Pith. sign in

REVIEW 3 major objections 5 minor 106 references

No-Free-Lunch Theories for Tensor-Network Machine Learning Models

T0 review · 3 major / 5 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 Tensor-network machine learning models cannot escape a no-free-lunch limit: averaged over all target unitaries, their generalization risk is bounded below by a floor set by the training-set size and the network's bond and physical…

desk verdict The 1D MPS bound is a clean, real result, but Theorems 1 and 2 overclaim: the proofs only handle product-structured training subspaces, not arbitrary linearly independent training sets. read the letter →

arxiv 2412.05674 v1 pith:M46NB45H submitted 2024-12-07 quant-ph cs.AIcs.DScs.LGstat.ML

Jing-Chuan Wu , Qi Ye , Dong-Ling Deng , Li-Wei Yu This is my paper · ORCID
classification quant-phcs.AIcs.DScs.LGstat.ML MSC 68Q3281P6882B2005A15 PACS 03.67.-a05.50.+q
keywords no-free-lunchtheoremtensornetworkmachinelearningmatrixproductstatesprojectedentangledpairgeneralizationriskpolyominoenumerationIsingpartitionfunctionquantum
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 aims to prove no-free-lunch theorems for machine learning models built from tensor networks: averaged over every possible target unitary and every possible training set, the prediction error on unseen tensor-network-encoded inputs cannot be made small unless the training set is large. For matrix product states it proves an explicit lower bound on the average risk, and for two-dimensional projected entangled-pair states it proves the analogous bound. Both bounds depend on the training-set size, the physical dimension, and the bond dimension, so the inner structure of the tensor network itself sets the generalization floor. This matters because tensor-network models are widely proposed for quantum-inspired learning, and the results say their sample requirements are governed by a rigorous no-free-lunch limit, not just by practical training difficulties.

What carries the argument

The argument turns on writing the learned unitary as $W = M^\dagger P_S = e^{i\theta} I_{t_k} \oplus Y$, so the training subspace is learned up to a phase and the complement is a free unitary, then expanding $W \otimes W^\dagger$ into five terms $Z_1,\ldots,Z_5$. Each term is a second-moment integral over random local unitaries of the tensor network, and those integrals are evaluated by mapping them to partition functions of classical Ising models on the network lattice: a one-dimensional transfer matrix for MPS, and a two-dimensional Ising partition function for PEPS. The paper avoids solving the 2D Ising model directly by counting the contributing spin configurations as directed polyominoes, using the generating function $G(q,p)$ for directed polyominoes to bound the partition function by $1 + c(0.7)^L$. The local unitary 2-design property and norm concentration of random tensor-network states justify replacing the risk integral with this partition-function calculation.

What would settle it

For a $2\times 2$ PEPS with $d=D=2$, enumerate all $2^{L^2}=16$ spin configurations, evaluate the exact average risk from the partition-function sum, and compare it with the Theorem 2 lower bound for $k=1,2,3$; if the exact value falls below the bound, the theorem or one of its assumptions fails.

Watch

Extended reading notes

Core claim

On its own terms, the paper establishes that learning an arbitrary unitary from tensor-network-encoded data is subject to a no-free-lunch limit. With the risk defined as the trace-norm distance between the target output $M|x\rangle$ and the learned output $P_S|x\rangle$ averaged over random local encoding unitaries, Theorem 1 lower-bounds the average risk for MPS inputs by the expression in Eq. (2), and Theorem 2 lower-bounds it for PEPS inputs by Eq. (3). The training sets are linearly independent and have size $t_k = d^n - d^{n-k}$ (MPS) or $t_k = d^{L^2} - d^{L^2-k}$ (PEPS); the bounds interpolate from near one for an empty training set to near zero only when the training set approaches the full Hilbert space. The central claim is therefore that no tensor-network learner can beat this dimension-dependent floor in the average case, and that the floor is set jointly by sample size and by the network's bond and physical dimensions.

Load-bearing premise

The load-bearing assumption is that a perfectly trained model acts as a single global phase on the entire training subspace and as an arbitrary unitary on a complementary subspace that is a product of whole local sites (an approximate square block in 2D); if the training subspace has a generic shape, the site-by-site factorization of the second-moment calculation is not established, so the stated bounds are not proven.

Editorial extensions

If this is right

  • An MPS- or PEPS-based learner with $t_k$ samples cannot push average risk below the stated floor; sample sizes approaching $d^n$ or $d^{L^2}$ are needed to make the average risk small.
  • The floor depends on bond dimension $D$ and physical dimension $d$, so the tensor network's internal structure, not merely the number of samples, controls generalization.
  • The same proof machinery yields no-free-lunch bounds for learning matrix product operators and for deep quantum neural networks that can be represented as MPSs.
  • In 2D, the polyomino counting method gives a rigorous bound without evaluating the 2D Ising partition function, so the approach transfers to other planar tensor-network geometries.
  • Empty training sets give average risk near one and complete training sets give average risk near zero, matching the classical no-free-lunch intuition that generalization is impossible without data.

Reading between the lines

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

  • A sympathetic reading suggests the bounds may carry over to other local tensor-network geometries whose second moments can be encoded as polyomino or similar lattice-animal counts, but the paper only proves the square-lattice PEPS case.
  • The results imply that entanglement in the encoded data does not rescue sample complexity in this averaged setting; the network's own dimensions enter the floor, unlike settings where entangled data can reduce error when measurements are plentiful.
  • Testing whether the bounds are tight, by exact enumeration on small PEPS lattices or larger MPS simulations with zero training error, would decide how much room remains for architecture-specific improvements.
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 / 5 minor

Summary. The paper formulates no-free-lunch (NFL) bounds for tensor-network machine learning models that learn a target unitary from data encoded in random MPS and PEPS states. The main results are Theorem 1, a lower bound on the average risk E_{M,S}[R_M(P_S)] for MPS-encoded data depending on training-set size t_k, physical dimension d, and bond dimension D, and Theorem 2, an analogous bound for PEPS-encoded data that additionally involves a polyomino enumeration bound and a factor (1+c(0.7)^L). The 1D proof is presented through a transfer-matrix calculation of the second moment of random MPSs, and the 2D proof maps the corresponding second moment to a partition function whose configurations are controlled by directed polyominoes. Numerical simulations for small MPS systems (n=4,5) are reported as supporting the monotonic decrease of the risk with training-set size.

Significance. If the theorems as stated were established, this would be a valuable contribution: it would give the first rigorous NFL-type limitation for tensor-network learning models with an explicit dependence on the tensor-network bond and physical dimensions, and it would extend the analysis to 2D PEPS via a nontrivial polyomino counting argument. The 1D transfer-matrix derivation is explicit and algebraically consistent, and it reproduces the expected empty- and full-training-set limits. The numerical experiments, while modest in scale, support the qualitative trend of the analytical bounds. However, the proofs as written establish the bounds only for a highly structured family of training subspaces, not for the arbitrary linearly independent training sets named in the theorems; this gap is load-bearing for the paper's central claim.

major comments (3)
  1. [SI §B, Eq. (S34); SII §D, Eq. (S67)] The proofs of Theorems 1 and 2 evaluate the average risk only for the special learned unitary W = e^{iθ}(I_d^{⊗n} − Σ^{⊗k}⊗I_d^{⊗(n−k)}) + Σ^{⊗k}⊗Y, whose training subspace is the span of all computational-basis states in which not all of the first k sites are in |d⟩. This is a tensor-product-structured subspace, not a generic t_k-dimensional subspace. For an arbitrary linearly independent training set S, the correct form of W is e^{iθ}Π_S ⊕ Y in a basis adapted to S; in the computational basis Π_S is generally not a tensor product, so the transfer-matrix and polyomino factorizations of tr[F({SWAP}) W⊗W†] are not available. No invariance argument is given to show that the average over M and S is independent of the detailed structure of S, and no probability distribution over training sets is defined. Consequently Eqs. (2) and (3) are not established for arbitrary linearly independent training sets as stated.
  2. [SII §D, Eqs. (S86)–(S91)] The 2D proof adds a geometric assumption that is absent from Theorem 2's statement: the k trained sites are amalgamated into an l×l square, and all nonzero ESSs outside the trained zone are assumed to be rooted on at most two boundaries of length l, leading to the factor (1+G(q_a,q_p^2))^{2l}. For an arbitrary placement of k trained sites on the L×L torus, the boundary of the trained region can have length far exceeding 2l, and ESSs can be rooted along several boundary segments or wind around the torus in ways not counted by Eq. (S86). The cycle-ESS bound in Eq. (S88) likewise depends on this specific geometry. Thus Theorem 2 is proven only for a restricted class of training-set geometries, not for the arbitrary training sets described in the theorem.
  3. [SII B–C, Theorem 3, Corollary 1] The decisive 2D estimates—the polyomino generating function G(q,p), the partition-function bound in Eq. (S63), and Corollary 1—are imported verbatim from Ref. [68] (described as 'Lemma 5 and Lemma 6 in Ref. [68]' and 'Theorem 1 in Ref. [68]'), while the main text says this paper introduces the combinatorial polyomino method. Because Theorem 2's proof rests on these imported bounds, the proof as presented is not self-contained, and the novelty attribution is inaccurate. The authors should include the necessary statements and proofs, or explicitly and accurately frame the contribution as applying the results of Ref. [68] to the NFL setting.
minor comments (5)
  1. [Main text, proof sketch after Theorem 1; SI Eq. (S34)] The matrix Σ is used in the main-text proof sketch without definition; it is defined only in the Supplemental Material. Please define Σ = diag(0,0,...,1) in the main text.
  2. [Main text, paragraph after Eq. (2)] The statement that the average risk is 'lower bounded by one' for the empty training set is imprecise, since the risk is at most one; the calculation in SI Eq. (S47) gives 1 − 1/d^n − ... ≈ 1. Rephrase as 'approaches one' in the relevant limit.
  3. [SII, proof sketch of Theorem 2] There is a typo: 'uniatry embedded PEPS' should read 'unitary embedded PEPS'.
  4. [SII B and throughout] The symbol D is used both for the bond dimension of the tensor network and for the polyomino counts D_{m,n}; please rename one of these to avoid ambiguity.
  5. [Numerical Results and Fig. S5] The numerical experiments do not report the number of random target unitaries or training sets used for the averages, nor error bars; please specify these details so that the claimed monotonic decrease can be assessed quantitatively.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the TN-NFL bounds follow from random-tensor 2-design and polyomino partition-function lemmas, with only a proof-scope gap in the training-subspace assumption.

full rationale

The 1D derivation is self-contained: the risk integral is rewritten as tr[F({SWAP}) W⊗W†] in Eq. (S27), the transfer matrix T is obtained directly from the 2-design integrals of the local MPS unitaries (Eqs. S20–S26), and the lower bound in Eq. (S46) is an algebraic trace bound. No parameter is fitted to the risk values being predicted, and the bound is not inserted by definition. The 2D derivation is not self-contained, but it is not circular: the key partition-function bounds (Theorem 3 and Corollary 1) are quoted from Ref. [68], which shares authors with this paper. That self-citation is load-bearing for Theorem 2, but Ref. [68] is a parameter-free general bound on the second moment of random PEPS with stated assumptions (D,d≥2 on an L×L periodic lattice) and contains no NFL or risk statement, so under the stated review rules it counts as independent support rather than a circular premise. The new boundary-rooted ESS counting (Eqs. S84–S91) is applied on top of that bound and does not redefine the target quantity. The genuine weakness is a scope gap, not circularity: the proofs set W = e^{iθ}I_{t_k}⊕Y with Y on an (n−k)-qudit subsystem (Eqs. S34 and S67), i.e., a product-form training subspace, while Theorems 1 and 2 state arbitrary linearly independent training sets. The sentence "The above results are independent of the training set S" is therefore not justified by an average over all S; this is an overgeneralization or rigor concern, not an equation reducing to its own input. The numerical results are consistency checks and are not used to derive the analytical bounds.

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

The central claim rests on five assumptions: unitary 2-design behaviour of random local tensors, state-norm concentration, a product-structure training subspace, the 2D square-zone amalgamation, and perfect training. The first, second, and fifth are standard in the cited literature. The third and fourth are introduced ad hoc in this paper and are the main unresolved gaps in the proofs. No free parameters are fitted to data; the constant c in Theorem 2 is an existential bound from Ref [68].

assumptions (5)
  • domain assumption At each site the random local unitaries of the unitary-embedded MPS/PEPS behave as approximate unitary 2-designs, i.e., moment integrals equal the Haar values used in Eqs (S20)-(S26) and (S56).
    Invoked in SI Eq. (S20) for MPS and Eq. (S56) for PEPS; standard in the random-tensor-network literature (Refs [67,88,68]), but it restricts the input distribution to random unitary-embedded tensor networks.
  • domain assumption The norm of the encoded state is exponentially concentrated around 1 (MPS) or concentrated with variance O(c(0.7)^L) (PEPS), allowing the risk function to be replaced by the simplified form.
    Used in the proof of Theorem 1 (SI Eq. S5, citing Ref [88]) and Theorem 2 (SI Eq. S52, citing Corollary 1 of Ref [68]).
  • ad hoc to paper The training subspace is a tensor-product subsystem: after training, M†P_S = e^{iθ}(I − Σ⊗k⊗I^{⊗(n−k)}) + Σ⊗k⊗Y with Y acting on an (n−k)-qudit subsystem, so the complement is a multi-site product space.
    Assumed in SI §B (after Eq. S34) and SII §D (after Eq. S67); load-bearing for the transfer-matrix factorization of F but not proven for arbitrary linearly independent training sets.
  • ad hoc to paper In 2D, the k trained sites can be amalgamated into an l×l square with l=⌈√k⌉ whose upper/left boundaries host the relevant ESS roots.
    Stated in SII §D around Fig. S3 and Eq. (S86); no rigorous argument is given that a generic k-site subsystem has this geometry.
  • domain assumption The hypothesis circuit PS can exactly realize the target unitary on the training set (perfect training, up to a global phase).
    Used in the derivation of W = M†P_S in both the main text and SI; standard in generalization analyses when the model is sufficiently expressive.

how reviews work

0 comments
Cite this review

Pith. "Pith review of No-Free-Lunch Theories for Tensor-Network Machine Learning Models." pith.science (2026). https://pith.science/paper/M46NB45H

@misc{pith2026241205674,
  author       = {Pith},
  title        = {Pith review of: No-Free-Lunch Theories for Tensor-Network Machine Learning Models},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/M46NB45H}},
  note         = {Machine review of arXiv:2412.05674}
}
read the original abstract

Tensor network machine learning models have shown remarkable versatility in tackling complex data-driven tasks, ranging from quantum many-body problems to classical pattern recognitions. Despite their promising performance, a comprehensive understanding of the underlying assumptions and limitations of these models is still lacking. In this work, we focus on the rigorous formulation of their no-free-lunch theorem -- essential yet notoriously challenging to formalize for specific tensor network machine learning models. In particular, we rigorously analyze the generalization risks of learning target output functions from input data encoded in tensor network states. We first prove a no-free-lunch theorem for machine learning models based on matrix product states, i.e., the one-dimensional tensor network states. Furthermore, we circumvent the challenging issue of calculating the partition function for two-dimensional Ising model, and prove the no-free-lunch theorem for the case of two-dimensional projected entangled-pair state, by introducing the combinatorial method associated to the "puzzle of polyominoes". Our findings reveal the intrinsic limitations of tensor network-based learning models in a rigorous fashion, and open up an avenue for future analytical exploration of both the strengths and limitations of quantum-inspired machine learning frameworks.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

106 extracted references · 48 canonical work pages

  1. [68]

    Z. Liu, Q. Ye, L.-W. Yu, L.-M. Duan, and D.-L. 7 Deng, Theory on variational high-dimensional tensor networks, arXiv:2303.17452 (2023)

    work page Pith review arXiv 2023

  2. [1]

    Or´ us, Tensor networks for complex quantum systems, Nat

    R. Or´ us, Tensor networks for complex quantum systems, Nat. Rev. Phys. 1, 538 (2019)

    2019

  3. [2]

    Biamonte, Lectures on quantum tensor networks, arXiv:1912.10049 (2019)

    J. Biamonte, Lectures on quantum tensor networks, arXiv:1912.10049 (2019)

    arXiv 2019

  4. [3]

    J. I. Cirac, D. P´ erez-Garc ´ ıa, N. Schuch, and F. Ver- straete, Matrix product states and projected entangled pair states: Concepts, symmetries, theorems, Rev. Mod. Phys. 93, 045003 (2021)

    2021

  5. [4]

    M. C. Ba˜ nuls, Tensor Network Algorithms: A Route Map, Annu. Rev. Condens. Matter Phys. 14, 173 (2023)

    2023

  6. [5]

    Rieser, F

    H.-M. Rieser, F. K¨ oster, and A. P. Raulf, Tensor net- works for quantum machine learning, Proc. Roy. Soc. A 10.1098/rspa.2023.0218 (2023)

    arXiv 2023

  7. [6]

    M. Wang, Y. Pan, Z. Xu, X. Yang, G. Li, and A. Ci- chocki, Tensor Networks Meet Neural Networks: A Sur- vey and Future Perspectives, arXiv:2302.09019 (2023)

    arXiv 2023

  8. [7]

    Cichocki, Tensor networks for big data analytics and large-scale optimization problems, arXiv:1407.3124 (2014)

    A. Cichocki, Tensor networks for big data analytics and large-scale optimization problems, arXiv:1407.3124 (2014)

    arXiv 2014

Show all 106 references
  1. [8]

    Novikov, D

    A. Novikov, D. Podoprikhin, A. Osokin, and D. P. Vetrov, Tensorizing Neural Networks, in Advances in NeuralIPS, Vol. 28 (2015)

    2015

  2. [9]

    Cichocki, N

    A. Cichocki, N. Lee, I. Oseledets, A.-H. Phan, Q. Zhao, and D. P. Mandic, Tensor Networks for Dimensionality Reduction and Large-scale Optimization: Part 1 Low- Rank Tensor Decompositions, Foundations and Trends in Machine Learning 9, 249 (2016)

    2016

  3. [10]

    Cichocki, A.-H

    A. Cichocki, A.-H. Phan, Q. Zhao, N. Lee, I. V. Os- eledets, M. Sugiyama, and D. Mandic, Tensor Net- works for Dimensionality Reduction and Large-scale Optimization: Part 2 Applications and Future Perspec- tives, Foundations and Trends in Machine Learning 9, 431 (2016)

    2016

  4. [11]

    Stoudenmire and D

    E. Stoudenmire and D. J. Schwab, Supervised learning with tensor networks, in Advances in NeuralIPS, edited by D. D. Lee, M. Sugiyama, U. V. Luxburg, I. Guyon, and R. Garnett (Curran Associates, Inc., 2016) pp. 4799–4807

    2016

  5. [12]

    Novikov, M

    A. Novikov, M. Trofimov, and I. Oseledets, Exponential machines, arXiv:1605.03795 (2016)

    2016 arXiv

  6. [13]

    Y. Liu, X. Zhang, M. Lewenstein, and S.-J. Ran, Entanglement-guided architectures of machine learning by quantum tensor network, arXiv:1803.09111 (2018)

    2018 arXiv

  7. [14]

    Z. Chen, K. Batselier, J. A. K. Suykens, and N. Wong, Parallelized Tensor Train Learning of Polynomial Clas- sifiers, IEEE Trans. Neural Netw. Learn. Syst. 29, 4621 (2018)

    2018

  8. [15]

    Levine, D

    Y. Levine, D. Yakira, N. Cohen, and A. Shashua, Deep learning and quantum entanglement: Fundamental con- nections with implications to network design, in ICLR (2018)

    2018

  9. [16]

    E. M. Stoudenmire, Learning relevant features of data with multi-scale tensor networks, Quantum Sci. Tech- nol. 3, 034003 (2018)

    2018

  10. [17]

    Z.-Y. Han, J. Wang, H. Fan, L. Wang, and P. Zhang, Unsupervised generative modeling using matrix product states, Phys. Rev. X 8, 031012 (2018)

    2018

  11. [18]

    Liu, S.-J

    D. Liu, S.-J. Ran, P. Wittek, C. Peng, R. B. Garc ´ ıa, G. Su, and M. Lewenstein, Machine learning by uni- tary tensor network of hierarchical tree structure, New J. Phys. 21, 073059 (2019)

    2019

  12. [19]

    Hayashi, T

    K. Hayashi, T. Yamaguchi, Y. Sugawara, and S.- i. Maeda, Exploring Unexplored Tensor Network De- compositions for Convolutional Neural Networks, in Advances in Neural Information Processing Systems, Vol. 32 (2019)

    2019

  13. [20]

    J. Liu, S. Li, J. Zhang, and P. Zhang, Tensor networks for unsupervised machine learning, arXiv:2106.12974v1 (2021)

    2021 arXiv

  14. [21]

    J. Chen, S. Cheng, H. Xie, L. Wang, and T. Xiang, Equivalence of restricted Boltzmann machines and ten- sor network states, Phys. Rev. B 97, 085104 (2018)

    2018

  15. [22]

    Levine, O

    Y. Levine, O. Sharir, N. Cohen, and A. Shashua, Quan- tum entanglement in deep learning architectures, Phys. Rev. Lett. 122, 065301 (2019)

    2019

  16. [23]

    A. S. Bhatia, M. K. Saggi, A. Kumar, and S. Jain, Matrix product state–based quantum classifier, Neural Comput. 31, 1499 (2019)

    2019

  17. [24]

    Efthymiou, J

    S. Efthymiou, J. Hidary, and S. Leichenauer, Tensornet- work for machine learning, arXiv:1906.06329 (2019)

    2019 arXiv

  18. [25]

    Huggins, P

    W. Huggins, P. Patil, B. Mitchell, K. B. Whaley, and E. M. Stoudenmire, Towards quantum machine learning with tensor networks, Quantum Sci. Technol. 4, 024001 (2019)

    2019

  19. [26]

    Glasser, N

    I. Glasser, N. Pancotti, and J. I. Cirac, From probabilis- tic graphical models to generalized tensor networks for supervised learning, IEEE Access 8, 68169 (2020)

    2020

  20. [27]

    J. Su, W. Byeon, J. Kossaifi, F. Huang, J. Kautz, and A. Anandkumar, Convolutional Tensor-Train LSTM for Spatio-Temporal Learning, in Advances in Neural Infor- mation Processing Systems, Vol. 33 (2020) pp. 13714– 13726

    2020

  21. [28]

    Sun, S.-J

    Z.-Z. Sun, S.-J. Ran, and G. Su, Tangent-space gradi- ent optimization of tensor network for machine learning, Phys. Rev. E 102, 012152 (2020)

    2020

  22. [29]

    Sun, Z.-F

    X. Sun, Z.-F. Gao, Z.-Y. Lu, J. Li, and Y. Yan, A Model Compression Method With Matrix Product Op- erators for Speech Enhancement, IEEE/ACM Trans. Audio Speech Lang. Process. 28, 2837 (2020)

    2020

  23. [30]

    S. Y.-C. Chen, C.-M. Huang, C.-W. Hsing, and Y.-J. Kao, Hybrid quantum-classical classifier based on tensor network and variational quantum circuit, arXiv:2011.14651 (2020)

    2020 arXiv

  24. [31]

    Z.-Z. Sun, C. Peng, D. Liu, S.-J. Ran, and G. Su, Gener- ative tensor network classification model for supervised machine learning, Phys. Rev. B 101, 075135 (2020)

    2020

  25. [32]

    J. Wang, C. Roberts, G. Vidal, and S. Le- ichenauer, Anomaly detection with tensor networks, arXiv:2006.02516 (2020). 6

    2020 arXiv

  26. [33]

    Z.-F. Gao, S. Cheng, R.-Q. He, Z. Y. Xie, H.-H. Zhao, Z.-Y. Lu, and T. Xiang, Compressing deep neural net- works by matrix product operators, Phys. Rev. Res. 2, 023300 (2020)

    2020

  27. [34]

    M. L. Wall, M. R. Abernathy, and G. Quiroz, Gener- ative machine learning with tensor networks: Bench- marks on near-term quantum computers, Phys. Rev. Res. 3, 023010 (2021)

    2021

  28. [35]

    Cheng, L

    S. Cheng, L. Wang, and P. Zhang, Supervised learning with projected entangled pair states, Phys. Rev. B 103, 125117 (2021)

    2021

  29. [36]

    M. N. da Costa, R. Attux, A. Cichocki, and J. M. Romano, Tensor-train networks for learning predictive modeling of multidimensional data, arXiv:2101.09184 (2021)

    2021 arXiv

  30. [37]

    Kardashin, A

    A. Kardashin, A. Uvarov, and J. Biamonte, Quantum Machine Learning Tensor Network States, Front. Phys. 8, 644 (2021)

    2021

  31. [38]

    Felser, M

    T. Felser, M. Trenti, L. Sestini, A. Gianelle, D. Zuliani, D. Lucchesi, and S. Montangero, Quantum-inspired ma- chine learning on high-energy physics data, npj Quan- tum Inf. 7, 1 (2021)

    2021

  32. [39]

    K. Wang, L. Xiao, W. Yi, S.-J. Ran, and P. Xue, Ex- perimental realization of a quantum image classifier via tensor-network-based machine learning, Photon. Res. 9, 2332 (2021)

    2021

  33. [40]

    Hawkins and Z

    C. Hawkins and Z. Zhang, Bayesian tensorized neural networks with automatic rank selection, Neurocomput- ing 453, 172 (2021)

    2021

  34. [41]

    C. Chen, K. Batselier, W. Yu, and N. Wong, Kernelized support tensor train machines, Pattern Recognit. 122, 108337 (2022)

    2022

  35. [42]

    Vieijra, L

    T. Vieijra, L. Vanderstraeten, and F. Verstraete, Gen- erative modeling with projected entangled-pair states, arXiv:2202.08177 (2022)

    2022 arXiv

  36. [43]

    X. Shi, Y. Shang, and C. Guo, Clustering using matrix product states, Phys. Rev. A 105, 052424 (2022)

    2022

  37. [44]

    Convy, W

    I. Convy, W. Huggins, H. Liao, and K. B. Whaley, Mutual information scaling for tensor network machine learning, Mach. Learn.: Sci. Technol. 3, 015017 (2022)

    2022

  38. [45]

    Metz and M

    F. Metz and M. Bukov, Self-correcting quantum many- body control using reinforcement learning with tensor networks, Nat. Mach. Intell. 5, 780 (2023)

    2023

  39. [46]

    H. Liao, I. Convy, Z. Yang, and K. B. Whaley, Decoher- ing tensor network quantum machine learning models, Quantum Mach. Intell. 5, 7 (2023)

    2023

  40. [47]

    Ran and G

    S.-J. Ran and G. Su, Tensor Networks for Interpretable and Efficient Quantum-Inspired Machine Learning, In- tell. Comput. 2, 0061 (2023)

    2023

  41. [48]

    D. Wu, R. Rossi, F. Vicentini, and G. Carleo, From tensor-network quantum states to tensorial recurrent neural networks, Phys. Rev. Res. 5, L032001 (2023)

    2023

  42. [49]

    Lopez-Piqueres, J

    J. Lopez-Piqueres, J. Chen, and A. Perdomo-Ortiz, Symmetric tensor networks for generative modeling and constrained combinatorial optimization, Mach. Learn.: Sci. Technol. 4, 035009 (2023)

    2023

  43. [50]

    Y.-M. Meng, J. Zhang, P. Zhang, C. Gao, and S.-J. Ran, Residual matrix product state for machine learning, Sci- Post Phys. 14, 142 (2023)

    2023

  44. [51]

    S. Shin, Y. S. Teo, and H. Jeong, Dequantizing quantum machine learning models using tensor networks, Phys. Rev. Res. 6, 023218 (2024)

    2024

  45. [52]

    Wesel and K

    F. Wesel and K. Batselier, Tensor Network- Constrained Kernel Machines as Gaussian Processes, arXiv:2403.19500 (2024)

    2024 arXiv

  46. [53]

    A. S. Bhatia and D. E. B. Neira, Federated Hierarchical Tensor Networks: A Collaborative Learning Quantum AI-Driven Framework for Healthcare, arXiv:2405.07735 (2024)

    2024 arXiv

  47. [54]

    Tomut, S

    A. Tomut, S. S. Jahromi, S. Singh, F. Ishtiaq, C. Mu˜ noz, P. S. Bajaj, A. Elborady, G. del Bimbo, M. Al- izadeh, D. Montero, P. Martin-Ramiro, M. Ibrahim, O. T. Alaoui, J. Malcolm, S. Mugel, and R. Orus, CompactifAI: Extreme Compression of Large Lan- guage Models using Quantu...

    2024 arXiv

  48. [55]

    Y. Teng, R. Samajdar, K. Van Kirk, F. Wilde, S. Sachdev, J. Eisert, R. Sweke, and K. Najafi, Learning topological states from randomized measure- ments using variational tensor network tomography, arXiv:2406.00193 (2024)

    2024

  49. [56]

    Bermejo, P

    P. Bermejo, P. Braccia, M. S. Rudolph, Z. Holmes, L. Cincio, and M. Cerezo, Quantum Convolutional Neu- ral Networks are (Effectively) Classically Simulable, arXiv:2408.12739 (2024)

    2024 arXiv

  50. [57]

    H. P. Casagrande, B. Xing, W. J. Munro, C. Guo, and D. Poletti, Tensor-Networks-based Learning of Proba- bilistic Cellular Automata Dynamics, arXiv:2404.11768 (2024)

    2024 arXiv

  51. [58]

    Chen and T

    H. Chen and T. Barthel, Machine learning with tree ten- sor networks, CP rank constraints, and tensor dropout, IEEE Trans. Pattern Anal. Mach. Intell. 46, 7825 (2024)

    2024

  52. [59]

    Z. Su, Y. Zhou, F. Mo, and J. G. Simonsen, Lan- guage Modeling Using Tensor Trains, arXiv:2405.04590 (2024)

    2024 arXiv

  53. [60]

    C. Guo, Z. Jie, W. Lu, and D. Poletti, Matrix product operators for sequence-to-sequence learning, Phys. Rev. E 98, 042114 (2018)

    2018

  54. [61]

    Meichanetzidis, S

    K. Meichanetzidis, S. Gogioso, G. De Felice, N. Chi- appori, A. Toumi, and B. Coecke, Quantum natural language processing on near-term quantum computers, arXiv:2005.04147 (2020)

    2020

  55. [62]

    Gao, Z.-Y

    X. Gao, Z.-Y. Zhang, and L.-M. Duan, A quantum ma- chine learning algorithm based on generative models, Sci. Adv. 4, eaat9004 (2018)

    2018

  56. [63]

    X. Gao, E. R. Anschuetz, S.-T. Wang, J. I. Cirac, and M. D. Lukin, Enhancing generative models via quantum correlations, Phys. Rev. X 12, 021037 (2022)

    2022

  57. [64]

    M. S. Rudolph, J. Miller, D. Motlagh, J. Chen, A. Acharya, and A. Perdomo-Ortiz, Synergistic pre- training of parametrized quantum circuits via tensor networks, Nat. Commun. 14, 8367 (2023)

    2023

  58. [65]

    Abadi, P

    M. Abadi, P. Barham, J. Chen, Z. Chen, A. Davis, J. Dean, M. Devin, S. Ghemawat, G. Irving, M. Isard, et al., TensorFlow: A system for large-scale machine learning, in Proceedings of the 12th USENIX Confer- ence on Operating Systems Design and Implementation, OSDI’16 (USENIX A...

    2016

  59. [66]

    Roberts, A

    C. Roberts, A. Milsted, M. Ganahl, A. Zalcman, B. Fontaine, Y. Zou, J. Hidary, G. Vidal, and S. Le- ichenauer, TensorNetwork: A Library for Physics and Machine Learning, arXiv:1905.01330 (2019)

    2019 arXiv

  60. [67]

    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

  61. [69]

    R. J. Garcia, C. Zhao, K. Bu, and A. Jaffe, Barren plateaus from learning scramblers with local cost func- tions, J. High Energ. Phys. 2023 (1), 90

    2023

  62. [70]

    Miao and T

    Q. Miao and T. Barthel, Isometric tensor network op- timization for extensive Hamiltonians is free of barren plateaus, Phys. Rev. A 109, L050402 (2024)

    2024

  63. [71]

    Barthel and Q

    T. Barthel and Q. Miao, Absence of barren plateaus and scaling of gradients in the energy optimization of isomet- ric tensor network states, arXiv:2304.00161 (2024)

    2024 arXiv

  64. [72]

    Strashko and E

    A. Strashko and E. M. Stoudenmire, Generalization and Overfitting in Matrix Product State Machine Learning Architectures, arXiv:2208.04372 (2022)

    2022 arXiv

  65. [73]

    Schaffer, A Conservation Law for Generalization Performance, in Machine Learning Proceedings 1994, edited by W

    C. Schaffer, A Conservation Law for Generalization Performance, in Machine Learning Proceedings 1994, edited by W. W. Cohen and H. Hirsh (San Francisco (CA), 1994) pp. 259–265

    1994

  66. [74]

    D. H. Wolpert, The Lack of A Priori Distinctions Be- tween Learning Algorithms, Neur. Comput. 8, 1341 (1996)

    1996

  67. [75]

    Wolpert and W

    D. Wolpert and W. Macready, No free lunch theorems for optimization, IEEE Trans. Evol. Computat. 1, 67 (1997)

    1997

  68. [76]

    Poland, K

    K. Poland, K. Beer, and T. J. Osborne, No free lunch for quantum machine learning, arXiv:2003.14103 (2020)

    2020 arXiv

  69. [77]

    Volkoff, Z

    T. Volkoff, Z. Holmes, and A. Sornborger, Uni- versal Compiling and (No-)Free-Lunch Theorems for Continuous-Variable Quantum Learning, PRX Quan- tum 2, 040327 (2021)

    2021

  70. [78]

    Sharma, M

    K. Sharma, M. Cerezo, Z. Holmes, L. Cincio, A. Sorn- borger, and P. J. Coles, Reformulation of the no-free- lunch theorem for entangled datasets, Phys. Rev. Lett. 128, 070501 (2022)

    2022

  71. [79]

    Brierley, No free lunch for Schr¨ odinger’s cat, Nat

    R. Brierley, No free lunch for Schr¨ odinger’s cat, Nat. Phys. 18, 373 (2022)

    2022

  72. [80]

    H. Zhao, L. Lewis, I. Kannan, Y. Quek, H.-Y. Huang, and M. C. Caro, Learning quantum states and unitaries of bounded gate complexity, arXiv:2310.19882 (2023)

    2023 arXiv

  73. [81]

    X. Wang, Y. Du, Z. Tu, Y. Luo, X. Yuan, and D. Tao, Transition role of entangled data in quantum machine learning, Nat. Commun. 15, 3716 (2024)

    2024

  74. [82]

    X. Wang, Y. Du, K. Liu, Y. Luo, B. Du, and D. Tao, Separable Power of Classical and Quantum Learning Protocols Through the Lens of No-Free-Lunch Theo- rem, arXiv:2405.07226 (2024)

    2024 arXiv

  75. [83]

    Perez-Garcia, F

    D. Perez-Garcia, F. Verstraete, M. M. Wolf, and J. I. Cirac, Matrix product state representations, Quantum Inf. Comput. 7, 401 (2007)

    2007

  76. [84]

    Gross and J

    D. Gross and J. Eisert, Quantum computational webs, Phys. Rev. A 82, 040303(R) (2010)

    2010

  77. [85]

    Garnerone, T

    S. Garnerone, T. R. de Oliveira, S. Haas, and P. Zanardi, Statistical properties of random matrix product states, Phys. Rev. A 82, 052312 (2010)

    2010

  78. [86]

    Garnerone, T

    S. Garnerone, T. R. de Oliveira, and P. Zanardi, Typ- icality in random matrix product states, Phys. Rev. A 81, 032336 (2010)

    2010

  79. [87]

    Collins, C

    B. Collins, C. E. Gonz´ alez-Guill´ en, and D. P´ erez- Garc ´ ıa, Matrix product states, random matrix theory and the principle of maximum entropy, arXiv:1201.6324 [quant-ph] (2012)

    2012 arXiv

  80. [88]

    Haferkamp, C

    J. Haferkamp, C. Bertoni, I. Roth, and J. Eisert, Emergent statistical mechanics from properties of dis- ordered random matrix product states, PRX Quantum 2, 040308 (2021)

    2021

  81. [89]

    M. C. Caro, H.-Y. Huang, M. Cerezo, K. Sharma, A. Sornborger, L. Cincio, and P. J. Coles, Generalization in quantum machine learning from few training data, Nat. Commun. 13, 4919 (2022)

    2022

  82. [90]

    See Supplemental Material at [URL will be inserted by publisher] for details on the notations and techniques, the proofs of both Theorem 1 and 2, and the numerical simulations

  83. [91]

    Verstraete, M

    F. Verstraete, M. M. Wolf, D. Perez-Garcia, and J. I. Cirac, Criticality, the area law, and the computational power of projected entangled pair states, Phys. Rev. Lett. 96, 220601 (2006)

    2006

  84. [92]

    Jordan, R

    J. Jordan, R. Or´ us, G. Vidal, F. Verstraete, and J. I. Cirac, Classical Simulation of Infinite-Size Quantum Lattice Systems in Two Spatial Dimensions, Phys. Rev. Lett. 101, 250602 (2008)

    2008

  85. [93]

    Z.-C. Gu, M. Levin, B. Swingle, and X.-G. Wen, Tensor- product representations for string-net condensed states, Phys. Rev. B 79, 085118 (2009)

    2009

  86. [94]

    Buerschaper, M

    O. Buerschaper, M. Aguado, and G. Vidal, Explicit tensor network representation for the ground states of string-net models, Phys. Rev. B 79, 085119 (2009)

    2009

  87. [95]

    Piroli and J

    L. Piroli and J. I. Cirac, Quantum Cellular Automata, Tensor Networks, and Area Laws, Phys. Rev. Lett.125, 190402 (2020)

    2020

  88. [96]

    Haghshenas, J

    R. Haghshenas, J. Gray, A. C. Potter, and G. K.-L. Chan, Variational Power of Quantum Circuit Tensor Networks, Phys. Rev. X 12, 011047 (2022)

    2022

  89. [97]

    Pan and P

    F. Pan and P. Zhang, Simulation of Quantum Circuits Using the Big-Batch Tensor Network Method, Phys. Rev. Lett. 128, 030501 (2022)

    2022

  90. [98]

    F. Pan, K. Chen, and P. Zhang, Solving the Sampling Problem of the Sycamore Quantum Circuits, Phys. Rev. Lett. 129, 090502 (2022)

    2022

  91. [99]

    Y. Wang, Y. E. Zhang, F. Pan, and P. Zhang, Tensor Network Message Passing, Phys. Rev. Lett.132, 117401 (2024)

    2024

  92. [100]

    Eisert, M

    J. Eisert, M. Cramer, and M. B. Plenio, Colloquium: Area laws for the entanglement entropy, Rev. Mod. Phys. 82, 277 (2010)

    2010

  93. [101]

    Schuch, M

    N. Schuch, M. M. Wolf, F. Verstraete, and J. I. Cirac, Computational Complexity of Projected Entangled Pair States, Phys. Rev. Lett. 98, 140506 (2007)

    2007

  94. [102]

    Bousquet-M´ elou, New enumerative results on two- dimensional directed animals, Discr

    M. Bousquet-M´ elou, New enumerative results on two- dimensional directed animals, Discr. Math. Proceedings of the 7th Conference on Formal Power Series and Al- gebraic Combinatorics, 180, 73 (1998)

    1998

  95. [103]

    Fishman, S

    M. Fishman, S. R. White, and E. M. Stoudenmire, The ITensor Software Library for Tensor Network Calcula- tions, SciPost Phys. Codebases , 4 (2022)

    2022

  96. [104]

    Fishman, S

    M. Fishman, S. R. White, and E. M. Stoudenmire, Codebase release 0.3 for ITensor, SciPost Phys. Code- bases , 4 (2022)

    2022

  97. [105]

    K. Beer, D. Bondarenko, T. Farrelly, T. J. Osborne, R. Salzmann, D. Scheiermann, and R. Wolf, Training deep quantum neural networks, Nat. Commun. 11, 808 (2020)

    2020

  98. [106]

    𝑌∗# 𝑌∗$𝑌∗% 𝑌∗

    X. Pan, Z. Lu, W. Wang, Z. Hua, Y. Xu, W. Li, W. Cai, X. Li, H. Wang, Y.-P. Song, C.-L. Zou, D.-L. Deng, and L. Sun, Deep quantum neural networks on a supercon- ducting processor, Nat. Commun. 14, 4006 (2023). 1 Supplemental Materials: No-Free-Lunch Theories for Tensor-Network...

    2023

Pith tools

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