Quantum Reservoir Computing (QRC) uses fixed quantum dynamics as a high-dimensional feature map. Still, standard pipelines flatten reservoir measurements and train a linear readout, ignoring the natural tensor structure across time steps, qubits, and observables. We propose a multilinear readout for quantum reservoir computing (ML-QRC), a tensor-structured neural readout compressed through mode-wise projections. Relative to a dense hidden-width-64 neural readout, this reduces the feature-to-hidden parameter count from $64TnF$ to $d(T+n+F)+64d^3$, giving a $12.1\times$ feature-block compression at $T=50$, $n=6$, $d=8$, and $F=n+\binom{n}{2}$. Across Mackey-Glass forecasting and TFIM phase classification, ML-QRC remains statistically competitive with Ridge on Mackey-Glass and obtains the numerically highest full-data TFIM accuracy among tested methods, while using $12.0\times$ fewer parameters than the full dense MLP. Its clearest benefit is improved low-data stability relative to neural MLP readouts on Mackey-Glass; this advantage does not transfer to TFIM, where Ridge performs best at 50\% data. Overall, ML-QRC is best viewed as a tensor-structured neural readout that compresses dense MLP readouts while preserving QRC tensor structure, rather than as a replacement for simple linear/Ridge readouts in the tested regime.
AI Summary
The paper asks whether tensor-structured readouts can compress neural quantum reservoir computing (QRC) readouts without losing predictive performance. It applies learned Tucker-style projections separately to the time, qubit, and observable axes of a reservoir-measurement tensor, followed by a width-64 ReLU network. For equal projected dimensions $d$, the feature-to-hidden parameter count becomes $d(T+n+F)+64d^3$, versus $64TnF$ for a dense readout.
Numerical experiments cover Mackey–Glass forecasting and transverse-field Ising phase classification, primarily using six-qubit simulated reservoirs, $T=50$, and 21 single- and two-qubit $Z$ observables, with analytical depolarizing noise. Comparisons include linear/Ridge readouts, dense and parameter-matched PCA-based MLPs, and fixed tensor compression. At $d=8$, the proposed network has 33,513 parameters versus 403,329 for the dense MLP, a $12.0\times$ reduction. Across ten seeds, full-data forecasting NMSE is $0.0086\pm0.0013$, comparable to Ridge’s $0.0085$ but above the parameter-matched MLP’s $0.0058\pm0.0011$. Full-data classification accuracy is $70.5\pm1.5\%$, overlapping competing neural readouts.
With 10% forecasting data, the tensor readout outperforms the tested MLPs but remains worse than linear/Ridge models; this neural-readout advantage does not transfer to reduced-data classification. Increasing projection dimension improves full-data performance but introduces cubic parameter growth, confounding rank and model-size effects. Stated limitations include untuned low-data MLP early stopping, split-dependent Ridge results, weaker scaling than the parameter-matched MLP, and the absence of finite-shot or hardware validation.
Some signs of AI-generated textIncrementalSmall numerics onlyPossible error
Central claim may not hold: The claimed 12.1× feature-block compression depends on an unexplained extra qubit axis: the explicitly listed 21 global observables supply 1,050 measurements over 50 steps, not the assumed 6,300.
The specified PCA-512 → 64 → 1 MLP-S contains 32,897 trainable parameters rather than the reported 33,537, although it remains approximately size-matched to ML-QRC-M.
The main-results text incorrectly identifies Ridge as best on full-data Mackey–Glass, whereas Table I reports lower NMSE for both MLP and MLP-S.
Written by gpt-6.1-sol from the paper's full text. Automated: it can be wrong.
Quantum Reservoir Computing (QRC) uses fixed quantum dynamics as a high-dimensional feature map. Still, standard pipelines flatten reservoir measurements and train a linear readout, ignoring the natural tensor structure across time steps, qubits, and observables. We propose a multilinear readout for quantum reservoir computing (ML-QRC), a tensor-structured neural readout compressed through mode-wise projections. Relative to a dense hidden-width-64 neural readout, this reduces the feature-to-hidden parameter count from 64TnF to d(T+n+F)+64d3, giving a 12.1× feature-block compression at T=50, n=6, d=8, and F=n+(2n). Across Mackey-Glass forecasting and TFIM phase classification, ML-QRC remains statistically competitive with Ridge on Mackey-Glass and obtains the numerically highest full-data TFIM accuracy among tested methods, while using 12.0× fewer parameters than the full dense MLP. Its clearest benefit is improved low-data stability relative to neural MLP readouts on Mackey-Glass; this advantage does not transfer to TFIM, where Ridge performs best at 50% data. Overall, ML-QRC is best viewed as a tensor-structured neural readout that compresses dense MLP readouts while preserving QRC tensor structure, rather than as a replacement for simple linear/Ridge readouts in the tested regime.
The paper asks whether tensor-structured readouts can compress neural quantum reservoir computing (QRC) readouts without losing predictive performance. It applies learned Tucker-style projections separately to the time, qubit, and observable axes of a reservoir-measurement tensor, followed by a width-64 ReLU network. For equal projected dimensions d, the feature-to-hidden parameter count becomes d(T+n+F)+64d3, versus 64TnF for a dense readout.
Numerical experiments cover Mackey–Glass forecasting and transverse-field Ising phase classification, primarily using six-qubit simulated reservoirs, T=50, and 21 single- and two-qubit Z observables, with analytical depolarizing noise. Comparisons include linear/Ridge readouts, dense and parameter-matched PCA-based MLPs, and fixed tensor compression. At d=8, the proposed network has 33,513 parameters versus 403,329 for the dense MLP, a 12.0× reduction. Across ten seeds, full-data forecasting NMSE is 0.0086±0.0013, comparable to Ridge’s 0.0085 but above the parameter-matched MLP’s 0.0058±0.0011. Full-data classification accuracy is 70.5±1.5%, overlapping competing neural readouts.
With 10% forecasting data, the tensor readout outperforms the tested MLPs but remains worse than linear/Ridge models; this neural-readout advantage does not transfer to reduced-data classification. Increasing projection dimension improves full-data performance but introduces cubic parameter growth, confounding rank and model-size effects. Stated limitations include untuned low-data MLP early stopping, split-dependent Ridge results, weaker scaling than the parameter-matched MLP, and the absence of finite-shot or hardware validation.
Quantum reservoir computing (QRC) offers a train-light approach to temporal learning by keeping the quantum dynamics fixed and training only a classical readout. Most QRC studies use linear readouts, potentially leaving nonlinear structure in measured reservoir states under-exploited. Inspired by recent reservoir-computing Kolmogorov–Arnold networks (RCKAN), we study Kolmogorov–Arnold Network (KAN)-style spline readouts for stateful QRC. Our QRCKAN architecture combines a refreshed-input, persistent-memory density-matrix reservoir with correlation observables, a shallow spline KAN readout, an optional current-state skip, and a residual formulation that adds a validation-weighted nonlinear correction to ridge regression. We summarize five seeded runs on Lorenz-63 and Mackey–Glass against matched linear QRC, stateless quantum feature maps, no-QRC controls, echo-state networks (ESNs), lag-based regressors, LSTM, and Transformer baselines. On Lorenz-63, residual QRCKAN reduces mean one-step NRMSE by 51.6% relative to QRC+Linear+Skip (0.000820 vs. 0.001696) and raises mean valid prediction time (VPT) from 102 to 218 steps. On Mackey–Glass, QRCKAN reduces mean one-step NRMSE by 78.2% and rollout NRMSE by 81.0%, while increasing mean VPT from 118 to 262 steps. Matched no-skip QRC–QFF comparisons and matched no-QRC controls support contributions from temporal state and reservoir features under the tested configurations. Because seed 0 informed architecture selection and is included in these summaries, all intervals and tests are exploratory rather than independent confirmatory inference. Strong ESNs remain superior in one-step accuracy. We claim an exploratory nonlinear-readout improvement within simulated QRC, not a general quantum advantage.
Quantum Reservoir Computing (QRC) exploits quantum dynamical systems as high-dimensional reservoirs for temporal sequence modeling. This work investigates QRC based on Rydberg atom arrays and evaluates its performance on both synthetic and real-world datasets. We analyze the effects of lattice geometry, readout mechanisms, and measurement shot count on predictive accuracy. Our results show that QRC achieves performance comparable to, and in some cases exceeding, classical Reservoir Computing (RC), particularly for tasks involving complex temporal dependencies. Reservoir geometry plays a size-dependent role: honeycomb lattices yield higher accuracy at small reservoir sizes, while square lattices scale more effectively as the number of atoms increases. Among readout strategies, Support Vector Regression consistently outperforms Ridge Regression and Neural Networks in extracting nonlinear structure from reservoir states. Increasing the number of measurement shots improves accuracy, with diminishing returns beyond a saturation threshold. Overall, this study delineates the conditions under which QRC offers advantages over classical RC and identifies regimes where classical approaches remain competitive. These findings provide practical guidance for optimizing QRC architectures and contribute to the development of quantum-enhanced time-series modeling.
Short-term load forecasting is essential for reliable energy management, but practical deployment on edge devices requires models that remain accurate under limited memory, finite measurement budgets, and hardware noise. This work proposes a hardware-efficient Quantum Reservoir Computing (QRC) framework for energy load forecasting, where a fixed quantum reservoir transforms temporal input windows into high-dimensional features and only a classical Elastic Net readout is trained. To reduce deployment cost, the trained readout is compressed using post-training fixed-point quantization at bit widths from 8 to 2 bits. The framework is evaluated on the Tetouan and Spain energy load datasets under exact statevector simulation, 512-shot finite sampling, and realistic hardware-noise models from IBM FakeTorino and IBM FakeMarrakesh. Results show that 6-bit readout precision preserves full-precision forecasting performance while reducing readout memory by 81.2%. Below this point, degradation becomes dataset dependent, with Tetouan showing stronger sensitivity and Spain degrading more gradually. Hardware-noise validation further shows that the trained readout transfers to noisy reservoir states without retraining. These findings support quantized QRC as a resource-aware forecasting approach for near-term quantum time-series applications.
Quantum reservoir computing (QRC) uses fixed quantum dynamics as a high-dimensional temporal feature map and trains only a lightweight classical readout. QRC is attractive for near-term quantum machine learning, but its performance depends strongly on architecture choices such as input encoding, reservoir depth, entanglement topology, measurement features, state-reset policy, feature construction, and readout regularization. We introduce \method, a simulator-based benchmark that formulates QRC design as constrained black-box architecture search and evaluates whether large language models can act as proposal controllers for this search problem. The benchmark compares five policies under identical evaluation budgets: random search, evolutionary search, Bayesian/TPE optimization, a feedback-based LLM agent, and \hybrid, which combines LLM proposals with memory, mutation, crossover, duplicate avoidance, and exploration. On NARMA10, Mackey-Glass forecasting, and temporal parity, \hybrid{} is the most consistent policy: it ranks first on NARMA10 and temporal parity and second on Mackey-Glass, narrowly behind evolutionary search. Under a 25-evaluation budget and three seeds, \hybrid{} improves over random search on all tasks, including a 23.6\% relative reduction in Mackey-Glass error. The results do not show that LLMs are universal QRC optimizers; rather, they show that generative models can be useful high-level controllers when embedded inside validated, reproducible hybrid search loops.
Due to rising electricity demand, accurate short-term load forecasting is increasingly important for grid stability and efficient energy management, particularly in resource-constrained edge settings. We present a hardware-efficient Quantum Reservoir Computing (QRC) framework based on a fixed, untrained quantum circuit with Chebyshev feature encoding, brickwork entanglement, and single- and two-qubit Pauli measurements, avoiding quantum backpropagation entirely. Using the Tetouan City Power Consumption dataset, we examine the effect of post-training fixed-point quantization on the classical readout layer, with the reservoir architecture selected through a genetic search over 18 candidate configurations. Under finite-shot evaluation, 8-bit and 6-bit quantization maintain forecasting accuracy within 1% of the FP32 baseline while reducing readout memory by 75% and 81%, respectively. These results suggest that quantized readout can improve the hardware efficiency and deployment practicality of QRC for memory-constrained energy forecasting.
Forecasting neural activity from short recordings remains a fundamental challenge. Reservoir computing may offer an efficient paradigm for temporal prediction, however classical reservoirs typically underperform in small-data regimes. Here we investigate whether quantum reservoir computing (QRC) can help overcome this limitation. Building on recent advances, we introduce a quantum reservoir based on a transverse-field Ising model, combined with heterogeneous quantum measurements and polynomial ridge regression. On a standard benchmark task, results show that the quantum reservoir outperforms a classical counterpart overall, with prediction accuracy strongly dependent on reservoir parameters. We further demonstrate feasibility by running the same task on quantum hardware. To assess performance on biological signals, we evaluate QRC on simulated human electroencephalography (EEG) data with a parallel reservoir architecture. On this challenging task, the tested quantum reservoir did not match the performance of the classical one, but it produced stable, convergent predictions. This is a meaningful first step toward forecasting of biologically realistic neural data using a quantum reservoir. Overall, our findings indicate that although current quantum hardware and parallel reservoir architectures do not yet surpass classical methods on complex neural signals, QRC can be executed on near-term devices and does converge with realistic EEG-like data. This work establishes a practical baseline for future algorithmic and hardware developments aimed at clinical time-series forecasting with quantum systems.