SigmaLayer/NullSeamGeneralN: the third horn at every N (D3b) #
Category: dynamical measurement — the general-N null-seam witness
(specs/BACKLOG.md D3b; the two-cell witness is NullSeamWitness.lean).
The construction #
N cells on the register circle, cut at the corpus's own cumulative Born
positions loSum r i (BornFibrePartition.lean) of a positive weight vector
r summing to 1. Each closed cell is realised as a closed ball
cellArc r i around the existing cell midpoint cellMid r i
(PointerWeights.lean), radius r i / 2, which hands the cell mass to
AddCircle.volume_closedBall with no bespoke measure computation; the open
cell is a rep-preimage (CircleFibre.lean), so measurability is free.
The landing amplitudes are plateau tents
cellTent r i θ = max 0 (cellGap r − infDist θ (cellArc r i)) with
cellGap = (min r)/2: value exactly cellGap on the whole closed cell,
decaying outside, vanishing at distance cellGap. Off the seam the active
cell's tent dominates strictly (it sits at the plateau, every other tent is
strictly below it, and at most one other tent is nonzero — the separation
estimates); at a seam point the two adjacent tents are equal — the kiss.
The record criterion momentMap > ½ therefore reads: record i exactly on
the open cell i, no record exactly on the N seam points.
The propagator's unitary is a single global formula — the rotation by π/2
in the plane spanned by the ready direction f₀ and the (normalised)
amplitude vector a ⊥ f₀:
M(a) = I − f₀f₀ᵀ − aaᵀ + af₀ᵀ − f₀aᵀ,
orthogonal for every unit a with a₀ = 0 (seamRotation_mem), with first
column a (the landing). Because M is a fixed polynomial in the
amplitudes, continuity of the propagator reduces to continuity of the
tents — there is no per-boundary gluing of plane rotations, and hence none of
the monodromy trouble a piecewise-rotation design would have at the wrap.
Relation to the two-cell witness #
NullSeamWitness.lean stands as the minimal exhibit (its (f₁,f₂)-plane
rotation is the N = 2 kissing-crossing in bespoke form); this module proves
the same closure shape for every N ≥ 2 and every weight vector: continuity,
measure invariance, exact records off an N-point null seam, and exact
Born mass r i per cell (nullSeamGeneralClosure). The scope notes of the
two-cell module apply verbatim: the exactness is at the Dirac-calibrated
ready point (the third horn's price — posMeasure_noRecord_pointer), and the
cell split r plays the Born-weight role without a preparation in the arena.
Cumulative-position extras #
loSum (BornFibrePartition.lean) carries the ordering lemmas
(loSum_add_le_loSum, loSum_add_self_le_one); the two small identities
below complete the tiling picture.
The canonical representative #
The tent half-width #
The tent half-width: half the smallest cell width. Every tent extends
exactly cellGap beyond its cell, so tents of cells that do not share a
boundary never overlap.
Equations
- CSD.RecordLayer.cellGap r = Finset.univ.inf' ⋯ r / 2
Instances For
Circle lemmas beyond the two-cell toolkit #
The cells #
Closed cell i: the closed ball of radius r i / 2 around the existing
cell midpoint cellMid r i (PointerWeights.lean). Equal to the arc image of
its lift interval (cellArc_eq_image).
Equations
- CSD.RecordLayer.cellArc r i = Metric.closedBall (CSD.RecordLayer.cellMid r i) (r i / 2)
Instances For
The open cell: the rep-preimage of the open CDF interval, so
measurability is definitional (compare circleCell).
Equations
- CSD.RecordLayer.openCell r i = CSD.RecordLayer.rep ⁻¹' Set.Ioo (CSD.RecordLayer.loSum r i) (CSD.RecordLayer.loSum r i + r i)
Instances For
The seam point at the left end of cell i. The N of them are the cell
boundaries (the right end of the last cell wraps to the left end of the
first).
Equations
- CSD.RecordLayer.seamPoint r i = ↑(CSD.RecordLayer.loSum r i)
Instances For
Every circle point lies in some closed cell (via its canonical representative, whose CDF interval is found by taking the largest cumulative position strictly below it).
The tents #
The plateau tent of cell i: value cellGap r on the whole closed cell,
decaying with the distance outside, vanishing at distance cellGap r.
Equations
- CSD.RecordLayer.cellTent r i θ = max 0 (CSD.RecordLayer.cellGap r - Metric.infDist θ (CSD.RecordLayer.cellArc r i))
Instances For
Vanishing at distance cellGap.
Positivity means the cell is within tent reach.
Separation: at most one foreign tent is ever within reach #
Tent-positivity at a lift, resolved into the two approach directions:
if cell k's tent is positive at coe s, then s is within cellGap of
the cell's lift interval either directly or around the wrap.
Foreign-tent approach bounds, cell to the left of the active one.
Foreign-tent approach bounds, cell to the right of the active one.
At most one foreign tent is within reach at any point of an open
cell. The twelve position/branch combinations each contradict the cumulative
ordering (loSum_add_le_loSum), the one-turn bound
(loSum_add_self_le_one), or the width floor (two_cellGap_le).
The amplitude vector and the record criterion #
The (unnormalised) landing amplitude vector: zero in the ready slot,
tent i in record slot i.succ.
Equations
- CSD.RecordLayer.tentVec r θ = WithLp.toLp 2 fun (k : Fin (N + 1)) => Fin.cases 0 (fun (i : Fin N) => ↑(CSD.RecordLayer.cellTent r i θ)) k
Instances For
The seam rotation #
The rotation by π/2 in the plane spanned by the ready direction f₀
and a unit amplitude vector a ⊥ f₀, over ℝ: first column a, first row
−aᵀ (off the corner), the record block I − aaᵀ.
Equations
Instances For
The seam rotation over ℂ: the real matrix, entrywise embedded.
Equations
Instances For
The first column of the seam rotation is the amplitude vector.
The propagator #
The total tent weight (the amplitude normaliser).
Equations
- CSD.RecordLayer.tentTotal r θ = √(∑ j : Fin N, CSD.RecordLayer.cellTent r j θ ^ 2)
Instances For
The normalised landing amplitudes: empty ready slot, tent/total in the
record slots.
Equations
- CSD.RecordLayer.seamAmp r θ k = Fin.cases 0 (fun (i : Fin N) => CSD.RecordLayer.cellTent r i θ / CSD.RecordLayer.tentTotal r θ) k
Instances For
The propagator's unitary at register point θ.
Equations
- CSD.RecordLayer.nullSeamGenUU hr hsum θ = ⟨CSD.RecordLayer.seamRotation (CSD.RecordLayer.seamAmp r θ), ⋯⟩
Instances For
The general-N null-seam propagator: register conserved, pointer
rotated by the seam rotation at the register's amplitudes.
Equations
- CSD.RecordLayer.nullSeamGenEvolve hr hsum y = (y.1, CSD.RecordLayer.nullSeamGenUU hr hsum y.1 • y.2)
Instances For
The landing identity: the propagator sends the calibrated ready state to the tent ray.
Records: exactly the open cells #
The record criterion: outcome i is recorded iff cell i's tent
strictly dominates all the others combined.
The seam: the kiss, and no record #
★ No record at any seam point, for any outcome — the kiss: two tents sit at the plateau, so no tent strictly dominates.
The outcome sets are exactly the open cells #
★ Exact Born, every cell: the record-i outcome set has measure
exactly r i.
★ The seam is null — indeed N points.
★★ Continuity and measure invariance #
Entrywise continuity of the seam rotation over the register.
The pointer component of the propagator is continuous.
★★ The propagator is continuous on the whole arena.
The arena's invariant measure: Haar on the register, Fubini–Study on the
pointer (the same product as the two-cell witness's nullSeamMeasure).
Equations
Instances For
★★ Measure invariance — a skew product: register conserved, every register slice acts by an FS-preserving unitary.
★★ The third horn at every N, bundled #
The general-N third horn: continuous, measure-preserving dynamics on
S¹ × ℂℙ^N whose records from the calibrated ready state are exact and
correct off an N-point null seam, with exact Born mass r i per cell.
The price is unchanged from the two-cell witness: Dirac calibration
(posMeasure_noRecord_pointer prices the alternative).
- continuity : Continuous (nullSeamGenEvolve hr hsum)
The propagator is continuous on the whole arena.
- invariant (q₀ : Pointer N) : MeasureTheory.MeasurePreserving (nullSeamGenEvolve hr hsum) (nullSeamGenMeasure q₀) (nullSeamGenMeasure q₀)
Measure invariance at every Fubini–Study base point.
- landing (i : Fin N) (θ : CircleFibre) : θ ∈ openCell r i → (nullSeamGenEvolve hr hsum (θ, readyState)).2 ∈ recordRegion i
Correct record on every open cell — exactly.
- seam_kiss (i j : Fin N) : (nullSeamGenEvolve hr hsum (seamPoint r i, readyState)).2 ∉ recordRegion j
No record at any seam point, for any outcome — the kiss.
The seam is null.
- born (i : Fin N) : MeasureTheory.volume {θ : CircleFibre | (nullSeamGenEvolve hr hsum (θ, readyState)).2 ∈ recordRegion i} = ENNReal.ofReal (r i)
Exact Born, every outcome.
Instances For
The uniform witness: N cells of weight 1/N each — non-vacuity of the
closure at every N ≥ 2.