Pith. sign in

REVIEW 3 major objections 6 minor 2 cited by

Numerical Algebraic Geometry for Energy Computations on Tensor Train Varieties

T0 review · 3 major / 6 minor · reviewed 2026-08-03 · deepseek-v4-flash

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

Pith's one-line read The constrained Rayleigh quotient over a tensor-train variety has a well-defined number of complex critical points, the Rayleigh-Ritz degree, and for small systems those points can be computed exhaustively, exposing when standard algorithms

desk verdict Solid birational-parametrization paper whose 'all critical points' claim overreaches; the authors disclose their own counterexamples, and the core theory holds. read the letter →

arxiv 2512.06939 v2 pith:BK2AXNL2 submitted 2025-12-07 math.AG math.OCphysics.chem-ph

Viktoriia Borovik , Hannah Friedman , Serkan Hoşten , Max Pfeffer This is my paper · ORCID
classification math.AGmath.OCphysics.chem-ph MSC 14Q2014M1265H1015A69
keywords tensortrainvarietiesRayleigh-RitzdegreeRayleighquotientcriticalpointshomotopycontinuationalternatinglinearschemeDMRGquantumchemistry
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 tries to settle a practical question: when a quantum-chemistry Hamiltonian is minimized over the low-rank tensor-train format, what does the optimization landscape actually look like? It defines the Rayleigh-Ritz degree — the number of complex critical points of the constrained Rayleigh quotient for a generic Hamiltonian — and proves it is well defined for tensor-train varieties. The authors give a birational parametrization of the tensor-train variety from a product of Grassmannians, turning the critical-point equations into a polynomial system that homotopy continuation can solve exhaustively for small systems. With the full critical-point list in hand, they benchmark the standard algorithms ALS and DMRG: ALS frequently lands in local minima with energies far from the global minimum, and DMRG's rank truncation means it does not converge to critical points of the fixed-rank problem at all. The payoff is the first exhaustive picture of where these heuristics can go wrong.

What carries the argument

The load-bearing object is the birational parametrization of the tensor-train manifold V=_{k,r} from a product of Grassmannians (Algorithm 1 and Theorem 3.3), built from successive XR decompositions — a factorization of a matrix into a row-selection factor and a coefficient matrix, valid under the assumption that the first r rows of each flattening are linearly independent. This converts the constrained Rayleigh quotient into an unconstrained rational function on the parameter space, whose critical equations form a polynomial system. The paper also introduces the Rayleigh-Ritz correspondence (the incidence variety of critical-point–Hamiltonian pairs) and its discriminant, the set of Hamilton

What would settle it

Construct a small Hamiltonian whose ground state lies on the tensor-train variety with the first r rows of some flattening linearly dependent, then run the parametrization-based homotopy solver and compare to direct eigenvector computation: if the solver's reported global minimum energy exceeds the true constrained minimum, the exhaustiveness claim fails.

Watch

Extended reading notes

Core claim

The central claim is that constrained Rayleigh-quotient critical points on a tensor-train variety can be computed completely for small systems, and that the count is governed by a constant, the Rayleigh-Ritz degree, independent of the generic Hamiltonian. This is achieved by identifying the tensor-train variety (when ranks satisfy certain inequalities) as a Segre product, by giving a birational map from a product of Grassmannians to the tensor-train manifold (Theorem 3.3), and by defining the Rayleigh-Ritz correspondence and discriminant, which describe Hamiltonians with a deficient number of critical points. Numerical homotopy continuation then computes all isolated complex critical points

Load-bearing premise

The exhaustive count is only guaranteed on the open set of tensors where, at every step, the first r rows of each flattened tensor are linearly independent; any critical point outside this open set is invisible to the computation, and the paper itself shows a case where the global minimum is one of the missing points.

Editorial extensions

If this is right

  • For small tensor-train and determinantal varieties, the Rayleigh-Ritz degree and the actual critical points are now known, giving a ground truth for testing any approximate eigensolver.
  • The Rayleigh-Ritz discriminant splits the space of Hamiltonians into regions with different numbers of real critical points; for example, P1×P1 has 4, 6, or 8 real critical points depending on the Hamiltonian.
  • ALS can converge to any local minimum, and the global minimum is not always the most frequent output; for random symmetric matrices the energy gap can be large.
  • DMRG with rank truncation is not a solver for the fixed-rank Rayleigh quotient, so its output should not be compared directly to manifold critical points.
  • When critical points lie outside the parametrized open set, the computed census is incomplete; Example 7.4 shows that the global minimum can be one of the missing points.

Reading between the lines

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

  • A natural next step is to compute the average number of real critical points using the discriminant's regions; the examples here show that counts vary widely, so a random Hamiltonian's landscape is far from typical.
  • Because the birational map misses critical points outside its open set, a robust implementation should union left-to-right and right-to-left XR/CX charts to capture low-energy states the current method cannot see.
  • If energy-adaptive truncation (rather than Frobenius-norm SVD truncation) is used inside DMRG, the algorithm might converge to fixed-rank critical points; the paper does not test this, leaving a concrete open modification.
  • The same Rayleigh-Ritz degree and discriminant framework could be applied to other tensor-network formats, such as matrix product operators or hierarchical Tucker varieties, to benchmark their optimizers.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

3 major / 6 minor

Summary. The paper studies Rayleigh-quotient minimization over tensor train (TT) varieties, a model problem in quantum-chemistry DMRG/ALS computations. It introduces the Rayleigh–Ritz degree as the number of complex critical points for generic symmetric H, defines the Rayleigh–Ritz discriminant, and proves a birational parametrization of the TT manifold from a product of Grassmannians (Algorithm 1/2, Theorem 3.3). The authors then use homotopy continuation to compute all critical points for small TT and determinantal varieties, report RR degrees and discriminant degrees, and benchmark ALS and DMRG against these exhaustive critical-point lists. The theoretical sections include several proved results (Lemma 2.2, Theorem 3.3, Theorem 5.1, Proposition 5.6) as well as conjectures, and the numerical pipeline is implemented in TensorTrainOptimization.jl with data on Zenodo.

Significance. If the claims hold, the paper provides a useful algebro-geometric framework for a problem of genuine computational interest: it formalizes the number of critical points of the constrained Rayleigh quotient, gives a parametrization that makes numerical homotopy methods applicable, and offers the first systematic comparison of ALS/DMRG with the complete critical set. The availability of reproducible code and data is a clear strength. The main theoretical contributions (RR correspondence, RR discriminant, birational parametrization) are mathematically substantial and appear sound in their local statements. However, the central numerical claim of exhaustive computation is undermined by the authors' own examples, as detailed below.

major comments (3)
  1. [§3, Theorem 3.3 and §7, Examples 7.3, 7.4] The birational parametrization of Theorem 3.3 is only a bijection between a Zariski-open U ⊆ V⁼_{k,r} and an open W of the Grassmannian product, defined by Assumption 3.1 (first r_i rows of each flattening independent). The homotopy pipeline of §7 solves the parametric critical equations on W, so it can only find critical points whose image lies in U. This is not a hypothetical restriction: Example 7.4 states that for k=(2)^3, r=(1)^2 only 44 of 48 critical points lie in W, with the global minimum among the four missing points; Example 7.3 reports that for H5 no extrema are found because the solution is expected to lie in the complement of W. Consequently, the statement in the abstract and §7 that the method computes 'all critical points' is not supported. This is structural to Algorithm 1, not a tuning issue: no parameter homotopy on W can reach V⁼_{k,r} \ U. The authors should either r
  2. [§7.3, Table 5 and Example 7.4] The benchmark conclusion that 'ALS frequently gets stuck in local minima with suboptimal energy' is based on comparing ALS outcomes with the computed critical-point set. Since the global minimum can lie outside W (as in Example 7.4), the comparison may systematically miss the true optimal low-energy points. Thus the strength of the empirical conclusions in Table 5 and the associated discussion is not justified by the data. Unless the missing critical points are computed by an independent method, the claims should be restricted to critical points lying in W, or explicitly qualified as applying only to the computed subset.
  3. [§5.2, Proposition 5.6 and §7.5] The proof of Proposition 5.6 asserts that the image of the ramification locus under the birational map ψ×Id is dense in R_ram^V, but the argument only establishes this for a dense open subset of the domain. A rigorous proof should show that the birational map induces a surjection (or at least a dominant map) on the ramification loci, not merely on a dense open piece. This is a gap in a theoretical statement that is used to justify the discriminant-degree computations in Example 7.5. I recommend either supplying a complete proof or stating the result as a heuristic/proposition with a weaker conclusion.
minor comments (6)
  1. [§2, after (1)] The notation V⁼_{k,r} appears in §3 but is not formally defined in a displayed equation; please define it explicitly. Also, the overline notation in V_{k,r} = im(Ψ) is used inconsistently with the open set U.
  2. [§4.1, Proposition 4.6] Typo: 'Lagragian' should be 'Lagrangian'.
  3. [§6, line before Definition 6.1] Typo: 'distnace' should be 'distance'. Also, the phrase 'essentially parametrizes' is vague; please specify the sense in which the BW correspondence is parametrized.
  4. [§5.1, Example 5.5] The statement that the RR correspondence of P¹×P¹ has degree 17 is given without a reference to the supplementary material; please include the computation or point to the relevant file in [4].
  5. [§7.2, Table 2] The table would be easier to read if the units of 'time' were stated (seconds, minutes, hours are mixed). Please unify or add a note.
  6. [References] Reference [5] is listed as 'J. Software for Algebra and Geometry' without page numbers; please complete the bibliographic data.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: RR degrees and discriminants are computed by independent homotopy solves; the open-set coverage gap is an acknowledged completeness limitation, not a construction-level identification.

full rationale

The derivation chain is self-contained at the level that matters for the circularity question. The RR degree is defined as the number of complex critical points of (6), and the numerical values in Tables 1-2 are obtained by monodromy/parameter homotopy on the critical equations (Example 7.1), not by inserting the reported number as an input. The keyword RRdeg=352 in Example 7.3 is a stopping/verification count for a degree already computed by the same independent solve, not a fitted parameter. Discriminant degrees are computed by intersecting the ramification locus with a line (Example 7.5), so the degree is an output of a polynomial system, not an assumed quantity. Theorem 3.3 and Algorithms 1-2 give a genuine birational parametrization U to W with proof; the exclusion of V minus U is a real coverage limitation that the authors flag explicitly (Example 7.3: H5 expected in complement of W; Example 7.4: only 44 of 48 critical points lie in W and the global minimum is among the missing four). That limitation affects the exhaustiveness of the numerics, but it is not circular: the equations solved are not the claimed answer by construction. Cited external results [38] for RR-degree bounds and BW equivalence are attributed and not by the present authors; the self-citations [2], [26], [43] are background proofs/algorithms or an optional improvement and are not load-bearing. The conjectures (2.8, 2.13, 5.14, 5.17, 5.18) are explicitly labeled as conjectures and are not used to force the numerical claims. No step equates a fitted input with a prediction, and no uniqueness conclusion is imported from the authors' own prior work.

Assumptions & free parameters 0 free parameters · 5 assumptions · 1 invented entities

The central claims rest on standard algebraic geometry, on the domain assumption that low-rank TT tensors are the relevant variational ansatz, and on unproved numerical completeness of homotopy/monodromy. Assumption 3.1 is the most fragile technical input because the paper's own examples show it can exclude the global minimizer.

assumptions (5)
  • domain assumption Tensor trains/matrix product states are the relevant low-rank ansatz for electronic structure ground states.
    The paper borrows from the quantum-chemistry tensor-network literature (DMRG, ALS) without proving that molecular ground states are well approximated at the TT ranks used in the experiments.
  • domain assumption Real symmetric matrices model physical Hamiltonians; the complex Hermitian case is only noted in Remark 4.1 and not analyzed.
    All theory uses ψ^T Hψ and the isotropic quadric ψ^Tψ=0; physical Hamiltonians are Hermitian with ψ†Hψ, so the numerical conclusions may not transfer automatically.
  • domain assumption Homotopy continuation and monodromy with the chosen stopping criterion find all isolated solutions of the critical equations.
    Section 7.2 relies on parameter continuation and monodromy heuristics; for non-Segre TT varieties the exactness is not certified and Proposition 7.2 gives only a lower bound.
  • ad hoc to paper Assumption 3.1: in every XR factorization the first r rows of the relevant flattening are linearly independent for all critical points of interest.
    This open-set condition makes Algorithm 1 work; the paper itself reports cases (Example 7.4, H5 in Example 7.3) where it fails for the global minimizer.
  • standard math Standard algebraic-geometry background: Zariski closure equals rank-constraint variety (Lemma 2.2), intersection-theoretic degree formulas, and irreducibility of incidence correspondences.
    Used throughout Sections 2-5; proofs are sketched and rely on standard references [11,15,36].
invented entities (1)
  • Rayleigh-Ritz discriminant Σ_V
    purpose: Set of symmetric matrices H for which the critical locus of the constrained Rayleigh quotient is infinite or nonreduced; introduced to study how real critical point counts change and to compute average RR degree.
    A mathematical definition rather than a physical entity; it has no empirical handle outside the paper and therefore carries no falsification burden, but it is a new construct introduced here for TT varieties.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Numerical Algebraic Geometry for Energy Computations on Tensor Train Varieties." pith.science (2026). https://pith.science/paper/BK2AXNL2

@misc{pith2026251206939,
  author       = {Pith},
  title        = {Pith review of: Numerical Algebraic Geometry for Energy Computations on Tensor Train Varieties},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/BK2AXNL2}},
  note         = {Machine review of arXiv:2512.06939}
}
read the original abstract

We study energy minimization problems in quantum chemistry through the lens of computational algebraic geometry. We focus on minimizing the Rayleigh quotient of a Hamiltonian over a tensor train variety. The complex critical points of this problem approximate eigenstates of the quantum system, with the global minimum approximating the ground state. We call the number of critical points the Rayleigh-Ritz degree. We first study the Rayleigh-Ritz degree and introduce the Rayleigh-Ritz discriminant, which describes Hamiltonians that lead to a deficient number of critical points. We then specialize this framework to tensor train varieties: we identify instances when they are Segre products of projective spaces, report what we know about their defining ideals, and present a birational parametrization from products of Grassmannians. We use homotopy continuation to compute all critical points of this optimization problem over various tensor train and determinantal varieties. Finally, we use these results to benchmark state-of-the-art methods, the Alternating Linear Scheme and Density Matrix Renormalization Group.

Figures

Figures reproduced from arXiv: 2512.06939 by the authors.

Figure 1
Figure 1. A tensor network diagram for binary tensors of order 4 of TT rank (1 [PITH_FULL_IMAGE:figures/full_fig_p004_1.png] view at source ↗

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 2 Pith papers

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

  1. Excitation spectra and rank tomography of linear matrix product tangent spaces

    cond-mat.quant-gas 2026-07 accept novelty 6.0 of 10

    A linear MPS tangent-space method for open-boundary systems is derived, and a particle-resolved rank tomography is introduced to diagnose the tangent space's expressivity limits.

  2. Algebraic Networks and Architectural Degenerations

    math.AG 2026-06 unverdicted novelty 6.0 of 10

    Develops algebraic geometry tools for monomial neural networks and proves the singular locus of neurovarieties is contained in the architectural degeneracy locus for fully connected networks with non-increasing widths...

Reference graph

Works this paper leans on

46 extracted references · 1 linked inside Pith · cited by 2 Pith papers

  1. [38]

    Salizzoni, L

    F. Salizzoni, L. Sodomaco, and J. Weigert. Nonlinear Rayleigh quotient optimization. 2025. arXiv:2510.17760

    arXiv 2025

  2. [1]

    Bezanson, A

    J. Bezanson, A. Edelman, S. Karpinski, and V. B. Shah. Julia: A fresh approach to numerical computing.SIAM Review, 59(1):65–98, 2017

    2017

  3. [2]

    Borovik.Flat degenerations and polynomial system solving in physics and chemistry

    V. Borovik.Flat degenerations and polynomial system solving in physics and chemistry. PhD thesis, Universit¨ at Osnabr¨ uck, 2025. OsnaDocs repository

    2025

  4. [3]

    Borovik and P

    V. Borovik and P. Breiding. A short proof for the parameter continuation theorem.Journal of Symbolic Computation, 127:102373, 2025

    2025

  5. [4]

    Borovik, H

    V. Borovik, H. Friedman, S. Ho¸ sten, and M. Pfeffer. Numerical algebraic geometry for energy computations on tensor train varieties: Supplementary material. 2025. doi:10.5281/zenodo. 17777072

    doi:10.5281/zenodo 2025

  6. [5]

    Breiding, B

    P. Breiding, B. Sturmfels, and K. Wang. Computing arrangements of hypersurfaces.Journal of Software for Algebra and Geometry, 15(1):11–27, 2025. 29

    2025

  7. [6]

    Breiding and S

    P. Breiding and S. Timme. HomotopyContinuation.jl: A Package for Homotopy Continuation in Julia. InMathematical Software – ICMS 2018, pages 458–465. Springer International Publishing, 2018

    2018

  8. [7]

    Buczy´ nska, J

    W. Buczy´ nska, J. Buczy´ nski, and M. Micha lek. The Hackbusch conjecture on tensor formats. Journal de Math´ ematiques Pures et Appliqu´ ees, 104(4):749–761, 2015

    2015

Show all 46 references
  1. [8]

    Draisma, E

    J. Draisma, E. Horobet ¸, G. Ottaviani, B. Sturmfels, and R. Thomas. The Euclidean distance degree of an algebraic variety.Foundations of Computational Mathematics, 16(1):99–149, 2016

    2016

  2. [9]

    Drineas, M

    P. Drineas, M. Mahoney, and S. Muthukrishnan. Subspace sampling and relative-error matrix approximation: column-row-based methods. InProceedings of the 14th Conference on Annual European Symposium, volume 14 ofESA’06, page 304–314, Berlin, Heidelberg, 2006. Springer- Verlag

    2006

  3. [10]

    T. Duff, C. Hill, A. Jensen, K. Lee, A. Leykin, and J. Sommars. Solving polynomial systems via homotopy continuation and monodromy.IMA Journal of Numerical Analysis, 39(3):1421–1446, 2018

    2018

  4. [11]

    Eisenbud and J

    D. Eisenbud and J. Harris.3264 and all that - a second course in algebraic geometry. Cambridge University Press, Cambridge, 2016

    2016

  5. [12]

    Falc´ o, W

    A. Falc´ o, W. Hackbusch, and A. Nouy. Geometry of tree-based tensor formats in tensor Banach spaces.Annali di Matematica Pura ed Applicata. Series IV, 202(5):2127–2144, 2023

    2023

  6. [13]

    Faulstich and M

    F. Faulstich and M. Oster. Coupled cluster theory: Toward an algebraic geometry formulation. SIAM Journal on Applied Algebra and Geometry, 8(1):138–188, 2024

    2024

  7. [14]

    Faulstich, B

    F. Faulstich, B. Sturmfels, and S. Sverrisd´ ottir. Algebraic varieties in quantum chemistry.Foun- dations of Computational Mathematics, 25(4):1167–1198, 2024

    2024

  8. [15]

    Fulton.Intersection theory, volume 2 ofErgebnisse der Mathematik und ihrer Grenzgebiete (3) [Results in Mathematics and Related Areas]

    W. Fulton.Intersection theory, volume 2 ofErgebnisse der Mathematik und ihrer Grenzgebiete (3) [Results in Mathematics and Related Areas]. Springer-Verlag, Berlin, 1998

    1998

  9. [16]

    Garstka, M

    M. Garstka, M. Cannon, and P. Goulart. COSMO: A conic operator splitting method for convex conic problems.Journal of Optimization Theory and Applications, 190(3):779–810, 2021

    2021

  10. [17]

    Grasedyck

    L. Grasedyck. Hierarchical singular value decomposition of tensors.SIAM Journal on Matrix Analysis and Applications, 31(4):2029–2054, 2010

  11. [18]

    Heinen, M

    S. Heinen, M. Schwilk, G. F. von Rudorff, and O. A. von Lilienfeld. Machine learning the computational cost of quantum chemistry.Machine Learning: Science and Technology, 1(2), 2020

    2020

  12. [19]

    Helgaker, P

    T. Helgaker, P. Jorgensen, and J. Olsen.Molecular Electronic-Structure Theory. Wiley, 2014

    2014

  13. [20]

    Holtz, T

    S. Holtz, T. Rohwedder, and R. Schneider. The alternating linear scheme for tensor optimization in the tensor train format.SIAM Journal on Scientific Computing, 34(2):A683–A713, 2012

    2012

  14. [21]

    Holtz, T

    S. Holtz, T. Rohwedder, and R. Schneider. On manifolds of tensors of fixed TT-rank.Numerische Mathematik, 120(4):701–731, 2012

    2012

  15. [22]

    Huh and B

    J. Huh and B. Sturmfels. Likelihood Geometry. InCombinatorial algebraic geometry, volume 2108 ofLecture Notes in Math., pages 63–117. Springer, Cham, 2014

    2014

  16. [23]

    N. V. Ilyushechkin. The discriminant of the characteristic polynomial of a normal matrix.Matem- aticheskie Zametki, 51(3):16–23, 143, 1992

    1992

  17. [24]

    Kahle, H

    T. Kahle, H. Schenck, B. Sturmfels, and M. Wiesmann. The likelihood correspondence. 2025. arXiv:2503.02536

    2025

  18. [25]

    Kayser, A

    L. Kayser, A. Kretschmer, and S. Telen. Logarithmic discriminants of hyperplane arrangements. Le Matematiche, 80(1), 2025

    2025

  19. [26]

    Krumnow, M

    C. Krumnow, M. Pfeffer, and A. Uschmajew. Computing eigenspaces with low rank constraints. SIAM Journal on Scientific Computing, 43(1):A586–A608, 2021. 30

    2021

  20. [27]

    J. M. Landsberg.Tensors: Geometry and Applications, volume 128 ofGraduate Studies in Math- ematics. American Mathematical Society, 2012

    2012

  21. [28]

    Legat, Chris Coey, Robin Deits, Joey Huchette, and Amelia Perry

    B. Legat, Chris Coey, Robin Deits, Joey Huchette, and Amelia Perry. Sum-of-squares optimization in Julia. InThe First Annual JuMP-dev Workshop, 2017

    2017

  22. [29]

    Lin and J

    L. Lin and J. Lu.A Mathematical Introduction to Electronic Structure Theory. Society for Industrial and Applied Mathematics, Philadelphia, 2019

    2019

  23. [30]

    Lubin, O

    M. Lubin, O. Dowson, J. Dias Garcia, Joey Huchette, Beno ˆ ıt Legat, and Juan Pablo Vielma. JuMP 1.0: Recent improvements to a modeling language for mathematical optimization.Mathe- matical Programming Computation, 15:581–589, 2023

    2023

  24. [31]

    A. P. Morgan and A. J. Sommese. Coefficient-parameter polynomial continuation.Applied Math- ematics and Computation, 29(2):123–160, 1989

    1989

  25. [32]

    OSCAR - Open Source Computer Algebra Research system, Version 1.5.0, 2025

    Team OSCAR. OSCAR - Open Source Computer Algebra Research system, Version 1.5.0, 2025

    2025

  26. [33]

    Oseledets

    I. Oseledets. Tensor-Train Decomposition.SIAM Journal on Scientific Computing, 33(5):2295– 2317, 2011

    2011

  27. [34]

    Oseledets and E

    I. Oseledets and E. Tyrtyshnikov. TT-cross approximation for multidimensional arrays.Linear Algebra and its Applications, 432(1):70–88, 2010

    2010

  28. [35]

    Ranestad

    K. Ranestad. Algebraic degree in nonlinear models. InCombinatorial, computational, and applied algebraic geometry: a tribute to Bernd Sturmfels, volume 111 ofProceedings of Symposia in Pure Mathematics, pages 287–305. American Mathematical Society, Providence, 2025

    2025

  29. [36]

    Restuccia and G

    G. Restuccia and G. Rinaldo. On certain classes of degree reverse lexicographic Gr¨ obner bases. International Mathematical Forum, 2(21-24):1053–1068, 2007

    2007

  30. [37]

    Salizzoni, L

    F. Salizzoni, L. Sodomaco, and J. Weigert. Nonlinear Kalman varieties. To appear

  31. [39]

    Schollw¨ ock

    U. Schollw¨ ock. The density-matrix renormalization group in the age of matrix product states. Annals of Physics, 326(1):96–192, 2011

    2011

  32. [40]

    N. H. Scott. A theorem on isotropic null vectors and its application to thermoelasticity.Proceedings of the Royal Society, 440(1909):431–442, 1993

    1909

  33. [41]

    Seynnaeve, J

    T. Seynnaeve, J. Jovcheva, and N. Vannieuwenhoven. Minimality of tree tensor network ranks. 2025.arXiv:2509.09463

    2025 arXiv

  34. [42]

    A. J. Sommese and C. W. Wampler.The Numerical Solution of Systems of Polynomials Arising in Engineering and Science. World Scientific, 2005

    2005

  35. [43]

    Szalay, Max Pfeffer, Valentin Murg, Gergely Barcza, Frank Verstraete, Reinhold Schneider, and ¨Ors Legeza

    S. Szalay, Max Pfeffer, Valentin Murg, Gergely Barcza, Frank Verstraete, Reinhold Schneider, and ¨Ors Legeza. Tensor product methods and entanglement optimization for ab initio quantum chemistry.International Journal of Quantum Chemistry, 115(19):1342–1391, 2015

    2015

  36. [44]

    L. N. Trefethen and D. Bau.Numerical Linear Algebra. SIAM, 1997

    1997

  37. [45]

    Tyrtyshnikov

    E. Tyrtyshnikov. Incomplete cross approximation in the mosaic-skeleton method.Computing, 64(4):367–380, 2000

    2000

  38. [46]

    S. R. White. Density matrix formulation for quantum renormalization groups.Physical Review Letters, 69:2863–2866, 1992. Authors’ addresses: Max Planck Institute for Mathematics in the Sciences Leipzig, Germanyborovik@mis.mpg.de University of California Berkeley, USAhannahfried...

    1992

Pith tools

Reviewed August 3, 2026 · model on record in the stance chip above.