Higher Equations of Motion (HEM)

• HEM is a family of constructions in mathematical physics where standard second-order equations are extended via higher derivatives or effective eliminations, with applications in gravity, quantum dynamics, and conformal field theories.
• In contexts like higher-derivative gravity and analytical mechanics, HEM replace simple Euler–Lagrange equations with complex structures that can lead to phenomena such as Ostrogradsky instability unless carefully managed.
• In quantum and conformal field theory settings, HEM emerge as effective equations or operator identities that encode hidden variables, symmetry properties, and resonance identities, providing deeper insights into quantum backreaction and conserved currents.

Higher Equations of Motion (HEM) denotes a family of constructions that recur across several areas of mathematical physics but are not identical in meaning. In higher-derivative gravity, the term refers to Euler–Lagrange equations containing derivatives of the fields beyond the standard second order; in canonical quantum theory it refers to effective equations for expectation values containing higher time derivatives after the elimination of quantum moments; and in Liouville-type conformal field theories it refers to operator identities built from null-vector structures of degenerate representations rather than to ordinary Euler–Lagrange equations (Kugo, 2021, Bojowald et al., 2012, Baverez et al., 2023). The common feature is that the dynamical content is organized beyond the simplest local second-order form, either through higher derivatives, effective elimination of auxiliary variables, or representation-theoretic identities.

1. Terminological scope and unifying features

Within the literature represented here, HEM has several established usages. For generally covariant higher-derivative gravity, one starts from an action of the form

$$S[\phi] = \int d^4x\; \mathcal{L}(\phi,\partial\phi,\partial^2\phi,\dots,\partial^N\phi),$$

so that the functional derivatives

$$E_j \equiv \frac{\delta S}{\delta\phi^j} = \sum_{n=0}^{N}(-)^n\partial_{\mu_1}\cdots\partial_{\mu_n}\,\mathcal{L}^{\mu_1\cdots\mu_n}_j$$

are higher equations of motion whenever derivatives of order greater than two occur (Kugo, 2021). In canonical quantum dynamics, HEM means effective equations for expectation values such as $q(t)$ that contain $\dddot q$, $q^{(4)}$, and higher derivatives after moments $G^{a,n}$ have been eliminated in semiclassical and adiabatic expansions (Bojowald et al., 2012). In analytical mechanics, the phrase covers genuine third-order and more generally $(2d-1)$-th order equations of motion derived from degenerate higher-derivative Lagrangians (Motohashi et al., 2014). In Liouville CFT and its supersymmetric extensions, HEM are exact operator identities relating descendants at degenerate momenta to other primary fields (Baverez et al., 2023, Ahn et al., 2010). Specialized usages also occur for higher-spin effective field equations in electromagnetic backgrounds and for higher $\alpha$-order fractal differential equations (Benakli, 10 Dec 2025, Golmankhaneh et al., 2024).

The shared structural motif is that the standard local equation is replaced by a more elaborate object. Sometimes the elaboration is literal differential order, as in higher-derivative mechanics and gravity. Sometimes it is an effective encoding of hidden variables, as in quantum backreaction. In conformal field theory, the “higher” character is representation-theoretic: higher-level null vectors generate operator identities that generalize the ordinary equation of motion. This suggests a useful umbrella description, but the term remains domain-dependent rather than universally standardized.

2. Higher-order classical dynamics and the problem of stability

In ordinary analytical mechanics with

$L=L(q_i,\dot q_i,t),$

the Euler–Lagrange equations are second order. For a Lagrangian depending on higher time derivatives,

$$L=L\bigl(q_i,\dot q_i,\ddot q_i,\dots,q_i^{(n)},t\bigr),$$

the equations of motion can be of order up to $$E_j \equiv \frac{\delta S}{\delta\phi^j} = \sum_{n=0}^{N}(-)^n\partial_{\mu_1}\cdots\partial_{\mu_n}\,\mathcal{L}^{\mu_1\cdots\mu_n}_j$$0. A central result for genuine odd-order dynamics is that the most general Lagrangian producing independent third-order equations for all variables can, up to total derivatives, be reduced to

$$E_j \equiv \frac{\delta S}{\delta\phi^j} = \sum_{n=0}^{N}(-)^n\partial_{\mu_1}\cdots\partial_{\mu_n}\,\mathcal{L}^{\mu_1\cdots\mu_n}_j$$1

with

$$E_j \equiv \frac{\delta S}{\delta\phi^j} = \sum_{n=0}^{N}(-)^n\partial_{\mu_1}\cdots\partial_{\mu_n}\,\mathcal{L}^{\mu_1\cdots\mu_n}_j$$2

The corresponding equations are genuinely third order, and the same pattern extends to general odd order $$E_j \equiv \frac{\delta S}{\delta\phi^j} = \sum_{n=0}^{N}(-)^n\partial_{\mu_1}\cdots\partial_{\mu_n}\,\mathcal{L}^{\mu_1\cdots\mu_n}_j$$3 (Motohashi et al., 2014).

The Hamiltonian analysis proceeds by introducing auxiliary variables and Lagrange multipliers, converting the system into a first-order constrained theory. For the third-order case there are $$E_j \equiv \frac{\delta S}{\delta\phi^j} = \sum_{n=0}^{N}(-)^n\partial_{\mu_1}\cdots\partial_{\mu_n}\,\mathcal{L}^{\mu_1\cdots\mu_n}_j$$4 primary constraints, all of them second class, leaving $$E_j \equiv \frac{\delta S}{\delta\phi^j} = \sum_{n=0}^{N}(-)^n\partial_{\mu_1}\cdots\partial_{\mu_n}\,\mathcal{L}^{\mu_1\cdots\mu_n}_j$$5 independent initial conditions, in agreement with the third-order character of the equations. The canonical Hamiltonian reduces on the constraint surface to

$$E_j \equiv \frac{\delta S}{\delta\phi^j} = \sum_{n=0}^{N}(-)^n\partial_{\mu_1}\cdots\partial_{\mu_n}\,\mathcal{L}^{\mu_1\cdots\mu_n}_j$$6

and is linear in the unconstrained canonical momenta $$E_j \equiv \frac{\delta S}{\delta\phi^j} = \sum_{n=0}^{N}(-)^n\partial_{\mu_1}\cdots\partial_{\mu_n}\,\mathcal{L}^{\mu_1\cdots\mu_n}_j$$7. Because these momenta are not removed by the constraints, the Hamiltonian is unbounded from below and above. The result is an Ostrogradsky-type instability, and the paper shows that this persists for general odd order $$E_j \equiv \frac{\delta S}{\delta\phi^j} = \sum_{n=0}^{N}(-)^n\partial_{\mu_1}\cdots\partial_{\mu_n}\,\mathcal{L}^{\mu_1\cdots\mu_n}_j$$8 (Motohashi et al., 2014).

An important caveat is that degeneracy by itself is not sufficient to imply instability. The example

$$E_j \equiv \frac{\delta S}{\delta\phi^j} = \sum_{n=0}^{N}(-)^n\partial_{\mu_1}\cdots\partial_{\mu_n}\,\mathcal{L}^{\mu_1\cdots\mu_n}_j$$9

is degenerate in the usual higher-derivative sense but yields only second-order equations of motion. The precise no-go statement is narrower: any degenerate Lagrangian that gives genuinely third-order equations of motion for all variables suffers from the Ostrogradsky instability. This distinction is central in later literature that separates harmless degeneracy from genuinely propagating higher-order sectors.

3. Effective quantum higher equations of motion

In canonically quantized systems, HEM arise not from a fundamental higher-derivative Lagrangian but from the elimination of quantum moments. For the anharmonic oscillator

$q(t)$0

the expectation values

$q(t)$1

do not evolve autonomously, because

$q(t)$2

depends on the moments

$q(t)$3

The paper treats $q(t)$4 as coordinates on an infinite-dimensional quantum phase space generated by the quantum Hamiltonian $q(t)$5 (Bojowald et al., 2012).

To obtain closed effective dynamics for $q(t)$6, the moments are solved approximately by combining a semiclassical expansion

$q(t)$7

with an adiabatic expansion implemented by $q(t)$8 and

$q(t)$9

At fixed adiabatic order, the moment equations become algebraic, and each $\dddot q$0 becomes an algebraic function of $\dddot q$1. After substitution back into the equation for $\dddot q$2, the effective equation acquires higher time derivatives. By fourth adiabatic order, terms up to $\dddot q$3 appear, and at the truncation studied in the paper the effective equation is a fourth-order ODE with corrections kept up to $\dddot q$4 (Bojowald et al., 2012).

The physical interpretation is that higher-derivative terms encode quantum backreaction and nonlocality in time. The full canonical system is still first order in time on the enlarged phase space of expectation values plus moments, so no new fundamental degrees of freedom are introduced. The higher-order equation for $\dddot q$5 is an effective encoding of those moment degrees of freedom. The paper therefore rejects a naive Ostrogradsky reading of these effective equations: the extra higher-derivative solutions that are nonanalytic in $\dddot q$6 are to be discarded, and only solutions consistent with the semiclassical and adiabatic expansions and with the uncertainty relations are admissible (Bojowald et al., 2012).

This framework is used to argue that quantum-gravity and cosmological effective equations should generically contain higher time derivatives. In particular, the paper states that higher-curvature terms in effective actions correspond canonically to higher-time-derivative corrections, and that effective loop quantum cosmology equations containing only modified functions of the Hubble parameter but no higher derivatives cannot be a faithful higher-curvature expansion unless special cancellations occur.

4. Generally covariant higher-derivative gravity and Maxwell-type reformulations

For generally covariant systems with arbitrarily high derivative fields, the higher equations of motion are the Euler–Lagrange expressions

$\dddot q$7

derived from a Lagrangian depending on $\dddot q$8. In metric theories this includes Einstein–Hilbert type terms, quadratic curvature terms such as $\dddot q$9 and $q^{(4)}$0, and higher covariant-derivative terms. When $q^{(4)}$1, the equations involve derivatives beyond second order and are therefore HEM in the literal higher-derivative sense (Kugo, 2021).

The central result is that general coordinate invariance imposes a tower of Noether identities, the $q^{(4)}$2-identities, which generalize the usual Bianchi identities. Using generalized both-side derivatives and a hierarchy of tensors $q^{(4)}$3, the paper shows that a suitable linear combination of the HEM can be rewritten in Maxwell-like divergence form. For the translation Noether current $q^{(4)}$4, one obtains

$q^{(4)}$5

where $q^{(4)}$6 is antisymmetric in $q^{(4)}$7. Thus a first-order divergence equation emerges from a particular linear combination of the higher-order field equations, with the energy–momentum tensor playing the role of the source (Kugo, 2021).

In the gauge-fixed quantum theory, formulated in a GL(4)-invariant de Donder–Nakanishi gauge, the same structure persists with an additional term involving the Nakanishi–Lautrup field $q^{(4)}$8. That term is shown to be BRS exact modulo a total derivative, and the Maxwell-type gravity equation can be written in a form exactly parallel to Yang–Mills theory. The same mechanism extends beyond translation symmetry to Nakanishi’s IOSp(8|8) choral symmetry. With

$q^{(4)}$9

the global IOSp(8|8) currents take the simple form

$G^{a,n}$0

and the paper derives corresponding Maxwell-type equations for suitable linear combinations of the HEM. In this formulation the BRS current appears as a special case of a larger symmetry current, and the higher equations are reorganized into divergence equations tied directly to Noether currents. The paper further states that this structure underlies proof strategies for existence theorems of massless spin-2 modes in general GC-invariant higher-derivative theories.

5. Liouville conformal field theory, resonance identities, and boundary HEM

In Liouville CFT, HEM are operator identities associated with degenerate Virasoro representations. They are not the same as the usual BPZ null-vector equations. For degenerate parameters $G^{a,n}$1, one has singular vector operators $G^{a,n}$2 and $G^{a,n}$3, and Zamolodchikov’s higher equations of motion take the schematic form

$G^{a,n}$4

where the prime denotes derivative with respect to the Liouville momentum. At level 2, the rigorous probabilistic construction on full Fock space shows that simple poles of the Poisson operator at $G^{a,n}$5 and $G^{a,n}$6 produce non-vanishing derivatives of top singular vectors, which are identified with primary fields of the same conformal weight. The paper derives explicit bulk identities for $G^{a,n}$7 and $G^{a,n}$8, and for the latter finds a “freezing” phenomenon: the coefficient vanishes for $G^{a,n}$9 (Baverez et al., 2023).

A complementary interpretation identifies Liouville HEM as resonance identities in conformal perturbation theory. In that formulation, Liouville theory is treated as a perturbation of a massless free boson, and the descendant operators

$(2d-1)$0

develop resonances at the degenerate values $(2d-1)$1. The residue reproduces the HEM,

$(2d-1)$2

so the higher equations of motion are reinterpreted as perturbatively exact operator resonance identities. The same paper shows that in sinh-Gordon the Liouville identities deform by extra resonance terms for odd $(2d-1)$3, while in sine-Gordon the resonance structure is more intricate because the target exponential fields may themselves be resonant (Lashkevich, 2011).

For boundary Liouville CFT, level-2 HEM were derived directly from Ward identities on the upper half-plane. The paper introduces derivatives with respect to boundary insertions and defines boundary descendants adapted to the probabilistic regularization. For the degenerate boundary fields with

$(2d-1)$4

the resulting higher equation of motion has the form

$(2d-1)$5

For $(2d-1)$6, the right-hand side is

$(2d-1)$7

while for $(2d-1)$8 and $(2d-1)$9 it is proportional to

$\alpha$0

times $\alpha$1; for $\alpha$2 it vanishes. When these coefficients vanish, the HEM reduce to the BPZ differential equations. The quadratic condition in $\alpha$3 is exactly the FZZ conic section in the boundary cosmological constants, and the paper presents the BPZ equations as a corollary of the higher equations of motion (Cerclé, 2024).

6. Supersymmetric, higher-spin, and fractal extensions

In $\alpha$4 superconformal Liouville field theory, the phrase HEM again denotes operator identities, now associated with degenerate representations of the $\alpha$5 superconformal algebra. The theory has three classes of degenerate fields; the HEM are in one-to-one correspondence with them and are enumerated, in addition to the $\alpha$6 charge $\alpha$7, by either the positive integer $\alpha$8 or the pair $\alpha$9. For class I one has identities of the form

$L=L(q_i,\dot q_i,t),$0

while class II gives charge-shifting relations such as

$L=L(q_i,\dot q_i,t),$1

The coefficients are determined from one-point functions on the Poincaré disk, and the paper checks that in the classical limit these HEM become relations among classical fields, with the lowest members reproducing the ordinary $L=L(q_i,\dot q_i,t),$2 Liouville equations of motion (Ahn et al., 2010).

A different extension appears in effective field theory for massive higher spins in constant electromagnetic backgrounds. There the “higher equations of motion” are equations for spin-2 and spin-$L=L(q_i,\dot q_i,t),$3 fields that preserve tracelessness, $L=L(q_i,\dot q_i,t),$4-trace, and transversality constraints while maintaining the correct number of propagating degrees of freedom. The point-like consistent systems require $L=L(q_i,\dot q_i,t),$5, but the EFT construction introduces specific operators proportional to $L=L(q_i,\dot q_i,t),$6 so that algebraic consistency can be restored perturbatively for arbitrary gyromagnetic ratio $L=L(q_i,\dot q_i,t),$7. For spin 2 this leads to induced non-transversality $L=L(q_i,\dot q_i,t),$8 fixed by the equations rather than by new initial data; for spin $L=L(q_i,\dot q_i,t),$9, a projector-based deformation with $$L=L\bigl(q_i,\dot q_i,\ddot q_i,\dots,q_i^{(n)},t\bigr),$$0 yields a consistent projected equation with generic $$L=L\bigl(q_i,\dot q_i,\ddot q_i,\dots,q_i^{(n)},t\bigr),$$1 and hyperbolic principal part in the EFT regime (Benakli, 10 Dec 2025).

Fractal calculus supplies yet another specialized meaning. In that framework, the fundamental derivative is the local fractal derivative

$$L=L\bigl(q_i,\dot q_i,\ddot q_i,\dots,q_i^{(n)},t\bigr),$$2

defined with respect to the staircase function $$L=L\bigl(q_i,\dot q_i,\ddot q_i,\dots,q_i^{(n)},t\bigr),$$3 of a fractal set $$L=L\bigl(q_i,\dot q_i,\ddot q_i,\dots,q_i^{(n)},t\bigr),$$4. Higher $$L=L\bigl(q_i,\dot q_i,\ddot q_i,\dots,q_i^{(n)},t\bigr),$$5-order equations are obtained by iteration:

$$L=L\bigl(q_i,\dot q_i,\ddot q_i,\dots,q_i^{(n)},t\bigr),$$6

The paper studies linear equations

$$L=L\bigl(q_i,\dot q_i,\ddot q_i,\dots,q_i^{(n)},t\bigr),$$7

states that the solution space has dimensionality $$L=L\bigl(q_i,\dot q_i,\ddot q_i,\dots,q_i^{(n)},t\bigr),$$8, and develops constant-coefficient solution theory, Wronskians, variation of parameters, and undetermined coefficients. Its mechanical example is a two-mass spring system in fractal time,

$$L=L\bigl(q_i,\dot q_i,\ddot q_i,\dots,q_i^{(n)},t\bigr),$$9

which reduces to the fourth-order fractal equation

$$E_j \equiv \frac{\delta S}{\delta\phi^j} = \sum_{n=0}^{N}(-)^n\partial_{\mu_1}\cdots\partial_{\mu_n}\,\mathcal{L}^{\mu_1\cdots\mu_n}_j$$00

Here HEM are literally higher-order equations of motion, but with differentiation taken with respect to fractal time rather than ordinary time (Golmankhaneh et al., 2024).

Taken together, these usages show that HEM is not a single doctrine but a recurrent structural idea. Depending on context, it may denote higher-derivative Euler–Lagrange equations, effective higher-time-derivative reductions, null-vector-based operator identities, or generalized dynamics on nonstandard geometric backgrounds. The literature consistently treats these equations as symmetry-sensitive objects: sometimes they expose hidden conservation laws, sometimes they encode eliminated degrees of freedom, and sometimes they identify descendants with other primaries.