Pith. sign in

REVIEW 2 major objections 4 minor 21 references

The First Variational Formula and the Ostrogradsky Formalism

T0 review · 2 major / 4 minor · reviewed 2026-08-06 · deepseek-v4-flash

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

Pith's one-line read Reading the boundary term in the first variation of a higher-derivative Lagrangian reproduces Ostrogradsky's canonical momenta and Hamiltonian without any ad hoc construction.

desk verdict A clean pedagogical derivation of Ostrogradsky's construction from the boundary term, worth teaching from, but the Sec. III uniqueness claim overreaches and should be softened. read the letter →

arxiv 2608.01491 v1 pith:QN3KEEIK submitted 2026-08-02 physics.ed-ph physics.class-ph

Drew Watson , Matthew Pontius , Charles Torre This is my paper · ORCID
classification physics.ed-phphysics.class-ph PACS 45.20.Jj
keywords Ostrogradskyformalismhigher-derivativeLagrangiansfirstvariationalformulaboundarytermcanonicalmomentaHamiltonianmechanicsPais–Uhlenbeckoscillatorinstability
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 aims to show that the Ostrogradsky Hamiltonian formalism for higher-derivative Lagrangians follows from one structural fact: the boundary term in the first variation of the Lagrangian already contains the canonical pairs. For a Lagrangian $L(q, q^{(1)}, \ldots, q^{(N)}, t)$, the variation splits into an Euler-Lagrange expression plus the time derivative of a boundary term $\Theta(\delta q)$. The coefficients of $\delta q, \delta \dot{q}, \ldots, \delta q^{(N-1)}$ in $\Theta$ are exactly the momenta conjugate to $Q_i = q^{(i-1)}$, and they match Ostrogradsky's definitions. This turns a recipe that usually appears clever and ad hoc into a mechanical consequence of integration by parts, and it gives students a direct route to Hamiltonians for systems such as the Pais–Uhlenbeck oscillator. If the derivation is right, the awkward-looking Ostrogradsky momenta need not be memorized; they can be read off the variation.

What carries the argument

The central object is the boundary term $\Theta(\delta q)$ in the first variational formula. The paper isolates it with the integration-by-parts identity $F\,\delta q^{(k)} = (-1)^k \frac{d^k F}{dt^k}\delta q + \frac{d}{dt}\left(\sum_{j=0}^{k-1} (-1)^j \frac{d^j F}{dt^j}\delta q^{(k-j-1)}\right)$, applied to each term $\frac{\partial L}{\partial q^{(k)}}\delta q^{(k)}$. This produces $\Theta$ as a sum already arranged like $\sum_i P_i \delta Q_i$, so the canonical coordinates and momenta are read off directly. Equivalently, the momenta are shifted variational derivatives $P_i = \delta L/\delta q^{(i)}$ with $\delta F/\delta $q^{{(s)}}$ = \sum_{i=0}^{N-s}\left(-\frac{d}{dt}\right)^i \frac{\partia

What would settle it

Compute $\Theta(\delta q)$ for a non-degenerate higher-derivative Lagrangian such as $L = \frac{1}{2}\ddot q^2 - \frac{1}{2}\omega^2 q^2$, read off $Q_1=q$, $Q_2=\dot q$, $P_1=-q^{(3)}$, $P_2=\ddot q$, and build $H=P_1Q_2+P_2A-L$ with $A=P_2$. Then check whether Hamilton's equations are equivalent to the Euler-Lagrange equation $q^{(4)}+\omega^2 q=0$. A single non-degenerate Lagrangian for which this equivalence fails would disprove the paper's central claim.

Watch

Extended reading notes

Core claim

The paper's central claim is that the first variational formula $\delta L = E(L)\delta q + \frac{d}{dt}\Theta(\delta q)$ contains the entire canonical structure of a higher-derivative mechanical system. For a non-degenerate Lagrangian depending on derivatives up to order $N$, the boundary term is $\Theta(\delta q) = \sum_{i=1}^N \left(\sum_{j=i}^N \left(-\frac{d}{dt}\right)^{j-i} \frac{\partial L}{\partial q^{(j)}}\right)\delta q^{(i-1)}$, which has the form $\sum_i P_i \delta Q_i$ with $Q_i = q^{(i-1)}$ and $P_i = \sum_{j=i}^N \left(-\frac{d}{dt}\right)^{j-i} \frac{\partial L}{\partial q^{(j)}}$. These $P_i$ are precisely Ostrogradsky's momenta, so the Hamiltonian $H = \sum_{i=1}^{N-1} P_i

Load-bearing premise

The load-bearing premise is that the Lagrangian depends non-degenerately on its highest derivative, $\partial^2 L/\partial (q^{(N)})^2 \neq 0$, so that $q^{(N)}$ can be solved for in terms of the canonical variables and the Hamiltonian can be built; degenerate higher-derivative Lagrangians are explicitly outside the paper's scope.

Editorial extensions

If this is right

  • For $N=1$, $\Theta = (\partial L/\partial \dot q)\delta q$, so the boundary-term rule reduces to the standard momentum definition and the usual Hamiltonian.
  • For $N=2$, it recovers $P_1 = \partial L/\partial \dot q - \frac{d}{dt}\partial L/\partial \ddot q$ and $P_2 = \partial L/\partial \ddot q$, the Ostrogradsky pairs used in the worked examples.
  • The boundary term organizes the $2N$ initial data of a generically $2N$-th-order Euler-Lagrange equation into $N$ canonical pairs, explaining the dimension of Ostrogradsky's phase space.
  • Because the Hamiltonian is linear in at least one canonical momentum for every non-degenerate higher-derivative Lagrangian, the construction reproduces the Ostrogradsky instability: the Pais–Uhlenbeck example has no ground state.
  • The accompanying code implements the derivation, so a Hamiltonian for a given non-degenerate higher-derivative Lagrangian can be produced by algorithm rather than by hand.

Reading between the lines

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

  • The same boundary-term logic could be applied to degenerate higher-derivative Lagrangians: $\Theta(\delta q)$ is still defined even when $\partial^2 L/\partial (q^{(N)})^2 = 0$, so it may constrain the constrained Hamiltonian analysis, but the paper does not pursue that.
  • Reading momenta as shifted variational derivatives, $P_i = \delta L/\delta q^{(i)}$, suggests a uniform computational rule for multi-degree-of-freedom and field-theory generalizations that could be tested on textbook Lagrangians.
  • If the boundary term is taken as the primary object, the Legendre transform becomes secondary: one could in principle derive Hamilton's equations directly from $\Theta$ without first constructing $H$, an extension the paper leaves implicit.
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

2 major / 4 minor

Summary. The paper presents a derivation of Ostrogradsky's Hamiltonian formalism for higher-derivative Lagrangians starting from the boundary term in the first variation. Section IV computes the boundary term for L(q, q^(1), ..., q^(N)) by repeated integration by parts and reads off canonical pairs Q_i = q^(i-1), P_i = Σ_{j≥i} (-d/dt)^{j-i} ∂L/∂q^(j), exactly the Ostrogradsky momenta. Worked examples (perturbed harmonic oscillator, Pais–Uhlenbeck oscillator), multi-variable formulas, and supplementary computer code are provided. The central integration-by-parts computation is correct and reproduces the standard Ostrogradsky results. Section III attempts to justify why the boundary term 'defines' canonical pairs; this justification contains an overclaim that is not load-bearing for the algebraic formula but is load-bearing for the paper's stated pedagogical thesis.

Significance. If the presentation is corrected, the paper offers a genuinely useful undergraduate-level route to the Ostrogradsky construction and makes explicit a connection between boundary terms and canonical momenta that is often left implicit. The derivation of the boundary term (Eqs. (59)–(68)) is self-contained and correct, the examples are accurate, and the accompanying code is a valuable supplement. The main weakness is Section III's claim of uniqueness: the boundary term does not uniquely determine canonical pairs, since rescalings Q_i → c_i Q_i, P_i → P_i/c_i preserve P_i δQ_i. The Ostrogradsky formulas are correct as a standard and natural convention, but not as a forced consequence of Eq. (56). This is a local, fixable gap rather than an error in the main computation.

major comments (2)
  1. [Section III, Eqs. (56)–(57)] The assertion that Eq. (56) 'is only satisfied if' (Q_i, P_i) = (q_i, ∂L/∂q̇_i) is false. For L = ½ q̇², take Q = 2q, P = q̇/2. Then P δQ = q̇ δq, so Eq. (56) holds, and with H = 2P² Hamilton's equations are equivalent to q̈ = 0. More generally, any rescaling Q_i → c_i Q_i, P_i → P_i/c_i preserves P_i δQ_i, and adding an exact form changes the boundary term by a total derivative without affecting the variational principle. Thus Eq. (56) fixes the momenta only after one chooses Q_i = q^(i-1). That choice is a convention, not a consequence. Please either prove uniqueness under a stated normalization or reframe the claim as selecting the natural/standard canonical pair.
  2. [Section IV, Eq. (69)] The identification of Θ with P_i δQ_i and the subsequent reading off of Eq. (69) rely on the flawed uniqueness argument in Section III. In the higher-derivative case the same rescaling freedom exists: with Q_i = c_i q^(i-1) and P_i = P_i^O/c_i, the boundary term is unchanged. The paper should state explicitly that Eq. (69) is the canonical choice obtained by taking Q_i = q^(i-1), which is the standard Ostrogradsky convention. The boundary term alone does not select this pair; the derivation yields the correct formulas only after that coordinate choice is made.
minor comments (4)
  1. [Section III, Eq. (53)] The notation L on both sides is confusing: the left side is the configuration-space Lagrangian and the right side is the phase-space Lagrangian of Eq. (49). Also, if the goal is to discuss general canonical transformations, the equality in Eq. (53) is stronger than necessary; the two Lagrangians may differ by a total derivative. This is related to the overclaim in Eqs. (56)–(57).
  2. [Section I and Discussion] The claim that 'no derivation of Ostrogradsky's construction exists in the modern literature which emphasizes the role of the boundary term' is too strong, given that Refs. [5–8] (Barnich–Henneaux–Schomblond, Lee–Wald, Crnkovic–Witten, Torre) already use boundary terms in covariant phase-space constructions. The paper's contribution is better framed as an elementary, self-contained exposition for mechanics rather than the first boundary-term-based derivation.
  3. [Section IV, Eq. (70)] The notation δF/δq^(s) for the variational derivative may be confused with the variation δF. Consider denoting this quantity by E_s(F) or a similar symbol, to avoid confusion with the variation of F.
  4. [Supplementary material] The manuscript mentions a computer program and a GitHub repository but gives little detail about the algorithm or its testing. If space permits, a short appendix describing the algorithm's handling of the non-degeneracy condition (12) and a few test cases would strengthen the reproducibility claim.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the boundary-term derivation is self-contained; an overstrong uniqueness claim in Sec. III is a correctness flaw, not a circular step.

full rationale

The paper's central derivation is self-contained. It computes the first variational formula for a higher-derivative Lagrangian by repeated integration by parts (Eq. 59), obtaining the Euler-Lagrange expression and a boundary term Θ (Eq. 63). The canonical momenta are then read off as coefficients of δq^(i-1) in Θ (Eq. 69). This is a direct algebraic identification, not a circular one: the boundary term is a function of L and its derivatives, and the momenta are defined by those derivatives. No fitted parameters are involved. The paper cites Woodard [13] for the equivalence of Hamilton and Euler-Lagrange equations and for the Ostrogradsky formulas, and it cites the authors' own prior work [5] for context, but neither citation is load-bearing for the derivation, which is performed in the paper itself. The only substantive issue is the uniqueness claim in Section III, where Eq. (56) is said to force the canonical pairing (57). As the skeptic notes, rescaling Q and P preserves the boundary term P δQ, so Eq. (56) does not uniquely determine (57). However, this is a logical gap in the justification of 'the boundary term defines' the canonical pairs, not a circular dependence of the conclusion on its premise. The Ostrogradsky formulas (69) remain a valid construction, and the paper's derivation of them from the boundary term is a genuine derivation, albeit one whose uniqueness is overstated. Thus there is no circularity: the result is not assumed as an input, and the derivation is independently checkable.

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

No free parameters are fitted; the parameters in examples (epsilon, omega, omega1, omega2) are model inputs. The derivation relies on standard integration by parts and summation identities, plus the stated non-degeneracy restriction. Section III's uniqueness assertion is an unproven extra premise that is stronger than needed.

assumptions (5)
  • standard math Integration by parts identity, Eq. (59): F delta q^{(k)} = (-1)^k (d^k F/dt^k) delta q + d/dt [sum_j (-1)^j (d^j F/dt^j) delta q^{(k-j-1)}].
    Used to convert delta L into the Euler-Lagrange expression plus a boundary term; this is the engine of the derivation.
  • standard math Summation identity, Eq. (60), converting sum_{j=1}^N sum_{i=1}^j to sum_{i=1}^N sum_{j=i}^N.
    Needed to arrange the boundary term into coefficients multiplying each delta q^{(i-1)}.
  • domain assumption Non-degeneracy condition, Eq. (12): partial^2 L / partial (q^{(N)})^2 != 0.
    Required to solve q^{(N)} = A(Q_i, P_N) and construct the Hamiltonian; degenerate Lagrangians are excluded in the Introduction.
  • ad hoc to paper Equality of boundary terms uniquely fixes the canonical pair, asserted around Eq. (56)-(57).
    The paper claims the boundary term is 'only satisfied if' Qi = qi and Pi = partial L / partial qdot_i, but canonical transformations preserve P dQ up to a total derivative, so this uniqueness is stronger than justified.
  • domain assumption Hamilton's equations and the Euler-Lagrange equation are equivalent for non-degenerate higher-derivative Lagrangians.
    The paper cites Woodard [13] for this equivalence, which is needed to claim the Hamilton equations reproduce Eq. (9).

how reviews work

0 comments
Cite this review

Pith. "Pith review of The First Variational Formula and the Ostrogradsky Formalism." pith.science (2026). https://pith.science/paper/QN3KEEIK

@misc{pith2026260801491,
  author       = {Pith},
  title        = {Pith review of: The First Variational Formula and the Ostrogradsky Formalism},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/QN3KEEIK}},
  note         = {Machine review of arXiv:2608.01491}
}
read the original abstract

We present a derivation at a level suitable for undergraduates of the Ostrogradsky formalism for Lagrangians in classical mechanics that depend upon an arbitrary number of time derivatives of the configuration. From the boundary term in the first variation of the Lagrangian we derive the Ostrogradsky formulas that define the Hamiltonian formulation of mechanical systems. Worked examples, exercises, and applications to the literature are also provided. An accompanying computer program that implements the formalism is discussed in the Supplementary Materials, and code for computing Hamiltonians via the Ostrogradsky formalism is provided in the Supplementary Materials and in a GitHub repository.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

21 extracted references · 19 canonical work pages

  1. [13]

    Pavˇ siˇ c, International Journal of Geometric Methods in Modern Physics13, 1630015 (2016)

    M. Pavˇ siˇ c, International Journal of Geometric Methods in Modern Physics13, 1630015 (2016)

    work page 2016

  2. [5]

    Lanczos,The Variational Principles of Mechanics(Dover Publications, 1986)

    C. Lanczos,The Variational Principles of Mechanics(Dover Publications, 1986)

    work page 1986

  3. [1]

    Goldstein,Classical Mechanics(Addison-Wesley, 1980)

    H. Goldstein,Classical Mechanics(Addison-Wesley, 1980)

    1980

  4. [2]

    Higher-order terms also appear in attempts to improve the ultraviolet behavior of quantum field theories, including higher- curvature modifications of gravity

    ˙x2 +ω 2 1ω2 2x2 ,(27) optionally supplemented by a self-interaction term− Λ 4 x4. Higher-order terms also appear in attempts to improve the ultraviolet behavior of quantum field theories, including higher- curvature modifications of gravity. The price of introducing such terms is that the Hamilto- nian formulation may acquire additional degrees of freedo...

    work page

  5. [3]

    boundary term

    ˙x−...x .(33) The non-degeneracy condition holds since ∂2L ∂¨x2 = 1̸= 0, and we can solve A≡¨x=P 2.(34) Substituting into the definition H=P 1Q2 +P 2A−L(Q 1, Q2, A),(35) 7 we obtain H(Q 1, Q2, P1, P2) =P 1Q2 + 1 2 P 2 2 + 1 2 (ω2 1 +ω 2 2)(Q2)2 − 1 2 ω2 1ω2 2(Q1)2.(36) If the interaction Λ 4 x4 is included in (27), then (36) is modified by the additional ...

    work page

  6. [4]

    L. D. Landau and E. M. Lifshitz,Mechanics(Butterworth-Heinemann, 1976)

    work page 1976

  7. [6]

    Whittaker,A Treatise on the Analytical Dynamics of Particles and Rigid Bodies: With an Introduction to the Problem of Three Bodies, 2nd ed

    E. Whittaker,A Treatise on the Analytical Dynamics of Particles and Rigid Bodies: With an Introduction to the Problem of Three Bodies, 2nd ed. (Cambridge University Press, 1917) see pp. 265–267

    work page 1917

  8. [7]

    C. G. Torre, Journal of Mathematical Physics33, 3802 (1992)

    work page 1992

Show all 21 references
  1. [8]

    Barnich, M

    G. Barnich, M. Henneaux, and C. Schomblond, Phys. Rev. D44, R939 (1991)

    1991

  2. [9]

    Lee and R

    J. Lee and R. M. Wald, Journal of Mathematical Physics31, 725 (1990)

    1990

  3. [10]

    Crnkovic and E

    C. Crnkovic and E. Witten, Covariant description of canonical formalism in geometrical the- ories (1986)

    1986

  4. [11]

    Woodard, Avoiding dark energy with 1/rmodifications of gravity, inThe Invisible Universe: Dark Matter and Dark Energy(Springer Berlin Heidelberg, 2007) pp

    R. Woodard, Avoiding dark energy with 1/rmodifications of gravity, inThe Invisible Universe: Dark Matter and Dark Energy(Springer Berlin Heidelberg, 2007) pp. 403–433

    2007

  5. [12]

    K. S. Stelle, Phys. Rev. D16, 953 (1977). 15

    1977

  6. [14]

    Ostrogradsky, Mem

    M. Ostrogradsky, Mem. Acad. St. Petersbourg6, 385 (1850)

  7. [15]

    R. P. Woodard, Scholarpedia10, 32243 (2015), arXiv:1506.02210 [hep-th]

    2015 arXiv

  8. [16]

    P. A. M. Dirac,Lectures on Quantum Mechanics(Dover Publications, 2001)

    2001

  9. [17]

    J. D. Brown, Universe8, 171 (2022)

    2022

  10. [18]

    J. D. Brown, American Journal of Physics91, 214 (2023). [17]https://github.com/Drew-Watson-117/hamiltonian

    2023

  11. [19]

    Smilga, Nuclear Physics B706, 598 (2005)

    A. Smilga, Nuclear Physics B706, 598 (2005)

    2005

  12. [20]

    I. B. Ilhan and A. Kovner, Phys. Rev. D88, 044045 (2013)

    2013

  13. [21]

    C. A. M. de Melo and I. F. de Souza, American Journal of Physics94, 230 (2026). 16

    2026

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.