Challenges
Datasets
Workspaces
Papers
Discussions
Leaderboard
Log inSign up
Challenges
Datasets
Workspaces
Papers
Discussions
Leaderboard
Blog
Job Board
Q3AS

© 2026 Aqora Quantum S.A.S.

TermsPrivacyLegal Notice

arXiv:2607.00235

Learning Low-Energy Subspace Overlaps in Many-Body Systems with Measurement-Based and Coherent Quantum Strategies

Shamminuj Aktar, Rishabh Bhardwaj, Tanmoy Bhattacharya, Stephan Eidenbenz

quant-phSubmitted Jun 30, 2026• Announced Jul 2, 2026• 0 citations
arXivHTMLPDF

Abstract

Predicting the overlap of quantum states with specified low-energy subspaces is a key diagnostic for quantum many-body dynamics, with direct applications in state preparation, subspace-based algorithms, and the study of thermalization. We study the supervised prediction of subspace overlaps O_K between time-evolved states and K-dimensional low-energy eigenspaces of a 10-qubit Heisenberg spin chain following a local perturbation. We compare two quantum information extraction strategies: measurement-based learning, in which classical shadow features are processed by convolutional neural networks, and coherent quantum learning, in which quantum convolutional neural networks process the state directly. We further introduce physics-informed variants for both approaches, including Hamiltonian-aware shadows and QCNN gates aligned with the Heisenberg exchange structure. Across five dataset configurations spanning weak, moderate, and strong quench regimes, physics-informed QCNNs achieve stable performance, with mean test-set coefficients of determination R^2 = 0.753-0.846. Shadow-based methods show stronger regime dependence: they outperform QCNNs in the moderate-quench regime, reaching R^2 = 0.886, but underperform in weak and strong quenches at default shot budgets, where the best shadow results are R^2 = 0.615 and 0.672, respectively. Hardware validation on Quantinuum and IBM noise models shows that arbitrary state preparation is the dominant limitation, requiring approximately 2,044 two-qubit gates and causing near-complete depolarization before inference. These results identify a regime-dependent tradeoff between measurement-based and coherent quantum learning, with shadow methods excelling when the target remains locally accessible and physics-informed QCNNs providing more robust performance across dynamical regimes.

Related papers

Uncovering quantum many-body scars with quantum machine learning

Jia-Jin Feng, Bingzhi Zhang, Zhi-Cheng Yang +1 more

Quantum many-body scars are rare eigenstates hidden within the chaotic spectra of many-body systems, representing a weak violation of the eigenstate thermalization hypothesis (ETH). Identifying these scars, as well as other non-thermal states in complex quantum systems, remains a significant challenge. Besides exact scar states, the nature of other non-thermal states lacking simple analytical characterization remains an open question. In this study, we employ tools from quantum machine learning—specifically, (enhanced) quantum convolutional neural networks (QCNNs), to explore hidden non-thermal states in chaotic many-body systems. Our simulations demonstrate that QCNNs achieve over 99% single-shot measurement accuracy in identifying all known scars. Furthermore, we successfully identify new non-thermal states in models such as the xorX model, the PXP model, and the far-coupling Su-Schrieffer-Heeger model. In the xorX model, some of these non-thermal states can be approximately described as spin-wave modes of specific quasiparticles. We further develop effective tight-binding Hamiltonians within the quasiparticle subspace to capture key features of these many-body eigenstates. Finally, we validate the performance of QCNNs on IBM quantum devices, achieving single-shot measurement accuracy exceeding 63% under real-world noise and errors, with the aid of error mitigation techniques. Our results underscore the potential of QCNNs to uncover hidden non-thermal states in quantum many-body systems.

arXiv:2409.07405•npj Quantum Information•Mar 11, 2025•10 citations

Quantum Convolutional Neural Networks are Effectively Classically Simulable

Pablo Bermejo, Paolo Braccia, Manuel S. Rudolph +3 more

Quantum convolutional neural networks (QCNNs) are widely regarded as a promising model for quantum machine learning (QML). In this work, we analyze the most widely used variants of these models (i.e., tracing out- and measurement-based QCNNs), and we relate their heuristic success to two facts. First, that when randomly initialized, they can only operate on the information encoded in low-bodyness measurements of their input states. And second, that they are commonly benchmarked on “locally easy” datasets whose states are precisely classifiable by the information encoded in these low-bodyness observables subspace. From these insights, we argue that the QCNN’s action on this subspace should be efficiently classically simulable. Indeed, we construct and train a purely classical QCNN surrogate—based on low-bodyness Pauli propagation, tensor networks, and classical shadows—that matches or outperforms standard QCNNs on all benchmark datasets and on up-to 1024 qubits, thereby empirically realizing our simulability claims. Our results can then be understood as highlighting a deeper symptom of QML: Models could only be showing heuristic success because they are benchmarked on simple problems, for which their action can be classically simulated. This insight points to the fact that nontrivial datasets are a truly necessary ingredient for moving forward with QML. To finish, we discuss how our results can be extrapolated to classically simulate other architectures.

arXiv:2408.12739•PRX Quantum•Feb 19, 2026•133 citations

Learning ground state observables from quantum computing experiments

Ben Jaderberg, Freya Shah, Minjun Jeon +3 more

Recent theoretical progress has established conditions under which machine learning models can efficiently predict ground-state properties of gapped local Hamiltonians when trained on quantum-generated data. Previous experimental demonstrations in this paradigm, however, have largely been limited to small systems or highly structured states, due to the difficulty of preparing many-body ground states on quantum processors. In this work, we demonstrate learning from experimental quantum data generated from approximate ground states of the two-dimensional Heisenberg XXZ model with system sizes up to 115 qubits. We construct a dataset of single-site expectation values, two-point correlations, and 12-body loop correlations across the antiferromagnetic phase. We then train neural networks on this data and show that they can accurately predict spatially resolved observables for previously unseen Hamiltonian parameters, both within the training distribution and in an out-of-distribution regime approaching the phase boundary. Our results demonstrate the practical realization of learning from quantum data for an interacting two-dimensional many-body system at scale, motivating a path toward regimes where quantum processors could provide training data beyond the reach of classical approximation methods.

quant-pharXiv:2606.15983•Jun 16, 2026•1 citation

Quantum Convolutional Neural Networks for Groundwater Heat Plume Prediction: A Surrogate Modeling Approach

Danyal Maheshwari, Julia Pelzer, Miriam Schulte

Quantum machine learning methods are increasingly explored for modeling complex environmental systems, including groundwater heat plume dynamics. In this work, we explore a Quantum Convolutional Neural Network (QCNN) as a surrogate model for predicting temperature variations in groundwater induced by geothermal heat pumps in the city of Munich. To comply with the scalability constraints of current quantum hardware, the original high-dimensional simulation output is reduced to a compact set of representative parameters that serve as training targets for the surrogate. The proposed QCNN architecture consists of a quantum convolutional layer, a quantum pooling layer, and a fully connected quantum readout stage. Convolution and pooling operations are realized via parameterized quantum circuits based on rotational gates and measurement-driven decoding, while a Hamiltonian-inspired feature-encoding scheme is used to prepare informative input states on the quantum device. We evaluate the QCNN across multiple execution backends, including an ideal statevector simulator, a noisy simulator, IBM's 127-qubit Kyiv quantum processor, and the same hardware augmented with advanced error-mitigation techniques. Realistic noise models are employed to approximate device behavior and to assess the impact of mitigation strategies. Model performance is benchmarked using mean squared error (MSE) on both training and testing sets. The results show that, although classical neural networks still achieve the highest predictive accuracy, the QCNN attains competitive and consistent performance on simulators and exhibits noticeable improvement under error-mitigated hardware conditions. These findings indicate that quantum-enhanced surrogate modeling is a promising direction for future groundwater temperature prediction as quantum hardware and error-mitigation techniques continue to mature.

quant-pharXiv:2606.23411•Jun 23, 2026•0 citations

Algorithmic Shadow Spectroscopy

Hans Hon Sang Chan, Richard Meister, Matthew L. Goh +1 more

We present shadow spectroscopy as a simulator-agnostic quantum algorithm for estimating energy gaps using very few circuit repetitions (shots) and no extra resources (ancilla qubits) beyond performing time evolution and measurements. The approach builds on the fundamental feature that every observable property of a quantum system must evolve according to the same harmonic components: we can reveal them by postprocessing classical shadows of time-evolved quantum states to extract a large number of time-periodic signals N o ∝ 10 8 , whose frequencies correspond to Hamiltonian energy differences with precision limited as ϵ ∝ 1 / T for simulation time T . We provide strong analytical guarantees that (a) quantum resources scale as O ( log  N o ) , while the classical computational complexity is linear O ( N o ) , (b) the signal-to-noise ratio increases with the number of processed signals as ∝ N o , and (c) spectral peak positions are immune to reasonable levels of noise. We demonstrate our approach on model spin systems and the excited-state conical intersection of molecular CH 2 and verify that our method is indeed intuitively easy to use in practice, robust against gate noise, amiable to a new type of algorithmic-error mitigation technique, and uses relatively few shots given a reasonable initial state is supplied—we demonstrate that even 10 shots per time step can be sufficient. Finally, we measured a high-quality, experimental shadow spectrum of a spin chain on readily available IBM quantum computers, achieving the same precision as in noise-free simulations without using any advanced error mitigation, and verified scalability in tensor-network simulations of up to 100-qubit systems.

arXiv:2212.11036•PRX Quantum•Mar 17, 2025•43 citations

Quantum Attention by Overlap Interference: Predicting Classical and Many-Body Quantum Sequences

Alessio Pecilli, Matteo Rosati

We propose a variational quantum implementation of self-attention (QSA)-the core operation in transformers and large language models-which predicts future elements of a sequence by forming overlap-weighted combinations of past data. At variance with previous approaches, our QSA realizes the required nonlinearity through interference of state overlaps and a degree-$k$ polynomial kernel, and estimates a loss based on Rényi-$1/2$ entropic functionals via two observables' expectation values, avoiding the decoding of amplitude-encoded predictions into classical probabilities. QSA also accommodates a constrained, trainable data-embedding tying state overlaps to data-level similarities. Its dominant end-to-end training complexity scales as $O\left(μ^{-1}k^2Td\right)$, versus $O\left(T d^{k+1}\right)$ of the fairest classical comparison, with $μ$ a training signal; we show numerically that this allows a complexity advantage in the regime where sequence length $T$ dominates the embedding size $d$. In simulations, our QSA-based quantum transformer learns sequence prediction on classical data and on many-body transverse-field Ising trajectories-establishing trainable attention as a practical primitive for quantum dynamical modeling.

quant-pharXiv:2602.06699•Feb 9, 2026•0 citations
Predicting the overlap of quantum states with specified low-energy subspaces is a key diagnostic for quantum many-body dynamics, with direct applications in state preparation, subspace-based algorithms, and the study of thermalization. We study the supervised prediction of subspace overlaps O_K between time-evolved states and K-dimensional low-energy eigenspaces of a 10-qubit Heisenberg spin chain following a local perturbation. We compare two quantum information extraction strategies: measurement-based learning, in which classical shadow features are processed by convolutional neural networks, and coherent quantum learning, in which quantum convolutional neural networks process the state directly. We further introduce physics-informed variants for both approaches, including Hamiltonian-aware shadows and QCNN gates aligned with the Heisenberg exchange structure. Across five dataset configurations spanning weak, moderate, and strong quench regimes, physics-informed QCNNs achieve stable performance, with mean test-set coefficients of determination R^2 = 0.753-0.846. Shadow-based methods show stronger regime dependence: they outperform QCNNs in the moderate-quench regime, reaching R^2 = 0.886, but underperform in weak and strong quenches at default shot budgets, where the best shadow results are R^2 = 0.615 and 0.672, respectively. Hardware validation on Quantinuum and IBM noise models shows that arbitrary state preparation is the dominant limitation, requiring approximately 2,044 two-qubit gates and causing near-complete depolarization before inference. These results identify a regime-dependent tradeoff between measurement-based and coherent quantum learning, with shadow methods excelling when the target remains locally accessible and physics-informed QCNNs providing more robust performance across dynamical regimes.
Quantum many-body scars are rare eigenstates hidden within the chaotic spectra of many-body systems, representing a weak violation of the eigenstate thermalization hypothesis (ETH). Identifying these scars, as well as other non-thermal states in complex quantum systems, remains a significant challenge. Besides exact scar states, the nature of other non-thermal states lacking simple analytical characterization remains an open question. In this study, we employ tools from quantum machine learning—specifically, (enhanced) quantum convolutional neural networks (QCNNs), to explore hidden non-thermal states in chaotic many-body systems. Our simulations demonstrate that QCNNs achieve over 99% single-shot measurement accuracy in identifying all known scars. Furthermore, we successfully identify new non-thermal states in models such as the xorX model, the PXP model, and the far-coupling Su-Schrieffer-Heeger model. In the xorX model, some of these non-thermal states can be approximately described as spin-wave modes of specific quasiparticles. We further develop effective tight-binding Hamiltonians within the quasiparticle subspace to capture key features of these many-body eigenstates. Finally, we validate the performance of QCNNs on IBM quantum devices, achieving single-shot measurement accuracy exceeding 63% under real-world noise and errors, with the aid of error mitigation techniques. Our results underscore the potential of QCNNs to uncover hidden non-thermal states in quantum many-body systems.
Quantum convolutional neural networks (QCNNs) are widely regarded as a promising model for quantum machine learning (QML). In this work, we analyze the most widely used variants of these models (i.e., tracing out- and measurement-based QCNNs), and we relate their heuristic success to two facts. First, that when randomly initialized, they can only operate on the information encoded in low-bodyness measurements of their input states. And second, that they are commonly benchmarked on “locally easy” datasets whose states are precisely classifiable by the information encoded in these low-bodyness observables subspace. From these insights, we argue that the QCNN’s action on this subspace should be efficiently classically simulable. Indeed, we construct and train a purely classical QCNN surrogate—based on low-bodyness Pauli propagation, tensor networks, and classical shadows—that matches or outperforms standard QCNNs on all benchmark datasets and on up-to 1024 qubits, thereby empirically realizing our simulability claims. Our results can then be understood as highlighting a deeper symptom of QML: Models could only be showing heuristic success because they are benchmarked on simple problems, for which their action can be classically simulated. This insight points to the fact that nontrivial datasets are a truly necessary ingredient for moving forward with QML. To finish, we discuss how our results can be extrapolated to classically simulate other architectures.
Recent theoretical progress has established conditions under which machine learning models can efficiently predict ground-state properties of gapped local Hamiltonians when trained on quantum-generated data. Previous experimental demonstrations in this paradigm, however, have largely been limited to small systems or highly structured states, due to the difficulty of preparing many-body ground states on quantum processors. In this work, we demonstrate learning from experimental quantum data generated from approximate ground states of the two-dimensional Heisenberg XXZ model with system sizes up to 115 qubits. We construct a dataset of single-site expectation values, two-point correlations, and 12-body loop correlations across the antiferromagnetic phase. We then train neural networks on this data and show that they can accurately predict spatially resolved observables for previously unseen Hamiltonian parameters, both within the training distribution and in an out-of-distribution regime approaching the phase boundary. Our results demonstrate the practical realization of learning from quantum data for an interacting two-dimensional many-body system at scale, motivating a path toward regimes where quantum processors could provide training data beyond the reach of classical approximation methods.
Quantum machine learning methods are increasingly explored for modeling complex environmental systems, including groundwater heat plume dynamics. In this work, we explore a Quantum Convolutional Neural Network (QCNN) as a surrogate model for predicting temperature variations in groundwater induced by geothermal heat pumps in the city of Munich. To comply with the scalability constraints of current quantum hardware, the original high-dimensional simulation output is reduced to a compact set of representative parameters that serve as training targets for the surrogate. The proposed QCNN architecture consists of a quantum convolutional layer, a quantum pooling layer, and a fully connected quantum readout stage. Convolution and pooling operations are realized via parameterized quantum circuits based on rotational gates and measurement-driven decoding, while a Hamiltonian-inspired feature-encoding scheme is used to prepare informative input states on the quantum device. We evaluate the QCNN across multiple execution backends, including an ideal statevector simulator, a noisy simulator, IBM's 127-qubit Kyiv quantum processor, and the same hardware augmented with advanced error-mitigation techniques. Realistic noise models are employed to approximate device behavior and to assess the impact of mitigation strategies. Model performance is benchmarked using mean squared error (MSE) on both training and testing sets. The results show that, although classical neural networks still achieve the highest predictive accuracy, the QCNN attains competitive and consistent performance on simulators and exhibits noticeable improvement under error-mitigated hardware conditions. These findings indicate that quantum-enhanced surrogate modeling is a promising direction for future groundwater temperature prediction as quantum hardware and error-mitigation techniques continue to mature.
We present shadow spectroscopy as a simulator-agnostic quantum algorithm for estimating energy gaps using very few circuit repetitions (shots) and no extra resources (ancilla qubits) beyond performing time evolution and measurements. The approach builds on the fundamental feature that every observable property of a quantum system must evolve according to the same harmonic components: we can reveal them by postprocessing classical shadows of time-evolved quantum states to extract a large number of time-periodic signals N o ∝ 10 8 , whose frequencies correspond to Hamiltonian energy differences with precision limited as ϵ ∝ 1 / T for simulation time T . We provide strong analytical guarantees that (a) quantum resources scale as O ( log  N o ) , while the classical computational complexity is linear O ( N o ) , (b) the signal-to-noise ratio increases with the number of processed signals as ∝ N o , and (c) spectral peak positions are immune to reasonable levels of noise. We demonstrate our approach on model spin systems and the excited-state conical intersection of molecular CH 2 and verify that our method is indeed intuitively easy to use in practice, robust against gate noise, amiable to a new type of algorithmic-error mitigation technique, and uses relatively few shots given a reasonable initial state is supplied—we demonstrate that even 10 shots per time step can be sufficient. Finally, we measured a high-quality, experimental shadow spectrum of a spin chain on readily available IBM quantum computers, achieving the same precision as in noise-free simulations without using any advanced error mitigation, and verified scalability in tensor-network simulations of up to 100-qubit systems.
We propose a variational quantum implementation of self-attention (QSA)-the core operation in transformers and large language models-which predicts future elements of a sequence by forming overlap-weighted combinations of past data. At variance with previous approaches, our QSA realizes the required nonlinearity through interference of state overlaps and a degree-kkk polynomial kernel, and estimates a loss based on Rényi-1/21/21/2 entropic functionals via two observables' expectation values, avoiding the decoding of amplitude-encoded predictions into classical probabilities. QSA also accommodates a constrained, trainable data-embedding tying state overlaps to data-level similarities. Its dominant end-to-end training complexity scales as O(μ−1k2Td)O\left(μ^{-1}k^2Td\right)O(μ−1k2Td), versus O(Tdk+1)O\left(T d^{k+1}\right)O(Tdk+1) of the fairest classical comparison, with μμμ a training signal; we show numerically that this allows a complexity advantage in the regime where sequence length TTT dominates the embedding size ddd. In simulations, our QSA-based quantum transformer learns sequence prediction on classical data and on many-body transverse-field Ising trajectories-establishing trainable attention as a practical primitive for quantum dynamical modeling.