An endofunctor needs somewhere to be an endofunctor of. The corpus applies the chain to phase vectors, to ordinals, to protein conformations, to atmospheric states and to spinal-cord lesions. Either there is a category general enough to contain all of them, or the chain is being applied by analogy and the analogy is the claim.
| Setting | Objects | Where it appears |
|---|---|---|
| Phase vectors | Fin 12 → ℝ | AXLE, PhaseVector; the D6 lock-in theorems |
| Ordinals | Ordinal | AXLE regeneration levels; the Mahlo hierarchy |
| Contact manifolds | Manifolds with a contact form | Vol II, contact realization |
| Applied domains | States of a physical or biological system | Vols III–VII |
These are not obviously one category. A functor from ordinals to phase vectors is not automatic, and a construction that is an endofunctor of both is a strong claim rather than a notational convenience.
Ask a class to plan a vacation. It sounds like fun, and that is the trap: fun is not what you booked. There are many moving parts and they have to gel. Flights, dates, money, visas, who is coming, what is open when you arrive — each one is an operation, not a value, and each constrains the next.
Notice what that rules out. Nobody believes there is a function from {destination, dates, budget} to a good trip. Two people with the identical list can produce different holidays, and the difference is in the order they did things: book the flight first and the hotel is constrained; book the hotel first and the flight is. Same components, different assembly, different result — which is the neodymium spheres of the previous section, and the basin boundary at r* of the section before that, stated in a language a first-year student already has.
Three things the frame gets right, and they are the three this chapter is about:
VALUE PREMISE A framework whose central object cannot be explained to a beginner in one paragraph is not yet understood by the person writing it. This paragraph is therefore a test of the chapter, not an ornament on it.
The LAW3M XIII submission brief states the system explicitly, on the contact 3-manifold this series has used since Volume II:
(ℝ³, α = dz − r²dθ), ε = 2
ṛ = r(1 − r²) + 2(r−1)e⁻ṫ
θ̇ = 1
ż = r² − 2(r−1)²e⁻ṫ
Γ = {r = 1, θ̇ = 1, ż = 1}, Lyapunov μ = −2
r* = 0.77594058 DOP853, rtol 1e−12, bisection tol 1e−7
That is a flow, not a function — an ODE with a basin boundary at r* separating initial conditions that reach the attractor from those that do not. DATA The value is computed, not asserted, and sits in the stated chain ε₀ = 1/3 < 2/3 < r* = 0.77594 < κ* = √(7/9) < 1.
So the path-dependence the chain has always claimed does not live in a weak composition. It lives in the basin structure: two routes are two initial conditions, and whether they land on Γ is decided by which side of r* they start.
Two hundred and sixteen neodymium spheres do not have an outcome; they have outcomes, plural, determined by the route along which they were assembled. Chain first then close it, or ring first then thicken it, and an identical starting set lands in different stable structures. That is hysteresis one can hold in the hand, and a function on a state space cannot represent it.
The composition question is equally tangible. Take three clusters. Join A to B, then attach C; then build B with C first and bring A. The results are not the same object — one snaps into a configuration one would call the same, after a rearrangement. That rearrangement is invertible and is not the identity. It is an associator, in the hand.
And the hexagon is already there: six around one, shells outward — G6, in a form that pours out of a tin. OPEN The artifact this chapter needs may not be Lean first. Same count, two assembly routes, photograph both, and state whether the results are equal or merely isomorphic.
Time addition is commutative: φs ∘ φt = φt ∘ φs. But the corpus treats [K, F] ≠ 0 as load-bearing — it is named as the algebraic source of the three-term Tribonacci recurrence, and CatGT derives it from a naturality failure. A single flow cannot supply both.
The likely resolution, and the first thing to test: the fold is exactly where the flow fails to be a diffeomorphism — rank-1 loss of injectivity, by Definition of F. So the composite is a flow only away from the fold locus, and the operator ordering survives as a statement about crossing the fold rather than about a product of arrows. If that holds, outcome 4 does not delete [K, F] ≠ 0; it explains why the fold is the only place it can come from.
Until that is settled, outcome 4 is a candidate, not a verdict.
Every later chapter — associativity, the associator, the pentagon, the 33-fold composite — is conditional on this one. Coherence data for a chain that has no category to live in is not a hard problem; it is an undefined one.
A named category with its objects and morphisms written down, and a stated verdict among the four outcomes above. Outcome 2, 3 or 4 closes the chapter as honestly as outcome 1; only silence fails.
Two artifacts would decide it, and neither is Lean. (a) Read LAW3M/law3m-brief.html and abstract-law3m-2026.html in full and re-run certify_rstar.py to confirm r* reproduces — that fixes whether the empirical object is the flow. (b) Run the two-route assembly, photograph both, and record equal-or-isomorphic. Then settle the [K, F] ≠ 0 tension above before writing any of it into Chapter 4.