---
title: Hamilton's Identity in Mathematical Contexts
url: https://www.emergentmind.com/topics/hamilton-s-identity
type: topic
---

# Hamilton's Identity in Mathematical Contexts

Searching arXiv for recent and relevant papers on the various mathematical usages of “Hamilton’s identity”.
In the literature, “Hamilton’s identity” is not a uniquely fixed term. Within the sources considered here it denotes, or is treated as denoting, several distinct constructions: the Cayley–Hamilton identity for endomorphisms, a Hamilton-type first integral for gradient solitons, the Jacobi identity of a Poisson bracket, an operator identity for the harmonic-oscillator Hamiltonian governed by Euler polynomials, and Hamilton’s composition law for turns in \(\mathrm{SU}(2)\) [2601.06136] [2507.12381] [1605.02471] [1508.04844] [1110.3127]. The common thread is structural constraint: a matrix satisfies an invariant polynomial, a soliton potential satisfies a conserved scalar relation, a Poisson bracket satisfies Jacobi, nested anti-commutators collapse to a special-function expression, or group multiplication is encoded by spherical arcs.

## 1. Terminological scope

The sources make the ambiguity explicit. One paper states that, if one asks about “Hamilton’s identity,” the relevant identity is precisely the Cayley–Hamilton identity for a second-order tensor. Another states that the relevant identity should indeed be understood as the Jacobi identity for the Poisson bracket. A third uses “Hamilton’s identity” as the name of a conserved quantity for complete gradient \(q\)-solitons. Two further papers discuss identities involving the Hamiltonian operator or Hamilton’s turns without presenting a universally fixed standalone theorem under that exact title [2601.06136] [1605.02471] [2507.12381] [1508.04844] [1110.3127].

| Context | Identity | Source |
|---|---|---|
| Matrix theory | \(A^m-\sigma_1A^{m-1}+\sigma_2A^{m-2}-\cdots+(-1)^m\sigma_m I=0\) | [2601.06136] |
| Poisson dynamics | \(\{F_1,\{F_2,F_3\}\}+\{F_2,\{F_3,F_1\}\}+\{F_3,\{F_1,F_2\}\}=0\) | [1605.02471] |
| Gradient \(q\)-solitons | \(|\nabla f|^2-\frac12\operatorname{tr}(q)-2\lambda f=C\) | [2507.12381] |
| Harmonic-oscillator operator algebra | \(\frac{1}{2^n}\{q,H\}_n=\frac12\{q,E_n(H+\tfrac12)\}\) | [1508.04844] |
| Hamilton turns | \(T(\hat{\mathbf n}_2,\hat{\mathbf n}_3)\,T(\hat{\mathbf n}_1,\hat{\mathbf n}_2)=T(\hat{\mathbf n}_1,\hat{\mathbf n}_3)\) | [1110.3127] |

This suggests that the expression is contextual rather than canonical. Its meaning is determined by the ambient subject: invariant theory, Poisson geometry, geometric analysis, operator algebra, or geometric representations of \(\mathrm{SU}(2)\).

## 2. Cayley–Hamilton identity in matrix theory

In the linear-algebraic usage, Hamilton’s identity is the Cayley–Hamilton theorem. For \(A\in \mathrm{End}(V)\), with \(V\) \(m\)-dimensional,
\[
p_A(\lambda)=\det(\lambda I-A), \qquad p_A(A)=0.
\]
If
\[
p_A(\lambda)=\lambda^m+c_1\lambda^{m-1}+\cdots+c_m,
\]
then
\[
A^m+c_1A^{m-1}+\cdots+c_m I=0.
\]
The invariant form preferred in the tensorial treatment writes the coefficients as principal invariants \(\sigma_k\), the elementary symmetric polynomials of the eigenvalues:
\[
A^m-\sigma_1A^{m-1}+\sigma_2A^{m-2}-\cdots+(-1)^m\sigma_m I=0,
\]
with \(c_k=(-1)^k\sigma_k\) [2601.06136].

The same paper encodes the \(\sigma_k\) by generalized-delta contractions,
\[
\sigma_k=\frac{1}{k!}\,\delta^{j_1\cdots j_k}_{i_1\cdots i_k} A^{i_1}_{j_1}\cdots A^{i_k}_{j_k}, \qquad k=1,\dots,m,
\]
and records the familiar low-order cases
\[
\sigma_1=\operatorname{tr}(A),\qquad
\sigma_2=\frac12\big[\operatorname{tr}^2(A)-\operatorname{tr}(A^2)\big],\qquad
\sigma_m=\det(A).
\]
In this formulation the coefficients are not ad hoc scalars appended to a matrix polynomial; they are isotropic invariants generated by contraction.

This formulation is standard in substance but unusual in emphasis. Rather than deriving the theorem from the characteristic polynomial as an auxiliary scalar object, the tensorial approach regards the matrix identity itself as primary and its coefficients as inherited from invariant contraction structure.

## 3. Dimensional syzygy and the geometric origin of Cayley–Hamilton

The geometric reconstruction of the Cayley–Hamilton identity starts from a dimension-dependent antisymmetry constraint rather than from \(p_A(\lambda)\). The fundamental fact is the impossibility of antisymmetrizing \(m+1\) indices in \(m\)-dimensional space:
\[
\delta^{j_1\cdots j_{m+1}}_{i_1\cdots i_{m+1}}=0
\qquad (\text{in }m\text{-dimensional space}),
\]
equivalently, in the paper’s shorthand, \(\varepsilon\otimes\varepsilon=0\). The authors call this vanishing generalized-delta relation the fundamental syzygy [2601.06136].

The key construction is to let the vanishing isotropic operator with two free indices act on \(A^{\otimes m}\):
\[
\delta^{p\, j_1\cdots j_m}_{q\, i_1\cdots i_m}A^{i_1}_{j_1}\cdots A^{i_m}_{j_m}.
\]
Because the generalized delta is identically zero in dimension \(m\), the whole expression vanishes. Its expansion, however, automatically produces a second-order tensor whose terms have the form “scalar invariant \(\times\) tensor power of \(A\).” Closed index loops contribute scalar isotropic invariants; the remaining open chain transports the free indices and contributes a power of \(A\). The appendix summarizes the result as
\[
\delta^{p\, j_1 \cdots j_m}_{q\, i_1 \cdots i_m} A^{i_1}_{j_1} \cdots A^{i_m}_{j_m}
= m! \sum_{k=0}^{m} (-1)^{m-k} \sigma_k (A^{m-k})^p_q,
\]
with \(\sigma_0:=1\). Since the left-hand side vanishes,
\[
\sum_{k=0}^{m}(-1)^k\sigma_kA^{m-k}=0.
\]

The \(m=2\) case is verified explicitly. Starting from
\[
\delta^{mbc}_{ijn}A_{ib}A_{jc}=0,
\]
termwise contraction yields
\[
2(A^2)_{mn}-2\operatorname{tr}(A)A_{mn}
+\big[\operatorname{tr}^2(A)-\operatorname{tr}(A^2)\big]\delta_{mn}=0.
\]
After division by \(2\) and the \(2\)-dimensional identity
\[
\det(A)=\frac12\big[\operatorname{tr}^2(A)-\operatorname{tr}(A^2)\big],
\]
one obtains
\[
A^2-\operatorname{tr}(A)A+\det(A)I=0.
\]

The dimension-independent proof combines Laplace expansion of the generalized Kronecker delta with the Newton–Girard identities
\[
k\sigma_k=\sum_{i=1}^{k}(-1)^{i-1}\sigma_{k-i}p_i, \qquad p_i=\operatorname{tr}(A^i).
\]
The conceptual claim is that the theorem’s degree, sign pattern, and number of terms are reflections of the ambient dimension, because they arise from the vanishing of an \((m+1)\)-fold antisymmetrization. The paper further suggests, but does not fully develop, that the same syzygy-based framework may organize invariants of higher-order tensors such as third-order piezoelectric tensors and fourth-order elastic tensors, precisely where ordinary characteristic-polynomial methods fail.

## 4. Noncommutative and quantum extensions

Several papers generalize the Cayley–Hamilton identity beyond commutative matrix algebras. For matrices over a ring \(R\) satisfying
\[
[x,y][u,v]=0,
\]
one obtains an invariant two-sided Cayley–Hamilton identity
\[
\sum_{0\le i,j\le n} A^i\,c_{i,j}\,A^j=0,
\]
with \(c_{i,j}=\lambda_i\lambda_j\) coming from the symmetric characteristic polynomial
\[
p(x)=\lambda_0+\lambda_1x+\cdots+\lambda_nx^n,\qquad \lambda_n=n!,
\]
and therefore
\[
c_{n,n}=(n!)^2.
\]
The mechanism is that left and right Cayley–Hamilton identities contain correction matrices \(C_i,D_i\in M_n([R,R])\); the ring identity forces every product \(C_iD_j\) to vanish, leaving a clean sandwich-coefficient identity [1106.3223].

In half-quantum linear algebra the theorem is absorbed into a broader Cayley–Hamilton–Newton framework. A half-quantum matrix \(M\) is defined, relative to a compatible pair \((R,\mathcal F)\), by
\[
S^{(2)}_{12}\,M_{\overline1}M_{\overline2}\,A^{(2)}_{12}=0.
\]
The resulting matrix identities use modified powers and quantum symmetric functions rather than ordinary powers and elementary symmetric polynomials. When the Hecke \(R\)-matrix is even of height \(n\), the half-quantum Cayley–Hamilton identity takes the form
\[
\sum_{k=0}^{n-1} (-1)^{\,n-k}\,\widetilde M^{\leftarrow\,n-k}\,\sigma_k(M)
+n_q\,\det\nolimits_q(M)\,{\cal D}=0.
\]
Here the constant term is not \((\det M)I\) but \(n_q\det_q(M){\cal D}\), and ordinary powers are replaced by braided powers such as \(\widetilde M^{\leftarrow k}\) [1303.3991].

A different extension studies quasi-identities on \(M_n(F)\), where the coefficients are arbitrary polynomial functions of the matrix entries rather than trace polynomials. In that setting the Cayley–Hamilton polynomial remains fundamental but is not universally generating. Positive results include: every one-variable quasi-identity is divisible by the one-variable Cayley–Hamilton polynomial \(q_n\); every multilinear quasi-identity of degree \(n\) is a scalar multiple of the polarized Cayley–Hamilton polynomial \(Q_n\); and for every quasi-identity \(P\) and every central polynomial \(c\) with zero constant term there exists \(m\in\mathbb N\) such that \(c^mP\in (Q_n)\). Negative results are equally explicit: there exist antisymmetric quasi-identities of degree \(n^2\) that do not follow from \(Q_n\), and the paper gives a complete description of a distinguished family of such counterexamples [1212.4597].

Taken together, these results preserve the central status of the Cayley–Hamilton identity while sharply enlarging its algebraic habitat. They also show that the classical monic scalar polynomial relation is only one member of a broader family of invariant, braided, or two-sided characteristic identities.

## 5. Jacobi identity as the structural condition of Hamiltonian dynamics

In Poisson dynamics, the identity treated as “Hamilton’s identity” is the Jacobi identity. The relevant class of brackets on \(\mathbb R^3\) is
\[
\{F,G\}=\boldsymbol{\alpha}\cdot \nabla F\times \nabla G,
\]
with \(\boldsymbol{\alpha}=\boldsymbol{\alpha}(x,y,z)\). This bracket is always bilinear, antisymmetric, and Leibniz. Antisymmetry alone implies energy conservation,
\[
\dot H=\{H,H\}=0.
\]
What distinguishes genuinely Hamiltonian dynamics is the Jacobi identity
\[
\{F_1,\{F_2,F_3\}\}+\{F_2,\{F_3,F_1\}\}+\{F_3,\{F_1,F_2\}\}=0,
\]
which in this \(\mathbb R^3\) setting is equivalent to
\[
\boldsymbol{\alpha}\cdot\nabla\times\boldsymbol{\alpha}=0
\]
[1605.02471].

The induced velocity field is
\[
\dot{\boldsymbol x}=\nabla H\times \boldsymbol{\alpha}.
\]
A central structural fact is the relation to Casimir invariants. A Casimir \(S\) satisfies \(\{S,F\}=0\) for all \(F\), which here becomes
\[
\boldsymbol{\alpha}\times \nabla S=0.
\]
Hence locally \(\boldsymbol{\alpha}=\lambda\nabla S\), and this representation implies the Jacobi condition. Conversely, any \(\boldsymbol{\alpha}\) satisfying \(\boldsymbol{\alpha}\cdot\nabla\times\boldsymbol{\alpha}=0\) can locally be written in that form. For these flows, Jacobi is therefore locally equivalent to the existence of a Casimir. The phase space is foliated by the invariant leaves \(S=\mathrm{const}\), and trajectories lie on intersections of \(H=\mathrm{const}\) and \(S=\mathrm{const}\).

The paper contrasts three regimes. In the Hamiltonian case, Jacobi holds, a Casimir exists, and the foliation blocks cross-leaf transport. In the almost-Poisson case, antisymmetry and energy conservation remain, but Jacobi fails:
\[
\boldsymbol{\alpha}\cdot\nabla\times\boldsymbol{\alpha}\neq 0.
\]
Then, in general, there is no Casimir and no invariant foliation; the dynamics may show attractors and what the paper calls fake dissipation. In the metriplectic case, one starts from a genuine Poisson bracket and adds a symmetric bracket,
\[
(F,G)=\boldsymbol{\alpha}\cdot \nabla F\times \nabla G
+\mu (\nabla F\times \nabla H)\cdot (\nabla H\times \nabla G),
\]
with \(\mu>0\). The resulting dynamics conserves \(H\) but produces monotone growth of an entropy-like Casimir \(S\):
\[
\frac{dS}{dt}=\mu \|\nabla S\times \nabla H\|^2\ge 0.
\]

The worked example takes
\[
H(x,y,z)=\cos(x+y+z), \qquad
\boldsymbol{\alpha}=(\cos x,\cos y,\sin z+\varepsilon \cos y).
\]
When \(\varepsilon=0\), the system is Hamiltonian and has Casimir
\[
S(x,y,z)=\sin x+\sin y-\cos z.
\]
When \(\varepsilon\neq 0\),
\[
\boldsymbol{\alpha}\cdot\nabla\times \boldsymbol{\alpha}=-\varepsilon \cos x\sin y,
\]
so Jacobi fails. The paper’s conclusion is that the Jacobi identity is fundamentally dynamical and geometric, not merely technical: it is the condition that makes an energy-preserving system genuinely Hamiltonian.

## 6. Hamilton’s identity for gradient solitons

In geometric analysis, “Hamilton’s identity” is used explicitly for a scalar conservation law on complete gradient \(q\)-solitons. The ambient flow is
\[
\frac{\partial}{\partial t}g=q,
\]
where \(q\) is a symmetric \((0,2)\)-tensor. A gradient \(q\)-soliton satisfies
\[
{\rm Hess}\,f=\lambda g+\frac12 q.
\]
If \(Q\) is the \((1,1)\)-tensor dual to \(q\), the paper singles out the tensor condition
\[
Q(\nabla f)=\frac12\nabla \operatorname{tr}(q),
\]
and proves that this is equivalent to the pointwise identity
\[
|\nabla f|^2-\frac12\operatorname{tr}(q)-2\lambda f=C,
\]
where \(C\) is constant. This equivalence is stated as Proposition 3.1 and is the paper’s Hamilton identity [2507.12381].

In the Ricci-soliton specialization \(q=-2\,{\rm Ric}\), one has \(\operatorname{tr}(q)=-2R\), so the identity becomes
\[
R+|\nabla f|^2-2\lambda f=C.
\]
The tensor condition correspondingly reduces to
\[
{\rm Ric}(\nabla f,\cdot)=\frac12\,dR(\cdot),
\]
which the paper identifies as a consequence of the contracted second Bianchi identity together with the Ricci soliton equation. In this sense the general \(q\)-soliton formula is a direct abstraction of Hamilton’s classical identity for gradient Ricci solitons.

The identity is used as a structural input for several downstream results. The paper derives trace-Laplacian formulas, including
\[
\Delta \operatorname{tr}(q)
=2\lambda \operatorname{tr}(q)+|q|^2+2(\mathrm{div}(q))(\nabla f),
\]
and
\[
\Delta_f \operatorname{tr}(q)
=2\lambda\operatorname{tr}(q)+|q|^2+\langle 2\mathrm{div}(Q)-\nabla \operatorname{tr}(q),\nabla f\rangle.
\]
It then uses Hamilton’s identity in compactness, rigidity, growth, volume, and maximum-principle arguments. Among the stated consequences are: if \((M,g,f)\) is a compact gradient \(q\)-soliton with \(\lambda\le 0\) satisfying the Hamilton condition, then \(f\) is constant; under explicit additional assumptions one obtains \(q\)-flatness; and under boundedness assumptions on
\[
F=\frac12|\nabla f|^2-\lambda f
\]
one obtains quadratic growth control for \(f\), polynomial upper volume growth, and the Omori–Yau maximum principle. The paper’s stated thesis is that many Ricci-soliton arguments depend primarily on this Hamilton-type first integral rather than on the specific tensor \({\rm Ric}\).

## 7. Operator identities and Hamilton turns

A different operator-theoretic usage appears for the canonical pair \((p,q)\) satisfying
\[
[q,p]=i,
\qquad
H=\frac12(p^2+q^2).
\]
The identity inferred by Bender and Bettencourt and proved in several ways by De Angelis and Vignat is
\[
\frac{1}{2^n}\{q,H\}_n=\frac12\{q,E_n(H+\tfrac12)\},
\]
where \(\{q,H\}_n\) is the \(n\)-fold nested anti-commutator and \(E_n\) is the Euler polynomial defined by
\[
\sum_{n\ge 0}\frac{E_n(x)}{n!}z^n=\frac{2e^{zx}}{e^z+1}.
\]
Equivalent forms include
\[
\{q,H-\tfrac12\}_n+\{q,H+\tfrac12\}_n=2\{q,H^n\},
\]
and the symbolic formulation
\[
\frac{1}{2^n}(\{q,H\}-1)^n+\frac{1}{2^n}(\{q,H\}+1)^n=\{q,H^n\}.
\]
The significance of the identity is that a hierarchy of nested anti-commutators collapses to a single Euler-polynomial expression in the Hamiltonian [1508.04844].

A geometric usage, associated with Hamilton’s quaternionic construction of turns, represents elements of \(\mathrm{SU}(2)\) by directed great-circle arcs on \(S^2\). For unit vectors \(\hat{\mathbf n}_1,\hat{\mathbf n}_2\),
\[
T(\hat{\mathbf n}_1,\hat{\mathbf n}_2)
=\hat{\mathbf n}_1\cdot \hat{\mathbf n}_2
-i(\hat{\mathbf n}_1\wedge \hat{\mathbf n}_2)\cdot\boldsymbol{\tau}.
\]
The central composition law is
\[
T(\hat{\mathbf n}_2,\hat{\mathbf n}_3)\,T(\hat{\mathbf n}_1,\hat{\mathbf n}_2)
= T(\hat{\mathbf n}_1,\hat{\mathbf n}_3),
\]
and more generally
\[
T(\hat{\mathbf n}_3,\hat{\mathbf n}_4)\,
T(\hat{\mathbf n}_2,\hat{\mathbf n}_3)\,
T(\hat{\mathbf n}_1,\hat{\mathbf n}_2)
= T(\hat{\mathbf n}_1,\hat{\mathbf n}_4).
\]
The paper does not introduce a separately labeled theorem called “Hamilton’s identity,” but it treats this composition law as the defining Hamiltonian construct: multiplication in \(\mathrm{SU}(2)\) becomes head-to-tail addition of directed spherical arcs, a non-Abelian analogue of the parallelogram law [1110.3127].

These operator and geometric usages widen the semantic range of the term. They also clarify a persistent pattern: whether the object is a matrix, a Hamiltonian, or a unitary gate, the “identity” attributed to Hamilton typically expresses a hidden closure law—polynomial, algebraic, dynamical, or geometric—that organizes an otherwise more complicated structure.

Source: https://www.emergentmind.com/topics/hamilton-s-identity