/-
Copyright 2026 The Formal Conjectures Authors.
Licensed under the Apache License, Version 2.0 (the "License");
you may not use this file except in compliance with the License.
You may obtain a copy of the License at
https://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software
distributed under the License is distributed on an "AS IS" BASIS,
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
See the License for the specific language governing permissions and
limitations under the License.
-/modulepublicimportFormalConjectures.Wikipedia.JacobianConjecturepublicimportFormalConjecturesUtil
The Poisson Conjecture
References:
[AvdE07] arxiv/math.0608009On the equivalence of the Jacobian, Dixmier and Poisson Conjectures in any characteristic
by Kossivi Adjamagbo, Arno van den Essen. Published as A proof of the equivalence of the
Dixmier, Jacobian and Poisson conjectures, Acta Math. Vietnam. 32 (2007), 205–214.
The Poisson Conjecture $PC_n$ ([AvdE07], Notations 5) asserts that, over a field of
characteristic zero, every endomorphism of the n-th canonical Poisson algebra
$P_n(K)$ — the polynomial algebra $K[X_1, \dots, X_{2n}]$ equipped with the canonical
Poisson bracket — is an automorphism.
By [AvdE07], Theorem 7 (the "United Conjectures Theorem"), for every n there is a chain
of implications
$$JC_{2n} \Longrightarrow PC_n \Longrightarrow DC_n \Longrightarrow JC_n,$$
where $JC_n$ is the Jacobian Conjecture in dimension n and $DC_n$ the Dixmier Conjecture
for the n-th Weyl algebra (the implication $DC_n \Longrightarrow JC_n$ being classical [BCW82]).
Since the Jacobian conjecture was disproved in dimension $3$ [Alp26] — and hence in every
dimension $n ≥ 3$ by padding with identity coordinates — $PC_n$ is false for all $n ≥ 3$:
composing the counterexample's failure of $JC_3$ through $PC_3 \Longrightarrow DC_3 \Longrightarrow JC_3$ refutes
$PC_3$, and directly, the cotangent (symplectic) lift of the counterexample map
([AvdE07], Theorem 1 in reverse) is a non-invertible Poisson endomorphism of $P_3(K)$.
$PC_2$ is false as well, by the explicit rank-two counterexample of [Lon26]. Only $n = 1$
remains open.
Indexing note: we index the $2n$ variables of $P_n(K)$ by Fin n ⊕ Fin n, with
Sum.inl i playing the role of $X_i$ and Sum.inr i the role of $X_{i+n}$ of [AvdE07],
so that the canonical bracket reads ${X_i, X_{i+n}} = 1$; see
MvPolynomial.poissonBracket.
A K-algebra endomorphism φ of the polynomial algebra in $2n$ variables is an
endomorphism of the n-th canonical Poisson algebra $P_n(K)$ if it preserves the canonical
Poisson bracket. Following [AvdE07], Lemma 2, it is equivalent to require preservation on
the generators only, which is the form used here (the brackets of generators are constants,
so preserving and intertwining them agree).
The Poisson Conjecture in dimension n ($PC_n$ in the notation of [AvdE07]): every
endomorphism of the n-th canonical Poisson algebra over K is an automorphism.
An endomorphism is an automorphism iff it is bijective, since the inverse of a bijective
algebra endomorphism preserving the Poisson bracket is again such.
The Poisson Conjecture ([AvdE07], the characteristic zero case): for every n, every
endomorphism of the n-th canonical Poisson algebra over a field K of characteristic
zero is an automorphism. This is false: the Jacobian conjecture fails in dimension $3$
[Alp26], hence so does $PC_3$ (via [AvdE07], Theorem 7, or directly via the symplectic
lift of the counterexample map).
The Poisson Conjecture in dimension $1$ ($PC_1$) is open. By [AvdE07], Theorem 7, it is
implied by the Jacobian conjecture in dimension $2$ (Keller's original problem, open) and
implies the Dixmier conjecture for the first Weyl algebra (Problem 1 of Dixmier (1968),
open).
The Poisson Conjecture is false in dimension $2$ ($PC_2$). [Lon26] gives four explicit
polynomials $R, T, D, S \in \mathbb{Q}[x, q, p, z]$ with
${R, D} = {T, S} = 1$ and ${R, T} = {R, S} = {D, T} = {D, S} = 0$, so that
$X_1 \mapsto R$, $X_2 \mapsto T$, $X_3 \mapsto D$, $X_4 \mapsto S$ is a Poisson endomorphism of
$P_2(K)$. It is not surjective: in the coordinates $(X_1, X_2, X_3, X_4)$ the two rational
points $(0, 0, 1/24, -1/8)$ and $(1, 2/3, 247/96, -89/64)$ have the same image under
$(R, T, D, S)$, so every polynomial in the image of the endomorphism takes equal values at
them, while $X_1$ does not. The formulas have rational coefficients and the two points are
rational, so the argument applies over every field of characteristic zero.
This does not settle any conjecture further down the chain of [AvdE07], Theorem 7, which only
propagates falsity upwards: $\lnot PC_2$ gives $\lnot JC_4$, already known from [Alp26].
$DC_2$, $JC_2$ (Keller's problem), $PC_1$ and $DC_1$ all remain open.
The Poisson Conjecture is false in dimension $3$: the cotangent (symplectic) lift
$\Phi(X_i) = F_i$, $\Phi(X_{i+3}) = \sum_j M_{ij} X_{j+3}$ with $M = (JF)^{-\top}$ of the
degree-$6$ counterexample map $F$ of [Alp26] (whose Jacobian determinant is the constant
$-2$, so $M$ is a polynomial matrix) is a Poisson endomorphism of $P_3(K)$ that is not
injective on points — e.g. it identifies $(0, 6, -142, 0, 0, 0)$ and $(1, 0, 2, 0, 0, 0)$
— and is therefore no automorphism. Alternatively, $PC_3$ fails by [AvdE07], Theorem 7,
as it implies the Jacobian conjecture in dimension $3$, contradicting [Alp26].