Frostyverse — H-093 Legal FrostNode nearest-curve audit
============================================================================
version:              1.0.93-legal-frostnode-nearest-curve-audit
patch:                frozen-h089-continuum-h007-preferred-node-vs-nearest-gr-and-fv-no-fit-p0-1
generated:            2026-08-15T20:40:52
python:               3.13.5
platform:             Linux-6.18.35-x86_64-with-glibc2.41
mode:                 VALIDATION H-093
frozen law:           sqrt(current-cell convergence) * sin(theta)^3
frozen K:             8.6816265415871847e-05
node completion:      inherited H-007 probabilistic exit-face capture
audit:                FV-preferred legal node vs nearest H-089 continuum and GR shelf
active R:             (9, 11)
deferred R:           (39, 45) intentionally disabled
reference:            exact snapped fractional-b GR, external scoring only
p:                    0.0 LOCKED
planetary holdout:    sealed

TEST 01 — H-093 contract: inherited node preference vs nearest GR/FV legal shelf
-------------------------------------------------------------------------------------
  frozen law                                             = sqrt(conv_mag) * sin(theta)^3
  frozen K                                               = 8.6816265415871847e-05
  frozen resolution normalization                        = R^2
  active audit resolutions                               = (9, 11)
  R39/R45                                                = DISABLED / reserved for later lean validation
  completed-node law                                     = inherited H-007 exit-face probabilities
  FV_PREFERRED node                                      = max inherited H-007 probability (diagnostic label only)
  broad legal set                                        = seven other corners of same active FrostCell
  strict capture support                                 = H-007 first-exit-face nodes only
  nearest-node rule inserted into FV mechanics           = NO
  GR labels in FV generation                             = NONE
  p                                                       = 0 LOCKED
  planetary/Jupiter holdout                              = SEALED
  TEST 01 VERDICT: PASS/FIREWALL

TEST 02 — construction identities: H-007 preference, legal support, nearest-corner geometry
-------------------------------------------------------------------------------------
  H-007 probability closure                              = 0.000e+00
  H-007 expectation error                                = 1.110e-16
  synthetic modal node == nearest support corner         = True
  exact +axis capture is +travel FrostNode only          = True
  GR used in construction identities                     = NO
  TEST 02 VERDICT: PASS/INSTRUMENT

TEST 03 — GR-CLOSED generation: H-089 continuum + inherited H-007 preferred-node route
-------------------------------------------------------------------------------------
  R=9 building p=0 field ...
    [direct field] h093-node-R9 radius=9.0000 K=10 n=181 nodes=3839 core=0.043 GiB ...
    [direct field] done: equation residual=3.727e-20 time=0.39s
    GR-CLOSED cases=24 | cached deformed nodes=2010 | residual=3.727e-20 | GR labels=SEALED
    internal preferred==FV-nearest support/broad          = 100.000000% / 100.000000%
    mean H-007 probability on FV-nearest support node     = 99.999933268%
    checkpoint saved after R=9
    R=9 field RELEASED before GR scoring
  R=11 building p=0 field ...
    [direct field] h093-node-R11 radius=11.0000 K=10 n=221 nodes=6643 core=0.078 GiB ...
    [direct field] done: equation residual=3.388e-20 time=0.54s
    GR-CLOSED cases=24 | cached deformed nodes=2442 | residual=3.388e-20 | GR labels=SEALED
    internal preferred==FV-nearest support/broad          = 100.000000% / 100.000000%
    mean H-007 probability on FV-nearest support node     = 99.999934467%
    checkpoint saved after R=11
    R=11 field RELEASED before GR scoring
  GR labels opened during node/continuum generation       = NO
  fresh labels used to alter capture or steering          = NO
  TEST 03 VERDICT: PASS/GR-CLOSED

TEST 04 — OPEN exact fractional-b GR ruler: is FV's preferred legal shelf the nearest one?
-------------------------------------------------------------------------------------
  R=9   preferred==GR-nearest support/broad = 100.000000% / 100.000000%
        preferred==FV-nearest support/broad = 100.000000% / 100.000000%
        GR-nearest == FV-nearest support/broad = 100.000000% / 100.000000%
  R=11  preferred==GR-nearest support/broad = 100.000000% / 100.000000%
        preferred==FV-nearest support/broad = 100.000000% / 100.000000%
        GR-nearest == FV-nearest support/broad = 100.000000% / 100.000000%
  TEST 04 VERDICT: PASS/NODE-NEAREST-SCORED

TEST 05 — inherited stochastic capture: how strongly does H-007 favor the nearest shelf?
-------------------------------------------------------------------------------------
  R=9   mean P(GR-nearest support)=99.999933268% | P(FV-nearest)=99.999933268%
        inherited H-007 nonpreferred probability mean/max = 0.667316 / 1.647157 per million
  R=11  mean P(GR-nearest support)=99.999934467% | P(FV-nearest)=99.999934467%
        inherited H-007 nonpreferred probability mean/max = 0.655334 / 1.615612 per million
  TEST 05 VERDICT: PASS/PROBABILITY-MEASURED

TEST 06 — b/orientation/reflection/seed microscope: is any nearest-node failure localized?
-------------------------------------------------------------------------------------
  case groups audited                                    = 12
  worst preferred==GR-nearest-support group              = 100.000000% at None
  TEST 06 VERDICT: PASS/LOCALIZATION-MEASURED

TEST 07 — decision: is FV choosing the best legal shelf around the smooth trajectory?
-------------------------------------------------------------------------------------
  preferred == GR-nearest H-007 support                  = 100.000000000%
  preferred == GR-nearest broad 7-node set               = 100.000000000%
  preferred == FV-nearest H-007 support                  = 100.000000000%
  preferred == FV-nearest broad 7-node set               = 100.000000000%
  GR-nearest == FV-nearest H-007 support                 = 100.000000000%
  GR-nearest == FV-nearest broad 7-node set              = 100.000000000%
  mean inherited H-007 probability on GR-nearest support = 99.999933927%
  expected nonpreferred stochastic captures / million    = 0.660726
  max preferred-vs-best-GR distance penalty (cell widths)= 0.000000e+00
  IMPORTANT                                               = H-007 is probabilistic; 'preferred always nearest' is not the same as 'every random capture always nearest'
  diagnosis                                               = STRONG SUPPORT: FV-PREFERRED H-007 EXIT-FACE NODE IS ALWAYS THE SAME NEAREST SHELF FOR GR AND THE H-089 FV CONTINUUM
  next route                                              = retain FrostNode discreteness; next run the lean H-089 continuum convergence ladder with R39/R45 before changing gravity mechanics
  p                                                       = 0 LOCKED
  planetary/Jupiter holdout                              = SEALED
  TEST 07 VERDICT: PASS/DECISION-RECORDED

H-093 SUMMARY / DECISION
------------------------
  TEST 01: PASS/FIREWALL
  TEST 02: PASS/INSTRUMENT
  TEST 03: PASS/GR-CLOSED
  TEST 04: PASS/NODE-NEAREST-SCORED
  TEST 05: PASS/PROBABILITY-MEASURED
  TEST 06: PASS/LOCALIZATION-MEASURED
  TEST 07: PASS/DECISION-RECORDED

  FV-preferred == GR-nearest H-007 support: 100.000000000%
  FV-preferred == FV-nearest H-007 support: 100.000000000%
  GR-nearest == FV-nearest support:          100.000000000%
  mean P(GR-nearest support):                99.999933927%
  expected stochastic nonpreferred / million: 0.660726
  diagnosis: STRONG SUPPORT: FV-PREFERRED H-007 EXIT-FACE NODE IS ALWAYS THE SAME NEAREST SHELF FOR GR AND THE H-089 FV CONTINUUM
  next route: retain FrostNode discreteness; next run the lean H-089 continuum convergence ladder with R39/R45 before changing gravity mechanics

  Files:
    h093results.txt
    h093_node_choice_summary.csv
    h093_node_choice_misses.csv
    h093_capture_probability_summary.csv
    h093_node_choice_by_case.csv
    h093_seed_control.csv

  H-093 VERDICT: LEGAL FROSTNODE NEAREST-CURVE AUDIT COMPLETE
  R39/R45 were intentionally not run. Planetary/Jupiter holdout remains sealed.

NOTE
----
This R9/R11 run is construction/functional validation only and is NOT scientific evidence.
No H-093 method, threshold, capture rule, or decision logic was changed in response to its GR score.
A separate targeted R25 smoke case also completed 400 transits successfully; its score was not saved as scientific output and did not alter the frozen test.
