Pith. sign in

REVIEW 3 major objections 5 minor 1 cited by

Carleman Linearization of Partial Differential Equations

T0 review · 3 major / 5 minor · reviewed 2026-08-12 · deepseek-v4-flash

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

Pith's one-line read This paper extends Carleman linearization from ODEs to quadratically nonlinear PDEs, embedding them exactly into infinite-dimensional linear PDE systems via continuous Kronecker powers.

desk verdict The core formal construction is correct and clearly exposed, but the advertised practical claims outrun the evidence, as the paper itself concedes. read the letter →

arxiv 2412.00014 v1 pith:UZEN2HAK submitted 2024-11-14 math.GM

Tamas Vaszary This is my paper · ORCID
classification math.GM MSC 35A2235Q5335Q83
keywords carlemanlinearizationpartialdifferentialequationsquadraticnonlinearitykroneckerproductvariablechangeoperatorburgersequationvlasovquantumsimulation
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

Dynamical systems with polynomial nonlinearity can be rewritten as infinite linear systems by Carleman linearization, a procedural method whose auxiliary variables are the Kronecker powers of the state vector. This paper shows the same construction works for partial differential equations with quadratic nonlinearities: one uses different copies of the coordinate variable in each factor of the Kronecker product, so the new variables are products of the field at different spatial points. The paper proves that these product variables obey a tridiagonal linear PDE system, and that truncating that system at finite order yields a finite linear PDE system with an explicit formal solution. The authors demonstrate the construction on Burgers' equation and on the Vlasov equation, and argue that the approach removes coordinate discretization error, a step toward solving nonlinear PDEs on analog quantum simulators.

What carries the argument

The continuous analogue of the Kronecker product: for vector fields $u(x)$, the $i$-th level variable is $u(x_1)\otimes \cdots \otimes u(x_i)$, a function on $i$ copies of the coordinate space. The paper also introduces the variable-change operator that replaces one coordinate variable by another, expressed both abstractly and as an integral against a Dirac delta; these operators let the quadratic term $F_2(x;w)[u(x)\otimes u(w)]$ be rewritten as a linear operator acting on the next level $y_{i+1}$ after suitable relabelling of coordinates. Together they give the tridiagonal linear structure and make the whole construction procedural.

What would settle it

Solve the linearized hierarchy for a smooth analytic solution of inviscid Burgers with known exact solution, and compare the first component of the truncated system (23) as $N$ grows against the exact solution: if the difference does not shrink to zero on the time interval where the exact solution is smooth and analytic, the exact-embedding claim is false. Alternatively, for the Vlasov example, check whether the truncated hierarchy preserves total mass; a violation that does not vanish as $N$ grows would also falsify exact embedding.

Watch

Extended reading notes

Core claim

The central claim is Proposition 3.1: for a PDE of the form $\partial u/\partial t = F_0(x) + F_1(x)u(x) + F_2(x;w)[u(x)\otimes u(w)]$, the auxiliary functions $y_i = u(x_1)\otimes \cdots \otimes u(x_i)$ evolve according to the tridiagonal linear PDE system $\partial y_i/\partial t = A^i_{i-1}(X_i) y_{i-1} + A^i_i(X_i) y_i + A^i_{i+1}(X_i; x_{i+1}) y_{i+1}$, with the explicit linear operators (16a)-(16c) built from $F_0$, $F_1$, $F_2$ and variable-change operators. It follows that every quadratically nonlinear PDE of this form is exactly embedded into an infinite-dimensional linear system, and truncating at level $N$ gives a finite linear PDE system with formal exponential solution (23). The paper applies this to Burgers' equation, where the inviscid case yields a Taylor-series solution in time, and to a two-species Vlasov–Poisson system.

Load-bearing premise

The construction assumes the infinite linear system can be cut off at a finite level without losing the solution of the original PDE: the truncated system must be well-posed and its solution must converge to the true solution as the cutoff $N$ grows, and the paper states that no error analysis has been performed.

Editorial extensions

If this is right

  • Any quadratically nonlinear PDE that fits the form (11) can be recast as an infinite linear PDE system, so methods and intuition for linear PDEs become applicable to nonlinear ones.
  • Truncating at order $N$ gives a finite linear PDE system whose formal solution is an exponential, so time integration can in principle be done without discretizing the coordinate space.
  • The inviscid Burgers example produces an explicit Taylor-series representation of the solution, suggesting a time-integrator built directly from the linearized hierarchy.
  • The Vlasov example shows the framework covers integro-differential nonlinearities, not just local ones, as long as they can be written in the quadratic form with dummy variables.
  • Because the linear system lives on continuous coordinate copies, it is a candidate for analog quantum simulation, eliminating coordinate discretization error entirely.

Reading between the lines

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

  • A natural testable extension is an error bound for the truncated hierarchy in the style of the ODE Carleman bounds, using a continuous analogue of the log-norm; the paper states this is the next step.
  • The framework may extend to polynomial nonlinearities of any degree by the same quadratization procedure used for ODEs, which the paper anticipates but does not prove.
  • If well-posedness of the linearized system can be established, the method could be compared numerically against spectral or finite-difference solvers on shock problems, where the Taylor-series form (33) is expected to have finite radius of convergence.
  • For kinetic equations like Vlasov, the exact linear embedding could provide a new route to deriving reduced models or closures by projecting the linear hierarchy.
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 generalizes Carleman linearization from ODEs to PDEs with quadratic nonlinearities of the form (11). It defines auxiliary variables y_i = u(x1) ⊗ ... ⊗ u(xi) and proves in Proposition 3.1 that their evolution is given by the tridiagonal infinite linear system (15) with operators (16a)-(16c). The system is then truncated at level N, yielding the finite linear PDE system (20)-(22) with a formal solution written as an exponential (23). Two examples are worked out, the Burgers equation and the Vlasov equation. The paper concludes by noting that an error analysis has not yet been performed and proposes this as future work.

Significance. If backed by a rigorous truncation and convergence analysis, the construction would provide a procedural, exact embedding of a class of quadratically nonlinear PDEs into a linear hierarchy, with potential applications to quantum simulation and numerical methods. The Leibniz-rule derivation in Proposition 3.1 is explicit and checkable by hand, and the two examples cover both a scalar conservation law and a kinetic equation. The paper is commendably honest in stating that no error analysis is performed, but this honesty also exposes the gap between the formal construction and the practical claims of solvability and elimination of discretization error.

major comments (3)
  1. [§3.2, Eq. (23)] The formal exponential solution z(X_N,t) = exp(A_N t) z0 + ... is only meaningful if the operator A_N(X_N), whose entries are unbounded partial differential and integral operators, generates a C0-semigroup on a suitable Banach space over R^{mN}. The manuscript specifies no such function space and proves no generation theorem, so (23) remains a formal expression rather than an analytical solution. Since the paper's stated practical advantage is the solvability of the truncated linear system, this gap is load-bearing.
  2. [§5, first two paragraphs] The paper explicitly states that 'an error analysis has not yet been performed' and that a truncation error bound 'is the next step to be done.' Without a bound on the difference between solutions of the infinite hierarchy (15) and the truncated system (20), or a convergence result as N→∞, the central practical claim that the truncation yields a usable computational method is not supported. The analogy to the ODE error bound of Ref. [10] is not a substitute for a PDE analysis, which must address unbounded spatial operators and the choice of function spaces.
  3. [§4.1, Eq. (33) and §3.1, Eq. (12)] For the inviscid Burgers equation (µ=0), which is presented as the first toy example, smooth initial data generally develop shocks in finite time. The hierarchy (12) requires u(x1,t)...u(xi,t) to be a classical function for each i, so the hierarchy and the Taylor series (33) cease to be well-defined at the shock time. The paper calls (33) a formal solution and suggests it 'may form the basis of a time-integrator,' but this example illustrates a concrete obstruction to the method's global validity that should be discussed or excluded.
minor comments (5)
  1. [§3.2] The domain of z is written as R^{m·n} × R≥0; it should be R^{mN} × R≥0 (or R^{m·N}), since X_N contains N copies of the m-dimensional coordinate variable.
  2. [Proposition 3.1 and Eq. (16a)] For i=1, the operator A^1_0 as defined involves the variable-change condition x_j = x_{j+1} for j≥1, but no x_2 exists in the one-level state; this case should be treated separately, as is implicitly done in (20) by making F0 an inhomogeneity.
  3. [Eq. (22)] The bottom-left block of the matrix A_N should be A^N_{N-1}, consistent with (20); as printed it appears to be A^{N-1}_N, which is the coupling in the second-to-last row.
  4. [§4.1] The sentence 'The F2 operator is not unique generally' raises a question: different choices of F2 lead to different operators A^i_{i+1} in the hierarchy, so the manuscript should state whether the resulting linear systems are equivalent, or at least that the non-uniqueness is harmless.
  5. [Abstract and §1] The claim that the method 'eliminate[s] coordinate discretization error completely' should be qualified in light of the admitted absence of an error analysis in Section 5; at present the claim is only conditional on the existence of a convergent truncation.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the PDE Carleman construction is an exact algebraic rewriting proved in line, with no fitted prediction or load-bearing self-citation.

full rationale

The central claim, Proposition 3.1, is a direct derivation rather than a circular prediction: the auxiliary variables y_i are defined independently in (12), and their time derivatives are computed by differentiating that definition, substituting the given PDE (11), and applying the Leibniz rule. The operators in (16a)-(16c) are then constructed with variable-change operators so that the resulting terms are exactly A_i^{i-1} y_{i-1}, A_i^i y_i, and A_i^{i+1} y_{i+1}. This is an identity by construction, but it is not a case of defining X in terms of Y and then claiming to predict Y: the paper's entire goal is to exhibit such an embedding, and the finite truncation (20) follows from dropping the y_{N+1} term. No parameter is fitted to data, no uniqueness theorem is imported from the author's own prior work, and no empirical result is renamed. The cited references [10], [15], and [17] are used for background, for the known ODE truncation error analysis, and for the historical continuous-Kronecker-product idea; the present paper re-derives its own formulas and explicitly notes that the analogous PDE error analysis has not yet been performed. That absence is an unverified correctness or convergence gap, not circularity. The formal exponential solution (23) is a restatement of the linearity of the truncated system and does not assume the PDE solution it is meant to represent. Accordingly, the paper warrants a circularity score of 0.

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

No free parameters are fitted. The framework introduces auxiliary product functions y_i, but these are defined directly from the state without new physical postulates. The main unstated burden is the assumed regularity and well-posedness of the linearized hierarchy.

assumptions (4)
  • domain assumption The solution u and initial data u0 are smooth enough that all products u(x1)...u(xi) are well-defined and the differential and integral operators in (16a)-(16c) can act.
    The hierarchy (12) and the formal exponential solution (23) require strong regularity; the paper notes this in Section 5 without proving conditions.
  • domain assumption The linear operator A_N(X_N) in (22) generates a semigroup, so the formal solution (23) is meaningful.
    No spectral or well-posedness analysis is given for the truncated linear PDE system.
  • domain assumption Every quadratically nonlinear PDE of interest can be written in the form (11) with time-independent F0, F1, F2.
    The two examples fit this form, but general PDEs may include explicit time or space dependent operator coefficients or non-polynomial nonlinearities not covered.
  • standard math Higher-degree polynomial nonlinearities can be reduced to quadratic ones as in [10, Section 3.3].
    The paper relies on this cited result to claim generality beyond quadratic nonlinearities.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Carleman Linearization of Partial Differential Equations." pith.science (2026). https://pith.science/paper/UZEN2HAK

@misc{pith2026241200014,
  author       = {Pith},
  title        = {Pith review of: Carleman Linearization of Partial Differential Equations},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/UZEN2HAK}},
  note         = {Machine review of arXiv:2412.00014}
}
read the original abstract

Carleman linearization is a technique that embeds systems of ordinary differential equations with polynomial nonlinearities into infinite dimensional linear systems in a procedural way. In this paper we generalize the method for systems of partial differential equations with quadratic nonlinearities, while maintaining the original structure of Carleman linearization. Furthermore, we apply our approach to Burger's equation and to the Vlasov equation as examples.

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

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

  1. Solving the Nonlinear Vlasov Equation on a Quantum Computer

    quant-ph 2024-11 conditional novelty 6.0 of 10

    For a 1+1 dimensional Vlasov model with Krook collisions, a Carleman linearization quantum algorithm has polynomially worse complexity than classical finite difference methods and requires unphysically large collision...

Reference graph

Works this paper leans on

22 extracted references · 6 canonical work pages · cited by 1 Pith paper

  1. [10]

    Explicit Error Bounds for Carleman Line arization,

    M. Forets and A. Pouly, “Explicit Error Bounds for Carleman Line arization,” 2017, https://doi.org/10.48550/arXiv.1711.02552. 11

    arXiv doi:10.48550/arxiv.1711.02552 2017

  2. [17]

    Kowalski and W.-H

    K. Kowalski and W.-H. Steeb, Nonlinear Dynamical Systems and Carleman Linearization . World Scientific, 1991, https://doi.org/10.1142/1347

    doi:10.1142/1347 1991

  3. [1]

    Quantum algorithm for lin - ear systems of equations,

    A. W. Harrow, A. Hassidim, and S. Lloyd, “Quantum algorithm for lin - ear systems of equations,” Phys. Rev. Lett. , vol. 103, p. 150502, 2009, https://doi.org/10.1103/PhysRevLett.103.150502

    doi:10.1103/physrevlett.103.150502 2009

  4. [2]

    A theory of quantum diffe rential equation solvers: limitations and fast-forwarding,

    D. An, J.-P. Liu, D. Wang, and Q. Zhao, “A theory of quantum diffe rential equation solvers: limitations and fast-forwarding,” 2023, https://doi.org/10.48550/arXiv.2211.05246

    arXiv doi:10.48550/arxiv.2211.05246 2023

  5. [3]

    An approximate fourier transform useful in q uantum factoring,

    D. Coppersmith, “An approximate fourier transform useful in q uantum factoring,” 2002, https://doi.org/10.48550/arXiv.quant-ph/0201067

    arXiv doi:10.48550/arxiv.quant-ph/0201067 2002

  6. [4]

    Hamiltonian Systems and Transformation in Hilber t Space,

    B. O. Koopman, “Hamiltonian Systems and Transformation in Hilber t Space,” Pro- ceedings of the National Academy of Sciences , vol. 17, no. 5, pp. 315–318, 1931, https://doi.org/10.1073/pnas.17.5.315

    doi:10.1073/pnas.17.5.315 1931

  7. [5]

    What Is the Koo pman Operator? A Simplified Treatment for Discrete-Time Systems,

    A. L. Bruce, V. M. Zeidan, and D. S. Bernstein, “What Is the Koo pman Operator? A Simplified Treatment for Discrete-Time Systems,” 2019 American Control Conference (ACC), pp. 1912–1917, 2019, https://doi.org/10.23919/ACC.2019.8814306

    arXiv 2019

  8. [6]

    Chaos as an intermittently forced linear system,

    S. L. Brunton, B. W. Brunton, J. L. Proctor, E. Kaiser, and J. N. Kutz, “Chaos as an intermittently forced linear system,” Nature Communications , vol. 8, no. 1, 2017, https://doi.org/10.1038/s41467-017-00030-8

    doi:10.1038/s41467-017-00030-8 2017

Show all 22 references
  1. [7]

    Modern Koopman Theory for Dynam- ical Systems,

    S. L. Brunton, M. Budiˇ si´ c, E. Kaiser, and J. N. Kutz, “Modern Koopman Theory for Dynam- ical Systems,” 2021, https://doi.org/10.48550/arXiv.2102.12086

    2021 doi arXiv

  2. [8]

    Application de la th´ eorie des ´ equations int´ egrale s lin´ eaires aux syst` emes d’´ equations diff´ erentielles non lin´ eaires,

    T. Carleman, “Application de la th´ eorie des ´ equations int´ egrale s lin´ eaires aux syst` emes d’´ equations diff´ erentielles non lin´ eaires,”Acta Mathematica , vol. 59, pp. 63–87, 1932, https://doi.org/10.1007/BF02546499

    1932 doi

  3. [9]

    Minisini, A

    J. Minisini, A. Rauh, and E. Hofer, Carleman Linearization for Approximate So- lutions of Nonlinear Control Problems: Part 1 - Theory , 2005, pp. 215–222. [Online]. Available: https://www.researchgate.net/publication/347889197 CARLEMAN LINEARIZATION FOR APPROXIMA

    2005

  4. [11]

    Carleman linearization of nonlinear systems and its finite-section approximations,

    A. Amini, C. Zheng, Q. Sun, and N. Motee, “Carleman linearization of nonlinear systems and its finite-section approximations,” Discrete and Continuous Dynamical Systems - B , vol. 30, no. 2, pp. 577–603, 2025, https://doi.org/10.3934/dcdsb.2024102

    2025 doi

  5. [12]

    Solving nonlinear aeronautical problems using the Carleman linearization method,

    B. W. Gaude, “Solving nonlinear aeronautical problems using the Carleman linearization method,” 2001. [Online]. Available: https://commons.erau.edu/db-theses/307/

    2001

  6. [13]

    Efficient quantum algorithm for dissipative nonlinear differential equ ations,

    J.-P. Liu, H. Øie Kolden, H. K. Krovi, N. F. Loureiro, K. Trivisa, an d A. M. Childs, “Efficient quantum algorithm for dissipative nonlinear differential equ ations,” Proceed- ings of the National Academy of Sciences , vol. 118, no. 35, p. e2026805118, 2021, https://doi.org/10.10...

    2021 doi

  7. [14]

    E fficient quantum algorithm for nonlinear reaction–diffusion equations and energy estimation,

    J.-P. Liu, D. An, D. Fang, J. Wang, G. H. Low, and S. Jordan, “E fficient quantum algorithm for nonlinear reaction–diffusion equations and energy estimation,” Quantum, vol. 5, p. 502, 2023, https://doi.org/10.1007/s00220-023-04857-9

    2023 doi

  8. [15]

    An efficient quant um algorithm for simulating polynomial differential equations,

    A. Surana, A. Gnanasekaran, and T. Sahai, “An efficient quant um algorithm for simulating polynomial differential equations,” 2023, https://doi.org/10.48550/arXiv.2212.10775

    2023 doi arXiv

  9. [16]

    Improved quantum algorithms for linear and nonlinear differential equations,

    H. Krovi, “Improved quantum algorithms for linear and nonlinear differential equations,” Quantum, vol. 7, p. 913, 2023, https://doi.org/10.22331/q-2023-02-02-913

    2023 doi

  10. [18]

    Infinite-dimensional Carleman linearization, the L ie series and optimal control of non-linear partial differential equations,

    S. P. Banks, “Infinite-dimensional Carleman linearization, the L ie series and optimal control of non-linear partial differential equations,” International Journal of Systems Science , vol. 23, no. 5, pp. 663–675, 1992, https://doi.org/10.1080/00207729208949241

    1992 doi

  11. [19]

    Global exact quadratization of continuous- time nonlinear control sys- tems,

    F. Carravetta, “Global exact quadratization of continuous- time nonlinear control sys- tems,” SIAM Journal on Control and Optimization , vol. 53, no. 1, pp. 235–261, 2015, https://doi.org/10.1137/130915418

    2015 doi

  12. [20]

    Exact and optimal quadra- tization of nonlinear finite-dimensional non-autonomous dynamical systems,

    A. Bychkov, O. Issan, G. Pogudin, and B. Kramer, “Exact and optimal quadra- tization of nonlinear finite-dimensional non-autonomous dynamical systems,” 2023, https://doi.org/10.48550/arXiv.2303.10285

    2023 doi arXiv

  13. [21]

    Optimal monomial quadratization f or ode systems,

    A. Bychkov and G. Pogudin, “Optimal monomial quadratization f or ode systems,” 2021, https://doi.org/10.48550/arXiv.2103.08013

    2021 doi arXiv

  14. [22]

    Dissipative quadratizations of polynomia l ODE systems

    Y. Cai and G. Pogudin, “Dissipative quadratizations of polynomia l ODE systems.” Springer Nature Switzerland, 2024, pp. 323–342, https://doi.org/10.1007/978-3-031-57249-4 16. 12

    2024 doi

Pith tools

Reviewed August 12, 2026 · model on record in the stance chip above.