Pith. sign in

REVIEW 4 major objections 4 minor 1 cited by

Sequence-Model-Guided Measurement Selection for Quantum State Learning

T0 review · 4 major / 4 minor · reviewed 2026-08-06 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read A transformer sequence model learns which quantum measurements to take next, beating random selection across property prediction, phase classification, and tomography.

desk verdict TGMS is a plausible adaptive measurement-selection framework with an intriguing boundary-finding result, but the 'consistently outperform' claim currently rests on noiseless simulations with only random-sampling as a baseline. read the letter →

arxiv 2507.09891 v1 pith:6ZVBVNVL submitted 2025-07-14 quant-ph cs.AIcs.LG

Jiaxin Huang , Yan Zhu , Giulio Chiribella , Ya-Dong Wu This is my paper
classification quant-phcs.AIcs.LG
keywords quantumstatelearningadaptivemeasurementselectiontransformersequencemodelPOVMmeasurementstomographytopologicalphasessymmetry-protectedordermachineformany-bodyphysics
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper claims that the question 'which measurement should I perform next on an unknown quantum state?' can be answered by a trained transformer sequence model, and that doing so beats the standard fallback of random measurement selection. The proposed model, TGMS, adaptively chooses one POVM measurement at a time based on the statistics of all previous measurements, and it is tested on spin-chain property prediction, unsupervised phase classification, and continuous-variable state tomography. Across these tasks the paper reports that TGMS-selected measurements reach the same or better accuracy with fewer measurement settings than uniform random sampling, and that for topological systems with open boundaries the model spontaneously concentrates measurements at the edges, even when the target property is a bulk quantity. If true, this provides a general, data-driven subroutine for quantum-state learning that reduces experimental measurement cost and can rediscover physical structure such as edge–bulk correspondence without being told about it.

What carries the argument

The central object is an encoder–decoder transformer. The encoder maps every available POVM $M_\theta$ to a latent vector $h_\theta$ once; the decoder maintains a state representation $h_x^{(t)}$ as the average of previous measurement embeddings and outcome statistics, assigns each unused measurement a utility score $u_\theta = \mathrm{Dec}(h_x^{(t)}, h_\theta)$, and samples the next measurement from the softmax of those scores. Training samples several candidate next measurements per step and rewards utilities that reduce the prediction loss $|y_x - f(\mathrm{data})|^2$, with a sliding window over time steps to keep gradient propagation depth constant. The pretrained predictor $f$ is fixed, so the transformer learns only the selection policy, making the approach a reusable subroutine for any quantum-state-learning workflow that supplies such a predictor.

What would settle it

Repeat the cluster-Ising, XXZ, and cat-state benchmarks with finite shot numbers per POVM setting, feeding empirical outcome frequencies instead of exact probabilities to the same TGMS model and the random baseline; if TGMS does not beat random sampling in prediction error or tomography infidelity at equal total shots, the central claim fails.

Watch

Extended reading notes

Core claim

The central discovery is that an adaptive measurement-selection policy can be learned end-to-end as a sequence model and that the learned policy transfers to new states. TGMS encodes the full set of available POVMs (positive-operator-valued measures, the general form of quantum measurements) into latent vectors, then at each round scores every unused measurement by a decoder conditioned on the average of previous measurement embeddings and their outcome statistics, and samples the next measurement from the softmax of these scores. It is trained to increase the utility of candidate measurements whose data reduce the prediction error of a fixed downstream predictor $f$, a multi-task network for spin chains and iterative maximum-likelihood estimation for tomography. In the paper's experiments, the same recipe yields higher prediction accuracy for spin correlations, entanglement entropies, and the many-body topological invariant; clean unsupervised separation of symmetry-protected and symmetry-broken phases with ten selected triplet measurements; and cat-state reconstruction with fidelity around $0.95$ using about half the phase-space points that random sampling needs. The authors interpret the pronounced edge preference on open-boundary topological chains, its disappearance on rings, and its persistence under qubit relabeling as evidence that the model discovered edge–bulk correspondence from local measurement statistics alone.

Load-bearing premise

The comparison assumes exact, noiseless measurement statistics for each chosen setting, and it assumes the fixed pretrained predictor is adequate; if realistic finite-shot noise degrades the learned strategies more than it degrades random sampling, the claimed consistent advantage is not established.

Editorial extensions

If this is right

  • Using TGMS-selected measurements lowers the number of measurement settings needed to reach a given prediction accuracy for spin correlations, entanglement entropies, and the many-body topological invariant, compared with uniform random sampling.
  • On cluster-Ising ground states, ten TGMS-selected triplet measurements yield state representations in which the symmetry-protected and symmetry-broken phases form distinct clusters, while random measurements do not separate them.
  • For cat-state tomography, the learned phase-space points reach fidelity around 0.95 with roughly half the measurement points that random sampling requires.
  • For open-boundary topological systems the learned policy concentrates measurements at the edges, even when the predicted quantity is a bulk property, and this edge preference disappears for periodic rings, showing the model infers boundary conditions from local data.
  • The trained policy transfers without retraining to out-of-distribution states, including ground states of a perturbed Hamiltonian and states generated by shallow random circuits, and retains its accuracy advantage.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • A direct test the paper leaves open is shot-noise robustness: retraining TGMS with empirical frequencies from finite measurement shots would likely be needed, and it is unresolved whether the learned policies stay better than random at realistic shot counts.
  • The boundary-focus result suggests a cheap experimental protocol for one-dimensional symmetry-protected phases, namely measuring a few edge triplets instead of the full string order parameter, which could be checked on current ion-trap or superconducting devices.
  • Because the policy is trained per state family on simulated data, deployment to a new device requires the measurement statistics and the predictor $f$ to match the training distribution; the paper does not quantify how much distribution shift the policy tolerates.
  • The same sequence-selection loop could be aimed at other measurement-optimization problems, such as fidelity estimation, cross-platform verification, or Hamiltonian learning, but those applications are not demonstrated here.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

4 major / 4 minor

Summary. The paper introduces TGMS, a transformer-based sequence model that adaptively selects POVM measurements for quantum state learning tasks. The model is trained to choose measurements that minimize the prediction error of a fixed predictor f, and it is tested on three families of tasks: property prediction and unsupervised phase classification for cluster-Ising and bond-alternating XXZ ground states, and quantum state tomography of optical cat states from Husimi-Q samples. The central claim is that TGMS-selected measurements consistently outperform uniformly random measurement selection across all tasks, and that for topological systems the model spontaneously prefers boundary measurements even when the target property is a bulk quantity. The paper also reports generalization to out-of-distribution states generated by random circuits and by a perturbed Hamiltonian.

Significance. If the claims hold, TGMS is a useful task-agnostic subroutine for reducing the number of measurement settings in quantum state learning, and the emergent boundary preference is an intriguing data-driven observation about edge-bulk correspondence. The experimental design is fair as an internal comparison: both TGMS and the random baseline use the same predictor f, and the boundary-preference phenomenon is tested under periodic boundary conditions and permuted qubit indices, which strengthens the interpretation. However, the evidence is simulation-based and largely noiseless, no error bars or significance tests are reported, and the only baseline is uniform random sampling. The central 'consistently outperform' claim is therefore not yet established at the level of experimental usefulness, although the internal comparison is suggestive and worth pursuing.

major comments (4)
  1. [Results A; Section C, Eq. (6)] The paper explicitly assumes noiseless measurement statistics: in the framework it says 'Ideally, if we ignore finite shot noise...', and in the tomography section it says 'we neglect the statistical errors introduced by finite-number shots of measurements, and hence each data di corresponds to the exact value of Q function.' Because the TGMS policy conditions its next choice on previous outcome statistics, with finite shots those inputs become noisy and a poor early choice can be amplified by the adaptive loop, whereas a non-adaptive random strategy does not suffer this feedback. The manuscript reports no finite-shot simulations or shot-count analysis, so the broad claim that TGMS 'consistently outperforms' random sampling is not yet established for realistic experimental data. This is load-bearing because the stated motivation is experimental data acquisition.
  2. [Figs. 2a, 3a, 4a, 5a, 6a] The accuracy-vs-measurements curves are reported without error bars, confidence intervals, or significance tests, and the number of random seeds or test states is not stated in the main text. Given that the claims are explicitly comparative ('outperform', 'consistently'), the absence of any statistical characterization makes it impossible to determine whether the observed gaps are robust or within fluctuation. The authors should provide repeated runs with standard deviations, or at least a clear statement of the number of independent repetitions and a significance test.
  3. [Section C] The tomography comparison is made in terms of infidelity versus the number of phase-space points, using exact Q-function values. The claim that TGMS achieves fidelity around 0.95 'using only half the measurement data' is not a demonstrated saving in experimental resources, because each phase-space point requires a number of shots and the total shot budget is never specified or compared. A meaningful resource comparison should be in total shots or total experiment time, including any overhead of the adaptive decision process.
  4. [Methods, Network training] Key training hyperparameters are not given in the main text: the number K of sampled measurements per step, the sliding-window parameters T1 and T2, the transformer depth and width, learning rate, batch size, and number of training states. The text only states that K samples are drawn and that T1 and T2 are progressively increased while keeping the window size fixed. Without these values, the experiments cannot be reproduced from the main text; if they appear only in the Supplementary Material, the main text should at least summarize the most important ones.
minor comments (4)
  1. [Section B.1] There is a duplicated paragraph: the text beginning 'Note that measurements at different three-qubit subsystems can be performed simultaneously...' appears twice in the same section, and the second occurrence is followed by 'For this reason, we also investigate...' although the content was already presented. Please remove the duplication.
  2. [Throughout] There are several typos: 'Spcefically' in Section B.2, 'patten' in Section B.1, 'phenomenons' in the Discussion, 'TMGS' in Section B.2 (should be TGMS), and 'BB' for batch normalization in the Methods section (should be BN).
  3. [Footnote [70]] The correlation measure between two measurement strategies is defined with a formula involving max and min over pairwise distances, but it is not clear whether the definition is symmetric or how it should be interpreted when the denominator is zero. Please clarify the definition.
  4. [Section B.1] The t-SNE visualizations in Figs. 2b and 4c would be more informative if the number of states and the random seed (or the statement that t-SNE is deterministic in the reported runs) were specified, since t-SNE projections can vary across runs.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: the TGMS advantage is an empirical held-out comparison under a fixed predictor, and the noiseless-data assumption is a scope limitation, not a circular reduction.

full rationale

The paper's central claim is an empirical comparison: a transformer-guided measurement selection model is trained to minimize the prediction error of a fixed predictor f on training states and then evaluated on held-out states against random measurement selection using the same f. Because the reported outperform result is measured on states not seen during TGMS training, including cluster-Ising ground states, bond-alternating XXZ ground states, perturbed-Hamiltonian ground states, random-circuit states, and cat states, the comparison is not a fitted-input-as-prediction loop. The boundary preference is an emergent property of the optimized policy, and the open-boundary versus periodic-boundary contrast in Figs. 2c and 3b provides a control that supports the interpretation. The reliance on a multi-task network from Ref. [38], which shares authors with this paper, is a mild self-citation, but it is not load-bearing for the circularity question: f is held fixed, the random baseline uses the same f, and the comparison is well-defined even if f is imperfect. The explicit noiseless idealizations, 'if we ignore finite shot noise' and 'we neglect the statistical errors introduced by finite-number shots of measurements', are stated scope limitations that weaken external validity under shot noise, but they do not make the prediction equivalent to the training input. No equation or fitted parameter is recycled as a prediction, and no uniqueness theorem or ansatz is imported via self-citation. Therefore the derivation chain is not circular.

Assumptions & free parameters 4 free parameters · 5 assumptions · 0 invented entities

The central claim depends on standard quantum measurement theory, an ideal noiseless-data assumption, the accuracy of numerical ground-state solvers, and a fixed pretrained predictor from the authors' earlier work. The learned neural network has many trainable parameters, but the only hand-chosen hyperparameters visible in the main text are listed above; several architecture details are deferred to the Supplementary Material. No new physical entities are introduced.

free parameters (4)
  • TGMS architecture hyperparameters
    Layer counts, embedding dimension dh, attention heads, and MLP widths are chosen by hand but not reported in the main text (deferred to Supplementary Material). The learned policy depends on them.
  • Training sampling constants K and sliding window T1, T2
    In Network training, K candidate measurements per step and the window [T1, T2] over which loss is accumulated are free choices; values are not given in the main text.
  • Truncated Hilbert dimension for CV tomography = 16
    Reconstruction of cat states assumes dimension d=16; infidelity numbers depend on this truncation.
  • Fixed number of measurement settings (10) in phase-clustering comparison = 10
    Both TGMS and random sampling are allowed 10 settings in the phase-clustering comparison; the comparison is defined relative to this choice.
assumptions (5)
  • standard math Born rule and POVM formalism govern measurement statistics.
    Used throughout Section A and Methods; no alternative measurement model is considered.
  • domain assumption Measurement outcome statistics are noiseless exact values.
    Locations: Section A ('Ideally, if we ignore finite shot noise...'), Methods for tomography ('we neglect the statistical errors introduced by finite-number shots of measurements').
  • domain assumption Ground states of cluster-Ising and XXZ Hamiltonians are computed accurately by exact diagonalization and DMRG.
    Methods, Data generation; DMRG is approximate for 50 and 100 qubits.
  • domain assumption The pretrained multi-task network from Ref. 38 is an adequate fixed predictor f for all property tasks.
    Section B.1: 'we employ a deep neural network, specifically a multi-task learning model [38]'; the TGMS advantage is measured with this f, not with a fresh predictor.
  • domain assumption In tomography, truncating to dimension 16 and using iterative MLE is a sufficient reconstruction procedure.
    Methods; infidelity values depend on this truncation and the noise-free assumption.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Sequence-Model-Guided Measurement Selection for Quantum State Learning." pith.science (2026). https://pith.science/paper/6ZVBVNVL

@misc{pith2026250709891,
  author       = {Pith},
  title        = {Pith review of: Sequence-Model-Guided Measurement Selection for Quantum State Learning},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/6ZVBVNVL}},
  note         = {Machine review of arXiv:2507.09891}
}
read the original abstract

Characterization of quantum systems from experimental data is a central problem in quantum science and technology. But which measurements should be used to gather data in the first place? While optimal measurement choices can be worked out for small quantum systems, the optimization becomes intractable as the system size grows large. To address this problem, we introduce a deep neural network with a sequence model architecture that searches for efficient measurement choices in a data-driven, adaptive manner. The model can be applied to a variety of tasks, including the prediction of linear and nonlinear properties of quantum states, as well as state clustering and state tomography tasks. In all these tasks, we find that the measurement choices identified by our neural network consistently outperform the uniformly random choice. Intriguingly, for topological quantum systems, our model tends to recommend measurements at the system's boundaries, even when the task is to predict bulk properties. This behavior suggests that the neural network may have independently discovered a connection between boundaries and bulk, without having been provided any built-in knowledge of quantum physics.

Figures

Figures reproduced from arXiv: 2507.09891 by the authors.

Figure 1
Figure 1. FIG. 1. The learning procedure is divided into three stages: measurement encoding, measurement selection and data acquisition. [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. FIG. 2. Prediction of properties for cluster Ising model [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. FIG. 3. Prediction of properties for cluster Ising model [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figures from the paper (3 more)
Figure 4
Figure 4. Figure 4: FIG. 4. Prediction of properties for states generated by ran [PITH_FULL_IMAGE:figures/full_fig_p005_4.png]
Figure 5
Figure 5. Figure 5: FIG. 5. Prediction of properties for bond-alternating XXZ model ground states. Subfig a shows the prediction accuracies of [PITH_FULL_IMAGE:figures/full_fig_p006_5.png]
Figure 6
Figure 6. Figure 6: FIG. 6. Tomography of cat states. Subfig [PITH_FULL_IMAGE:figures/full_fig_p007_6.png]

Discussion (0). Sign in to comment.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. Artificial intelligence for representing and characterizing quantum systems

    quant-ph 2025-09 unverdicted novelty 1.0 of 10

    A review organizes AI-based quantum system characterization into ML, deep learning, and language model paradigms, covering property prediction and implicit state reconstruction.

Reference graph

Works this paper leans on

89 extracted references · 68 canonical work pages · cited by 1 Pith paper

  1. [1]

    Cluster-Ising model ground states We apply our transformer-based model for predicting quantum properties, using the ground states of a one- dimensional quantum system with local interactions as a testbed. Specifically, we first consider the ground states of a nine-qubit cluster-Ising model with open bound- aries [61] H open cI = − N −2X i=1 σz i σx i+1σz ...

  2. [2]

    Bond-alternating XXZ model ground states Now we consider a learning task of a larger-sized quan- tum system. Spcefically, we consider the ground states of a 50-qubit bond-alternating XXZ model [49] HXXZ =J N/2X i=1 σx 2i−1σx 2i + σy 2i−1σy 2i + δσ z 2i−1σz 2i + J ′ N/2−1X i=1 σx 2iσx 2i+1 + σy 2iσy 2i+1 + δσ z 2iσz 2i+1 . (4) There are three different pha...

  3. [3]

    Learning quantum systems,

    Valentin Gebhart, Raffaele Santagati, Antonio Andrea Gentile, Erik M Gauger, David Craig, Natalia Ares, Leonardo Banchi, Florian Marquardt, Luca Pezze, and 10 Cristian Bonato, “Learning quantum systems,” Nature Reviews Physics 5, 141–156 (2023)

    2023

  4. [4]

    Provably efficient ma- chine learning for quantum many-body problems,

    Hsin-Yuan Huang, Richard Kueng, Giacomo Torlai, Vic- tor V Albert, and John Preskill, “Provably efficient ma- chine learning for quantum many-body problems,” Sci- ence 377, eabk3333 (2022)

    2022

  5. [5]

    Improved machine learning algorithm for predicting ground state properties,

    Laura Lewis, Hsin-Yuan Huang, Viet T Tran, Sebastian Lehner, Richard Kueng, and John Preskill, “Improved machine learning algorithm for predicting ground state properties,” Nat. Commun. 15, 895 (2024)

    2024

  6. [6]

    Ex- ponentially improved efficient machine learning for quan- tum many-body states with provable guarantees,

    Yanming Che, Clemens Gneiting, and Franco Nori, “Ex- ponentially improved efficient machine learning for quan- tum many-body states with provable guarantees,” Phys. Rev. Res. 6, 033035 (2024)

    2024

  7. [7]

    Machine learning on quantum experimental data toward solving quantum many-body problems,

    Gyungmin Cho and Dohun Kim, “Machine learning on quantum experimental data toward solving quantum many-body problems,” Nat. Commun. 15, 7552 (2024)

    2024

  8. [8]

    Learning symmetry-protected topological order from trapped-ion experiments

    Nicolas Sadoune, Ivan Pogorelov, Claire L Edmunds, Giuliano Giudici, Giacomo Giudice, Christian D Marciniak, Martin Ringbauer, Thomas Monz, and Lode Pollet, “Learning symmetry-protected topological order from trapped-ion experiments,” arXiv preprint arXiv:2408.05017 (2024)

    work page Pith review arXiv 2024

Show all 89 references
  1. [9]

    Neural-network quantum state tomography,

    Giacomo Torlai, Guglielmo Mazzola, Juan Carrasquilla, Matthias Troyer, Roger Melko, and Giuseppe Carleo, “Neural-network quantum state tomography,” Nat. Phys. 14, 447–450 (2018)

    2018

  2. [10]

    Reconstructing quantum states with generative models,

    Juan Carrasquilla, Giacomo Torlai, Roger G Melko, and Leandro Aolita, “Reconstructing quantum states with generative models,” Nat. Mach. Intell. 1, 155–161 (2019)

    2019

  3. [11]

    Experimental quantum homodyne tomography via machine learning,

    Egor S Tiunov, VV Tiunova, Alexander E Ulanov, AI Lvovsky, and Aleksey K Fedorov, “Experimental quantum homodyne tomography via machine learning,” Optica 7, 448–454 (2020)

    2020

  4. [12]

    Quantum state tomography with conditional generative adversarial networks,

    Shahnawaz Ahmed, Carlos S´ anchez Mu˜ noz, Franco Nori, and Anton Frisk Kockum, “Quantum state tomography with conditional generative adversarial networks,” Phys. Rev. Lett. 127, 140502 (2021)

    2021

  5. [13]

    Attention-based quantum tomography,

    Peter Cha, Paul Ginsparg, Felix Wu, Juan Carrasquilla, Peter L McMahon, and Eun-Ah Kim, “Attention-based quantum tomography,” Mach. Learn.: Sci. Tech. 3, 01LT01 (2021)

    2021

  6. [14]

    Quantum state tomography inspired by language modeling,

    Lu Zhong, Chu Guo, and Xiaoting Wang, “Quantum state tomography inspired by language modeling,” arXiv preprint arXiv:2212.04940 (2022)

    2022 arXiv

  7. [15]

    Tomography of quantum states from structured measurements via quantum-aware trans- former,

    Hailan Ma, Zhenhong Sun, Daoyi Dong, Chunlin Chen, and Herschel Rabitz, “Tomography of quantum states from structured measurements via quantum-aware trans- former,” arXiv preprint arXiv:2305.05433 (2023)

    2023 arXiv

  8. [16]

    En- hancing quantum state tomography via resource-efficient attention-based neural networks,

    Adriano Macarone Palmieri, Guillem M¨ uller-Rigat, Anubhav Kumar Srivastava, Maciej Lewenstein, Grze- gorz Rajchel-Mieldzio´ c, and Marcin P lodzie´ n, “En- hancing quantum state tomography via resource-efficient attention-based neural networks,” Phys. Rev. Res. 6, 033248 (2024)

    2024

  9. [17]

    Neural-shadow quantum state tomog- raphy,

    Victor Wei, W. A. Coish, Pooya Ronagh, and Chris- tine A. Muschik, “Neural-shadow quantum state tomog- raphy,” Phys. Rev. Res. 6, 023250 (2024)

    2024

  10. [18]

    Efficient quantum state sample tomography with basis- dependent neural networks,

    Alistair W. R. Smith, Johnnie Gray, and M. S. Kim, “Efficient quantum state sample tomography with basis- dependent neural networks,” PRX Quantum 2, 020348 (2021)

    2021

  11. [19]

    Direct fi- delity estimation of quantum states using machine learn- ing,

    Xiaoqian Zhang, Maolin Luo, Zhaodi Wen, Qin Feng, Shengshi Pang, Weiqi Luo, and Xiaoqi Zhou, “Direct fi- delity estimation of quantum states using machine learn- ing,” Phys. Rev. Lett. 127, 130503 (2021)

    2021

  12. [20]

    Predicting properties of quantum sys- tems with conditional generative models,

    Haoxiang Wang, Maurice Weber, Josh Izaac, and Cedric Yen-Yu Lin, “Predicting properties of quantum sys- tems with conditional generative models,” arXiv preprint arXiv:2211.16943 (2022)

    2022 arXiv

  13. [21]

    Trans- former quantum state: A multipurpose model for quan- tum many-body problems,

    Yuan-Hang Zhang and Massimiliano Di Ventra, “Trans- former quantum state: A multipurpose model for quan- tum many-body problems,” Phys. Rev. B 107, 075147 (2023)

    2023

  14. [22]

    Quantum similarity testing with con- volutional neural networks,

    Ya-Dong Wu, Yan Zhu, Ge Bai, Yuexuan Wang, and Giulio Chiribella, “Quantum similarity testing with con- volutional neural networks,” Phys. Rev. Lett. 130, 210601 (2023)

    2023

  15. [23]

    Deep learning of quan- tum entanglement from incomplete measurements,

    Dominik Koutn` y, Laia Gin´ es, Magdalena Mocza la- Dusanowska, Sven H¨ ofling, Christian Schneider, Ana Predojevi´ c, and Miroslav Jeˇ zek, “Deep learning of quan- tum entanglement from incomplete measurements,” Sci. Adv. 9, eadd7131 (2023)

    2023

  16. [24]

    Towards llm4qpe: Unsupervised pretrain- ing of quantum property estimation and a benchmark,

    Yehui Tang, Hao Xiong, Nianzu Yang, Tailong Xiao, and Junchi Yan, “Towards llm4qpe: Unsupervised pretrain- ing of quantum property estimation and a benchmark,” in The Twelfth International Conference on Learning Representations

  17. [25]

    Unified quantum state tomography and hamiltonian learning: A language-translation-like ap- proach for quantum systems,

    Zheng An, Jiahui Wu, Muchun Yang, D. L. Zhou, and Bei Zeng, “Unified quantum state tomography and hamiltonian learning: A language-translation-like ap- proach for quantum systems,” Phys. Rev. Appl. 21, 014037 (2024)

    2024

  18. [26]

    Correlation-pattern-based continu- ous variable entanglement detection through neural net- works,

    Xiaoting Gao, Mathieu Isoard, Fengxiao Sun, Carlos E. Lopetegui, Yu Xiang, Valentina Parigi, Qiongyi He, and Mattia Walschaers, “Correlation-pattern-based continu- ous variable entanglement detection through neural net- works,” Phys. Rev. Lett. 132, 220202 (2024)

    2024

  19. [27]

    Machine learning phases of matter,

    Juan Carrasquilla and Roger G Melko, “Machine learning phases of matter,” Nat. Phys. 13, 431–434 (2017)

    2017

  20. [28]

    Learning phase transitions by confusion,

    Evert PL Van Nieuwenburg, Ye-Hua Liu, and Sebas- tian D Huber, “Learning phase transitions by confusion,” Nat. Phys. 13, 435–439 (2017)

    2017

  21. [29]

    Discrim- inative cooperative networks for detecting phase transi- tions,

    Ye-Hua Liu and Evert P. L. van Nieuwenburg, “Discrim- inative cooperative networks for detecting phase transi- tions,” Phys. Rev. Lett. 120, 176401 (2018)

    2018

  22. [30]

    Vector field divergence of predictive model output as indication of phase transi- tions,

    Frank Sch¨ afer and Niels L¨ orch, “Vector field divergence of predictive model output as indication of phase transi- tions,” Phys. Rev. E 99, 062107 (2019)

    2019

  23. [31]

    Unsuper- vised identification of topological phase transitions using predictive models,

    Eliska Greplova, Agnes Valenti, Gregor Boschung, Frank Sch¨ afer, Niels L¨ orch, and Sebastian D Huber, “Unsuper- vised identification of topological phase transitions using predictive models,” New Journal of Physics 22, 045003 (2020)

    2020

  24. [32]

    Interpretable and unsupervised phase clas- sification,

    Julian Arnold, Frank Sch¨ afer, MartinˇZonda, and Axel U. J. Lode, “Interpretable and unsupervised phase clas- sification,” Phys. Rev. Res. 3, 033052 (2021)

    2021

  25. [33]

    Adversarial machine learning phases of matter,

    Si Jiang, Sirui Lu, and Dong-Ling Deng, “Adversarial machine learning phases of matter,” Quantum Frontiers 2, 15 (2023)

    2023

  26. [34]

    Neural-network decoders for mea- surement induced phase transitions,

    Hossein Dehghani, Ali Lavasani, Mohammad Hafezi, and Michael J Gullans, “Neural-network decoders for mea- surement induced phase transitions,” Nat. Commun. 14, 2918 (2023)

    2023

  27. [35]

    Attention to quantum complexity,

    Hyejin Kim, Yiqing Zhou, Yichen Xu, Kaarthik Varma, Amir H Karamlou, Ilan T Rosen, Jesse C Hoke, Chao Wan, Jin Peng Zhou, William D Oliver, et al., “Attention to quantum complexity,” arXiv preprint 11 arXiv:2405.11632 (2024)

    2024

  28. [36]

    Discovering physical concepts with neural networks,

    Raban Iten, Tony Metger, Henrik Wilming, L ´ ıdia del Rio, and Renato Renner, “Discovering physical concepts with neural networks,” Phys. Rev. Lett. 124, 010508 (2020)

    2020

  29. [37]

    Flexible learn- ing of quantum states with generative query neural net- works,

    Yan Zhu, Ya-Dong Wu, Ge Bai, Dong-Sheng Wang, Yuexuan Wang, and Giulio Chiribella, “Flexible learn- ing of quantum states with generative query neural net- works,” Nat. Commun. 13, 6222 (2022)

    2022

  30. [38]

    Intelligent certification for quantum simulators via machine learning,

    Tailong Xiao, Jingzheng Huang, Hongjing Li, Jianping Fan, and Guihua Zeng, “Intelligent certification for quantum simulators via machine learning,” NPJ Quan- tum Inf. 8, 138 (2022)

    2022

  31. [39]

    Multimodal deep representation learning for quantum cross-platform verification,

    Yang Qian, Yuxuan Du, Zhenliang He, Min-Hsiu Hsieh, and Dacheng Tao, “Multimodal deep representation learning for quantum cross-platform verification,” Phys. Rev. Lett. 133, 130601 (2024)

    2024

  32. [40]

    Learning quantum properties from short- range correlations using multi-task networks,

    Ya-Dong Wu, Yan Zhu, Yuexuan Wang, and Giulio Chiribella, “Learning quantum properties from short- range correlations using multi-task networks,” Nat. Com- mun. 15, 8796 (2024)

    2024

  33. [41]

    Shadowgpt: Learning to solve quantum many-body problems from randomized measurements,

    Jian Yao and Yi-Zhuang You, “Shadowgpt: Learning to solve quantum many-body problems from randomized measurements,” arXiv:2411.03285 (2024)

    2024 arXiv

  34. [42]

    Direct en- tanglement detection of quantum systems using machine learning,

    Yulei Huang, Liangyu Che, Chao Wei, Feng Xu, Xin- fang Nie, Jun Li, Dawei Lu, and Tao Xin, “Direct en- tanglement detection of quantum systems using machine learning,” npj Quantum Information 11, 29 (2025)

    2025

  35. [43]

    Rydberggpt,

    David Fitzek, Yi Hong Teoh, Hin Pok Fung, Gebremed- hin A Dagnew, Ejaaz Merali, M Schuyler Moss, Benjamin MacLellan, and Roger G Melko, “Rydberggpt,” arXiv preprint arXiv:2405.21052 (2024)

    2024 arXiv

  36. [44]

    Adaptive quantum state tomography improves accuracy quadratically,

    D. H. Mahler, Lee A. Rozema, Ardavan Darabi, Christo- pher Ferrie, Robin Blume-Kohout, and A. M. Steinberg, “Adaptive quantum state tomography improves accuracy quadratically,” Phys. Rev. Lett. 111, 183601 (2013)

    2013

  37. [45]

    Self-guided quantum tomography,

    Christopher Ferrie, “Self-guided quantum tomography,” Phys. Rev. Lett. 113, 190404 (2014)

    2014

  38. [46]

    Adaptive quan- tum state tomography via linear regression estimation: Theory and two-qubit experiment,

    Bo Qi, Zhibo Hou, Yuanlong Wang, Daoyi Dong, Han- Sen Zhong, Li Li, Guo-Yong Xiang, Howard M Wiseman, Chuan-Feng Li, and Guang-Can Guo, “Adaptive quan- tum state tomography via linear regression estimation: Theory and two-qubit experiment,” NPJ Quantum Inf. 3, 19 (2017)

    2017

  39. [47]

    Probing r´ enyi en- tanglement entropy via randomized measurements,

    Tiff Brydges, Andreas Elben, Petar Jurcevic, Beno ˆ ıt Ver- mersch, Christine Maier, Ben P Lanyon, Peter Zoller, Rainer Blatt, and Christian F Roos, “Probing r´ enyi en- tanglement entropy via randomized measurements,” Sci- ence 364, 260–263 (2019)

    2019

  40. [48]

    Predicting many properties of a quantum system from very few measurements,

    Hsin-Yuan Huang, Richard Kueng, and John Preskill, “Predicting many properties of a quantum system from very few measurements,” Nat. Phys. 16, 1050–1057 (2020)

    2020

  41. [49]

    Cross-platform verification of intermediate scale quantum devices,

    Andreas Elben, Beno ˆ ıt Vermersch, Rick van Bij- nen, Christian Kokail, Tiff Brydges, Christine Maier, Manoj K. Joshi, Rainer Blatt, Christian F. Roos, and Peter Zoller, “Cross-platform verification of intermediate scale quantum devices,” Phys. Rev. Lett. 124, 010504 (2020)

    2020

  42. [50]

    Mixed- state entanglement from local randomized measure- ments,

    Andreas Elben, Richard Kueng, Hsin-Yuan (Robert) Huang, Rick van Bijnen, Christian Kokail, Marcello Dalmonte, Pasquale Calabrese, Barbara Kraus, John Preskill, Peter Zoller, and Beno ˆ ıt Vermersch, “Mixed- state entanglement from local randomized measure- ments,” Phys. Rev. L...

    2020

  43. [51]

    Many-body topological invariants from random- ized measurements in synthetic quantum matter,

    Andreas Elben, Jinlong Yu, Guanyu Zhu, Mohammad Hafezi, Frank Pollmann, Peter Zoller, and Beno ˆ ıt Verm- ersch, “Many-body topological invariants from random- ized measurements in synthetic quantum matter,” Sci. Adv. 6, eaaz3666 (2020)

    2020

  44. [52]

    The randomized measurement toolbox,

    Andreas Elben, Steven T Flammia, Hsin-Yuan Huang, Richard Kueng, John Preskill, Beno ˆ ıt Vermersch, and Peter Zoller, “The randomized measurement toolbox,” Nat. Rev. Phys. 5, 9–24 (2023)

    2023

  45. [53]

    Demon- stration of robust and efficient quantum property learn- ing with shallow shadows,

    Hong-Ye Hu, Andi Gu, Swarnadeep Majumder, Hang Ren, Yipei Zhang, Derek S Wang, Yi-Zhuang You, Zlatko Minev, Susanne F Yelin, and Alireza Seif, “Demon- stration of robust and efficient quantum property learn- ing with shallow shadows,” Nature Communications 16, 2943 (2025)

    2025

  46. [54]

    Direct fidelity esti- mation from few pauli measurements,

    Steven T. Flammia and Yi-Kai Liu, “Direct fidelity esti- mation from few pauli measurements,” Phys. Rev. Lett. 106, 230501 (2011)

    2011

  47. [55]

    Importance sampling of randomized measurements for probing entanglement,

    Aniket Rath, Rick van Bijnen, Andreas Elben, Peter Zoller, and Beno ˆ ıt Vermersch, “Importance sampling of randomized measurements for probing entanglement,” Phys. Rev. Lett. 127, 200503 (2021)

    2021

  48. [56]

    En- hanced estimation of quantum properties with common randomized measurements,

    Beno ˆ ıt Vermersch, Aniket Rath, Bharathan Sundar, Cyril Branciard, John Preskill, and Andreas Elben, “En- hanced estimation of quantum properties with common randomized measurements,” PRX Quantum 5, 010352 (2024)

    2024

  49. [57]

    Attention is all you need,

    Ashish Vaswani, Noam Shazeer, Niki Parmar, Jakob Uszkoreit, Llion Jones, Aidan N Gomez, Lukasz Kaiser, and Illia Polosukhin, “Attention is all you need,” Ad- vances in neural information processing systems 30 (2017)

    2017

  50. [58]

    Adap- tive quantum state tomography with neural networks,

    Yihui Quek, Stanislav Fort, and Hui Khoon Ng, “Adap- tive quantum state tomography with neural networks,” NPJ Quantum Inf. 7, 105 (2021)

    2021

  51. [59]

    Adaptive quantum state tomography with active learn- ing,

    Hannah Lange, Matjaˇ z Kebriˇ c, Maximilian Buser, Ul- rich Schollw¨ ock, Fabian Grusdt, and Annabelle Bohrdt, “Adaptive quantum state tomography with active learn- ing,” Quantum 7, 1129 (2023)

    2023

  52. [60]

    Tensor- entanglement-filtering renormalization approach and symmetry-protected topological order,

    Zheng-Cheng Gu and Xiao-Gang Wen, “Tensor- entanglement-filtering renormalization approach and symmetry-protected topological order,” Phys. Rev. B 80, 155131 (2009)

    2009

  53. [61]

    Detection of symmetry-protected topological phases in one dimen- sion,

    Frank Pollmann and Ari M. Turner, “Detection of symmetry-protected topological phases in one dimen- sion,” Phys. Rev. B 86, 125441 (2012)

    2012

  54. [62]

    Symmetry protection of topological phases in one-dimensional quantum spin systems,

    Frank Pollmann, Erez Berg, Ari M. Turner, and Masaki Oshikawa, “Symmetry protection of topological phases in one-dimensional quantum spin systems,” Phys. Rev. B 85, 075125 (2012)

    2012

  55. [63]

    Statistical mechanics of the cluster ising model,

    Pietro Smacchia, Luigi Amico, Paolo Facchi, Rosario Fazio, Giuseppe Florio, Saverio Pascazio, and Vlatko Ve- dral, “Statistical mechanics of the cluster ising model,” Phys. Rev. A 84, 022304 (2011)

    2011

  56. [64]

    Efficient tomography of a quantum many-body system,

    BP Lanyon, C Maier, Milan Holz¨ apfel, Tillmann Baum- gratz, C Hempel, P Jurcevic, Ish Dhand, AS Buyskikh, AJ Daley, Marcus Cramer, et al., “Efficient tomography of a quantum many-body system,” Nat. Phys. 13, 1158– 1162 (2017)

    2017

  57. [65]

    Many-body entropies and entanglement from polynomi- ally many local measurements,

    Beno ˆ ıt Vermersch, Marko Ljubotina, J. Ignacio Cirac, Peter Zoller, Maksym Serbyn, and Lorenzo Piroli, “Many-body entropies and entanglement from polynomi- ally many local measurements,” Phys. Rev. X 14, 031035 (2024). 12

    2024

  58. [66]

    De- tecting genuine multipartite entanglement in multi-qubit devices with restricted measurements,

    Nicky Kai Hong Li, Xi Dai, Manuel H Mu˜ noz-Arias, Kevin Reuer, Marcus Huber, and Nicolai Friis, “De- tecting genuine multipartite entanglement in multi-qubit devices with restricted measurements,” arXiv preprint arXiv:2504.21076 (2025)

    2025

  59. [67]

    Diagnosing 2d symmetry protected topological states via mixed state anomaly,

    Chao Xu, Yunlong Zang, Yixin Ma, Yingfei Gu, and Shenghan Jiang, “Diagnosing 2d symmetry protected topological states via mixed state anomaly,” arXiv preprint arXiv:2506.13096 (2025)

    2025 arXiv

  60. [68]

    Quan- tum convolutional neural networks,

    Iris Cong, Soonwon Choi, and Mikhail D Lukin, “Quan- tum convolutional neural networks,” Nat. Phys. 15, 1273–1278 (2019)

    2019

  61. [69]

    Realizing quantum convolutional neural networks on a super- conducting quantum processor to recognize quantum phases,

    Johannes Herrmann, Sergi Masot Llima, Ants Remm, Petr Zapletal, Nathan A McMahon, Colin Scarato, Fran¸ cois Swiadek, Christian Kraglund Andersen, Christoph Hellings, Sebastian Krinner, et al., “Realizing quantum convolutional neural networks on a super- conducting quantum proc...

    2022

  62. [70]

    Observation of entangled states of a fully controlled 20-qubit system,

    Nicolai Friis, Oliver Marty, Christine Maier, Cornelius Hempel, Milan Holz¨ apfel, Petar Jurcevic, Martin B. Ple- nio, Marcus Huber, Christian Roos, Rainer Blatt, and Ben Lanyon, “Observation of entangled states of a fully controlled 20-qubit system,” Phys. Rev. X 8, 021012 (2018)

    2018

  63. [71]

    Probing entanglement in a 2d hard-core bose– hubbard lattice,

    Amir H Karamlou, Ilan T Rosen, Sarah E Muschinske, Cora N Barrett, Agustin Di Paolo, Leon Ding, Patrick M Harrington, Max Hays, Rabindra Das, David K Kim, et al., “Probing entanglement in a 2d hard-core bose– hubbard lattice,” Nature , 1–6 (2024)

    2024

  64. [72]

    The measure- ments of two states are represented by two normal- ized 27-dimensional vectors fi and fj after bin count- ing

    Here we neglect the spin sites and the order of qubit- triplet Pauli measurement sequences. The measure- ments of two states are represented by two normal- ized 27-dimensional vectors fi and fj after bin count- ing. Their correlation is defined as Corr( fi, fj) := maxi,j {di,j...

  65. [73]

    Quantitative tomography for contin- uous variable quantum systems,

    Olivier Landon-Cardinal, Luke C. G. Govia, and Aashish A. Clerk, “Quantitative tomography for contin- uous variable quantum systems,” Phys. Rev. Lett. 120, 090501 (2018)

    2018

  66. [74]

    Observation of quantum state collapse and revival due to the single-photon kerr effect,

    Gerhard Kirchmair, Brian Vlastakis, Zaki Leghtas, Si- mon E Nigg, Hanhee Paik, Eran Ginossar, Mazyar Mir- rahimi, Luigi Frunzio, Steven M Girvin, and Robert J Schoelkopf, “Observation of quantum state collapse and revival due to the single-photon kerr effect,” Nature 495, 205–...

    2013

  67. [75]

    Iterative maximum-likelihood re- construction in quantum homodyne tomography,

    Alexander I Lvovsky, “Iterative maximum-likelihood re- construction in quantum homodyne tomography,” Jour- nal of Optics B: Quantum and Semiclassical Optics 6, S556 (2004)

    2004

  68. [76]

    Dynamically protected cat-qubits: a new paradigm for universal quantum computation,

    Mazyar Mirrahimi, Zaki Leghtas, Victor V Albert, Steven Touzard, Robert J Schoelkopf, Liang Jiang, and Michel H Devoret, “Dynamically protected cat-qubits: a new paradigm for universal quantum computation,” New Journal of Physics 16, 045014 (2014)

    2014

  69. [77]

    Interpretable correlator trans- former for image-like quantum matter data,

    Abhinav Suresh, Henning Schl¨ omer, Baran Hashemi, and Annabelle Bohrdt, “Interpretable correlator trans- former for image-like quantum matter data,” Machine Learning: Science and Technology 6, 025006 (2025)

    2025

  70. [78]

    Foundation neural-network quantum states,

    Riccardo Rende, Luciano Loris Viteritti, Federico Becca, Antonello Scardicchio, Alessandro Laio, and Giuseppe Carleo, “Foundation neural-network quantum states,” arXiv preprint arXiv:2502.09488 (2025)

    2025 arXiv

  71. [79]

    Operationally meaningful representations of physical systems in neural networks,

    Hendrik Poulsen Nautrup, Tony Metger, Raban Iten, Sofiene Jerbi, Lea M Trenkwalder, Henrik Wilming, Hans J Briegel, and Renato Renner, “Operationally meaningful representations of physical systems in neural networks,” Machine Learning: Science and Technology 3, 045025 (2022)

    2022

  72. [80]

    On scientific understanding with artificial intelligence,

    Mario Krenn, Robert Pollice, Si Yue Guo, Matteo Aldeghi, Alba Cervera-Lierta, Pascal Friederich, Gabriel dos Passos Gomes, Florian H¨ ase, Adrian Jinich, Ak- shatKumar Nigam, et al., “On scientific understanding with artificial intelligence,” Nature Reviews Physics 4, 761–769 (2022)

    2022

  73. [81]

    Scientific discovery in the age of artificial intelligence,

    Hanchen Wang, Tianfan Fu, Yuanqi Du, Wenhao Gao, Kexin Huang, Ziming Liu, Payal Chandak, Shengchao Liu, Peter Van Katwyk, Andreea Deac, et al., “Scientific discovery in the age of artificial intelligence,” Nature620, 47–60 (2023)

    2023

  74. [82]

    Interpretable ma- chine learning in physics: A review,

    Sebastian Johann Wetzel, Seungwoong Ha, Raban Iten, Miriam Klopotek, and Ziming Liu, “Interpretable ma- chine learning in physics: A review,” arXiv preprint arXiv:2503.23616 (2025)

    2025 arXiv

  75. [83]

    Agents for self- driving laboratories applied to quantum computing,

    Shuxiang Cao, Zijian Zhang, Mohammed Alghadeer, Si- mone D Fasciati, Michele Piscitelli, Mustafa Bakr, Pe- ter Leek, and Al´ an Aspuru-Guzik, “Agents for self- driving laboratories applied to quantum computing,” arXiv preprint arXiv:2412.07978 (2024)

    2024

  76. [84]

    Boundary anomaly detection in two-dimensional sub- system symmetry-protected topological phases,

    Ke Ding, Hao-Ran Zhang, Bai-Ting Liu, and Shuo Yang, “Boundary anomaly detection in two-dimensional sub- system symmetry-protected topological phases,” Phys. Rev. B 111, 205125 (2025)

    2025

  77. [85]

    The density-matrix renormalization group,

    Ulrich Schollw¨ ock, “The density-matrix renormalization group,” Reviews of modern physics 77, 259–315 (2005)

    2005

  78. [86]

    Qutip 5: The quantum toolbox in python,

    Neill Lambert, Eric Gigu` ere, Paul Menczel, Boxi Li, Patrick Hopf, Gerardo Su´ arez, Marc Gali, Jake Lish- man, Rushiraj Gadhvi, Rochisha Agarwal, Asier Galicia, Nathan Shammah, Paul Nation, J. R. Johansson, Shah- nawaz Ahmed, Simon Cross, Alexander Pitchford, and Franco Nori...

    2024 arXiv

  79. [87]

    Deep residual learning for image recognition,

    Kaiming He, Xiangyu Zhang, Shaoqing Ren, and Jian Sun, “Deep residual learning for image recognition,” in Proceedings of the IEEE conference on computer vision and pattern recognition(2016) pp. 770–778

    2016

  80. [88]

    Batch normaliza- tion: Accelerating deep network training by reducing internal covariate shift,

    Sergey Ioffe and Christian Szegedy, “Batch normaliza- tion: Accelerating deep network training by reducing internal covariate shift,” in International conference on machine learning (pmlr, 2015) pp. 448–456

    2015

  81. [89]

    Probabilistic interpretation of feedfor- ward classification network outputs, with relationships to statistical pattern recognition,

    John S Bridle, “Probabilistic interpretation of feedfor- ward classification network outputs, with relationships to statistical pattern recognition,” in Neurocomputing: Al- gorithms, architectures and applications(Springer, 1990) pp. 227–236

    1990

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.