Book 3 · The Mini-Beast · Chapter 5 of 44
The dm³ System & Contact Normal Form
The geometric invariant shared across rooms.
α = dz - y dx
OrientationThe Invariant the Rooms Share
Two systems are the same dm³ system when they reduce to the same normal form. Not when they look alike, and not when a metaphor survives translation. This chapter states the form and the three numbers that index it.
Darboux’s theorem is the licence for the whole construction: every contact form is, locally, the same contact form. There are no local invariants of a contact structure to distinguish one from another. Whatever differs between two contact manifolds differs globally, or in the dynamics carried on top of the structure — never in the local geometry.
ConsequenceDomain-Agnosticism Is a Theorem, Not an Aspiration
If the four rooms each carry a contact structure, then locally they carry the same contact structure, and the only question left is what the dynamics do on it. This cuts both ways, and the second edge is the important one: finding the same local contact geometry in four domains is not by itself evidence of a connection between them. Darboux hands that over for free.
The content of Book 3 therefore has to live in the dynamics — in the specific normal form below, in the rank-one Hessian degeneracy at the fold, and in the numerical bands for (μ_max, ω, β). Those do not come free, and they are what can be wrong.
The Normal FormThree Equations, Three Invariants
In a tubular neighbourhood of the post-transition circle Γ (a helix in the full space, see the toy system below), every dm³ system is locally equivalent to:
The triple (μ_max, ω, β) comprises the canonical invariants of a dm³ system. Every instantiation — biological, plasma, financial, neural — reduces to this form. Identifying a new system means measuring three numbers, not proposing a new model.
μ_max — the radial rate
Negative in every orbit measured so far, which is the statement that the radial coordinate contracts onto the cycle while z > 0. The coefficient μ_max(1 − e−βz) changes sign at z = 0, so for z < 0 the same system expands. It is read off as a reconnection rate in plasma, a mean-reversion rate under stress in markets, a relaxation rate in the biological orbits. Its magnitude also enters d_f.
ω — the angular frequency
The dominant cycle frequency of the room. Between the two endpoints named here it spans about four orders of magnitude (a ratio near 6×103) — from 2π/86400 for the circadian clock to 0.45 rad/s for neural oscillations — without changing the structure of the equations. This spread is the strongest available evidence that the form is not an artefact of timescale.
β — the fold sharpness
β controls how abruptly (1 − e^−βz) saturates, and therefore how sharp the fold is. Empirically it sits in a narrow band, 1.6 ≤ β ≤ 2.4, across all six orbits. Nothing in the derivation requires that narrowness; it is an observation, and a target for anyone trying to break the framework.
RecognitionHow to See It in Data
Recognising the normal form in a new dataset is a three-step procedure, and it is the same procedure in every room.
- Build the manifold. Choose coordinates that span the state space, and equip them with the metric induced by the system’s own energy or information functional — the MHD energy Hessian in plasma, the Fisher information metric in markets.
- Check the Morse condition. Critical points non-degenerate away from the transition; Hessian rank dropping by exactly one at it. If rank drops by more, the system is not a single fold and the normal form does not apply.
- Fit three numbers. Extract μ_max from the post-transition decay envelope, ω from the dominant spectral peak, β from the saturation profile. Then test the residual against O(ρ²).
If the residual is not O(ρ²), the system is not in the normal form and no amount of reparameterisation will put it there. That is a negative result worth publishing.
The Toy SystemWhere the Framework Is Actually Sharp
Behind the four rooms sits an explicit toy system, and it is the only place in the framework where every claim can be checked to the last digit. It is worth meeting it before trusting anything fitted.
On r = 1 the radial rate is zero and ż = 1, so r = 1 is an attracting circle that the orbit travels as a helix in (r, θ, z): z advances without return, and there is no closed orbit in M. A Grönwall estimate gives a symmetric perturbation size ε₀ = 1/3. That is a size, not a basin. High-precision DOP853 integration at z(0) = 0 puts the inner boundary at r★ = 0.77594058, not at 2/3; a start at r = 0.70, inside |r − 1| < 1/3, does not converge. The basin is a set in (r, z), and r★ moves with z(0).
The gap is not a matter of sharpness. The coupling term 2(r − 1)e⁻ẓ is sign-aware — it behaves differently above and below r = 1 — so the basin is asymmetric and a symmetric Lyapunov argument cannot see it. The symmetric estimate does not merely lose sharpness; it misclassifies orbits.
The inner edge has been located numerically to eight digits and not proved. The Lean files that carry the surrounding structure include book2/lean/StabilityRadius.lean (τ·ε₀ = 2/3); the audit of 2026-10-03 did not re-read them, and a search did not locate a Lean statement of the basin result over (2/3, 1), so this page makes no claim about one. The integrator and the failing cases are public.
BridgesWhere This Connects
- Pointers · XII Bienal SBM, UFRN, 3–7 Aug 2026The basin question put to contact geometry and Reeb dynamics, and to planar vector fields via the Newton polyhedron.
- Book 6 · Elliptic / Poisson FoundationsThe analytic foundations the normal-form estimates lean on.
- GrönwallThe symmetric estimate itself — where 1/3 comes from, and why it cannot see an asymmetric basin.
- Book 6 · Cardiac Rhythms — Wenckebach, Lyapunov, and T*The normal form in a clinical setting, with T* rather than κ* as the exposed quantity.