Spectral splitting of the saddle-center linearization
ProvedBirkhoffGlobalSection.saddle_center_spectral_splittingdynamical-systemssymplectic-geometry
At , the saddle-center linearization splits: there are an elliptic speed , a hyperbolic rate , and four independent directions with
This is pure linear algebra of the explicit Hessian at : the characteristic polynomial has one imaginary and one real pair. It packages the rotation fact driving all subsequent index growth, with no analysis.
Preamble
import Definitions.Def_BirkhoffGlobalSection_AmbientRotation
Formal statement
namespace BirkhoffGlobalSection
theorem saddle_center_spectral_splitting :
∃ ω lam : ℝ, 0 < ω ∧ 0 < lam ∧ ∃ e₁ e₂ f₁ f₂ : Phase,
TangentialHessian.qI.mulVec
(((!![-16, 0, 0, 1; 0, 8, -1, 0; 0, -1, 1, 0; 1, 0, 0, 1] :
Matrix (Fin 4) (Fin 4) ℝ).mulVec e₁)) = ω • e₂ ∧
TangentialHessian.qI.mulVec
(((!![-16, 0, 0, 1; 0, 8, -1, 0; 0, -1, 1, 0; 1, 0, 0, 1] :
Matrix (Fin 4) (Fin 4) ℝ).mulVec e₂)) = -ω • e₁ ∧
TangentialHessian.qI.mulVec
(((!![-16, 0, 0, 1; 0, 8, -1, 0; 0, -1, 1, 0; 1, 0, 0, 1] :
Matrix (Fin 4) (Fin 4) ℝ).mulVec f₁)) = lam • f₁ ∧
TangentialHessian.qI.mulVec
(((!![-16, 0, 0, 1; 0, 8, -1, 0; 0, -1, 1, 0; 1, 0, 0, 1] :
Matrix (Fin 4) (Fin 4) ℝ).mulVec f₂)) = -lam • f₂ ∧
LinearIndependent ℝ ![e₁, e₂, f₁, f₂] := by sorry
end BirkhoffGlobalSection
Source
Liu--Salomao, https://arxiv.org/html/2506.17867v2#S7, Eq. (7.7) and the proof of Theorem 1.8 (the linearized flow at the saddle-center splits into a hyperbolic and an elliptic part, and the elliptic part forces index growth). Gutt, https://arxiv.org/pdf/1307.7239, Eq. (9). H0 is the Hessian of the platform Levi-Civita Hamiltonian at mu = 1/2, c = 2.