Shadow Faraday Zero-Out · v0.1 · pre-registered
Do local shadows recover Faraday induction?
A minimal, falsification-first experiment. We define a local, gauge-invariant shadow projection 𝒫shadow and ask whether it is sufficient to close Faraday's law of induction algebraically — without reconstructing the electromagnetic potential Aμ globally. The result lands in one of three pre-registered zones: clean structural zero, named quarantine, or bounded failure. All three ship a receipt. For the AI-safety translation of this same apparatus, see the shadow-method bridge essay.
The Pre-Registered Claim
One algebraic question, three landing zones, all of them publishable.
The Sundog method is allergic to ambiguity, not to negative results. We register the predicate before the algebra runs, then publish whichever of structural zero, named quarantine, or bounded failure falls out. Faraday is the calibration object: too well-understood to fudge, too foundational to ignore.
This page is a receipt surface: Phase 3 now lands Branch A on the registered classical vacuum domain, and Phase 4 verification / falsification passes 5/5. Phase 5 chapter close is signed, and the page now carries the public metadata, share card, sitemap entry, and internal path.
The Object · I
Antisymmetric, two scalars, gauge-invariant.
All of classical electromagnetism, including Faraday's law, Gauss's law for magnetism, and Lorentz invariance under boosts, compresses into a single 4×4 antisymmetric tensor Fμν = ∂μAν − ∂νAμ and exactly two Lorentz scalars. The shadow apparatus must respect the same symmetries.
| 0 | Ex/c | Ey/c | Ez/c |
| −Ex/c | 0 | −Bz | By |
| −Ey/c | Bz | 0 | −Bx |
| −Ez/c | −By | Bx | 0 |
Invariant I · scalar density
Invariant II · pseudoscalar · CP-odd
The Apparatus · II
𝒫shadow is the plaquette holonomy on A.
Phase 2 fixes the operator as a two-tier coordinate plaquette holonomy. For a point x, a coordinate plaquette ωμν(x, ε) of edge ε, and the closed boundary loop ∂ω, the finite-stencil operator is the Wilson-loop-style line integral of the potential one-form:
Stokes' theorem turns the line integral of A on the closed loop into the surface integral of F on the plaquette — and F is gauge-invariant by construction. The operator never inspects a globally consistent Aμ: it reads A only on the four edges of one plaquette.
The pre-registered audit predicate is deceptively simple, and load-bearing:
The proof on the registered domain is a single line: ∮∂ω dλ = 0 for any closed loop ∂ω with smooth λ, so the gauge term drops out of the holonomy by Stokes. The audit fails by name outside that domain: five quarantine hooks — regularity, topology, monopole, operator-stencil commutator, motional EMF — are enumerated in the Phase 2 receipt before any Phase 3 algebra runs. The discipline lineage is Isotrophy K_facet v0.3h, whose audit chain landed 20 structural-zero receipts and the named O_617 quarantine on exactly this pre-registration pattern.
Phase 2 receipt: docs/faraday/SHADOW_FARADAY.md ▸ Phase 2: Local Shadow Projection Operator. Phase 2 sign-off decisions are recorded there: coordinate plaquettes, point-limit gate plus finite-stencil locality receipt, no Floquet/twist rescue in Phase 3, and the two-tier operator retained with roles locked.
The Battery · III
Three falsifiers, pre-committed and now run.
Each falsifier is designed to break the claim in a pre-named way. Anything else — any unexpected residual, any silent failure — is the receipt worth writing down. These were registered before Phase 3 algebra ran. Phase 4 passed 5/5: constant B, a source-free plane wave, a nonlocal residual, a monopole quarantine, and gauge invariance.
Non-local probe
Deliberately widen the stencil beyond the locality radius.
observed: residualArtificial monopole
Inject a source term forbidden by the source-free assumption.
observed: quarantineGauge after projection
Apply A → A + dλ after 𝒫shadow has run.
observed: invariantThe Pipeline · IV
Symbol table → public page · six phases · each gated on a signed receipt.
Same discipline as Isotrophy v0.3 and the three-body controller: each phase exits when its receipt is signed off, not when it feels done. Phase 3 — the core zero-out — is the load-bearing algebra; Phase 4 is the support battery; Phase 5 closes the chapter before public dissemination.
The Outcomes · V
Three landing zones. All of them ship.
Pre-registered framing turns every possible Phase 3 result into a public receipt. The discipline is the same one that produced Isotrophy K_facet v0.3h's 20 structural zeros + 1 named quarantine: name the success predicate before the algebra runs; if the result is a partial, the partial is the artifact.
Structural identity
Shadow projection is sufficient; Faraday induction closes locally with no residual. The strongest claim — and the most demanding receipt.
Topological / boundary term
One precisely-named survivor — boundary at infinity, Aharonov–Bohm phase, or a documented global obstruction. Closure is conditional; the failure is articulate.
Reconstruction survives
Shadow projection is insufficient at order X. The bound is publishable, the prior is updated, the scope is corrected.
The Boundary · VI · Phase 7
Three steps off the clean domain — each named before the algebra ran.
Phase 7 does not broaden Branch A; it audits the first
boundary around it. Three nearby cases were registered with
their expected landings before any computation, then
derived by hand and mirrored by a seconds-scale battery
(npm run faraday:phase7). All three landed where
pre-registered — no surprise residual in any clean,
contractible computation.
Electric sources are invisible
Add a real current (Jμ ≠ 0) but keep dF = 0: the Faraday residual stays an exact zero. Electric sources live in the inhomogeneous law d⋆F = J, never in dF — closure is source-tolerant in the narrow Bianchi sense only, not a full-Maxwell claim.
A named magnetic flux
Insert a registered magnetic charge (dF = Km ≠ 0): the survivor is exactly the magnetic flux Qm through the audited volume — named and accounted, not a clean-domain failure.
The Aharonov–Bohm survivor
Encircle a solenoid where F = 0: the local field reads nothing, yet the loop holonomy ∮A = Φ survives — the enclosed flux, an exact topological invariant (H1).
Phase 7 receipt:
docs/faraday/FARADAY_PHASE7_BOUNDARY.md
· branches A7-clean-control / B7-magnetic-source / B7-topology
· support battery npm run faraday:phase7.
Classical only; single-valued λ keeps Φ exactly
gauge-invariant. For the cross-substrate reading — the first
exact shadow-decides-without-reconstructing
separation in the portfolio, bounded to topological scale —
see the public body-resistance vocabulary.
The Completion · VII · Phase 8
The other half of Maxwell — and why Aharonov–Bohm was the only sharp witness.
Faraday is the homogeneous half (dF = 0, an
identity). Phase 8 turns to the inhomogeneous half
d⋆F = J — the sourced, metric-bearing side —
through a dual shadow
𝒫shadowdual = ∮⋆F. Three
registered cases, hand-derived and mirrored by a seconds-scale
battery (npm run faraday:phase8), landed exactly
where pre-registered.
The dual of the monopole
The dual shadow recovers Gauss's law: ∮⋆F = Qenc. Electric charge lives in d⋆F = J exactly where magnetic charge broke dF = 0 in Phase 7 — the Hodge star ⋆ is the swap, and the place the metric finally enters.
Displacement current closes the loop
With a time-varying field the dual shadow returns ∮B·d𝓁 = Ienc + d/dt ΦE — conduction plus the distinctively-Maxwell displacement current. The battery matches to thirteen digits.
No new sharp separation
On a topological domain the field splits into a sourced piece (determined by its charge) and a harmonic piece (= the Phase 7 Aharonov–Bohm flux). The dual shadow reads the charge; the loop shadow reads the flux — orthogonal sectors.
Phase 8 receipt:
docs/faraday/FARADAY_PHASE8_BOUNDARY.md
· branches A8-dual-closure / B8-harmonic-survivor · support
battery npm run faraday:phase8. The dual-shadow
closure recovers the integral form of the inhomogeneous law —
it is not a re-derivation of Maxwell. Static /
given-source domain; free radiation, retarded dynamics, and
curved spacetime are named deferrals.
Where This Sits
Faraday is the calibration receipt; Generality is the stress test.
Shadow Faraday is the clean-room version of the Sundog pattern: define a local shadow, pre-register the branch table, let the algebra land, and keep the boundary named. The result closes because classical electromagnetism is already understood; the contribution is the audit shape, not a new law of physics.
The High-Stakes Generality page asks what happens when the same discipline is carried into domains where the answer is not known in advance or the result is bounded-null: Navier-Stokes C1, Riemann, Yang-Mills, P-vs-NP, ARC-AGI, and Three-Body. Read Faraday as the method receipt, then Generality as the harder transfer matrix.
Claim Boundary
Not a universal result.
The experiment closed Branch A on the registered classical-vacuum domain. Phase 7 audited the first boundary off it — electric sources, magnetic sources, and the Aharonov–Bohm topology — each landing as a named survivor, not a broadened claim. Distributional fields, curved spacetime, quantum EM, plasma, and moving surfaces remain named limitations, not silent claims.
Not a derivation of Maxwell.
Phase 8 closed the sourced side too — the dual shadow ∮⋆F = Qenc recovers Gauss and Ampère–Maxwell — but closure recovers the integral form of an already-given law. It is not a re-derivation of the Maxwell system from shadow primitives.
Not a claim about AI.
For the AI-safety translation of the same apparatus, see the dedicated bridge essay at /safety-method, marked SKETCH and load-bearing on this Branch A receipt as its worked example.
Inspection Trail