---
title: Armould Calculus for Lie Normal Forms
url: https://www.emergentmind.com/topics/armould-calculus
type: topic
---

# Armould Calculus for Lie Normal Forms

Armould calculus, in the sense suggested by the Lie-algebraic presentation of Écalle’s mould formalism in "Normalization in Lie algebras via mould calculus and applications" [1604.07142], denotes the use of moulds—families of scalar coefficients indexed by words—to encode and solve normalization problems in complete filtered Lie algebras. Its defining feature is the reduction of iterated commutators, homological equations, and resonance constraints to an explicit mould equation whose solutions are universal: they depend only on the combinatorics of words and the eigenvalue map of the unperturbed operator, not on the specific dynamical realization. In this framework, normal forms for vector fields, Hamiltonians, averaging problems, and quantum operators are obtained by coupling universal mould coefficients with problem-specific Lie comoulds.

## 1. Abstract normalization problem

The basic setting is a field \(k\) of characteristic zero and a complete filtered Lie algebra \((\mathcal L,[\,,\,])\) endowed with a decreasing filtration
\[
\mathcal L=\mathcal L_{\ge 0}\supset \mathcal L_{\ge 1}\supset \mathcal L_{\ge 2}\supset \cdots,
\qquad
[\mathcal L_{\ge m},\mathcal L_{\ge n}]\subset \mathcal L_{\ge m+n},
\]
with \(\bigcap_m \mathcal L_{\ge m}=\{0\}\). The associated order function \(\operatorname{ord}(X)\) is the largest \(m\) such that \(X\in \mathcal L_{\ge m}\), and completeness is taken with respect to the metric
\[
d(X,Y):=2^{-\operatorname{ord}(X-Y)}.
\]
This makes infinite sums and exponentials such as \(e^{\mathrm{ad}_Y}\) meaningful [1604.07142].

The normalization problem starts from an element \(X_0\in\mathcal L\) and a perturbation \(B\in\mathcal L_{\ge 1}\) decomposed along eigenvectors of \(\mathrm{ad}_{X_0}\),
\[
B=\sum_{n\in\mathcal N} B_n,
\qquad
[X_0,B_n]=\lambda(n)B_n,
\]
where \(\lambda:\mathcal N\to k\) is the eigenvalue map and \((B_n)_{n\in\mathcal N}\) is formally summable. One seeks \(Y,Z\in\mathcal L_{\ge 1}\) such that
\[
e^{\mathrm{ad}_Y}(X_0+B)=X_0+Z,
\qquad
[X_0,Z]=0.
\]
Then \(\Phi=e^{\mathrm{ad}_Y}\) is a Lie algebra automorphism conjugating \(X_0+B\) to a normal form commuting with \(X_0\).

The central Lie-algebraic result is the existence of universal scalar families
\[
F^{\lambda_1,\dots,\lambda_r},
\qquad
G^{\lambda_1,\dots,\lambda_r}\in k,
\]
such that
\[
Z=\sum_{r\ge 1}\sum_{n_1,\dots,n_r\in\mathcal N}
\frac{1}{r}
F^{\lambda(n_1),\dots,\lambda(n_r)}
[B_{n_r},[\dots,[B_{n_2},B_{n_1}]\dots]],
\]
\[
Y=\sum_{r\ge 1}\sum_{n_1,\dots,n_r\in\mathcal N}
\frac{1}{r}
G^{\lambda(n_1),\dots,\lambda(n_r)}
[B_{n_r},[\dots,[B_{n_2},B_{n_1}]\dots]],
\]
and these satisfy
\[
[X_0,Z]=0,
\qquad
e^{\mathrm{ad}_Y}(X_0+B)=X_0+Z.
\]
The coefficients \(F\) and \(G\) are moulds in Écalle’s terminology. Their universality is the first sense in which Armould calculus is “armoured”: the same coefficient system can be deployed across multiple normalization problems.

## 2. Mould algebra, comoulds, and shuffle structures

Fix a nonempty alphabet \(\mathcal N\). Its free monoid \(\mathcal N^\bullet\) consists of finite words \(n=n_1\cdots n_r\), with empty word \(\emptyset\) as unit. A mould is a map \(M:\mathcal N^\bullet\to k\), written \(M^n\) at the word \(n\). The vector space \(k^{\mathcal N^\bullet}\) becomes an associative algebra under convolution,
\[
(M\times N)^n:=\sum_{n=ab} M^aN^b,
\]
where the sum runs over all factorizations \(n=ab\). The multiplicative unit is \(1^\bullet\), defined by \(1^\emptyset=1\) and \(1^n=0\) for \(n\neq\emptyset\).

Given a family \((B_n)_{n\in\mathcal N}\) in \(\mathcal L\), one constructs two associated objects. The associative comould \(B_\bullet\) sends a word \(n=n_1\cdots n_r\) to
\[
B_n:=B_{n_r}\cdots B_{n_1}
\]
in an associative algebra \(A\) containing \(\mathcal L\) as a Lie subalgebra. The Lie comould \(B_{[\bullet]}\) sends a nonempty word to the iterated bracket
\[
B_{[n]}:=[B_{n_r},[\dots,[B_{n_2},B_{n_1}]\dots]],
\qquad
B_{[\emptyset]}:=0.
\]
For any mould \(M\), the mould expansion
\[
M\cdot B_{[\bullet]}:=\sum_{n\neq\emptyset} M^n B_{[n]}
\]
is formally summable in the complete filtered setting.

The key combinatorics is controlled by shuffling. For two words \(a,b\), the shuffling coefficient \(\operatorname{sh}(a,b;n)\) counts the shuffles of \(a\) and \(b\) yielding \(n\). A mould \(M\) is alternal if
\[
M^\emptyset=0,
\qquad
\sum_n \operatorname{sh}(a,b;n)\,M^n=0
\quad\text{for all nonempty words }a,b,
\]
and a mould \(S\) is symmetral if
\[
S^\emptyset=1,
\qquad
\sum_n \operatorname{sh}(a,b;n)\,S^n=S^aS^b
\quad\text{for all nonempty }a,b.
\]
One has
\[
\mathrm{Sym}^\bullet(\mathcal N)=\{e^M: M\in \mathrm{Alt}^\bullet(\mathcal N)\},
\]
and \(\mathrm{Sym}^\bullet(\mathcal N)\) is a group for mould multiplication [1604.07142].

Alternality is the structurally relevant condition for Lie normalization. For any finite alternal mould \(M\),
\[
M\cdot B_{[\bullet]}=M\cdot B_\bullet,
\]
and the map \(M\mapsto M\cdot B_{[\bullet]}\) is a Lie algebra anti-morphism from alternal moulds to \(\mathcal L\). This is the mechanism that turns word-indexed scalar coefficients into iterated Lie-bracket formulae.

## 3. Mould equations and gauge classification

The eigenvalue map extends additively from letters to words by
\[
\lambda(n_1\cdots n_r):=\lambda(n_1)+\cdots+\lambda(n_r),
\qquad
\lambda(\emptyset):=0.
\]
This defines a derivation \(V_\lambda\) on moulds,
\[
(V_\lambda M)^n:=\lambda(n)M^n.
\]
A mould is \(\lambda\)-resonant if \(V_\lambda M=0\), equivalently if \(M^n\neq 0\) only when \(\lambda(n)=0\).

The Lie normalization problem is transferred to mould algebra through two identities:
\[
[X_0,M\cdot B_{[\bullet]}]=(V_\lambda M)\cdot B_{[\bullet]},
\]
and, for alternal \(M\),
\[
e^{\mathrm{ad}_{M\cdot B_{[\bullet]}}}X_0
=
X_0-\bigl(e^{-M}\times V_\lambda(e^M)\bigr)\cdot B_{[\bullet]}.
\]
If one writes
\[
Z=F\cdot B_{[\bullet]},
\qquad
Y=G\cdot B_{[\bullet]},
\]
then the normal-form relations become the mould equation
\[
V_\lambda F=0,
\qquad
V_\lambda(e^G)=I^\bullet\times e^G-e^G\times F,
\]
where \(I^\bullet\) is the mould supported on one-letter words.

This reduction is the decisive step of Armould calculus: the Lie-theoretic problem is reformulated as a universal algebraic problem on word-indexed coefficients. Theorem B classifies all alternal solutions. For each \(\lambda\)-resonant alternal mould \(A^\bullet\), there exists a unique pair \((F^\bullet,G^\bullet)\) solving the mould equation such that the gauge generator satisfies
\[
J_\lambda(G):=e^{-G}\times V_1(e^G)\big|_{\lambda=0}=A^\bullet,
\]
where \(V_1M^n=r(n)M^n\). Conversely, every alternal solution arises this way. Thus the set of solutions is an affine space modelled on resonant alternal moulds [1604.07142].

Writing \(S=e^G\), the equivalent system is
\[
V_\lambda F=0,
\qquad
V_\lambda S=I^\bullet\times S-S\times F,
\qquad
S\in\mathrm{Sym}^\bullet(\mathcal N).
\]
The proof provides explicit recursive formulas for \(F\), \(S\), and \(G=\log S\). In the generic nonresonant case, where \(\lambda(n)\neq 0\) for all nonempty words, the solution is unique, all \(F^n=0\), and \(S\) admits a closed formula. The paper relates this formula to a generalization of the classical Dynkin idempotent in the Hopf algebra of quasi-symmetric functions.

## 4. Gauge group, ambiguity, and classification of normal forms

The resonant symmetral moulds
\[
\mathrm{Sym}^\bullet_{\lambda=0}(\mathcal N)
:=
\{K\in\mathrm{Sym}^\bullet(\mathcal N)\mid V_\lambda K=0\}
\]
form the gauge group. If \((F,S)\) is one solution of the mould equation, any \(K\in\mathrm{Sym}^\bullet_{\lambda=0}(\mathcal N)\) yields another solution by
\[
(F',S')=(\mathrm{Ad}_K F,S\times K)
=
(K^{-1}\times F\times K,\;S\times K).
\]
The action is free and transitive: any two solutions are related by a unique gauge transformation.

On the Lie-algebra side, this gauge freedom matches the non-uniqueness of the normalizing transformation. If \((Z,Y)\) solves
\[
e^{\mathrm{ad}_Y}(X_0+B)=X_0+Z,
\qquad
[X_0,Z]=0,
\]
and if \(W\in\mathcal L_{\ge 1}\) satisfies \([X_0,W]=0\), then
\[
Z':=e^{\mathrm{ad}_W}Z,
\qquad
Y':=\mathrm{BCH}(W,Y)
\]
defines another solution. The ambiguity in choosing a normal form is therefore not arbitrary; it is canonically organized by the gauge group [1604.07142].

This directly addresses a recurrent misconception in normal form theory, namely that the non-uniqueness of normal forms is merely ad hoc. In the mould formalism it is a precise gauge phenomenon. The paper further notes that a canonical choice such as zero gauge, \(A=0\), selects one distinguished normal form, identified in the vector-field setting with Écalle’s “royal prenormal form.” Alternative gauges, such as requiring the resonant part \(G_{\lambda=0}=0\), produce different but gauge-equivalent normal forms.

## 5. Dynamical realizations

The same abstract scheme applies once one chooses a Lie algebra, a filtration, and an eigenvector decomposition of the perturbation under \(\mathrm{ad}_{X_0}\). This uniformity is one of the principal claims of the framework [1604.07142].

For Poincaré–Dulac theory, \(\mathcal L\) is the Lie algebra of formal vector fields
\[
X=\sum_{j=1}^N v_j(z)\,\partial_{z_j}
\]
with no constant term, filtered by degree in \(z\). With
\[
X_0=\sum_{j=1}^N \omega_j z_j\partial_{z_j},
\]
each monomial vector field \(z^k\partial_{z_j}\) is an eigenvector of \(\mathrm{ad}_{X_0}\) with eigenvalue \(\langle k,\omega\rangle-\omega_j\). Grouping monomials by eigenvalue produces the \(B_\lambda\), and the resulting \(Z\) satisfies \([X_0,Z]=0\). Here that means precisely that only resonant monomials, with \(\langle k,\omega\rangle-\omega_j=0\), remain. Thus \(X_0+Z\) is a Poincaré–Dulac normal form. In the nonresonant case, \(Z=0\), so the system is formally linearizable.

For classical Birkhoff normal forms, \(\mathcal L\) is the Poisson algebra of formal Hamiltonians in \((x,y)\in k^{2d}\), filtered by degree. Taking
\[
H_0(x,y)=\sum_{j=1}^d \omega_j\frac{x_j^2+y_j^2}{2},
\]
and passing to complex coordinates,
\[
z_j=\frac{x_j+iy_j}{\sqrt 2},
\qquad
w_j=\frac{x_j-iy_j}{i\sqrt 2},
\]
one obtains
\[
\{H_0,z^kw^\ell\}=i\langle k-\ell,\omega\rangle z^kw^\ell.
\]
Grouping monomials by \(n=k-\ell\in\mathbb Z^d\) gives eigenvectors \(B_n\) with \(\lambda(n)=i\langle n,\omega\rangle\). The resulting \(Z\) satisfies \(\{H_0,Z\}=0\), hence is a Birkhoff normal form. If \(\omega\) is strongly nonresonant, then \(Z\) is a formal series in the actions \(I_j=(x_j^2+y_j^2)/2\).

For multiphase averaging, the slow–fast vector field
\[
X=
\sum_{j=1}^d (\omega_j+\varepsilon f_j(\varphi,I,\varepsilon))\,\partial_{\varphi_j}
+
\sum_{k=1}^N \varepsilon g_k(\varphi,I,\varepsilon)\,\partial_{I_k}
\]
has unperturbed part
\[
X_0=\sum_{j=1}^d \omega_j\partial_{\varphi_j}.
\]
Fourier modes \(e^{i\langle n,\varphi\rangle}\) are eigenvectors of \(\mathrm{ad}_{X_0}\) with eigenvalue \(i\langle n,\omega\rangle\). The normal form \(Z\) contains only resonant Fourier modes. If \(\omega\) is strongly nonresonant, then \(Z\) is independent of \(\varphi\), and the fast angles are formally eliminated to infinite order.

For quantum Birkhoff normal forms, the Lie algebra consists of formal perturbations \(X=X_0+B\) in \(\mathcal E[[\varepsilon]]\), where \(\mathcal E\) is the algebra of finite-column operators and
\[
[A,B]_{\mathrm{qu}}:=\frac{1}{ih}(AB-BA).
\]
If \(X_0e_k=E_ke_k\), then each rank-one operator \(|e_\ell\rangle\langle e_k|\) is an eigenvector of \(\mathrm{ad}_{X_0}\) with eigenvalue \(E_\ell-E_k\). Grouping matrix elements by common energy difference gives the decomposition needed for the mould formalism. The normal form \(Z\) then satisfies \([X_0,Z]_{\mathrm{qu}}=0\), so \(Z\) is block-diagonal with respect to the spectral decomposition of \(X_0\); when the spectrum is simple, it is diagonal. If \(X_0\) is self-adjoint and \(B\) is symmetric, the construction respects the \(*\)-structure, and \(\exp(ihY)\) is a formal unitary conjugating \(X\) to its quantum normal form.

## 6. Semi-classical correspondence and universality

A particularly sharp application concerns harmonic oscillators and the relation between classical and quantum Birkhoff normal forms. Classically,
\[
H_0(x,\xi)=\sum_{j=1}^d \frac{\omega_j}{2}(\xi_j^2+x_j^2),
\]
while the corresponding Weyl-quantized operator on \(L^2(\mathbb R^d)\) is
\[
X_0=\sum_{j=1}^d\Bigl(-\frac{h^2}{2}\partial_{x_j}^2+\frac{\omega_j^2x_j^2}{2}\Bigr).
\]
Its spectrum is
\[
E_k(h)=h\bigl\langle k+(1/2,\dots,1/2),\omega\bigr\rangle,
\qquad
k\in\mathbb N^d,
\]
so
\[
E_\ell(h)-E_k(h)=h\langle n,\omega\rangle,
\qquad
n=\ell-k\in\mathbb Z^d.
\]
Crucially, the eigenvalue map entering the mould equation is \(\lambda(n)=i\langle n,\omega\rangle\), independent of \(h\). The same mould \(F\) therefore controls both the classical and quantum normalizations [1604.07142].

If \(B^{\mathrm{cl}}\) is a classical perturbation, formal in \(\varepsilon\) with coefficients in the Schwartz class, then its decomposition into eigencomponents \(B^{\mathrm{cl}}_n\) yields the classical Birkhoff normal form
\[
H_0+Z^{\mathrm{cl}},
\qquad
Z^{\mathrm{cl}}=F\cdot B^{\mathrm{cl}}_{[\bullet]}.
\]
After Weyl quantization,
\[
X=X_0+B^{\mathrm{qu}},
\qquad
B^{\mathrm{qu}}=\mathrm{Op}^W(B^{\mathrm{cl}}),
\]
the corresponding quantum normal form is
\[
X_0+Z^{\mathrm{qu}},
\qquad
Z^{\mathrm{qu}}=F\cdot B^{\mathrm{qu}}_{[\bullet]}.
\]
The key analytic input is that the symbol of the quantum commutator tends to the Poisson bracket as \(h\to 0\), and iteratively the symbols of the quantum iterated commutators converge to the classical iterated Poisson brackets. Consequently,
\[
\sigma(Z^{\mathrm{qu}})\xrightarrow[h\to 0]{} Z^{\mathrm{cl}}
\]
termwise in the formal parameter \(\varepsilon\).

The paper emphasizes that this semi-classical convergence requires no Diophantine condition on \(\omega\). Earlier results by Graffi–Paul and by Degli Esposti–Graffi–Herczynski did require such a condition. Here the absence of a Diophantine hypothesis is tied to the algebraic character of the mould construction: the same universal coefficients are determined solely by \(\lambda(n)=i\langle n,\omega\rangle\), without small-divisor estimates.

More broadly, the framework adopts Écalle’s mould calculus, in the Lie-algebraic form developed by Paul–Sauzin, but shifts the entire argument to an abstract Lie algebra rather than an operator algebra. The paper proves the needed properties in a self-contained manner, with some technical help from dimoulds. Its principal universal claim is that the heavy algebraic work—solving the mould equation, handling resonances, and organizing gauge freedom—is done once at the level of moulds. Plugging the resulting coefficients into different Lie comoulds then produces Poincaré–Dulac normal forms, classical Birkhoff normal forms, multiphase averaging, quantum Birkhoff normal forms, and the semi-classical correspondence. This is the precise content of the “armoured” power of mould, or Armould, calculus.

Source: https://www.emergentmind.com/topics/armould-calculus