pith. sign in

arxiv: 2604.20180 · v1 · submitted 2026-04-22 · 🪐 quant-ph

Tensor network surrogate models for variational quantum computation

Ryo Watanabe , Dries Sels , Joseph Tindall This is my paper

Pith reviewed 2026-05-10 00:45 UTC · model grok-4.3

classification 🪐 quant-ph
keywords tensor networksvariational quantum algorithmsQAOAIsing spin glasssurrogate models2D latticesquantum simulation
0
0 comments X

The pith

Tensor networks simulate deep QAOA circuits on large 2D lattices and train better variational parameters than small-to-large transfer.

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

The paper establishes that a two-dimensional tensor-network ansatz accurately models variational quantum algorithms such as QAOA on heavy-hexagonal and square lattices for Ising spin-glass problems. This provides an efficient classical benchmark for circuit performance at system sizes an order of magnitude beyond direct quantum simulation. Training parameters directly within the tensor-network framework on larger instances produces lower-energy samples by escaping local minima that arise when parameters are simply transferred from small systems. Entanglement growth stays manageable with moderate bond dimensions, keeping the entire procedure classically tractable. The same tensor-network approach also reveals that parameter-concentration effects persist on square lattices, though at higher sampling cost.

Core claim

We adopt a two-dimensional tensor-network (TN) ansatz to simulate variational quantum algorithms on two-dimensional qubit architectures, demonstrating its capability to accurately simulate deep circuits through the Quantum Approximate Optimization Algorithm (QAOA) applied to Ising spin-glass problems on heavy-hexagonal and square lattices. For heavy-hexagonal problems with up to three-body interactions, parameters trained on small instances and transferred to systems an order of magnitude larger improve the sampled energy distribution only up to intermediate depths, indicating a fundamental limit of parameter concentration as a transfer strategy. By extending the training itself with TN simu

What carries the argument

The two-dimensional tensor-network ansatz with moderate bond dimension, which reproduces the entanglement structure and sampling statistics of QAOA circuits for the Ising problems.

If this is right

  • Parameters transferred from small to large heavy-hexagonal instances improve sampled energies only up to intermediate circuit depths.
  • Performing the variational training itself inside the TN simulation on larger sizes avoids local minima and yields lower energies.
  • Entanglement growth and importance sampling remain controlled enough for classical feasibility at moderate bond dimension.
  • Parameter concentration persists on square lattices, though reliable sampling becomes substantially more expensive.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The surrogate approach could support iterative hybrid loops in which classical TN optimization scales variational training beyond current quantum hardware sizes.
  • Observed depth-dependent limits on parameter transfer imply that optimal circuit depths may need to be chosen with system size in mind.
  • The framework could be applied to other variational algorithms or lattice connectivities to test how widely the surrogate training benefit holds.

Load-bearing premise

The two-dimensional tensor-network ansatz with moderate bond dimension accurately captures the entanglement structure and sampling statistics of the QAOA circuits for the Ising spin-glass problems considered.

What would settle it

Running the same TN simulations at substantially higher bond dimension on the largest lattices and checking whether the obtained energy distributions or trained parameters change would directly test whether the moderate-bond ansatz remains accurate.

Figures

Figures reproduced from arXiv: 2604.20180 by Dries Sels, Joseph Tindall, Ryo Watanabe.

Figure 1
Figure 1. Figure 1: FIG. 1. Schematic diagrams of heavy-hexagonal lattices [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 3
Figure 3. Figure 3: FIG. 3. Dependence on the norm-MPS rank [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: FIG. 4. Bipartite entanglement behavior during the QAOA [PITH_FULL_IMAGE:figures/full_fig_p006_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: FIG. 5. (a)–(c) Histograms of sampled energies for the [PITH_FULL_IMAGE:figures/full_fig_p007_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: FIG. 6. Sampling results for a [PITH_FULL_IMAGE:figures/full_fig_p008_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: FIG. 7. Sampling results for a [PITH_FULL_IMAGE:figures/full_fig_p008_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: FIG. 8. (a) Histogram of sampled energies for the [PITH_FULL_IMAGE:figures/full_fig_p011_8.png] view at source ↗
Figure 9
Figure 9. Figure 9: FIG. 9. Approximate bipartite entanglement entropy [PITH_FULL_IMAGE:figures/full_fig_p012_9.png] view at source ↗
read the original abstract

We adopt a two-dimensional tensor-network (TN) ansatz to simulate variational quantum algorithms on two-dimensional qubit architectures, demonstrating its capability to accurately simulate deep circuits through the Quantum Approximate Optimization Algorithm (QAOA) applied to Ising spin-glass problems on heavy-hexagonal and square lattices. For heavy-hexagonal problems with up to three-body interactions, parameters trained on small instances and transferred to systems an order of magnitude larger improve the sampled energy distribution only up to intermediate depths, indicating a fundamental limit of parameter concentration as a transfer strategy. By extending the training itself with TN simulations on larger system sizes, we avoid local minima and obtain lower-energy samples. Analyses of entanglement growth and importance sampling show that the simulation remains classically feasible with moderate bond dimension. We find that parameter concentration also persists on square lattices, albeit at substantially higher computational cost to perform reliable sampling. Overall, our TN framework not only provides an efficient and controlled framework for benchmarking variational quantum algorithms on two-dimensional lattices, but also serves as an effective surrogate model for training variational algorithms.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

3 major / 2 minor

Summary. The paper adopts a two-dimensional tensor-network (TN) ansatz to simulate variational quantum algorithms on 2D qubit lattices, focusing on QAOA applied to Ising spin-glass problems with up to three-body interactions on heavy-hexagonal and square lattices. It demonstrates that parameters trained on small instances can be transferred to systems an order of magnitude larger, improving sampled energy distributions up to intermediate circuit depths (limited by parameter concentration), while extending TN-based training to larger sizes avoids local minima and yields lower-energy samples. Entanglement growth and importance sampling analyses indicate that moderate bond dimensions keep the simulation classically feasible. The TN framework is positioned as both an efficient benchmarking tool for VQAs on 2D lattices and an effective surrogate model for training variational algorithms.

Significance. If the TN approximations are shown to faithfully reproduce the target QAOA statistics, the work would provide a valuable classical surrogate for studying and optimizing VQAs beyond exact simulability on 2D lattices, with direct implications for understanding parameter concentration and improving variational training. The entanglement and sampling analyses are a strength, as they explicitly tie computational cost to circuit properties and support the feasibility claim.

major comments (3)
  1. [Abstract] Abstract and results sections: The central claim that the 2D TN serves as an effective surrogate model for training variational algorithms requires that moderate bond dimension accurately captures the QAOA entanglement structure and output sampling statistics for the Ising instances considered, yet no quantitative validation metrics (fidelity, energy error, or distribution overlap) against exact statevector simulations are reported even on the smallest lattices where such comparisons are feasible.
  2. [Results] Heavy-hexagonal lattice results: The reported improvement in sampled energy distributions from transferred parameters (only up to intermediate depths) lacks explicit baselines (e.g., random initialization or direct large-system optimization) and error quantification, making it difficult to assess whether the gains are statistically significant or robust to the TN approximation.
  3. [Entanglement analysis] Entanglement growth analysis: While the paper uses entanglement growth to justify moderate bond dimension, specific values of bond dimension employed, convergence tests with increasing bond dimension, and the resulting approximation error on observables are not quantified, which is load-bearing for the feasibility and surrogate claims.
minor comments (2)
  1. [Methods] Clarify the precise TN contraction scheme and importance sampling procedure in the methods section to allow reproducibility.
  2. [Figures] Include error bars or statistical details on the sampled energy distributions in all relevant figures.

Simulated Author's Rebuttal

3 responses · 0 unresolved

We thank the referee for their constructive and detailed comments, which highlight important aspects for strengthening the validation of our tensor-network surrogate approach. We have revised the manuscript to incorporate quantitative metrics, baselines, and convergence analyses as requested, thereby providing stronger support for the claims regarding accuracy and feasibility.

read point-by-point responses

  1. Referee: [Abstract] Abstract and results sections: The central claim that the 2D TN serves as an effective surrogate model for training variational algorithms requires that moderate bond dimension accurately captures the QAOA entanglement structure and output sampling statistics for the Ising instances considered, yet no quantitative validation metrics (fidelity, energy error, or distribution overlap) against exact statevector simulations are reported even on the smallest lattices where such comparisons are feasible.

    Authors: We agree that quantitative validation against exact simulations is necessary to substantiate the surrogate-model claim. In the revised manuscript we add direct comparisons on the smallest lattices (2x2 and 3x3 heavy-hexagonal) where exact state-vector simulation remains feasible. We report state fidelity, relative energy error, and distribution overlap (total-variation distance) between the TN and exact output statistics at representative depths. These metrics confirm that the moderate bond dimensions employed (D=4–6) reproduce the relevant entanglement structure and sampling statistics to high accuracy, thereby supporting the use of the TN as a faithful surrogate. revision: yes

  2. Referee: [Results] Heavy-hexagonal lattice results: The reported improvement in sampled energy distributions from transferred parameters (only up to intermediate depths) lacks explicit baselines (e.g., random initialization or direct large-system optimization) and error quantification, making it difficult to assess whether the gains are statistically significant or robust to the TN approximation.

    Authors: We have added the requested baselines and error quantification. The revised results section now includes (i) random-parameter initialization on the large instances and (ii) direct TN optimization performed on the same large systems. Statistical uncertainties are reported from multiple independent optimization runs (typically 10–20 seeds). The transferred-parameter improvements remain statistically significant up to intermediate depths; beyond this regime the gains saturate, consistent with the parameter-concentration phenomenon we analyze. The added baselines also demonstrate that the observed gains exceed the scale of the TN approximation error. revision: yes

  3. Referee: [Entanglement analysis] Entanglement growth analysis: While the paper uses entanglement growth to justify moderate bond dimension, specific values of bond dimension employed, convergence tests with increasing bond dimension, and the resulting approximation error on observables are not quantified, which is load-bearing for the feasibility and surrogate claims.

    Authors: We have expanded the entanglement and sampling section to provide the missing quantification. We now state the bond dimensions used for each depth and lattice size, include convergence plots of energy and magnetization versus increasing bond dimension (showing saturation at moderate D), and report the absolute approximation error on these observables for the small lattices where exact reference data exist (error <1 % at the D values employed). These additions directly link the entanglement-growth analysis to the computational cost and to the accuracy of the surrogate model. revision: yes

Circularity Check

0 steps flagged

No significant circularity in the TN surrogate model for QAOA

full rationale

The paper's claims rest on numerical simulations of QAOA circuits using a 2D tensor-network ansatz, with direct analyses of entanglement growth, importance sampling, and energy distributions on small-to-large lattices. No equations or steps reduce predictions by construction to fitted parameters, self-citations, or ansatzes imported from the authors' prior work. The surrogate training is performed explicitly via TN contraction on the target system sizes, and feasibility is justified by entanglement scaling rather than any self-referential definition. This is a standard application of TN methods to benchmark and extend variational algorithms, self-contained against external exact methods on small instances.

Axiom & Free-Parameter Ledger

1 free parameters · 1 axioms · 0 invented entities

The approach rests on standard tensor-network approximations for quantum circuit simulation and the assumption that moderate bond dimension suffices for the entanglement generated by the QAOA circuits studied.

free parameters (1)
  • bond dimension
    Chosen according to observed entanglement growth to keep classical simulation tractable; not fitted to target data in the usual sense.
axioms (1)
  • domain assumption Tensor networks with moderate bond dimension can faithfully represent the quantum states produced by QAOA circuits on 2D lattices for the spin-glass instances considered.
    Invoked throughout the simulation and sampling analyses; standard in the tensor-network literature for quantum many-body systems.

pith-pipeline@v0.9.0 · 5473 in / 1473 out tokens · 41131 ms · 2026-05-10T00:45:27.714711+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Forward citations

Cited by 1 Pith paper

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

  1. Local tensor-train surrogates for quantum learning models

    quant-ph 2026-04 unverdicted novelty 7.0

    Local tensor-train surrogates approximate quantum machine learning models via Taylor polynomials and tensor networks, delivering polynomial parameter scaling and explicit generalization bounds controlled by patch radius.

Reference graph

Works this paper leans on

55 extracted references · 55 canonical work pages · cited by 1 Pith paper · 6 internal anchors

  1. [1]

    The data are obtained with bond dimensionχ= 32, amplitude-MPS rankR m =χ, and norm-MPS rankRM = 1

    (b)–(e) Distributions of importance sampling weights ωfor( γ∗,β∗), p = 10,50, and100, respectively; the mean values of the weights are illustrated by vertical dashed lines, and the variances are indicated in the upper-left corner of each panel. The data are obtained with bond dimensionχ= 32, amplitude-MPS rankR m =χ, and norm-MPS rankRM = 1. contrast, Fig...

    work page

  2. [2]

    P. W. Shor, Polynomial-time algorithms for prime factor- ization and discrete logarithms on a quantum computer, SIAM Journal on Computing26, 1484 (1997)

    work page 1997

  3. [3]

    R. P. Feynman, Simulating physics with computers, Inter- national Journal of Theoretical Physics21, 467 (1982)

    work page 1982

  4. [4]

    Lloyd, Universal quantum simulators, Science273, 1073 (1996)

    S. Lloyd, Universal quantum simulators, Science273, 1073 (1996)

    work page 1996

  5. [5]

    P. W. Shor, Scheme for reducing decoherence in quantum computer memory, Phys. Rev. A52, R2493 (1995)

    work page 1995

  6. [6]

    Y. Kim, A. Eddins, S. Anand, K. X. Wei, E. van den Berg, S. Rosenblatt, H. Nayfeh, Y. Wu, M. Zaletel,et al., Evidence for the utility of quantum computing before fault tolerance, Nature618, 500 (2023)

    work page 2023

  7. [7]

    A. D. King, J. Raymond, T. Lanting, R. Harris, A. Zucca, F. Altomare, A. J. Berkley, K. Boothby, S. Ejtemaee, et al., Quantum critical dynamics in a 5,000-qubit pro- grammable spin glass, Nature617, 61 (2023)

    work page 2023

  8. [8]

    A. D. King, A. Nocera, M. M. Rams, J. Dziarmaga, R. Wiersema, W. Bernoudy, J. Raymond, N. Kaushal, N. Heinsdorf,et al., Beyond-classical computation in quan- tum simulation, Science388, 199 (2025)

    work page 2025

  9. [9]

    Robledo-Moreno, M

    J. Robledo-Moreno, M. Motta, H. Haas, A. Javadi-Abhari, P. Jurcevic, W. Kirby, S. Martiel, K. Sharma, S. Sharma, et al., Chemistry beyond the scale of exact diagonalization on a quantum-centric supercomputer, Science Advances 11, eadu9991 (2025)

    work page 2025

  10. [10]

    Tindall, M

    J. Tindall, M. Fishman, E. M. Stoudenmire, and D. Sels, Efficient tensor network simulation of ibm’s eagle kicked ising experiment, PRX Quantum5, 010308 (2024)

    work page 2024

  11. [11]

    Patra, S

    S. Patra, S. S. Jahromi, S. Singh, and R. Orús, Efficient tensor network simulation of ibm’s largest quantum pro- cessors, Phys. Rev. Res.6, 013326 (2024)

    work page 2024

  12. [12]

    Begušić, J

    T. Begušić, J. Gray, and G. K.-L. Chan, Fast and con- verged classical simulations of evidence for the utility of quantum computing before fault tolerance, Science Advances10, eadk4321 (2024)

    work page 2024

  13. [13]

    Tindall, A

    J. Tindall, A. Mello, M. Fishman, M. Stoudenmire, and D. Sels, Dynamics of disordered quantum systems with two- and three-dimensional tensor networks (2025), arXiv:2503.05693 [quant-ph]

    work page internal anchor Pith review arXiv 2025

  14. [14]

    O., Searle, A., Tindall, J

    F. Hasselgren, M. O. Al-Hasso, A. Searle, J. Tindall, and M. von der Leyen, Probabilistic computing optimization of complex spin-glass topologies (2025), arXiv:2510.23419 [cond-mat.dis-nn]

    work page arXiv 2025

  15. [15]

    D. A. Abanin, R. Acharya, L. Aghababaie-Beni, G. Aigeldinger, A. Ajoy, R. Alcaraz, I. Aleiner, T. I. An- dersen, M. Ansmann,et al., Observation of constructive interference at the edge of quantum ergodicity, Nature 646, 825 (2025)

    work page 2025

  16. [16]

    A Quantum Approximate Optimization Algorithm

    E. Farhi, J. Goldstone, and S. Gutmann, A quantum ap- proximate optimization algorithm (2014), arXiv:1411.4028 [quant-ph]

    work page internal anchor Pith review Pith/arXiv arXiv 2014

  17. [17]

    Cerezo, A

    M. Cerezo, A. Arrasmith, R. Babbush, S. C. Benjamin, S. Endo, K. Fujii, J. R. McClean, K. Mitarai, X. Yuan, et al., Variational quantum algorithms, Nature Reviews Physics3, 625 (2021)

    work page 2021

  18. [18]

    Peruzzo, J

    A. Peruzzo, J. McClean, P. Shadbolt, M.-H. Yung, X.-Q. Zhou, P. J. Love, A. Aspuru-Guzik, and J. L. O’Brien, A variational eigenvalue solver on a photonic quantum processor, Nature Communications5, 4213 (2014)

    work page 2014

  19. [19]

    Mitarai, M

    K. Mitarai, M. Negoro, M. Kitagawa, and K. Fujii, Quan- tum circuit learning, Phys. Rev. A98, 032309 (2018)

    work page 2018

  20. [20]

    Farhi, J

    E. Farhi, J. Goldstone, S. Gutmann, J. Lapan, A. Lund- gren, and D. Preda, A quantum adiabatic evolution al- gorithm applied to random instances of an np-complete problem, Science292, 472 (2001)

    work page 2001

  21. [21]

    Begušić, K

    T. Begušić, K. Hejazi, and G. K.-L. Chan, Simulating quantum circuit expectation values by clifford perturba- tion theory, The Journal of Chemical Physics162, 154110 (2025)

    work page 2025

  22. [22]

    M. S. Rudolph, E. Fontana, Z. Holmes, and L. Cincio, Classical surrogate simulation of quantum systems with lowesa (2023), arXiv:2308.09109 [quant-ph]

    work page arXiv 2023

  23. [23]

    Y. Shao, F. Wei, S. Cheng, and Z. Liu, Simulating noisy variational quantum algorithms: A polynomial approach, Phys. Rev. Lett.133, 120603 (2024)

    work page 2024

  24. [24]

    Quantum Convolutional Neural Networks are Effectively Classically Simulable

    P. Bermejo, P. Braccia, M. S. Rudolph, Z. Holmes, L. Cincio, and M. Cerezo, Quantum convolutional neu- ral networks are effectively classically simulable (2026), arXiv:2408.12739 [quant-ph]

    work page internal anchor Pith review Pith/arXiv arXiv 2026

  25. [25]

    The Computational Complexity of Linear Optics

    S. Aaronson and A. Arkhipov, The computational com- plexity of linear optics (2010), arXiv:1011.3245 [quant-ph]

    work page Pith review arXiv 2010

  26. [26]

    A. P. Lund, M. J. Bremner, and T. C. Ralph, Quan- tum sampling problems, bosonsampling and quantum supremacy, npj Quantum Information3, 15 (2017)

    work page 2017

  27. [27]

    Arute, K

    F. Arute, K. Arya, R. Babbush, D. Bacon, J. C. Bardin, R. Barends, R. Biswas, S. Boixo, F. G. S. L. Brandao, et al., Quantum supremacy using a programmable super- conducting processor, Nature574, 505 (2019)

    work page 2019

  28. [28]

    Zlokapa, B

    A. Zlokapa, B. Villalonga, S. Boixo, and D. A. Lidar, Boundaries of quantum supremacy via random circuit sampling, npj Quantum Information9, 36 (2023)

    work page 2023

  29. [29]

    A.Morvan, B.Villalonga, X.Mi, S.Mandrà, A.Bengtsson, P. V. Klimov, Z. Chen, S. Hong, C. Erickson,et al., Phase transitions in random circuit sampling, Nature634, 328 (2024)

    work page 2024

  30. [30]

    Pelofske, A

    E. Pelofske, A. Bärtschi, and S. Eidenbenz, Quantum annealing vs. qaoa: 127 qubit higher-order ising prob- lems on nisq computers, inHigh Performance Computing (Springer Nature Switzerland, 2023) pp. 240–258

    work page 2023

  31. [31]

    Pelofske, A

    E. Pelofske, A. Bärtschi, and S. Eidenbenz, Short-depth qaoa circuits and quantum annealing on higher-order ising models, npj Quantum Information10, 30 (2024)

    work page 2024

  32. [32]

    Manca, S

    V. Akshay, D. Rabinovich, E. Campos, and J. Bia- monte, Parameter concentrations in quantum approxi- mate optimization, Physical Review A104, 10.1103/phys- reva.104.l010401 (2021)

    work page doi:10.1103/phys- 2021

  33. [33]

    A. Apte, S. H. Sureshbabu, R. Shaydulin, S. Boulebnane, Z. He, D. Herman, J. Sud, and M. Pistoia, Iterative inter- polation schedules for quantum approximate optimization algorithm (2025), arXiv:2504.01694 [quant-ph]

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  34. [34]

    Tindall and M

    J. Tindall and M. Fishman, Gauging tensor networks with belief propagation, SciPost Phys.15, 222 (2023)

    work page 2023

  35. [35]

    Alkabetz and I

    R. Alkabetz and I. Arad, Tensor networks contraction and the belief propagation algorithm, Phys. Rev. Res.3, 023073 (2021)

    work page 2021

  36. [36]

    Vieijra, J

    T. Vieijra, J. Haegeman, F. Verstraete, and L. Van- derstraeten, Direct sampling of projected entangled-pair states, Phys. Rev. B104, 235141 (2021)

    work page 2021

  37. [37]

    M. S. Rudolph and J. Tindall, Simulating and sampling from quantum circuits with 2d tensor networks (2025), arXiv:2507.11424 [quant-ph]

    work page arXiv 2025

  38. [38]

    Pelofske, A

    E. Pelofske, A. Bärtschi, L. Cincio, J. Golden, and S. Ei- denbenz, Scaling whole-chip qaoa for higher-order ising 11 spin glass models on heavy-hex graphs, npj Quantum Information10, 1 (2024)

    work page 2024

  39. [39]

    Sachdeva, G

    N. Sachdeva, G. S. Hartnett, S. Maity, S. Marsh, Y. Wang, A. Winick, R. Dougherty, D. Canuto, Y. Q. Chong, et al.,Quantumoptimizationusinga127-qubitgate-model ibm quantum computer can outperform quantum anneal- ers for nontrivial binary optimization problems (2024), arXiv:2406.01743 [quant-ph]

    work page internal anchor Pith review arXiv 2024

  40. [40]

    Acharya, D

    R. Acharya, D. A. Abanin, L. Aghababaie-Beni, I. Aleiner, T. I. Andersen, M. Ansmann, F. Arute, K. Arya, A. Asfaw, et al., Quantum error correction below the surface code threshold, Nature638, 920 (2025)

    work page 2025

  41. [41]

    Farhi, J

    E. Farhi, J. Goldstone, S. Gutmann, and L. Zhou, The Quantum Approximate Optimization Algorithm and the Sherrington-Kirkpatrick Model at Infinite Size, Quantum 6, 759 (2022)

    work page 2022

  42. [42]

    F. G. S. L. Brandao, M. Broughton, E. Farhi, S. Gut- mann, and H. Neven, For fixed control parameters the quantum approximate optimization algorithm’s objective function value concentrates for typical instances (2018), arXiv:1812.04170 [quant-ph]

    work page Pith review arXiv 2018

  43. [43]

    Boulebnane and A

    S. Boulebnane and A. Montanaro, Solving boolean sat- isfiability problems with the quantum approximate opti- mization algorithm (2022), arXiv:2208.06909 [quant-ph]

    work page arXiv 2022

  44. [44]

    M. J. Powell, The bobyqa algorithm for bound constrained optimization without derivatives, Cambridge NA Report NA2009/06, University of Cambridge, Cambridge26, 26 (2009)

    work page 2009

  45. [45]

    Leifer and D

    M. Leifer and D. Poulin, Quantum graphical models and belief propagation, Annals of Physics323, 1899 (2008)

    work page 2008

  46. [46]

    Sahu and B

    S. Sahu and B. Swingle, Efficient tensor network simu- lation of quantum many-body physics on sparse graphs (2022), arXiv:2206.04701 [quant-ph]

    work page arXiv 2022

  47. [47]

    C. Guo, D. Poletti, and I. Arad, Block belief propagation algorithm for two-dimensional tensor networks, Phys. Rev. B108, 125111 (2023)

    work page 2023

  48. [48]

    S. S. Jahromi and R. Orús, Universal tensor-network algorithm for any infinite lattice, Phys. Rev. B99, 195105 (2019)

    work page 2019

  49. [49]

    H. C. Jiang, Z. Y. Weng, and T. Xiang, Accurate determi- nation of tensor network state of quantum lattice models in two dimensions, Phys. Rev. Lett.101, 090603 (2008)

    work page 2008

  50. [50]

    S. T. Tokdar and R. E. Kass, Importance sampling: a review, WIREs Computational Statistics2, 54 (2010)

    work page 2010

  51. [51]

    Pelofske, M

    E. Pelofske, M. Rams, A. Bärtschi, P. Czarnik, P. Brac- cia, L. Cincio, and S. Eidenbenz, Evaluating the lim- its of qaoa parameter transfer at high-rounds on sparse ising models with geometrically local cubic terms (2025), arXiv:2509.13528 [quant-ph]

    work page internal anchor Pith review arXiv 2025

  52. [52]

    Lerch, R

    S. Lerch, R. Puig, M. S. Rudolph, A. Angrisani, T. Jones, M. Cerezo, S. Thanasilp, and Z. Holmes, Efficient quantum-enhanced classical simulation for patches of quantum landscapes (2024), arXiv:2411.19896 [quant-ph]

    work page arXiv 2024

  53. [53]

    Watanabe, J

    R. Watanabe, J. Tindall, S. Miyakoshi, and H. Ueda, Quantum-inspired algorithm for classical spin hamil- tonians based on matrix product operators (2026), arXiv:2602.05224 [quant-ph]

    work page arXiv 2026

  54. [54]

    C. Fan, M. Shen, Z. Nussinov, Z. Liu, Y. Sun, and Y.-Y. Liu, Searching for spin glass ground states through deep reinforcement learning, Nature Communications14, 725 (2023)

    work page 2023

  55. [55]

    opti- mal

    M. Fishman, S. R. White, and E. M. Stoudenmire, The ITensor Software Library for Tensor Network Calculations, SciPost Phys. Codebases , 4 (2022). FIG. 8. (a)Histogram ofsampled energiesfor the27-qubit0th ibm_genevainstance in Ref. [29], where the minimum energy is C(z∗) =−42. The labels are the same as in Fig. 2(a). (b)-(c) Distributions of importance sam...

    work page 2022