Pith. sign in

REVIEW 3 major objections 4 minor 54 references

Advantages of density in tensor network geometries for gradient based training

T0 review · 3 major / 4 minor · reviewed 2026-08-11 · deepseek-v4-flash

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

Pith's one-line read Denser tensor network geometries train faster and fit random quantum states better than sparse ones, even when the sparse network has enough memory to represent the target exactly.

desk verdict A useful empirical study with a genuine compactification trick, but the central density claim is undercut by an undisclosed surrogate geometry in the target-generation step. read the letter →

arxiv 2412.17497 v1 pith:OIXFSXDX submitted 2024-12-23 quant-ph

Sergi Masot-Llima , Artur Garcia-Saez This is my paper · ORCID
classification quant-ph
keywords tensornetworksgradient-basedtraininggeometrybonddimensioninfidelitybarrenplateausmatrixproductstatescompact
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

This paper studies gradient-based training of tensor networks used as machine-learning models. It trains several network geometries, from a one-dimensional chain to star-shaped trees and a two-dimensional grid, to encode the same random quantum state, and measures the final infidelity. The central claim is that denser connectivity, meaning shorter paths between tensors, leads to lower infidelity and higher success rates, and that this advantage persists even when the bond dimension is large enough to represent the target state exactly. The authors conclude that how a tensor network connects the information it stores matters more than how much information it can store. They also introduce a compactification procedure for tree-shaped networks that reduces memory and improves training, and they document practical speedups from GPU acceleration on a large supercomputer for the largest system sizes.

What carries the argument

The surrogate training protocol is the central tool: a random tensor network with a chosen geometry and bond dimension is contracted into a single dense tensor, producing a target quantum state with controlled entanglement but no visible network structure; the model tensor network is then trained against this single surrogate through one contraction per step rather than through statistical sampling. This isolates the effect of geometry from sampling artifacts. The paper also uses a compactification procedure, contracting the outer branches of a tree network until a bond of maximal dimension is reached, which lowers total memory while slightly increasing contraction cost; node-to-node distance in the network graph serves as the operational measure of density.

What would settle it

Generate the same training benchmark using targets that are exactly Haar-random statevectors, and separately using targets produced by contracting an explicitly sparse surrogate network such as a low-bond-dimension MPS, then compare dense and sparse model geometries; if sparse models match dense ones on either target family, the density advantage is an artifact of the target-generation procedure rather than a property of gradient training.

Watch

Extended reading notes

Core claim

For gradient-based training of tensor networks to encode random quantum states, network geometry dominates representational capacity. When the target state is generated with a fixed amount of controlled entanglement and the model network is trained by minimizing a log-infidelity loss with L-BFGS-B and automatic differentiation, the final infidelity decreases as the network becomes more densely connected, as measured by the maximum distance between nodes. Dense structures such as the star geometry and PEPS reach low infidelity at low bond dimension, while MPS and large-diameter trees need to reach the maximal Schmidt bond dimension and, in some cases, still fail to converge. Increasing the bond dimension beyond what is needed for exact representation does not improve the infidelity and, for sparse structures, lowers the fraction of successful trainings. The compact version of a tree network, obtained by contracting peripheral bonds with dimension below the maximum, both reduces total memory and trains to better infidelity in fewer iterations.

Load-bearing premise

The surrogate targets are meant to be representative random states with controlled entanglement, but the paper does not specify the geometry or bond dimension of the surrogate network used to generate the main targets, so the density advantage could partly reflect the hidden structure of the targets themselves.

Editorial extensions

If this is right

  • If geometry rather than capacity controls trainability, then practitioners should choose tensor-network ansatze whose connectivity mirrors the presumed correlation structure of the target data, rather than merely increasing bond dimension.
  • Barren-plateau effects in tensor-network training should be viewed as a continuous function of network density: MPS sits at the sparse extreme, while star and PEPS geometries are largely immune in the tested regime.
  • The compactification procedure gives a memory reduction for tree tensor networks with many leaves, and in the tested cases it also improves training quality and speed, so it can be used to raise the achievable precision under a fixed memory budget.
  • Because the best geometry depends on the entanglement structure of the task, automatic structural optimization of tree tensor networks, referenced in the paper, becomes a natural complement to density-oriented training.
  • For large enough system sizes, GPU-accelerated nodes yield a time and energy advantage over CPU-only runs, provided double precision is used; single precision degrades the achievable infidelity by one to two orders of magnitude.

Reading between the lines

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

  • A testable extension would be to generate surrogate targets from deliberately sparse networks, such as low-bond-dimension MPS, and check whether dense geometries still win; if the density advantage disappears for sparse targets, the benefit is tied to matching geometry to target entanglement rather than to density per se.
  • The result suggests that for classical machine-learning tensor train (MPS-style) models, reshaping the data layout to shorten effective paths, for example through feature rearrangement, could improve gradient training without changing the total parameter count.
  • The observed failure of MPS at maximal bond dimension hints that the loss landscape, not representational power, is the bottleneck; quantifying the spectral properties of the Hessian for different geometries would make the mechanism precise.
  • Because compactification effectively increases density, the same procedure could be applied to loop-containing geometries such as PEPS, though the paper only demonstrates it for tree networks.
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

3 major / 4 minor

Summary. The paper studies gradient-based training of tensor networks with different geometries (MPS, tree-like structures, Star networks, and PEPS) whose task is to encode random quantum states prepared by contracting a random surrogate tensor network into a single dense tensor. The main empirical claim is that more densely connected geometries reach lower infidelities and higher success rates than sparser ones, even when the bond dimension is large enough to represent the target exactly, and that connectivity matters more than total information capacity. The paper also introduces a 'compactification' procedure for tree tensor networks, reports that compact versions train better and use less memory, and presents HPC/GPU timing comparisons. The authors conclude that TN geometry affects trainability beyond representational capacity and connect the density trend to the appearance of barren plateaus.

Significance. If the empirical trend is robust, the paper provides a useful design principle for tensor-network machine learning: for gradient-based training, connectivity can be more important than bond dimension or total parameter count. The compactification observation is a clean, analytically motivated memory-reduction trick for tree networks, and the link to the barren-plateau literature is timely. The main limitation is that the central numerical claim rests on an underspecified benchmark: the surrogate geometry used to generate the target states is never stated for the main experiments, and the reported comparisons do not control for total parameter count or include error bars. The strength of the paper is therefore conditional on additional specification and control experiments, but the core question is well posed and the compactification derivation is a positive contribution.

major comments (3)
  1. [Sec. II C and Sec. III] The surrogate geometry used to generate the target states is never specified for the main n=12 and n=16 experiments. The text says targets are prepared by contracting a random tensor network 'with a given structure ansatz' and then explicitly defers 'the study of encoding different structures into the surrogate' to future work. If the surrogate was a dense geometry, the dense models have an unfair prior match; if the surrogate geometry was matched to each model geometry, different models are trained on different targets and the cross-geometry comparison in Figs. 4 and 5 is not well defined. The manuscript must state the surrogate geometry (and its bond dimension) for every experiment, confirm that it is held fixed across all model geometries, and ideally repeat the main comparison with several surrogate geometries to show that the density trend is not an artifact of the target generation procedure. No code or data are released to resolve this ambiguity.
  2. [Figs. 4 and 5] The central infidelity-versus-density trend is reported without error bars or seed statistics. Figure 4 shows single-point infidelities for n=12 and Figure 5a shows single-point infidelities for n=16, while Figure 5b reports only the number of trainings crossing a 10^-3 threshold and Figure 7 reports best and median runs for a single n=16 example. Because the main claim is comparative and the text emphasizes that MPS can fail even after 400 trials, the paper should report the distribution of infidelities over random initializations for each geometry and each bond dimension, including the number of trials, the median, and a dispersion measure. Without this, the reader cannot judge whether the density ordering is statistically meaningful or driven by a few lucky runs.
  3. [Sec. IV and Sec. V] The main comparisons fix the bond dimension (or the size of the largest tensor) rather than the total number of parameters. At fixed bond dimension, denser geometries contain more tensor entries, so the observed advantage could be explained by parameter count rather than by connectivity. The paper's central conclusion that 'the way the TN structure connects the information that it stores is much more important than the total amount of information it can store' requires at least one control experiment with matched total parameter count, e.g., reducing the bond dimension of dense structures so that their total number of trainable parameters is comparable to that of a sparser geometry. Without such a control, the stated distinction between connectivity and information capacity is not established.
minor comments (4)
  1. [Eq. (3)] The fidelity F is defined as an overlap rather than as the absolute square typically used for pure states; please clarify whether the states are taken to be real or whether F denotes the modulus of the overlap, since the loss function in Eq. (4) takes a logarithm of F.
  2. [Sec. II B and Fig. 1] The 'density' of a geometry is quantified by the maximum node distance, but the graph-theoretic definition of distance for PEPS (which contains loops) is only described informally in the figure caption; a precise definition would help reproducibility.
  3. [Sec. III] The hardware description contains an incomplete specification: the accelerated-node CPU is described as 'two Intel Sapphire Rapids 8460Y+ with 40 cores at' with the clock speed missing.
  4. [Sec. V] There are typographical errors such as 'compactification ot tree TN' and 'repreentation' in Sec. II B; a careful proofread is needed.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: the density–trainability trend is an empirical training outcome, not a quantity forced by construction; the unspecified surrogate geometry is a reproducibility caveat, not a circular step.

full rationale

The derivation chain is self-contained: the main result is an empirical comparison of infidelities obtained by training fixed TN geometries against a surrogate-generated target, with no fitted parameters and no quantity defined in terms of the outcome it is used to explain. The compactification bound in Sec. II D follows from explicit tensor-dimension counting; the barren-plateau discussion in Sec. II E invokes independent literature [15,16]; and the only self-citation [42] concerns HPC memory tradeoffs and is not load-bearing. The one genuine weakness is that Sec. II C leaves 'for future work the study of encoding different structures into the surrogate to see if we can recover them,' and Section III does not state which surrogate geometry produced the n=12 and n=16 targets; if that geometry coincided with a compared dense ansatz, the benchmark could in principle be biased. But the paper never asserts that the surrogate is dense, and the density trend is not an identity with any input: it is a measured outcome that could have gone the other way. Under the rule that speculation about an unstated surrogate geometry is not circularity, no step in the paper's argument reduces by construction to its own inputs.

Assumptions & free parameters 3 free parameters · 6 assumptions · 0 invented entities

The paper's central trend is empirical and depends on several design choices: the loss function, the success threshold, and the fixed-bond-dimension comparison. The most important unvalidated premise is that surrogate-generated targets do not favor dense geometries. No free constants are fitted to produce the main result.

free parameters (3)
  • Success threshold = 10^{-3}
    Hand-chosen to define successful trainings in Fig. 5b; affects the reported success rates.
  • Number of training trials = 100 to 400
    Trial counts vary by experiment (Fig. 5 uses 400, Fig. 7 uses 100); best and median statistics depend on this choice.
  • Loss function = L = (log(F) - 1)^2
    Selected after testing L = I^2 gave worse results (Sec. III); the objective is a design choice, not a fitted constant.
assumptions (6)
  • standard math Schmidt decomposition: any bipartition of an n-qubit state has Schmidt rank at most 2^{n/2}, so a tensor network with bond dimension 2^{n/2} can represent any state exactly.
    Used to justify that chi = chi_S gives sufficient representational capacity (Sec. II A, Eq. 1, and Fig. 4 caption).
  • standard math A tree tensor network with all virtual bonds of dimension at least the Schmidt ranks can represent any target state.
    Relies on the tree structure and the capacity bound; invoked to claim that training failure at chi = chi_S is an optimization issue, not a capacity issue.
  • domain assumption Surrogate targets generated by contracting a random tensor network with fixed chi are representative random quantum states with controlled entanglement and do not systematically favor dense model geometries.
    Central to the benchmark design (Sec. II C, Fig. 2); the paper does not specify the surrogate geometry used in the main experiments.
  • ad hoc to paper Comparing geometries at fixed bond dimension, rather than fixed total parameter count, isolates the effect of network density.
    The conclusion that connectivity matters more than information content (Sec. V) depends on this comparison choice.
  • domain assumption L-BFGS-B with automatic differentiation and the selected loss function is an adequate optimizer for this training task.
    Optimizer and loss are fixed by the authors (Sec. III); no systematic comparison with other optimizers is reported.
  • domain assumption The reported best and median runs over tens to hundreds of trials represent the training behavior of each geometry.
    Used in Figs. 4-7; trial counts vary and initialization details are not given.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Advantages of density in tensor network geometries for gradient based training." pith.science (2026). https://pith.science/paper/OIXFSXDX

@misc{pith2026241217497,
  author       = {Pith},
  title        = {Pith review of: Advantages of density in tensor network geometries for gradient based training},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/OIXFSXDX}},
  note         = {Machine review of arXiv:2412.17497}
}
read the original abstract

Tensor networks are a very powerful data structure tool originating from quantum system simulations. In recent years, they have seen increased use in machine learning, mostly in trainings with gradient-based techniques, due to their flexibility and performance exploiting hardware acceleration. As ans\"atze, tensor networks can be used with flexible geometries, and it is known that for highly regular ones their dimensionality has a large impact in performance and representation power. For heterogeneous structures, however, these effects are not completely characterized. In this article, we train tensor networks with different geometries to encode a random quantum state, and see that densely connected structures achieve better infidelities than more sparse structures, with higher success rates and less time. Additionally, we give some general insight on how to improve memory requirements on these sparse structures and its impact on the trainings. Finally, as we use HPC resources for the calculations, we discuss the requirements for this approach and showcase performance improvements with GPU acceleration on a last-generation supercomputer.

Figures

Figures reproduced from arXiv: 2412.17497 by the authors.

Figure 1
Figure 1. FIG. 1. TN geometries used in this work, with a few tree-like [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. FIG. 2. Usual training process of a TN model using samples [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. This method can be trivially generalized to any [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figures from the paper (5 more)
Figure 3
Figure 3. Figure 3: FIG. 3. Example using an MPS of the “compact” TN approach, where bonds smaller than [PITH_FULL_IMAGE:figures/full_fig_p005_3.png]
Figure 4
Figure 4. Figure 4: FIG. 4. Training performance measured with infidelity as a [PITH_FULL_IMAGE:figures/full_fig_p006_4.png]
Figure 5
Figure 5. Figure 5: FIG. 5. Training performance measured with infidelity as a [PITH_FULL_IMAGE:figures/full_fig_p007_5.png]
Figure 6
Figure 6. Figure 6: FIG. 6. Training performance measured with infidelity as a [PITH_FULL_IMAGE:figures/full_fig_p007_6.png]
Figure 8
Figure 8. Figure 8: FIG. 8. Time cost of the training in this work for different [PITH_FULL_IMAGE:figures/full_fig_p008_8.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

54 extracted references · 30 canonical work pages

  1. [1]

    J. I. Cirac, D. Perez-Garcia, N. Schuch, and F. Verstraete, Matrix Product States and Projected Entangled Pair States: Concepts, Symmetries, and Theorems, Reviews of Modern Physics 93, 045003 (2021), 2011.12127

    arXiv 2021

  2. [2]

    I. L. Markov and Y. Shi, Simulating quantum computa- tion by contracting tensor networks, SIAM Journal on Computing 38, 963–981 (2008)

    work page 2008

  3. [3]

    Gray and G

    J. Gray and G. K.-L. Chan, Hyperoptimized approxi- mate contraction of tensor networks with arbitrary geom- etry, Physical Review X 14, 10.1103/physrevx.14.011009 (2024)

    doi:10.1103/physrevx.14.011009 2024

  4. [4]

    Vidal, Efficient classical simulation of slightly entan- gled quantum computations, Physical Review Letters 91, 10.1103/physrevlett.91.147902 (2003)

    G. Vidal, Efficient classical simulation of slightly entan- gled quantum computations, Physical Review Letters 91, 10.1103/physrevlett.91.147902 (2003)

    doi:10.1103/physrevlett.91.147902 2003

  5. [5]

    I. P. McCulloch, Infinite size density matrix renormaliza- tion group, revisited (2008), 0804.2509

    arXiv 2008

  6. [6]

    Vidal, Entanglement Renormalization, Physical Re- view Letters 99, 220405 (2007)

    G. Vidal, Entanglement Renormalization, Physical Re- view Letters 99, 220405 (2007)

    work page 2007

  7. [7]

    Verstraete and J

    F. Verstraete and J. I. Cirac, Renormalization algorithms for quantum-many body systems in two and higher dimen- sions (2004), arXiv:cond-mat/0407066 [cond-mat.str-el]

    arXiv 2004

  8. [8]

    Beguˇ si´ c, J

    T. Beguˇ si´ c, 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 Advances 10, 10.1126/sciadv.adk4321 (2024)

    doi:10.1126/sciadv.adk4321 2024

Show all 54 references
  1. [9]

    Verstraete, V

    F. Verstraete, V. Murg, and J. Cirac, Matrix product states, projected entangled pair states, and variational renormalization group methods for quantum spin systems, Advances in Physics 57, 143–224 (2008)

    2008

  2. [10]

    Schollw¨ ock, The density-matrix renormalization group in the age of matrix product states, Annals of Physics 326, 96–192 (2011)

    U. Schollw¨ ock, The density-matrix renormalization group in the age of matrix product states, Annals of Physics 326, 96–192 (2011)

    2011

  3. [11]

    E. M. Stoudenmire and D. J. Schwab, Supervised learning with quantum-inspired tensor networks (2017), arXiv:1605.05775 [stat.ML]

    2017 arXiv

  4. [12]

    J. Wang, C. Roberts, G. Vidal, and S. Leichenauer, Anomaly detection with tensor networks (2020), arXiv:2006.02516 [cs.LG]

    2020 arXiv

  5. [13]

    Evenbly, A practical guide to the numerical implemen- tation of tensor networks i: Contractions, decompositions and gauge freedom (2022), arXiv:2202.02138 [quant-ph]

    G. Evenbly, A practical guide to the numerical implemen- tation of tensor networks i: Contractions, decompositions and gauge freedom (2022), arXiv:2202.02138 [quant-ph]

    2022 arXiv

  6. [14]

    Zhao, R.-G

    Y.-Q. Zhao, R.-G. Li, J.-Z. Jiang, C. Li, H.-Z. Li, E.-D. Wang, W.-F. Gong, X. Zhang, and Z.-Q. Wei, Simulation of quantum computing on classical supercomputers with tensor-network edge cutting, Phys. Rev. A 104, 032603 (2021)

    2021

  7. [15]

    Cervero Mart ´ ın, K

    E. Cervero Mart ´ ın, K. Plekhanov, and M. Lubasch, Bar- ren plateaus in quantum tensor network optimization, Quantum 7, 974 (2023)

    2023

  8. [16]

    size of largest tensor

    – perhaps unsurprisingly considering the connection of the structures to quantum computing. The structure of the TN in use affects greatly the appearance of plateaus [15, 48], with MPS being the most susceptible. This ef- fect should be considered together with our findings on...

    2022

  9. [17]

    Menczer, M

    A. Menczer, M. van Damme, A. Rask, L. Hunting- ton, J. Hammond, S. S. Xantheas, M. Ganahl, and ¨Ors Legeza, Parallel implementation of the density ma- trix renormalization group method achieving a quarter petaflops performance on a single dgx-h100 gpu node (2024), arXiv:2407.0...

    2024 arXiv

  10. [18]

    Liu, L.-W

    Z. Liu, L.-W. Yu, L.-M. Duan, and D.-L. Deng, Presence and absence of barren plateaus in tensor-network based machine learning, Phys. Rev. Lett. 129, 270501 (2022)

    2022

  11. [19]

    Horodecki, P

    R. Horodecki, P. Horodecki, M. Horodecki, and K. Horodecki, Quantum entanglement, Rev. Mod. Phys. 10 81, 865 (2009)

    2009

  12. [20]

    Benenti, G

    G. Benenti, G. Casati, I. Davide Rossini (Universita di Pisa, and G. Strini), Entanglement and non-classical cor- relations, in Principles of Quantum Computation and Information (World Scientifics, 2018) Chap. Chapter 6, pp. 241–286

    2018

  13. [21]

    M. A. Nielsen and I. L. Chuang, Introduction to quantum mechanics, in Quantum Computation and Quantum Infor- mation: 10th Anniversary Edition(Cambridge University Press, 2010) pp. 60–119

    2010

  14. [22]

    Antenna” structure that reduces the distance between sites with respect to an MPS, with- out allowing tensors with more than 3 virtual indices, whereas“Balanced

    example to illustrate this, which is a 1-dimensional chain and thus the simplest example of a non-trivial tensor network. Here, the high-dimensional tensor T with n physical bonds is equivalent to the contraction of n tensors as: T i1...in = X k1...kn−1 (T1)i1 k1 (T2)i2k1 k2 ....

  15. [23]

    Horodecki, Lukasz Rudnicki, and K

    P. Horodecki, Lukasz Rudnicki, and K. ˙Zyczkowski, Multi- partite entanglement (2024), arXiv:2409.04566 [quant-ph]

    2024 arXiv

  16. [24]

    Orus, A Practical Introduction to Tensor Networks: Matrix Product States and Projected Entangled Pair States, Annals of Physics 349, 117 (2014), 1306.2164

    R. Orus, A Practical Introduction to Tensor Networks: Matrix Product States and Projected Entangled Pair States, Annals of Physics 349, 117 (2014), 1306.2164

    2014 arXiv

  17. [25]

    Sharma, P

    M. Sharma, P. P. Markopoulos, E. Saber, M. S. Asif, and A. Prater-Bennette, Convolutional auto-encoder with tensor-train factorization, in 2021 IEEE/CVF Interna- tional Conference on Computer Vision Workshops (IC- CVW) (2021) pp. 198–206

    2021

  18. [26]

    Z. Su, Y. Zhou, F. Mo, and J. G. Simonsen, Language modeling using tensor trains (2024), arXiv:2405.04590 [cs.CL]

    2024 arXiv

  19. [27]

    I. V. Oseledets, Tensor-train decomposition, SIAM Journal on Scientific Computing 33, 2295 (2011), https://doi.org/10.1137/090752286

    2011 doi

  20. [28]

    Or´ us, Tensor networks for complex quantum systems, Nature Reviews Physics 1, 538–550 (2019)

    R. Or´ us, Tensor networks for complex quantum systems, Nature Reviews Physics 1, 538–550 (2019)

    2019

  21. [29]

    Affleck, T

    I. Affleck, T. Kennedy, E. H. Lieb, and H. Tasaki, Rigorous results on valence-bond ground states in antiferromagnets, Physical Review Letters 59, 799 (1987)

    1987

  22. [30]

    Evenbly and G

    G. Evenbly and G. Vidal, Tensor Network States and Geometry, Journal of Statistical Physics 145, 891 (2011)

    2011

  23. [31]

    Eisert, M

    J. Eisert, M. Cramer, and M. B. Plenio, Area laws for the entanglement entropy - a review, Reviews of Modern Physics 82, 277 (2010), 0808.3773

    2010 arXiv

  24. [32]

    Shi, L.-M

    Y.-Y. Shi, L.-M. Duan, and G. Vidal, Classical simula- tion of quantum many-body systems with a tree tensor network, Phys. Rev. A 74, 022320 (2006)

    2006

  25. [33]

    Vidal, Entanglement renormalization, Phys

    G. Vidal, Entanglement renormalization, Phys. Rev. Lett. 99, 220405 (2007)

    2007

  26. [34]

    Okunishi, H

    K. Okunishi, H. Ueda, and T. Nishino, Entan- glement bipartitioning and tree tensor networks, Progress of Theoretical and Experimental Physics 2023, 023A02 (2023), https://academic.oup.com/ptep/article- pdf/2023/2/023A02/49294184/ptad018.pdf

    2023

  27. [35]

    Haghshenas, M

    R. Haghshenas, M. J. O’Rourke, and G. K.-L. Chan, Con- version of projected entangled pair states into a canonical form, Physical Review B 100, 054404 (2019)

    2019

  28. [36]

    Hyatt and E

    K. Hyatt and E. M. Stoudenmire, Dmrg approach to optimizing two-dimensional tensor networks (2020), arXiv:1908.08833 [cond-mat.str-el]

    2020 arXiv

  29. [37]

    Hikihara, H

    T. Hikihara, H. Ueda, K. Okunishi, K. Harada, and T. Nishino, Automatic structural optimization of tree tensor networks, Phys. Rev. Res. 5, 013031 (2023)

    2023

  30. [38]

    Hikihara, H

    T. Hikihara, H. Ueda, K. Okunishi, K. Harada, and T. Nishino, Visualization of entanglement geometry by structural optimization of tree tensor network (2024), arXiv:2401.16000 [cond-mat.stat-mech]

    2024

  31. [39]

    J. Lu, P. Gong, J. Ye, J. Zhang, and C. Zhang, A survey on machine learning from few samples (2023), arXiv:2009.02653 [cs.LG]

    2023 arXiv

  32. [40]

    Fuksa, M

    J. Fuksa, M. G¨ otte, I. Roth, and J. Eisert, A quan- tum inspired approach to learning dynamical laws from data—block-sparsity and gauge-mediated weight sharing, Machine Learning: Science and Technology 5, 025064 (2024)

    2024

  33. [41]

    Cerezo, G

    M. Cerezo, G. Verdon, H.-Y. Huang, L. Cincio, and P. J. Coles, Challenges and opportunities in quantum ma- chine learning, Nature Computational Science 2, 567–576 (2022)

    2022

  34. [42]

    X.-C. Wu, S. Di, F. Cappello, H. Finkel, Y. Alexeev, and F. T. Chong, Memory-efficient quantum circuit simulation by using lossy data compression (2018), arXiv:1811.05630 [quant-ph]

    2018 arXiv

  35. [43]

    F. Pan, H. Gu, L. Kuang, B. Liu, and P. Zhang, Efficient quantum circuit simulation by tensor network methods on modern gpus (2024), arXiv:2310.03978 [quant-ph]

    2024 arXiv

  36. [44]

    Sanchez-Ramirez, J

    S. Sanchez-Ramirez, J. Conejero, F. Lordan, A. Quer- alt, T. Cortes, R. M. Badia, and A. Garcia-Saez, Rosnet: A block tensor algebra library for out-of-core quantum computing simulation, in 2021 IEEE/ACM Second In- ternational Workshop on Quantum Computing Software (QCS) (IEEE...

    2021

  37. [45]

    Vidal, Efficient simulation of one-dimensional quantum many-body systems, Phys

    G. Vidal, Efficient simulation of one-dimensional quantum many-body systems, Phys. Rev. Lett. 93, 040502 (2004)

    2004

  38. [46]

    Hashizume, J

    T. Hashizume, J. C. Halimeh, and I. P. McCulloch, Hybrid infinite time-evolving block decimation algorithm for long- range multidimensional quantum many-body systems, Phys. Rev. B 102, 035115 (2020)

    2020

  39. [47]

    J. R. McClean, S. Boixo, V. N. Smelyanskiy, R. Bab- bush, and H. Neven, Barren plateaus in quantum neural network training landscapes, Nature Communications 9, 10.1038/s41467-018-07090-4 (2018)

    2018 doi

  40. [48]

    Cerezo, M

    M. Cerezo, M. Larocca, D. Garc ´ ıa-Mart ´ ın, N. L. Diaz, P. Braccia, E. Fontana, M. S. Rudolph, P. Bermejo, A. Ijaz, S. Thanasilp, E. R. Anschuetz, and Z. Holmes, Does provable absence of barren plateaus imply classi- cal simulability? or, why we need to rethink variational ...

    2024 arXiv

  41. [49]

    Schuld and N

    M. Schuld and N. Killoran, Is quantum advantage the right goal for quantum machine learning?, PRX Quantum 3, 10.1103/prxquantum.3.030101 (2022)

    2022 doi

  42. [50]

    Basheer, Y

    A. Basheer, Y. Feng, C. Ferrie, S. Li, and H. Pashayan, On the trainability and classical simulability of learning ma- trix product states variationally (2024), arXiv:2409.10055 [quant-ph]

    2024 arXiv

  43. [51]

    A. G. Baydin, B. A. Pearlmutter, A. A. Radul, and J. M. Siskind, Automatic differentiation in machine learning: a survey (2018), arXiv:1502.05767 [cs.SC]

    2018 arXiv

  44. [52]

    Gray, quimb: A python package for quantum informa- tion and many-body calculations, Journal of Open Source Software 3, 819 (2018)

    J. Gray, quimb: A python package for quantum informa- tion and many-body calculations, Journal of Open Source Software 3, 819 (2018)

    2018

  45. [53]

    Bradbury, R

    J. Bradbury, R. Frostig, P. Hawkins, M. J. Johnson, C. Leary, D. Maclaurin, G. Necula, A. Paszke, J. Van- derPlas, S. Wanderman-Milne, and Q. Zhang, JAX: com- posable transformations of Python+NumPy programs (2018)

    2018

  46. [54]

    https://www.bsc.es/marenostrum/marenostrum-5,

Pith tools

Reviewed August 11, 2026 · model on record in the stance chip above.