Classification of Hamiltonian operators and Hamiltonian pairs in the formal variational calculus beyond the Gel'fand–Dorfman affine class

The Gel'fand–Dorfman paper of 1981 characterized, within the formal variational calculus, all Hamiltonian differential operators whose coefficients are affine in the field variables — linear in the fields plus constant-coefficient terms — through an exact correspondence with graded infinite-dimensional Lie algebras: the admissible constant-coefficient deformations are the 2-cocycles on the associated algebra (trivial cocycles corresponding to the constant shifts that introduce a spectral parameter), and Hamiltonian pairs correspond to compatible pairs of such structures. What this leaves open is that operators affine in the fields do not exhaust the Hamiltonian operators needed for Hamiltonian systems, and that no classification of general Hamiltonian operators or Hamiltonian pairs is known. The general classification problem remains open: only low orders and special structure classes are classified, the literature describes the general problem as very difficult, and in at least one natural restriction a discrete classification is obstructed by nontrivial moduli.

Background

An evolution equation $u_t=K(u)$ for fields $u_i(x)$ is called Hamiltonian if it can be written as a matrix differential operator $H$ applied to the variational (Euler–Lagrange) derivative of a Hamiltonian functional $\int f\,dx$ whose density is a differential polynomial — a polynomial in the fields and finitely many of their $x$-derivatives; this polynomial setting is the formal variational calculus, the standard framework for the Hamiltonian theory of evolution equations. Such an $H$ is a Hamiltonian operator if it is skew-symmetric and the induced bracket on functionals, ${f,g}_H = \int(\delta f/\delta u)^T H(\delta g/\delta u)\,dx$, satisfies the Jacobi identity. Gel'fand and Dorfman (1979, 1980) proved that the Jacobi identity is equivalent to the vanishing of the variational Schouten bracket — the extension of the Lie bracket of vector fields to antisymmetric multivector fields — of the bivector determined by $H$: the condition $[H,H]=0$. Two operators forming a Hamiltonian pair (each Hamiltonian, with $[H_1,H_2]=0$) make the system bi-Hamiltonian; by a theorem of Magri (1978), a bi-Hamiltonian system then carries an infinite hierarchy of commuting symmetries and conserved quantities in involution, which is the structural mechanism behind soliton integrability. The Korteweg–de Vries equation (KdV) is the prototype: it is bi-Hamiltonian with respect to the pair $D^3+2uD+u_x$ and $D$, where $D=d/dx$.

In that paper, Gel'fand and Dorfman posed the task of describing all symplectic (equivalently Poisson) structures whose Hamiltonian operator is linear in the field variables, $H_{ij} = \sum_{k,m} u_k a_{ij}^{km} D^m$ with constant coefficients $a_{ij}^{km}$, possibly plus a constant-coefficient operator $K$. They proved (Theorems 1.1–1.5 of the paper) that such an operator is Hamiltonian if and only if its coefficients define a Lie algebra structure: concretely, a graded infinite-dimensional Lie algebra on a space $W$ with basis elements $e_{i\lambda}$ ($\lambda$ real or integer) and commutator $[e_{i\lambda}, e_{j\mu}] = \sum_k \varphi_{ij}^k(\lambda,\mu) e_{k,\lambda+\mu}$ with polynomial structural functions $\varphi_{ij}$. Adding a constant-coefficient operator $K$ preserves Hamiltonicity exactly when $K$ defines an invariant pairing on this Lie algebra that satisfies the 2-cocycle identity (a skew form $\omega$ with $\omega([a,b],c)+\omega([b,c],a)+\omega([c,a],b)=0$). The cocycle is cohomologically trivial precisely when $H+K$ results from $H$ by the constant shift $u_k \mapsto u_k+c_k$, and such trivial cocycles are of interest because this shift corresponds to introducing a spectral parameter — the auxiliary parameter in the Lax representation of an integrable equation; in the KdV example the pencil $D^3+(2u+\mu)D+u_x$ interpolates between the two KdV Hamiltonian structures. The paper's examples recover the classical families: operators from current algebras (the Lie algebra of maps of a line into a finite-dimensional Lie algebra) with invariant metrics, the Gelfand–Dikii operators generating the higher KdV equations (whose Lie-algebraic meaning had been found by Adler and by Lebedev–Manin), the Benney-moments operator (the Hamiltonian structure of Benney's system of moment equations), and commutative rings with Frobenius-type pairings. The abstract machinery of the paper (complexes over Lie algebras and their factor-complexes) was completed in the authors' 1982 sequel, which showed that the classical Yang–Baxter equation of Lie theory — the quadratic consistency equation for a solution matrix (an $R$-matrix) on a Lie algebra, central to the theory of quantum groups — is itself a Hamiltonicity condition in this sense, and in Dorfman's theory of Dirac structures — subspaces of $TM \oplus T^*M$ unifying Poisson and presymplectic geometry — which is now the standard framework.

What the 1981 framework does not cover, and explicitly acknowledges, is everything beyond the affine class: Hamiltonian operators whose coefficients depend nonlinearly on the fields or on their derivatives, at arbitrary differential order. The basic example beyond it is a bracket of hydrodynamic type, a first-order operator with coefficients depending on the fields alone; Dubrovin and Novikov proved that a nondegenerate such bracket is equivalent to a flat pseudo-Riemannian metric with its Levi-Civita connection, a result that originated the theory of Frobenius manifolds — manifolds whose tangent spaces carry a commutative, associative multiplication compatible with a flat metric, the geometric structure underlying two-dimensional topological field theory and quantum cohomology. Higher-order operators with field-dependent leading coefficients (for example third-order brackets) and compatible pairs are governed by far larger systems of conditions, and the algebraic structures extracted from the affine case — Novikov algebras, meaning non-associative algebras defined by the identities that encode first-order Hamiltonicity, and their relatives — are themselves the object of an active classification program. The open question descending from the 1981 paper is therefore the general classification problem: which Hamiltonian operators and Hamiltonian pairs exist in the formal variational calculus, beyond the affine-in-fields class characterized by the Gel'fand–Dorfman theorems.

Problem Statement

The problem is to classify Hamiltonian operators and Hamiltonian pairs in the formal variational calculus (the calculus of local functionals whose densities are polynomials in the fields $u_k(x)$ and finitely many of their $x$-derivatives) beyond the class treated by Gel'fand and Dorfman (1981), that is, beyond operators whose coefficients are affine in the field variables (linear in the fields $u_k$ plus constant-coefficient terms). Concretely: (1) determine normal forms, up to reparametrization of the dependent variables, of skew-symmetric matrix differential operators $H$ of arbitrary differential order, with coefficients that are arbitrary differential polynomials in the fields (polynomials in the $u_k$ and finitely many of their $x$-derivatives), satisfying the condition $[H,H]=0$, where $[\,,\,]$ is the variational Schouten bracket (the extension of the Lie bracket of vector fields to antisymmetric multivector fields; its vanishing on the bivector determined by $H$ is equivalent to the Jacobi identity for the bracket that $H$ induces on functionals); in particular, decide for which classes of operators a discrete or parametric list of normal forms exists at all, and where genuinely varying moduli (continuous families of pairwise inequivalent forms) are unavoidable; (2) classify Hamiltonian pairs $(H_1,H_2)$ — pairs of compatible Hamiltonian operators, that is, operators each satisfying $[H,H]=0$ with $[H_1,H_2]=0$ — beyond the cases covered by the 1981 theorems, in particular beyond pairs in which the operators are affine in the fields or one of them has constant coefficients; (3) clarify, within the general theory, the role of the deformations highlighted in the source: adding a constant-coefficient operator $K$ to a linear-in-the-fields Hamiltonian operator $H$ preserves Hamiltonicity exactly when $K$ defines a 2-cocycle (a skew form $\omega$ on the associated infinite-dimensional Lie algebra satisfying $\omega([a,b],c)+\omega([b,c],a)+\omega([c,a],b)=0$), this cocycle is cohomologically trivial exactly under the constant shifts $u_k \mapsto u_k+c_k$, which correspond to introducing a spectral parameter (an auxiliary parameter in the Lax representation of an integrable equation), and it remains to determine what analogues of this deformation mechanism persist for operators beyond the affine class. Sub-questions (1) and (2) are jointly posed; (3) is a clarifying sub-question about the deformation mechanism that is subordinate to, and does not replace, (1) and (2). A complete answer determines all Poisson brackets of differential type on formal spaces of fields, which is the structural backbone of bi-Hamiltonian integrable hierarchies. The first-order nondegenerate case (equivalent to flat pseudo-Riemannian metrics) and the third-order differential-geometric case are already classified in the literature; the general problem, at arbitrary order and with arbitrary nonlinear field-dependence, is open.

References

1. I. M. Gel'fand and I. Ya. Dorfman, “Hamiltonian operators and infinite-dimensional Lie algebras”, Functional Analysis and Its Applications 15(3):173–187 (1981), DOI 10.1007/BF01089922, https://doi.org/10.1007/BF01089922; Russian original full text at https://www.mathnet.ru/eng/faa1729
2. I. M. Gel'fand and I. Ya. Dorfman, “Hamiltonian operators and the classical Yang—Baxter equation”, Functional Analysis and Its Applications 16(4):241–248 (1982), DOI 10.1007/BF01077846, https://doi.org/10.1007/BF01077846; Russian original full text at https://www.mathnet.ru/eng/faa1663
3. I. M. Gel'fand and I. Ya. Dorfman, “Hamiltonian operators and algebraic structures related to them”, Functional Analysis and Its Applications 13(4):248–262 (1979), DOI 10.1007/BF01078363, https://doi.org/10.1007/BF01078363
4. I. M. Gel'fand and I. Ya. Dorfman, “The Schouten bracket and Hamiltonian operators”, Functional Analysis and Its Applications 14(3):223–226 (1980), DOI 10.1007/BF01086188, https://doi.org/10.1007/BF01086188
5. D. R. Lebedev and Yu. I. Manin, “Gel'fand–Dikii Hamiltonian operator and the coadjoint representation of the Volterra group”, Functional Analysis and Its Applications 13(4):268–273 (1979), DOI 10.1007/BF01078365, https://doi.org/10.1007/BF01078365
6. M. Adler, “On a trace functional for formal pseudo-differential operators and the symplectic structure of the Korteweg–de Vries type equations”, Inventiones Mathematicae 50(3):219–248 (1978), DOI 10.1007/BF01410079, https://doi.org/10.1007/BF01410079
7. F. Magri, “A simple model of the integrable Hamiltonian equation”, Journal of Mathematical Physics 19(5):1156–1162 (1978), DOI 10.1063/1.523777, https://doi.org/10.1063/1.523777
8. I. Ya. Dorfman, “Dirac structures of integrable evolution equations”, Physics Letters A 125(5):240–246 (1987), DOI 10.1016/0375-9601(87)90201-5, https://doi.org/10.1016/0375-9601(87)90201-5
9. I. Dorfman, Dirac Structures and Integrability of Nonlinear Evolution Equations, Nonlinear Science: Theory and Applications, Wiley, New York, 1993, vii+176 pp., ISBN 0-471-93893-9 (bibliographic record as printed in the review by Olver, Bulletin of the American Mathematical Society 31(2):305–308), https://www.ams.org/journals/bull/1994-31-02/S0273-0979-1994-00539-7/
10. P. J. Olver, “Darboux' theorem for Hamiltonian differential operators”, Journal of Differential Equations 71(1):10–33 (1988), DOI 10.1016/0022-0396(88)90036-8, https://doi.org/10.1016/0022-0396(88)90036-8
11. P. J. Olver, “Book review: Dirac structures and integrability of nonlinear evolution equations”, Bulletin of the American Mathematical Society (New Series) 31(2):305–308 (1994), DOI 10.1090/S0273-0979-1994-00539-7, https://doi.org/10.1090/S0273-0979-1994-00539-7
12. B. A. Dubrovin and S. P. Novikov, “Hydrodynamics of weakly deformed soliton lattices. Differential geometry and Hamiltonian theory”, Russian Mathematical Surveys 44(6):35–124 (1989), DOI 10.1070/RM1989V044N06ABEH002300, https://doi.org/10.1070/RM1989V044N06ABEH002300
13. G. V. Potëmin, “On third-order Poisson brackets of differential geometry”, Russian Mathematical Surveys 52(3):617–618 (1997), DOI 10.1070/RM1997v052n03ABEH001817, https://doi.org/10.1070/RM1997v052n03ABEH001817; Math-Net record at https://www.mathnet.ru/eng/rm857
14. A. V. Balandin and G. V. Potemin, “On non-degenerate differential-geometric Poisson brackets of third order”, Russian Mathematical Surveys 56(5):976–977 (2001), DOI 10.1070/RM2001V056N05ABEH000441, https://doi.org/10.1070/RM2001V056N05ABEH000441
15. O. I. Mokhov, “The classification of nonsingular multidimensional Dubrovin–Novikov brackets”, Functional Analysis and Its Applications 42(1):33–44 (2008), DOI 10.1007/s10688-008-0004-8, https://doi.org/10.1007/s10688-008-0004-8
16. M. V. Pavlov, “The description of pairs of compatible first-order differential geometric Poisson brackets”, Theoretical and Mathematical Physics 142(2):244–258 (2005), DOI 10.1007/s11232-005-0063-4, https://doi.org/10.1007/s11232-005-0063-4
17. J. T. Ferguson, “Second-order deformations of hydrodynamic-type Poisson brackets”, Glasgow Mathematical Journal 51(A):75–82 (2009), DOI 10.1017/S0017089508004795, https://doi.org/10.1017/S0017089508004795
18. E. Getzler, “A Darboux theorem for Hamiltonian operators in the formal calculus of variations”, Duke Mathematical Journal 111(3):535–560 (2002), DOI 10.1215/S0012-7094-02-11136-3, arXiv:math/0002164, https://doi.org/10.1215/S0012-7094-02-11136-3
19. P. Vergallo and R. Vitolo, “Projective geometry of homogeneous second-order Hamiltonian operators”, Nonlinearity 36(10):5311–5333 (2023), DOI 10.1088/1361-6544/acf269, https://doi.org/10.1088/1361-6544/acf269
20. M. Gubbiotti, B. van Geemen and P. Vergallo, “Line geometry of pairs of second-order Hamiltonian operators and quasilinear systems”, Proceedings of the Royal Society A 480(2303):20240280 (2024), DOI 10.1098/rspa.2024.0280, https://doi.org/10.1098/rspa.2024.0280
21. L. A. Bokut, Y. Chen and Z. Zhang, “On free Gelfand–Dorfman–Novikov–Poisson algebras and a PBW theorem”, Journal of Algebra 500:153–170 (2018), DOI 10.1016/j.jalgebra.2016.12.006, https://doi.org/10.1016/j.jalgebra.2016.12.006
22. Y. Hong, C. Bai and L. Guo, “A bialgebra theory of Gel'fand–Dorfman algebras with applications to Lie conformal bialgebras”, Canadian Journal of Mathematics, pp. 1–33 (published online 11 July 2025), DOI 10.4153/S0008414X25101181, https://doi.org/10.4153/S0008414X25101181