Provide a complete smooth-category proof of the Eliasson normal form

Prove, in full generality in the smooth category, the Eliasson normal form theorem for integrable systems near non-degenerate singular points, establishing local symplectic coordinates and commuting quadratic normal forms that linearize the system, and, in particular, prove the existence of the local diffeomorphism g composing F with the model Q under the stated hypotheses (including the focus–focus and hyperbolic cases).

Background

The Eliasson normal form asserts that near any non-degenerate singular point of an integrable system (M,ω,F), there exist local symplectic coordinates in which the components of F Poisson-commute with a model of quadratic forms (elliptic, hyperbolic, focus–focus, and regular blocks). In the non-hyperbolic case, one further expects a local diffeomorphism g so that g∘F equals the quadratic model.

While the theorem is proved in various special cases (analytic case; fully elliptic; focus–focus in dimension four; elliptic/hyperbolic in dimension two), the author notes that a complete proof in the smooth category for all cases does not exist in the literature, highlighting a foundational gap.

References

To my knowledge, there does not exist a complete proof of this theorem anywhere in the literature. It was originally proved only in the analytic case by Vey, and since then various special cases have been proved in the smooth case: completely elliptic in all dimensions, focus-focus in dimension four, and in all cases (hyperbolic and elliptic) in dimension two. There is also an equivariant version, which is proved using Theorem~\ref{thm:eliasson}. This is also discussed around Theorem 2.1 in and in Remark 4.16.

Integrable systems with symmetries: toric, semitoric, and beyond (2510.05337 - Palmer, 6 Oct 2025) in Remark following Theorem (Eliasson normal form), Section 4.2