channel is justified by introducing a model based on pure outputs. The consequent implementation enables the solution of complex non-linear tasks, such as the Mackey-Glass time series, with as few as $N=5$ atoms. We establish a rigorous framework for engineering dissipative quantum reservoirs required for intrinsic temporal processing on near-term quantum devices.
AI Summary
The paper asks how neutral-atom arrays can retain temporal information intrinsically for quantum reservoir computing, rather than through classical input windows. It simulates globally driven, one-dimensional Rydberg arrays using an input-dependent Hamiltonian and local Markovian Lindblad dissipation. Single-site observables and two-body correlators supply features for a ridge-regression readout. Trace-distance tests assess initial-state forgetting and sensitivity to input histories; short-term-memory benchmarks compare dissipation channels, Hamiltonian parameters, encoding schemes, array size, and time multiplexing. Amplitude damping gives the highest memory capacity among the tested channels. For $N=4$ and $\gamma=0.4\,\mu\mathrm{s}^{-1}$, amplitude- and phase-damped reservoirs forget initial-state differences to below $0.1\%$ after approximately $20\,\mu\mathrm{s}$. Phase damping supports only transient memory: transverse driving restores mixing, but the asymptotic state remains input-independent and maximally mixed. A single-qubit pure-output classification and Liouvillian analysis support this interpretation. Memory improves at a parameter point identified spectrally as an edge-of-chaos regime. Detuning encoding outperforms Rabi-amplitude encoding, while time-multiplexing gains saturate around five virtual nodes in the tested configuration. Benchmarks include NARMA, Mackey–Glass, and Santa Fe forecasting. Mackey–Glass one-step test SMSE reaches approximately $10^{-5}$ with four atoms; a five-atom reservoir maintains SMSE below $0.1$ through a 20-step prediction horizon. On Santa Fe, dissipative operation without buffering achieves $0.23\pm0.05$, comparable to a unitary-reset baseline using seven-step windows. The results are simulations. Dissipation is phenomenological rather than microscopically derived for interacting atoms, and the single-qubit channel classification is explicitly used only as a heuristic guide to many-body dynamics.
Central claim may not hold: The claimed rigorous reservoir framework is supported by autonomous single-qubit results and sampled numerical tests, not general guarantees for the input-driven many-body dynamics. The driven amplitude-damping reservoir cannot have the pure, input-dependent stationary state invoked in the Results discussion, because its only possible pure stationary state is the all-ground state and the transverse drive makes that state nonstationary. The edge-of-chaos comparison changes several parameters simultaneously and therefore does not establish the asserted causal mechanism or optimality. The main text and appendix specify detuning half-ranges differing by a factor of approximately 9.425, leaving the simulated operating range ambiguous. The phase-damping spectral matrix uses transverse decay rates of γ although the stated master equation gives 2γ.
Written by gpt-6.1-sol from the paper's full text. Automated: it can be wrong.
- Comments
- 17 pages, 8 figures