Independent adversarial audit: classical pedal controls
UTC checkpoint: 2026-10-16T06:17:08Z
Completion: 100% of this bounded algebraic audit; target theorem progress not assessed.
Disposition: PASS WITH EXPLICIT CONVENTION CAVEATS. No algebraic defect found in
exact_classical_controls.py or either supplied receipt. No historical claim audited.

1. Strongest verified result

Let real nonzero normals n_i and real supports h_i define an ordered cycle
n_i dot z = h_i, with det(n_i,n_{i+1}) nonzero. Let T_i be consecutive
intersections, Q_i(M) orthogonal feet, and A', B_M their signed shoelace
areas in precisely this order. Set

  w_i = 2 det(n_i,n_{i+1})(n_i dot n_{i+1})
        / (|n_i|^2 |n_{i+1}|^2),
  S = sum_i w_i,  W = sum_i w_i T_i.

Then, including self-crossing cycles and external M,

  4(2 B_M - A') = sum_i w_i |M-T_i|^2,
  B_M = B_0 + S |M|^2/8 - M dot W/4.

This proves the requested formula for h_i=1 without a convexity assumption.

Checkable algebraic proof:
Put t_i=T_i-M and q_i=Q_i(M)-M. Because n_i dot t_i=h_i-n_i dot M,
q_i is the projection of t_i onto n_i. Direct two-dimensional expansion gives

  2 det(q_i,q_{i+1}) - (w_i/2)|t_i|^2
    = det(q_i,t_i) - det(q_{i+1},t_i).

On summing, the right side is
sum_i det(q_i,t_i-t_{i-1}) = sum_i det(t_{i-1},t_i) = 2A',
because q_i,t_i,t_{i-1} are collinear. The left side is
4B_M - (1/2)sum_i w_i|t_i|^2. Rearrangement proves the identity and
expansion proves the second formula.

If the ordered polygon is centered at 0 and paired by
T_{i+N/2}=-T_i with opposite side normals, then
w_{i+N/2}=w_i, so W=0 and B_M=B_0+S|M|^2/8.
Central symmetry alone does not cancel W about an arbitrary origin:
for a center C the correct formula is W=C S and
B_M=B_C+S|M-C|^2/8. The independent translated hexagon has
S=4, C=(2/3,-5/7), and W=(8/3,-20/7), a concrete falsifier of an
unqualified zero-moment claim.

2. Sign, ordering, and denominator restrictions

- delta_i must mean the oriented turn from n_i to n_{i+1}, modulo pi.
  Then w_i=sin(2 delta_i). An unsigned interior angle generally reverses
  this sign. Independent normal sign changes do not alter w_i.
- A' and B_M are signed areas in the same side cycle. Reversal negates
  both areas and the weight sum. Taking absolute areas, sorting vertices,
  or replacing a star's signed area by its filled area is unsupported.
- Nonzero real normals and nonparallel consecutive lines are required.
  The formula does not define a vertex for a parallel or coincident pair;
  its zero weight must not be interpreted as removing an infinite vertex.
  Nonconsecutive parallels, right angles (zero weight), and feet outside
  side segments cause no problem.
- Central pairing must hold in the ordered polygon, not merely for an
  unordered collection of vertices or lines.

3. Exact excluded-parity orbit controls

Outer ellipse: x^2/4+y^2=1.
Inner proper confocal caustic: (5/16)x^2+5y^2=1, lambda=4/5 in (0,1).

Orbit                  A       A'      B_0     AA'     A'B_0
(2,0),(0,1),...         4       8       4       32      32
(4,1),(-4,1),... /sqrt5 16/5    10      5       32      50

Both cycles have four distinct vertices and are primitive period 4.
Each chord is tangent to the stated inner ellipse, the tangency point
lies strictly inside the chord segment, and the physical reflection
direction is correct. Contact segment parameters are
(1/5,4/5,1/5,4/5) and (1/2,1/2,1/2,1/2), respectively.
Every weight vanishes individually, so B_M=A'/2 for every M in these
two cycles. The values are genuine geometric controls; they refute
an unrestricted inference from Steiner plus constant AA' to constant
A'B_0. They are excluded-parity examples and provide no counterexample
to a theorem restricted to another parity.

4. Other controls and exact remaining gap

The trapezoid (0,0),(2,0),(1,1),(0,1) has
w=(-1,1,0,0), S=0, W=(-1,1), A'=3/2, and

  B_(x,y) = 1/2 + x/4 - y/4.

Thus S=0 permits nonconstant linear pedal-area levels and does not
supply the finite center W/S. If S=W=0 the area is constant instead.
When S is nonzero, completing the square does give center W/S.

For x+y=1, the origin's foot is (1/2,1/2). Under (x,y)->(2x,y)
it maps to (1,1/2); the transformed line x/2+y=1 has Euclidean foot
(2/5,4/5). The affine-projection noncommutation control is correct.

The radial identity does not prove any phase-family product invariant.
For a centered family, A'B_M=A'B_0+|M|^2 A'S/8; invariance for every M
requires both coefficient invariants. The finite controls supply no
missing Poncelet-family invariant and establish no target all-N result.

5. Reproduction and evidence

verify_classical_controls.py uses only the Python standard library.
It independently checks every polynomial coefficient of the general
identity in M for 448 rational directed line cycles (224 cycles in both
orientations), plus an explicit self-crossing pentagram and reversal,
centered and translated central hexagons, both N=4 orbits, the full
trapezoid polynomial, affine projection, and two denominator failures.
It also reruns the parent source and exactly reproduces both receipts.
Input SHA-256 values were unchanged before and after verification.
VERIFY_RESULTS.json and VERIFY_RESULTS_OPTIMIZED.json record exact values
and input hashes. All 23 validation conditions use explicit exception
checks; no Python assert remains. Normal and -O runs match on every
result field except the UTC timestamp. A deliberately false condition
is rejected with the expected failure in both modes.
MODE_ENFORCEMENT_RECEIPT.json records these enforcement checks.
HASH_RECEIPT.json records local artifact hashes.

No other research-family artifacts were read. No Git state, source
artifact, branch, index, checkout, remote, or external contact was changed.

