Evidence

Hypotheses

• H-001

  An RK4 twin of the damped oscillator reproduces the closed form on [0,20].

  open

• H-002

  One-step F remains superior to persistence under isotropic process noise.

  open

• H-003

  CUSUM detects a known N(0,1)→N(2.5,1) shift with delay ≤ 30 and no early FA.

  open

• H-004

  A rolling-mean drift alarm detects linear mean drift before t=150.

  open

• H-005

  A radial anomaly score attains AUROC ≥ 0.9 and does not lose to max-|z|.

  open

• H-006

  The architecture detects Kuramoto synchronization at high K and not at low K.

  open

• H-008

  The compiler implements [n]_q including the q→1 removable discontinuity.

  open

• H-Q01

  q-factorial recurrence, 7! at q=1, and e_q(z)E_q(−z)=1 hold in the kernel.

  open

• H-Q02

  The Jackson derivative of x^n equals [n]_q x^{n−1} on a declared grid.

  open

• H-Q03

  Phillips q-Bernstein reproduces constants and affine functions.

  open

• H-Q04

  H and X are involutions; H|0⟩=|+⟩; S(pure)=0 and S(I/2)=1 bit.

  open

• H-Q05

  The Barnsley FIF interpolates its knots; d=0 is piecewise linear.

  open

• H-Q06

  A Bell |Φ⁺⟩ twin violates CHSH; a product twin does not.

  open

• H-Q07

  The first sign change of Hardy Z(t) matches the published ordinate 14.1347…

  open

• H-Q08

  A 2D leapfrog wave stays inside the CFL bound with finite energy.

  open

• H-A01

  The FAO-56 bucket conserves water over a 180-day generated season.

  open

• H-A02

  Ra at equator equinox equals (24·60/π) G_sc; Hargreaves ET0 is monotone in Tmax and zero at ΔT=0.

  open

• H-A03

  CUSUM on 1−K_s detects an irrigation cutoff at day 90 without early false alarms.

  open

• H-A04

  Under Doorenbos–Kassam, full irrigation Ya/Ym exceeds rainfed by more than 0.12 on the generated season.

  open

• H-A05

  A +4 °C Tmax bias lifts seasonal ET0 by more than 20 mm versus the unheated twin.

  open

• H-S01

  Every registered instantiation declares primitives from L and a maturity consistent with implementation.

  open

• H-S02

  Operational Z on a damped oscillator is finite, deterministic, and of length 201.

  open

• H-S03

  The becoming loop is an 8-cycle mapping onto the eight spine strata.

  open

• H-S04

  The relation graph has no self-loops, all endpoints exist, and ssl contains lsra.

  open

• H-S05

  L0 leaves have no route; implemented instruments sit above L0 and have routes.

  open

• H-B01

  Schwarzschild r_s is linear in M; photon/r_s = 3/2; ISCO/r_s = 3; r_s(M_☉) ∈ (2940, 2970) m.

  open

• H-B02

  Bekenstein–Hawking entropy obeys the area law: S(2M)/S(M) = 4.

  open

• H-B03

  Hawking temperature scales as 1/M; T_H(M_☉) ∈ (5e-8, 8e-8) K.

  open

• H-B04

  Operational B: κ=1 is identity, κ=0 absorbs, compress(0.5) keeps half the components.

  open

• H-B05

  Filtering an oscillator Twin at κ=0.35 strictly lowers operational mean drift.

  open

• H-B06

  Ideal Casimir pressure is attractive and scales as a⁻⁴.

  open

• H-X01

  The formula engine evaluates Ledger!D2 as tons/acres = 3 and applies packing_factor on D5.

  open

• H-X02

  Ledger!D5 depends on C5, B5, and Assumptions!B1; packing fans out to packed yield cells.

  open

• H-X03

  CUSUM on the yield column alarms between 2023 and 2026.

  open

• H-X04

  Integrity layer flags the unsupported 2027 claim, the hidden packing factor, and a hardcoded forecast cell.

  open

• H-X05

  The Valley Grain audit is deterministic and reconstructs the planted story.

  open

• H-X06

  A line-dictionary Codex reconstructs a repetitive corpus losslessly with ratio > 2, and a high-entropy corpus losslessly without a miracle ratio.

  open

• H-O01

  Organs are uniquely typed with role, consumes, and produces; the human is not an organ; admission is a 10-step contract.

  open

• H-O02

  Every live instrument is cited by at least one organ.

  open

• H-O03

  CODE PRESS remains L0 and does not claim Excel 2030 as the factory.

  open

• H-O04

  The organ graph has no self-loops; emergence returns ideas; the human is at the intake.

  open

• H-O05

  No organ produces measured-evidence; L0 organs are not live.

  open

• H-P01

  With RF off, mean bunch energy is conserved through the compact linac lattice.

  open

• H-P02

  On-crest RF raises mean energy by approximately VRF.

  open

• H-P03

  A focusing quadrupole reduces rms in its focusing plane versus a drift-only lattice.

  open

• H-P04

  WHAT IF QF-1 +2% produces a finite differential in orbit or emittance.

  open

• H-P05

  A tight collimator raises loss versus a wide aperture on the same bunch.

  open

• H-L01

  A tissue twin is deterministic: two runs on the same seed agree after 8 steps.

  open

• H-L02

  Every grid site remains occupied; fate changes, count does not.

  open

• H-L03

  Selective drug pressure reduces malignant count versus untreated on the same seed.

  open

• H-L04

  After a declared injury, regeneration score rises over 28 morphogen steps.

  open

• H-L05

  Gray–Scott B develops spatial variance on a 32² grid.

  open

• H-M01

  A closed six-store hydrologic cycle conserves water; irrigation is a declared source.

  open

• H-M02

  A saturated 1D Darcy column matches q = −K (h_R − h_L)/L.

  open

• H-M03

  A shaded, wetter patch is cooler and more active than an exposed control on the same forcing.

  open

• H-M04

  A declared habitat corridor raises algebraic connectivity of the Fresno mesh.

  open

• H-M05

  Injection above a pressure threshold raises effective permeability and flux versus baseline.

  open

• H-N01

  On a coupled navigator twin, change, growth, and becoming remain distinct.

  open

• H-N02

  Product, geometric, and mean hope formulations are not identical on a non-uniform HV.

  open

• H-N03

  When hope-like variables enter the generator, Model E beats state-only persistence on a held-out split.

  open

• H-N04

  When HV is shuffled on a control world, Model E does not win a miracle over persistence.

  open

• H-N05

  Raising agency at a declared time reduces distance to F* versus a no-op on the coupled twin.

  open

Lineage

Failure registry

Empty on this seed. That is a kernel result, not a license to skip the registry. Failed hypotheses remain part of the research record when they occur.